vvEPA
United States
Environmental Protection
Agency
Six-Year Review 3 Technical Support
Document for Long-Term 2 Enhanced
Surface Water Treatment Rule

-------
Office of Water (4607M)
EP A-810-R-16-011
December 2016
www. epa. gov/ safewater

-------
Disclaimer
This document is not a regulation. It is not legally enforceable, and does not confer legal rights
or impose legal obligations on any party, including EPA, states, or the regulated community.
While EPA has made every effort to ensure the accuracy of any references to statutory or
regulatory requirements, the obligations of the interested stakeholders are determined by statutes,
regulations or other legally binding requirements, not this document. In the event of a conflict
between the information in this document and any statute or regulation, this document would not
be controlling.

-------
This page intentionally left blank.

-------
Table of Contents
Appendices	iv
List of Exhibits	v
Acronyms	vii
1	Introduction	1-1
1.1	Purpose of This Document	1-1
1.2	Brief History and Overview of the Six-Year Review and Retrospective Review
of Existing Regulations	1-1
1.2.1	Six-Year Review	1-1
1.2.2	Retrospective Review of Existing Regulations	1-2
1.3	Summary of the LT2 Regulatory Review Efforts	1-2
1.4	Other Six-Year Review 3 Efforts	1-4
2	Review Protocol	2-1
3	History of the Long Term 2 Enhanced Surface Water Treatment Rule	3-1
3.1	Statutory Authority	3-1
3.2	Summary of the Rule	3-1
3.2.1	History of the LT2 Promulgation	3-2
3.2.2	Monitoring and Treatment Requirements for Unfiltered Systems	3-9
3.2.3	Requirements for Existing Uncovered Finished Water Reservoirs	3-10
3.2.4	Disinfection Profiling and Benchmarking Requirements	3-10
3.2.5	Implementation Timeline	3-10
4	Health Effects Information	4-1
4.1	Cryptosporidium	4-3
4.1.1	Infectivity	4-4
4.1.2	Morbidity	4-20
4.1.3	Mortality	4-25
4.2	Giardia	4-28
4.3	Viruses	4-30
4.4	Other Pathogens	4-31
4.4.1	Fungi	4-31
4.4.2	Protozoa	4-32
4.4.3	Bacteria	4-32
4.5	Summary	4-35
Six-Year Review 3	i	December 2016
Technical Support Document for LT2

-------
5	Cryptosporidium Analytical Methods Information	5-1
5.1	Performance of Method 1623.1	5-2
5.1.1	Single-Laboratory Side-by-Side Comparison of Method 1623 with
Method 1623.1	5-2
5.1.2	Four-Laboratory Side-by-Side Comparison of Method 1623 with Method
1623.1 	5-4
5.1.3	Fourteen-Laboratory Method 1623.1 Validation Data	5-5
5.1.4	Summary	5-6
5.2	Other Cryptosporidium Detection Techniques and Suggested Improvements	5-7
5.3	Analysis and Recoveries of Cryptosporidium Isolates	5-9
5.4	Conclusion	5-9
6	Occurrence and Exposure Information	6-1
6.1	Cryptosporidium	6-3
6.1.1	Summary of Round 1 Occurrence Data	6-3
6.1.2	Predictive Modeling for Round 2	6-9
6.2	E. coli Indicator to Predict Cryptosporidium Occurrence	6-15
6.2.1	Background	6-15
6.2.2	Data Cleaning Process	6-16
6.2.3	Analysis	6-19
6.2.4	Results and Discussion	6-26
6.3	Cooccurrence of Cryptosporidium and Other Pathogens of Concern	6-31
6.3.1 Giardia and Cryptosporidium Cooccurrence from ICR Supplemental
Survey Data	6-31
6.4	Summary	6-35
7	LT2 Microbial Toolbox and Other Tools	7-1
7.1	Summary of Data on Toolbox Options and Treatment Credits	7-2
7.2	Treatment Technology Usage	7-7
7.3	Microbial Toolbox Tools	7-8
7.3.1	Watershed Control Program	7-8
7.3.2	Alternative Source/Intake Management	7-12
7.3.3	Presedimentation Basin with Coagulation	7-13
7.3.4	Two-stage Lime Softening	7-14
7.3.5	Bank Filtration	7-15
7.3.6	Combined Filter Performance	7-18
7.3.7	Individual Filter Performance	7-19
Six-Year Review 3	ii	December 2016
Technical Support Document for LT2

-------
7.3.8	Demonstration of Performance of Treatment Process(es)	7-20
7.3.9	Bag or Cartridge Filters	7-22
7.3.10	Membrane Filtration	7-23
7.3.11	Second Stage Filtration	7-27
7.3.12	Slow Sand Filtration	7-28
7.3.13	Chlorine Dioxide	7-30
7.3.14	Ozone	7-31
7.3.15	UV Disinfection	7-33
7.4 Summary	7-44
8	Uncovered Finished Water Reservoirs	8-1
8.1	Background on the LT2 Uncovered Finished Water Reservoir Requirements	8-1
8.2	Background on Uncovered Finished Water Reservoirs	8-1
8.3	Summary of Information Supporting the LT2 Requirements	8-2
8.4	Information Available Since the Promulgation of the LT2	8-3
8.4.1	Permanent Solutions Taken	8-5
8.4.2	Effectiveness of Permanent Solutions	8-5
8.4.3	Temporary Solutions Taken	8-5
8.4.4	Effectiveness of Temporary Solutions	8-6
8.5	Implementation Issues Related to the LT2 Cover/Treat Requirements	8-6
9	References	9-1
Six-Year Review 3	Hi	December 2016
Technical Support Document for LT2

-------
Appendices
Appendix A: Data for Methods 1623 and 1623.1 Cryptosporidium Recoveries
Appendix B: Occurrence and Exposure
Appendix C: Toolbox Option Usage and Related Implementation Issues
Six-Year Review 3
Technical Support Document for LT2
iv
December 2016

-------
List of Exhibits
2	Review Protocol
Exhibit 2.1 Six-Year Review Protocol Overview and Major Categories of
Revise/Take No Action Outcomes	2-2
3	History of the Long Term 2 Enhanced Surface Water Treatment Rule
Exhibit 3.1 Bin Classifications and Treatment Requirements for Filtered Systems	3-6
Exhibit 3.2 Microbial Toolbox Components for the LT2	3-8
Exhibit 3.3 Implementation Timeline for the LT2 for Filtered Systems	3-11
4	Health Effects Information
Exhibit 4.1 Characteristics of the LT2 EA Primary Model and Six Alternative
Models	4-5
Exhibit 4.2 Dose Response Relation, as a Function of Dose and IgG Level	4-10
Exhibit 4.3 A: Dose Response Relations for the Four Isolates (TAMU, Iowa, UCP
and Moredun); B: Quantile Contours of the Predicted Dose Response
Relation Generalized from the Four Curves in A; C: Low Dose
Extrapolated Dose Response Relations for the Four Isolates	4-12
Exhibit 4.4 Best Fit Models and Optimized Parameter Values for Cryptosporidium
Isolates (CAMRA)	4-13
Exhibit 4.5 Fitting Dose-Response Curves of Infection Probability for Healthy Adult
Volunteers and Intake of Cryptosporidium Oocysts	4-14
Exhibit 4.6 Cryptosporidium Outbreaks Associated with Drinking Water, by Year:
Waterborne Disease and Outbreak Surveillance System, United States
2005-2010 (CDC, 2008; 2011; 2013; 2015a)	4-24
Exhibit 4.7 Giardia Outbreaks Associated with Drinking Water, by Year:
Waterborne Disease and Outbreak Surveillance System, United States
2005-2010 (CDC, 2011; 2013)	4-29
5	Cryptosporidium Analytical Methods Information
Exhibit 5.1 Observed Recovery at a Single Laboratory, Using One Source Water and
Three Artificial Matrices	5-3
Exhibit 5.2 Observed Recovery at a Single Laboratory, Using Nine Source Waters	5-4
Exhibit 5.3 Observed Recovery at Four Laboratories, Using Three Source Waters	5-4
Exhibit 5.4 Method 1623.1 Validation Data for 14 Laboratories, Reagent Water; N =
56	5-6
Exhibit 5.5 Method 1623.1 Validation Data for 14 Laboratories, Source Water; N =
53	5-6
6	Occurrence and Exposure Information
Exhibit 6.1 System Size and Round 1 Sampling Schedule	6-1
Exhibit 6.2 Cryptosporidium Round 1 Monitoring Participation	6-5
Exhibit 6.3 Cryptosporidium Round 1 Summary Statistics	6-5
Exhibit 6.4 Cryptosporidium Round 1 Summary Statistics by Plant	6-5
Exhibit 6.5 Cryptosporidium Round 1 Summary Statistics by Source Water Type	6-6
Six-Year Review 3
Technical Support Document for LT2
v
December 2016

-------
Exhibit 6.6 Binning Results for Filtered Systems > 10,000 People	6-7
Exhibit 6.7 Binning Results for Plants in Grandfathered Systems > 10,000 People	6-9
Exhibit 6.8 Modeled Round 2 Outcomes Using Method 1623, by Source Water Type .... 6-12
Exhibit 6.9 Modeled Round 2 Outcomes Using Method 1623.1, by Source Water
Type	6-13
Exhibit 6.10 Modeled Round 2 Outcomes by Source Water Type and Method	6-14
Exhibit 6.11 Plants in Bins 2-4 under Alternative Scenarios on Occurrence
Distribution	6-15
Exhibit 6.12 Definition of Variables Used in Analysis	6-19
Exhibit 6.13 Criteria for Reservoirs and Lakes Using Original Cleaning Procedure	6-20
Exhibit 6.14 Criteria for Rivers and Flowing Streams Using Original Cleaning
Procedure	6-21
Exhibit 6.15 Criteria for All Samples Using Original Cleaning Procedure	6-22
Exhibit 6.16 Criteria for Reservoirs and Lakes Using Revised Cleaning Procedure	6-23
Exhibit 6.17 Criteria for Rivers and Streams Using Revised Cleaning Procedure	6-24
Exhibit 6.18 Criteria for All Samples Using Revised Cleaning Procedure	6-25
Exhibit 6.19 E. coli Trigger Analysis Results for Small Plants Using Original
Cleaning Procedures	6-27
Exhibit 6.20 E. coli Trigger Analysis Results for Small Plants Using Revised
Cleaning Procedures	6-28
Exhibit 6.21 2010 Trigger Analysis Results	6-29
Exhibit 6.22 Percent Reduction in Plants Required to Monitor with Alternate Trigger
Levels	6-30
Exhibit 6.23 Distribution of ICRSS Source Waters by System Size and Water Type	6-31
Exhibit 6.24 ICRSS Summary Statistics for Cryptosporidium and Giardia	6-31
Exhibit 6.25 Scatterplot of Observed Mean Concentrations for 87 Source Waters	6-32
Exhibit 6.26 Deviance Information Criterion Model Results	6-33
7	LT2 Microbial Toolbox and Other Tools
Exhibit 7.1 Summary of New Information on the LT2 Microbial Toolbox Options	7-3
Exhibit 7.2 Comparative Effectiveness of Tools for Different Organisms	7-4
Exhibit 7.3 Microbial Toolbox Tool Usage	7-8
Exhibit 7.4 Summary of UV Findings	7-38
8	Uncovered Finished Water Reservoirs
Exhibit 8.1 Systems with Remaining UCFWRs as of December 2015	8-2
Technical Support Document for LT2

-------
Acronyms
AIDS
Acquired Immunodeficiency Syndrome
ASDWA
Association of State Drinking Water Administrators
AWOP
Area-Wide Optimization Program
AWWA
American Water Works Association
BAT
Best Available Technology
BCDPW
Baltimore City Department of Public Works
BF
Bank Filtration
CAMRA
Center for Advanced Microbial Risk Assessment
CCL
Contaminant Candidate List
CDC
Centers for Disease Control and Prevention
CFE
Combined Filter Effluent
CFSE
Carboxyfluorescein Diacetate Succinimidyl Ester
CFU
Colony-Forming Units
CT
the product of the residual disinfectant concentration "C" in

milligrams/liter (mg/L) and contact time "T" in minutes
CWRWS
Central Wyoming Regional Water System
DBP
Disinfection Byproducts
D/DBPR
Disinfectants/Disinfection Byproducts Rule
DCTS
Data Collection and Tracking System
DEC
Decimal Elimination Capacity
DIC
Deviance Information Criterion
DMS
Dyed Microsphere
DNA
Deoxyribonucleic Acid
DOC
Dissolved Organic Carbon
DOP
Demonstration of Performance
EA
Economic Analysis
EHEC
Enterohemorrhagic E. coli
EMC
Event Mean Concentration
EO
Executive Order
EPA
United States Environmental Protection Agency
FA
Factor Analysis
FAC
Federal Advisory Committee
FCV
Feline Calicivirus
FDM-MPN
Focus Detection Method-Most-Probable-Number
FPUD
Fallbrook Public Utility District
FR
Federal Register
FS
River/Stream
GAC
Granular Activated Carbon
gpm
Gallons Per Minute
GWUDI
Ground Water Under the Direct Influence
HEA
Health Effects Assessment
HESD
Health Effects Support Document
HPC
Heterotrophic Plate Count
ICR
Information Collection Rule
Six-Year Review 3
Technical Support Document for LT2
vii
December 2016

-------
ICRSS
Information Collection Rule Supplemental Survey
ID50
Median Infective Dose
IDEM
Indiana Department of Environmental Management
IESWTR
Interim Enhanced Surface Water Treatment Rule
IFA
Immunofluorescent Antibody and Microscopy Assay
IFE
Individual Filter Effluent
IgG
Immunoglobulin G
IMS
Immunomagnetic Separation
IRIS
Integrated Risk Information System
JAGS
Just Another Gibbs Sampler
LADWP
Los Angeles Department of Water and Power
LED
Light Emitting Diode
LP
Low Pressure
LPHO
Low Pressure, High Output
LR
Lake/Reservoir
LRV
Log Removal Values
LT1
Long Term 1 Enhanced Surface Water Treatment Rule
LT2
Long Term 2 Enhanced Surface Water Treatment Rule
MDBP
Microbial and Disinfection Byproducts
MAC
Mycobacterium avium Complex
MCL
Maximum Contaminant Level
MCLG
Maximum Contaminant Level Goal
MCMC
Markov Chain Monte Carlo
MDBK
Madin-Darby Bovine Kidney
MF
Microfiltration
mgd
Millions of Gallons per Day
mg/L
Milligrams/Liter
mL
Milliliters
MLE
Maximum-Likelihood Estimation
MP
Medium Pressure
MPA
Microscopic Particulate Analysis
MRAA
Maximum Running Annual Average
MRSA
Methicillin-Resistant Staphylococcus Aureus
NaHMP
Sodium Hexametaphosphate
NDWAC
National Drinking Water Advisory Council
NF
Nanofiltration
NOM
Natural Organic Matter
NPDWR
National Primary Drinking Water Regulation
NTU
Nephelometric Turbidity Units
NYCDEP
New York City Department of Environmental Protection
O&M
Operations and Maintenance
OPP
Office of Pesticide Programs
OW
Office of Water
PAC
Powdered Activated Carbon
PCB
Pathogen Catchment Budget
PCR
Polymerase Chain Reaction
Six-Year Review 3
Technical Support Document for LT2
December 2016

-------
PMA
Propidium Monoazide
POU
Point-Of-Use
PQL
Practical Quantitation Limit
PSL
Polystyrene Latex
PT
Proficiency Testing
PUV
Pulsed UV
PWS
Public Water System
PWSA
Pittsburgh Water and Sewer Authority
QA
Quality Assurance
QC
Quality Control
QMRA
Quantitative Microbial Risk Assessment
qPCR
Quantitative PCR
RAA
Running Annual Average
RED
Reduction Equivalent Dose
RMP
Risk Mitigation Plan
RNA
Ribonucleic Acid
RO
Reverse Osmosis
RNA
Ribonucleic Acid
rRNA
Ribosomal Ribonucleic Acid
RT-PCR
Reverse Transcription-PCR
SCADA
Supervisory Control and Data Acquisition
SDWA
Safe Drinking Water Act
SDWIS
Safe Drinking Water Information System
SPU
Seattle Public Utilities
SSRC
Spores of Sulfite-Reducing Clostridia
SWTR
Surface Water Treatment Rule
SYR1
Six-Year Review 1
SYR2
Six-Year Review 2
SYR3
Six-Year Review 3
TCR
Total Coliform Rule
THM
Trihalomethane
TNRS
Tennessee River Sediment
TOC
Total Organic Carbon
TPU
Tacoma Public Utilities
TSS
Total Suspended Solids
TT
Treatment Technique
UCFWR
Uncovered Finished Water Reservoirs
UF
Ultrafiltration
UPS
Uninterruptible power supply
USEPA
United States Environmental Protection Agency
UV
Ultraviolet
UVA
Ultraviolet A
UVDGM
Ultraviolet Disinfection Guidance Manual
WBDO
Waterborne Disease Outbreaks
WCP
Watershed Control Program
WHO
World Health Organization
Six-Year Review 3
Technical Support Document for LT2
ix
December 2016

-------
WTP
Water Treatment Plant
Six-Year Review 3	x	December 2016
Technical Support Document for LT2

-------
1 Introduction
1.1	Purpose of This Document
The purpose of this document is to present technical information the U.S. Environmental
Protection Agency (EPA) analyzed as part of the ongoing Six-Year Review 3 (SYR3) of the
Long Term 2 Enhanced Surface Water Treatment Rule (LT2) as well as the Retrospective
Review of the LT2 under Executive Order 13563. The Agency used the information presented in
this document to formulate its determination of whether it would consider any changes to the
LT2.
This introduction provides an overview of Six-Year Review and Retrospective Review
requirements, and a summary of the SYR3 and Retrospective Review of the LT2. The end of this
introductory section presents a brief overview of the content of the remaining chapters in this
document.
1.2	Brief History and Overview of the Six-Year Review and Retrospective Review of
Existing Regulations
1.2.1 Six-Year Review
Section 1412(b)(9) of the 1996 Safe Drinking Water Act (SDWA) Amendments requires EPA to
conduct a review of each existing National Primary Drinking Water Regulation (NPDWR) at
least once every six years and revise each as appropriate. Additionally, the SDWA specifies that
any revision of a NPDWR "shall maintain, or provide for greater, protection of the health of
persons." To date, EPA has completed two rounds of Six-Year Reviews, referred to as the Six-
Year Review 1 (SYR1) and the Six-Year Review 2 (SYR2). The EPA Administrator signed the
notice announcing the results of the SYR1 on July 11, 2003, and the notice was published in the
Federal Register (FR) on July 18, 2003 (USEPA, 2003a). The EPA Administrator signed the
notice announcing the results of the SYR2 on December 17, 2009, and the notice was published
in the FR on March 29, 2010 (USEPA, 2010a).
A decision to revise an NPDWR starts a regulatory process that involves more detailed analyses
concerning health effects, costs, benefits, contaminant occurrence and other topics. At any point
in this process, EPA may find that regulatory revisions are not appropriate and may discontinue
regulatory revision efforts. Review of that NPDWR would, however, remain part of future Six-
Year Reviews. Similarly, a determination to "take no action at this time" means only that EPA
does not believe that regulatory changes to a particular NPDWR are appropriate at the current
time based on health effects, analytical methods, treatment data, ongoing scientific reviews,
priority or other reasons. EPA may decide in future Six-Year Reviews that regulatory changes
are appropriate (USEPA, 2009a).
Under the SYR1, EPA identified only the Total Coliform Rule (TCR) as a candidate NPDWR
for revisions, while EPA identified four NPDWRs under the SYR2 as candidates for revision:
acrylamide, epichlorohydrin, tetrachloroethylene and trichloroethylene.
Six-Year Review 3
Technical Support Document for LT2
1-1
December 2016

-------
1.2.2 Retrospective Review of Existing Regulations
In August 2011, EPA published its final plan for conducting periodic retrospective reviews of
existing regulations, prepared in response to Executive Order 13563. The order required each
federal agency to develop a plan "consistent with law and its resources and regulatory priorities.
Under the final plan, the Agency will periodically review its existing significant regulations to
determine whether any such regulations should be modified, streamlined, expanded or repealed
so as to make the Agency's regulatory program more effective or less burdensome in achieving
the regulatory objectives" (USEPA, 201 la). In its plan, EPA identified 35 regulations, including
the LT2, for inclusion in the first round of retrospective reviews. The plan stated that, "EPA
intends to evaluate effective and practical approaches that may maintain, or provide greater
protection of, the water treated by public water systems (PWSs) and stored prior to distribution
to consumers. EPA plans to conduct this review expeditiously to protect public health while
considering innovations and flexibility as called for in Executive Order (EO) 13563." (USEPA,
2011a)
This Agency-wide effort, separate from the SDWA Six-Year Review process, aims to better
understand the impacts of its regulations and, as noted above, identify ways to improve and make
them less burdensome.
EPA completed its detailed review of the LT2 and at this time believes that it is not a candidate
for regulatory revision.
1.3 Summary of the LT2 Regulatory Review Efforts
As part of the LT2 regulatory review, EPA assessed and analyzed information presented in this
document regarding health effects and risks, monitoring methods, occurrence, the use of E. coli
as a screen for small systems, the microbial toolbox and uncovered finished water reservoirs
(UCFWRs), to evaluate whether there are new or additional ways to manage risk while assuring
equivalent or improved public health protection.
EPA has developed and implemented protocols for ensuring a systematic approach is taken to
conduct each of the Six-Year Reviews. EPA carried out an initial assessment of the application
of the current protocol to the LT2 SYR3; EPA presents its consideration of the Six-Year Review
Protocol Decision Tree for the LT2 and explains how EPA mapped the protocol to this LT2
Technical Support Document in Chapter 2.
EPA provides in Chapter 3 a history of the development of the LT2, a summary of the LT2
requirements, and information on the statutory authority EPA used to develop the LT2.
EPA evaluated available information on Cryptosporidium and other pathogens of concern that
could potentially be present in source waters and UCFWRs. As part of this analysis, EPA noted
new information on Cryptosporidium species that have recently been linked to human infection
as part of this analysis. EPA presents a more complete discussion of potential pathogens of
concern and health effects in Chapter 4 of this document.
In January 2012 EPA published a revision to Method 1623, the method used to determine the
source water occurrence of Cryptosporidium and Giardia. Method 1623.1 encompasses
Six-Year Review 3
Technical Support Document for LT2
1-2
December 2016

-------
improvements to Method 1623 regarding separation of Cryptosporidium oocysts from extraneous
material and the removal of interfering substances. Method 1623.1 has been shown to increase
recovery efficiencies for Cryptosporidium oocysts in some complex sample matrices. As part of
this review, EPA determined that the impact of Method 1623.1 was not sufficient to justify
requiring systems to use this method, but that systems could use it for compliance with the
additional source water monitoring (e.g., a second round required under the LT2). EPA also
evaluated other methods for determining the occurrence of Cryptosporidium in water samples,
including polymerase chain reaction (PGR) and cell culture methods. Because of concerns with
these methods (e.g., cell culture would detect some, but not all Cryptosporidium species), EPA
believes that the use of these methods for LT2 compliance could amount to backsliding, which
the SDWA specifically prohibits (i.e.. Section 1412(b)(9) requires that revisions maintain or
provide for greater protection of health). EPA provides more detail on these analyses in Chapter
5 of this document.
EPA conducted analyses on the LT2 source water monitoring requirements, and the
Cryptosporidium occurrence data collected during the first round of monitoring, to determine the
value of proceeding with the second round. While the data generated by the first round of source
water monitoring indicated a lower occurrence rate than EPA predicted when it promulgated the
LT2, EPA believes there is value in conducting the second round of monitoring to capture
temporal changes in source water Cryptosporidium occurrence. To support its determination of
the value of the second round of monitoring, EPA developed predictions of the numbers of
additional systems that would be assigned to treatment bins as a result of Round 2 monitoring.
EPA provides more detail on these analyses in Chapter 6 of this document.
Systems serving 100,000 or more people are defined as Schedule 1 systems, while systems
serving 50,000 to 99,999 people, and those serving 10,000 to 49,999 people are defined as
Schedule 2 and Schedule 3 systems, respectively. EPA analyzed the Cryptosporidium and E. coli
data collected by Schedule 1-3 systems during Round 1 monitoring to determine the usefulness
of E. coli as a screen for Schedule 4 (small) systems, that is, to determine whether coli
occurrence correlated with Cryptosporidium occurrence. EPA determined that coli is an
effective screen for determining the need for Schedule 4 systems to conduct Cryptosporidium
monitoring. As a result, EPA will continue to allow Schedule 4 systems to use E. coli as a screen
for Cryptosporidium monitoring for the second round of LT2 monitoring. Where E. coli
concentrations are below certain levels, Cryptosporidium monitoring will not be required for
small systems. EPA provides more detail on this analysis in Chapter 6 of this document.
Based on the source water monitoring results, systems are placed into one of four categories of
additional treatment requirement (i.e., bins). Systems in Bin 1 require no additional treatment
beyond existing requirements. Systems in Bins 2, 3 or 4 select from a microbial toolbox of
options for ensuring Cryptosporidium source protection and management, removal or
inactivation. EPA reviewed and analyzed available information on the microbial toolbox tools
and other risk mitigation strategies to determine the need to revise the toolbox credits awarded
for the use of the tools, and to determine the need to change the tools available in the microbial
toolbox. Based on this analysis, EPA determined that no changes are warranted to the credits
awarded to the tools in the microbial toolbox, because limited data exists that would support any
changes. Also, stakeholders identified and provided feedback to EPA on implementation issues
related to some of the tools from the microbial toolbox. EPA believes there is enough flexibility
Six-Year Review 3
Technical Support Document for LT2
1-3
December 2016

-------
in the microbial toolbox that changes to the available tools are not necessary to address these
implementation concerns. EPA provides more detail on these analyses in Chapter 7 of this
document.
The LT2 included disinfection profiling and benchmarking requirements that apply if a PWS
proposes to make a significant change to its disinfection practice as it implements the Stage 2
Disinfection Byproducts Rule and the LT2. EPA analyzed new disinfection studies to inform a
determination of whether existing CT (the product of the residual disinfectant concentration "C"
in milligrams/liter (mg/L) and contact time "T" in minutes) tables are still protective. EPA
provides a detailed discussion of the results and implications of these more recent studies in the
Six-Year Review 3 Technical Support Document for Microbial Contaminant Regulations
(USEPA, 2016a).
To review the UCFWR requirements of the LT2, EPA collected information related to the
potential risks posed by UCFWRs, and measures taken to address those risks. EPA also received
stakeholder feedback during a public meeting devoted to UCFWRs. Based on the available
information, EPA was unable to identify any alternative risk mitigation measures that are as
effective as the cover or treat requirements of the LT2. EPA also notes that many PWSs have
already addressed their UCFWRs. As a result, EPA believes that the cover or treat requirements
that pertain to UCFWRs continue to be appropriate. EPA provides more detail on these analyses
in Chapter 8 of this document.
EPA held three public meetings as part of the LT2 regulatory review. On December 7, 2011,
EPA hosted a public meeting to discuss improvements to the Cryptosporidium analytical method
and to provide an update on the LT2's source water monitoring results. The two main objectives
of this meeting were: (1) to start the LT2 regulatory review process, and (2) to meet the
recommendation of the microbial and disinfection byproducts (MDBP) federal advisory
committee (FAC) to have public meetings following the first round of monitoring. On April 24,
2012, EPA hosted a second public meeting to discuss information that could inform the
regulatory review of the LT2 UCFWR requirement. The main objectives of this meeting were to:
(1)	provide background information on the UCFWR requirement and discuss/solicit public input,
(2)	provide an overview of the SYR3 process, and (3) engage in a scientific and technical
discussion on data and information related to occurrence of Cryptosporidium, Giardia, viruses
and other pathogens/indicators in UCFWRs; perspectives on public health risks; strategies to
control or remove contaminants in UCFWRs; and potential assessment. On November 15, 2012,
EPA hosted a third public meeting to discuss the review process, monitoring, occurrence,
binning and microbial toolbox information. Summaries of these three public meetings, along
with presentations made during the meetings, are available at
www.epa.uov/d wsi x vearreview/review-1 t2-rule . Findings from the meetings are also discussed
in more detail in applicable chapters of this document.
1.4 Other Six-Year Review 3 Efforts
The SYR3 examines rules that address chemical contaminants/indicators, radiological
contaminants, microbiological contaminants/indicators, disinfection byproducts
(DBPs)/indicators and disinfectant residuals. In addition to the LT2, the specific regulations
under review include the Chemical Phase Rules (Inorganic, Synthetic Organic and Volatile
Six-Year Review 3
Technical Support Document for LT2
1-4
December 2016

-------
Organic Chemicals); the Surface Water Treatment Rule (SWTR); the Interim Enhanced Surface
Water Treatment Rule (IESWTR); the Long Term 1 Enhanced Surface Water Treatment Rule
(LT1); the Ground Water Rule; the Filter Backwash Recycling Rule; the Stage 1
Disinfectants/Disinfection Byproducts Rule (D/DBPR); and the Stage 2 D/DBPR. This
document covers the review of only the LT2; the reviews of other regulations included within the
SYR3 are described in separate documents.
Six-Year Review 3
Technical Support Document for LT2
1-5
December 2016

-------
2 Review Protocol
This chapter provides an overview of the process the Agency used to review the National
Primary Drinking Water Regulations (NPDWRs) discussed in the Six-Year Review 3. The
protocol document, "EPA Protocol for the Third Review of Existing National Primary Drinking
Water Regulations," contains a detailed description of the process the Agency used to review the
NPDWRs (USEPA, 2016b). The foundation of this protocol was developed for the Six-Year
Review 1 based on the recommendations of the National Drinking Water Advisory Council
(NDWAC) (2000). This Six-Year Review 3 process is very similar to the process implemented
during the Six-Year Review 1 and the Six-Year Review 2, with some clarifications to the
elements related to the review of NPDWRs included in the MDBP rules.
Exhibit 2.1 presents an overview of the Six-Year Review protocol and major categories of
review outcomes. The protocol is broken down into a series of questions about whether there is
new information for a contaminant that suggests it is appropriate to revise one or more of the
NPDWRs. The two major outcomes of the detailed review are either:
(1)	the NPDWR is not appropriate for revision and no action is necessary at this time, or
(2)	the NPDWR is a candidate for revision.
Individual regulatory provisions of NPDWRs that are evaluated as part of the Six-Year Review
are: maximum contaminant level goals (MCLGs), maximum contaminant levels (MCLs),
maximum residual disinfectant level goals (MRDLGs), maximum residual disinfectant levels
(MRDLs), treatment techniques, other treatment technologies and regulatory requirements (e.g.,
monitoring). The MCL provisions are not applicable for evaluation of the microbial
contaminants regulations which establish treatment technique requirements in lieu of MCLs. The
MRDLG and MRDL provisions are only applicable for evaluation of the DBP rules as part of the
Six-Year Review.
The review elements that EPA considered for each NPDWR during the Six-Year Review 3
include the following: initial review, health effects, analytical feasibility, occurrence and
exposure, treatment feasibility, risk balancing, and other regulatory revisions. Further
information about these review elements are described in the protocol document (USEPA,
2016b).
Six-Year Review 3
Technical Support Document for LT2
2-1
December 2016

-------
Exhibit 2.1 Six-Year Review Protocol Overview and Major Categories of
Revise/Take No Action Outcomes
Yes
No
Yes
No
No
Outcome:
No action
at this time
Uncertain - emerging
information
Yes
No
Yes
No
Ongoing or planned HEA
No new information
NPDWR reviewed in recent or ongoing action?
Regulatory action ongoing
or recently completed
NPDWRs Under Review
Data gaps/emerging
information
Low priority - No meaningful
opportunity
Health effects assessment (HEA)
in process or planned? *
Data sufficient to support
regulatory revision?
New information to suggest possible changes (i.e.,
to an MCLG, MCL, Treatment Technique and/or
other regulatory revisions)?
Meaningful opportunity for health risk reduction for
persons served by PWS and/or cost savings while
maintaining/improving public health protection?
* Contaminants with an HEA in process that have an MCL based on practical
quantitation limit and are greater than MCLG are passed to the next question to
evaluate potential to revise the MCL.
The Initial Review branch of the protocol identifies NPDWRs with recent or ongoing actions and
excludes them from the review process to prevent duplicative agency efforts (USEPA, 2016b).
The cutoff date for the NPDWRs reviewed under the SYR3 was August 2008. Based on the
Initial Review, EPA excluded the Aircraft Drinking Water Rule, which was promulgated in
2009, and the Revised Total Coliform Rule (RTCR) (the revision of the 1989 TCR), which was
promulgated in 2013. Further, since most of the 1989 Total Coliform Rule (TCR) requirements
were replaced by the 2013 RTCR, the 1989 TCR was excluded from the Six-Year Review.
Outcome:
Candidate
for Revision
Six-Year Review 3
Technical Support Document for LT2
2-2
December 2016

-------
3 History of the Long Term 2 Enhanced Surface Water Treatment Rule
This chapter provides a brief history of the statutory authority EPA used to develop the Long
Term 2 Enhanced Surface Water Treatment Rule (LT2) and of the development of the LT2, as
well as description of the regulatory requirements that are part of the LT2.
3.1	Statutory Authority
The 1974 Safe Drinking Water Act (SDWA) authorized EPA to protect public health by
regulating the nation's public drinking water supply. Although the SDWA was amended slightly
in 1977, 1979 and 1980, the most significant changes occurred when the SDWA was
reauthorized in 1986 and amended in 1996. To safeguard public health, the 1986 amendments
required EPA to set maximum contaminant level goals (MCLGs) and maximum contaminant
levels (MCLs) for 83 contaminants. The 1986 amendments authorized EPA to require treatment
techniques (TTs) instead of MCLs where appropriate. EPA was also required to establish
regulations for disinfection of all public water supplies and to specify filtration requirements for
water systems that draw their water from surface sources (US EPA, 1991). The disinfection and
filtration requirements were intended to protect the public from potential adverse health effects
due to exposure to Giardia lamblia, viruses, Legionella, heterotrophic bacteria and other
pathogens that would be removed by those TTs.
3.2	Summary of the Rule
EPA convened the Stage 2 MDBP federal advisory committee (FAC) in March 1999 to evaluate
new information and develop recommendations for the LT2 and the Stage 2
Disinfectants/Disinfection Byproducts Rule (Stage 2 D/DBPR). The FAC was comprised of
representatives from EPA, state and local public health and regulatory agencies, local elected
officials, Indian Tribes, drinking water suppliers, chemical and equipment manufacturers, and
public interest groups. The FAC members signed an Agreement in Principle in September 2000
(USEPA, 2000) stating consensus recommendations of the group. For the LT2, the Committee
recommended the following:
(1)	Supplemental risk-targeted Cryptosporidium treatment by filtered public water
systems (PWSs) with higher source water contaminant levels as shown by monitoring
results,
(2)	Cryptosporidium inactivation by all unfiltered PWSs, which must meet overall
treatment requirements using a minimum of two disinfectants,
(3)	A toolbox of treatment and control processes for PWSs to comply with
Cryptosporidium treatment requirements,
(4)	Reduced monitoring burden for small filtered PWSs,
(5)	Future monitoring to confirm or revise source water quality assessments, and
(6)	Development of guidance for ultraviolet (UV) disinfection and other toolbox
components.
Cover or treat existing uncovered finished water reservoirs (UCFWRs) or implement risk
mitigation plans (RMPs).
Six-Year Review 3
Technical Support Document for LT2
3-1
December 2016

-------
The primary intent of the LT2 is to supplement existing microbial treatment requirements for
systems where additional public health protection is needed due to elevated source water
Cryptosporidium concentrations. The LT2 requires filtered systems to monitor their source water
for Cryptosporidium and/or Escherichia coli during two different rounds of monitoring. Each
round of monitoring lasts one to two years, depending on system size. Systems must conduct a
second round of monitoring to determine if source water conditions changed significantly.
Larger systems (those serving at least 10,000 people) are required to conduct their second round
of monitoring six years after submitting their bin calculation, approximately 6 V2 years after
completing their initial round of monitoring. Filtered systems serving fewer than 10,000 people
may have as many as eight years after completing their first round to begin their second round of
source water monitoring. While larger systems must monitor their source water for
Cryptosporidium, smaller systems (those serving fewer than 10,000 people) can monitor for is.
coli first; if their E. coli levels exceed established thresholds, the smaller systems must then
conduct Cryptosporidium monitoring. Systems may opt out of the source water monitoring
requirements of the LT2 if they provide 5.5-log Cryptosporidium inactivation or removal. Based
on the Cryptosporidium results, filtered systems must meet one of four levels of treatment for
Cryptosporidium (with the first level requiring no additional treatment). Unfiltered systems must
also monitor for Cryptosporidium; as with filtered systems the duration of each round of
monitoring depends on system size. All unfiltered systems must achieve 2-log (99 percent) or 3-
log (99.9 percent) Cryptosporidium inactivation, depending on their source water
Cryptosporidium levels. The LT2 also requires systems with UCFWRs either to cover the
reservoirs or provide additional treatment to the water exiting the reservoir. The LT2's
provisions are described in more detail below. The first round of source water monitoring is
complete. The second round began in 2015.
Most of the requirements in the final LT2 reflect consensus recommendations from the Stage 2
MDBP FAC. However, EPA did not include provisions for RMPs for UCFWRs in the final LT2
because EPA determined (after reviewing public comments on the Proposed LT2 (USEPA,
2003b)) that an RMP would not provide equivalent public health protection to covering or
treating the UCFWRs. Consequently, an RMP would not meet the statutory provision for a TT to
prevent adverse health effects from pathogens like Giardia and Cryptosporidium to the extent
feasible (SDWA section 1412(b)(7)(A)).
3.2.1 History of the LT2 Promulgation
EPA promulgated the final LT2 requirements on January 5, 2006.1 The LT2 built upon the
Interim Enhanced Surface Water Treatment Rule (IESWTR) and the Long Term 1 Enhanced
Surface Water Treatment Rule (LT1) by improving control of microbial pathogens, specifically
the contaminant Cryptosporidium.2 The LT2, in conjunction with the Stage 2 D/DBPR,3
addresses the trade-off between competing risks that are posed by the simultaneous control of
microbial pathogens and DBPs. The disinfectants commonly used to kill microorganisms react
1	LT2 (USEPA, 2006a).
2	IESWTR (USEPA, 1998), LT1 (USEPA, 2002).
3	Stage 2 D/DBPR (USEPA, 2006b).
Six-Year Review 3
Technical Support Document for LT2
3-2
December 2016

-------
with naturally occurring organic and inorganic matter in source water, forming DBPs that present
potential risks for cancer and reproductive and developmental health effects. In order to balance
the risks posed by DBPs and microbial pathogens, and to make it easier for water systems to
comply with both rules, EPA promulgated the LT2 concurrently with the Stage 2 D/DBPR. The
LT2 applies to all PWSs (i.e., 15 service connections or 25 people served for at least 60 days per
year) that use surface water or ground water under the direct influence of surface water
(GWUDI) as a source.
3.2.1.1	Monitoring and Treatment Requirements for Filtered Systems
Under the LT2, systems serving 10,000 or more people were required to first monitor source
water Cryptosporidium concentrations. The required level of source protection, removal or
inactivation increases for systems in higher bins (i.e., increasing source water concentrations of
Cryptosporidium).
The LT2 includes source water monitoring and bin classification exemptions for all filtered
systems that provided, or will provide, 5.5-log treatment4 for Cryptosporidium by the date on
which they are required to comply with additional Cryptosporidium treatment requirements. To
meet the requirement for 5.5-log treatment, systems using conventional treatment, diatomaceous
earth filtration or slow sand filtration must provide 2.5-log additional treatment beyond what
they are assumed to currently provide, and systems using direct filtration must provide 3-log
additional treatment. The requirements for 5.5 log treatment for those systems using alternative
filtration technologies are determined by the state.
3.2.1.2	Initial Monitoring for Bin Classification—Systems Serving at Least
10,000 People5
Schedule 1-3 filtered systems were required to monitor their raw water sources for
Cryptosporidium at each plant at least once per month for a minimum of two years beginning
October 2006.6 Bin classification was based on one of the following:
•	The highest 12-month running annual average (RAA) Cryptosporidium concentration (in
oocysts per liter) if samples were taken monthly (24 samples total), or;
•	The 2-year mean Cryptosporidium concentration. The facility could conduct monitoring
twice per month for 24 months (48 samples total) or perform additional sampling and
4 The term "log removal" is used when the contaminant is eliminated by way of filtration; "log inactivation" is used
when oocysts are killed by disinfection. The term "log treatment" encompasses both removal and inactivation, and is
used to reflect the fact that under the LT2, treatment will be achieved using a combination of filtration, disinfection,
and other non-traditional methods.
5The monitoring and treatment requirements for wholesale systems—i.e., those that sell water only to other
systems—was dependent on the population served by the largest system in the combined distribution system.
6 The largest systems (serving 100,000 people or more) were required to begin monitoring October 2006. Systems
serving 50,000 to 99,999 people were required to begin monitoring April 2007; systems serving 10,000 to 49,999
people were required to begin monitoring April 2008.
Six-Year Review 3
Technical Support Document for LT2
3-3
December 2016

-------
include these results in the calculation of the mean, but the additional samples had to be
evenly distributed over the 2-year monitoring period.
Cryptosporidium analysis was required to be conducted in accordance with EPA Method
1622/1623 using a sample volume of at least 10 liters.7 Samples were also required to be
analyzed for E. coli and turbidity. EPA analyzed the E. coli and turbidity data to determine
whether the data could help predict Cryptosporidium occurrence.
Systems with historical Cryptosporidium data that were equivalent in sample number, frequency
and quality to data required under the LT2 Round 1 monitoring were allowed to use these data to
determine bin classification in lieu of conducting additional Cryptosporidium monitoring, if the
state approved the use of these data. Systems with less than two years of Cryptosporidium data
were also allowed to grandfather their data and collect less than two full years of
Cryptosporidium samples, if the data they had previously collected were equivalent in sample
number, frequency and quality of data required under the LT2 Round 1 monitoring, represented
the months when Round 1 sampling would not be taking place, and the state approved the use of
the data. These are referred to as grandfathered data.
Systems and their laboratories submitted monitoring results to EPA by entering the data into an
EPA database known as Data Collection and Tracking System (DCTS). Systems that
grandfathered data were not required to use DCTS, and additional systems also submitted their
data directly to the primacy agency, not through DCTS. EPA prepared quarterly files containing
the data and delivered those files to the states/primacy agencies via their EPA regional offices.
Systems were responsible for determining their bin placements and reporting their bins to their
states/primacy agencies for approval within six months of completing their monitoring.
7 Systems must meet all requirements of the analytical methods for Cryptosporidium, which include analysis of two
matrix spiked samples.
Six-Year Review 3
Technical Support Document for LT2
December 2016
3-4

-------
3.2.1.3	Initial Monitoring for Bin Classification—Systems Serving Fewer than
10,000 People
The LT2 required small filtered systems (defined as Schedule 4) to conduct E. coli source water
monitoring starting in October 2008, two years after the first large systems initiated source water
Cryptosporidium monitoring. EPA delayed the small system monitoring schedule to allow
incorporation of information on E. coli and turbidity collected by the medium and large systems
into the monitoring requirements as necessary. Based on the Schedule 1-3 systems' data
analyses, EPA determined that turbidity does not appear to be very informative regarding
Cryptosporidium occurrence, whereas E. coli does have merit in this regard. EPA provides more
discussion in Chapter 6, Section 6.2 of this document. Therefore, Schedule 4 systems conducted
one year of biweekly E. coli source water monitoring and were required to conduct
Cryptosporidium monitoring if E. coli concentrations exceeded the following levels8:
•	An annual mean concentration greater than 10 is. coli per 100 milliliters (mL) for lake
and reservoir source waters,
•	An annual mean concentration greater than 50 E. coli per 100 mL for flowing stream
source waters, or
•	Alternative trigger levels of 100 E. coli per 100 mL for source waters drawing from both
lakes/reservoirs and flowing streams.9
EPA assumed that filtered Schedule 4 systems that did not exceed these levels had a
Cryptosporidium concentration of less than 0.075 oocysts/L, and these systems were placed in
Bin 1 (Exhibit 3.1). Schedule 4 systems that exceeded the E. coli levels mentioned above were
required to conduct Cryptosporidium monitoring twice per month for a 1-year period or monthly
for a 2-year period, beginning in April 2010, six months after the conclusion of E. coli
monitoring. Bin classification for Schedule 4 systems conducting Cryptosporidium monitoring
was determined by the highest 12-month RAA.
Schedule 4 systems and their laboratories submitted monitoring data to EPA by either submitting
the data directly to their state/primacy agency or by entering the data into DCTS. Systems were
responsible for determining their bin placements and reporting their bins to their state/primacy
agency for approval within six months of completing their monitoring.
3.2.1.4	Bins and Treatment Requirements—All System Sizes
Exhibit 3.1 presents the bins for filtered systems according to the type of treatment already in
place. Systems must meet Cryptosporidium treatment requirements by using one or a
combination of the treatment options in the "microbial toolbox," or by demonstrating
performance equivalent to, or exceeding the required treatment. The LT2 requires systems to
8	Small systems were also required to conduct Cryptosporidium monitoring if they failed to conduct E. coli
monitoring or if they elected to proceed directly to Cryptosporidium monitoring.
9	Alternative trigger levels were identified in USEPA (2010b).
Six-Year Review 3
Technical Support Document for LT2
3-5
December 2016

-------
meet the treatment requirements associated with their bin placement within three years after first
being assigned to a bin. States/primacy agencies may grant systems a 2-year extension to comply
if capital investments are necessary. Systems must report to the state their bin placement within
six months of completing their source water monitoring.
Exhibit 3.1 Bin Classifications and Treatment Requirements for Filtered Systems
If your source
water
Cryptosporidium
concentration
(oocysts/L) is ...
Your bin
classification
is ...
And if you use the following filtration treatment in
full compliance with existing regulations, then
your additional treatment requirements are ...
Conventional
Filtration
Direct
Filtration
Slow Sand or
Diatomaceous
Earth
Filtration
Alternative
Filtration
Technologies
< 0.075
1
No additional
treatment
No
additional
treatment
No additional
treatment
No additional
treatment
> 0.075 and < 1.0
2
1 -log
treatment
1.5-log
treatment
1 -log treatment
As determined
by the state1
> 1.0 and < 3.0
32
2-log
treatment
2.5-log
treatment
2-log treatment
As determined
by the state3
>3.0
42
2.5-log
treatment
3-log
treatment
2.5-log
treatment
As determined
by the state4
Notes:
1)	Total Cryptosporidium treatment must be at least 4.0-log.
2)	Systems must achieve at least 1 -log of the required treatment using ozone, chlorine dioxide, UV light,
membranes, bag/cartridge filters or bank filtration (BF).
3)	Total Cryptosporidium treatment must be at least 5.0-log.
4)	Total Cryptosporidium treatment must be at least 5.5-log.
The total Cryptosporidium treatment required for systems in Bins 2, 3 and 4 is 4.0-log, 5.0-log
and 5.5-log, respectively. EPA based the additional treatment requirements in Exhibit 3.1 on
information obtained after EPA promulgated the IESWTR and the LT1 that indicates that
conventional, slow sand and diatomaceous earth filtration plants in compliance with the
IESWTR or the LT1 achieve an average of 3-log removal of Cryptosporidium across all plants.
The IESWTR and the LT1 require systems to achieve 2-log removal; EPA based this
requirement on the information on the minimum removal expected from these types of filtration
(USEPA, 2006a). Therefore, systems with conventional, slow sand and diatomaceous earth
filtration plants will require an additional 1.0- to 2.5-log treatment to meet the total treatment
requirement, depending on the bin in which the systems are placed.
In the LT2 EPA determined that direct filtration plants achieve an average 2.5-log removal of
Cryptosporidium (USEPA, 2006a). The removal is less than the removal in conventional
filtration because direct filtration lacks a sedimentation process. Consequently, under the LT2,
direct filtration plants in Bins 2-4 must provide 0.5-log more in additional treatment than
conventional plants to meet the total Cryptosporidium treatment requirement.
Six-Year Review 3
Technical Support Document for LT2
3-6
December 2016

-------
3.2.1.5 Microbial Toolbox for Meeting Additional Treatment Requirements
To meet the Cryptosporidium treatment requirements for the bin in which they are classified,
filtered systems select from a "toolbox" of treatment or management options. Exhibit 3.2 lists the
treatment and management strategies comprising the microbial toolbox, and Chapter 7 includes a
discussion of these strategies in more detail. EPA prescribed each option in the toolbox with a
certain amount of log treatment credit, which systems can apply toward their total treatment
requirements. [Systems do not get the log credit automatically when they install these
technologies; systems must show that they are meeting certain operational or performance
criteria specific to each technology in order to receive credit.] Log treatment credit under the
existing rules (e.g., the IESWTR and the LT1) works in a similar process. Systems already using
ozone, chlorine dioxide, UV light or membranes, in addition to conventional treatment prior to
the promulgation of the LT2, can receive credit for those technologies toward meeting bin
requirements if they meet the LT2 criteria for the chosen technology. Systems currently using
chorine and/or chloramines do not receive credits for these disinfectants under the LT2 because "
Six-Year Review 3
Technical Support Document for LT2
3-7
December 2016

-------
Exhibit 3.2 Microbial Toolbox Components for the LT210
Toolbox Option
Log Treatment Credit
Source Toolbox Components
Watershed control program
0.5
Alternative source/intake
management
No presumptive credit. Systems may conduct
simultaneous monitoring for treatment bin
classification at alternative intake locations or
under alternative intake management
strategies.1
Pre-Filtration Toolbox Components
Presedimentation basin with
coagulation
0.5
Two-stage lime softening
0.5
BF
0.5 or 1.0, depending on setback
Treatment Performance Toolbox Components
Combined filter performance
0.5
Individual filter performance
1.0
Demonstration of performance (DOP)
State approved2
Additional Filtration Toolbox Components
Bag filters
2.0 as individual and 2.5 for two in series
Cartridge filters
2.0 as individual and 2.5 for two in series
Membrane filtration
As demonstrated3
Second stage filtration
0.5
Slow sand filters
2.5
Inactivation Toolbox Components
Chlorine dioxide
As demonstrated4
Ozone
As demonstrated4
UV
As demonstrated2
Notes:
1)	Exhibit 3.3 of the Microbial Toolbox Guidance Manual contains additional information
(USEPA, 2010c).
2)	The state must approve the method used to demonstrate performance and must approve
the log credit claimed by the system.
3)	EPA based the credit for membrane filtration and UV on the results of equipment-specific
testing.
4)	EPA based the credit for chlorine dioxide and ozone on CT values achieved (CT is the
product of disinfectant concentration and contact time).
Six-Year Review 3
Technical Support Document for LT2
3-8
December 2016

-------
3.2.1.6 Reassessment and Future Monitoring
Six years after initial bin classification, systems must conduct a second round of monitoring to
reassess source water conditions for bin assignments. Systems that provide a total of 5.5-log
treatment for Cryptosporidium are not subject to future monitoring if they provide that treatment
by the compliance date for Round 1 monitoring.
For those Schedule 1-3 systems not providing a total of 5.5-log treatment for Cryptosporidium,
the second round of monitoring began no later than October 2016.11 For Schedule 4 systems, the
second round of monitoring begins no later than October 2019.
In addition to the second round of monitoring described above, in their primacy application the
state/primacy agency will describe how they will assess any significant changes in the watershed
and source water as part of the sanitary survey process.
3.2.2 Monitoring and Treatment Requirements for Unfiltered Systems
Prior to the LT2 promulgation, unfiltered systems were required to address Cryptosporidium
through their watershed control plans but did not have specific log inactivation requirements for
Cryptosporidium. The LT2 established new treatment requirements for unfiltered systems,
except that unfiltered systems that already have 3-log Cryptosporidium treatment in place prior
to the compliance date are exempt from source water monitoring and additional Cryptosporidium
inactivation requirements.
Unfiltered Schedule 1-3 systems were required to monitor Cryptosporidium in their source water
monthly for at least two years, and unfiltered Schedule 4 systems had to monitor
Cryptosporidium twice a month for 12 months or monthly for 24 months. All unfiltered systems
determined their treatment requirements based on the arithmetic mean Cryptosporidium
concentration. Systems with an average Cryptosporidium concentration of less than or equal to
0.01 oocysts/L must provide 2-log Cryptosporidium inactivation. If their average concentration
is greater than 0.01 oocysts/L, they must provide 3-log inactivation.
EPA based the monitoring for unfiltered systems on the same schedule as monitoring for filtered
systems, although unfiltered systems are not required to monitor E. coli or turbidity. As with the
filtered systems, unfiltered systems must conduct a second round of Cryptosporidium monitoring
six years after the initial bin assignment, unless they provide 3-log treatment for
Cryptosporidium.
In addition to the new Cryptosporidium inactivation requirements, the LT2 requires unfiltered
systems to continue to meet the filtration avoidance criteria under the 1989 Surface Water
Treatment Rule (SWTR) and to continue to provide inactivation for Giardia and viruses.
10	In order for a water system to receive Cryptosporidium treatment credit for using a toolbox option, the system
must comply with the operational, monitoring, and reporting requirements associated with the toolbox option that
EPA established in the LT2 or by the primacy agency.
11	Schedule 1 systems began monitoring no later than April 2015. Schedule 2 systems began monitoring no later
than October 2015; Schedule 3 systems began monitoring no later than October 2016.
Six-Year Review 3
Technical Support Document for LT2
3-9
December 2016

-------
Systems must meet the overall inactivation requirements (i.e., 4-log virus, 3-log Giardia, and 2-
or 3-log Cryptosporidium) using a minimum of two disinfectants. Additionally, each of two
disinfectants must meet the total inactivation for one of the three pathogens. For example, a
system could use UV to inactivate 2-log of Cryptosporidium and Giardia and use chlorine to
inactivate 4-log of viruses and 1-log of Giardia.
3.2.3	Requirements for Existing Uncovered Finished Water Reservoirs
The LT2 builds on the IESWTR and the LT1, which require covers only for new finished water
reservoirs. The LT2 established requirements for all systems with existing UCFWRs. Systems
must either cover the reservoir or treat the water exiting the reservoir to the distribution system to
achieve 2-log Cryptosporidium, 3-log Giardia lamblia and 4-log virus inactivation.
3.2.4	Disinfection Profiling and Benchmarking Requirements
A disinfection profile is a graphical representation of a system's level of Giardia and virus
inactivation measured during the course of one or more year(s). A benchmark is the lowest
monthly average of microbial inactivation during the disinfection profile period. The LT2
includes disinfection profiling and benchmarking requirements that apply if a public water
system (PWS) proposes to make a significant change to its disinfection practice as it implements
the Stage 2 Disinfectants and Disinfection Byproducts Rule and the LT2 [40 CFR 141.708-709],
The LT2 defines a significant change as a change to the point of disinfection, a change in the
disinfectant used, a change to the disinfection process or any other modification that the state
identifies as significant.
EPA developed profiling and benchmarking requirements under the Interim Enhanced Surface
Water Treatment Rule and the LT1 and extended the requirements to the LT2 to address similar
risk-balancing tradeoffs between the control of microbial pathogens and disinfection byproducts
(DBPs).
The LT2 requires these systems to prepare a disinfection profile that characterizes current levels
of Giardia lamblia and virus inactivation throughout the plant over the course of one year. Prior
to making the change, the system must calculate a benchmark and consult with the state
regarding how the proposed change will affect the current disinfection level.
A detailed discussion of disinfection profiling and benchmarking is described in the Six-Year
Review 3 (SYR3) Technical Support Document for Microbial Contaminant Regulations
(USEPA, 2016a).
3.2.5 Implementation Timeline
Exhibit 3.3 shows the timeline of the LT2 activities for filtered systems as described in previous
sections. The schedule for monitoring and compliance with treatment requirements differs by
population served.
Six-Year Review 3
Technical Support Document for LT2
3-10
December 2016

-------
Exhibit 3.3 Implementation Timeline for the LT2 for Filtered Systems
Systems Serving > 100,000 People
Crypto Monitoring
Oct 2006 - Oct 2008
Treatment Installation
Apr 2009 - Apr 2012
Possible Extension
Apr 2012 - Apr 2014
2 Round Crypto
Monitoring
Apr 2015 - Apr 2017
2nd Round Treatment
Installation
Oct 2017 - Oct 2020
2nd Round Possible
Extension
Oct 2020-Oct 2022
Bin Determination
Oct 2008-Apr 2009
2 Round Bin Determination
Apr 2017-Oct 2017
Systems Serving 50,000-99,999 People
Crypto Monitoring
Apr 2007 - Apr 2009
Treatment Installation
Oct 2009 - Oct 2012
Possible Extension
Oct2012-Oct 2014
2 Round Crypto
Monitoring
Oct2015-Oct 2017
2nd Round Treatment
Installation
Apr 2018 - Apr 2021
2nd Round Possible
Extension
Apr 2021-Apr 2023
Bin Determination
Apr 2009-Oct 2009
2 Round Bin Determination
Oct 2017 - Apr 2018
Systems Serving 10,000-49,999 People
Crypto Monitoring
Apr 2008 - Apr 2010
Treatment Installation
Oct 2010-Oct 2013
Possible Extension
Oct 2013 - Oct 2015
2 Round Crypto
Monitoring
Oct 2016 - Oct 2018
2 Round Treatment
Installation
Apr 2019-Apr 2022
2nd Round Possible
Extension
Apr 2022-Apr 2024
Bin Determination
Apr 2010 - Oct 2010
2 Round Bin Determination
Oct 2018 - Apr 2019
Oct 2006
Systems Serving < 10,000 People4
£. Coll

Crypto

Mon.

Mon.

Oct 2008-

Apr 2010 -

Oct 2009

Apr 2011

Treatment Installation
Oct 2011 - Oct 2014
Possible Extension
Oct 2014 - Oct 2016

2nd Round

2nd Round


E. Coli

Crypto


Mon.

Mon.


Oct 2017-

Apr 2019 -


Oct 2018

Apr 2020

2nd Round Treatment
Installation
Oct 2020-Oct 2023
Oct 2025
2 Round Possible
Extension
Oct 2023 - Oct 2025
Bin Determination	2 Round Bin Determination
Apr 2011 - Oct 2011	Apr 2020 - Oct 2020
*Systems following this schedule have the option to conduct lKand 2^ round Crypto monitoring over a one or two year period. Note that the remainder of the schedule would shift accordingly.
Six-Year Review 3
Technical Support Document for LT2
3-11
December 2016

-------
4 Health Effects Information
EPA uses the Health Effects Branch of the Six-Year Review Protocol primarily to assess whether
scientific information suggests a potential for revision of a maximum contaminant level goal
(MCLG) under an existing rule. MCLGs are nonenforceable health goals set at a level at which
no known or anticipated adverse health effects occur and which allow an adequate margin of
safety.
The 1996 Safe Drinking Water Act (SDWA) amendments require special consideration of all
sensitive populations (e.g., infants, children, pregnant women, elderly, individuals with a history
of serious illness) in the development of drinking water regulations. In addition, EPA's plan for
Retrospective Review of Existing Regulations criteria for review outlines consideration of effect
of the regulation on children. New studies about the effect of Cryptosporidium on children
further support EPA's conclusion in the Long Term 2 Enhanced Surface Water Treatment Rule
(LT2) (USEPA, 2006a) that Cryptosporidium may have a disproportionate effect on children.
Since the promulgation of the LT2, there is additional evidence that asymptomatic
Cryptosporidium infection in children can lead to reduced and delayed growth. The LT2 states
that while Cryptosporidium may have a disproportionate effect on children, available data were
not adequate to distinctly assess the health risk for children resulting from Cryptosporidium-
contaminated drinking water. No new data have been found to allow for that assessment.
Therefore, the approach to assessing children's risk as part of the Six-Year Review 3 (SYR3)
process would be the same as used to support the LT2 (USEPA, 2006a). In assessing risk to
children when evaluating regulatory alternatives for the LT2, EPA assumed the same risk for
children as for the population as a whole. This is consistent with Executive Order (EO) 13045
"Protection of Children from Environmental Health Risks and Safety Risks." The Agency
explained in the LT2 preamble why the planned regulation is preferable to other potentially
effective and reasonably feasible alternatives considered by EPA. EPA concluded that result of
the LT2 will be a reduction in the risk of illness for the entire population, including children
(USEPA, 2006a).
Under the Six-Year Review Protocol Health Effects Branch, EPA also considers whether the
peer-reviewed literature findings include health effects that are significantly different from those
considered previously that could lead to a nomination for a new agency HEA.
With respect to the LT2, EPA did not establish any new MCLGs when the LT2 was
promulgated. However, EPA established MCLGs for several pathogenic microorganisms under
various preceding versions of the Surface Water Treatment Rules that led up to the LT2. Under
the 1989 Surface Water Treatment Rule (SWTR), EPA established MCLGs of zero for Giardia
lamblia12, viruses and Legionella (USEPA, 1989a). At the same time, EPA promulgated the
related Total Coliform Rule (TCR) (USEPA, 1989b) which established an MCLG of zero for all
coliform bacteria, including fecal coliforms and Escherichia coli specifically. Under the 1998
12 The current preferred taxonomic name is Giardia duodenalis, with Giardia lamblia and Giardia intestinalis as
synonymous names. However, Giardia lamblia was the name used to establish the MCLG in 1989. Elsewhere in
this document this pathogen will be referred to as Giardia spp. or simply Giardia unless discussing information on
an individual species.
Six-Year Review 3
Technical Support Document for LT2
4-1
December 2016

-------
Interim Enhanced Surface Water Treatment Rule (USEPA, 1998), EPA established an MCLG of
zero for Cryptosporidium.
EPA established treatment technique (TT) requirements for all the surface water treatment rules
including the LT2 in lieu of finished water maximum contaminant levels (MCLs) for the
pathogens being addressed by the MCLGs noted earlier in this section of the document. EPA
used a TT for the LT2 because it recognized that it was not economically or technologically
feasible to determine the level of Cryptosporidium in finished water for the purpose of
compliance with a finished water standard (i.e., MCL) (USEPA, 2006a). This limitation has been
recognized since the time of the original 1989 SWTR where EPA similarly concluded that a TT
rule was needed for Giardia, Legionella and virus noting that: (1) the only analytical methods
which are available require levels of expertise that many utility personnel do not have; (2)
analysis by independent laboratories is generally very expensive; and (3) systems would have to
monitor inordinately large and frequent samples of water to ensure that the occurrence of these
pathogens is not of health risk significance (e.g., failure to detect Giardia in one or a few
samples provides no assurance that Giardia do not occur at significant levels in the water
supply).
One specific focus of the LT2 was to require the use of risk-targeted additional TTs for all public
water systems (PWSs) that use surface water sources, including ground water under the direct
influence of surface water (GWUDI), and have relatively high Cryptosporidium levels in their
source waters. In addition, the LT2 also addressed specific treatment requirements for uncovered
finished water reservoirs (UCFWRs), for which the health effect concerns extend beyond
Cryptosporidium to other pathogens.
Because the LT2 uses a TT requirement rather than an MCL, EPA also considered new health
effects information under the TT Analysis Branch of the Six-Year Review Protocol to help
inform the decision as to whether a meaningful opportunity exists for any TT revisions to lower
health risks.
EPA has organized this health effects information chapter around the pathogens that the LT2 and
its predecessor surface water treatment rules are designed to address:
•	4.1 Cryptosporidium,
•	4.2 Giardia,
•	4.3 Viruses, and
•	4.4 Other pathogens.
Because the risk-based TT requirements of the LT2 focus primarily on Cryptosporidium risks
associated with exposure from filtered and unfiltered systems, the bulk of the discussion
presented in this chapter regarding new health effects information involves that pathogen.
However, EPA also reviewed new information on other pathogens of concern to determine
whether additional measures are warranted to provide public health protection from those
pathogens, particularly in the context of the UCFWR components of the LT2.
Six-Year Review 3
Technical Support Document for LT2
4-2
December 2016

-------
4.1 Cryptosporidium
The risk assessment for Cryptosporidium that EPA conducted in support of the LT2 was
structured in accordance with EPA's general framework for conducting health risk assessments
for environmental contaminants.
The first two risk-assessment framework components are most concerned with health effects
information, the subject of this chapter. The hazard identification component discussed in this
chapter addresses the specific adverse health effects associated with Cryptosporidium, while the
dose-response assessment addresses the relationship between the magnitude of exposure to
Cryptosporidium and the likelihood of those adverse health effects occurring.
Because the critical adverse health effects associated with Cryptosporidium are well known—
severe gastrointestinal illness that can in some cases be fatal, especially for certain susceptible
subpopulations—EPA focused on gathering and reviewing new information concerning various
elements of the dose-response assessment component of the Cryptosporidium health risk
assessment for the LT2 effort rather than on the hazard identification component.
As described in detail in the LT2 Economic Analysis (EA) (USEPA, 2005a), the dose-response
assessment model that was developed for Cryptosporidium to support the LT2 has three distinct
parts:
•	Infectivity: addresses the probability of an individual becoming infected given ingestion
of one or more oocysts;
•	Morbidity: addresses the probability of an infected individual becoming ill and
experiencing the gastrointestinal illness symptoms of cryptosporidiosis, and;
•	Mortality: addresses the probability of a case of cryptosporidiosis being fatal to an
individual.
For this review of the LT2, EPA focused on the following key aspects of the dose-response
assessment and the risk characterization:
•	Are there additional human challenge studies that can also be used to parameterize the
infectivity dose-response function?
•	Are there studies suggesting alternative model forms for infectivity different from the
exponential model used in the 2005 LT2 EA that EPA should consider?
•	Are there studies or data suggesting alternative approaches to considering immunity in
the exposed population for the dose-response modeling of infectivity?
•	Are there studies or data that provide additional information on the assumption
concerning the "v" ratio (the ratio of the fraction of infectious oocysts in the environment
(numerator) to the fraction of infectious oocysts in doses tested in clinical challenge
studies (denominator)) of infectious oocysts?
Six-Year Review 3
Technical Support Document for LT2
4-3
December 2016

-------
•	Are there studies or data that provide additional information on the estimates of the
morbidity rate (risk of illness given an infection)?
•	Are there studies or data that provide additional information on the estimates of mortality
rate (risk of death given an illness)?
For this review effort, EPA conducted literature searches to identify published information
relevant to Cryptosporidium infectivity, morbidity and mortality that has become available since
EPA promulgated the LT2.
The following sections present and discuss the approach EPA used to characterize
Cryptosporidium infectivity, morbidity and mortality at the time of the LT2 promulgation, as
well as new information that has become available since that time that EPA could use to update
the assessment of Cryptosporidium health risks for the Six-Year Review.
4.1.1 Infectivity
In the LT2 EA, EPA used several variations of a basic dose-response model for infectivity that
were intended individually, and in combination, to reflect the uncertainty associated with
modeling the probability of a Cryptosporidium infection occurring in humans given a dose
(number of oocysts) ingested by an individual.
The basic dose-response model used for infectivity in the LT2 EA is the exponential model
having the form:
Pi= 1 -Q-dr
where Pi is the probability of an individual becoming infected given ingestion of a dose "d" of
oocysts. The model parameter "r", which is estimated from fitting the model to human challenge
data, has a value between zero and one and is effectively a measure of the probability that an
infectious organism that has been ingested will survive and cause an infection in the host.
At the time of the LT2 proposal in 2003, EPA used human challenge study data that were then
available from three different C. parvum isolates (TAMU, Iowa and UCP) to parameterize the
infectivity dose-response model. Following the proposal, EPA acquired data for two additional
C. parvum isolates (Moredun and 16W), as well as for one C. hominis isolate (TU502). EPA
used the human challenge data from these six isolates in a variety of ways to develop alternative
infectivity models that were used as the primary model for the risk assessment, and as
alternatives to capture the underlying uncertainty in the data and the model selections.
Exhibit 4.1 taken from the LT2 EA summarizes the key characteristics of those model variations.
The models differ with respect to the model form (as noted above); the number and type of
Cryptosporidium isolates used to estimate the r and, where used, the y parameters for those
models; the assumed distributional form characterizing the r parameter; and the underlying
assumptions with respect to the source of variability in the r parameter obtained from the human
challenge data, namely, organism variability or host variability.
Six-Year Review 3
Technical Support Document for LT2
4-4
December 2016

-------
Exhibit 4.1 Characteristics of the LT2 EA Primary Model and Six Alternative
Models
Model
Data Sets
Functional Form
Distribution
of r
Assumptions
Mean
Risk1
Primary Model
Two models
I

-------
Alternative Model 1 and two had the form and assumptions of Alternative Model 2 as shown in
Exhibit 4.1. In one set of Model 1 and Model 2, the r parameters were based on human challenge
results for two isolates (Iowa and TAMU) while the other set of Model 1 and 2 used results for
three isolates (Iowa, TAMU and UCP). Again, EPA combined these four model variations to
form the single primary model that was used in the LT2 EA.
In addition to the primary model, EPA also separately modeled infectivity using the six model
forms and assumptions indicated in Exhibit 4.1, and in those cases using the human challenge
data from all six of the isolates identified previously.
(For a more detailed discussion of the LT2 EA infectivity modeling, the reader is referred to
Chapter 5 and Appendix N of the LT2 EA.)
One important aspect of the infectivity modeling to note is that in the modeling procedures
addressed earlier in this section of the document to arrive at the estimates of the model
parameters r and y, EPA assumed that all of the oocysts ingested by the individuals in the human
challenge studies were "viable and infectious." This is not, however, assumed to be the case
with respect to ingestions of oocysts in drinking water derived from environmental sources.
Therefore, in the application of the infectivity dose response models to estimating infectivity in
the population exposed from drinking water, EPA included an additional parameter in the model
as shown below for the basic exponential model form:
Pl= 1 -Q-dvr
where "v" is the fraction of environmental oocysts that are "viable and infectious". EPA included
this value in the modeling as a distribution reflecting uncertainty in the expected value for this
parameter. The range of those values varied depending upon the dataset used for the
environmental oocysts (15 to 25 percent for the Information Collection Rule (ICR) data, 30 to 50
percent for the Information Collection Rule Supplemental Survey (ICRSS) data) and based on
the fraction of "empty" oocysts observed in those studies. [Note that v can also be viewed as a
"viability fraction" where the numerator is the fraction of environmental oocysts considered to
be viable and infectious, and the denominator the fraction in the human challenge studies
considered to be viable and infectious which, as noted above, is assumed to be 1. This
assumption is discussed further below.]
Six-Year Review 3
Technical Support Document for LT2
4-6
December 2016

-------
For EPA's LT2 review of new information relevant to the infectivity portion of the
Cryptosporidium dose-response model, the Agency focused on three main elements:
•	The review of new human challenge studies (including the role of immunity),
•	The review of new quantitative dose-response model forms (also including the role of
immunity), and
•	The review of new information on oocyst viability studies.
4.1.1.1 Review of New Human Challenge Studies
A microbial human challenge study is the deliberate exposure of human volunteers to a range of
doses of a microorganism to evaluate infectivity. The human challenge study data EPA evaluated
and used in the LT2 EA included data from a total of six Cryptosporidium isolates. Those
isolates included Iowa (DuPont et al., 1995; Okhuysen et al., 1999), TAMU (Okhuysen et al.,
1999), UCP (Okhuysen et al., 1999), Moredun (Okhuysen et al., 2002), TU502 (Chappell et al.,
2006) (note that this is an updated citation for the then-unpublished feeding study data for this C.
hominis isolate) and 16W (the data for this C. parvum isolate have not been published in a peer-
reviewed manuscript).
In the human challenge studies EPA used for the LT2 development, human volunteers who had
low levels of pre-existing anti-Cryptosporidium immunoglobulin G (IgG) antibodies13 were
initially challenged with Iowa, UCP, TAMU, TU502 and Moredun isolates. In these trials, the
ID50S (the median infective dose) varied substantially, indicating potential differences in
infectivity among these isolates (Okhuysen et al., 1999; Okhuysen et al., 2002). Nineteen of the
29 volunteers who had been challenged with the Iowa isolate were rechallenged after one year,
with a single dose of 500 oocysts. Of these 19, infection was observed in 15 in the first challenge
and no infection was observed in four. The rechallenge resulted in the same frequency of
diarrhea symptoms as the first challenge, indicating a lack of protective immunity against
infection and illness, although the number of oocysts shed by infected volunteers was lower than
in the first challenge (Okhuysen et al., 1998). When the same Iowa isolate was tested in a
different group of volunteers with and without pre-existing serum IgG to C. parvum, the ID50
was 14-20 fold higher in the volunteers with pre-existing IgG, and illness was significantly
associated with the highest doses (DuPont et al., 1995; Chappell et al., 1999). Based on these
observations, high IgG levels could indicate a recent Cryptosporidium infection and may be
associated with a protective response, but with a likely duration of less than 12 months.
The literature review EPA conducted for the Six-Year Review revealed several new challenge
studies published after the LT2, and one published prior to the LT2 but not explicitly considered
previously. In this latter study, Chappell et al. (1999) evaluated the infectivity of C. parvum
(Iowa strain) in healthy adults with pre-existing anti-C. parvum serum IgG. In this study, 17
13 IgG antibodies are associated primarily with the secondary immune response and are indicative of a past exposure
to the antigen of interest and provide a potential measure of the extent and duration of immunity from subsequent
exposures.
Six-Year Review 3
Technical Support Document for LT2
4-7
December 2016

-------
healthy adults were challenged with 500-50,000 oocysts. Infection and diarrhea were associated
with the higher challenge doses. The ID50 was 1,880 oocysts, >20-fold higher than in
seronegative volunteers from previous challenge studies with this strain (DuPont et al., 1995;
Okhuysen et al., 1999). In other words, it took a greater number of oocysts to produce an
infection in adults with pre-existing anti-C. parvum serum IgG. Fecal oocysts were detected in
only 7 of 13 individuals (54 percent) with clinical cryptosporidiosis, indicating that the host
response may effectively decrease the number of oocysts produced. Subjects with the highest
IgG levels prior to challenge had little to no increase in IgG following challenge, whereas
volunteers with lower IgG levels showed significant postchallenge increases. This suggests that
an upper limit of serum IgG was present in some subjects, while others were further stimulated
by an additional exposure. These data imply that prior exposure to C. parvum may provide some
protection from infection and illness.
Chappell et al. (2011) examined the infectivity of C. meleagridis in healthy adults. (Note that C.
meleagridis was not among the species or isolates used in the LT2 risk assessment.) Five
volunteers were challenged with 105 C. meleagridis oocysts and monitored during six weeks for
fecal oocysts and clinical manifestations. All five volunteers had evidence of infection by either
clinical or microbiological measures, or both. Four volunteers had diarrhea yielding an 80
percent illness rate. Three had detectable fecal oocysts. Two of the four volunteers with diarrhea
had detectable oocysts, and two did not. One volunteer had no unformed stools or symptoms
while shedding low, but detectable, numbers of oocysts. Fecal deoxyribonucleic acid (DNA)
from two volunteers was amplified by using a polymerase chain reaction (PCR) specific for the
Cryptosporidium small subunit ribosomal ribonucleic acid (rRNA) gene. Nucleotide sequence of
these amplicons was diagnostic for C. meleagridis. All infections were self-limited and oocysts
were cleared within 12 days of challenge. These studies establish that healthy adults can be
infected and become ill from ingestion of C. meleagridis oocysts. These findings are consistent
with a study that evaluated the epidemiology and symptoms of cryptosporidiosis patients in
Sweden, reporting cases due to C. parvum (n=l 11), C. hominis (n=65), C. meleagridis (n=l 1), C.
felis (n=2), C. chipmunk genotype I (n=2) and C. viatorum (n=2). That study also reported that
clinical manifestations differed slightly by species with diarrhea lasting longer in C. parvum
cases compared to C. hominis and C. meleagridis cases (Insulander et al., 2012). A similar
finding was also reported in Ethiopia, where symptoms of diarrhea where longer in patients
infected with C. parvum as compared with C. hominis (Adamu et al., 2014. PLOS Neg. Trop.
Dis 8:e2831).
As part of an EPA research grant, Chappell et al. (2003) examined the potential for C. muris to
infect healthy adults. (Note that C. muris was not among the species or isolates used in the LT2
risk assessment.) Each of six volunteers was challenged with 105 C. muris oocysts and monitored
for six weeks for infection and/or illness. All six became infected, and two experienced a
diarrheal illness. The number of total oocysts shed during the study ranged from 6.7 x 106 to 4.1
x 108, and was higher (mean = 2.8 x 108) in volunteers with diarrhea than in asymptomatic
shedders (mean = 4.4 x 107). Follow-up fecal examinations at seven months postchallenge on
five of the six volunteers revealed that three volunteers were still positive for fecal oocysts.
These data establish the susceptibility of healthy humans to C. muris infection. C. muris illness
was self-limiting in two cases and resulted in persistent, asymptomatic infections in three persons
for seven months before treatment cleared the infection.
Six-Year Review 3
Technical Support Document for LT2
4-8
December 2016

-------
Sheoran et al. (2012) conducted a challenge study with gnotobiotic piglets to investigate cross-
protection from C. hominis and C. parvum. After >3 days of recovery from C. hominis infection,
the piglets were completely protected against subsequent challenge with C. hominis but only
partially against challenge with C. parvum, as compared with age-matched control animals
challenged with either species. All C. hominis-C. parvum group piglets shed oocysts 3-4 days
after the challenge infection, whereas none of the C. hominis-C. hominis group shed any oocysts
for the duration of the experiment. However, total oocyst counts of the null-C. parvum group
(range, 98-225) were significantly higher than that of the C. hominis-C. parvum group. Based on
these results the authors conclude that C. hominis-specific immunity was sufficient to completely
protect against challenge with the same species but insufficient to provide the same level of
protection against C. parvum.
4.1.1.2 Review of Key Dose Response Modeling Studies
EPA evaluated the key dose response modeling studies to determine if data exist that suggest
EPA should consider revisions to the TT requirements of the LT2. Teunis et al. (2002a)
reevaluated three of the Cryptosporidium isolates (Iowa, TAMU, UCP) that had been evaluated
previously in challenge studies (DuPont et al., 1995; Okhuysen et al., 1999) and that had been
reported to have different IDsos. Those previous results, the findings of which are based on
serologically negative healthy volunteers, indicated substantial variation in their infectivity for
humans with IDsos of 1,042 for UCP, 87 for the Iowa isolate and 9 for TAMU. In this
investigation, both within- and between-isolate variability were considered. The results indicate
that in these controlled conditions there is substantial variation in infectivity among these
isolates, although greater heterogeneity may exist among unspecified environmental strains. This
analysis yields a wider estimate of the posterior range of the infection risk, reflecting additional
uncertainty. Based on this work, starting from a discrete number of pathogens, the resultant dose
response model would be a binomial model with Beta heterogeneity (a Beta-binomial model)
both within and between isolates.
Teunis et al. (2002b) adapted the hit theory model of microbial infection to incorporate
covariables, characterizing the immune status of the susceptible host. The probability of any
single oocyst in the inoculum causing infection appears to depend on pre-existing IgG levels.
Since these can be measured in human populations, the authors concluded that the IgG-
dependence of the dose-response relation can be used to assess the distribution of susceptibility
to infection and illness by Cryptosporidium and could be easily applied in quantitative risk
analysis.
There are several possible ways to account for immunity and/or differential susceptibility to
infection in a quantitative microbial risk assessment (QMRA). One possibility suggested by
Teunis et al. (2002b) and implemented in the World Health Organization (WHO) risk assessment
for Cryptosporidium (2009) is to use individual IgG levels as a proxy for susceptibility to partly
control for host variation in the general population. To incorporate a covariable like IgG level
into the dose response relationship, one can make the dose response relationship dependent on
that covariable. One approach would be to use a logistic relation nested into the exponential dose
response model. For example, when data from the two studies with the Iowa isolate are
combined, one has data on infection in subjects with a wide range of IgG levels. Thus, one
possibility would be to fit the IgG-dependent dose response model results in a function for the
Six-Year Review 3
Technical Support Document for LT2
4-9
December 2016

-------
probability of infection which shows that there is a protective effect associated with high levels
of (prechallenge) IgG. When such a model is applied to subjects with high pre-existing IgG,
there is decreasing infectivity with increasing IgG levels. Exhibit 4.2 shows the results from one
such dose response relation, as a function of dose and IgG level (WHO, 2009). Exhibit 3.2 shows
the maximum likelihood dose response relationship fit to both low and high IgG data, a function
of dose and pre-existing IgG levels.
Exhibit 4.2 Dose Response Relation, as a Function of Dose and IgG Level
Englehardt and Swartout (2004) present a hierarchical predictive population dose-response
Bayesian assessment for C. parvum for the infection endpoint. The authors used available data
on the infectivity of three isolates of C. parvum (Iowa, TAMU and UCP) to adjust, by bootstrap
analysis, for sensitive and antibody-positive subpopulations not proportionately represented in
the data. The authors used the diverse mean infectivities of the isolates to obtain a predictive
distribution for population infectivity, which was used to obtain the predictive population dose-
response function. The result is a distribution of unconditional probability of infection, based on
available dose-response information. Information includes theoretical and empirical evidence for
the conditional beta-Poisson parametric dose-response function. Results indicate that a dose of
6x 10~6 oocysts per exposure, and assuming 365 exposures per year, corresponds to 10~4
infections per person per year. By comparison, the corresponding estimate of daily oocyst
exposure resulting in a 10"4 annual infection risk using the "best estimates" of parameter values
and exposure assumptions in the EA would be 1.5 x 10"5 to 7.6 x 10"6oocysts, where EPA used
factors of 0.2 and 0.4, respectively, to reflect the fraction of oocysts that are viable and
infectious. Englehardt and Swartout (2004) did not include consideration of viability in their
analysis, implying a factor of 1.0. If this factor is set tol.O in the EPA EA analysis, the daily
oocyst intake corresponding to an annual infection risk of 10"4 would be 3.0 x 10"6.
WHO conducted a risk assessment for Cryptosporidium in drinking water (WHO, 2009) that
used the results from the Cryptosporidium human challenge studies conducted at the University
of Texas as the basis for their dose response relationship (Okhuysen et al., 1999; Okhuysen et al.,
2002). Four isolates (Iowa, TAMU, UCP and Moredun) formed the basis of this work for
Six-Year Review 3
Technical Support Document for LT2
4-10
December 2016

-------
individuals with and without pre-existing C. parvum IgG (Chappell et al., 1999). To obtain a
dose-response relationship, WHO analyzed the infection data as a binary (yes/no) response, with
the single hit model for microbial infection. The modeling approach is a two-level model, the
lower hierarchical level representing variation in isolates, and the upper level representing
variation among hosts. WHO used the beta-Poisson dose response model with the parameters
(a ,P) taken from (joint) distributions, describing the "between isolates" variation. This approach
generalizes the heterogeneous infectivity to the "group" level for isolates, treating the four data
sets as an n=4 sample from the presumed population of environmental oocysts. The distribution
may be interpreted as the frequency distribution for dose response relations from the oocysts.
Sampling from this distribution produces a predictive dose response relation which covers a wide
range of infectivities. However, even with this wide range of infectivity, at low doses the
exposure part of dose response relations becomes dominant: overall uncertainty in the predicted
infectivity may be about 10-fold (Exhibit 4.3).
Six-Year Review 3
Technical Support Document for LT2
4-11
December 2016

-------
Exhibit 4.3 A: Dose Response Relations for the Four Isolates (TAMU, Iowa, UCP
and Moredun); B: Quantile Contours of the Predicted Dose Response Relation
Generalized from the Four Curves in A; C: Low Dose Extrapolated Dose
Response Relations for the Four Isolates
1
0.8
0.6
f?nf(dose)
0,4
0.2
1
0.8
Fjydose)
0.2
0
TAMU
Iowa
UCP
Moredun
1a1 io° io1 102 103 104 105 106
dose
A
1CT1 10° 10' 102 103 10» 10fi 106
dose
B
0.01—
log10(dose)
The Center for Advanced Microbial Risk Assessment (CAMRA) has a QMRA wiki for current
quantitative information and knowledge developed for QMRA (Enger, 2013). The dose response
section of that wiki contains CAMRA's summary and analysis for C. parvum and C. hominis
dose response. Although this wiki has not been subject to the same peer review as manuscripts,
the wiki provides best fit models and optimized parameter values (obtained via 10,000 bootstrap
iterations) for the five published Cryptosporidium isolates. EPA presents a summary of those
results in Exhibit 4.4.
Six-Year Review 3
Technical Support Document for LT2
4-12
December 2016

-------
Exhibit 4.4 Best Fit Models and Optimized Parameter Values for Cryptosporidium
Isolates (CAMRA)
Strain
Best Fit Model
Optimized
Parameter(s)
id50
Probability of Infection
for Dose=1 Oocyst
TAMU
Exponential
k =0.0572
12.1
0.06
TU502
beta-Poisson
a=0.27 N50=16.8
16.8
0.14
Iowa
Exponential
k =0.00419
165
0.004
Moredun
beta-Poisson
a=0.114 N50=455
455
0.07
UCP
beta-Poisson
a=0.145 N50=179
179
0.07
Mitchell-Blackwood (2010) used a Bayesian hierarchical modeling approach to evaluate the
dose-response relationship for six Cryptosporidium strains (based on the human challenge study
data from the University of Texas). The results indicated that individual (nonhierarchical)
models rank highest and that three of the data sets (UCP, TU502 and Moredun) do not fit the
exponential model. Moreover, maximum-likelihood estimation (MLE) estimates indicated that
goodness of fit was achieved by the beta-Poisson model for all data sets, but the improvement
over the exponential model was not statistically significant for the data sets that fit the
exponential-based model. Parameter value estimates for TAMU, UCP and Iowa were consistent
with the MLE and hierarchical models reported by Messner et al. (2001). All individual models
generally agree with the MLE estimates for the Cryptosporidium spp. isolates.
An et al. (2011) conducted a quantitative health risk assessment of Cryptosporidium in rivers in
China that included an extensive and novel dose response assessment based on five strains of
Cryptosporidium. They indicate that Cryptosporidium infectivity modeling has typically
neglected the virulence differential among strains. Different strains of pathogens normally
coexist in the environment, and some strains may not be known, making it impossible to predict
the infection rate of each strain using its specific dose-response curve. To solve this problem, An
et al. used mixed dose-response curves to predict the infection rate at a specific dose by bootstrap
resampling, based on all of the existing dose-response patterns of different Cryptosporidium
strains. Thus, the authors assumed that all strains agreed with the same dose-response curve type,
with their specific parameter values (a and P) based on the difference in virulence, with the curve
type selected according to the model selection criteria. They fitted the dose-response curves
using the beta-Poisson function and obtained curves for different isolates based on groups of
parameters (a, P), which they resampled using the bootstrapping method with 2000 trials. EPA
provides the resultant dose response relationship graphically in Exhibit 4.5. The red circles in
Exhibit 3.5 indicate the observed dataset from the healthy adult volunteers, and the blue lines
represent the bootstrap predicted curves. The bootstrap resampling curves converge with
decreasing Cryptosporidium dose, which agrees well with previously published dose-response
relationships for individual strains.
Six-Year Review 3
Technical Support Document for LT2
4-13
December 2016

-------
Exhibit 4.5 Fitting Dose-Response Curves of Infection Probability for Healthy
Adult Volunteers and Intake of Cryptosporidium Oocysts
The following discussion uses the probability that ingestion of a single organism will cause
infection (equivalent to the dose response parameter (r) from an exponential distribution or the
expected value of r) to compare the relative level of infectivity used in the LT2 risk assessment
to other recently published assessments.
As summarized in this section, An et al. (2011) conducted a quantitative risk assessment of
Cryptosporidium in rivers in China that included an extensive dose response assessment based
on five strains of Cryptosporidium. To address the issue of multiple Cryptosporidium strains
potentially present in the environment they used mixed beta-Poisson dose-response curves based
on bootstrap resampling. Inspection of the resultant family of dose response curves (Exhibit 4.5)
indicates that the range of probabilities that a single organism will cause infection (that is, where
the log dose in Exhibit 4.5 is 0) is similar to mean values from the LT2 analysis, although the
resultant range of r values (probability of infection due to ingestion of one oocyst) is narrower
than the range used in the LT2 analysis.
Cummins et al. (2010) employed a quantitative Monte Carlo simulation model to evaluate the
annual risk of infection from Cryptosporidium in tap water in Ireland. In this study an
exponential dose response model was used with an r = 0.00526 (as reported in Pouillot et al.
(2004)). This relative level of probability that a single organism will cause infection is lower than
that used in the LT2 EA, and corresponds more closely to the infectivity of the TAMU isolate
than an unknown mixture of isolates based on all of the available challenge study data.
The CAMRA QMRA wiki contains a summary and analysis for C. parvum and C. hominis dose
response. Comparing the optimized dose response relationships for each of five Cryptosporidium
strains to the infectivity of the "environmental strains" used in the LT2 analysis indicates that the
o
-4 -2 0 2 4 6 8 10 12 14
Dose(D)( natural logarithm)
Six-Year Review 3
Technical Support Document for LT2
4-14
December 2016

-------
relative level of infectivity used in the LT2 is at the upper range of the infectivity for the
optimized individual strains (range 0.004 to 0.14).
WHO conducted a risk assessment for Cryptosporidium in drinking water (WHO, 2009) using
the results from the Cryptosporidium human challenge studies for four isolates (Iowa, TAMU,
UCP and Moredun) as a basis. As summarized previously, the modeling approach was a two-
level model, the lower hierarchical level representing variation in isolates, and the upper level
representing variation among hosts. The beta-Poisson dose response model is used with the
parameters (a ,(3) taken from (joint) distributions, describing the "between isolates" variation.
This approach generalizes the heterogeneous infectivity to the "group" level for isolates, treating
the four data sets as an n=4 sample from the presumed population of environmental oocysts. This
distribution produces a predictive dose response relation that covers a wide range of infectivities
and appears to suggest substantial uncertainty, as also suggested by the LT2 dose response
modeling. The median infectivity and associated uncertainty predicted by this modeling
approach are similar to that suggested by the set of models employed in the LT2 assessment.
4.1.1.3 Review of New Studies on Viable and Infectious Oocysts
Johnson, Rochelle and their colleagues (Johnson et al., 2010; Rochelle et al., 2012) conducted a
study to determine the prevalence of infectious Cryptosporidium in conventionally treated
drinking water. The authors analyzed a total of 370 samples comprising 349,053 L of drinking
water. Sample volumes ranged from 84 to 2,282 L, with an average of 943 L (the average
recovery efficiency was 71 percent for Easy Seed oocysts, 42 percent for ColorSeed oocysts and
80 percent (range 3-200 percent) for matrix spikes using freshly shed oocysts). None of the 370
finished water samples produced infections that were detected by the cell culture
immunofluorescence microscopy assay. In freshly shed C. parvum oocysts, the mean infectivity
was 10.8 percent. Based on these results, the authors report that even in freshly shed oocysts,
only a small portion of the oocysts are capable of initiating infection. They suggest a typical
range of 5-15 percent (5-15 foci per 100 oocysts inoculated onto the cell monolayer).
Gennaccaro et al. (2003) collected water samples throughout several water reclamation facilities
and analyzed them for the presence of infectious C. parvum by the focus detection method-most-
probable-number (FDM-MPN) cell culture technique. Their results revealed the presence of C.
parvum oocysts in 67 percent of the 15 final disinfected effluent samples, but infectious oocysts
in only 40 percent of those samples. In 100 liters of the sampled final disinfected effluents with
oocysts present, the authors found an average of 28 total oocysts and average of 7 infectious
oocysts, indicating that roughly 25 percent of the oocysts were infectious. Huffman et al. (2006)
conducted a follow up study to evaluate the presence and infectivity of cysts and oocysts in
primary effluent and reclaimed water using cell culture FDM-MPN for the detection of
infectious Cryptosporidium oocysts and animal infectivity analysis for the detection of infectious
Giardia cysts. A total of approximately 120 oocysts were determined to be present in the
composite reclaimed water sample of 1,870 L, but the infectivity of the oocysts could not be
assessed because of the high number of algal cells on the cell culture monolayer. For primary
effluent the authors report an MPN of 105 infectious oocysts/100 L.
Johnson et al. (2012) compared the three most commonly used assays for detecting
Six-Year Review 3
Technical Support Document for LT2
4-15
December 2016

-------
Cryptosporidium spp. infections in cell culture: immunofluorescent antibody and microscopy
assay (IFA), PCR targeting Cryptosporidium spp.-specific DNA, and reverse transcriptase PCR
(RT-PCR) targeting Cryptosporidium spp.-specific messenger ribonucleic acid (RNA). They
used a mean oocyst infectivity of 14 percent to compare assays. This value is similar to other
studies using the same definition of infectivity, which have reported C. parvum infectivity values
of 8 to 9.5 percent (Bukhari et al., 2007), 10 to 22 percent (Sifuentes and Di Giovanni, 2007) and
5 to 14 percent (Rochelle et al., 2001). All assays detected infection of flow cytometry-
enumerated C. parvum oocysts, including assays with low doses of one, three or five oocysts.
The RT-PCR assay, IFA and PCR assay detected infection in 23 percent, 25 percent and 51
percent of monolayers inoculated with three C. parvum oocysts and 10 percent, 9 percent and 16
percent of monolayers inoculated with one oocyst, respectively. All methods also detected
infection with C. hominis although the authors noted that C. hominis was found to be less
infective than C. parvum in all three assays. The PCR assay was the most sensitive, but it had the
highest frequency of false positives. IFA was the only infection detection assay that did not
produce false positives. Recent studies by EPA showed a much improved cell culture/IFA assay,
resulting in much higher infectivity rates (>80 percent) than previously reported (Varughese et
al., 2014). Based on these results, the authors suggest that cell culture with IFA detection is the
most appropriate method for routine and sensitive detection of infectious C. parvum and C.
hominis in drinking water.
Keegan et al. (2008) evaluated the effect of the water treatment chemical aluminum sulfate
(alum) alone and in conjunction with chlorine or chloramine disinfection on Cryptosporidium
oocyst infectivity to determine whether the treatment processes were more effective on aged
oocysts. They used an assay that combined cell culture and real-time PCR techniques. The
infectivity of fresh and temperature-aged oocysts (stored up to six months at 4 or 15°C) was
unaffected by exposure to a range of alum doses in standard jar test procedures and dissolved air
flotation processes, and by subsequent exposure to chlorine or chloramine.
Kar et al. (2011) conducted a comparative evaluation of time-dependent changes in the viability
of purified C. parvum oocysts by means of different excystation methods. Pretreatment with
hypochlorite markedly enhanced the excystation of younger oocyst samples but did not increase
excystation rates of 9- or 12-month-old oocysts. A cell culture-PCR assay was consistent with
excystation trials including oocyst pretreatment. Depending on the excystation method (with or
without pretreatment), viability of sporozoites is not necessarily linked with the excystation rate,
especially for short-term storage of oocysts.
Lalancette et al. (2010) reported a dual direct detection method using differential
immunofluorescent staining that allows detection of both oocysts and cell culture infection foci.
The key trigger for oocyst stimulation was acidification. Addition of a low concentration of D-
glucose (50 mM) to the infection media increased rates of infectivity, while a higher dose (300
mM) was inhibitory. With this method, the authors reported that it was possible to determine the
numbers of total and infectious oocysts for a given sample in a single analysis.
Chappell et al. (2011) studied the infectivity of C. meleagridis oocysts (TU1867) in Madin-
Darby bovine kidney (MDBK) cell monolayer and compared the results with a C. parvum isolate
(GCH1). The data demonstrated that C. meleagridis was capable of infecting MDBK cells.
Infectivity of TU1867 and GCH1 was proportional to the inoculating dose. At 106 oocysts per
Six-Year Review 3
Technical Support Document for LT2
4-16
December 2016

-------
well, the infectivity between the two isolates was comparable but differed at the lower doses.
Chappell et al. also evaluated four concentrations of C. muris oocysts representing oocyst:cell
ratios of 1:1, 2:1, 3:1 and 4:1 by adding them to HCT-8 cell cultures and allowing them to
develop for approximately 18 hours. All concentrations established infection in the cells. These
data confirmed that human enterocytes are susceptible to C. muris infections. Further, the degree
of infectivity was dose-dependent. The authors compared the growth rate for C. muris to that of
C. parvum and C. hominis oocysts. All three Cryptosporidium species appeared to be roughly
equal in their capacity to invade enterocytes, as similar numbers of foci were seen. C. parvum
and C. hominis developed at a similar rate. In comparison, C. muris developed more slowly over
the entire course of the experiment and reached only 67 percent of C. parvum/C. hominis growth
rates at 48 hours. These data suggest that the cycle time for C. muris may be longer than for the
other two species or that merozoite attrition may be higher. Nevertheless, these data indicate
potential uncertainty in the infectivity to humans in Cryptosporidium spp. enumerated from
environmental samples.
Feng et al. (2006) describe a method that tracks and quantifies the early phase of attachment and
invasion of C. parvum sporozoites using a fluorescent dye. They labeled newly excysted
sporozoites with the amine-reactive fluorescein probe carboxyfluorescein diacetate succinimidyl
ester (CFSE) using an optimized protocol. The initial invasion of cells by labeled parasites was
detected with confocal microscopy. The authors quantified the infection of cells by flow
cytometry. Comparative analysis of infection of cells with CFSE-labeled and unlabeled
sporozoites showed that the infectivity of C. parvum was not affected by CFSE labeling.
Quantitative analysis showed that C. parvum isolates were considerably more invasive than C.
hominis isolate TU502. CFSE labeling permitted the tracking of the initial invasion of C.
parvum.
The literature summarized above revealed that additional information is available to estimate the
proportion of infectious Cryptosporidium relative to total Cryptosporidium present in source
waters. However, comparison of results from the various studies requires careful consideration,
in part because it is possible that the proportion of infectious environmental oocysts may vary
widely from site to site (Schets et al., 2005; Lalancette et al., 2010), and because researchers use
a variety of methods to define infective oocyst status, and the results from the different methods
may not be directly comparable. For example, Rochelle et al. (2012) suggest a typical infectivity
range of 5-15 percent and report mean infectivity of 14 percent. These findings are similar to
other studies using the same definition of infectivity, which have reported C. parvum infectivity
values of 8 to 9.5 percent (Bukhari et al., 2007), 10 to 22 percent (Sifuentes and Di Giovanni,
2007) and 5 to 14 percent (Rochelle et al., 2001). However, studies by Varughese et al. (2014)
reported much higher infectivity rates 80-90 percent, in several cell lines evaluated, including
those used in previous studies. On the other hand, the Cryptosporidium challenge studies used
oocyst excystation to confirm infectivity. Li et al. (2010a) used the infection status of neonatal
mouse pups to test oocyst infectivity, and Garvey et al. (2010) reported on the use of in vitro
HCT-8 cell culture-quantitative PCR (qPCR) assay and the in vivo severe combined
immunodeficiency-mouse bioassay for evaluating critical factors that reduce or eliminate
infectivity of C. parvum. In that respect, Garvey et al. (2010) reported that the use of this HCT-8
cell culture assay is equivalent to using the mouse-based infectivity assay. Further complicating
interpretation of the available data, Theodos et al. (1998) report that the infectivity results from
the gnotobiotic pig assay model appear to more closely mimic human infection than either cell
Six-Year Review 3
Technical Support Document for LT2
4-17
December 2016

-------
culture or other animal models. Finally, EPA notes that most of the literature either focuses on
methods development or reports on the proportion of samples for a particular source that are
positive for infectious Cryptosporidium. For example, 25 percent of source water sites, 7 percent
of surface water samples and 7 percent of backwash samples were reported to be positive for
infectious oocysts (Di Giovanni et al., 1999), 1.4 percent of filtered, finished drinking water
samples were reported to be positive for infectious Cryptosporidium, and 27 percent (22 of 82)
of treatment plants reported at least one positive sample (Aboytes et al., 2004) and no detectable
levels of infectious Cryptosporidium oocysts in 370 samples collected from 14 drinking water
treatment plants (WTPs) in the US (Johnson et al., 2010).
Another potentially important consideration is the proportion of Cryptosporidium that was or
was assumed to be infectious during the challenge studies. As indicated previously, the human
challenge study data evaluated in the LT2 EA included data from a total of six Cryptosporidium
strains (isolates). During these studies, the Iowa, UCP, TAMU and Moredun oocysts were shown
to have an excystation rate of 85 percent or greater at the time of challenge (Okhuysen et al.,
1999; 2002). For the TU502 study, excystation rates of oocysts delivered to volunteers were
between 67 percent and 80 percent at the time of volunteer challenge depending on the oocyst
batch (Chappell et al., 2006). As indicated previously, the LT2 risk assessment incorporated an
assumption that all oocysts in the challenge studies were infectious (recognizing that this
assumption was meant to offset the percent of oocysts that were not recovered). Assuming that
the excystation rate is a reasonable proxy for the proportion of Cryptosporidium that is infectious
(Smith et al., 2005), this assumption appears reasonable.
Outbreaks and human cryptosporidiosis cases have been related to handling of animals or living
near them (Dreelin et al., 2014; Jiang et al., 2014; Lange et al., 2014; DeSilva et al., 2015;
Zahedi et al., 2015). Animal waste in run off into source water has been identified as the source
of Cryptosporidium outbreaks (DeSilva et al., 2015). Zahedi et al. (2015) in a meta-analysis
conclude that with the increasing human settlement encroaching upon wildlife habitats, the
chance increases for Cryptosporidium in animal waste run-off to occur in drinking water
catchment areas such as source water reservoirs or UCFWRs increases thus increasing the
chance of human infection. In addition to the Cryptosporidium species already referenced in this
section, other species involved in zoonotic transmission are C. andersoni, C. ubiquitum, C. canis,
C. erinacei and C. cuniculus found in such animals as cattle, goats, hedgehogs, rabbits, dogs and
rodents (Chalmers et al., 2011; Jiang et al., 2014; Zahedi et al., 2015). C. meleagridis, identified
in this section, is found in migratory birds (Zahedi et al., 2015).
4.1.1.4 Summary of Findings from New Data / Information Relevant to Key Infectivity
Dose-Response Issues
The following is a brief summary of findings for the key aspects related to the infectivity portion
of the dose-response modeling of Cryptosporidium based on the information presented above.
• Are there additional human challenge studies that can also be used to parameterize the
infectivity dose-response function?
Although there is a limited amount of human challenge study data that were not
previously incorporated into the LT2 risk assessment, it is not expected that these data
would influence the LT2 dose response relationship in a significant way. The set of six
Six-Year Review 3
Technical Support Document for LT2
4-18
December 2016

-------
studies used in the LT2 represents the most comprehensive set of human feeding studies
that has been used to date for the purposes of dose response modeling. With respect to
revision to the LT2 dose response modeling, one possibility is to augment this set with
the results from Chappell et al. (2011), who established that healthy adults can be
infected and become ill from ingestion of C. meleagridis and C. muris oocysts. In these
studies, the authors examined the infectivity of C. meleagridis and C. muris in healthy
adults—five volunteers were challenged with a single dose of C. meleagridis and six
volunteers were challenged with a single dose of C. muris (105 oocysts in each case). The
results indicate that four of the C. meleagridis volunteers had diarrhea and three had
detectable fecal oocysts, and all six of the C. muris volunteers became infected, with two
experiencing diarrheal illness. These data could be added to the database of challenge
study data, although they likely would only be valuable if aggregated with the other
challenge study data since there is only one data point (dose) for each Cryptosporidium
strain. Moreover, it is extremely unlikely that addition of these data would impact the
dose response relationship to a meaningful degree at the low doses that are of particular
interest for estimating risks from drinking water.
•	Are there studies suggesting alternative model forms from the exponential model used in
the LT2 that should be considered?
Several researchers have published modified or enhanced versions of the dose response
relationships based on the Cryptosporidium human challenge studies. The general
consensus seems to be that the IDsos vary from strain to strain and that there is variability
among hosts also. Few researchers have modeled the potential range of infectivity
associated with an unknown environmentally relevant exposure to Cryptosporidium as
was done in the LT2 analysis. Comparison of the reported dose response relationships in
the literature and used in peer reviewed risk assessments with the LT2 expected value for
"r" for an unknown environmental isolate indicates that the LT2 values appear quite
reasonable and are on the health protective side of the spectrum of those used. There are
three notable exceptions—An et al. (2011) and WHO (2009) did model the presumed or
unknown population of environmental oocysts. The resultant family of dose response
curves reported by An et al. (2011) is similar to those used in the LT2 EA, although the
resultant range of "r" values is narrower than the range used in the LT2 analysis, and the
median infectivity and associated uncertainty predicted by the WHO (2009) dose
response modeling are similar to those suggested by the set of models employed in the
LT2 assessment. The third exception, which is only a research suggestion, is a norovirus
dose-response modelling research paper by Messner et al. (2014) that recommends a
simpler model than the LT2 group of models be applied to pathogens like
Cryptosporidium spp., which generates an acute, strong immune response in exposed
humans.
•	Are there studies or data suggesting alternative approaches to considering immunity in
the exposed population?
EPA identified limited additional data to suggest alternative approaches to considering
the potential risk-based implications of immunity to infection and/or differential
immunity in the population. The approach suggested by Teunis et al. (2002b) and
implemented in the WHO risk assessment report (WHO, 2009) using individual IgG
levels as a proxy for susceptibility to partly control host variation in the general
Six-Year Review 3	4-19	December 2016
Technical Support Document for LT2

-------
population is interesting and may be further considered if EPA conducts any future
revisions to the LT2 risk assessment. Currently, sufficiently robust and detailed
population level IgG data do not appear to be available to implement this approach in the
LT2 risk assessment. However, this modeling approach may be valuable for more
research-oriented modeling, and may provide insights with respect to population level
risks, risks due to rare but potentially severe events (treatment plant failures or large
precipitation events, for example), or the potential efficacy of risk mitigation efforts.
• Are there studies or data that provide additional information on the assumption
concerning the "v" ratio of infectious oocysts?
The "v" ratio is the ratio of the fraction of infectious oocysts in the environment
(numerator) to the fraction of infectious oocysts in doses tested in clinical challenge
studies (denominator). In the LT2 risk assessment, EPA modeled the numerator as a
triangular distribution (minimum, 30 percent, likeliest 40 percent and maximum 50
percent) based on data from cell culture oocyst infectivity studies, and it assumed that the
denominator was unity based on oocyst excystation. EPA based that estimate on the
physical structure of the oocysts observed and research data from cell culture oocyst
infectivity studies. LeChevallier et al. (2003) found that when using method 1622/23 and
cell culture assays, 60 of the method 1622/23 samples and 22 of the cell culture
infectivity assay samples tested positive, resulting in an estimate of 37 percent infectious
oocysts.
There are data available in the literature that EPA could use to refine the characterizations
for both the numerator and denominator in the "v" ratio. However, substantial uncertainty
exists with respect to the interpretation of these data for the purposes of revising the LT2
risk assessment and it is unlikely any revision would result in meaningful changes in
treatment limits. With respect to the numerator, Rochelle et al. (2012) suggest a typical
infectivity range of 5-15 percent based on their studies, which used a cell culture
definition of infectivity. Garvey et al. (2010) reported that the use of HCT-8 cell culture
assay is equivalent to using the mouse-based infectivity assay. But, Theodos et al. (1998)
report that the infectivity results from the gnotobiotic pig assay model appear to more
closely mimic human infection than either cell culture or other animal models. With
respect to the denominator, the Cryptosporidium challenge studies used oocyst
excystation as a surrogate to confirm infectivity, and reported excystation rates of 67-85
percent or greater at the time of challenge.
Based on the above information and data, it is apparent that diverse methods have been
used to report oocyst infectivity and if these reported results are to be used to modify the
"v" ratio in the LT2 risk assessment, careful consideration would be needed to ensure that
the data used for the numerator and denominator are compatible and/or directly
comparable. It is not known at this time, if it is possible to directly and rigorously
compare the data available to characterize the "v" ratio.
4.1.2 Morbidity
As described at the beginning of Section 4.1, the second component of the Cryptosporidium
dose-response model used for the LT2 addresses the probability of an infected individual
Six-Year Review 3
Technical Support Document for LT2
4-20
December 2016

-------
experiencing cryptosporidiosis (the probability of having an illness given an infection). This
"morbidity" component of the dose-response model recognizes that not all infections with
Cryptosporidium result in the gastrointestinal illness or other effects associated with this
pathogen. There are some individuals who may have an asymptomatic infection where the
individual sheds the organism in feces but exhibits no signs of adverse effects.
The dose-response function implemented in the LT2 risk and benefits assessment is an expansion
of the basic infectivity exponential dose response model described previously having the form:
Pi= 1 -Q-dvr
where d is the dose of Cryptosporidium ingested, v is the fraction of Cryptosporidium ingested
expected to be viable and infectious, and r is the exponential model parameter obtained from the
human feeding studies. EPA notes that this model form considers the risk of an infection given a
single dose of as done in the human feeding studies. Because EPA focused the LT2 analysis
on estimating endemic cases of cryptosporidiosis from repeated potential exposure to relatively
low levels of Cryptosporidium in drinking water over time, the dose-response model EPA used
to include morbidity was also modified to address repeated exposures that could result in illness
over the course of one year.
The expanded form of the dose-response model to address the annual endemic morbidity risk
used for the LT2 was:
Pm = M* (1 _e_c*/*v*r*n)
Here, Pm is the probability of an individual having one (or more) illness(es) over the course of a
year, M is the morbidity factor or fraction of infections that result in an illness, C * I are
concentration of Cryptosporidium in drinking water (C) and the amount of water ingested per
day (/), v is the fraction of Cryptosporidium ingested expected to be viable and infectious, r is
the exponential model parameter obtained from the human feeding studies, and n is the number
of days per year that the individual consumes that drinking water.
EPA discussed the v and r parameters in the model in the preceding section on infectivity. The C,
/ and n factors in the morbidity dose-response model are reflective of information on the
occurrence and exposure to Cryptosporidium and are beyond the focus of this chapter. The
remainder of this section therefore focuses on the morbidity factor, M.
EPA analyzed available literature at the time it was developing the LT2 and identified several
human challenge studies with data that addressed the fraction of those who became infected who
also showed signs of illness. The interpretation of these data is somewhat complicated by the
immune status of the study population relative to the population at large. Some of the
preliminary human ingestion trials were conducted on healthy individuals with no evidence of
previous C. parvum infection (DuPont et al., 1995). Other studies challenged individuals with
existing antibodies or rechallenged those who had participated in earlier studies.
DuPont et al. (1995) found that 39 percent of those infected had clinical cryptosporidiosis. Haas
et al. (1996) provided information based on the same data also suggesting a morbidity rate of 39
percent, but also computed 95 percent confidence limits of 19 and 62 percent. Another study
Six-Year Review 3	4-21	December 2016
Technical Support Document for LT2

-------
found that after repeated exposure to C. parvum (Iowa strain), the morbidity rate was the same as
for the initial exposure in reinfected subjects (Okhuysen et al., 1998). Okhuysen et al. (1998)
also found that 58 percent of their subjects who received doses of Cryptosporidium developed
diarrhea, which is an underestimate of morbidity since symptoms other than diarrhea contribute
to the morbidity rate. However, these subjects were given doses higher than those projected in
water supplies.
Given these results, as well as morbidity variability associated with C. parvum during reported
outbreaks, EPA recognized that the morbidity rate may vary with the type of strain to which a
population is exposed, as well as with the immune status of the exposed population. However,
the prevalence of strains and the immune status of the population are unknown, and therefore,
EPA could not explicitly factor these into the LT2 risk assessment.
Recognizing this, EPA included a characterization of uncertainty around the value for morbidity
(M) used in the dose-response model. However, the quality of available data at that time did not
support making more than a generalized estimate of the range and nature of uncertainty. The
underlying data supported the use of only a distribution with a central tendency and provided
information to establish reasonable bounds. As a result, EPA modeled morbidity as an uncertain
variable having a triangular distribution.
Analysis of the reviewed research resulted in a mode (central tendency) of 50 percent, a lower
bound of 30 percent and an upper bound of 70 percent for the triangular uncertainty distribution
of M. EPA identified the following limitation in the research and considered this in the derivation
of the above values: the Okhuysen et al. (1998) results based on diarrheal rates are probably an
underestimate.
The central tendency (mode) for the distribution used in the risk assessment model is 50 percent.
This is a bit below the Okhuysen et al. (1998) results (58 percent), but above the values estimated
by DuPont et al. (1995) and Haas et al. (1996) (39 percent). These studies used the Iowa isolate,
and a simple average of them results in a value of 48.5 percent. The mode was rounded up to 50
percent to account for the apparent underestimation of these studies, as noted above.
The upper bound for the distribution used in the risk assessment model is 70 percent. The upper
bound was set above the 95 percent confidence limit of 62 percent estimated by Haas et al.
(1996). This reflects that the absolute limit of the triangular distribution would reasonably be
above that 95 percent confidence limit and the apparent underestimation of these studies, as noted
above. The difference in the upper bound (70 percent) and the Haas et al. 95 percent confidence
limit (62 percent) represents only 3 percent of the triangular distribution, indicating that the upper
tail of the triangular distribution is comparable to the upper portion of Haas's distribution.
The lower bound for the distribution used in the risk assessment model is 30 percent. The lower
bound was set higher than the 19 percent estimated by Haas et al. (1996). While this bound does
not encompass the lower 95 percent confidence level in the distribution used in the risk
assessment, it does account for the apparent underestimation in the studies.
Some studies indicate that the morbidity rates increase at higher doses (DuPont et al., 1995).
However, for this risk assessment, the morbidity rate is independent of dose. After examining the
Six-Year Review 3
Technical Support Document for LT2
4-22
December 2016

-------
potential impact, EPA determined that a higher-morbidity-at-higher-dose effect was not directly
relevant to this analysis. As noted previously, the morbidity dose-response model is structured to
quantify the endemic rate of illness from persistent but relatively low levels of Cryptosporidium,
not the higher levels that might result in a large outbreak. The underlying dose data, both as
measured and modeled, reflect at most a few oocysts per day for individuals, and generally no
more than one. In the risk assessment model, the portion of the risk posed by the small portion of
the population ingesting even an expected number of 2 oocysts/L is negligible; and the portion of
the risk posed by people ingesting three or more oocysts/L is virtually zero. Thus, the results of
the analysis would not be affected by using increased morbidity rates with significantly higher
doses since these doses are not anticipated to occur under normal operating conditions.
As noted in Section 4.1, EPA conducted a literature search to try to identify studies published
after the LT2 providing additional information on morbidity and the morbidity factor. Although a
small number of studies as identified in the September 2013 literature search addressed aspects
of Cryptosporidium morbidity, none of those studies presented new data that further informed
the estimate of the morbidity factor, M, or the uncertainty range central tendency and bounds
used in the LT2 analysis.
There was one study found that developed a dose-response relationship using human challenge
data and the illness (diarrhea) endpoint directly rather than using infectivity and applying a
morbidity factor to those results as was done in the EA. Englehardt and Swartout (2006)
developed a conditional parametric dose-response function for GI illness (diarrhea) using human
challenge data for five C. parvum isolates (Iowa, TAMU, UCP, Moredun and Peru). The
function is a generalized beta-Poisson illness dose-response relationship for the population as a
whole. Use of this form is demonstrated in a predictive Bayesian dose-response assessment for
cryptosporidiosis. The authors reported that a daily exposure to 5.0 x 10 7 oocysts for 365 days
would result in an illness risk of 10 4, It is important to note that this result of 5 x 10"7oocysts per
daily exposure for an annual illness risk of 10"4appears to be inconsistent with the results these
authors reported in Englehardt and Swartout (2004) of 6 x 10"6 oocysts per daily exposure for an
annual infection rate of 10"4. One might expect (as Englehardt and Swartout commented
themselves in the 2004 study) that the number of oocysts per exposure for the illness endpoint
would be greater than the number for the infection endpoint, whereas their results here indicate a
daily exposure to oocysts that is an order of magnitude less for illness than for exposure. The
authors did not comment on this in their 2006 paper. It is possible that this difference is partly
related to the inclusion of data for the two additional strains (Moredun and Peru) in the modeling
done for the illness endpoint that were not included in the modeling done for the infection
endpoint. The illness rate for those two strains at the various doses at which they were tested
always exceeded the rates observed for the other three strains.
Another illness endpoint or endpoints could be considered in future dose-response models. Rehn
et al. (2015) report that gastrointestinal and joint symptoms can occur several months after being
infected with Cryptosporidium, and suggest that Cryptosporidium be considered a cause of some
of these unexplained symptoms. Not considering these later occurring endpoints may
underestimate the morbidity from Cryptosporidium.
Although not directly useful for characterizing the morbidity factor, M, in the dose-response
model, EPA did identify some additional recent information on the incidence of
Six-Year Review 3
Technical Support Document for LT2
4-23
December 2016

-------
cryptosporidiosis. In the most recent documented U.S. drinking water-associated
Cryptosporidium outbreak data available (Exhibit 4.6), there was one drinking water-associated
outbreak caused, in part, by Cryptosporidium (out of 36 total outbreaks) between 2007 and 2008;
Cryptosporidium was just one of the multiple etiologic agents identified with this outbreak
(CDC, 2011). Centers for Disease Control and Prevention (CDC) reported one drinking water-
associated outbreak of cryptosporidiosis between 2005 and 2006 (out of 20 total outbreaks)
(CDC, 2008). There were two Cryptosporidium related outbreaks reported from 2009 to 2010 by
CDC (2013). One was at a vacation / rental property in Vermont in January 2010 involving 34
AGI cases (Cryptosporidium sp) and the other was at a private residence in Pennsylvania in July
2010 involving 10 cases (Campylobacter jejuni and Cryptosporidium spp). In 2013, a
community water supply outbreak occurred in Baker City, Oregon (CDC, 2015a) with an
estimated 2780 illnesses (C. parvum) (DeSilva, 2015). This was the first cryptosporidiosis
outbreak in almost 20 years reported from a US PWS (DeSilva, 2015). The estimated number of
illnesses were calculated based on a survey of residents that yielded an attack rate of 28 percent
which was applied to the population of the city. The incident demonstrates the importance of
multiple barrier protection required by the LT2. The source of the incident was considered run-
off of cattle feaces into the watershed that provided Baker City with its drinking water supply
(DeSilva, 2015). The Baker City water system had not yet taken measures to treat for
Cryptosporidium under the LT2 because it had not yet been required to so. While the water
system used chlorine to disinfect, it did not filter or use ultraviolet (UV) to treat its water
(DeSilva, 2015).
Exhibit 4.6 Cryptosporidium Outbreaks Associated with Drinking Water, by Year:
Waterborne Disease and Outbreak Surveillance System, United States 2005-2010
(CDC, 2008; 2011; 2013; 2015a)
State
Year
Etiology
Cases
Water System
Water Source
Oregon
2013
Cryptosporidium
23 (lab
confirmed, but
2780 estimated)
Community
Surface water from
mountainous terrain
to underground
transmission lines
Illinois
2008
Cryptosporidium;
Giardia: Shigella sonnei
41
Community
Lake
Ohio
2006
Cryptosporidium
10
Community
Well
U.S. Cryptosporidium surveillance data indicated a morbidity rate of 2.5 cases of
cryptosporidiosis per 100,000 population and 2.9 cases of cryptosporidiosis per 100,000
population in 2009 and 2010, respectively (CDC, 2012). A total of 2.6 percent and 3.3 percent of
cases were reported to be associated with a detected outbreak in 2009 and 2010, respectively
(CDC, 2012). Rates from 2005 to 2010 ranged from 2.3-3.9 cases per 100,000 population,
peaking in 2007 (CDC, 2010). A decrease in large cryptosporidiosis outbreaks was reported in
2009	and 2010. The annual proportion of cases associated with outbreaks reported in 2009 and
2010	were the lowest since national reporting began in 1995 (CDC, 2012).
The CDC database substantially understates the number of identified and reported outbreaks and
the actual incidence of waterborne disease cases (Craun and Calderon, 1996; National Research
Council, 1997). Many factors contribute to this under-reporting. Detection, investigation and
Six-Year Review 3
Technical Support Document for LT2
4-24
December 2016

-------
reporting of outbreaks is often incomplete and the level of surveillance varies across states and
localities. Outbreaks often go undetected because many people experiencing gastrointestinal
illness do not seek medical attention. Corso et al. (2003) reported that during the 1993 outbreak
in Milwaukee, medical care was sought in approximately 12 percent of all cryptosporidiosis
cases. In cases where those who are ill seek medical attention, the testing may not identify the
pathogen. Physicians and patients often lack sufficient information to attribute gastrointestinal
illness to any specific origin, such as drinking water, and few states have an active outbreak
surveillance program (USEPA, 2006a).
Limitations in analytical methods may lead to pathogens not being detected even if they are
present in drinking water being investigated as the potential source of an outbreak. As a result,
outbreaks may not be traced to a drinking water source. In addition, an unknown but probably
significant portion of waterborne disease is endemic (i.e., isolated cases not associated with an
outbreak) and, thus, is even more difficult to recognize (USEPA, 2006a).
Due to this underreporting, the actual incidence of cryptosporidiosis associated with drinking
water is unknown. However, investigators have extrapolated indications of this incidence rate
from different sources. Mead et al. (1999) estimated approximately 300,000 total physician visits
involving cryptosporidiosis annually, with 90 percent of these attributed to drinking water,
recreational water and secondary transmission.
CDC's surveillance data for 2001 show 1.5 laboratory-diagnosed cryptosporidiosis cases per
100,000 people (CDC, 2002). During the 1993 Milwaukee cryptosporidiosis outbreak only 739
out of an estimated 403,000 cases were laboratory-confirmed (MacKenzie et al., 1994), or only
one of every 545 cases.
Socioeconomic factors may affect Cryptosporidium infections by placing some individuals in
locations where food or water contamination is more common and more frequent contact with
livestock is likely (Becker et al., 2015). Food inadequacy was found to be a strong predictor of
serological level; individuals in homes reporting some food inadequacy had a higher likelihood
of Cryptosporidium seropositivity. Becker et al. report that there may be a relationship between
food inadequacy and susceptibility to infection through a decrease in the nutritional level which
then reduced immunity of the individual (Katona and Katona-Apte, 2008). Non-white older
individuals and individuals born outside the US were also identified as being at greater risk for
cryptosporidiosis (Becker et al., 2015).
4.1.3 Mortality
As noted in Section 4.1, the third dose-response relationship component EPA used in the LT2 EA
analysis is the mortality factor, that is, the probability of fatality given that an illness has
occurred. This final factor is simply applied to the number of illnesses per year predicted from
the morbidity dose-response function described above to estimate the number of deaths per year
related to cryptosporidiosis.
No general data existed on the rate of mortality from cryptosporidiosis in the United States at the
time of the LT2. To derive mortality estimates, EPA used data from the Milwaukee outbreak,
recognizing that the majority of the associated deaths occurred among the subpopulation with
Six-Year Review 3
Technical Support Document for LT2
4-25
December 2016

-------
immunodeficiency. EPA adjusted those observed fatality data to reflect changes in rates of
illnesses and advanced treatments that have lessened mortality among persons living with
acquired immunodeficiency syndrome (AIDS). EPA used further adjustments to reflect the
differences between the populations of those living in areas served by filtered systems and those
served by unfiltered systems. Since there is considerable uncertainty around the mortality rate
ultimately used in the dose-response model, EPA also conducted a sensitivity analysis that varied
the AIDS mortality rate by +/- 50 percent.
The starting point for the analysis was the mortality rate associated with the Milwaukee
Cryptosporidium outbreak. In that outbreak, 54 people died who had cryptosporidiosis listed on
their death certificates. Of those, 46 also had AIDS listed as an underlying cause of death (Hoxie
et al., 1997). The Milwaukee outbreak had an estimated 403,000 cases of illness (Kramer et al.,
1996). The unadjusted rate for Cryptosporidium mortality among those with AIDS is thus 46
deaths/403,000 illnesses, or 11.41 deaths/100,000 illnesses. The rate for deaths of those not
having AIDS is thus 8 deaths/403,000 illnesses, or 1.98 deaths/100,000 illnesses.
EPA made no further adjustments to the mortality rate for those not having AIDS. A review of
available statistics showed that data to compare the incidence of the other underlying illnesses
(coccidiosis (presumably cryptosporidiosis), viral hepatitis, brain tumor, heart failure and
alcoholic cirrhosis of the liver) between Milwaukee in 1993 and the nation in 1999 or 2000 were
generally unavailable. Even comparison of proxy data (death rates rather than incidence) proved
of little value. Data for Milwaukee were, in general, inconclusive; too few cases were reported to
make statistics meaningful. One factor that could affect the mortality rate for those not having
AIDS is age. Hoxie et al. (1997) did not provide data on the age of those who died in the
outbreak. Although Naumova et al. (2003) found that the rate of gastroenteritis during (and prior)
to the outbreak increased with age, there was no information on whether the elderly have a higher
mortality rate from cryptosporidiosis.
EPA adjusted the Milwaukee mortality rate for those with AIDS to account for the decrease in
the mortality rate among people with AIDS from the time of the Milwaukee incident to 2001 (the
most recent year with comparable data), and the difference in the Milwaukee AIDS population in
1993 to the national AIDS population in 2001. EPA describes these adjustments below; the
adjusted calculation is:
Deaths/100,000 illnesses in the Milwaukee outbreak (11.41) x factor to adjust
for lessened mortality over time among persons with AIDS x factor to adjust
for changes in the prevalence of AIDS in the general population = AIDS-
related deaths per 100,000 cryptosporidiosis illnesses.
The mortality rate for AIDS declined between 1993 and when the LT2 was promulgated due to
the use of combination retroviral therapies and other factors. Combination retroviral therapy
raises the CD4+ cell count, enabling people with AIDS to better fight off infection. Correlations
have been shown between cryptosporidiosis in AIDS patients and CD4+ counts (Pozio et al.,
1997; Inungu et al., 2000). The AIDS mortality rate in 2001 was 4,845 deaths per 100,000 AIDS
population (17,402 deaths in a population of 359,141) (CDC, 2002). In 1993, this rate was
25,963 per 100,000 (45,271 deaths in a population of 173,772) (CDC, 2001). The ratio of these
Six-Year Review 3
Technical Support Document for LT2
4-26
December 2016

-------
rates is 18.4 percent, that is, the rate of deaths among AIDS patients for all reasons in 2001 was
only 18.4 percent of what it was in 1993.
A second adjustment accounted for the difference in the percent of the national population that
was living with AIDS in 2001 and the percent of the Milwaukee population that was living with
AIDS in 1993. EPA calculated this adjustment separately for areas served by unfiltered systems
and filtered systems. EPA used an approximation of the value for populations served by unfiltered
systems, the percentage of the population living with AIDS, which is 0.196 percent (62,349 in a
population of 31,859,141). As an approximation of the percentage in areas served by filtered
systems, EPA used national estimates, less than had been accounted for by unfiltered systems.
The rate for filtered systems is 0.118 percent (based on an AIDS population of 299,912 in a
population base of 253,234,672) (US Census Bureau, 2001; CDC, 2002).
EPA used the percentages of people living with AIDS in 2001 served by filtered and unfiltered
systems separately to adjust and update the 1993 incidence rate of AIDS in Wisconsin. The data
on AIDS incidence and population should represent the same location; however, the areas for
which data are available do not match the exact geography of the areas served. Nevertheless, the
ratios that come from this approach are still useful as approximations, and their use is an
improvement over not including adjustments for these factors at all. In Wisconsin in 1993, the
percentage of the population that had AIDS was 0.017 (862 persons with AIDS in a population of
5,044,318). Extrapolating the Wisconsin data to all populations served by unfiltered and filtered
systems, gives a factor of 11.45 for unfiltered systems (0.196 percent/0.017 percent) and a factor
of 6.93 (0.118 percent/0.017 percent) for filtered systems. The incidence of people living with
AIDS in 2001 in areas served by unfiltered systems is 11.45 times the incidence in Wisconsin in
1993. Similarly, there are 6.93 times as many people living with AIDS in 2001 and served by
filtered systems as there were in Wisconsin in 1993.
Using the Milwaukee mortality rate for those with AIDS and the adjustment factors described
above, the final mortality rate for unfiltered systems (expressed as deaths per 100,000
cryptosporidiosis illnesses) is 24.07 (11.41 x 18.4 percent x 11.45). Similarly, for filtered
systems, the mortality rate for those with AIDS is 14.56 deaths per 100,000 cryptosporidiosis
illnesses.
The risk assessment model uses a combination of mortality rates for those with and those without
AIDS. Thus, adding together these rates yields an overall mortality rate for unfiltered systems of
26.05 deaths per 100,000 cryptosporidiosis illnesses (24.07 AIDS + 1.98 non-AIDS). For filtered
systems, this figure is 16.53 deaths per 100,000 cryptosporidiosis illnesses (14.65 AIDS + 1.98
non-AIDS). These mortality factors are constants in the model (in other words, no uncertainty is
attributed to these parameters).
The mortality rate from the Milwaukee outbreak may not reflect the overall mortality rates from
low-level endemic exposure. The estimated levels of Cryptosporidium in the finished water
supplies during the Milwaukee outbreak were much higher than the levels expected in systems
complying with the surface water treatment rules. It is not known, however, whether the higher
level of Cryptosporidium in the water supply could have resulted in a higher mortality rate than
that expected from much lower endemic exposures, that is, if mortality rates increased more than
proportionately at higher dose levels.
Six-Year Review 3
Technical Support Document for LT2
4-27
December 2016

-------
No data were available at the time of the LT2 to support that hypothesis; data are available to
indicate only a higher probability of infection resulting from higher ingested doses, as mentioned
in the preceding section on morbidity. In an outbreak in Las Vegas, similar mortality rates were
observed in AIDS patients (52.6 percent of deaths were in AIDS patients in Las Vegas compared
with 68 percent among AIDS patients in Milwaukee). These similar rates were observed despite
the hypothesis that the drinking water had been contaminated over an extended period of time
with intermittent low levels of oocysts, unlike Milwaukee's massive contamination (Rose, 1997).
A study by Hunter et al. (2001) suggests that the level of endemic diarrhea from all sources was
underestimated in the Milwaukee incident, leading to an overestimation of the number of
diarrheal illnesses due to cryptosporidiosis. A lower estimate of illness would consequently raise
the mortality rate per case of illness by holding deaths constant as illnesses decreased. However,
there was no consensus on whether to accept the Hunter et al. conclusions at the time of the LT2
analysis, and therefore the model used the Hoxie et al. 1997 illness estimates (cited previously)
for the Milwaukee outbreak.
As noted in Section 4.1, EPA conducted a literature search to try to identify studies published
after the LT2 providing additional information on mortality from cryptosporidiosis and the
mortality factors used in the dose-response model. EPA found only one paper on
Cryptosporidium mortality, but this was an animal study that did not provide any useful
additional information relevant to human mortality from cryptosporidiosis.
EPA found no new health effects information that would suggest any need to consider a change
from the MCLG of zero for Cryptosporidium or for a more stringent inactivation target.
4.2 Giardia
Giardia spp. are single-cell protozoan parasites that live in the intestines of infected humans and
animals and can be introduced by fecal contamination into source waters that are used for
drinking water. Giardia survives in the environment as cysts, similar to Cryptosporidium. As
indicated at the beginning of this chapter, EPA established an MCLG of zero for Giardia under
the 1989 SWTR, as well as filtration and disinfection TT requirements for filtered and unfiltered
systems to achieve a minimum of 3-log removal and/or inactivation. Additional protections
against Giardia also resulted from the control measures established under the subsequent
revisions to the Surface Water Treatment Rules, including the LT2, that were focused on
Cryptosporidium.
Because Giardia can infect some animals, those animals can serve as vectors for Giardia
transmission. In this way, Giardia may contaminate finished drinking water in UCFWRs if those
reservoirs allow access to animals.
CDC (2015b) states that the acute symptoms of Giardia infections, known as giardiasis, include
diarrhea, flatulence, greasy stools, stomach and abdominal cramps, nausea, vomiting and
dehydration. Symptoms may last for one to two weeks, but can persist longer, and other health
endpoints can occur in some cases, especially among children. CDC (2015b) also notes that
some individuals infected with Giardia are asymptomatic.
Six-Year Review 3
Technical Support Document for LT2
4-28
December 2016

-------
CDC (2015b) states that giardiasis is still the most common intestinal parasitic disease affecting
humans in the United States. While the usual route of exposure leading to infection is oral, not
all Giardia infections result from drinking water (e.g., some are passed hand to mouth or through
food).
Giardiasis is a nationally notifiable gastrointestinal illness for which CDC collects data and
reports periodically on both endemic and outbreak cases. The first recognized waterborne disease
outbreak (WBDO) of giardiasis was documented in 1965. Subsequent information (Craun, 1988;
Craun et al., 2010) indicated that the most prevalent deficiencies in drinking water system
associated with outbreaks (>70 percent) were related to undisinfected ground water, unfiltered
systems with inadequate disinfection or systems with filtration failures.
In the most recent drinking water-associated Giardia outbreak data available (Exhibit 4.7), CDC
reported three outbreaks of giardiasis (out of a total of 33 outbreaks) from 2009 to 2010 (CDC,
2013). From 2007 to 2008, CDC reported three drinking water-associated outbreaks caused by
Giardia (out of 36 total outbreaks); one of the three outbreaks had multiple etiologic agent types
identified (CDC, 2011). CDC reported no drinking water-associated outbreaks of giardiasis
between 2005 and 2006 (out of 20 total outbreaks) (CDC, 2008).
Exhibit 4.7 Giardia Outbreaks Associated with Drinking Water, by Year:
Waterborne Disease and Outbreak Surveillance System, United States 2005-2010
(CDC, 2011; 2013)
State
Year
Etiology
Cases
Water System
Water
Source
Minnesota
2010
Giardia
6
Transient noncommunity
Well
Utah
2009
Giardia
8
Community
Well, surface
water
Illinois
2008
Giardia: Cryptosporidium;
Shigella sonnei
41
Community
Lake
California
2007
Giardia
46
Noncommunity
Spring
New
Hampshire
2007
Giardia
35
Community
Well
Daly et al. (2010) conducted a cohort study to identify the risk factors for giardiasis in the 2007
outbreak, included in Exhibit 4.7, which occurred in a small community system in New
Hampshire. Their analysis confirmed that consumption of tap water was significantly associated
with illness. The authors further determined that the likely source was a well that had been
brought online without regulatory approval and was located closer to a surface water source than
regulations permit, specifically, 50 feet. Removal of that well from the system was followed by
cessation of the outbreak.
Six-Year Review 3
Technical Support Document for LT2
4-29
December 2016

-------
EPA found no new health effects information that would suggest any need to consider a change
from the MCLG of zero for Giardia or for a more stringent inactivation target.
4.3 Viruses
Viruses are infectious agents that replicate only inside the living cells of other organisms.
Viruses are very small entities 18 to 120 nm in size, compared with bacteria, which are generally
over 1,000 nm, and protozoa, such as Giardia and Cryptosporidium, that are generally on the
order of 10,000 nm.
The viruses of most concern from drinking water sources are those considered to be "waterborne
enteric viruses." Such viruses may reach hosts through consumption or contact with water, infect
and replicate within the gastrointestinal tract of hosts, are then shed in extremely high numbers in
the feces of infected individuals, and can then reenter the aquatic environment via sewage,
leaking septic systems and other related routes. The American Water Works Association
(AWW A) (2006) indicates that there are more than 120 different enteric viruses known to infect
humans.
Fong and Lipp (2005) identified the common groups of waterborne enteric viruses of concern as
belonging to the following families: Picornaviridae (polioviruses, enteroviruses, Coxsackie
viruses, hepatitis A virus and echoviruses); Adenoviridae (adenoviruses); Caliciviridae
(noroviruses, caliciviruses, astroviruses14 and small round-structured viruses); and Reoviridae
(reoviruses and rotaviruses).
Bosch (1998) also identified human enteric viruses that can be transmitted by water according to
genus as follows: Enterovirus (poliovirus, Coxsackie virus, echovirus); Hepatovirus (hepatitis A
virus); Reovirus; Rotavirus; Mastadenovirus (human adenovirus); Calicivirus (Norwalk virus,
small round-structured virus, hepatitis E); Astrovirus, Parvovirus; Coronavirus; and Torovirus.
When EPA promulgated the SWTR in 1989, it established an MCLG of zero for all viruses and
required that all surface water systems provide 4-log (99.99 percent) removal/inactivation of
viruses. In 2006, EPA extended the requirements to provide 4-log inactivation of viruses to
systems using ground water if monitoring of those systems demonstrated fecal contamination in
the source waters under the Ground Water Rule.
Enteric virus infections in humans are associated primarily with diarrhea and self-limiting
gastroenteritis. However, they may also cause respiratory infections, conjunctivitis, hepatitis and
diseases that have high mortality rates, such as aseptic meningitis, encephalitis and paralysis. In
addition, some enteric viruses have been linked to chronic diseases such as myocarditis and
insulin-dependent diabetes (Fong and Lipp, 2005). Some of these outcomes may be more severe
in immunocompromised individuals. Although most enteric viruses cause mild or asymptomatic
14 The genomic and sub genomic organization of astrovirus and its polyprotein processing led to the proposal of the
new family Astroviridae, separated from the families Picornaviridae and Caliciviridae, within the positive-sense
single-stranded RNA (ssRNA) viruses. In 1995, the International Committee for the Taxonomy of Viruses (ICTV)
definitively established the Astroviridae family in their sixth report. See Bosch et al. (2014).
Six-Year Review 3
Technical Support Document for LT2
4-30
December 2016

-------
infections, they can cause a wide range of serious and life-threatening illnesses, especially in
children (Nwachuku and Gerba, 2006). It is also important to note that while some enteric
viruses have a relatively low rate of infectivity, for others it is quite high. Bosch (1998) indicates
that asymptomatic infections are common for enteric viruses and that the development of clinical
symptoms depends on factors such as age of the infected individual, immunological status, and
strain and virulence of the pathogen.
Unlike Cryptosporidium and Giardia, most enteric viruses are host-specific, although there are
some known exceptions to this where viruses can infect both avian species and humans.
Available reviews on waterborne enteric viruses (Bosch, 1998; Fong and Lipp, 2005; AWWA,
2006) provided no information indicating that fecal matter from other animals can serve as a
vector for the waterborne enteric viruses that affect humans. EPA is unaware of information that
suggests that birds or other mammals are vectors for enteric viruses entering UCFWRs as is the
case for Cryptosporidium, Giardia or certain bacterial pathogens. EPA found no new health
effects information that would suggest any need to consider a change from the MCLG of zero for
viruses or for a more stringent inactivation target.
4.4 Other Pathogens
Pathogenic microorganisms in addition to Cryptosporidium, Giardia and viruses that may be of
concern in UCFWRs, which is also addressed by the LT2 are discussed in the following sections.
4.4.1 Fungi
4.4.1.1 Microsporidia
Microsporidia refers to a very large group of spore-forming parasites once considered protozoans
or protists but are now classified as fungi. Microsporidia includes over a thousand species.
Microsporidia are found in insects, fish and mammals, including humans. There are about a
dozen different types of Microsporidia known to infect humans. The common human pathogenic
species is E. bieneusi. Microsporidia are considered to be opportunistic pathogens and cause a
variety of adverse health outcomes including diarrhea, hepatitis, peritonitis, conjunctivitis,
sinusitis, myositis, encephalitis, renal failure, keratoconjunctivitis and blindness. EPA found no
information on waterborne outbreaks of Microsporidia in the United States, but Cotte et al.
(1999) reported an incident involving a waterborne outbreak in France in 1995 affecting 200
people. There was no evidence of fecal contamination of the water and the explanation for the
outbreak was not determined. AWWA (2006) notes that pilot-scale conventional treatment with
alum coagulation, flocculation, sedimentation and filtration has been shown to be ineffective,
removing only 1- to 1.5-log of seeded spores. Disinfection with chlorine or UV irradiation may
be accomplished, but with variable results, depending upon the species. Given the ubiquity of
Microsporidia throughout the environment and the variety of species serving as hosts, there are
likely many vectors for these organisms by which they could reach UCFWRs.
Six-Year Review 3
Technical Support Document for LT2
4-31
December 2016

-------
4.4.2 Protozoa
4.4.2.1 Toxoplasma gondii
Toxoplasma gondii is an obligate intracellular parasitic protozoan that is responsible for the
disease toxoplasmosis. T. gondii is able to infect all warm-blooded animals, and is one of the
most common human parasites worldwide. T. gondii is usually transmitted by ingestion of food
or water that has been contaminated with oocyst containing cat feces. Cats are the only definitive
host for this organism, that is, it can only reproduce sexually within the intestines of cats. Feral
cats can have higher infection rates than domestic cats, depending on geographic locations,
because they prey on T. gondii infected birds and small mammals (AWW A, 2006). Although
toxoplasmosis is typically asymptomatic in healthy adults, it is a lifelong infection that persists in
the central nervous system of the host. Severe morbidity and even death, in healthy and
immunocompromised individuals have also been reported. Moreover, T. gondii can multiply in
and be transmitted across the placenta, where it can cause neurological effects, loss of vision,
hearing impairments and death in children exposed in utero. T. gondii appears to be resistant to
disinfectants such as chlorine. Infection from this parasite have been suggested to have a role in
schizophrenia and other neurological disorders in animals and humans (Hinze-Selch, 2007; Vyas
et al., 2007).
Epidemiological studies on drinking water related toxoplasmosis are limited, since, unlike many
enteric pathogens, does not cause any overt acute GI illnesses such as diarrhea and thus difficult
to identify. Any reports of water-associated toxoplasmosis outbreaks are likely to be
underestimated. A recent study by Kreuger et al. (2014) showed individuals without at-home
water treatment devices served by public or private water companies and individuals with or
without at-home water treatment devices served by wells had significantly higher odds of T.
gondii seropositivity compared to individuals with at-home water treatment devices served by
public or private water companies.
T. gondii is also resistant to disinfectants such as chlorine, however, the oocysts are relatively
susceptible to UV treatment (Ware et al., 2010). Because of its unique host relationship with
cats, T. gondii could contaminate UCFWRs if cats, particularly feral cats, are able to enter the
UCFWR premises. Such was the case for the 1995 waterborne toxoplasmosis outbreak in
Canada, were contamination by oocyst containing feces from domestic cats and/or cougars living
around the drinking water reservoirs (Aramini et al., 1999; Bowie et al., 1997).
4.4.3 Bacteria
4.4.3.1 Escherichia coli
E. coli is a species of bacteria that comprises a large number of strains of both pathogenic and
nonpathogenic organisms. E. coli is a major natural constituent of the intestinal flora of
mammals, including humans. It is the predominant member of the fecal coliform group of
bacteria. While the majority of E. coli strains are harmless, there are several strains that are
pathogenic and cause severe gastrointestinal illness. One of the pathogenic strains, known as
enterohemorrhagic E. coli (EHEC) has been recognized in recent years as a particular concern
with respect to drinking water exposure. Infections with EHEC can cause intestinal
Six-Year Review 3
Technical Support Document for LT2
4-32
December 2016

-------
hemorrhaging and can progress to hemolytic uremic syndrome, resulting in kidney failure and
death. Of particular concern among the EHEC strains is the serotype 0157:H7, which has been
implicated in over half the EHEC outbreaks (predominately from food sources), resulting in over
70,000 illnesses and several dozen deaths per year. Nevertheless, there have been several
significant waterborne outbreaks involving 0157:H7, resulting in large numbers of illnesses and
several deaths. In the years 1991 to 2002, E. coli 0157:H7 was associated with approximately 5
percent of WBDOs (Craun et al., 2006).
E. coli can survive in the aquatic environment. As noted above, E. coli is normally present in the
intestines of mammals, and therefore fecal contamination of source waters (and water stored in
UCFWRs) can result in E. coli entering drinking water. E. coli in general (including EHEC) can
be inactivated with standard drinking water disinfection practices.
E. coli has an MCLG of zero, which was initially established under the 1989 TCR. Under the
2013 Revised TCR, presence of E. coli is an MCL violation requiring actions to be taken by the
system. Similarly, the detection of E. coli in source water monitoring of undisinfected ground
water systems triggers a requirement for further actions under the 2006 Ground Water Rule. It
should be noted that the 2-/3-/4-log inactivation requirements under the LT2 for
Cryptosporidium, Giardia and viruses leaving UCFWRs also should provide adequate control
for coli as well.
4.4.3.2 Salmonella enterica
Salmonella spp. comprises a large number of bacteria found in soil, water, plants and the normal
intestinal flora of animals, including humans. There are several pathogenic species of
Salmonella, including S. typhi, S. paratyphi and S. typhimurium. S. typhi and S. paratyphi
colonize only in humans, and therefore infection with those species indicates a vector involving
contamination with human feces. Other strains, including S. typhimurium, can be found in a
variety of domestic and wild animals, including birds and some other nonmammalian species.
Salmonellosis typically involves self-limiting gastrointestinal symptoms (diarrhea, fever,
abdominal pain) but can progress to more serious effects, including meningitis, endocarditis and
pneumonia.
Most Salmonella outbreaks are foodborne, but there are documented large waterborne outbreaks
involving S. typhimurium, notably the 1993 outbreak in Gideon, Missouri, where over 600
people became ill and seven people died. The cause of this outbreak was a defective storage tank,
which allowed birds to enter the tank. The system was not disinfected at the time of the outbreak.
Salmonella was also identified to have caused 19 percent (20/105) of the non-legionellosis
drinking water outbreaks reported in the US from 1971 to 2006 (Craun et al., 2010).
Salmonella can be inactivated with standard drinking water disinfection practices. As noted,
some pathogenic species of Salmonella, notably S. typhimurium, can be transmitted by a variety
of animals, including birds, which may serve as vectors for these organisms to reach drinking
water sources and UCFWRs. It should be noted that the 2-/3-/4-log inactivation requirements
under the LT2 for Cryptosporidium, Giardia and viruses leaving UCFWRs would likely provide
adequate control for Salmonella as well.
Six-Year Review 3
Technical Support Document for LT2
4-33
December 2016

-------
4.4.3.3 Klebsiella
Klebsiella are coliform bacteria. Most strains are harmless to humans, but there are some human
pathogenic forms. Although Klebsiella can be found in the intestinal tract of some animals,
including humans, it is not ubiquitous in intestinal flora in the same way as E. coli. Of particular
note among the human pathogenic forms is Klebsiella pneumonia, which can cause infections in
the respiratory system, nose, throat and genitourinary tract.
Most outbreaks of Klebsiella have involved water in medical facilities, and no community
waterborne outbreaks have been reported (AWW A, 2006). However, AWWA (2006) also notes
that most of the water systems having distribution system coliform occurrences indicate that the
predominant organism was a form of Klebsiella. Although Klebsiella can generally be controlled
by normal disinfection practices, this organism can encapsulate, which can provide some
resistance to disinfection, which may be of concern in UCFWRs.
4.4.3.4	Pseudomonas
Pseudomonas is a genus of bacteria that includes a large number of species. The most significant
pathogenic form from a drinking water perspective is Pseudomonas aeruginosa. It is an
opportunistic pathogen generally associated with hospitalized patients.
Pseudomonas is rarely found in the human intestinal tract. However, it can survive in surface
water, ground water and bottled water. P. aeruginosa has been associated with skin and inner ear
infections related to its presence in water in hot tubs and other communal water facilities. They
are of potential concern in UCFWRs.
4.4.3.5	Staphylococcus
Staphylococcus, and particularly the species S. aureus, S. epidermis and S. saprophyticus, are
opportunistic human pathogens, primarily associated with infections of the skin, although
ingested organisms can cause gastrointestinal infections. S. aureus can also cause meningitis, as
well as vomiting and diarrhea caused by an endotoxin produced by the organism when it grows
in food (AWWA, 2006). A recently recognized pathogen of particular concern is the highly
antibiotic-resistant form called MRSA: methicillin-resistant S. aureus, which has become
endemic in health care facilities.
Although in humans staphylococci are predominately found on skin, they can also be present in
the gastrointestinal tract and therefore may be found in human feces. They are able to survive for
weeks in water provided nutrients are present. AWWA (2006) notes that there are no known
waterborne outbreaks involving staphylococci. Because of the emerging concerns with respect to
MRSA, more attention is being given to the potential for human exposure from tap water, but
currently no specific data appear to be available (Reynolds, 2013). Based on the host specificity
of human pathogenic staphylococci, vectors for their entry into source waters for drinking water
or UCFWRs would likely require direct contact with human skin or feces. It should be noted that
the 2-/3-/4-log inactivation requirements under the LT2 for Cryptosporidium, Giardia and
viruses leaving UCFWRs would likely provide adequate control for Staphylococcus as well.
Six-Year Review 3
Technical Support Document for LT2
4-34
December 2016

-------
4.4.3.6	Campylobacter
Campylobacter is a group of bacteria including Campylobacter jejuni, Campylobacter coli,
Campylobacter lari and Campylobacter upsaliensis. They are typically found in the intestinal
tract of certain mammals. They are the major cause of foodborne diarrhoeal illness in humans
and are the most common bacteria that cause gastroenteritis worldwide (WHO, 2011) Typically
Campylobacter is transmitted from contaminated foods or water. (U.S. National Library of
Medicine, 2015)
Campylobacter strains that infect humans have been found in migratory birds including those
from the US mid-Atlantic (Broman et al., 2004; Keller and Shriver, 2014).
4.4.3.7	Cyanobacteria
Cyanobacteria commonly referred to as "blue-green algae", survive in marine and freshwater
environments. Under certain conditions, cyanobacteria can proliferate causing blooms.
Cyanobacterial blooms tend to occur in warmer water, still or slowly moving water, and nutrient
enriched waters, specifically with nitrogen and phosphorus. Since UCFWRs generally have short
hydraulic retention times and low nutrient concentrations, cyanobacteria growth is less likely to
occur in UCFWRs than in source water reservoirs. However, some concern has been expressed
about using phosphate-based corrosion inhibitors as a potential nutrient source for cyanobacterial
growth.
Cyanobacteria can produce toxins that can cause a variety of adverse health impacts such as liver
and kidney toxicity, gastrointestinal and central nervous system effects if ingested. Currently
there are no federal regulations for cyanobacteria or their toxins in drinking water, however
cyanotoxins are listed on the draft Contaminant Candidate List (CCL) 4 including system
anatoxin-a, microcystin-LR and the cylindrospermopsin. In June 2015, EPA released Health
Advisories for microcystins and cylindrospermopsin. Health Effects Support Documents
(HESDs) for microcystins, cylindrospermopsin and anatoxin-a. It was determined at that time
that there was insufficient information on the health impacts associated with cyanotoxins, the
HAs and the HESDs. Please see http://www.epa.gov/nutrient-policy-data/health-and-ecological-
effects.
4.5 Summary
While EPA found new data and information that supports the contaminants of concern for the
LT2, EPA does not believe that these data and information, when examined in conjunction with
the data and information that support the original LT2 requirements, suggest a revision to the
LT2. New health effects information continues to support Cryptosporidium and E. coli
monitoring for source waters. New Cryptosporidium dose-response information and information
on emerging concerns with a wide range of Cryptosporidium species in particular, emphasizes
the importance of the LT2 source water monitoring requirements. Similarly, the risks from
contamination of UCFWRs by a wide range of pathogens, including Cryptosporidium, Giardia,
E. coli, viruses and others continues to support the LT2 requirements for UCFWRs.
Six-Year Review 3
Technical Support Document for LT2
4-35
December 2016

-------
5 Cryptosporidium Analytical Methods Information
Reliable quantification of Cryptosporidium oocysts in water is challenging, and the recovery
efficiency from public water system (PWS) sources varies from site to site. Diagnostic
techniques have been developed for clinical samples and applied to detection of protozoa in
environmental samples but with limited success, given the low concentrations of oocysts
compared with other particulates typically found in drinking water sources (Staggs et al., 2013).
Numerous investigators have published data demonstrating the spatial and temporal influence of
chemical, biological, hydrological and meteorological conditions on recovery (Kuhn et al., 2002;
Feng et al., 2003; Francy, 2004; Ongerth, 2013; Rosen et al., 2014). Highly variable watershed
conditions are best approached with analytical methods that include options for various
procedural components to enhance data quality and consistency.
Improved collection procedures and the use of immunomagnetic separation (IMS) are the key
advances developed to detect oocysts in water during the last two decades. Sample collection
procedural improvements include increases in sample volume, allowing samples to represent ten
to hundreds of liters of water. The Envirocheck® HV Sampling Capsule, the Filta-Max® foam
filter and the portable continuous-flow centrifuge have been validated in multiple laboratories
(USEPA, 2012a). IMS improved the isolation of oocysts compared to previous techniques that
left interfering debris and biota. Not only was the recovery efficiency of oocysts significantly
improved; but visual interference was reduced, resulting in fewer false negatives and less eye
fatigue for the microscopists. IMS continues to be the isolation technique of choice for preparing
environmental samples for different targets of interest, such as fungi and bacteria (Yakub and
Stadterman-Knauer, 2004).
The Agreement in Principle prepared by the Federal Advisory Committee (FAC) in 2000 to
support development of the Stage 2 Microbial and Disinfection Byproducts Regulations
encouraged EPA to continue investigating advances that might allow for better detection of
pathogens. Before the implementation of the Long Term 2 Enhanced Surface Water Treatment
Rule (LT2), EPA developed Methods 1622 and 1623 (USEPA, 2005b; 2005c) to achieve higher
recovery rates and lower inter- and intra- laboratory variability compared to previous methods.
Specific improvements included incorporation of the more effective filters, IMS, and the addition
of 4,6-diamidino-2-phenylindole stain for microscopic analysis. The performance of Methods
1622 and 1623 was tested through single-laboratory studies and validated through multiple-
laboratory studies (USEPA, 2001a).
EPA continued to evaluate new information available on Cryptosporidium analytical methods
after publishing the LT2. Standard operating procedures were requested from laboratories having
proficiency test results within the top third of laboratories approved for the LT2 analyses.
Twenty-eight different method steps used by these laboratories were compared, and a technique
that rinses away extraneous debris was common to the majority of these skilled laboratories.
EPA also investigated procedures used in the research community and identified two additional
potential improvements: 1) adding a dispersant, sodium hexametaphosphate (NaHMP), to better
separate Cryptosporidium oocysts from extraneous particles (Rhodes et al., 2012); and 2)
applying a heat dissociation step (Ware et al., 2003) to more effectively release the organisms
from the beads, which may be useful when handling complex water matrices (Shaw et al., 2008).
Building on Method 1623, EPA added these advances and updated quality control (QC) criteria
Six-Year Review 3
Technical Support Document for LT2
5-1
December 2016

-------
based on laboratory performance demonstrated during the LT2 Round 1 monitoring. The
resulting procedural alternative was published as Method 1623.1 (USEPA, 2012a).
In 2011 EPA hosted a public meeting (USEPA, 201 lb)15 to present, among other LT2-related
information, the improvements included in Method 1623.1. Additional discussion at the meeting
addressed the advantages, limitations and future considerations of existing molecular methods
for detection of and genotyping Cryptosporidium.
Section 5.1 describes the performance of Method 1623.1. Section 5.2 characterizes additional
detection techniques for Cryptosporidium. Section 5.3 describes research on oocyst recovery
using different isolates for spiking water samples. Conclusions on EPA's review of
Cryptosporidium detection methods for the LT2 are presented in Section 5.4.
5.1 Performance of Method 1623.1
This section describes the performance of Method 1623.1, evaluated by analyzing spiked
Cryptosporidium recovery data generated with 30 different PWS sources. For the LT2 Round 1
monitoring, approximately 80 percent of the Cryptosporidium analyses were performed using the
Envirocheck HV® filter capsule, and the majority of laboratories used this option. For this
reason, the Envirocheck HV® filter capsule was used for both the side-by-side comparisons of
Methods 1623 and 1623.1 and validation testing of Method 1623.1 (USEPA, 2012b). Because
the data were generated using only one of the three method-approved equipment options for
sample filtration, the recoveries for EPA Method 1623.1 included in this chapter do not
necessarily represent recoveries that might be observed if other approved filter methods were to
be used. Additional data on the impact of different filtration equipment and recoveries for other
EPA Method 1623.1 options are needed to determine their impact on recovery rates.
EPA compiled data from side-by-side comparisons of the two methods into three exhibits
showing Cryptosporidium recovery values from different matrices. Exhibit 5.1 through Exhibit
5.3 show the difference in recoveries based on both single- and four-laboratory side-by-side data.
The box in each exhibit represents the interquartile range. The line in the middle of each box is
the median recovery. The lines extending from each box show the adjacent values, and the dots
(as appear in Exhibit 5.3), are observations beyond the adjacent values. The upper adjacent value
is the largest value that is less than or equal to the third quartile plus 1.5 times the interquartile
range. The lower adjacent value is the point greater than or equal to the first quartile minus 1.5
times the interquartile range. Exhibit 5.4 and Exhibit 5.5 show the distribution of recovery values
generated in 14 laboratories for both reagent water and source waters during validation of
Method 1623.1. Raw data for the side-by-side comparisons are found in Appendix A, Exhibits
A.l - A.3.
5.1.1 Single-Laboratory Side-by-Side Comparison of Method 1623 with Method 1623.1
Recoveries of spiked oocysts from reagent water, were comparable for both methods including
the mean, median and standard deviation (Appendix A, Exhibit A.l). This result is not surprising
15 Slide presentations from this meeting are available at www.epa.gov/dwsixyearreview/review-lt2-rule
Six-Year Review 3
Technical Support Document for LT2
5-2
December 2016

-------
considering that the primary improvements in Method 1623.1 were intended to remove
interferences found in some source water matrices and not expected to be present in reagent
water. Importantly, this result shows that method changes in Method 1623.1 (e.g., addition of
NaHMP and the additional wash step) had no negative effect on oocyst recoveries from the
reagent water that laboratories typically use for QC.
Recoveries of spiked oocysts from Ohio River samples and three artificial matrices demonstrated
a substantial improvement for an artificial clay matrix (Exhibit 5.1; Appendix A, Exhibit A.2).
The artificial matrices were prepared using reagent water mixed with 0.1 g diatomaceous earth,
0.2 g Tennessee River sediment (TNRS) or 0.2 g clean clay. Samples associated with each of the
matrices had an increase in mean recovery using Method 1623.1.
Exhibit 5.1 Observed Recovery at a Single Laboratory, Using One Source Water
and Three Artificial Matrices
0
§ 60-
O
0
cn
0
O)
ra
1	40-
o
0
Q_
"O
0
t
0 20-
oo
_Q
O
o-
Clay
Tennessee River Sediment
Ohio River
D.E.
I I EPA Method 1623	EPA Method 1623.1
Recoveries of spiked oocysts in source water samples from nine utilities located around the
United States could be grouped into two categories based on the difference of the mean between
Methods 1623 and 1623.1 (Exhibit 5.2; Appendix A, Exhibit A.3):
1.	Substantial difference in mean recovery between the methods using source waters from
North Carolina, Mississippi and Massachusetts, and
2.	Moderate difference in mean recovery between the methods using source water from the
Ohio River, two sites in Texas (Texas-1 and Texas-2), Colorado, Missouri and Montana.
The observed recovery rates for Method 1623 for three matrices (source waters from North
Carolina, Mississippi and Massachusetts) were well below the recoveries for the other six
matrices (Ohio, Texas-1, Colorado, Montana, Texas-2 and Missouri). Using Method 1623.1, the
observed recovery rates for these three matrices were similar to the recoveries for the other six
matrices. Turbidity alone does not appear to have a clear relationship to either method (Appendix
A, Exhibit A.3).
Six-Year Review 3
Technical Support Document for LT2
5-3
December 2016

-------
Exhibit 5.2 Observed Recovery at a Single Laboratory, Using Nine Source Waters
100-
80-
60-
40-
I
I
T
_Q
o
20-
0-
MA
NC MS TX-2 OH TX-1 CO MO MT
I	I EPA Method 1623	EPA Method 1623.1
5.1.2 Four-Laboratory Side-by-Side Comparison of Method 1623 with Method 1623.1
Four laboratories measured recovery for Methods 1623 and 1623.1 using samples from source
waters in Michigan, Montana and Ohio (Exhibit 5.3; Appendix A; Exhibit A.4). The
improvement in oocyst recovery between Method 1623 and 1623.1 ranged from approximately 5
percent to approximately 23 percent for the three matrices and appears to depend on the matrix.
Exhibit 5.3 Observed Recovery at Four Laboratories, Using Three Source Waters
100-
60-
40-
20-
Michigan
Montana
Ohio
I I EPA Method 1623	EPA Method 1623.1
Six-Year Review 3
Technical Support Document for LT2
5-4
December 2016

-------
5.1.3 Fourteen-Laboratory Method 1623.1 Validation Data
EPA conducted a multi-lab oratory validation study of Method 1623.1 (USEPA, 2012b). Fourteen
laboratories were asked to select a PWS source from which oocyst recoveries were
representative of the matrices tested in their laboratory, with the exception of one laboratory that
was asked to select a ground water under the direct influence (GWUDI) of surface water. The
collection sites included six lakes/reservoirs, seven rivers and the GWUDI source. Oocyst
suspensions were provided to each laboratory and they were instructed to spike five replicates of
source water and five replicates of reagent water. The recoveries from the matrices used in the
side-by-side studies are consistent with recoveries from the matrices used in the validation study
(Exhibit 5.4 and Exhibit 5.5). The reagent-water recovery data was used to develop QC criteria
for laboratory performance. The distribution of recovery was estimated by using random effects
ANOVA. The recovery limits were estimated as the 5th percentile of the predictive distribution.
A lower limit of 33 percent was established as acceptable recovery of Cryptosporidium detected
in reagent water with Method 1623.1. The current lower limit for acceptable recovery with
Method 1623 is 22 percent, though a revised limit of 33 percent has been proposed and is under
consideration. (Note that the original lower limit for Method 1623 recovery was updated during
the first round of the LT2 based on an evaluation of proficiency test data available at the time.
The revision under consideration was based on a larger, more current set of proficiency testing
(PT) data.)
Six-Year Review 3
Technical Support Document for LT2
5-5
December 2016

-------
Exhibit 5.4 Method 1623.1 Validation Data for 14 Laboratories,
Reagent Water; N = 56
Observed Percentage Recovery
Exhibit 5.5 Method 1623.1 Validation Data for 14 Laboratories,
Source Water; N = 53
0.030-
Observed Percentage Recovery
5.1.4 Summary
Although the number of observations in source water is relatively small, it appears that
the degree of recovery for either method is highly dependent upon the source matrix, since some
matrices show a substantial difference in mean recovery and other matrices show only a
moderate difference in mean recovery. The data were generated in a subset of Cryptosporidium
Six-Year Review 3
Technical Support Document for LT2
5-6
December 2016

-------
laboratories and may not be indicative of results in the larger lab community. The difference in
recovery was more pronounced for those samples with a low initial recovery with Method 1623.
The degree to which the broad range of natural conditions in drinking water sources is
represented by the water conditions in the side-by-side and validation data is not clear. PWSs
now have a choice of using either the established Methods 1622 and 1623, or the alternative test
procedure, Method 1623.1. Since the degree of improvement achieved by use of Method 1623.1
appears to vary from matrix to matrix, PWSs with good recoveries using Method 1623 may not
see much improvement with a change to 1623.1.
5.2 Other Cryptosporidium Detection Techniques and Suggested Improvements
Depending on future analytical needs, whether for species resolution, quantitative accuracy or
environmental trends; molecular-based methods may have a significant role in future analyses
for Cryptosporidium. Identification of Cryptosporidium genotypes could be useful for microbial
risk assessment models and watershed management decisions (Ware et al., 2013). For example, a
large multiyear study that analyzed nearly 690 raw surface water samples for Cryptosporidium
numbers and genotypes found a link between watershed use and the risk of human infection
(Wilkes etal., 2013).
Polymerase chain reaction (PCR) relies upon in vitro enzyme-mediated amplification of
Cryptosporidium-specific nucleic acids in order to facilitate identification in water samples. This
technique should offer significant endpoint sensitivity as well as the possibility of distinguishing
subtle differences among discrete strains of parasites. Real-time PCR, or qPCR or (qPCR),
allows for real-time analysis of the quantity of deoxyribonucleic acid (DNA) product in each
cycle of the conventional PCR reaction. Measurement of the quantity of DNA fragment copies
created during the PCR reaction relies on the fluorescence emitted by a fluorochrome molecule
bound to a DNA probe used to detect the amplified DNA. The assay detects and differentiates C.
hominis and C. parvum in both clinical and environmental samples. In another study, Staggs et
al. also reported that qPCR detection sensitivities depend highly on the target genes, with pPCR
assays targeting multi-copy gene regions providing better sensitivities (Staggs et al., 2013).
Sunnotel et al. (2006) compared the sensitivities of real-time PCR and nested PCR usingl8S
ribosomal ribonucleic acid (rRNA) as a target. The authors determined that qPCR was routinely
capable of detecting three Cryptosporidium oocysts. Yang et al. (2013) tested a qPCR assay
targeting a unique Cryptosporidium-specific protein-coding gene and found it to be more
sensitive and specific than an 18S nested PCR assay. In addition, the use of qPCR reduces the
risk of cross-contamination between samples, which is a greater concern for nested PCR assays.
PCR analysis of microscope slides as an extension of Method 1623 has been developed to
identify the Cryptosporidium species that reportedly cause the most infection in humans as well
as animal-associated species. Di Giovanni et al. (2010) developed a streamlined genotyping
assay for use after quantitation by microscopy that could be readily used by water quality
laboratories to genotype Cryptosporidium parvum, Cryptosporidium hominis and
Cryptosporidium meleagridis, which appear to readily infect immunocompetent humans. They
incorporated a single-round multiplex heat shock protein 70 (hsp 70) and 18S rRNA protocol for
use with conventional and real-time PCR instruments combined with high resolution melt
analysis capabilities. Ware et al. (2013) used a Poisson distribution analysis to estimate the
relative target densities and limits of detection, they found that 18 oocysts (Cryptosporidium
Six-Year Review 3
Technical Support Document for LT2
5-7
December 2016

-------
parvum, Iowa strain) are required for a 95 percent probability of detecting a single oocyst on a
slide. Conversely, Di Giovanni et al. (2010) observed an overall 83 percent positive detection
rate for slides seeded with a single oocyst.
Since PCR is capable of detecting the genetic material of both live and dead microorganisms, a
number of studies have been conducted to distinguish viable from nonviable oocysts. Brescia et
al. (2009) developed a Cryptosporidium propidium monoazide (PMA)-PCR assay that includes
PMA treatment prior to PCR analysis in order to prevent the amplification of DNA from dead
oocysts. The results from their initial investigation demonstrated that PMA penetrates only dead
oocysts and blocks amplification of their DNA. The authors suggest that CryptoPMA-PCR assay
is an attractive approach to specifically detect and genotype viable Cryptosporidium oocysts in
the water. Another procedure, reverse transcription-PCR, relies on ribonucleic acid, which can
provide information on viability (Skotarczak, 2010).
Cell culture is also used to assess the viability of Cryptosporidium oocysts and has been used as
an alternative to infectivity studies. In the cell culture assay, the infection process and asexual
reproduction of Cryptosporidium occur. Several cell lines have been used; e.g., the lines from
human enterocytes (HCT8) and colon cells (CaCo2) (Rochelle et al., 1997). Recently, a more
sensitive culture assay for detecting C. parvum using the human small intestinal line FHs 74 Int
has been developed (Varughese et al., 2014). Immunofluorescence (foci method) or PCR is used
to detect the presence of Cryptosporidium in the cell culture.
Feng et al. (2011) describe a method that tracks and quantifies the early phase of attachment and
invasion of C. parvum sporozoites using a fluorescent dye. The authors labeled newly excysted
sporozoites with the amine-reactive fluorescein probe carboxyfluorescein diacetate succinimidyl
esters (CFSE). They detected the initial invasion of cells by labeled parasites with fluorescent or
confocal microscopy and quantified the infection of cells by flow cytometry. Comparative
analysis of infection of cells with CFSE-labeled and unlabeled sporozoites showed that the
infectivity of C. parvum was not affected by CFSE labeling. Quantitative analysis showed that C.
parvum isolates were considerably more invasive than C. hominis isolate TU502. The differences
can be attributed to host-parasite adaptation, particularly C. parvum, which could account for this
difference.
A number of the aforementioned methods/techniques are of a research nature or are early in
development. Additional development and validation work would be needed before they could
be applied to a national monitoring effort.
A method improvement suggested by some experts is to increase the frequency of matrix spike
samples because of the difficulty to determine Cryptosporidium prevalence with highly variable
watershed conditions in different geographic areas (Ongerth, 2013). The LT2 requires a
minimum frequency of one matrix spiked sample per 20 field samples for each individual source
analyzed. PWSs may sample more frequently, based on the recovery of Cryptosporidium oocysts
in their site-specific samples, to better estimate oocyst concentration.
Six-Year Review 3
Technical Support Document for LT2
5-8
December 2016

-------
5.3	Analysis and Recoveries of Cryptosporidium Isolates
The Iowa isolate of C. parvum has been used extensively for research and QC efforts associated
with the detection of Cryptosporidium in water. Previous studies reported genetic differences
among several Iowa isolates propagated in different laboratories (Cama et al., 2006). To
investigate method efficacy to recover Iowa isolates from different laboratories, Villegas et al.
(2010) compared recoveries of four different C. parvum Iowa isolates using Method 1623. The
range of recovery for the different isolates from suspensions with TNRS (Tennessee River
Sediment, National Institute of Standards and Technology's Standard Reference Material 8406),
was 9.8 to 49 percent. Similar results were also observed in high clay content raw surface water.
It is unknown if a variation in recoveries would be observed with waterborne oocysts of different
origins.
This variation in the ability to recover oocysts from the different Iowa isolates within the genus
of Cryptosporidium may possibly bias the results from analyses of samples containing a mixture
of Cryptosporidium genotypes. Xiao and Ryan (2008) describe a problem of detecting only the
dominant genotype because of the inherent nature of exponential amplification by PCR, and the
requirement of a substantial amount of PCR product needed for detection. They further describe
the challenges of PCR inhibitors for environmental samples but still stress the importance of
genotyping to assess public health importance and source tracking for watershed management.
5.4	Conclusion
In addition to EPA Methods 1622 and 1623, which are identified for Cryptosporidium analyses
in the LT2 [40 CFR 141.704], EPA has approved EPAMethod 1623.1 as another option for the
LT2 analyses pursuant to the June 28, 2012 Federal Register Notice, "Expedited Approval of
Alternative Test Procedures for the Analysis of Contaminants Under the Safe Drinking Water
Act" (USEPA, 2012c). Method modifications, such as those incorporated into Method 1623.1,
have been used to improve Cryptosporidium recovery in some matrices. As the data discussed in
this section indicate, however, those modifications may not improve Cryptosporidium oocyst
recovery in all source waters. Statistical analyses of water quality data (Rosen et al., 2014)
suggest several different constituents, individually or in combination, may be correlated with
recovery. Without more observations from source waters with variable composition, we cannot
quantify the impact of EPA Method 1623.1 over the previous methods. In the meantime, EPA
Methods 1622, 1623 and 1623.1 continue to be options that may be used by the PWSs for the
LT2 Round 2.
Six-Year Review 3
Technical Support Document for LT2
5-9
December 2016

-------
6 Occurrence and Exposure Information
The Long Term 2 Enhanced Surface Water Treatment Rule (LT2) requires source water
Cryptosporidium and E. coli monitoring for all filtered surface water sources, including ground
water under the direct influence of surface water (GWUDI) systems. Unfiltered systems of all
sizes are required to monitor for Cryptosporidium only. The LT2 allows small filtered systems to
perform an indicator screening analysis using /•]. coli, with subsequent Cryptosporidium
monitoring required only if source water E. coli concentrations exceed annual mean trigger
levels [40 CFR 141.701], If the monitoring results do not exceed the applicable trigger, these
systems are not required to perform Cryptosporidium monitoring. Small systems are those
serving fewer than 10,000 people.
Round 1 monitoring began in October 2006 and was completed in April 2012. Specific
beginning and completion dates were established according to a schedule based on population
served and, for small systems, the E. coli monitoring results (see Exhibit 6.1).
Exhibit 6.1 System Size and Round 1 Sampling Schedule
System Size (population served)
Schedule
Sampling Start Date*
>100,000
1
October 2006
50,000 to 99,999
2
April 2007
10,000 to 49,999
3
April 2008
< 10,000 and monitoring for E. coli
4
October 2008
< 10,000 and monitoring for
Cryptosporidium
4
April 2010
* No later than listed date.
EPA classified filtered water systems into one of four treatment categories, or "bins," based on
the results of the source water monitoring [40 CFR 141.710], The bin classification determines
the degree of additional Cryptosporidium treatment, if any, the filtered water system must
provide. Unfiltered systems must provide either 2-1 og or 3-log Cryptosporidium inactivation,
depending on the measured source water Cryptosporidium concentrations. Round 2 monitoring
began in April 2015 for systems serving at least 100,000 people. Other systems serving 10,000 or
more people began their second round of monitoring later in 2015 or in 2016 depending on
system size. Small systems will begin the second round of monitoring in 2017 or 2018.
EPA developed and managed a Data Collection and Tracking System (DCTS) as a central
repository of Round 1 occurrence data. EPA pulled data from the DCTS and posted the data on
EPA's website (USEPA, 2016c).
Microbial occurrence from data sets described later in this section informed source water bin
criteria for the LT2. "Occurrence" is the term used to describe the nationwide distribution of
measured concentrations of organisms. Occurrence data on source water E. coli and
Cryptosporidium concentrations collected by large systems in the Round 1 monitoring effort
informed the indicator screening analysis approach for small systems that EPA conducted under
this review. The occurrence information is discussed later in this chapter.
Six-Year Review 3
Technical Support Document for LT2
6-1
December 2016

-------
For the LT2 development, EPA used five data sets to characterize the occurrence of
Cryptosporidium, Giardia and coliforms in source water: the ICR filtered systems data set and
the ICR unfiltered systems data set, both collected from 1997-1998, and the ICR Supplemental
Survey (ICRSS) for large water systems data set, the ICRSS for medium systems data set, and
the ICRSS for small systems date set, all collected from 1999-2000. EPA developed a statistical
model and fit these four data sets to the model to characterize from each a national distribution of
source water Cryptosporidium concentrations, reflecting both the variability from one plant to
another and the uncertainty in the estimates. EPA discusses this statistical model in Chapter 4
and also in Appendix B of the Occurrence and Exposure Assessment for the Long Term 2
Enhanced Surface Water Treatment Rule (USEPA, 2005d).
The statistical model used a database containing the ICR data for 18 monthly samples from
approximately 300 large systems (surface water and GWUDI systems serving a population of
over 100,000 people), representing approximately 500 plants. It also used the ICRSS database
containing semimonthly data collected over 12 months from 47 large systems, 40 medium
systems and 40 small systems (coliform data only). The data analysis and statistical calculations
used data from both databases on Cryptosporidium, Giardia, viruses (ICR data only) and
coliforms.
From these data, EPA developed a hierarchical model using Bayesian parameter estimation
techniques to describe the uncertainty associated with individual assays, the variability of
occurrence over space and time, and the contributions of source water type, turbidity and time of
year. This model used observed data to better characterize the national distribution of protozoa
occurrence in source water.
Following the LT2 promulgation and implementation of much of Round 1 monitoring, EPA
hosted a public meeting, on December 7, 2011, to discuss the LT2 Cryptosporidium analytical
method improvements and to provide a source water monitoring update. After giving an
overview of the LT2 requirements, EPA and other presenters discussed the following topic areas.
•	The LT2 Round 1 Cryptosporidium occurrence, particularly preliminary data analysis
comparing occurrence data from EPA's DCTS to the ICR and ICRSS data.
•	The LT2 Round 1 Cryptosporidium matrix spike recovery, particularly comparison of
Round 1 recovery data to the data projections made during the development of the LT2
Economic Analysis (EA).
EPA hosted an additional public meeting on November 15, 2012, in Washington, DC, to discuss
the LT2 review process, monitoring, occurrence, binning and microbial toolbox information.
Topics raised and discussed during the meeting included the following.
•	The LT2 Round 1 occurrence analysis and binning estimates.
o The quality of the data in DCTS and the data cleanup needed prior to analysis.
o Lack of information for small systems (i.e., those with fewer than 10,000 people
served, or "Schedule 4" systems) since they did not have to report to DCTS.
Six-Year Review 3
Technical Support Document for LT2
6-2
December 2016

-------
o Possible improvements to the reporting process for Round 2.
•	E. coli effectiveness as a small system screen.
o Analyses using the latest DCTS data on E. coli and Cryptosporidium
concentrations in large system sources.
o Analysis of available plant data for Schedule 4 systems to determine the
percentage of plants in Bin 2 and above compared with plants in Schedule 1-3
systems. See Exhibit 6.1 for a cross-reference between system size and schedule.
•	Cryptosporidium occurrence variations and matrix spike recovery.
o Implications of the detection numbers and whether they accurately represent the
true number of Cryptosporidium oocysts in the water.
o Recovery rates and variability (both in source water samples over time and within
laboratory analyses).
o Poor quality water and lower recovery rates resulting in lower bin classifications.
o Value of these results without a determination of whether the Cryptosporidium
found is infective to humans.
•	Predicted results and implications for Round 2 results.
o Round 2 monitoring and data collection will be managed by states rather than
through DCTS. This could present challenges for those who want to do a national
analysis of data.
o Whether the next generation of the Safe Drinking Water Information System
(SDWIS), called SDWIS Prime, will be ready to accept the Round 2 data.
The following sections of this chapter present a summary of the data on Cryptosporidium
collected from Round 1 of the LT2, predictions for Round 2 under a variety of scenarios (e.g.,
using Method 1623.1 instead of Method 1623), and analysis of E. coli as a trigger for
Cryptosporidium monitoring (as well as some material on co-occurrence). A final section
considers Giardia and Cryptosporidium co-occurrence based on ICRSS data and modeling
results.
6.1 Cryptosporidium
6.1.1 Summary of Round 1 Occurrence Data
At the November 2012 public meeting, EPA gave a presentation on updates to the LT2 Round 1
Cryptosporidium occurrence and binning estimates since the December 2011 meeting. The
intention was to address the following questions.
Six-Year Review 3
Technical Support Document for LT2
6-3
December 2016

-------
1.	How representative are the Round 1 monitoring data?
2.	To what extent has Cryptosporidium occurrence changed over time?
3.	What is the status of system bin classification?
The primary data source for the information presented was DCTS, the data system used by
public water systems (PWSs) serving 10,000 or more people, and some smaller systems to
upload Round 1 data. DCTS allowed users to download Round 1 monitoring data, create an
automatic Round 1 binning report, obtain a list of systems using grandfathered data, and generate
a list of systems intending to provide treatment instead of monitoring. In addition to pulling data
from DCTS, EPA worked with states to validate DCTS information. In particular, EPA was
interested in identifying a list of systems in Bin 2 or above, as well as a list of systems that
intended to provide treatment instead of conducting Round 1 monitoring.
6.1.1.1 DCTS Round 1 Monitoring Data and Binning Results
To conduct its analyses, EPA used DCTS data pull from April 2012, which contained 44,944
records representing all system sizes. As discussed in Chapter 1, the LT2 required all medium
and large systems (those serving 10,000 or more people) to submit their data to DCTS. Schedule
4 systems (those serving less than 10,000 people) were not required to submit their data to DCTS
(although some did), so Round 1 monitoring data for small systems were incomplete. EPA
included this limited Schedule 4 data in its analysis. EPA reviewed this information to remove
redundant records and/or contested records, flag data with quality concerns and remove
unnecessary data fields. EPA included additional data fields to clarify some potential data quality
issues. EPA posted the original and "cleaned-up" datasets on the EPA website at
https://www.epa.gov/dwsixyearreview/long-term-2-enhanced-surface-water-treatment-lt2-rule-
round- 1 -source-water.
Approximately 95 percent of the almost 45,000 records were from filtered water systems, with
the remaining 5 percent from unfiltered systems or from systems with unknown filtration status.
The breakdown by schedule appears in Exhibit 6.2. Over half (686 of 1,323) of the plants are in
Schedule 3 systems, and about a quarter (284 of 1,323) of the plants are in Schedule 1 systems.
Plants at filtered systems serving 10,000 or more people in Round 1 represent 80 percent (1,381
of 1,733)16 of the estimated number of plants at filtered systems serving 10,000 or more people
in the LT2 EA. Plants at filtered systems serving fewer than 10,000 people in Round 1 represent
3.4 percent (191 of 5,57s)1 of the estimated number of unfiltered system plants serving fewer
than 10,000 people in the LT2 EA, likely because small systems were not required to submit data
to DCTS and the use of the E. coli trigger excluded systems with low occurrence. EPA describes
the baseline conditions in Chapter 4 of the LT2 EA (USEPA, 2005a).
Exhibit 6.2 shows the Cryptosporidium occurrence summary statistics by monitoring schedule.
16 Based on baseline number of filtered plants in the LT2 EA (USEPA, 2005a).
Six-Year Review 3
Technical Support Document for LT2
6-4
December 2016

-------
Exhibit 6.2 Cryptosporidium Round 1 Monitoring Participation
Schedule
Systems1
Plants
Field
Samples
Matrix
Spikes
Records
1
284
403
10,634
825
11,459
2
167
219
5,679
455
6,134
3
686
759
18,641
1,523
20,164
4
186
191
4,486
346
4,832
Total
1,323
1,572
39,440
3,149
42,589
Note:
1) Includes only plants having at least six field sample results.
Exhibit 6.3 presents the Cryptosporidium monitoring summary statistics. The exhibit shows that
mean Cryptosporidium concentrations in oocysts per liter tend to be greater as system size
decreases.
Exhibit 6.3 Cryptosporidium Round 1 Summary Statistics
Schedule
Mean1
% Nondetections
1
0.00962
94.6%
(10,064 of 10,634)
2
0.0127
93.5%
(5,308 of 5,679)
3
0.0165
93.1%
(17,346 of 18,641)
42
0.0239
88.3%
(3,959 of 4,486)
All
0.0149
93.0%
(36,677 of 39,440)
Notes:
1)	Arithmetic mean using zero for nondetections.
2)	Does not include systems that met E. coli trigger level and avoided Cryptosporidium monitoring or
systems that did not report to DCTS.
Exhibit 6.4 presents Cryptosporidium summary statistics by plant for each monitoring schedule.
Across schedules, no trends are apparent in either the percentage of systems detecting no
Cryptosporidium or the percentage with means at or above 0.075 oocysts per liter (the level at
which additional treatment is required). Cryptosporidium occurrence varies significantly from
plant to plant, and from laboratory to laboratory. The probability of zero recovery of
Cryptosporidium oocysts also varies significantly by laboratory.
Exhibit 6.4 Cryptosporidium Round 1 Summary Statistics by Plant
Schedule
Number of Plants
Number of Plants
with No
Detections (%)
Number of
Plants2 with
Means at or
Above 0.075
oocysts/L (%)
1
403
240 (60%)
12 (3.0%)
2
219
108 (49%)
5 (2.3%)
Six-Year Review 3
Technical Support Document for LT2
6-5
December 2016

-------
Schedule
Number of Plants
Number of Plants
with No
Detections (%)
Number of
Plants2 with
Means at or
Above 0.075
oocysts/L (%)
3
759
388 (51%)
32 (4.2%)
4
191
84 (44%)
13 (6.8%)
All
1,572
820 (52%)
62 (3.9%)
Notes:
1)	Includes only plants having at least six field sample results.
2)	Based on plant mean, not running annual average (RAA).
Exhibit 6.5 provides Cryptosporidium summary statistics by source water type. This exhibit
shows that approximately 84 percent (1,319 of 1,572) of the plants use either a lake/reservoir or a
river/stream as their water source. Between 6 and 7 percent (103 of 1,572) of the plants use
GWUDI. Among the 62 plants with average levels of Cryptosporidium detections at or above
0.075 oocysts per liter, most (46) have flowing river or stream source waters. Among plants with
no detections, most (458 of 829) have lake or reservoir source waters.
Exhibit 6.5 Cryptosporidium Round 1 Summary Statistics by Source Water Type



Number of


Number of Plants
Plants with
Water Type
Number of Plants
with No
Detections (%)
Means at or
Above 0.075
oocysts/L (%)
Lake/Reservoir
709
458 (65%)
8 (1%)
(LR)



River/Stream (FS)
610
211 (35%)
46 (8%)
Both (LR & FS)
47
23 (49%)
3 (6%)
GWUDM-LR
33
24 (73%)
1 (3%)
GWUDM-FS
70
51 (73%)
2 (3%)
NA2
103
53 (51%)
2 (2%)
All
1,572
820 (52%)
62 (3.9%)
Notes:
1)	GWUDI = ground water under direct influence of surface water.
2)	NA = not available. Source water type was not specified.
The information in Exhibit 6.3 through Exhibit 6.5 can be compared to historic summary
occurrence statistics for the ICRSS, which consisted of 47 plants from large filtered systems
serving at least 100,000 people and 40 plants from medium filtered systems serving between
10,000 to 100,000 people. All 87 plants were sampled twice per month for 12 months using
Method 1622 or 1623. However, seven of the plants from large systems were selected as
"certainty samples" based on prior information about those locations, whereas the other 40 plants
from large systems and the 40 plants from medium systems were selected randomly. For support
of the LT2, only data from the 40 randomly selected plants from large systems were used along
with the data from the 40 randomly selected plants at medium systems. The ICRSS monitoring
results are summarized below (USEPA, 2005a; 2005b).
Six-Year Review 3
Technical Support Document for LT2
6-6
December 2016

-------
• There were 1,920 source water samples.
•	86 percent of samples were nondetections.
•	Plant mean Cryptosporidium concentration = 0.060 oocysts/L in source waters with
detections
•	12 plants (15 percent) had no detections.
•	Approximately 14 percent of plants had means of at least 0.075 oocysts/L.
The comparison of the ICRSS and the DCTS Round 1 data shows that overall occurrence of
Cryptosporidium for Round 1 monitoring was considerably lower than in the ICRSS used in the
LT2 EA for the LT2 Round 1 prediction.
•	More nondetections in Round 1 (93 percent) vs. the ICRSS (86 percent).
•	More plants with no detections in Round 1 (52 percent) vs. the ICRSS (15 percent).
•	Lower overall average concentrations in Round 1 (0.015 oocysts/L) vs. the ICRSS (0.060
oocysts/L).
•	Smaller percentage of source waters with mean concentrations of at least 0.075 oocysts/L
in Round 1 (3.9 percent) vs. the ICRSS (14 percent).
At the November 2012 public meeting, EPA also presented information on the bins in which
water systems were placed based on their source water monitoring results. The data presented
came from three sources: DCTS binning report, grandfathered and "missing" system information
(provided by EPA regions and states), and information on systems providing treatment instead of
monitoring.
The treatment bin classifications established for filtered PWSs (Chapter 3, Exhibit 3.1) are used
to determine whether additional treatment is needed. Treatment plants classified as Bin 1 are not
required to implement additional treatment; treatment plants classified as Bins 2 through 4 are
required to implement increasing levels of treatment [40 CFR 141.71 1], The LT2 does not
require filtered systems to conduct source water monitoring if the system will provide a total of
at least 5.5-log of treatment for Cryptosporidium, equivalent to meeting the treatment
requirements of Bin 4.
Exhibit 6.6 presents the binning results for filtered systems serving more than 10,000 people
based on DCTS and non-DCTS data sources. Based on a total of 1,733 filtered plants in the
monitoring baseline in the LT2 EA (USEPA, 2005a), the percent of plants in an action bin (i .e..
Bins 2 to 4) is estimated to be 7.1 percent.
Exhibit 6.6 Binning Results for Filtered Systems > 10,000 People
Data Source
Bin 2
Bin 3
Bin 4
DCTS
80
1
0
Six-Year Review 3
Technical Support Document for LT2
6-7
December 2016

-------
Data Source
Bin 2
Bin 3
Bin 4
Non-DCTS
41
1
0
Total
121
2
0
A total of 204 filtered systems submitted their intent to provide 5.5-log treatment instead of
monitoring (equivalent to Bin 4). This includes 21 systems serving 10,000 or more people and
183 systems serving fewer than 10,000 people. Fifteen unfiltered systems submitted their intent
to provide 3-log treatment instead of monitoring. This includes two systems serving more than
10,000 people and 13 systems serving fewer than 10,000 people. Fifty-one systems had unknown
filtration status. The actual Cryptosporidium concentrations are unknown for these systems.
6.1.1.2 Grandfathered Data
The LT2 allowed PWSs to use state-approved previously collected Cryptosporidium data either
in lieu of, or in addition to, results generated during the LT2 Round 1 monitoring. PWSs that
opted to use "grandfathered" data were required to produce and report data equivalent to the data
collected during Round 1 monitoring. If the grandfathered data satisfied this requirement, the
systems could use these data for bin determination.
Eight hundred twelve systems submitted at least one grandfathered sample result to DCTS. Out
of the 812 systems, 642 served more than 10,000 people, 169 served fewer than 10,000 people
and one was of unknown size. The number of samples each system requested to grandfather
ranged from one to 175. Some systems had a mixture of grandfathered data and Round 1
monitoring data. In addition to Cryptosporidium data, some systems also provided optional E.
coli and Giardia data in their submittals.
The grandfathered data submitted to DCTS were mostly in PDF format, with some in Microsoft
Excel spreadsheets for bin concentration calculations. Some data packages also contained state
approval letters for the LT2 grandfathered data and bin classification. Due to limitations on
ability to process grandfathered data in PDF file format, EPA did not use the grandfathered data
to analyze the Cryptosporidium occurrence or evaluate the E. coli trigger levels (these analyses
are described in Sections 6.1.2 and 6.2). However, EPA reviewed the available bin calculation
spreadsheets and bin classification letters in DCTS to verify the binning results provided by EPA
regions/states and to identify additional plants to be placed in action bins. Exhibit 6.7 presents
the binning results for grandfathered systems serving 10,000 or more people. The binning results
for grandfathered data were included as part of the non-DCTS data sources in Exhibit 6.6. Based
on a total of 642 grandfathered systems serving 10,000 or more people, EPA estimates that 42
systems (or 6.5 percent of systems using grandfathered data) will be in an action bin.
Six-Year Review 3
Technical Support Document for LT2
6-8
December 2016

-------
Exhibit 6.7 Binning Results for Plants in Grandfathered Systems > 10,000 People
Data Source
Bin 2
Bin 3
Bin 4
DCTS binning report1
12
0
0
EPA/states
26
1
0
Grandfathered data
3
0
0
reports



Total
41
1
0
Note:
1) Plants shown on DCTS binning report as having a mixture of grandfathered data and
Round 1 monitoring data.
6.1.2 Predictive Modeling for Round 2
To predict the results of an additional round of monitoring, including the effects of possibly
using Method 1623.1 instead of Method 1623, EPA developed a mathematical model of
Cryptosporidium occurrence that is informed by the Round 1 data and based on the model
structure described in the LT2 EA. The following sections describe the additional data cleaning
conducted, provide an overview of the modeling approach, and discuss the predictions for Round
2 under a variety of scenarios (e.g., whether labs use Method 1623 or Method 1623.1 and
whether the underlying occurrence is greater or lesser than the underlying occurrence distribution
from Round 1).
The LT2 DCTS data on Cryptosporidium concentrations in source water and matrix spike
samples used in the analysis include: sample locations (broken down by PWS and source water),
system schedule, sample dates, sample volumes and the number of oocysts detected in each
sample.
EPA further refined the data set in a few simple steps:
1.	EPA excluded from the analyses any observations that were contested by the Agency;
2.	EPA also excluded data from unfiltered systems and Schedule 4 systems because
procedural differences in Cryptosporidium monitoring and binning requirements make
their data incomparable to the data from other systems;
3.	EPA excluded systems without data on source water type because they do not provide
enough information to enable an analysis that includes source water type effects;
4.	EPA dropped matrix spike data with fewer than 80 oocysts spiked to increase confidence
in the counts, because lower counts are not typically available commercially, and;
5.	EPA dropped from the data sample locations with too few observations.
EPA also tracked the source water type for each sampling location so that it could break out
simulation results by source water type. EPA generated numerical indices for some variables,
and partitioned samples into source water field samples and matrix spike samples.
Six-Year Review 3
Technical Support Document for LT2
6-9
December 2016

-------
6.1.2.1 Overview of Modeling Approach
The Cryptosporidium occurrence model is hierarchical in structure, with several parameters at
the highest level, about 1,250 parameters at midlevel and more than 60,000 parameters at the
lowest level. The model is broadly described in this section.
High level parameters include central values (means) and variances (or precisions) that describe
how parameters at the midlevel are distributed. EPA also included effects for source water type
and system size.
The midlevel parameters are for specific sampling points and laboratories. For sampling points
(1,158 in number), the parameters are the means (averages) of log concentration. These sampling
point means are normally distributed, based on higher-level parameters. For laboratories (49 in
number), the parameters are means of log-odds of recovery and log-odds of zero recovery. These
parameters are normally distributed across labs, based on higher-level parameters. In a particular
laboratory, EPA assumed the log-odds of recovery (recovery / (1 - recovery)) to be normally
distributed across samples, except for those occasions when the recovery is exactly zero.
The log-odds of zero recovery is lab-specific and EPA assumed it to be normally distributed
across laboratories. Zero recovery is a special case, where no oocysts can be counted, regardless
of the number present in the original sample. Cases of zero recovery are rare but explain the
occasional case where zero oocysts are counted for a spiked sample. On equally rare occasions,
zero recovery may explain why the count for a field sample is zero. Most field samples with zero
counts are due to the limited volumes assayed, the low concentrations in source water, and
recoveries that are generally less than 100 percent, but rarely exactly zero.
At the lowest level are more than 60,000 parameters. For each sample, there are two parameters:
recovery (expected probability of detection, shared by each and every oocyst that may be present
in the sample) and the concentration of Cryptosporidium in the source water from which the
sample was drawn. Recoveries at this level are distributed so that their log-odds are normally
distributed around lab-specific means. Concentrations at this level are lognormally distributed,
centered on sampling point means.
The number of oocysts counted for a particular field sample depends on three things:
•	The volume assayed,
•	The recovery (again, expected probability of detection that is shared by every oocyst
present in the sample), and
•	The concentration of oocysts in the source water at the time of sampling.
The expected number of oocysts counted in a field sample is the product of the volume assayed,
the concentration at the time of sampling, and recovery. Because oocysts are assumed to be
randomly dispersed in the source water and because any oocysts in the volume assayed share a
common probability of detection (recovery), the actual number counted is a Poisson random
variable.
Six-Year Review 3
Technical Support Document for LT2
6-10
December 2016

-------
The expected number of oocysts counted in a matrix spike is the product of the number spiked
into the sample and the sample-specific recovery. Because each oocyst is counted with a
probability that is shared by other oocysts in the sample (recovery), the actual number counted is
a binomial random variable whose parameters are the number spiked and the unobserved
recovery.
EPA employed Bayesian Markov Chain Monte Carlo (MCMC) methods to estimate the model
parameters. EPA used noninformative- prior distributions17 to estimate the model's parameters.
The resulting MCMC sample was used to produce summaries about the model parameters and to
indicate uncertainty about them. EPA drew from the MCMC sample to simulate potential
outcomes for a Round 2 monitoring, as described in the next section.
Simulations
In order to separately identify the underlying distribution of concentrations from the distribution
of recovery rates, the preceding model estimation made an important assumption: recovery of
oocysts from matrix spike samples was assumed to be representative of the recovery of oocysts
in field samples. Additional assumptions were needed to predict Round 2 outcomes. One is that
DCTS sampling points are representative of a larger population of sampling points. Another is
that the between- and within-location variances (of log concentration) are stable over time, given
any systematic increases or decreases in Cryptosporidium occurrence.
EPA ran the Just Another Gibbs Sampler (JAGS) package in "R" in parallel fashion to generate
several independent MCMC samples. Following burn-in periods (i.e., time for program to adjust
to data inputs to avoid inputting highly unlikely data), sampled parameter values were thinned to
reduce autocorrelation and, therefore, behave as independent draws from the joint posterior
distribution. EPA confirmed convergence and combined the thinned samples for use in
predicting Round 2 outcomes.
For simulations of Round 2 Cryptosporidium occurrence, sampling point parameters were used
to draw 24 monthly Cryptosporidium concentrations for each sampling point in the data set. EPA
randomly assigned source water types and labs, based on the Round 1 data. For example, if half
of a sampling point's samples went to Lab A during Round 1, then each simulated Round 2
sample for that sampling point went to Lab A with a probability of 0.5.
6.1.2.2 Predictions of Round 2 Occurrence and Binning
Modeled Round 2 Outcomes Using Method 1623
The number of oocysts detected for simulated samples was then processed according to the same
algorithm used on the Round 1 data to classify plants into bins. The Round 2 simulations
assumed that each system drew one sample per month for each sampling location for 24
consecutive months. For each sampling point, the algorithm aggregated (via simple averaging)
17 Priors were either normal with very small precision (0.0001) or uniform over a wide range. Posterior samples
were examined to ensure that these priors had negligible effect on the outcome.
Six-Year Review 3
Technical Support Document for LT2
6-11
December 2016

-------
samples that occur within the same calendar month. For plants with multiple sampling points, the
plant's monthly mean was the simple average of the monthly means across the sampling points.
The annual Cryptosporidium concentration average for the plant could then be computed for
each month by taking the average of the 12 consecutive calendar months prior, ignoring skipped
months (i.e., if there were no data in one of the 12 calendar months, then the annual average was
computed from the remaining 11 calendar months). This annual average was computed for each
month using its respective rolling window of 12 consecutive preceding months. EPA used the
maximum of these rolling annual averages, defined as the maximum running annual average
(MRAA), to determine each plant's placement into one of the four bins. Although plants with 48
or more samples (per sampling point) in the data were binned according to their simple average,
instead of the MRAA, the Round 2 simulations assumed that all sampling points drew one
sample per month for 24 consecutive months.
These simulations predicted the binning outcome for plants that were previously placed in Bin 1
during Round 1, assuming the use of Method 1623 in Round 2 (and assuming no change in the
underlying distribution of Cryptosporidium occurrence). For the 1,046 plants that were placed in
Bin 1 in Round 1, predicted placements for Round 2, using Method 1623, are detailed in Exhibit
6.8.
Exhibit 6.8 Modeled Round 2 Outcomes Using Method 1623, by Source Water
Type
Of 1,046 Plants in Bin 1 in Round 11
Water Type
Plants in Bin 1
Plants in Bin 2
Plants in Bin 3
Plants in Bin 4
Reservoirs/Lakes
543.4
[537, 549]
10.5
[5,17]
0
[0,1]
0
[0,0]
Flowing Streams/Rivers
360.5
[351, 370]
30.3
[21, 40]
0.1
[0,1]
0
[0,0]
Both (R/L and S/R)
26.6
[24, 28]
1.3
[0,3]
0
[0,0]
0
[0,0]
GWUDI - R/L
17
[16, 17]
0
[0,1]
0
[0,0]
0
[0,0]
GWUDI - S/R
54.1
[52, 56]
1.9
[0,4]
0
[0,0]
0
[0,0]
All
1,001.7
[991, 1011]
44.1
[34, 54]
0.2
[0,1]
0.0
[0,0]
Note:
1) The numbers in brackets are the 90 percent credible intervals.
EPA predicted that most of the plants that were in Bin 1 during Round 1 would remain in Bin 1
under Method 1623 in Round 2. Only 4.2 percent (44.3 of 1,046) of plants originally in Bin 1 are
predicted to fall into Bins 2-4 following Round 2. Most of those newly placed in Bins 2-4 are
plants with flowing stream/river source waters.
About 7.8 percent (30.4 of 390.9) of plants with flowing stream/river source waters can be
expected to move from Bin 1 to Bins 2-4 from Round 1 to Round 2, whereas less than 5 percent
of plants with other water types are expected to change from Bin 1 to Bins 2-4. On average, 1.9
percent (10.5 of 543.4) of plants with reservoir/lake sources are predicted to move from Bin 1 to
Six-Year Review 3
Technical Support Document for LT2
6-12
December 2016

-------
Bins 2-4 and 4.7 percent (1.3 of 27.9) of plants with source type "both" are expected to move
from Bin 1 to Bins 2-4.
Modeled Round 2 Outcomes Using Method 1623.1
The predicted outcomes of Round 2 for plants placed in Bin 1 during Round 1, assuming the use
of Method 1623.1 for Round 2 (and assuming no change in Cryptosporidium occurrence levels),
are described here. For the 1,046 plants that were placed in Bin 1 for Round 1, predicted
placements using Method 1623.1 during Round 2 are detailed in Exhibit 6.9.
Exhibit 6.9 Modeled Round 2 Outcomes Using Method 1623.1, by Source Water
Type
Of 1,046 Plants in Bin 1 in Round 11
Water Type
Plants in Bin 1
Plants in Bin 2
Plants in Bin 3
Plants in Bin 4
Reservoirs/Lakes
533.4
[525, 541]
20.5
[13, 29]
0.1
[0,1]
0
[0,0]
Flowing
Streams/Rivers
333.9
[323, 346]
56.7
[45, 68]
0.3
[0,2]
0
[0,0]
Both (R/L and S/R)
25.5
[23, 28]
2.5
[0, 5]
0
[0, 0]
0
[0,0]
GWUDI - R/L
16.9
[16, 17]
0.1
[0,1]
0
[0,0]
0
[0,0]
GWUDI - S/R
52.9
[50, 55]
3.1
[1,6]
0
[0,0]
0
[0,0]
All
962.7
[951, 975]
82.9
[71, 95]
0.4
[0,2]
0.0
[0,0]
Note:
1) The numbers in brackets are the 90 percent credible intervals.
EPA predicted that, with Method 1623.1, most plants in Bin 1 for Round 1 would remain in Bin
1 during Round 2. However, 8.0 percent (83.3 of 1,046) plants originally in Bin 1 were placed in
Bins 2-4 for Round 2 under Method 1623.1, with the greatest proportion of change coming from
plants with mostly flowing stream/river source waters.
About 14.6 percent (57.0 of 390.9) of plants with flowing stream/river source waters can be
expected to move from Bin 1 to Bins 2-4 from Round 1 to Round 2, whereas less than 10
percent of plants with other water types are expected to change from Bin 1 to Bins 2-4. On
average, 3.7 percent (20.6 of 554) of plants with mostly reservoir/lake source waters would move
from Bin 1 to Bins 2-4, and 8.9 percent (2.5 of 28) of plants with source type "both" R/L and
S/R would move from Bin 1 to Bins 2-4.
Comparing Method 1623 and Method 1623.1 for Modeled Round 2 Outcomes
Because Method 1623.1 has a greater recovery rate than Method 1623 for some water matrices, a
greater number of plants' Cryptosporidium concentrations would fall in Bins 2-4 under Method
1623.1, and more plants for each source water type would be placed in Bin 1 under Method
1623.
Six-Year Review 3
Technical Support Document for LT2
6-13
December 2016

-------
Of plants that were in Bin 1 during Round 1, 44.3 plants can expect to be in Bins 2-4 under
Method 1623, and 83.3 plants can expect to be in Bins 2-4 under Method 1623.1; the relative
percentage of systems moving into Bin 2 with the newer method versus the older method is
roughly 8 percent vs. 4 percent (the statistical significance of that difference is apparent from the
non-overlapping of their credible intervals). But even with Method 1623.1, few plants with
reservoir/lake type source waters would be assigned to Bins 2-4. Most of the changes would
occur for plants with mostly flowing stream/river source waters. EPA provides a side-by-side
comparison of Round 2 results using Methods 1623 and 1623.1 in Exhibit 6.10.
Exhibit 6.10 Modeled Round 2 Outcomes by Source Water Type and Method
Round 2 Simulations of Binning for the 1,046 Plants in Bin 1 in Round V
Water Type
Method 1623
Method 1623.1
Plants in Bin 1
Plants in Bins
2-4
Plants in Bin
1
Plants in Bins 2-4
Reservoirs/Lakes
543.4
[537, 549]
10.5
[5,18]
533.4
[525, 541]
20.6
[13, 30]
Flowing Streams
360.5
[351, 370]
30.4
[21, 41]
333.9
[323, 346]
57.0
[45, 70]
All (includes
GWUDI & both)
1,001.7
[991, 1011]
44.3
[34, 55]
962.7
[951, 975]
83.3
[71, 97]
Note:
1) The numbers in brackets are the 90 percent credible intervals.
Predictive Modeling on Alternative Scenarios on Occurrence Distribution
Consideration is also given to what Round 2 predictions would look like if Cryptosporidium
levels were three times higher than estimated for Round 1 or if Cryptosporidium levels were only
one-third of those levels. EPA based the three-fold level and one-third level estimations on a
comparison of the Round 1 data with the ICRSS data. EPA presents the results of Round 2
simulations under alternative scenarios on occurrence distribution in Exhibit 6.11.
Six-Year Review 3
Technical Support Document for LT2
6-14
December 2016

-------
Exhibit 6.11 Plants in Bins 2-4 under Alternative Scenarios on
Occurrence Distribution
Round 2 Simulations of Binning for the 1,046 Plants in Bin 1 in Round 1
Water Type
Plants in Bin 1
in Round 1
% in Bins 2-4, Method
1623
% in Bins 2-4, Method
1623.1
x(3)
x(1)
x(1/3)
x(3)
x(1)
x(1/3)
Reservoirs/Lakes
554
8.0%
1.9%
0.3%
12.9%
3.7%
0.6%
Flowing Streams
391
24.2%
7.8%
1.3%
35.2%
14.6%
3.2%
All
(includes GWUDI &
both)
1,046
14.3%
4.2%
0.7%
21.5%
8.0%
1.6%
If the underlying Cryptosporidium concentrations were three times higher during Round 2 than
Round 1, use of either Method 1623 or Method 1623.1 would result in a greater proportion of
plants falling into Bins 2-4. Under Method 1623, a 3-fold increase in Cryptosporidium
concentration would roughly triple the number of plants in Bins 2-4. Under Method 1623.1, a 3-
fold increase in Cryptosporidium count would more than double the number of plants in Bins 2-
4, with roughly triple the number of reservoir/lake source water plants being placed in Bins 2-4
and roughly double the number of flowing stream source water plants being placed in Bins 2-4.
However, Method 1623.1, with its greater baseline recovery rate, would still result in greater
overall numbers of plants in Bins 2-4 than Method 1623.
If Cryptosporidium counts were reduced to one-third of the Round 1 levels, then use of either
method would result in a decline in the number of plants in Bins 2-4. Under Method 1623, one-
third of the concentration would yield less than one-fifth of the number of plants in Bins 2-4 (for
both reservoir/lake source water plants and flowing stream source water plants). Under Method
1623.1, one-third of the concentration would also yield about one-fourth of the number of plants
in Bins 2-4. Again, use of Method 1623.1 would still consistently place a greater proportion of
plants in Bins 2-4 than Method 1623.
6.2 E. coli Indicator to Predict Cryptosporidium Occurrence
6.2.1 Background
As part of the review of the LT2, EPA reviewed the E. coli trigger level used to determine when
small plants (serving fewer than 10,000 people) were required to monitor for Cryptosporidium.
The LT2 required large surface water systems serving 10,000 or more people to monitor for E.
coli and Cryptosporidium in their source water monthly for two years to allow EPA to better
evaluate the appropriateness of the trigger levels set in the LT2. Due to the high cost of
Cryptosporidium monitoring, the LT2 allowed small systems to monitor for E. coli as an
indicator for Cryptosporidium. These small systems could monitor coli on a twice- monthly
basis for one year first. If the average E. coli concentration exceeded 10 colony-forming units
(cfu) per 100 milliliters (mL) for reservoirs and lakes or 50 cfu/100 mL for flowing streams, as
per the regulatory language, the system was required to monitor for Cryptosporidium. The LT2
staggered the monitoring schedules for four different size categories of systems. Systems serving
at least 100,000 people began monitoring first, followed by systems serving 50,000 to 99,999
Six-Year Review 3
Technical Support Document for LT2
6-15
December 2016

-------
people, followed by systems serving 10,000 to 49,999 people, followed by the small systems.
The small systems began monitoring two years after the systems serving at least 100,000 began
monitoring.
In 2010, EPA performed an analysis using data collected during the first year of
Cryptosporidium and is. coli monitoring from systems serving 100,000 or more people. The
intent of the 2010 analysis was to compare concentrations of Cryptosporidium and E. coli to
determine the adequacy of the trigger level in the LT2. The analysis examined the paired
Cryptosporidium and E. coli data from monitoring at WTPs in systems serving 100,000 or more
people. EPA compared the trigger levels in the LT2 with a range of alternative trigger levels
higher than those specified in the LT2 based on two criteria. The first criterion was the number
of systems with Cryptosporidium levels with a concentration greater than or equal to 0.075
oocysts/L (these systems would require additional treatment). The second criterion was the
number of small systems required to monitor for Cryptosporidium based on E. coli
concentrations. This was estimated based on the occurrence and concentrations of E. coli from
the large system monitoring, The 2010 analysis showed that increasing the E. coli trigger levels
from the levels originally required by the LT2 (E. coli values of 10 and 50 cfu/100 mL for
reservoirs/lakes and flowing streams, respectively) to 100 cfu/100 mL for both types of sources
would result in a 50-percent decrease in the number of plants required to monitor for
Cryptosporidium but only a 10-percent reduction in the number of plants using sources with high
Cryptosporidium levels from being triggered into monitoring. Based on this analysis, EPA issued
guidance to states allowing an alternative trigger level for small systems of 100 cfu/100 mL for
both flowing streams and lakes or reservoirs. Some states, however, are prohibited from allowing
the alternative trigger level by state law.
The purpose of the current analysis, which is described below, is to determine the adequacy of
the 2010 alternative E. coli trigger level using the two years of available monitoring data for all
large WTPs using Method 1623. The current analysis examines several other E. coli trigger
levels for comparison.
6.2.2 Data Cleaning Process
Data EPA used in the analysis are from DCTS, with Cryptosporidium and E. coli data extracted
separately (USEPA, 2016c). EPA cleaned the data to remove invalid, nonrepresentative or
otherwise questionable data (the original data are also available at the same website). For
example, EPA deleted data from inactive or seasonal water systems. Cryptosporidium and E. coli
samples collected on the same date from the same plant were then paired and used to perform the
analysis. More information on the data cleaning procedures used for EPA analysis is available on
EPA's website (USEPA, 2009b).
The first step in the data cleaning process was to calculate the E. coli concentrations of samples
for which the lab entered raw data into the spreadsheet. For E. coli data, labs had a choice of
entering the calculated concentration of E. coli or of entering the raw data from the analytical
method. Calculations were conducted using the EPA-approved methods in the LT2 [40 CFR
141.704], Additionally, EPA omitted a total of four samples from the analysis because the
database contained insufficient data to calculate a concentration.
Six-Year Review 3
Technical Support Document for LT2
6-16
December 2016

-------
After EPA calculated the E. coli concentrations, the Agency cleaned the data to remove data that
could lead to erroneous or biased results. EPA used two separate data cleaning procedures and
compared the results for each to see how they differed. The first data cleaning procedure exactly
replicated the procedure used in the 2010 analysis to determine if the additional year of data
changed the 2010 analysis in any way. Since the original 2010 analysis used in the guidance for
the alternative trigger used only one year of monitoring data, some of the cleaning criteria were
intended to ensure the validity of the additional data. With the data set now complete, some of
the criteria were no longer applicable, so EPA conducted a second data cleaning procedure to
assess the effect of changing these criteria. Therefore, the current analysis includes two sets of
results, one for each data cleaning procedure used.
EPA removed data based on lab certification for Cryptosporidium, sample status and method
reporting limits. Specifically, the following data were removed as part of the first data cleaning
procedure.
•	Samples analyzed by a laboratory that withdrew from EPA's Cryptosporidium
Laboratory Approval Program in September 2007.
•	Any samples with a status code of "contested," unless the sample collection date was the
only issue contested.
•	Any samples with a status code of "entered," because this means that the samples had not
been reviewed by the laboratory or the PWS for correctness.
•	Any samples with a status code of "lab approved," since these had not yet been reviewed
by the PWS for correctness. These were removed unless the E. coli count was determined
to be zero. EPA assumed if the result were zero, the PWS would not contest a sample.
While it is possible allowing the zero results and not positive results might have
introduced some bias, it allowed the greatest number of data points while not ignoring the
possibility of error in the unreviewed samples.
•	E. coli data for which the results were listed as "greater than" a number between 0 and
500. These data were considered to be uninformative.
•	E. coli data for which the result was listed as "less than" a number greater than 10. The
method detection limit on these samples was too high to provide meaningful results.
•	Cryptosporidium matrix spike samples.
•	Samples from systems listed as unfiltered or with no data in the plant filtration column.
The E. coli trigger does not apply to unfiltered systems.
•	Samples from PWSs that were inactive or seasonal (these systems would have had
incomplete data and would require different methods for calculating averages).
For the second data cleaning procedure, fewer data were removed based on sample status. The
only status-based results that were removed were those data that EPA contested or removed.
Six-Year Review 3
Technical Support Document for LT2
6-17
December 2016

-------
EPA assumed that if the systems or laboratories desired to contest a sample, they had had
sufficient time to do so. Also, since the laboratory that withdrew was approved up until the point
it withdrew from the program, EPA kept its data. Data for small systems were also eliminated
from the second data cleaning procedure because all small systems did not report data to DCTS
and this could bias the results. Additionally, in the first data cleaning procedure, EPA
categorized source water type based on the source water used in the first month of sampling. All
Water Treatment Plants (WTPs), however, do not use the same source water type every month
but may switch sources seasonally. Two hundred seventy-four plants were classified as either
using both river and lake sources or switching sources during the sampling period. In the second
data cleaning procedure, EPA categorized these plants by the source type they used more than 50
percent of the time. If no source type was used more than 50 percent of the time the plant was
categorized as both.
In addition to removing these data, EPA performed several other steps to ready the data for
analysis. These steps were completed for both data cleaning procedures:
•	Units of measure were removed from the worksheet column used for the actual sample
concentration value after confirming the correct units;
•	Sample concentrations listed as "ND" were changed to a value of "0;"
•	Sample concentrations listed as a number less than 10 were changed to a value of "0,"
and;
•	If the sample concentration was listed as a number greater than 500, the concentration
was set to that number.
Setting nondetection results and results below the reporting limit to zero does not signify that
there were no microbes in the sample but is rather a mathematical simplification for purposes of
the analysis. While it may introduce some bias by skewing mean E. coli concentrations lower,
EPA applied the change consistently throughout the analysis, and it is consistent with how data
would be handled during sampling.
6.2.2.1 Data Pairing and Calculation
After EPA calculated the E. coli concentrations and cleaned the data, it paired E. coli and
Cryptosporidium samples collected on the same day at the same WTP. If more than one sample
was collected for a given microbiological analysis at a given plant on the same day, those values
were averaged and the average value was considered the concentration on that day. EPA created
two sets of paired data, one for each set of data cleaning criteria. The data cleaning and pairing
operation resulted in 29,741 paired samples representing 1,356 plants using the original cleaning
criteria. Of the 1,356 WTPs, 697 plants used reservoir/lake sources of supply, 618 plants used
flowing streams and 41 plants used both types of supply. For the second set of cleaning criteria,
there were 25,426 paired samples representing 1,372 WTPs. The number of paired samples per
plant ranged from one to 105. Of the 1,372 WTPs, 702 plants used reservoir/lake supplies, 617
used flowing streams and 53 used both types of supply.
Six-Year Review 3
Technical Support Document for LT2
6-18
December 2016

-------
6.2.3 Analysis
EPA used paired samples to calculate overall plant averages for E. coli and Cryptosporidium.
Average counts were calculated according to methods required under the LT2. For E. coli, a
straight average was calculated by summing the results of all samples collected for a given plant
and dividing by the number of samples. For Cryptosporidium, the LT2 states that if 48 or more
samples are taken, then a straight average can be used. If fewer than 48 samples are taken, the
MRAA must be used. Therefore, a straight average was used to calculate plant averages for
plants with 48 or more samples. For plants with fewer than 48 samples, calculation of the
MRAAs was complicated by the fact that many plants did not have continuous data, either
because of the data cleaning procedure or because of missed samples. Therefore, for plants with
fewer than 12 samples, a simple average was used, as this would be equivalent to a RAA. For
plants with 12 to 47 samples, EPA determined the first and last sample dates and calculated
RAAs for each one year period beginning with the first sample. Each RAA was calculated using
all samples collected in that one year period. The last RAA calculated was the one ending with
the last sample. The highest of these RAAs, the MRAA, was selected as the plant average. This
is consistent with the LT2 language, which requires systems to use the MRAA for treatment
determination.
6.2.3.1 Criterion Values
Criteria used in this analysis are from Pope et al. (2002). Mathematically these criteria can be
expressed by considering the true and false positives and negatives, as shown in Exhibit 6.12.
Appendix B provides the detailed observed data for specific trigger values corresponding to the
variables in Exhibit 5.12.
Exhibit 6.12 Definition of Variables Used in Analysis

Cryptosporidium
concentration < 0.075
oocysts/L
Cryptosporidium
concentration > 0.075
oocysts/L
E. coli concentration > trigger value
B
D
E. coli concentration < trigger value
A
C
Based on this definition of variables the eight criteria calculated can be defined as:
•	False positives = B/(B+D);
•	False negatives = C/(A+C);
•	Sensitivity = D/(C+D);
•	Specificity = A/(A+B);
•	Plants Protectively Classified = (A+B+D)/(A+B+C+D);
•	Plants Incorrectly Classified = C/(A+B+C+D);
Six-Year Review 3
Technical Support Document for LT2
6-19
December 2016

-------
•	Plants Correctly Exempted from Cryptosporidium Monitoring = A/(A+B+C+D), and;
•	Plants Required to Monitor for Cryptosporidium = (B+D)/(A+B+C+D).
After EPA calculated the plant averages for E. coli and Cryptosporidium, EPA used the data to
calculate values for the criteria above, for E. coli trigger levels of 10, 50, 75, 100, 150 and 200
cfu/100 mL. The formulas used to calculate each of these criterion values are provided in the
bullets following Exhibit 6.12.
These criterion values were calculated for each of the two sets of data cleaning criteria. Exhibit
6.13 through Exhibit 6.15 show the calculated criterion values for each of the trigger levels
examined for the original data cleaning criteria for each source water type.
Exhibit 6.16 through Exhibit 6.18 show the calculated criterion values for the revised cleaning
criteria, again by source water type.
Exhibit 6.13 Criteria for Reservoirs and Lakes Using Original Cleaning Procedure
Levels of Testing Based on Various Trigger Levels

E. coli Trigger Level (cfu/100 mL)
Criterion
10
50
75
100
150
200
B/(B+D) (False
Positives)
95.51%
92.91%
91.18%
90.28%
90.00%
84.62%
C/(A+C) (False
Negatives)
1.56%
1.93%
1.85%
2.08%
2.44%
2.38%
D/(C+D) (Sensitivity)
70.00%
45.00%
45.00%
35.00%
20.00%
20.00%
A/(A+B) (Specificity)
55.98%
82.57%
86.26%
90.40%
94.68%
96.75%
(A+B+D)/(A+B+C+D)
(Plants Protectively
Classified)
99.14%
98.42%
98.42%
98.13%
97.70%
97.70%
C/(A+B+C+D) (Plants
Incorrectly Classified
0.86%
1.58%
1.58%
1.87%
2.30%
2.30%
A/(A+B+C+D) (Plants
Correctly Exempted
from Cryptosporidium
Monitoring)
54.38%
80.20%
83.79%
87.80%
91.97%
93.97%
(B+D)/(A+B+C+D)
(Plants Required to
Monitor for
Cryptosporidium)
44.76%
18.22%
14.63%
10.33%
5.74%
3.73%
Six-Year Review 3
Technical Support Document for LT2
6-20
December 2016

-------
Exhibit 6.14 Criteria for Rivers and Flowing Streams Using Original Cleaning
Procedure
Levels of Testing Based on Various Trigger Levels
E. coli Trigger Level (cfu/100 mL)
Criterion
10
50
75
100
150
200
B/(B+D) (False
Positives)
84.98%
81.13%
79.68%
77.15%
76.55%
72.97%
C/(A+C) (False
Negatives)
1.52%
3.04%
3.90%
3.99%
5.61%
5.77%
D/(C+D) (Sensitivity)
97.33%
89.33%
84.00%
81.33%
70.67%
66.67%
A/(A+B) (Specificity)
23.94%
46.96%
54.51%
62.06%
68.14%
75.14%
(A+B+D)/(A+B+C+D)
(Plants Protectively
Classified)
99.68%
98.71%
98.06%
97.73%
96.44%
95.95%
C/(A+B+C+D) (Plants
Incorrectly Classified
0.32%
1.29%
1.94%
2.27%
3.56%
4.05%
A/(A+B+C+D) (Plants
Correctly Exempted
from Cryptosporidium
Monitoring)
21.04%
41.26%
47.90%
54.53%
59.87%
66.02%
(B+D)/(A+B+C+D)
(Plants Required to
Monitor for
78.64%
57.44%
50.16%
43.20%
36.57%
29.94%
Cryptosporidium)






Six-Year Review 3
Technical Support Document for LT2
6-21
December 2016

-------
Exhibit 6.15 Criteria for All Samples Using Original Cleaning Procedure
Levels of Testing Based on Various Trigger Levels

E. coli Trigger Level (cfu/100 mL)
Criterion
10
50
75
100
150
200
B/(B+D) (False
Positives)
88.92%
83.98%
82.23%
79.48%
78.31%
74.07%
C/(A+C) (False
Negatives)
1.50%
2.32%
2.57%
2.77%
3.69%
3.77%
D/(C+D) (Sensitivity)
91.92%
79.80%
75.76%
71.72%
59.60%
56.57%
A/(A+B) (Specificity)
41.93%
67.06%
72.39%
78.12%
83.05%
87.27%
(A+B+D)/(A+B+C+D)
(Plants Protectively
Classified)
99.41%
98.53%
98.23%
97.94%
97.05%
96.83%
C/(A+B+C+D) (Plants
Incorrectly Classified
0.59%
1.47%
1.77%
2.06%
2.95%
3.17%
A/(A+B+C+D) (Plants
Correctly Exempted
from Cryptosporidium
Monitoring)
38.86%
62.17%
67.11%
72.42%
76.99%
80.90%
(B+D)/(A+B+C+D)
(Plants Required to
Monitor for
Cryptosporidium)
60.55%
36.36%
31.12%
25.52%
20.06%
15.93%
Six-Year Review 3
Technical Support Document for LT2
6-22
December 2016

-------
Exhibit 6.16 Criteria for Reservoirs and Lakes Using Revised Cleaning Procedure
Levels of Testing Based on Various Trigger Levels

E. coli Trigger Level (cfu/100 mL)
Criterion
10
50
75
100
150
200
B/(B+D) (False
Positives)
94.72%
92.09%
90.20%
89.47%
90.00%
82.61%
C/(A+C) (False
Negatives)
0.58%
1.78%
1.83%
2.08%
2.57%
2.50%
D/(C+D) (Sensitivity)
90.48%
52.38%
47.62%
38.10%
19.05%
19.05%
A/(A+B) (Specificity)
49.93%
81.20%
86.49%
90.01%
94.71%
97.21%
(A+B+D)/(A+B+C+D)
(Plants Protectively
Classified)
99.72%
98.58%
98.43%
98.15%
97.58%
97.58%
C/(A+B+C+D) (Plants
Incorrectly Classified
0.28%
1.42%
1.57%
1.85%
2.42%
2.42%
A/(A+B+C+D) (Plants
Correctly Exempted
from Cryptosporidium
Monitoring)
48.43%
78.77%
83.90%
87.32%
91.88%
94.30%
(B+D)/(A+B+C+D)
(Plants Required to
Monitor for
Cryptosporidium)
51.28%
19.80%
14.53%
10.83%
5.70%
3.28%
Six-Year Review 3
Technical Support Document for LT2
6-23
December 2016

-------
Exhibit 6.17 Criteria for Rivers and Streams Using Revised Cleaning Procedure
Levels of Testing Based on Various Trigger Levels

E. coli Trigger Level (cfu/100 mL)
Criterion
10
50
75
100
150
200
B/(B+D) (False
Positives)
87.53%
84.06%
82.84%
81.03%
80.86%
77.78%
C/(A+C) (False
Negatives)
0.83%
2.94%
3.50%
4.12%
5.64%
5.61%
D/(C+D) (Sensitivity)
98.41%
87.30%
82.54%
76.19%
63.49%
60.32%
A/(A+B) (Specificity)
21.48%
47.65%
54.69%
63.00%
69.49%
75.99%
(A+B+D)/(A+B+C+D)
(Plants Protectively
Classified)
99.84%
98.70%
98.22%
97.57%
96.27%
95.95%
C/(A+B+C+D) (Plants
Incorrectly Classified
0.16%
1.30%
1.78%
2.43%
3.73%
4.05%
A/(A+B+C+D) (Plants
Correctly Exempted
from Cryptosporidium
Monitoring)
19.29%
42.79%
49.11%
56.56%
62.40%
68.23%
(B+D)/(A+B+C+D)
(Plants Required to
Monitor for
Cryptosporidium)
80.55%
55.92%
49.11%
41.00%
33.87%
27.71%
Six-Year Review 3
Technical Support Document for LT2
6-24
December 2016

-------
Exhibit 6.18 Criteria for All Samples Using Revised Cleaning Procedure
Levels of Testing Based on Various Trigger Levels

E. coli Trigger Level (cfu/100 mL)
Criterion
10
50
75
100
150
200
B/(B+D) (False
Positives)
90.36%
86.17%
84.38%
82.54%
81.78%
78.00%
C/(A+C) (False
Negatives)
0.83%
2.41%
2.62%
3.00%
3.86%
3.92%
D/(C+D) (Sensitivity)
95.56%
76.67%
72.22%
65.56%
52.22%
48.89%
A/(A+B) (Specificity)
37.13%
66.46%
72.62%
78.24%
83.54%
87.83%
(A+B+D)/(A+B+C+D)
(Plants Protectively
Classified)
99.71%
98.47%
98.18%
97.74%
96.87%
96.65%
C/(A+B+C+D) (Plants
Incorrectly Classified
0.29%
1.53%
1.82%
2.26%
3.13%
3.35%
A/(A+B+C+D) (Plants
Correctly Exempted
from Cryptosporidium
Monitoring)
34.69%
62.10%
67.86%
73.10%
78.06%
82.07%
(B+D)/(A+B+C+D)
(Plants Required to
Monitor for
Cryptosporidium)
65.01%
36.37%
30.32%
24.64%
18.80%
14.58%
As expected, E. coli and Cryptosporidium values are higher in flowing streams than they are in
reservoirs and lakes. The qualitative trends, however, are similar for both source types. For both
source types specificity (the proportion of Cryptosporidium negatives that are correctly identified
by E. coli negatives) increases with higher trigger levels, while sensitivity (the proportion of
Cryptosporidium positives that are correctly identified by E. coli positives) decreases. Selecting
the proper trigger level involves balancing the number of plants with high Cryptosporidium
concentrations properly triggered into monitoring versus those with low Cryptosporidium
concentrations correctly allowed to avoid monitoring.
There does not seem to be any significant difference between criteria calculated based on the
original cleaning procedure (Exhibit 6.13 through Exhibit 6.15) and criteria calculated based on
the revised cleaning procedure (Exhibit 6.16 through Exhibit 6.18), and the two procedures lead
to similar trends and conclusions. There are slight differences in the criterion values, but they do
not seem to point to any clear bias. The revised cleaning procedure shows slightly higher false
positives and false negatives with a lower specificity than the original cleaning procedure. The
differences appear to be mostly in the river and stream category. There are two possible causes
for any bias. The first is the exclusion of the Schedule 4 systems in the revised cleaning
procedure. Schedule 4 systems were most likely triggered into Cryptosporidium monitoring by
high E. coli concentrations, so they tend to have higher E. coli and Cryptosporidium
concentrations. This could account for the slightly lower specificity in the revised cleaning
Six-Year Review 3
Technical Support Document for LT2
6-25
December 2016

-------
procedure criterion values, as some of the higher E. coli and Cryptosporidium samples were
eliminated, leading to less specificity. The other possible cause of bias may be the inclusion of
the laboratory and of previously unreviewed data in the revised cleaning procedure data. If these
data were skewed, that might also account for some difference.
In general, the sensitivity is very high at the lower E. coli trigger levels. This means that at the
lower trigger levels, most plants that have sources of supply with Cryptosporidium
concentrations greater than the binning level (i.e., Bin 1 requires a Cryptosporidium
concentration < 0.075 oocysts per liter) would have been triggered into monitoring based on their
E. coli data. This trend is not as pronounced in reservoirs and lakes as it is in flowing streams.
The number of plants required to monitor and the number of plants incorrectly classified,
however, are also higher at the lower trigger values. As the trigger level is increased, the
specificity drops, and the number of plants correctly exempted from monitoring increases.
6.2.4 Results and Discussion
The effectiveness of the alternative E. coli trigger can be thought of using two criteria.
1.	Sensitivity. Plants that have high Cryptosporidium concentrations (>0.075 oocysts/L)
would be correctly triggered into monitoring based on their E. coli monitoring.
2.	The total number of plants triggered into Cryptosporidium monitoring. Ideally, only those
plants with high Cryptosporidium would be triggered into monitoring.
The most effective trigger level is that which minimizes the plants required to unnecessarily
monitor while maximizing the number of plants with high Cryptosporidium that are triggered
into monitoring.
6.2.4.1 Plants Monitoring Versus Plants Triggered Into Treatment
Using the calculated criterion values for each E. coli trigger level and scaling up by the total
number of small WTPs for each source water type, EPA completed the following estimates.
•	The total number of small plants that would be required to monitor based on E. coli
results.
•	The total number of small plants with Cryptosporidium greater than or equal to 0.075
oocysts/L that would be correctly assigned to treatment bins.
EPA determined the total number of small plants of each source water type by assuming that
small systems have the same percentage of different source water types as the systems in DCTS
and multiplying by the total number of small surface water systems from EPA's State Drinking
Water Information System database. This analysis assumes that the underlying Cryptosporidium
and E. coli concentrations are the same for large and small systems. EPA multiplied the
percentage of plants required to monitor for each source water type by the total number of small
systems for that source water type to obtain the total number of small plants that would be
required to monitor for Cryptosporidium. Likewise, the percentage of small plants with greater
than or equal to 0.075 oocysts/L for each source water type was multiplied by the total number of
Six-Year Review 3
Technical Support Document for LT2
6-26
December 2016

-------
small plants of that source water type to obtain the total number of small plants with high
Cryptosporidium concentrations. Exhibit 6.19 shows the number of high-( Cryptosporidium plants
triggered into monitoring versus the total number of plants triggered into monitoring for both
lakes and reservoirs and flowing stream sources, based on the original data cleaning procedure.
Exhibit 6.20 shows the number of high-( Cryptosporidium plants triggered into monitoring versus
the total number of plants triggered into monitoring for both lakes and reservoirs and flowing
stream sources using the revised cleaning procedure.
Exhibit 6.19 E. coli Trigger Analysis Results for Small Plants Using Original
Cleaning Procedures
o
M-
o
"O
 0.075 oocysts/L
250
200
150
100
50
100 CFU/lOOmL
50 CFU/lOOmL
-~ -Reservoir & Lakes
-¦-Flowing Streams
A LT2 Trigger Limits
Alternative Trigger
~ "~' ^ 100 CFU
	A;
10 CFU/100mL
100 CFU/lOOmL
200 400 600 800 1,000 1,200 1,400 1,600 1,800
Plants Triggered into Monitoring
Six-Year Review 3
Technical Support Document for LT2
6-27
December 2016

-------
Exhibit 6.20 E. coli Trigger Analysis Results for Small Plants Using Revised
Cleaning Procedures
250
Effectiveness of E. coli Monitoring for Identifying High-Cryptosporidium Plants
S 200
50 CFU/100 mL
100 CFU/100 mL
-Reservoirs& Lakes
o 150
LT2 Trigger Limit
Alternative Trigger
100
10 CFU/100 mL
100 CFU/100 ad,
200
400
600
800 1000 1200
Plants Triggered into Monitoring
1400
1600
1800
Six-Year Review 3
Technical Support Document for LT2
6-28
December 2016

-------
As the trigger level drops, an increased number of plants with high Cryptosporidium required to
monitor increases. While each lower trigger level has a larger number of high Cryptosporidium
plants triggered, the increase is greatest between the 150 and 100 cfu/100 mL trigger values for
both the original and revised data cleaning procedures. The revised cleaning procedures
described in Section 6.2.2 resulted in 26 plants being triggered into Cryptosporidium monitoring
at the alternative trigger level for lakes and reservoirs, compared to 23 plants with the original
cleaning procedures. For flowing streams the opposite trend was observed with the revised
cleaning procedures resulting in 156 plants triggered and the original cleaning procedures
resulting in 198 plants triggered.
6.2.4.2 Analysis of Current Trigger Level and Comparison to Previous Data
Exhibit 6.21 shows the same graph that was prepared for the original alternative trigger analysis
in 2010.
Exhibit 6.21 2010 Trigger Analysis Results
Effectiveness of Monitoring to Capture H'\gh-Cryptosporidium Plants
250
=J
CT
CD
QC

+¦»
c
¦4-J 200
LAKE/ RESERVOIR
150
FLOWING STREAM
_nj
Q_
• Trigger Limits under LT2
100
• Alternative Guidance Trigger
Limits
500
1,000
1,500
2,000
2,500
3,000
3,500
Plants Required to Monitor for Crypto
Qualitatively the graphs are very similar to the graphs from the current analysis. They are both
flatter toward the right hand side of the graph and show an inflection point near the alternative
trigger level of 100 cfu/100 mL.
6.2.4.3 Recommended Trigger Level
As both the original 2010 graphs and those completed for this analysis show inflection points at a
trigger level of 100 cfu/100 mL, the alternative trigger level is supported by the data collected
Six-Year Review 3	6-29	December 2016
Technical Support Document for LT2

-------
during the first round of Cryptosporidium monitoring. Exhibit 6.22 shows the reductions in each
of the criteria between the LT2 trigger level and a trigger level of 75, 100 or 150 cfu/100 mL.
Exhibit 6.22 Percent Reduction in Plants Required to Monitor with Alternate
Trigger Levels
Reduction in Plants Required to Monitor from Rule Trigger
Levels1 to Alternate Trigger Levels2

Criterion
Alternate Trigger
75
100
150
Original
Cleaning
Procedure
H ig h-Cryptosporidium
Plants Triggered
Reservoirs/Lakes
36%
50%
71%
Flowing Streams
6%
9%
21%
Plants Required to
Monitor
Reservoirs/Lakes
67%
77%
87%
Flowing Streams
13%
25%
36%
Revised
Cleaning
Procedure
H ig h-Cryptosporidium
Plants Triggered
Reservoirs/Lakes
47%
58%
79%
Flowing Streams
5%
13%
27%
Plants Required to
Monitor
Reservoirs/Lakes
72%
79%
89%
Flowing Streams
12%
27%
39%
Notes:
1)	10 cfu/100 mL for reservoirs/lakes and 50 cfu/100 mL for flowing streams.
2)	75, 100 or 150 cfu/100 mL for all types of supply sources.
Comparing the original LT2 trigger values to the alternative trigger value of 100 cfu/100 mL
reveals a reduction in plants required to monitor by 77 percent for reservoirs/lakes and 25
percent for flowing streams based on the original cleaning procedures. The reduction is 79
percent for reservoirs and lakes and 27 percent for flowing streams based on the revised cleaning
procedures. This results in a 50-percent reduction in high-Cryptosporidium plants being triggered
for reservoirs and lakes and 9 percent for flowing streams for the original cleaning procedures
and a 58-percent reduction for reservoirs and lakes and a 13-percent difference for rivers and
streams based on the revised data cleaning procedures.
Using the original cleaning procedure, a trigger value of 150 cfu/100 mL would result in a 71-
percent drop in high Cryptosporidium plants being triggered for lakes and reservoirs and 21
percent for flowing streams while reducing the number of plants monitoring by 87 and 36
percent, respectively. Using the revised cleaning procedure a trigger value of 150 cfu/100 mL
would result in a 79-percent drop in high Cryptosporidium plants triggered for reservoirs and
lakes and 27-percent reduction for rivers and streams, while reducing the number of plants
monitoring by 89 and 39 percent, respectively.
The inflection point at the alternative trigger value of 100 cfu/100 mL indicates this is where the
incremental number of high-Cryptosporidium plants correctly triggered is maximized versus
minimizing the number of small plants required to perform Cryptosporidium monitoring.
Moreover, use of the 100 cfu/100 mL alternative indicator allows for a large reduction in plants
required to monitor for Cryptosporidium, with only a small reduction in plants with high
Cryptosporidium not being required to monitor.
Six-Year Review 3
Technical Support Document for LT2
6-30
December 2016

-------
6.3 Cooccurrence of Cryptosporidium and Other Pathogens of Concern
6.3.1 Giardia and Cryptosporidium Cooccurrence from ICR Supplemental Survey Data
6.3.1.1 Background on the ICR Supplemental Survey
The ICRSS required Method 1623 for measurement of Cryptosporidium and Giardia in the
surface water sources of 87 drinking water systems. Forty of the systems were medium sized
(serving between 10,000 and 100,000 people) and 47 were large (serving at least 100,000
people). Each system's source water was categorized as one of three types (flowing streams,
reservoirs/lakes or both). The distribution of sources by size and water type is shown in Exhibit
6.23.
Exhibit 6.23 Distribution of ICRSS Source Waters by System Size and Water Type
Water Type
System Size
Medium
Large
Flowing Stream (Type 1)
17
16
Lake/Reservoir (Type 2)
19
29
Both Flowing Stream and
Lake/Reservoir (Type 3)
4
2

Total = 40
Total = 47
6.3.1.2 Summary of ICRSS Data
Over a 12-month period, beginning in March 1999, participating systems sampled each source
twice per month for Cryptosporidium, although some systems assayed extra samples while
others had missing samples. Testing for Giardia began in July 1999; samples taken earlier were
assayed only for Cryptosporidium. A total of 2,086 samples were assayed (average of 24 samples
per source) for Cryptosporidium and 1,350 of these samples were also assayed for Giardia
(average of 15.5 per source). Exhibit 6.24 provides some additional summary statistics.
Exhibit 6.24 ICRSS Summary Statistics for Cryptosporidium and Giardia
Statistic
Cryptosporidium
Summary Value
Giardia
Summary Value
Average Volume Assayed
10.11 L
10.17 L
Total Oocysts Counted
1057
3255
Average Concentration
0.050/L
0.237/L
(total count / total volume)


Percent of Samples with
86.4%
67.1%
No Detections


Percent of Sources with
20.7%
18.4%
No Detections


Exhibit 6.25 is a scatterplot showing the observed mean concentrations for the 87 sources. A
small constant (0.005) was added so sources with no detections could be displayed. Some points
overlap others. In particular, the point in the lower left represents all of the sources having no
Six-Year Review 3
Technical Support Document for LT2
6-31
December 2016

-------
detections for either Cryptosporidium or Giardia. Exhibit 6.25 suggests a positive correlation
between Cryptosporidium and Giardia and shows greater overall occurrence for flowing streams
than for lakes and reservoirs.
Exhibit 6.25 Scatterplot of Observed Mean Concentrations for 87 Source Waters
1.00"
Q
: 0.10 H
Size
~	Large
A Medium
WaterType
~	BOTH
TO	*	A A A	* FL STREAM
°	*	* LAKE/RES
0.01 -
0.01	0.10	1.00
CryptoMean + Delta
6.3.1.3 Statistical Modeling
EPA employed a simple model to describe the co-occurrence of Cryptosporidium and Giardia.
For each source water, EPA assumed recoverable protozoa of the two types are independently
lognormally distributed over time. The number counted in a sample volume of V liters assayed is
assumed to be a Poisson random variable with parameter V * C, where C is the unobserved
concentration of recoverable oocysts in the waterbody at the time of sampling. Between source
waters, EPA assumed the mean concentrations of Cryptosporidium and Giardia to be
lognormally distributed with a variance-covariance structure. This simple model therefore
includes random effects for source waters, but no effects for water type or for system size. The
model includes a correlation coefficient to reveal the degree to which levels of the two protozoa
are correlated across sources. EPA also evaluated two more complex models. The first of these
includes effects for water type (two additional parameters), but not for system size, while the
second includes an effect for each combination of water type and system size (five additional
parameters).
Six-Year Review 3
Technical Support Document for LT2
6-32
December 2016

-------
EPA estimated parameters using Bayesian MCMC methods. Deviance information criterion
(DIC), a measure of model quality, is monitored for each model. Higher DIC values indicate
decreasing model quality. Accordingly, the results, shown in Exhibit 6.26, reveal the more
complex models (Model 2 and Model 3) perform better than the simple model (Model 1), which
had no effects for water type or system size. Moreover, Model 2 has the lowest DIC and is
judged to be the best of the three models, though the DIC for Model 3 is close to Model 2.
Exhibit 6.26 Deviance Information Criterion Model Results
Model Name
Features
DIC
Model 1
Effects for Sources
4662.5
Model 2
Model 1 + Effects for
Water Types
4648.3
Model 3
Model 2 + Effects for
System Size
4649.7
The following summarizes the estimates (posterior means) for several high-level parameters.
•	The effect of Type 2 (Lake/Reservoir) water on Cryptosporidium is -1.716.
o Reservoirs and lakes have significantly less Cryptosporidium than flowing
streams.
o Overall, the levels in reservoirs and lakes are lower by about a factor of 5.5
(exp(1.716)).
•	The effect of Type 3 (both flowing stream and lake/reservoir) water on Cryptosporidium
is-1.496.
o Sources of type "both" have significantly less Cryptosporidium than flowing
streams.
o Overall levels in sources of type "both" are lower by about a factor of 4.5
(exp(1.496)).
o Levels in reservoirs and lakes are not significantly different from levels in
category "both."
•	The effect of Type 2 (lake/reservoir) water on Giardia is -2.938.
o Reservoirs and lakes have significantly less Giardia than flowing streams.
o Overall, the levels in reservoirs and lakes are lower by about a factor of 19
(exp(2.938)).
•	The effect of Type 3 (both flowing stream and lake/reservoir) water on Giardia is -2.516.
o Sources of type "both" have significantly less Giardia than flowing streams.
Six-Year Review 3
Technical Support Document for LT2
6-33
December 2016

-------
o Overall levels in sources of type "both" are lower by about a factor of 12
(exp(2.516)).
o Levels in reservoirs and lakes are not significantly different from levels in
category "both."
•	The correlation coefficient is 0.637.
o Significantly greater than zero (positive correlation between average levels of
Cryptosporidium and Giardia).
o Giardia levels explain nearly two-thirds of the between-source variance of
Cryptosporidium. Likewise, Cryptosporidium levels explain nearly two-thirds of
the between-source variance of Giardia levels.
•	Within-location standard deviation of {^{Cryptosporidium concentration) is 2.001.
o For an individual source water, the ratio of the 97.5th percentile concentration to
2.5th percentile concentration is about 2600.
•	Within-location standard deviation of In {Giardia concentration) is 1.396.
o For an individual source water, the ratio of the 97.5th percentile concentration to
2.5th percentile concentration is about 240.
o Giardia has lower within-source variability (on a log scale) than does
Cryptosporidium.
6.3.1.4 Results
The positive correlation of Cryptosporidium and Giardia indicates that locations with high
Cryptosporidium occurrence tend also to have high Giardia levels. Moreover, by reducing
Cryptosporidium exposures, plants whose Round 1 Cryptosporidium data place them in Bins 2-4
may also be reducing higher-than-average Giardia exposures. Median model parameter estimate
calculations based on the model; using those estimates led to the following findings.
•	Occurrences of both Giardia and Cryptosporidium are greater in flowing streams than in
reservoirs and lakes.
o On average, flowing streams have nearly 20 times as much Giardia as do
reservoirs and lakes.
o On average, flowing streams have nearly six times as much Cryptosporidium as
do reservoirs and lakes.
•	Across sources of either type, Cryptosporidium and Giardia are positively correlated,
signifying that sources with high Cryptosporidium occurrence tend to also have high
Giardia occurrence.
Six-Year Review 3
Technical Support Document for LT2
6-34
December 2016

-------
o Of all Cryptosporidium in flowing streams, more than 80 percent occur in streams
with mean concentrations greater than 0.075 oocysts/L. Most flowing streams
(approximately 60 percent of them) have mean concentrations less than 0.075
oocysts/L.
o Nearly 80 percent of the Giardia in flowing streams occur in the 40 percent of
locations where mean Cryptosporidium concentrations exceed 0.075 oocysts/L.
o Of all Cryptosporidium in reservoirs and lakes, about 30 percent occur in those
with mean concentrations greater than 0.075 oocysts/L, but only about 2 percent
of reservoirs and lakes have mean concentrations greater than 0.075 oocysts/L.
o About 25 percent of the Giardia in reservoirs and lakes occur in the 2 percent of
reservoirs and lakes where mean Cryptosporidium concentrations exceed 0.075
oocysts/L.
• In both water types, average Giardia occurrence exceeds that of Cryptosporidium.
o In flowing streams, the average Giardia concentration is about nine times the
average Cryptosporidium concentration.
o In reservoirs and lakes, the average Giardia concentration is about three times the
average Cryptosporidium concentration.
EPA based these findings on median parameter values and did not consider parameter
uncertainty. The ratios and percentages stated are best estimates, based on the ICRSS data. Data
from Round 1 monitoring reveal that Cryptosporidium occurrence levels are significantly lower
than they were at the time of the ICRSS. Without assays for both Cryptosporidium and Giardia,
however, the Round 1 data are not useful for testing the model described above. Levels of
Giardia may have changed, due to the same unknown factors that led to the reduction in
Cryptosporidium. Also, the strength of the correlation may have changed, but new data would be
needed to inform EPA's understanding of the correlation.
6.4 Summary
In developing the LT2, EPA relied on data from surveys conducted in earlier years (1997-1998
for the ICR survey and 1999-2000 for the ICRSS). Statistical modeling of those data revealed
differing occurrence levels due to true differences in occurrence, systematic differences in the
measurement methods or other factors. Recognizing that the LT2 would not be implemented for
some time, EPA considered the differences as evidence that occurrence levels could vary from
year-to-year. Costs and benefits were derived under these alternative models of occurrence to
represent uncertainty about future occurrence.
In the LT2 EA (USEPA, 2005a), EPA based estimated benefits only on reduced exposure to
Cryptosporidium, although treatment and other actions designed to reduce Cryptosporidium
exposures were also assumed to reduce exposures to other pathogens. A more recent analysis of
ICRSS data identified a strong correlation between average levels of Cryptosporidium and
Giardia. This suggests that the LT2 controls aimed primarily at Cryptosporidium removal may
Six-Year Review 3
Technical Support Document for LT2
6-35
December 2016

-------
be effective at lowering exposures to Giardia as well. The benefit of these additional reductions
in Giardia in finished water is believed to be supplemental to existing treatment, however, due to
the effectiveness of chlorine and other disinfectants used for Giardia inactivation.
The Round 1 monitoring data reveal that Cryptosporidium occurrence levels are significantly
lower than predicted by any of the earlier, survey-based models. The numbers of plants falling
into action bins (Bins 2, 3 and 4) was smaller than expected, so both the costs and benefits
related to binning and treatment are lower than predicted in the LT2 EA. It is not clear, however,
whether the low occurrence in Round 1 is due to random year-to-year differences or part of a
systematic decline that can be expected to continue through Round 2 monitoring. The rationale
for Round 2 remains valid: to determine where source water occurrence has changed to the
extent that changes in treatment are needed.
EPA also used the Round 1 monitoring data to better understand the utility of E. coli as a trigger
for Cryptosporidium monitoring by small systems. The complete Round 1 dataset supports the
alternative trigger levels that were based on data available in 2010. Compared to the original LT2
trigger levels (10 cfu/100 mL for reservoirs/lakes and 50 cfu/100 mL for flowing streams), the
alternative trigger (100 cfu/100 mL for all source waters) greatly reduces the number of plants
required to monitor for Cryptosporidium, with only a small reduction of monitoring among
plants with high levels of Cryptosporidium.
Finally, EPA used statistical models based on the Round 1 data to gauge the sensitivity of Round
2 predictions to systematic changes in recovery (due to using Method 1623.1 rather than
1622/1623) and occurrence. As expected, the number of plants requiring additional treatment
increases with systematic increases in occurrence and recovery and decreases with systematic
decreases in occurrence and recovery.
Six-Year Review 3
Technical Support Document for LT2
6-36
December 2016

-------
7
LT2 Microbial Toolbox and Other Tools
Public water systems (PWSs) required to provide additional Cryptosporidium treatment under
the Long Term 2 Enhanced Surface Water Treatment Rule (LT2) can select from a variety of
microbial toolbox treatment and management strategy options. The LT2 Toolbox Guidance
Manual (USEPA, 2010c) provides technical information on applying the toolbox options. This
section serves to analyze information on the use and effectiveness of risk mitigation tools in the
toolbox. This chapter focuses on the relevant new information published since 2006, but includes
some older papers as needed for context.
Where information is available this section identifies and describes the types of challenges
encountered and how they were overcome by water systems and state programs when
implementing the microbial toolbox requirements of the LT2. The implementation issues
described are based on input from various EPA offices and regions, along with the review of
several related documents (e.g., utility reports, journal publications). This document also
provides additional input received from stakeholders during public stakeholder meetings held by
EPA.
EPA's literature review efforts focused on obtaining case studies of water systems placed in Bins
2, 3 and 4 illustrating challenges with implementation and approaches to their resolution.
Additional information EPA reviewed identifies, in some cases, which toolbox components
systems are implementing. EPA also contacted nine utilities assigned to either Bins 2 or 3
regarding the use of the microbial toolbox tools or other tools. The contacted utilities all must
provide additional treatment under the LT2. Appendix C summarizes the information gathered.
Background on the LT2 Microbial Toolbox Requirements
The LT2 microbial toolbox provides technical information on applying the "toolbox" of
Cryptosporidium treatment and management strategies that are part of the Long Term 2
Enhanced Surface Water Treatment Rule. This toolbox, referred to in the LT2 as the microbial
toolbox, covers the following categories: source protection, pre-filtration, treatment performance,
additional filtration and inactivation options. The microbial and disinfection byproducts (MDBP)
federal advisory committee (FAC) recommended the toolbox options as part of a multiple barrier
treatment process that provides a number of protective "layers" against drinking water
contamination. These options emphasize using more than one method to minimize, remove or
inactivate microorganisms and minimize disinfection byproducts (DBPs).
Under the LT2, systems receive prescribed treatment credits by meeting conditions for one
toolbox option or a combination of options. States are responsible for establishing performance
criteria for many of the options and for developing monitoring and reporting requirements to
determine compliance with the established criteria. Technical guidance documents and
implementation guidance are available to assist with understanding the toolbox options, site-
specific validation or demonstrated performance considerations, and compliance monitoring
recommendations (USEPA, 2006c; 2007; 2010c).
Background on the LT2 Microbial Toolbox Implementation Issues
Six-Year Review 3
Technical Support Document for LT2
7-1
December 2016

-------
The compliance dates for implementing toolbox options resulting from Round 1 monitoring
range from April 2012 for systems on Schedule 1 to October 2014 for systems on Schedule 4. If
additional time is necessary for capital improvements, the compliance date may be extended (if
approved by the state) by up to two years [40 CFR 141.713(c)], In addition, systems on Schedule
4 that are conducting Escherichia coli sampling prior to any Cryptosporidium sampling may
have their compliance schedule extended to October 2016. Systems should have already
completed their Round 1 source water sampling, determined if they are required to install
treatment, and selected their toolbox options. However, installation or implementation of the
options may still be in progress in some cases. Therefore, the following is a preliminary
summary of implementation issues reported to EPA regarding the microbial toolbox options.
During the November 15, 2012 meeting held by EPA the microbial toolbox was discussed, along
with two states' perspectives on toolbox options, and results from a water industry survey
(Cornwell et al., 2012). Input from state and water utility representatives included:
•	Some tools are too complex for systems to implement and states to review;
•	Some tools have capital cost or site issues that limit their use;
•	In some cases, states cannot use guidance documents for compliance purposes. The
Ultraviolet Disinfection Guidance Manual (UVDGM), in particular, may be too
complicated and difficult to use, and;
•	States may also benefit from access to a designated expert who can provide technical
assistance for each technology or tool.
The degree to which implementation issues have been identified varies by the option in the
microbial toolbox. Systems vary widely in their use of the different tools available in the
microbial toolbox. This document discusses each of the microbial toolbox options in order of
their appearance in the LT2. The discussion of the toolbox options is followed by a discussion of
ceramic membranes and other potential tools that are not part of the microbial toolbox.
7.1 Summary of Data on Toolbox Options and Treatment Credits
Exhibit 7.1 provides a summary of the toolbox options and the review findings. For each toolbox
option, the table indicates whether the literature reviewed provided new information on risk
reduction from Cryptosporidium, risk reduction from other pathogens of concern, and design and
implementation criteria. For each "yes" entry in this table, there is a summary of the literature
reviewed in the following subsections.
Six-Year Review 3
Technical Support Document for LT2
7-2
December 2016

-------
Exhibit 7.1 Summary of New Information on the LT2 Microbial Toolbox Options
Microbial Toolbox Option
Is There New Information Related to:
Risk Reduction
from
Cryptosporidium?
Risk Reduction
from Other
Pathogens of
Concern?
Design and
Implementation
Criteria?
Watershed Control Program
(WCP)
Yes
See Section 7.3.1
Yes
See Section 7.3.1
No
Alternative Source / Intake
Management
Yes
See Section 7.3.2
Yes
See Section 7.3.2
No
Presedimentation Basin with
Coagulation
No
No
No
Two-stage Lime Softening
No
No
No
Bank Filtration (BF)
Yes
See Section 7.3.5
Yes
See Section 7.3.5
Yes
See Section 7.3.5
Combined Filter Performance
No
No
No
Individual Filter Performance
No
No
No
Demonstration of Performance
(DOP)
Yes
See Section 7.3.8
No
Yes
See Section 7.3.8
Bag or Cartridge Filters
(individual filters)1
Yes
See Section 7.3.9
Yes
See Section 7.3.9
Yes
See Section 7.3.9
Bag or Cartridge Filters (in
series)1
Yes
See Section 7.3.9
Yes
See Section 7.3.9
Yes
See Section 7.3.9
Membrane Filtration
Yes
See Section 7.3.10
Yes
See Section 7.3.10
Yes
See Section 7.3.10
Second Stage Filtration
Yes
See Section 7.3.11
Yes
See Section 7.3.11
Yes
See Section 7.3.11
Slow Sand Filters
Yes
See Section 7.3.12
Yes
See Section 7.3.12
Yes
See Section 7.3.12
Chlorine Dioxide
No
Yes
See Section 7.3.13
Yes
See Section 7.3.13
Ozone
Yes
See Section 7.3.14
Yes
See Section 7.3.14
Yes
See Section 7.3.14
Ultraviolet (UV) Disinfection
Yes
See Section 7.3.15
Yes
See Section 7.3.15
Yes
See Section 7.3.15
Note:
1) EPA did not identify articles specific to bag or cartridge filters within a treatment plant as a post-treatment
process for obtaining additional removal credit under the LT2. The treatment units discussed in the literature
review were composite, point-of-use units, and are described in Section 7.3.9.
Six-Year Review 3
Technical Support Document for LT2
7-3
December 2016

-------
Exhibit 7.2 provides a summary of each of the LT2 microbial toolbox options and compares the
public health protection for Cryptosporidium relative to the protection each tool provides for
other organisms. For bacteria and viruses the data presented are for those species of potential
drinking water concern that are known to be the most resistant to the disinfectants. This
comparison can help provide regulators and the public with insight into the secondary benefits of
application of the LT2 toolbox tools beyond Cryptosporidium reduction. For several of the
toolbox options, the literature reviewed did not provide information to assess the relative
protection provided for other organisms compared to Cryptosporidium. This chapter provides a
discussion of the relative protection of each of the tools in the relevant section, where available.
Exhibit 7.2 Comparative Effectiveness of Tools for Different Organisms
Toolbox
Option
Cryptosporidium
(credit for
toolbox option)
Organism size
4 to 6 microns'1
Giardia
Organism size
7 to 14 microns1
Bacteria
Organism size
0.2 to 5 microns2
Viruses
Organism size
0.005 to 0.1
microns2
Source Toolbox Components
WCP
0.5-log
No information found indicating relative protection by
organism type.
Alternative
Source/ Intake
Management
Not prescribed
No information found indicating relative protection by
organism type.
Prefiltration Toolbox Components
Presedimentation
Basin
w/Coagulation
0.5-log
No information found indicating relative protection by
organism type.
Two-stage Lime
Softening
0.5-log
No information found indicating relative protection by
organism type.
BF
0.5 to 1 -log
No information found indicating relative protection by
organism type.
Treatment Performance Toolbox Components
Combined Filter
Performance
0.5-log
No information found indicating relative protection by
organism type.
Individual Filter
Performance
0.5-log
No information found indicating relative protection by
organism type.
DOP
State decision
Relative protection cannot be determined as
demonstration could be for a wide variety of treatment
technologies.
Additional Filtration Toolbox Options
Bag or Cartridge
Filters—individual
(Pore size down
to 0.2 jjm for
cartridge and 1
jjm for bag3)
<2-log
Similar
effectiveness as
Cryptosporidium
based on pore size.
Less effective than
for
Cryptosporidium.1
Less effective
than for
Cryptosporidium.
Not reliably
capable of 4-log
virus removal.1
Bag or Cartridge
Filters—series
<2.5-log
Similar to
effectiveness for
Cryptosporidium
based on pore size.
Less effective than
for
Cryptosporidium.1
Less effective
than for
Cryptosporidium.
Not reliably
capable of 4-log
virus removal.1
Six-Year Review 3
Technical Support Document for LT2
7-4
December 2016

-------
Toolbox
Option
Cryptosporidium
(credit for
toolbox option)
Organism size
4 to 6 microns'1
Giardia
Organism size
7 to 14 microns1
Bacteria
Organism size
0.2 to 5 microns2
Viruses
Organism size
0.005 to 0.1
microns2
Membrane
Filtration—
Reverse Osmosis
(Nominal pore
size 0.0001 jjm1)
Log removal credit
based on
demonstration
(Capable of >2-log
Cryptosporidium
removal1)
Capable of >3-log
Giardia removal.1
Capable of
removal of
bacteria, but not
absolute.1
Capable of >4-log
virus removal.1
Membrane
Filtration—
Nanofiltration
(Nominal pore
size 0.001 jjm1)
Log removal credit
based on
demonstration
(Capable of >2-log
Cryptosporidium
removal1)
Capable of >3-log
Giardia removal.1
Capable of
removal of
bacteria, but not
absolute.1
Capable of >4-log
virus removal.1
Membrane
Filtration—
Ultrafiltration
(Nominal pore
size 0.01 jjm1)
Log removal credit
based on
demonstration
(Capable of >2-log
Cryptosporidium
removal1)
Capable of >3-log
Giardia removal.1
Capable of at least
6-log removal of
all known
bacteria.1
Capable of >4-log
virus removal.1
Direct integrity
testing infeasible
(USEPA, 2005e).
Membrane
Filtration—
Microfiltration
(Nominal pore
size 0.1 jjm1)
Log removal credit
based on
demonstration
(Capable of >2-log
Cryptosporidium
removal1)
Capable of >3-log
Giardia removal.1
Capable of partial
bacterial removal.1
Less effective
than for
Cryptosporidium.
Not reliably
capable of 4-log
virus removal.1
Second Stage
Filtration—
Granular
Activated Carbon
(GAC)
0.5-log
Capable of removal
(2.1-log per Hijnen
etal., 2011a).
Not reliably
capable of
bacteria removal
(per Hijnen et al.,
2011a).
Not reliably
capable of 4-log
virus removal
(per Hijnen et al.,
2011a).
Second Stage
Filtration—Sand,
dual media
0.5-log
No information found indicating relative protection by
organism type.
Slow Sand Filters
2.5 to 3.0-log
Similar to
effectiveness for
Cryptosporidium.1
Less effective than
for
Cryptosporidium.1
Better than 99.4%
removal of total
conforms (Cleasby
et al., 1984) and
85-99% removal
of conforms
(Bellamy et al.,
1985)
Less effective
than for
Cryptosporidium.
Not reliably
capable of 4-log
virus removal
(Per Hijnen et al.,
2011b; 2011c;
2011d).
Six-Year Review 3
Technical Support Document for LT2
7-5
December 2016

-------
Toolbox
Option
Cryptosporidium
(credit for
toolbox option)
Organism size
4 to 6 microns'1
Giardia
Organism size
7 to 14 microns1
Bacteria
Organism size
0.2 to 5 microns2
Viruses
Organism size
0.005 to 0.1
microns2
Inactivation Toolbox Components4
Chlorine Dioxide
2-log with CT of
858 mg-min/L at
5°C
More effective than
for
Cryptosporidium.
3-log with CT of
26.0 mg-min/L at
5°C
More effective
than for
Cryptosporidium.
2-log with CT of
0.4-0.75 mg-min/L
at 5°C 5
More effective for
Hepatitis A virus
than for
Cryptosporidium.1
4-log with CT of
33.4 mg-min/L at
5°C
Ozone
2-log with CT of
32 mg-min/L at
5°C
More effective than
for
Cryptosporidium.
3-log with CT of 1.9
mg-min/L at 5°C
More effective
than for
Cryptosporidium.
2-log with CT of
0.02 mg-min/L at
5°C 5
More effective for
Polio virus than
for
Cryptosporidium.
8
4-log with CT of
1.2 mg-min/L at
5°C
UV Disinfection
2-log at 5.8
mJ/cm2
(4-log at 22
mJ/cm2)
Similar to
effectiveness for
Cryptosporidium.
3-log at 11 mJ/cm2
(4-log at 22
mJ/cm2)
Similar to
effectiveness for
Cryptosporidium
4-log at 20 mJ/cm2
6
Less effective for
Adenovirus than
for
Cryptosporidium
9
4-log at 186
mJ/cm2
Notes:
1)	Source: American Waterworks Association (AWWA), 1999.
2)	Source: Vance, 2002.
3)	Source: Filtra Systems, 2009.
4)	Inactivation tool effectiveness compared based on CT requirements and CT information in literature (CT is
the product of disinfectant concentration and contact time). Sources: Hoff, 1986; USEPA, 1991; WHO 2004;
USEPA. 2006c.
5)	Source: Hoff, 1986.
6)	Source: LeChevallier and Au, 2004.
7)	CT values for virus inactivation by chlorine dioxide based on Hepatitis A virus. Source: USEPA 1991
8)	CT values for virus inactivation by ozone based on polio virus with a factor of safety of 3 applied and
additional factors of safety based on temperature. Source: USEPA 1991.
9)	CT values for virus inactivation based on adenovirus, considered the most UV-resistant virus of concern in
drinking water. Source: USEPA, 2006c.
For toolbox options categorized as source toolbox components (i.e., WCP and alternative
source/intake management), EPA found no literature that provided a comparison of the treatment
capability for Cryptosporidium to that for other organisms of concern.
For toolbox options categorized as prefiltration toolbox components, treatment performance
toolbox components and additional filtration toolbox options:
• For some technologies, EPA found no literature that provided a comparison of the
treatment capability for Cryptosporidium to that for other organisms of concern;
Six-Year Review 3
Technical Support Document for LT2
7-6
December 2016

-------
•	Where possible, each toolbox option was rated as more effective than, less effective than
or of similar effectiveness to that for Cryptosporidium. Effectiveness was based on the
technology's ability to reliably remove the organism of concern to the log reduction
required, and;
•	To facilitate comparison of treatment capability, organism sizes are noted at the top of the
table, and nominal pore sizes have been provided where applicable (bag, cartridge and
membranes) in the "Toolbox Option" column.
For inactivation toolbox options:
•	Where possible, EPA rated each toolbox option as more effective than, less effective than
or of similar effectiveness to that for Cryptosporidium. For the inactivation tools, EPA
based this rating on the CT or dose required to achieve required inactivation, and;
•	To facilitate a comparison of treatment capability for disinfectants, EPA provides the CT
or UV dose necessary for required log-inactivation based on the Surface Water Treatment
Rule (SWTR) and the LT2 in Exhibit 7.2.
7.2 Treatment Technology Usage
EPA collected information on the toolbox tools used by 96 utilities and presented it at the public
meeting on November 15, 2012 (Finn, 2012). EPA collected the information to better understand
which tools were presenting challenges to utilities. Exhibit 7.3 provides the percentage of those
96 utilities that use each tool. (See also Appendix C for updated toolbox option information
collected since 2012.) Combined filter performance and individual filter performance were the
most-used tools, with 37.5 percent and 34.4 percent of systems using those tools, respectively.
UV disinfection was used by 19.8 percent of the 96 utilities. EPA found no information on the
use of two-stage lime softening and slow sand filters. McTigue and Cornwell (2013) found that
approximately one-third of 24 utilities responding to a survey they conducted used UV
disinfection to achieve compliance, while five used either the combined filter performance or
individual filter performance tool. They also found that three plants each used membranes, BF
and a WCP, the latter being used in combination with another tool. Five utilities also used DOP.
The authors cited the small footprint required, ease in retrofitting relative to other toolbox
options, and the ability to use UV at most treatment facilities as reasons for the preferred use of
UV disinfection. Cross and Bunton (2012) also reported that UV disinfection has been the tool of
choice for the Bin 2 systems in Iowa. The ability to fit the technology into the existing treatment
facility footprint was cited as a benefit of the tool, as were the relatively inexpensive cost of the
treatment and the additional safety factor obtained. Considering the frequency with which these
tools are used, implementation issues associated with UV, individual filter performance and
combined filter performance have the potential for a large impact on system implementation
nationally.
Six-Year Review 3
Technical Support Document for LT2
7-7
December 2016

-------
Exhibit 7.3 Microbial Toolbox Tool Usage
Please cite presentation from public meeting
Toolbox Option
Percentage of Systems Using the Tool1
Watershed Control Program
10.4%
Alternative Intake/Source Management
3.1%
Presedimentation Basin with Coagulation
2.1%
Two-Stage Lime Softening
No information available
Bank Filtration
3.1%
Combined Filter Performance
37.5%
Individual Filter Performance
34.4%
Demonstration of Performance
3.1%
Bag or Cartridge Filters
1.0%
Membrane Filtration
15.6%
Second Stage Filtration
1.0%
Slow Sand Filters
No information available
Chlorine Dioxide
1.0%
Ozone
2.1%
UV
19.8%
Note:
1) Percentage of 96 PWSs using specific tools based on information obtained from the EPA regions and
states. Some PWS reports indicate they plan to use a particular tool or that they use a tool, but it is unclear
whether they claim credit for the LT2 compliance purposes. Some may also use more than one tool.
7.3 Microbial Toolbox Tools
For each of the toolbox tools, this section presents an overview of the LT2 requirements, key
background information on the tool from the LT2, new information on each tool, and
implementation issues associated with the tool (if any). This section presents selected case
studies of water systems for implementing the following toolbox options: WCP, BF, combined
and individual filter performance, DOP, membrane filtration, chlorine dioxide and UV
disinfection.
7.3.1 Watershed Control Program
7.3.1.1 Overview of the LT2 Watershed Control Program Requirements
According to the Surface Water Treatment Rule Guidance Manual, a WCP is 'a surveillance and
monitoring program which is conducted to protect the quality of a surface water source'.
(USEPA, 1991). Watershed control, also referred to as catchment control in much of the
literature published outside the United States, is the first barrier used by many drinking water
systems in a multiple-barrier approach to risk management (Cinque and Jayasuriya, 2010).
Filtered PWSs using surface water sources can obtain a 0.5-log credit for Cryptosporidium
reduction by developing and implementing a state-approved WCP [40 CFR 141.716(a)], Systems
with existing source water protection efforts that were in place on or before January 5, 2006, can
seek credit for this toolbox option as long as the program meets the requirements discussed in the
LT2 [40 CFR 141.716(a)(3)], Unfiltered systems are not eligible for this toolbox option.
Six-Year Review 3
Technical Support Document for LT2
7-8
December 2016

-------
The WCP plan elements and demonstration include identification of potential Cryptosporidium
sources, prioritization of the identified sources, development of control measures to address the
prioritized sources, and continuation of these efforts in the future. To maintain 0.5-log reduction
credit, the system is required to submit an annual program status report to the state and conduct
watershed sanitary surveys every 3 years for community water systems and every five years for
noncommunity water systems.
7.3.1.2	Key Watershed Control Program Information from the LT2
The LT2 assigns a 0.5-log credit for state-approved WCPs consisting of required elements that
include: an area of influence of Cryptosporidium or fecal contamination affecting the treatment
plant intake; identification of potential and actual sources of Cryptosporidium and an assessment
of the relative impact of the sources on the system's source water quality; an analysis of the
effectiveness and feasibility of control measures that could reduce Cryptosporidium loading to
the system's source water; and goals and specific actions to be undertaken to reduce oocyst
concentrations by the watershed partners and their roles in the plan [40 CFR 141.716(a)], In
addition, the system must provide an annual program status report to the state and perform
regular watershed sanitary surveys at a frequency specified in the LT2. Unfiltered PWSs are not
eligible for this credit; they must maintain a WCP that minimizes the potential for contamination
by Cryptosporidium under 40 CFR 141.71.
The MDBP FAC recommended 0.5-log Cryptosporidium treatment credit for a WCP.
7.3.1.3	New Watershed Control Program Information
There was no new literature publicly available for review that specifically addressed the amount
of removal achieved by a WCP; however, the literature that was reviewed provided resources for
managing and estimating water quality conditions in watersheds. This section discusses current
literature addressing WCPs as an existing toolbox option.
A study conducted by Phillip et al. (2008) in the Republic of Trinidad & Tobago investigated the
relative importance of three main environmental categories of sources of Cryptosporidium in
watersheds: urban, agriculture and wildlife. The results of the study were used to assist in
prioritizing efforts to manage Cryptosporidium contamination of drinking water supplies. The
study included 19 sampling sites within three watersheds; the authors analyzed 243 raw water
samples for Cryptosporidium. The results of the study indicated that urban and wildlife are the
two most important sources of Cryptosporidium in the study area. The contribution of
Cryptosporidium from agricultural sources appeared to be minor (Philip et al., 2008).
A study conducted by Signor et al. (2005) in the Adelaide Hills of southern Australia evaluated
the use of event mean concentration (EMC) to quantify the effect of runoff events on water
quality in watersheds. EMC is a method that uses the flow-weighted average concentration of a
contaminant over the duration of a single runoff event to describe water quality during the runoff
event. The concept of EMC has been used extensively in the engineering of urban stormwater to
evaluate and quantify the effect of a high flow event on contaminant and pollutant loadings.
Signor et al. (2005) evaluated the EMC concept using flow and water quality data during several
runoff events in an urban watershed. The study found that within the watershed, contaminant
Six-Year Review 3
Technical Support Document for LT2
7-9
December 2016

-------
concentrations fluctuated over the duration of runoff events, indicating that the use of single
samples to estimate water quality during events may not be an adequate approach. Because the
EMC is based on the flow-weighted average concentration of contaminants, it takes into account
the fluctuation of microbiological concentrations. Signor et al. indicated that the results of the
EMC provided a good step to more detailed modeling of the effect of a runoff event on
contaminant concentrations in a watershed. EMC has the potential to support the WCP tool from
the microbial toolbox. However, since it is a diagnostic tool, rather than a tool that can result in
changes to source water concentrations, it would not be useful as a stand-alone tool in the
microbial toolbox.
Cinque and Jayasuriya (2010) conducted a study of the West Tarago River watershed located
east of Melbourne, Australia, to evaluate water quality data associated with watershed processes
such as surface runoff and erosion. The study involved two techniques, factor analysis (FA) and
EMC, to determine which processes within a watershed have the greatest impact on water
quality. FA is a multivariate statistic typically used to evaluate spatial and temporal changes in
water quality and to determine trends. In this study, the researchers used FA to interpret water
quality data and relate it back to watershed processes. They compared EMCs to baseline
microbiological concentrations to evaluate the effect of rainfall on contaminant concentrations.
FA was successful in identifying the most significant processes within the watershed: surface
runoff and erosion. The EMC evaluation indicated that the highest risk to water quality is during
rainfall, when the concentrations of pollutants increase significantly. The study found that total
coliform levels were not related to surface runoff in this watershed, and E. coli was not strongly
related to erosion. Therefore, neither total coliforms nor E. coli were adequate indicators of
pathogens for this watershed. Cinque and Jayasuriya (2010) recommended that Clostridium
perfringens and Enterococcus always be sampled, as they appeared to be related to erosion and
were better indicators of pathogens. The results of the study indicated that it is important for
systems to monitor watersheds during as many rainfall events as possible in order to gain
knowledge of the watershed processes and associated water quality. Like EMCs, FA will not
remove contaminants from drinking water or prevent their occurrence, but it may be useful for
some systems as a diagnostic tool in support of WCPs within the LT2 microbial toolbox.
Ferguson et al. (2007) developed a process-based mathematical model or pathogen catchment
budget (PCB) to predict Cryptosporidium, Giardia and E. coli loads within watersheds. The
model evaluates the processes that affect the generation and transport of microorganisms from
humans and animals using land use and flow data, and watershed-specific information including
point sources such as sewage treatment plants. The authors discussed how the PCB results can be
used to determine watersheds' pathogen loading and determine the watershed(s) with the highest
potential to deliver pathogens to the reservoirs. Systems can then prioritize and implement
control measures for the reduction of pathogen risks to drinking water. The results of the PCB
can be used as inputs to a hydrodynamic model developed by Hipsey et al. (2005).
Six-Year Review 3
Technical Support Document for LT2
7-10
December 2016

-------
7.3.1.4 Implementation Issues Associated with Watershed Control Programs
Few systems have chosen the WCP option. Some reasons for this may be the tight time schedule
and time-consuming efforts needed for development and implementation, difficulty meeting the
stringent elements required for the WCP (e.g., identifying potential Cryptosporidium sources),
and the relatively low 0.5-log removal credit assigned to this option.
Regarding the timeframe for development and implementation, in order to receive credit, a
system must have a WCP plan approved and in place within 3 years after the Cryptosporidium
sampling and bin assignments are complete. All systems choosing this option are required to
notify their state two years or more before their treatment compliance date. Depending on system
size, proposed WCPs were due to states no later than one year prior to their treatment
compliance date (April 2011, October 2011, October 2012 or October 2013).
Cross and Bunton (2012) reported that no systems in Iowa requested this option and credited the
lack of interest to difficulties addressing all of the required elements in a large watershed, the
large number of point sources of Cryptosporidium in agricultural areas, problems implementing
control strategies on private property, the need for ongoing oversight, and the labor-intensive
requirements for achieving 0.5-log reduction. Cornwell et al. (2012) conveyed their survey
findings that many states did not embrace this option and indicated they would not approve it,
and many systems did not consider implementing the option due to the challenges of
uncontrolled watersheds.
There are many potential sources of Cryptosporidium in watersheds, including wastewater
treatment plant discharges and nonpoint sources associated with animal feces. Although few
systems are using the WCP toolbox option, some are still implementing or considering source
water protection efforts, but have indicated a primary challenge and concern is with the control
of human activity within the watershed, particularly eliminating unauthorized entry (Tacoma
Public Utilities (TPU), 2010; City of Bend, 2012).
Another implementation issue raised was the need to address potential impacts of natural
disasters, such as fires, on source water quality and Cryptosporidium concentrations. Wildfires
have the potential to significantly affect physical and chemical characteristics of soils and
surface waters that can, in turn, negatively affect drinking water utilities. Recent reports on
wildfires have emphasized the effect of water quality parameters on water treatability but have
not directly investigated Cryptosporidium concentrations. Source water quality may be degraded
after a wildfire due to erosion, which can increase turbidity, or the addition of organic carbon and
nutrients. Emelko et al. (2011) found that streams in burned watersheds had 95th percentile
turbidity values of 15.3 nephelometric turbidity units (NTU) and total suspended solids (TSS) of
4.6 milligrams/liter (mg/L) as compared to streams in unburned watersheds that had 95th
percentile turbidity values of 5.1 NTU and TSS of 3.8 mg/L. A study conducted in Australia
(Smith et al., 2011) reviewed the impacts of wildfires on water quality within reservoirs. Based
on a turbidity guideline value of 1 NTU, the findings of the study indicated an increase in
turbidity after a wildfire of 26,000-fold, 600-fold and 280-fold for three reservoirs. Following a
wildfire, there is an enhanced potential for Cryptosporidium oocysts to be transported with
eroded sediment. The implications of fires in watersheds to the LT2 implementation would be
Six-Year Review 3
Technical Support Document for LT2
7-11
December 2016

-------
site-specific and depend on the location and severity of the fire as well as the consequences of
precipitation runoff events.
7.3.2 Alternative Source/Intake Management
7.3.2.1	Overview of the LT2 Alternative Source/Intake Management Requirements
Under the alternative source/intake option, systems may conduct their Cryptosporidium source
water monitoring at an alternative intake (in the current source or in an alternative source). Or
they may monitor using a different procedure for the timing or level of withdrawal from the same
source. (Systems must still conduct source water monitoring at their current intake using current
procedures.) Water systems can use the results of such alternative monitoring to determine their
bin classification. With state approval, the system may choose the source, intake location or
intake procedure it will use based on bin classification results [40 CFR 141.716(b)], This section
discusses literature addressing management of alternative sources and intakes relevant to the
LT2. EPA found no publicly available literature on systems installing a new intake or having
multiple intakes to be placed in a lower bin; however, the literature that was reviewed provided
resources and discussed intake management in general and potential risk of pathogen exposure to
consumers.
7.3.2.2	Key Alternative Source/Intake Management Information from the LT2
The LT2 does not assign a prescribed Cryptosporidium removal credit for this option. PWSs may
conduct simultaneous monitoring for treatment bin classification at alternative intake locations or
under alternative intake management strategies if approved by the state [40 CFR 141.716(b)],
The MDBP FAC recommended that PWSs be allowed to modify their plant intakes to comply
with the LT2.
7.3.2.3	New Alternative Source/Intake Management Information
A technical conference presentation by Ndong et al. (2011) discussed a study of Missisquoi Bay
on Lake Champlain in Quebec, Canada, that considered the vulnerability of a water treatment
plant (WTP) based on stratification and mixing within the supply reservoir, migration of
cyanobacterial blooms, and intake position and depth. The study results indicated that weather
conditions, particularly wind direction, influence migration of cyanobacteria. The authors
suggested the development of a hydrodynamic model that considers parameters such as wind
strength, predominant wind direction, diurnal stratification and light penetration in the water as
an early warning tool to avoid operational issues associated with cyanobacterial blooms. In a
related presentation, De Boutray (2011) presented a study that continued to evaluate the
vulnerability of the same WTP to cyanobacteria and the processes that affect the spatial and
temporal distribution and abundance of cyanobacteria. De Boutray (2011) used a hydrodynamic
model (DYRESM-CAEDYM) that considered weather conditions, reservoir temperature and
stratification to estimate the distribution of cyanobacteria and determine a suitable location and
depth for a new intake for the WTP.
Astrom et al. (2007) evaluated the use of E. coli as an indicator of "pathogen-rich" water at the
river intake for the Alelyckan Water Treatment Plant in Sweden. The Alelyckan Water
Six-Year Review 3
Technical Support Document for LT2
7-12
December 2016

-------
Treatment Plant in the city of Goteborg is supplied by the river Gota alv, which also supplies a
lake reservoir (Delsjon), the source water for the city's second WTP. Before the study, the water
system used E. coli data to determine when the system should not use the river intake for the
Alelyckan Water Treatment Plant so as to limit the impact of contaminated water from
wastewater sources on the WTP. The closing of the intake serves as a microbial barrier by
protecting the WTP from increased levels of pathogens. The study found that using E. coli and
establishing a threshold level for E. coli for intake closure was not as effective as receiving
notices about microbial discharges upstream of the intake.
Astrom et al. (2009) evaluated pathogen loading on a river during dry and wet weather
conditions. The study estimated theoretical concentrations of pathogens (E. coli, spores of
Clostridia, somatic coliphages, norovirus, Giardia and Cryptosporidium) in the river from
wastewater treatment plant effluents and sewer overflows using Monte Carlo simulations. They
compared the simulated concentrations to measured river water concentrations. The findings
indicated that the microbial load from all wastewater discharge points substantially increased
during wet weather conditions and that simulated pathogen concentrations compare well to
measured concentrations. The authors concluded that simulated concentrations can be used to
estimate pathogen density in raw water to model the risk to drinking water consumers. Water
systems may consider using such a tool to determine when to avoid using a specific intake,
relying more heavily on an alternative source if available, or making use of stored water during a
high pathogen event, if possible.
Brookes et al. (2004) presented a risk management framework for reservoirs that considers the
fate and transport of pathogens. The authors noted the need for a high level of knowledge and
understanding of the particular system being analyzed. For example, it is important to understand
how a reservoir's hydrodynamics affect pathogen transport and inactivation. Furthermore,
Brookes et al. emphasized the need to optimize sampling programs to gather the maximum
amount of information from a small number of samples. Antenucci et al. (2005) drew similar
conclusions after developing and testing models to help utilities analyze pathogen distribution in
supply reservoirs. Water systems may be able to use such a model to avoid using a specific
intake during a high pathogen event if alternative sources or adequate storage are options for
interim supplies.
7.3.2.4 Implementation Issues Associated with Alternative Source/Intake Management
State regulators from the states of Iowa and New Mexico expressed concerns with the timing and
the potential investment needed to incorporate this tool into the existing treatment processes.
(Cross and Bunton, 2012).
7.3.3 Presedimentation Basin with Coagulation
7.3.3.1 Overview of the LT2 Presedimentation Basin with Coagulation Requirements
Presedimentation is a preliminary treatment process used to remove gravel, sand and other
particulate material from the source water through settling before the water enters the treatment
plant. Presedimentation is not included as a process in the definitions of conventional or direct
filtration in 40 CFR 141.2. The LT2 assigns a 0.5-log credit to Cryptosporidium removal by
Six-Year Review 3
Technical Support Document for LT2
7-13
December 2016

-------
continuously operated presedimentation that treats all flow reaching the treatment plant,
continuously adds a coagulant to the presedimentation basin, and achieves either at least 0.5-log
mean reduction of influent turbidity or complies with state-approved performance criteria [40
CFR 141.717(a)],
7.3.3.2	Key Presedimentation Basin with Coagulation Information from the LT2
The MDBP FAC recommended that PWSs be allowed to achieve 0.5-log Cryptosporidium
treatment credit for presedimentation with coagulation. EPA reviewed published studies of
bench-, pilot- and full-scale processes on Cryptosporidium and aerobic spore removal (Payment
and Franco, 1993; Kelley et al., 1995; Patania et al., 1995; States et al., 1997; Edzwald and
Kelly, 1998; Dugan et al., 2001) in developing the LT2.
7.3.3.3	New Presedimentation Basin with Coagulation Information
EPA found no new information in the literature on this particular tool that would support a
potential change to the credits.
7.3.3.4	Implementation Issues Associated with Presedimentation Basin with Coagulation
Cross and Bunton (2012) stated that systems without existing presedimentation basins would
require capital and sufficient land area to build the unit process. They reported two systems in
Iowa with existing presedimentation basins that treat 100 percent of the flow but do not add a
coagulant prior to their basins. In order for those systems to implement this tool, they would need
to install coagulant feed and sludge removal systems (Cross and Bunton, 2012). McTigue and
Cornwell (2013) reported three reasons why utility personnel did not choose this option for LT2
compliance: availability of land, reluctance to add a coagulant to their pre-sedimentation basin,
and discomfort with the requirement of continuous basin operation.
Cornwell et al. (2012) reported that there was confusion regarding systems that already add a
coagulant ahead of their presedimentation basins and whether they could get credit for their
existing processes.
7.3.4 Two-stage Lime Softening
7.3.4.1 Overview of the LT2 Two-stage Lime Softening Requirements
Lime softening is the use of lime and other chemicals for chemical precipitation to reduce
hardness and enhance clarification prior to filtration. The LT2 provides a 0.5-log credit for two-
stage lime softening where chemical addition and hardness precipitation occur in both stages [40
CFR 141.717(b)], To obtain this credit, all plant flow must pass through both stages. The LT2
credits single-stage softening as equivalent to conventional treatment, which receives 2-log
treatment credit. A DOP credit is also allowed under the LT2 if the system meets state-approved
criteria.
Six-Year Review 3
Technical Support Document for LT2
7-14
December 2016

-------
7.3.4.2	Key Two-stage Lime Softening Information from the LT2
In developing the LT2, EPA reviewed data by Logsdon et al. (1994) and data collected by PWSs
on removal of aerobic spores as an indicator of Cryptosporidium response to the treatment
processes. These studies indicated a lime softening plant could achieve greater than 0.5-log
Cryptosporidium removal during routine operations.
7.3.4.3	New Two-stage Lime Softening Information
EPA found no new information in the literature on this particular tool.
7.3.4.4	Implementation Issues Associated with Two-stage Lime Softening
Cross and Bunton (2012) reported that one system in Iowa utilizes lime softening but as a split-
treatment configuration. The presenters added that in order to receive credit, plants that currently
employ single-stage lime softening would have to either acquire additional space and capital to
expand their process or reduce their treatment capacity to meet two-stage requirements. In
addition, the authors point out that two-stage lime softening may result in additional chemical
and sludge removal costs.
7.3.5 Bank Filtration
7.3.5.1	Overview of the LT2 Bank Filtration Requirements
BF is a surface water pretreatment process that uses the bed or bank of a surface water body and
the adjacent aquifer as a natural filter. BF induces or enhances natural surface water infiltration
and recovers the surface water from a subsurface collector.
Because of the difficulty of directly measuring Cryptosporidium removal (due to the relatively
low oocyst concentrations typically present in surface water and ground water), in developing the
LT2, EPA reviewed BF studies that measured the removal of surrogates such as aerobic and
anaerobic bacterial endospores (Havelaar et al., 1995; Rice et al., 1996; Pang et al., 2005; Arora
et al., 2000; Medema et al., 2000; Wang et al., 2001). In addition, EPA reviewed studies of the
transport of Cryptosporidium through soil materials in laboratory column studies (Harter et al.,
2000).
7.3.5.2	Key Bank Filtration Information from the LT2
Criteria specified in the LT2 allow from 0.5- to 1.0-log Cryptosporidium reduction credit. The
LT2 log credit varies based on setbacks from the surface water body. To earn the credit, the PWS
must meet additional requirements regarding the type of collector, certain aquifer characteristics
including the percent of fines, and a maximum turbidity limit.
States may award more than 1.0-log credit based on a site-specific DOP, as described in the LT2
Toolbox Guidance Manual (USEPA, 2010c). The LT2 establishes criteria for such a study and
requires collection of data on the removal of Cryptosporidium or a surrogate, including related
hydrogeologic and water quality parameters for the full range of operating conditions. The study
also must include sampling from both the production well(s) and screened monitoring well(s)
Six-Year Review 3
Technical Support Document for LT2
7-15
December 2016

-------
located along the shortest flow-path. Guidance on site-specific BF studies provides analytical
methods for measuring aerobic and anaerobic bacterial spores, which may serve as surrogates for
Cryptosporidium removal.
The LT2 specifies the following removal credits based on site-specific conditions [40 CFR
141.717(c)],
•	0.5-log credit for a 25-foot setback.
•	1.0-log credit for 50-foot setback.
The credit applies to horizontal and vertical wells only, and the aquifer must be unconsolidated
sand containing at least 10 percent fines (as defined in the LT2). The average turbidity in wells
must be less than 1 NTU. PWSs using existing wells followed by filtration must monitor the well
effluent to determine bin classification and are not eligible for additional credit.
For several years prior to the development of the LT2 Toolbox Guidance Manual, primacy
agencies assessed water sources for designation as ground water under the direct influence
(GWUDI). They also assessed sources for pathogen removal credits through BF as alternative
filtration technologies under the Interim Enhanced Surface Water Treatment Rule (IESWTR) or
the Long Term 1 Enhanced Surface Water Treatment Rule (LT1).
7.3.5.3 New Bank Filtration Information
Farkas et al. (2015) found that systems practicing riverbank filtration saw a lower rate of
seroprevalence 21 and 23 percent to Cryptosporidium antigens than systems using surface water
sources (49-61 percent). This supports Cryptosporidium removal credits for BF.
Since the award of BF Cryptosporidium removal credit via a DOP is implemented via guidance
rather than regulation, states have flexibility to vary the procedure to better match local
conditions. Because these were addressed in guidance some challenges in implementation may
result. Some challenges had been identified with BF prior to the LT2. These challenges include:
•	Some systems may not collect sufficient data to address seasonal variability. EPA
guidance recommends collecting 18 months' worth of data. However, systems can collect
less data, since EPA cannot require 18 months under guidance. For example, the Kansas
City, Kansas, Board of Public Utilities conducted a 6-month spore study in 2005 to
demonstrate 2-log removal of Cryptosporidium to comply with the IESWTR (Berger,
2006). This may limit the understanding of the amount of removal achieved by systems.
•	Some systems may use predictive colloid filtration modeling to estimate removal
efficiencies of Cryptosporidium using BF. Faulkner et al. (2010) studied the colloid
attachment rates for three different sandy riverbank sediments. For filtration in high-
energy rivers, C. parvum removal was not found to be dependent upon the sand texture,
carbon content or bacterial colony forming units. In addition, straining was not an
important mechanism in Cryptosporidium removal.
Six-Year Review 3
Technical Support Document for LT2
7-16
December 2016

-------
•	Some systems may collect particle removal data based on particles that unambiguously
provide removal estimates. The recommendation in the LT2 Toolbox Guidance Manual is
for systems to collect data on total coliforms, total aerobic spores and diatoms from
microscopic particulate analysis (MPA). However, some systems collected data on only
some of these parameters, or on other parameters. Removal credit based, in part, on total
algae removal, MP A values (Abbaszadegan et al., 2011) or total particle count data may
not adequately illustrate the levels of removal a system achieves. The presence of green
algae in a ground water collector is an indicator of direct surface water influence in the
current MP A, and some green algae are similar in size to Giardia and Cryptosporidium.
However, Abbaszadegan et al. (2011) reported that green algae are not a good indicator
of surface water because the variability in the size and shape of algae species affects how
they are transported through an aquifer. Also, total particle counts differ depending on the
pumping well's status (van Beek et al., 2010).
•	Some systems have received removal credits for BF based on well location or pumping
rate. Assigning credits based on this information may not accurately reflect the removal
rates achieved by BF for these systems. The LT2 Toolbox Guidance Manual recommends
collecting bioparticle data to estimate credits (USEPA, 2010c).
•	Some systems have received removal credits for BF based on removal assumptions
achieved by combining estimates for several unit processes into one estimate. For
example, the Iowa Department of Natural Resources awarded Sioux City, Iowa, 4.0-log
removal credit for Giardia and 3.5-log removal credit for Cryptosporidium for the
combined treatment of the source water through BF and the WTP (Mach et al., 2003;
Abbaszadegan et al., 2011). The LT2 Toolbox Guidance Manual recommends that
removal be based strictly on BF performance, rather than on the combined unit processes
of the BF and treatment plant components (USEPA, 2010c).
At least three sites, Central Wyoming Regional Water System in Casper, Wyoming (CWRWS),
(Gollnitz et al., 2005), the City of Kennewick, Washington (Gollnitz et al., 2007) and the Greater
Cincinnati Water Works in Cincinnati, Ohio, conducted DOP studies consistent with the LT2
Toolbox Guidance Manual. (Cincinnati has both surface and ground water sources. Its DOP
study was undertaken on its ground water source as a precautionary measure.) At all three sites,
the DOP was undertaken using total aerobic spores as the surrogate bioparticle, consistent with
current guidance. In all cases, the studies appeared to appropriately demonstrate
Cryptosporidium removal, despite the absence of measurable Cryptosporidium.
7.3.5.4 Implementation Issues Associated with Bank Filtration
In some locations, BF provides sufficient Cryptosporidium reduction to be used as a toolbox
option for compliance. An implementation issue identified for the BF toolbox option is the
difficulty of demonstrating more than 2-log Cryptosporidium removal credit using BF.
Although the LT2 Toolbox Guidance Manual recommends diatom presence/absence in wells as a
validation check on log removal estimates based on aerobic spores, it is not clear whether states
or utilities have applied this recommendation. EPA guidance suggests that diatom presence is a
qualitative measure that log removal is inefficient but does not recommend any diatom count or
Six-Year Review 3
Technical Support Document for LT2
7-17
December 2016

-------
occurrence frequency (USEPA, 2010c). Diatom speciation may be particularly helpful because
only a few diatom taxa are small enough to be considered relevant in Giardia or
Cryptosporidium removal studies. Because an MPA does not typically report diatom species,
additional microscopy work would be necessary to improve the applicability of the MPA to BF
evaluations.
Among the sites most studied by EPA, the CWRWS was able to demonstrate only 2-1 og
reduction of Cryptosporidium (Gollnitz et al., 2005). At the CWRWS site, EPA estimated that
the typical aerobic spore concentration was 10,000 per liter in the river (1,000 per 100 milliliter
[mL]s) and 10 per liter (natural background) or above in the wells. Thus, the river spore
concentration and natural background occurrence limited the ability of CWRWS to demonstrate
more than 2-log removal.
7.3.6 Combined Filter Performance
7.3.6.1	Overview of the LT2 Combined Filter Performance Requirements
Combined filter effluent (CFE) is the water from the combination of two or more individual
water treatment filters run in parallel (after mixing). PWSs using conventional or direct filtration
systems can obtain an additional 0.5-log Cryptosporidium treatment credit for CFE with turbidity
less than or equal to 0.15 NTU in at least 95 percent of measurements each month [40 CFR
141.718(a)], based on sampling at least every four hours using approved methods. This is in
comparison to the existing requirement of not exceeding 0.3 NTU in 95 percent of the CFE
measurements each month and not exceeding 1 NTU in any single sample [40 CFR 141.173(a)
and 141.551 (a)-(b)].
7.3.6.2	Key Combined Filter Performance Information from the LT2
In developing the LT2, EPA estimated that PWSs complying with the existing rules would
typically operate with filter effluent turbidity between 0.1-0.2 NTU and would typically meet the
0.15 NTU standard through operations at less than 0.1 NTU. Studies by Patania et al. (1995),
Emelko et al. (1999) and Dugan et al. (2001) observed the average removal of Cryptosporidium
to be 0.5- to 1.2-log greater when filter effluent turbidity was less than 0.1 NTU than when
effluent turbidity was between 0.1-0.2 NTU.
7.3.6.3	New Combined Filter Performance Information
EPA found no new information in the literature on this particular tool.
7.3.6.4	Implementation Issues Associated with Combined Filter Performance
Cross and Bunton (2012) voiced concerns regarding the combined and individual filter
performance tools. Some concerns were related to data integrity, accuracy of the tools and
oversight needed to ensure the validity of the credit over time. Examples of data integrity issues
include the exclusion of brief turbidity spikes in reported data, relying on proper turbidimeter
calibrations, and relying on proper supervisory control and data acquisition (SCADA)
programming. Another potential issue is reporting. Systems may simply state in their monthly
operating report that the criteria are being met even if they aren't (Cornwell et al., 2012).
Six-Year Review 3
Technical Support Document for LT2
7-18
December 2016

-------
Cross and Bunton (2012) added that they had difficulty in documenting the basis for the log
removal credits. They stated that the State of Iowa is reluctant to award this credit because of this
and also because Iowa has adopted national Area-Wide Optimization Program (AWOP) goals
that have more stringent turbidity criteria than the requirements for this toolbox option.
The AWOP goals for CFE and individual filter effluent (IFE) are < 0.10 NTU in 95 percent of
the daily maximum readings collected at 1-minute intervals, with a maximum of < 0.30 NTU as
the maximum of all daily readings.
In comparison, the combined filter performance credit under the LT2 for 0.5-log removal
requires the CFE turbidity measurements taken for any month at each plant to be < 0.15 NTU in
at least 95 percent of the measurements. An additional 0.5-log Cryptosporidium removal credit
can be awarded for combined filter performance for any month if both of the following IFE
turbidity requirements are met.
1.	IFE turbidity must be less than 0.15 NTU in at least 95 percent of values recorded at each
filter in each month, excluding the 15-minute period following return to service from a
filter backwash.
2.	No individual filter may have a measured turbidity greater than 0.3 NTU in two
consecutive measurements taken 15 minutes apart.
Cornwell et al. (2012) reported that utilities in the Partnership for Safe Water actively sought the
combined and individual filter performance tools. After assessing their plant and filter
performances, some systems determined they needed changes in operations and improvements to
their existing plants, including upgraded monitoring equipment, SCADA and components of the
filters. Costs for the upgrades reportedly ranged from $20,000 for new turbidimeters to $4
million for full filter upgrades. Systems also continue to incur costs due to more frequent
backwashing and increased filter-to-waste intervals (Cornwell et al., 2012).
7.3.7 Individual Filter Performance
7.3.7.1	Overview of the LT2 Individual Filter Performance Requirements
IFE is the water exiting an individual water treatment filter. PWSs using conventional or direct
filtration systems can obtain an additional 0.5-log Cryptosporidium treatment credit for IFE
turbidity less than or equal to 0.15 NTU in at least 95 percent of measurements each month in
each filter, if the turbidity is never greater than 0.3 NTU in two consecutive measurements taken
15 minutes apart in any filter [40 CFR 141.718(b)], IFE turbidity monitoring must occur
continuously, and results must be recorded every 15 minutes. This credit is in addition to 0.5-log
combined filter performance credit, but is not required to obtain 0.5-log combined filter
performance credit.
7.3.7.2	Key Individual Filter Performance Information from the LT2
The LT2 individual filter performance credit reflects the goals of Phase IV of the Partnership for
Safe Water—a voluntary cooperative program involving PWSs, professional associations, and
Six-Year Review 3
Technical Support Document for LT2
7-19
December 2016

-------
federal and state regulatory agencies that seek to increase protection from microbial
contaminants by optimizing WTP performance.
7.3.7.3	New Individual Filter Performance Information
EPA found no new information in the literature on this particular tool.
7.3.7.4	Implementation Issues Associated with Individual Filter Performance
Issues related to combined filter performance discussed above are also applicable to the
individual filter performance tool.
7.3.8 Demonstration of Performance of Treatment Process(es)
7.3.8.1	Overview of the LT2 Demonstration of Performance Requirements
The purpose of the DOP toolbox option is to allow a system to demonstrate that a plant, or a unit
process within a plant, should receive a higher Cryptosporidium removal/inactivation credit than
presumptively awarded under the LT2. Specific DOP protocols may vary by state because they
are established and/or approved by the state. The EPA LT2 Toolbox Guidance Manual
recommends that to demonstrate a higher level of Cryptosporidium treatment and thereby receive
a higher treatment credit for compliance, systems conduct a site-specific study accounting for all
expected operating conditions. The state has the discretion to determine ongoing monitoring
and/or performance requirements to ensure conditions under which the DOP was awarded are
maintained during routine operations. The applicable section of the LT2 is 40 CFR 141.718(c).
DOP requires testing and demonstration of the site-specific treatment process, which could
include a description of maintenance activities or equipment failure, performance in response to
variations in flow rate and raw water quality, data on historic water quality, and a contingency
plan for achieving compliance.
7.3.8.2	Key Demonstration of Performance Information from the LT2
The LT2 allows Cryptosporidium treatment credit to be awarded to a unit process or treatment
train based on a demonstration to the state using a state-approved protocol [40 CFR 141.718(c)],
The option reflects a recommendation by the MDBP FAC that the LT2 allow site-specific testing
both to establish Cryptosporidium treatment credit above the prescribed credit for toolbox tools
and to demonstrate Cryptosporidium removal for technologies not listed in the microbial
toolbox.
7.3.8.3	New Demonstration of Performance Information
One study discussed demonstration results for a treatment plant (clarification and filtration);
another examined potential surrogates for Cryptosporidium for demonstration of oocyst removal.
Brown and Cornwell (2007) used naturally occurring aerobic spore-forming bacteria to
demonstrate a WTP's ability to remove Cryptosporidium. The authors proposed that
Cryptosporidium is more readily removed by treatment processes such as clarification and
Six-Year Review 3
Technical Support Document for LT2
7-20
December 2016

-------
filtration than are spores, making spores a conservative choice as a challenge organism. Also,
they reported that spores are ubiquitous in most surface water sources and are typically found at
levels high enough (10,000 spores/L needed) to test for 4-log removal.
Pang et al. (2012) evaluated coated microspheres as a potential surrogate for Cryptosporidium to
quantify oocyst removal by granular media filtration. The authors studied glycoprotein-coated
microspheres and two forms of biotin-coated microspheres (one contained an amine compound)
as test surrogates because they displayed similar properties to Cryptosporidium. They also
studied unmodified microspheres for comparison purposes. The results showed that for tests
using 0.78-mm sand, the coated microspheres typically achieved the same log removal
(approximately 3-log) as oocysts, compared to approximately 2-log removal for the unmodified
microspheres. For tests using 1.37-mm sand, the glycoprotein-coated microspheres and oocysts
achieved 2- to 2.3-log removal, compared to 1.5-log removal for the unmodified microspheres.
Of the three coated microspheres, glycoprotein-coated microspheres produced the most
comparable results to Cryptosporidium for removal via sand filtration.
McTigue and Cornwell (2013) presented results from a utility survey on lessons learned from the
use of the LT2 toolbox. The authors contacted utilities classified in Bin 2 and investigated their
use of the toolbox. Based on the systems that selected DOP, McTigue and Cornwell (2013)
reported successful use of aerobic spore-forming bacteria as a surrogate for Cryptosporidium at
three utilities, with a fourth utility using particle counts.
7.3.8.4 Implementation Issues Associated with Demonstration of Performance
As described in the LT2 Toolbox Guidance Manual, "where a system can demonstrate that a
plant, or a unit process within a plant, consistently achieves a Cryptosporidium treatment
efficiency greater than the presumptive credit specified in the LT2, the state may allow the
system to receive a higher Cryptosporidium treatment credit for compliance with the LT2." To
demonstrate the higher level of Cryptosporidium treatment, systems must conduct a site-specific
study using a protocol approved by the state.
DOP is site-specific, and completing the DOP may be beyond the capability of many systems.
Cross and Bunton (2012) reported this toolbox option requires extensive monitoring and a
continuing high level of management at the treatment plant, as well as extensive review by the
state. Cornwell et al. (2012) found that some utilities misread guidance and believed
Cryptosporidium was to be spiked for the demonstration study, while other utilities felt the DOP
criteria were too complex. They also reported some states were willing to review the DOP but
were not enthusiastic, and that systems had limited expertise and resources available to design
and implement the DOP. They also noted that guidance on the use of surrogates could be
clarified to remedy the impression that it is unlikely that the criteria could be met.
EPA attempted to locate state-developed protocols for DOP but found no information. Therefore,
it is unclear if states have developed DOP protocols, and if so, when the protocols were issued
and whether they are made available to the public. While EPA's LT2 Toolbox Guidance Manual
introduces basic DOP concepts for treatment achieved through physical removal, without state-
developed protocols issued in a timely manner, a system would not be aware of all the required
elements to include in the DOP or the length of time needed for the study (i.e., pilot test).
Six-Year Review 3
Technical Support Document for LT2
7-21
December 2016

-------
The schedule for LT2 compliance may have been a reason larger systems did not select this
option, since systems serving 100,000 persons or more were to have complied with the treatment
requirements by April 2012, and systems serving 50,000 to 99,999 persons by October 2012.
7.3.9 Bag or Cartridge Filters
7.3.9.1	Overview of the LT2 Bag or Cartridge Filters Requirements
Bag filters and cartridge filters use engineered porous filtration media with pressure to remove
particulate matter larger than 1 micrometer in size [40 CFR 141.2], The LT2 allows up to 2-log
Cryptosporidium treatment credit for individual bag or cartridge filters, and up to 2.5-log credit
when used in series [40 CFR 141.719(a)], To obtain these credits, filters must undergo challenge
testing to demonstrate removal efficiency with an applied safety factor. The challenge testing is
product-specific but not site-specific, and challenge testing criteria are specified in the LT2.
7.3.9.2	Key Bag or Cartridge Filters Information from the LT2
In developing the LT2, EPA reviewed several studies and concluded both bag or cartridge filters
exhibit variable removal efficiency, and no correlation between the pore size rating assigned by
the manufacturer and the removal efficiency of the filter was apparent.
7.3.9.3	New Bag or Cartridge Filters Information
Literature reviewed for bag or cartridge filters included one study that discussed performance
testing using a composite cartridge filter, and another study that examined Cryptosporidium and
E. coli removal, as well as surrogates for Cryptosporidium and viruses, using two types of point-
of-entry/point-of-use (POU) devices.
Muhammad et al. (2010) evaluated a composite cartridge filter that provides physical filtration as
well as adsorption and UV disinfection. The authors conducted tests at flows ranging from 11 to
15 gallon per minute (gpm) to challenge the device using physical, chemical and biological
constituents (turbidity, particle counts, B. subtilis spores, E. coli, MS2 bacteriophage,
polystyrene latex (PSL) beads, methyl tert-butyl ether), "super-chlorination," total
trihalomethanes (THM), haloacetic acids and diazinon.) With respect to the biological-related
results, the overall cartridge (including UV) showed potential for removal/inactivation of
turbidity, PSL beads and E. coli but did not perform adequately for B. subtilis spores and MS2
bacteriophage. The system's performance with the PSL beads depended on the condition of the
filter, showing a decline under dirty filter conditions. E. coli had an average log
removal/inactivation of 3.35 compared toB. subtilis spores with an average log
removal/inactivation of 1.75. Muhammad et al. (2010) concluded that the higher log reduction
for E. coli was due to their vegetative cells being more vulnerable to UV disinfection than the
spore form of B. subtilis. The authors noted the low log removal/inactivation of 1.1 for MS2
indicated the need for a higher UV dose.
Muhammad et al. (2008) also studied surrogates for Cryptosporidium removal in POU systems.
In this experiment, the challenge organisms tested included Cryptosporidium oocysts, B. subtilis
spores, PSL beads, E. coli and MS2 bacteriophage. The authors tested two filter systems at 0.5
gpm. The first system contained a pleated prefilter, an activated carbon filter, a cationic polymer,
Six-Year Review 3	7-22	December 2016
Technical Support Document for LT2

-------
a cationic silver complex, a thermoplastic binder and a pH-altering material. The second system
contained a molded carbon block filter, a binder and an optional component containing silver to
suppress bacterial growth. Both systems achieved greater than 5-log removal of all challenge
organisms with the exception of PSL beads which achieved 3.14- and 3.56-log removal for the
first and second system, respectively. The filters most efficiently removed Cryptosporidium,
achieving greater than 8.8-log removal. Since the authors determined that PSL beads were an
overly conservative surrogate, they concluded that B. subtilis was a more reasonable and
effective surrogate for removal of Cryptosporidium oocysts.
7.3.9.4 Implementation Issues Associated with Bag or Cartridge Filters
Cross and Bunton (2012) reported that no systems in the state of Iowa were using this tool to
meet compliance. The authors believed the challenge testing requirement was a deterrent for
systems in Iowa. However, this does not necessarily represent systems in other states.
7.3.10 Membrane Filtration
7.3.10.1	Overview of the LT2 Membrane Filtration Requirements
Membrane filtration is an engineered separation process that rejects particulate matter larger than
1 micrometer through size exclusion driven by pressure or a vacuum [40 CFR 141.2], Chapter 14
of the LT2 Toolbox Guidance Manual and EPA's Membrane Filtration Guidance Manual
(USEPA, 2005e) discuss a number of different types of membrane materials and module system
designs for different classes of membranes. These include microfiltration (MF), nanofiltration
(NF), ultrafiltration (UF) and reverse osmosis (RO) treatment processes. Since the publication of
the guidance, new technologies (e.g., ceramic membranes) have entered the market.
EPA's current guidance manuals introduce ceramic membranes as a possible membrane filtration
treatment material. In order for a system to receive removal credit for Cryptosporidium, the LT2
requires that a membrane filtration system meet the following three criteria:
1.	The process must comply with the definition of membrane filtration stipulated by the
LT2;
2.	The removal efficiency of a membrane filtration process must be established through a
product-specific challenge test and ongoing, site-specific direct integrity testing during
system operation, and;
3.	The membrane filtration system must undergo periodic direct integrity testing and
continuous indirect integrity monitoring during operation.
7.3.10.2	Key Membrane Filtration Information from the LT2
EPA based the criteria for awarding credit to membrane filtration processes on data
demonstrating the Cryptosporidium removal efficiency of membrane filtration processes, a
critical evaluation of available integrity monitoring techniques, and a study of state approaches to
regulating membrane filtration for pathogen removal, as summarized in the report Low Pressure
Six-Year Review 3
Technical Support Document for LT2
7-23
December 2016

-------
Membrane Filtration for Pathogen Removal: Application, Implementation, and Regulatory
Issues (USEPA, 2001b).
7.3.10.3 New Membrane Filtration Information
Polymeric Membranes
In addition to removing Cryptosporidium, polymeric membranes have the ability to remove other
contaminants. Systems using membranes to comply with the LT2 realize the secondary benefit of
this removal. Literature reviewed below addressed cyanobacteria and their associated toxins, f-
specific ribonucleic acid (RNA) bacteriophages, viruses and submicron bacterial pathogens.
Literature also discussed the use of a membrane integrity testing computer model.
Dixon et al. (2012) evaluated the performance of MF, UF and NF membranes used in
conjunction with coagulation and powdered activated carbon (PAC) for the treatment of
cyanobacteria and their associated toxins in order to achieve taste, odor and toxin control. The
study included bench, pilot and full-scale studies. The study concluded that integrated membrane
systems can be effectively used for the removal of cyanobacteria, and NF can be an efficient
treatment process for the removal of extracellular cyanobacterial toxin (e.g., microcystin-LA)
and can be effective as the final step in an integrated membrane system. Cyanobacteria cells
were completely removed using UF membranes alone and when used in conjunction with
coagulation; alum was the least effective coagulant for this purpose. The authors reported that
extracellular toxins were more effectively treated by PAC addition; however, coagulation
hindered this adsorption process in some instances.
Newcombe et al. (2009) also studied the ability of membrane systems to remove cyanobacteria
and their associated toxins. They acknowledged that, based on cell size (1 micron or larger) and
membrane pore sizes, NF and RO membranes could remove cyanobacterial cells. However, for
MF and UF membranes, removal may be affected by pore size variations between
manufacturers. Regarding dissolved cyanobacterial toxins released from damaged cells, the
researchers concluded NF and RO membranes are expected to remove these toxins while UF and
MF membranes are not. The authors noted that some removal has been observed for UF and MF
membranes but it was reported that it was most likely due to adsorption of the toxin to the
membrane surface.
Langlet et al. (2008) studied four genogroups of the f-specific RNA bacteriophages (MS2, GA,
QP and SP) to determine their effectiveness as viral surrogates for membrane filtration testing.
The concern with surrogates is their potential to aggregate under certain conditions resulting in
an overestimation of membrane removal efficiencies. The study tested removal of the phages at
varying pH (1.5 to 7.5) and electrolyte concentrations (1 mM and 100 mM). For GA and SP
phages, the authors observed aggregation for the full range of pH and ionic strength conditions.
They did not observe aggregation for QP, in a solution with a pH greater than 3 and a small ionic
strength (1 mM), but they did observe aggregation with a large ionic strength over the full range
of pH conditions. For MS2, the study observed no aggregation for either ionic strength condition
when the pH was greater than 4. Based on the results, the authors concluded that MS2 represents
a worst-case scenario and, therefore, would be a good viral surrogate for membrane filtration.
Six-Year Review 3
Technical Support Document for LT2
7-24
December 2016

-------
Jacangelo et al. (2006) investigated the development of a systematic performance testing
protocol and specification for MF and UF membranes with respect to the removal of viruses and
submicron bacterial pathogens in lab, bench and full-scale studies. The authors used MS2 and
PRDl bacteriophages as viral surrogates for membrane challenge testing. The study concluded
that MS2 was a better surrogate than PRDl because PRDl achieved higher log removal values
(LRVs). The study also showed that the removals observed were dependent on the type of
membrane and organism tested as well as water quality and operational conditions. The study
recommended that a membrane's ability to remove microorganisms should be determined by
testing, not based on the reported nominal pore size.
Brehant et al. (2008) developed a model for predicting microbial LRV to aid water system
operators when conducting integrity testing of membranes. The authors tested the model in both
bench- and full-scale UF membrane applications. Based on the theoretical equations, the model
complied with the LT2 direct integrity test requirements of a 3-|im or less resolution. However,
the 3 |im requirement could not be confirmed in the field due to experimental limitations (the
laser method could not produce holes smaller than 20 |im). The model proved to be highly
sensitive in that it could detect one defective membrane fiber out of more than 700,000 fibers,
guaranteeing at least 4-log removal efficiency. The authors concluded that the model can be used
as an automated tool for optimizing membrane fiber repair schedules while maintaining
treatment requirements.
Ferrer et al. (2015) examined substituting conventional pre-treatment consisting of dioxi-
chlorination, coagulation/flocculation, settling and sand filtration with direct UF. While UF was
proven to efficiently remove bacterial indicators, the removal of small viruses such as some
small bacteriophages and human viruses was lower than conventional pre-treatment.
Ceramic Membranes
Ceramic membranes are a type of artificial membrane made from inorganic materials such as
alumina, titania, zirconia oxides or some glassy materials. Pore size can vary but is typically 0.1
|im. Unlike traditional polymeric membranes, ceramic membranes have higher mechanical
strength and are more resistant to chemicals and high temperature. Such characteristics give
ceramic membranes a longer life and allow higher pressures, higher fluxes, higher recovery rates
and more vigorous backwashing (Freeman and Shorney-Darby, 2011; Amy and Ha, 2012).
Literature reviewed below discussed operational strategies for ceramic membranes and removal
efficiencies for E. coli, algae, THM precursors and some indicator organisms.
Gaulinger (2007) studied coagulation as a pretreatment step to reduce fouling of ceramic
membranes. He found that at low coagulant doses, coagulation time was very important and at
higher doses coagulant time was less critical. In addition, Gaulinger determined an optimum
coagulant dose that limited fouling and that increasing the pH reduced membrane fouling, but the
effect was small. Since ceramic membranes are more chemical-resistant, ozone treatment can be
applied prior to the membranes (Amy and Ha, 2012). Amy and Ha (2012), had preliminary
findings that showed that the use of ozone prior to the membranes can enhance membrane flux
and recovery by degradation of natural organic matter (NOM) and microorganisms.
Six-Year Review 3
Technical Support Document for LT2
7-25
December 2016

-------
Bottino et al. (2001) conducted a pilot-scale test to determine operational performance and
removal efficiencies of ceramic membranes for E. coll, coliforms and eight algal organisms
including Asterionella, Ceratium, Cyclotella, Melosira, Navicula, Nitzchia, Oscillatoria and
Synedra. The study did not provide LRVs but did discuss laboratory results that showed only
Cyclotella was detected in the permeate at approximately 1 percent of the original concentration.
None of the other organisms were detected.
Ciora and Liu (2003) discussed pilot-scale tests (over 1,000 hours in operation) using ceramic
membranes that yielded 50- to 70-percent THM precursor removal. They also reported a separate
laboratory study that showed greater than 3- to 4-log removal of MS2 bacteriophage-spiked
distilled water. The authors noted that higher virus removal efficiencies are expected in full-scale
plants since significant virus loading is found on micron- and submicron-sized suspended solids,
which are readily removed.
Rajagopalan (2001) conducted a pilot-scale test at a water system in Illinois to determine if
ceramic membranes with a nominal pore size of 0.2 microns could treat water with moderate to
poor water quality to meet drinking water standards. The study focused on testing operational
parameters (e.g., flux, pressure, recovery) and water quality, with heterotrophic bacteria using
the heterotrophic plate count (HPC) method, coliforms and Pseudomonas as microbial
indicators. Rajagopalan reported issues with membrane fouling but noted that the membranes
achieved 4.79-log removal of HPC, and Pseudomonas was not detected in the permeate. Results
were not reported for coliforms.
The California Department of Public Health awarded 4-log removal credit for Giardia and
Cryptosporidium to a ceramic membrane (Freeman and Shorney-Darby, 2011).
Contracts are in place for constructing two ceramic membrane facilities: a 2.5 mgd (millions of
gallons per day) facility in Watsonville, California, and a 10 mgd facility in Parker, Colorado.
7.3.10.4 Implementation Issues Associated with Membrane Filtration
In the feedback provided while EPA gathered information on microbial toolbox usage, EPA
Regions and states raised concerns regarding a lack of data and a consensus on how to use the
data in evaluating membrane applications.
A Texas water system is using an unspecified type of membrane filtration to meet additional
treatment requirements and has found that having to perform direct integrity testing on each unit
cost the plant five hours (30 minutes each on 10 units) of production time each day. This has
proven to be a potentially critical loss of production during summer months, as well as man-hour
costs.
Cross and Bunton (2012) stated systems are resistant to conducting direct integrity testing and
providing the data necessary to maintain the log removal credit. They reported that five systems
in Iowa employ membranes but none have sought credit for their processes.
Six-Year Review 3
Technical Support Document for LT2
7-26
December 2016

-------
7.3.11 Second Stage Filtration
7.3.11.1	Overview of the LT2 Second Stage Filtration Requirements
Second stage filtration is a separate second stage of granular media filtration that follows a first
stage of granular media filtration. For example, a granular media filter that follows a
conventional treatment or direct filtration plant would be considered second stage filtration. The
LT2 allows a 0.5-log Cryptosporidium treatment credit for a second separate granular media
filtration stage if the treatment train includes coagulation prior to the first filter, all of the water
flows through both filtration stages, and the state approves the treatment credit [40 CFR
141.719(c)],
7.3.11.2	Key Second Stage Filtration Information from the LT2
EPA believes secondary filters may remove Cryptosporidium that were destabilized but not
trapped in the primary filters or that were trapped but subsequently detached from the primary
filters prior to backwash. EPA believes studies that support Cryptosporidium removal by single-
stage granular media filtration also support second stage removal credit because the same
removal mechanisms are the operative factors (USEPA, 2006a). Data on removal of aerobic
spores through GAC filters following conventional treatment also support this option—those
data indicated GAC filters exceeded 0.5-log reduction (USEPA, 2006a).
7.3.11.3	New Second Stage Filtration Information
Hijnen et al. (201 la) determined whether GAC filtration following an in-line filtration treatment
system (coagulation and rapid sand filtration) was capable of removing viruses, bacteria, Giardia
and Cryptosporidium. The pilot-scale study examined fresh GAC media and "loaded" GAC
media (40,000 bed volumes). The test organisms included MS2 bacteriophage for viruses, E. coli
and spores of Clostridium bifermentans for bacteria, C. parvum oocysts and G. lamblia cysts.
The researchers inoculated a rapid sand filtration effluent stream with high concentrations of
organisms prior to the GAC filter unit. The results showed that GAC filtration is not effective at
removing viruses and has limited effectiveness for removing bacteria (less than 0.1- to 1.1-log
removal), but it is effective for the removal of oocysts and cysts. The LRVs for C. parvum were
2.7 and 1.2 for the fresh and loaded GAC filters, respectively. The LRV for G. lamblia was 2.1
for both fresh and loaded GAC filters. The authors cited another study by Patania, which
discussed lower LRVs for C. parvum and G. lamblia when the GAC filter was operated at higher
filtration rates of 12 to 18 meters/hour (Patania et. al., 1995).
7.3.11.4	Implementation Issues Associated with Second Stage Filtration
Cross and Bunton (2012) commented that to implement second stage filtration that can treat 100
percent of the plant flow may require a large capital investment. They reported that only one
system in Iowa had second stage filtration capability.
Six-Year Review 3
Technical Support Document for LT2
7-27
December 2016

-------
7.3.12 Slow Sand Filtration
7.3.12.1	Overview of the LT2 Slow Sand Filtration Requirements
Slow sand filtration is an engineering process that uses a bed of sand to filter raw water at a low
velocity [40 CFR 141.2], The LT2 prescribes a 2.5-log Cryptosporidium treatment credit when
systems use slow sand filtration as a secondary filtration stage following a primary filtration
process [40 CFR 141.719(d)], To obtain this credit, there must be no disinfectant residual in the
influent water to the slow sand filtration process, both filtration stages must treat 100 percent of
the treatment plant flow, and the state must approve the treatment credit based on an assessment
of the design characteristics of the filtration process.
7.3.12.2	Key Slow Sand Filtration Information from the LT2
EPA based this option on several studies demonstrating significant Cryptosporidium removal
using slow sand filtration (Schuler and Ghosh, 1991; Hall et al., 1994; Timms et al., 1995). Slow
sand filtration as a primary process receives a prescribed 3-log Cryptosporidium treatment credit.
The LT2 allows 2.5-log credit as a secondary stage due to the uncertainties regarding the
performance of slow sand as a secondary filtration step.
7.3.12.3	New Slow Sand Filtration Information
Several pilot- and lab-scale studies evaluated the performance of slow sand filters in removing
microorganisms and identified key design and operational parameters that influence the system's
removal efficiency.
Unger and Collins (2008) conducted a series of lab-scale sand column studies to understand
microbial removal mechanisms in slow sand filters. Although empty bed contact time and grain
size had some effect on removal of E. coli, the studies confirmed that E. coli removal occurs
primarily at the schmutzdecke, the thin layer on the top of the sand bed in which a dense
population of microorganisms develops. However, the authors pointed out that a deep bed will
help mitigate the reduction in microbial removal after filter scraping, when the schmutzdecke is
not fully developed. The study also showed that temperature can play a role in E. coli removal
with warm (24°C) biological columns outperforming colder (8°C) columns. In examining the
removal mechanisms in the schmutzdecke, Unger and Collins determined that biological activity
(as measured by respiration) correlated to E. coli removal, but extracellular polymeric substances
excreted by the biofilm (which enhance "stickiness" of the filter media) did not. They also
postulated that in addition to limited direct removal of E. coli, protistan predation may indirectly
contribute to E. coli removal by accelerating the development of the biofilm.
DeLoyde et al. (2006) also found that filter maturation (i.e., development of the schmutzdecke
layer) affected the removal of Cryptosporidium oocysts, but filter bed depth and temperature
were not important factors. Dullemont et al. (2006) conducted a series of pilot-scale studies to
assess the removal of microorganisms by slow sand filtration. Using MS2 bacteriophage as an
indicator of virus removal, the authors observed that the bacteriophage had the lowest removal
rate amongst the microorganisms tested and as a result was considered the most critical
microorganism for the assessment of slow sand filter performance. The removal of viruses and
Six-Year Review 3
Technical Support Document for LT2
7-28
December 2016

-------
bacteria was not dependent on seed concentrations and was much higher at higher temperatures.
They observed high removal rates of C. parvum in mature filters, which could be the result of
predation or degradation. Clostridium spore and centric diatom removals were less efficient than
oocyst removals. Thus, Dullemont et al. found Clostridium spores and centric diatoms are not
useful organisms for modeling the removal of oocysts by slow sand filtration.
DeLoyde et al. (2006) stated that numerous previous studies have shown greater than 4.0-log
reduction for Cryptosporidium. DeLoyde et al. (2006) conducted nine bench-scale challenges
and concluded that slow sand filtration provides significant removal of Cryptosporidium and
Giardia. Complete Giardia removal was observed in two of the nine challenges during the
course of this study.
Bichai et al. (2014) took experimental measurements from GAC and slow sand filtration tests
using high influent concentrations of Cryptosporidium (1.3 x 106 and 3.3 x 104 oocysts/L) and
Giardia (4.8 x 104 cysts/L). The results preliminarily indicated lower transportation and survival
ratios in slow sand filtration when compared to GAC filters. The researchers calculated the
probability of infection due to internalized (oo)cysts in the filtered water. Under the likeliest
environmental conditions, risks were found to fall below the tolerable risk target of 10"4
infections per person per year.
In a series of studies, Hijnen et al. (201 lb; 201 lc; 201 Id) assessed pathogen removal in slow
sand filters and considered the use of surrogate organisms. Hijnen et al. (201 Id) concluded that
the decimal elimination capacity (DEC), a concept analogous to log reduction, of slow sand
filters for MS2 bacteriophage was 1.5- to 2.0-log, making these organisms a good surrogate for
virus removal. Similarly, thermotolerant coliforms experienced 2- to 3-log removal, and were
considered conservative surrogates for pathogenic bacteria. However, the authors did not
consider DECs of 2- to 3-log for spores of sulfite-reducing Clostridia (SSRC) to be good
surrogates for Cryptosporidium or Giardia, which were rated at >5-log DEC. Furthermore,
Hijnen et al. (201 Id) found C. perfringens spores and Stephandiscus hantzschii to be too
conservative to be good surrogates for Cryptosporidium removal with the DEC for these
organisms of 3.6- and 1.8-log, respectively {Cryptosporidium removal at a DEC of 4.7-log).
Hijnen et al. (201 lb; 201 lc) also considered Clostridium spore survival in both water and in the
sand filter and found greater than 10-year survival time; and concluded that the longevity of this
organism makes it too conservative to be used as a process indicator for Cryptosporidium.
7.3.12.4 Implementation Issues Associated with Slow Sand Filtration
Cross and Bunton (2012) stated that there are two systems in the state of New Mexico that utilize
slow sand filtration, but neither system uses its slow sand process for compliance with the LT2.
Cross and Bunton also noted that Iowa has no systems that employ slow sand filtration.
Six-Year Review 3
Technical Support Document for LT2
7-29
December 2016

-------
7.3.13 Chlorine Dioxide
7.3.13.1	Overview of the LT2 Chlorine Dioxide Disinfection Requirements
Chlorine dioxide is a strong oxidant used in the disinfection of drinking water (AWW A, 2000).
Chlorine dioxide is an effective toolbox disinfection option that can assist with DBP control,
achieve CT credit, and, when used properly, result in low levels of chlorite.
To achieve Cryptosporidium treatment credit using chlorine dioxide, PWSs must measure the
water temperature, disinfectant contact time and residual disinfectant concentration at least once
each day and determine the log inactivation credit using CT tables provided in the LT2 [40 CFR
141.720], The state may also approve alternative CT values based on a site-specific study that
follows a state-approved protocol.
7.3.13.2	Key Chlorine Dioxide Disinfection Information from the LT2
EPA based the CT values for ozone and chlorine dioxide on analyses by Clark et al. (2002a,
2002b18) with additional procedures to assess confidence bounds.
7.3.13.3	New Chlorine Dioxide Disinfection Information
Gates et al. (2009) contained an update on the science of chlorine dioxide use in drinking water
treatment. The report included a detailed literature review of the scientific, technological and
operational approaches used internationally. Gates et al. compared European and North
American practices regarding chlorine dioxide production (including effectiveness and safety
procedures), applications and regulations. The authors included case histories and provided
insight into production technologies, process controls and residual monitoring. Gates et al.
concluded that in some cases, chlorine dioxide can be an effective and inexpensive drinking
water disinfection method, providing inactivation of microorganisms and an overall reduction in
DBP formation. According to Gates et al., no adverse effects to human health have been reported
when chlorine dioxide is generated and used properly.
Thurston-Enriquez et al. (2005) conducted bench-scale experiments using buffered, disinfectant
demand-free water under varying pH and temperature conditions to determine the effectiveness
of chlorine dioxide for the inactivation of enteric adenovirus type 40 (AD40) and feline
calicivirus (FCV). The authors observed that at a pH of 8 and temperature of 15°C, chlorine
dioxide achieved over 4-1 og virus inactivation within 15 seconds for both AD40 and FCV.
However, at a pH of 6 and temperature of 5°C, chlorine dioxide achieved 4-log inactivation of
AD40 after 3 minutes and only 3.6-log inactivation for FCV during the experiment.
Corona-Vasquez et al. (2002) studied the inactivation of C. parvum with chlorine dioxide as
primary and then chlorine as the secondary disinfectant. They did not find support for any added
benefit for this sequential approach as had been reported for ozone with chlorine. The authors
18 Note that this study was subsequently published in 2003.
Six-Year Review 3
Technical Support Document for LT2
7-30
December 2016

-------
conclude that the findings support the use of a simpler CT approach for Cryptosporidium
inactivation requirements with chlorine dioxide.
7.3.13.4 Implementation Issues Associated with Chlorine Dioxide Disinfection
Cross and Bunton (2012) reported that the major disadvantage of using chlorine dioxide as a tool
is the ongoing monitoring, sampling and analysis requirements. They stated that eight systems in
Iowa use chlorine dioxide for DBP control, not to comply with the LT2.
7.3.14 Ozone
7.3.14.1	Overview of the LT2 Ozone Disinfection Requirements
Ozone is a strong oxidant used in the disinfection of drinking water (AWW A, 2000). To achieve
Cryptosporidium treatment credit using ozone, PWSs must measure the water temperature,
disinfectant contact time and residual disinfectant concentration at least once each day and
determine the log inactivation credit using CT tables provided in the LT2 [40 CFR 141.720], The
state may also approve alternative CT values based on a site-specific study that follows a state
approved protocol.
7.3.14.2	Key Ozone Disinfection Information from the LT2
EPA based the CT values for ozone and chlorine dioxide on analyses by Clark et al. (2002a,
2002b19) with additional procedures to assess confidence bounds.
7.3.14.3	New Ozone Disinfection Information
Literature on ozonation showed that raw water quality may affect treatment efficiency and
bench-scale studies appear to overestimate ozone's ability to inactivate microorganisms. Full-
scale experimentation may better demonstrate ozone's efficiency at inactivating contaminants of
interest at different sites.
EPA examined several studies on the effectiveness of ozone for treating various pathogens.
Lanao et al. (2008) conducted bench-scale studies for treating C. perfringens using three
different advanced oxidation treatments: ozone, ozone/hydrogen peroxide and ozone/titanium
dioxide. C. perfringens was inactivated more rapidly by ozone/hydrogen peroxide than by ozone
and ozone/titanium dioxide, especially in the experiments conducted on water with low NOM
levels. Lanao et al. believed NOM competes with microorganisms for the consumption of
disinfectants.
Hijnen et al. (201 le) studied the use of SSRC as a surrogate for Cryptosporidium inactivation by
ozone. The study found that the inactivation rate constants for SSRC were on the same order of
magnitude as those for Cryptosporidium. Hijnen et al. concluded that SSRC is an appropriate
19 Note that this study was subsequently published in 2003.
Six-Year Review 3
Technical Support Document for LT2
7-31
December 2016

-------
tool to study ozone disinfection processes, but that further research was needed to determine
whether SSRC can be used as a surrogate for Cryptosporidium.
Pereira et al. (2008) studied chemical disinfectants for inactivation of Cryptosporidium. The
study found ozone to be an effective disinfectant for the inactivation of Cryptosporidium, with
100 percent inactivation reported at concentrations as low as 24 mg/L. Ozone performed better
than chlorine dioxide and hypochlorous acid under the experimental conditions. However,
Sivaganesan and Marinas (2005) suggest taking oocyst lot variability into account when
assessing CT requirements.
Vonder Haar et al. (2010) studied the inactivation kinetics of Coxsackie virus B5 by ozone at the
bench scale. The study revealed that neither pH nor temperature of the raw water measurably
affected the inactivation kinetics of the highly virulent virus. Ozone achieved approximately 99.9
percent inactivation of Coxsackie virus B5 at 0.003 min*mg/L at pH 7.4 and 14°C.
Smeets et al. (2006) studied the inactivation of E. coli by ozone under bench-scale and full-scale
hydraulic conditions and concluded that inactivation potential is overestimated by bench-scale
studies (based on the LT2 T10 calculations), possibly due to the poor mixing of ozone. Smeets et
al. recommended conducting full-scale ozonation studies before increasing the ozone dose.
Alvarez et al. (2010) reported that high ozone doses are typically needed to destroy microcystins
within cyanobacteria due to their cellular material. Alvarez et al. noted that, according to
previous studies, ozone was very effective in oxidizing microcystin-LR toxins at the lab-scale
and the only factor appearing to affect its performance was pH. They also mentioned that another
study noted total organic carbon (TOC) and dissolved organic carbon (DOC) can influence the
effects of ozone and can cause its rapid depletion (Hoeger et al., 2002). In a bench-scale study
using raw water from two Florida sites, Alvarez et al. (2010) aimed to determine the fundamental
parameters needed to design and operate an ozone system for the oxidation of algal toxins and
disinfection. Alvarez et al. concluded that, although several water quality parameters appeared to
influence the kinetics of the oxidation process, only pH and ozone dose had significant effects on
microcystin-LR in oxidation scenarios similar to those present in a WTP. They observed 100-
percent destruction of microcystin-LR during the study at ozone doses of 0.5 to 1.1 mg/L, with
the exact dose depending on the raw water quality. Ozone was more effective at pH levels lower
than 6. Doses as low as 0.4 mg/L achieved 97-percent oxidation of microcystin-LR under acidic
conditions. Alvarez et al. also concluded that ozone doses and contact times typically used for
disinfection could be adequate for the oxidation of microcystin-LR under normal treatment
conditions.
Rodriguez et al. (2007) stated that results of the effects of pH and DOC from previous studies on
cyanotoxin oxidation depend on the specific source water's oxidant consumption and thus cannot
be applied to other waters. Rodriguez et al. compared the oxidation of microcystin-LR,
cylindrospermopsin and anatoxin-a by multiple oxidants in the same source water. They used
static-dose testing and dynamic time-resolved experiments to study oxidation processes. Results
showed that ozone can effectively oxidize all three of the toxins studied.
Cheng et al. (2009) studied the oxidation of cylindrospermopsin and the inactivation of
Cylindrospermopsis raciborskii by ozone and other disinfectants. Cheng et al. reported that
Six-Year Review 3
Technical Support Document for LT2
7-32
December 2016

-------
exposure to 1 mg min/L provided complete oxidation of cylindrospermopsin and complete
inactivation of C. raciborskii.
7.3.14.4 Implementation Issues Associated with Ozone Disinfection
Cornwell et al. (2012) indicated that ozone has not been widely used by Bin 2 systems to comply
with additional treatment requirements, stating that the cost for the process is not competitive
(especially in cold climates), that bromate formation may preclude the use of ozone, and that the
practicality of calculating CT is limited if a state does not allow online residual monitoring.
Cross and Bunton (2012) reported that systems in Iowa and New Mexico using ozone have not
requested credit for LT2 compliance.
7.3.15 UV Disinfection
7.3.15.1	Overview of the LT2 UV Disinfection Requirements
UV disinfection is a process used to inactivate microbes through the use of UV light, resulting in
the disruption of the microbes' metabolic activities (AWW A, 2000). PWSs may use UV
disinfection to comply with Cryptosporidium treatment requirements in the LT2 and Giardia and
virus treatment requirements in other rules. The LT2 requires reactor validation testing to
establish the UV dose and associated operating conditions under which the reactor will deliver
the UV dose required for the desired log inactivation of the target organism. In order to receive
treatment credit for UV disinfection, systems must monitor these parameters, as well as any
additional parameters designated by the state. The system must operate such that at least 95
percent of the water delivered to the public during each month was produced under the validated
conditions for the required dose [40 CFR 141.720(d)(3)],
7.3.15.2	Key UV Disinfection Information from the LT2
Data used to develop the UV disinfection credits under the LT2 were based on UV light applied
at a wavelength of 254 nm, as delivered by a low pressure (LP) mercury vapor lamp. The LT2
allows other lamp types through validation testing, as described in the UVDGM (USEPA,
2006c). UV dose requirements in the LT2 range from 1.6 mJ/cm2 for 0.5-log credit for
Cryptosporidium to 186 mJ/cm2for 4.0-log inactivation of viruses [40 CFR 141.720(d)(1)],
In a study using LP UV lamps, and based on in vivo gerbil infectivity, Linden et al. (2002) found
greater than 4-log inactivation of G. lamblia cysts at low UV doses (10 J/m2). They also report
no light or dark repair of the irradiated G. lamblia cysts. The study concluded that G. lamblia
appeared to be substantially more sensitive to LP UV irradiation than other human pathogens
including E. coli, adenovirus 40 and C. parvum.
7.3.15.3	New UV Disinfection Information
This section presents an analysis of new studies on the inactivation of Cryptosporidium and other
pathogens by UV disinfection. EPA discusses UV disinfection more extensively in this section
because UV is one of the primary technologies water systems have used to comply with the LT2
Six-Year Review 3
Technical Support Document for LT2
7-33
December 2016

-------
and because the literature review identified numerous publications on UV disinfection that have
come out since the promulgation of the LT2.
Hubel (2007) suggested that redundant equipment, standby power, uninterruptible power supply
(UPS) devices, and active series compensators can improve electrical reliability and lead to less
downtime. Wright et al. (2009) also made suggestions for sensor requirements including
placement to improve reactor performance monitoring.
Inactivation of Pathogens by Low Pressure versus Medium Pressure UV Lamps
A concern raised in some recent literature is the high UV dose requirement (e.g., 186 mJ/cm2)
for 4-log inactivation of viruses in the LT2. EPA based the 186 mJ/cm2 minimum dose
requirement on studies performed with LP mercury vapor lamps on adenovirus, the most UV-
resistant pathogenic virus (USEPA, 2006a). Since then, several studies have compared
inactivation of adenovirus by monochromatic LP to inactivation by polychromatic medium
pressure (MP) UV light (Linden et al., 2007; Eischeid et al., 2009; Eischeid and Linden, 2009;
Linden et al., 2009; Shin et al., 2009a; Guo et al., 2010). These studies provide experimental
evidence that 4-log inactivation of adenovirus can be achieved at lower doses by MP lamps than
by LP lamps (Exhibit 7.4).
Other studies have addressed LP versus MP inactivation of bacteria and protozoa. Shin et al.
(2009b) found MP UV to be highly effective for Giardia inactivation. Zimmer-Thomas et al.
(2007) found that deoxyribonucleic acid (DNA) repair of E. coli was substantially less for MP
exposure compared to LP exposure (Exhibit 7.4). However, Sakai et al. (2011) found no
substantial difference between the LP and MP inactivation of Microcystis aeruginosa, and
Bohrerova and Linden (2006) found no statistical difference between LP and MP inactivation of
Mycobacterium terrae.
Six-Year Review 3
Technical Support Document for LT2
7-34
December 2016

-------
Inactivation of Cryptosporidium by UV
Clancy et al. (2004), noting that all previous studies on UV inactivation of C. parvum had been
on the same strain (Iowa strain), conducted a study on five separate strains to determine if similar
findings would be achieved with this expanded group of organisms. The study was based on
mouse infectivity and utilized LP monochromatic UV lamps. All five strains were found to be
highly susceptible to UV light with low doses (10 mJ/cm2) achieving at least 4-log inactivation.
Mahmud et al. (2006) conducted a study to determine if Cryptosporidium oocysts in
conventionally filtered drinking water will aggregate and, therefore, be more resistant to UV
disinfection. UV exposure was conducted using a collimated beam apparatus with an LP mercury
arc lamp. Results from these tests suggest that if Cryptosporidium oocysts are present in poorly
filtered drinking water, they are likely to be aggregated to some extent and will exhibit a greater
resistance to UV disinfection than oocysts seeded in clean water. The log inactivation at 5
mJ/cm2 was 1.6-log and 2.4-log for the poorly and well-filtered water, respectively. The
UVDGM recommends doses of 5.8 and 8.5 mJ/cm2 to achieve 2.0- and 2.5-log Cryptosporidium
inactivation credit, respectively (USEPA, 2006c).
Lee et al. (2008) demonstrated that pulsed UV (PUV) light is effective against Cryptosporidium,
but at higher laboratory doses than those required by LP and MP lamps. (PUV is not included in
the UVDGM.) The study found that maximum inactivation (4.9-log) was achieved when oocysts
were irradiated 20 cm from the source for 60 seconds at a PUV dose of 278 mJ/cm2 In addition,
at least 5 seconds of exposure at no more than 40 cm from PUV achieved 2-log reduction of
oocyst infectivity with a dose of 15 mJ/cm2 and a 3-log reduction with a dose of 23 mJ/cm2 It is
important to note that Lee et al. did not perform these experiments using a collimated beam
apparatus as recommended in the UVDGM, and thus, results should be viewed with caution.
Sivaganesan and Sivaganesan (2005) expanded upon the studies of Cryptosporidium inactivation
by investigating the effect of lot variability in determining the required UV radiation. The
authors conclude that "for 90 percent inactivation (or 2 log-inactivation) of C. parvum oocysts
with UV radiation, the minimum UV dose requirement is about 49 percent higher when a
simultaneous modeling with lot variability is used in the data analysis. The corresponding
minimum UV is about 75 percent higher for 99.9 percent (or 3 log-inactivation) inactivation of
C. parvum oocysts." These estimates were developed using a hierarchical Bayesian model with
posterior distributions developed using a Markov Chain Monte Carlo (MCMC) method.
In addition, Hijnen et al. (2006) published a review paper on inactivation credit of UV radiation
for viruses, bacteria and protozoans in water. Six of the studies reviewed indicated that 3.0-log
inactivation of Cryptosporidium could be achieved with MP UV doses in the range of 0.5 to 6.1
mJ/cm2; while four other studies indicated 3.0-log inactivation could be achieved with LP UV
doses in the range of 0.9 to 13.1 mJ/cm2 The LT2 dose requirement for 3-log inactivation of
Cryptosporidium is 12 mJ/cm2
Inactivation of Other Pathogens by UV
This section provides a summary of new information related to UV inactivation of pathogens
other than Cryptosporidium.
Six-Year Review 3
Technical Support Document for LT2
7-35
December 2016

-------
Other Protozoa
Recent studies have shown that UV irradiation also inactivates other protozoa, such as Giardia,
Toxoplasma gondii, Acanthamoeba and Naegleria fowleri, some of which exhibited similar dose
responses compared to Cryptosporidium. Shin et al. (2009b) studied Giardia inactivation by
polychromatic MP UV light. They found that for phosphate-buffered saline solution samples,
1.53-log inactivation was achieved at a dose of 0.5 mJ/cm2, and over 2.65-log inactivation was
achieved at 1 mJ/cm2 They reached the detection limit (>3.74-log inactivation) for filtered
drinking water samples at a dose of 1 mJ/cm2 Li et al. (2008) also studied Giardia lamblia
inactivation. While this study focused on the ability of the trophozoite to reactivate following
exposure to UV, the authors suggest the findings may have implications on criteria for UV
disinfection of G. lamblia.
Ware et al. (2010) reported the findings of a cell culture and mouse bioassay study, indicating 1-
log inactivation of T. gondii oocysts was achieved at an LP UV dose of 4 mJ/cm2 and a 3-log
inactivation was achieved at 10 mJ/cm2
Hijnen et al. (2006) studied the relationship between UV irradiation and the inactivation of
waterborne microorganisms. Their research showed Acanthamoeba is highly resistant to
disinfection by UV irradiation.
Sarkar and Gerba (2012) studied the inactivation of N. fowleri cysts and trophozoites by UV light
and found that the trophozoite form was readily inactivated by LP UV disinfection, requiring a
dose of 13, 18 and 24 mJ/cm2 for 2-, 3- and 4-log inactivation, respectively. The cyst form
required an LP dose of 63, 104 and 121 mJ/cm2 for 2-, 3- and 4-log inactivation, respectively.
The authors postulated that N. fowleri may have greater resistance to UV inactivation relative to
C. parvum because it is a free-living organism in surface water, where it is exposed to UV light
and, therefore, may have more developed DNA repair enzymes (Sarkar and Gerba, 2012).
Viruses
Adenoviruses are known to be more resistant to UV light than other pathogens, including other
viruses. Linden et al. (2007) studied enteric (Ad40) and respiratory (Ad2) adenovirus and found
MP to be more effective than LP UV. Full spectrum MP lamps achieved 4-log inactivation of
Ad40 at less than 60 mJ/cm2 The authors also studied surface discharge PUV lamps for
inactivation of Ad40 and concluded that PUV achieved 4-log inactivation of Ad40 at less than 40
mJ/cm2 For the inactivation of Ad2, results showed that full spectrum MP lamps achieved
greater than 4-log removal at 40 mJ/cm2 and at 60 mJ/cm2 Shin et al. (2009a) compared
inactivation of adenovirus by LP and MP UV light and found that after 14 days of exposure, MP
inactivation was 2.5 times more effective than LP inactivation, with the dose calculated to
achieve 4-log inactivation at 160 mJ/cm2 for LP and 63 mJ/cm2 for MP. Linden et al. (2009)
presented results of a full-scale MP-UV reactor validation of 4-log inactivation at < 100 mJ/cm2
using a live adenovirus challenge. Eischeid et al. (2011) conducted a literature review to better
understand the effects of UV irradiation on adenovirus inactivation. The researchers noted that
the majority of research conducted to date has used monochromatic (254 nm) LP UV
disinfection. Results from more recent studies using polychromatic UV sources along with
Six-Year Review 3
Technical Support Document for LT2
7-36
December 2016

-------
alternative assay methods show that adenovirus may not be as resistant to UV light as previously
understood.
Guo et al. (2010) tested three host cells for inactivation of adenovirus by LP and MP UV and
found that LP UV doses required for 4-log inactivation ranged from 123 to 182 mJ/cm2,
compared to MP UV doses of 65 to 90 mJ/cm2 A study by Eischeid and Linden (2009) explored
the possibility that MP irradiation may be more effective due to its ability to damage viral
proteins in addition to the viral DNA. However, the findings were not supportive of this
hypothesis at the UV doses used in water treatment (Eischeid and Linden, 2009). Linden et al.
(2011) recommended that, in light of new information on MP lamps, EPA should develop
separate dose requirements for inactivation of viruses by MP UV.
Only one study on UV inactivation of Coxsackie virus was identified. Vonder Haar et al. (2010)
found that 4-log inactivation of Coxsackie virus was achieved at LP lamp doses of 28 to 36
mJ/cm2, a level substantially lower than the 160 mJ/cm2 calculated for adenovirus by Linden et
al. (2007) discussed above.
Bacteria
The literature review identified many studies regarding UV inactivation of bacteria of potential
concern. For example, findings by Sun and Liu (2009) on the effectiveness of UV treatment on
E. coli, total bacteria counts, B. subtilis and MS2 bacteriophage show the required UV fluence
for 4-log inactivation to be 10, 5, 82 and 80 mJ/cm2, respectively, for LP lamps. UV irradiation
was effective against all four microorganisms. However, B. subtilis was more UV-resistant than
the other bacteria. The authors also pointed out that other studies have reported a higher UV
resistance for environmental bacteria and bacterial spores versus lab-cultured organisms (Sun
and Liu, 2009).
Hu et al. (2012) studied LP UV inactivation of several bacteria {Shigella dysenteriae, Salmonella
typhimurium and E. coli) and viruses (human rotavirus, MS2 and T4) and found the viruses to be
more resistant to UV light than the bacteria. Of the organisms studied, S. dysenteriae was found
to be the most sensitive to UV (5.71-log inactivation with an LP UV dose of 40 mJ/cm2).
New information is also available on inactivation of coliform bacteria and fungi in biofilms (Li
et al., 2010b, Murphy et al., 2008). Li et al. (2010b) studied the enhanced germicidal effects of
low-frequency psoralen plus ultraviolet-A-light emitting diode (LED) on biofilms that were
developed using E. coli and the fungus Candida albicans. Psoralen is a chemical that when
exposed to UV-A causes DNA damage or cell death. The biofilms were irradiated by continuous
and PUV light, and both achieved at least 90 percent inactivation of each organism. However,
PUV light produced significantly better results (over 95 percent of both microorganisms were
inactivated) than continuous UV light (Li et al., 2010b). Hotze et al. (2009) studied the use of
photosynthesized fullerol suspensions containing bacteriophage exposed to ultraviolet A (UVA)
light for possible optimization of viral inactivation.
Murphy et al. (2008) studied the effects of LP UV light in combination with a chemical
disinfectant (chlorine, chlorine dioxide and chloramines), compared to chemical disinfection
alone, on E. coli growth and persistence. The study included conditions representing those in
Six-Year Review 3
Technical Support Document for LT2
7-37
December 2016

-------
treatment plant effluent and distribution system biofilms. The authors concluded that, in general,
greater log inactivation was achieved with sequential disinfection compared to chemical
disinfection alone (UV log reduction results were not reported separately) (Murphy et al., 2008).
However, the authors acknowledged that some UV/chlorine-based disinfectant combinations
worked better than others. The combination of UV irradiation prior to chlorine resulted in a
longer persistence of E. coli, which suggested that some small colonies of E. coli repaired
themselves after exposure to UV treatment prior to chlorine disinfection (Murphy et al., 2008).
With UV and chlorine dioxide, one reactor showed E. coli reappearing during low disinfectant
doses. For UV and chloramines, E. coli was not detected in the effluent but was detected in the
biofilm (Murphy et al., 2008).
Hayes et al. (2008) studied the LP UV inactivation of Mycobacterium avium complex (MAC)
microorganisms, which are opportunistic human pathogens resistant to chlorination. They found
4-log inactivation achieved at less than 20 mJ/cm2for all organisms tested.
Cylindrospermopsis raciborskii is a cyanobacterium that produces the toxin cylindrospermopsin.
Cheng et al. (2009) studied the effect of free chlorine, ozone, chlorine dioxide, chloramine,
permanganate and LP UV light on both the bacteria and the toxin. They found that free chlorine
and ozone were effective against both, but UV irradiation was only effective at doses
substantially higher than would be typical for water treatment disinfection.
Ou et al. (2012) studied the impacts of LP UV irradiation on the photosynthetic capacity,
survival and recovery of Microcystis aeruginosa and investigated the risk of microcystin release
during UV irradiation. They found limited degradation at 140 mJ/cm2, and increased toxin
release. At higher doses UV light can efficiently destroy the photosynthetic capacity of the
organism, but with even higher toxin release.
Summary of UV Findings
Exhibit 7.4 provides a summary of the findings reported in this section.
Exhibit 7.4 Summary of UV Findings
Study
Organism
Type of
Lamp
Dose
(mJ/cm2)
Log
Inactivation
Other Information
Cryptosporidium (5.8 mJ/cm2 required for 2-log Cryptosporidium inactivation credit)
Mahmud et al.
2006
Cryptosporidium
LP
5
1.6
Poorly filtered water
2.4
Well-filtered water
Lee et al. 2008
Cryptosporidium
Pulsed
278
4.9
20 cm for 60 s
23
3
40 cm for 5 s
15
2
40 cm for 5 s
Hijnen et al.
2006
Cryptosporidium
MP
0.5-6.1
3
Summary of six studies
LP
0.9-13.1
3
Summary of four studies
Other Organisms - Protozoa (11 mJ/cm2 required for 3-log Giardia inactivation credit)
Shin et al.
2009b
G. lamblia
MP
0.5
1.53
Phosphate-buffered saline
samples
1.0
>2.65
Phosphate-buffered saline
samples
1
>3.74
Filtered drinking water
samples
Six-Year Review 3	7-38	December 2016
Technical Support Document for LT2

-------
Study
Organism
Type of
Lamp
Dose
(mJ/cm2)
Log
Inactivation
Other Information
Li et al. 2008
G. lamblia
LP
10-100
Not reported

Ware et al. 2010
T. gondii
LP
4
1

10
3
Sakar and
Gerba 2012
N. fowleri
LP
104
3

121
4
Other Organisms - Viruses (186 mJ/cm2 required for 4-log virus inactivation credit)
Linden et al.
2007
Adenovirus
LP
60
2

MP
<60
4

Pulsed
<40
4

Shin et al.
2009a
Adenovirus
MP
63
4
14 days
LP
160
Linden et al.
2009
Adenovirus
MP
<100
4

Guo et al. 2010
Adenovirus
LP
123-182
4

MP
65-90
Vonder Haar et
al. 2010
Coxsackie virus
LP
28-36
4

Other Organisms - Bacteria
Sun and Liu
2009
E. coli
LP
10
4
Environmental bacteria
more resistance to UV than
lab cultured bacteria
Total bacteria
counts
5
B. subtilis
82
MS2 bacteriophage
80
Huet al. 2012
Bacteria and
viruses
LP
Range
Range

Li et al. 2010b
E. coli and C.
albicans (a fungus)
UVA-LED
Continuous
0.28
90%
inactivation

UVA -LED
Pulsed
0.28
95%
inactivation

Murphy et al.
2008
E. coli
LP
90-100
UV with
chemical
disinfection

Hayes et al.
2008
MAC
LP
20
4
No light or dark repair of
W41 at four hours
Cheng et al.
2009
Cylindrospermopsis
raciborskii
LP


Degradation achieved but
not at typical WTP levels
Ou et al. 2012
Microcystis
aeruginosa
LP
<140

Degradation limited
>140

Photosynthetic capacity
destroyed
Sakai et al.
2011
M. aeruginosa
LP and MP
30-180
2
May result in toxin release
7.3.15.4 Implementation Issues Associated with UV Disinfection
UV is a relatively new technology compared to conventional disinfection and there have been a
number of issues with implementing UV for Cryptosporidium treatment. Water systems have
experienced operational issues with UV systems and some have had difficulty meeting the
criteria in the UVDGM. The documented operational issues include the following:
Six-Year Review 3
Technical Support Document for LT2
7-39
December 2016

-------
•	Monitoring the UV dose delivered,
•	Verifying inactivation for viruses,
•	Verifying that pre-existing UV installations meet Cryptosporidium inactivation
requirements, and
•	Maintenance of reactors.
Systems have reported challenges regarding UV monitoring requirements. The LT2 requires UV
reactors to be validated to determine the operating conditions that yield the UV dose needed to
achieve log inactivation credit [40 CFR 141.720(d)(2)], Parameters must include flow rate, UV
intensity as measured by a UV sensor, and UV lamp status (on or off).
Operating costs have been an issue for some systems. Depending on the frequency of lamp
replacement, the cost to replace aging lamps can be significant. Lengthening lamp replacement
times can result in lower costs but may affect inactivation credits. In addition, continuous
operation of the UV lamps can lead to high power costs. Using dose pacing can reduce power
use but can also result in lower Cryptosporidium inactivation and may result in greater than 5
percent off-spec operation.
Cornwell et al. (2012) and Cross and Bunton (2012) both reported the time-intensive review
required by primacy agencies and the validation requirements for UV as difficulties in approving
the technology. Cornwell et al. (2012) also reported difficulties systems have had with the
monthly reporting requirements established by states and felt the reporting was overburdening
due to states' unfamiliarity with the processes. For the Citizens Energy Group/Water of
Indianapolis UV system, Indiana Department of Environmental Management (IDEM) approval
was delayed due to a third party reactor validation requirement (Moran, 2013). Additionally,
Cornwell et al. state that some states must follow Recommended Standards for Water Works,
which conflicts with the UVDGM on dosage requirements.
UV Implementation Issues Described in the Literature
Some literature suggested a need for improving/updating the validation protocol in the 2006
UVDGM. Other important topics in the literature include the precision and accuracy of
validation testing, challenge organisms, recommendations regarding UV sensors, fouling and
other relevant issues.
Accuracy and Precision of Validation Testing
Since the publication of the UVDGM in 2006, advances in UV validation methods have
improved the accuracy and precision of validation testing (Heath et al., 2009). The first
validation conducted in 2003 yielded a UV dose monitoring equation that fit the measured MS2
reduction equivalent dose (RED) with an R2 value of 0.90. Validation testing conducted in
fall/winter 2008 yielded UV dose-monitoring equations that predict MS2 and T1 REDs,
accounting for the RED bias (described below), with R2 values typically above 0.98.
Six-Year Review 3
Technical Support Document for LT2
7-40
December 2016

-------
The improvements include updated dose monitoring algorithms that better represent the true
relationship between RED and independent variables (flow rate, UV transmittance and lamp UV
output). They also account for the impact of microbe UV sensitivity on RED (RED bias), the use
of combined aging and fouling index values to minimize fouling during validation testing, and
the use of tighter criteria for the bounds of test microbe UV dose response.
Heath et al. (2009) state that the UVDGM empirical equation provides a good fit between RED,
flow rate, UV transmittance, UV sensor readings and lamp on/off status. However, the authors
state that since the publication of the 2006 guidance, there have been advancements in UV
validation testing. They provided what they believe to be a more accurate equation of the RED of
the modeled data based on the relations between RED and the dependent variables that better
reflect the performance of the reactor.
According to Heath et al. (2009), revising the quality assurance/quality control (QA/QC) process
would allow less uncertainty in the measured dose response. A smaller margin of error would
provide more accurate delivery by the UV reactor.
Challenge Microorganisms and Nonbiological Alternatives for Validation Testing
One study by Wright et al. (2011), noted that the challenge organism MS2 bacteriophage
typically used in validation testing is significantly more resistant to UV light than
Cryptosporidium. Use of MS2 bacteriophage can lead to overly conservative reactor designs,
oversized equipment and increased costs.
The LT2 requires a UV dose of 186 mJ/cm2 for 4-1 og inactivation of viruses, based on the highly
UV-resistant adenovirus. This high UV dose presents multiple challenges. First, to ensure that
the required doses are reliably achieved, reactors must be designed with a safety factor or
validation factor (typically between 1.2 and 3.0). This puts dosage requirements at a very high
level. A second concern is associated with challenge organisms used for validation testing. A
common challenge organism is the MS2 bacteriophage, which has a lower UV resistance than
adenovirus. To demonstrate the ability to achieve 4-log inactivation of adenovirus, the MS2
bacteriophage must achieve a 7.4- to 10.3-log inactivation, which is operationally difficult to
achieve (Yates et al., 2006; Petri and Odegaard, 2008).
Issues with validation testing have been a subject of research, and several ideas have been
proposed to improve the validation process. A number of studies suggest alternate challenge
organisms for reactor validation (Fallon et al., 2007; Wright et al., 2007; Bandy, 2010). Several
studies (Petri and Odegaard, 2008; Gall et al., 2010; Rochelle et al., 2010) suggest using other
challenge organisms for validating high log reduction or inactivation of viruses. Wright et al.
(2009) suggested using different sensors and the application of scaling factors. Yates et al.
(2006) reviewed validation methods and suggested options such as changing the UV wavelength
and using different analytical techniques.
EPA identified research on new challenge microorganisms for the validation of UV reactors at
the UV doses required for virus inactivation credit (Rochelle et al., 2010; Hargy et al., 2011).
The primary objective of the Rochelle et al. (2010) study was to identify a challenge organism
that effectively models a UV reactor's virus inactivation efficiency. While the authors identified
Six-Year Review 3
Technical Support Document for LT2
7-41
December 2016

-------
a variety of highly UV-resistant microbes, the microbes that were most resistant and most
suitable as challenge organisms were spores of native isolates of B. pumilus. The spore's
response to UV radiation could be manipulated by varying the concentration of manganese in the
culture medium. This provided the ability to vary the dose response to be close to that of the
adenovirus 2 (Ad2) dose response curve or to be more resistant than adenoviruses. The ability to
culture the organism with the desired dose response makes B. pumilus spores a promising
alternative challenge microbe for validating UV reactors for virus inactivation credit (Rochelle et
al., 2010). Hargy et al. (2011) supported these findings, noting that B. pumilus spores overcome
the logistical problems of using coliphage MS2 at high UV doses.
Petri et al. (2011) presented findings from an experiment that used Aspergillus niger as a high-
resistance challenge organism for validating low pressure, high output (LPHO) UV reactors for
4-log virus inactivation. To overcome the RED bias, Petri et al. included MS2, T1 and T7
bacteriophages in the study. The study involved four UV reactors, denoted as small, medium,
large and extra-large. The study concluded that high UV transmittance is needed at a higher flow
rate, but 4-log inactivation was achieved for the challenge organisms.
Prior to the 2011 study, Petri and Odegaard (2008) also tested A. niger as a challenge organism
for LP and MP UV reactors. The authors showed that A. niger spores can be used to demonstrate
high-dose UV reactor challenges for both LP and MP UV reactors. However, even at high UV
doses (100 to 450 mJ/cm2), the 2008 experiment achieved only up to 2.7-log inactivation and not
the desired 4-log inactivation.
In addition to testing challenge microorganisms, Rochelle et al. (2010) evaluated the use of dyed
microspheres for UV reactor validation. Results showed that Lagrangian actinometry, based on
dyed microsphere (DMS) fluorescence measurements, characterized the dose distribution in UV
reactors. In addition, at higher flow rates, there was a correlation between spore inactivation and
DMS predicted inactivation. The authors noted that the correlation was weaker at lower flow
rates. The study concluded that while further testing is needed to better define the method's
uncertainty, Lagrangian actinometry should be considered for UV reactor validation.
An earlier study by Scheible et al. (2008) supported Rochelle's conclusions regarding
Lagrangian actinometry. They stated that experiments conducted to date (up to 2008) confirmed
the method's ability to yield accurate dose distribution measurements over wide ranges of
operating conditions and reactor types.
For systems, particularly small systems, that installed UV prior to the LT2, it has been difficult
to apply for credit to meet the LT2 requirements. Many of the UV reactors in these systems are
very small, were not initially validated, and may not have monitoring capabilities.
Nicholson et al. (2011) surveyed six large systems that employed UV treatment and found all
had SCADA systems that could monitor parameters to verify whether the system is operating
within validated conditions. Therefore, if manufacturers can provide validation parameters,
systems can generally monitor these parameters. However, smaller systems may not have this
advanced technology. These systems do not always have SCADA systems and may not have UV
sensors to monitor the necessary parameters. Some combination of higher doses and on/off
monitoring may be a possible method of addressing these smaller system issues.
Six-Year Review 3
Technical Support Document for LT2
7-42
December 2016

-------
Sensors
Another issue when monitoring for UV dose is related to the sensors that measure the lamp
intensity. There is a wide variety of reference sensors used to calibrate sensors at the plant. The
variation was found to be significant enough that some systems did not meet the requirements as
laid out in the UVDGM (Wright et al., 2009). Sensor readings can also degrade over time and are
affected by their position relative to the lamp. Also, lamp output degrades over time and the
degradation is not uniform. Many systems did not perform regular checks of reactor sensors
using reference sensors (Wright et al., 2009).
Wright et al. (2009) provided recommendations to standardize the design and performance of
UV sensor systems. The authors determined that it would not be feasible to propose standard
reference sensors with fixed physical dimensions and optical properties, and they instead
proposed guidance that would allow the suppliers some latitude in how they achieve the
standards for primary reference, reference and duty sensors.
Fouling during Operation of UV Reactor Systems
Reactor maintenance can also be an issue. Lamp fouling can occur in certain waters and in some
cases has caused wiper blades to stick, resulting in broken lamps (Nicholson et al., 2011).
Systems need to follow manufacturers' recommendations on performing preventive maintenance
and using replacement bulbs emitting correct wavelength for UV systems.
Substances in water can foul the external surfaces of UV lamp sleeves and reduce the
transmittance of UV light through the sleeve into the water. Fouling presents a design and
operational challenge for UV systems. A year-long study performed by Talbot et al. (2011)
investigated fouling tendencies of UV reactors for an unfiltered system. They examined two UV
MP reactors and one LPHO reactor. The results showed that fouling occurred only in the MP
reactors during the late summer and early fall when the raw water source contained elevated
levels of iron and manganese. For the remainder of the year, fouling was not an issue. The LPHO
reactor did not show signs of fouling even with elevated minerals in the water. This study also
examined the effect of prechlorination on the UV reactors. While the LPHO reactor showed
some signs of fouling, the MP reactors demonstrated "rapid and severe" fouling. The authors
concluded that chlorination of unfiltered water prior to UV treatment can dramatically increase
fouling, and if the system were to use an MP UV system, automatic wipers would be needed
during the fall months.
Off-Specification Operation
Design and operation of a UV system under the LT2 requires that 95 percent of the water
produced by a UV system be produced under validated conditions. This allows only 5 percent of
water to be produced during "off-spec" operation. Some systems have reported difficulty
meeting the 95 percent requirement because of power issues and limitations. For example, since
UV lamps have a relatively long warm-up time, even short power interruptions can lead to
significant reductions in inactivation achieved. One study found that even reactors operating
within the validated conditions 95 percent of the time could still see reduced log inactivation
(Wright et al., 2007).
Six-Year Review 3
Technical Support Document for LT2
7-43
December 2016

-------
Hubel (2007) conducted a study to evaluate the reliability of UV disinfection systems. The study
found that downtime significantly reduced inactivation. Hubel discussed a system operating with
1-percent downtime. Assuming constant flow and 100 percent survival during a 1-percent off-
spec operation, the system cannot achieve higher than 2-log pathogen inactivation. Continuously
operating with a downtime of 0.01 percent or less provides a net disinfection closer to the
validated system credit but Hubel considered this unrealistic, even with standby power and a
UPS. Due to the effects of downtime, Hubel does not believe inactivation of greater than 3-log is
realistic. Hubel also states that warm up time for LP systems is 4-7 minutes for cold start and 2-
7 minutes for warm start. For MP systems, the warm up time is 1-5 minutes for a cold start and
4-10 minutes for a warm start. Potential mitigation measures recommended by Hubel to increase
UV disinfection system reliability and minimize downtime include redundant equipment,
standby power with an automatic transfer switch, active series compensators and UPSs.
7.4 Summary
PWSs have a variety of options available in the microbial toolbox if required to provide
additional Cryptosporidium treatment under the LT2. These tools include treatment and
management strategies including source protection, enhanced treatment performance, pre-
filtration treatment, additional filtration and disinfection/inactivation options. Anecdotal
information provided to EPA indicates that currently the most commonly implemented tools are
combined or individual filter performance, UV disinfection, membrane filtration and WCPs.
Literature that has come available since the publication of the LT2 was reviewed along with key
information that supported the LT2 requirements to assess the use of each risk mitigation tool,
their effectiveness and identify implementation issues. EPA believes that these new information,
when taken in the context of the information that supported the original rule, support the existing
LT2 microbial toolbox requirements and credits.
Six-Year Review 3
Technical Support Document for LT2
7-44
December 2016

-------
8
Uncovered Finished Water Reservoirs
8.1	Background on the LT2 Uncovered Finished Water Reservoir Requirements
Under the Long Term 2 Enhanced Surface Water Treatment Rule (LT2), public water systems
(PWSs) with uncovered finished water reservoirs (UCFWRs) must either cover the storage
facility or treat the water leaving the storage facility to achieve inactivation and/or removal of 4-
log virus, 3-log Giardia lamblia and 2-log Cryptosporidium using a protocol approved by the
state [40 CFR 141.714] (USEPA, 2006a).
On August 11, 2003, EPA proposed as part of the LT2 that all PWSs using any UCFWRs must
cover them, treat the water leaving the reservoir to achieve at least 4-log virus inactivation using
a protocol approved by the state, or have a state-approved risk mitigation plan (RMP) that
addresses physical access and site security, surface water runoff, animal and bird waste, and an
ongoing water quality assessment (USEPA, 2003b). EPA changed the inactivation option in the
final LT2, such that systems treating water leaving the UCFWRs must provide 4-log virus, 3-log
G. lamblia and 2-log Cryptosporidium inactivation and/or removal (USEPA, 2006a). EPA also
eliminated the option of the state-approved RMP from the final LT2. EPA found a plan with
control measures for all sources of contamination that may affect UCFWRs would not be
feasible; therefore, an RMP would not provide public health protection equivalent to covering or
treating the water exiting the UCFWRs (USEPA, 2006a).
Section 8.2 provides background information on UCFWRs and the status of remaining reservoirs
as of December 2015. Section 8.3 summarizes information that supported the LT2 requirements
for PWSs with UCFWRs. Information that has become available since the promulgation of the
LT2, including discussions of permanent and temporary solutions and their respective
effectiveness, is discussed in Section 8.4. Section 8.5 provides a discussion of implementation
issues related to the LT2 requirements to cover or treat UCFWRs.
8.2	Background on Uncovered Finished Water Reservoirs
Some PWSs have used UCFWRs for water storage for decades, with their use being questioned
since 1930 due to concerns about their susceptibility to contamination (LeChevallier et al.,
1997). The storage of treated drinking water in open reservoirs can lead to significant water
quality degradation and health risks to consumers (USEPA, 1999). Examples of such water
quality degradation include increases in algal cells, coliform bacteria, heterotrophic bacteria,
turbidity, particulates, disinfection byproducts (DBPs), metals, taste and odor issues, insect
larvae, viruses, Giardia, Cryptosporidium and nitrate (USEPA, 1999). Contamination of
reservoirs occurs through surface water runoff, bird and animal wastes, human activity, algal
growth, insects and fish, and airborne deposition.
Many systems have taken measures to cover these reservoirs, treat the water leaving the
reservoirs, replace them with other storage facilities (e.g., ground level storage) or take them out
of service. When EPA promulgated the LT2 on January 5, 2006, there were 81 UCFWRs still in
use by PWSs in the United States. As of December 2015, there were only 24 UCFWRs among
12 PWSs in use. All PWSs with UCFWRs in the United States are under administrative orders or
compliance agreements to cover or treat their reservoirs (Exhibit 8.1).
Six-Year Review 3
Technical Support Document for LT2
8-1
December 2016

-------
Exhibit 8.1 Systems with Remaining UCFWRs as of December 2015
Location
Reservoir
Compliance Strategy
Ticonderoga, NY
Chilson
Storage Tank or elimination
Rochester, NY
Highland
Ultraviolet (UV) Treatment
Cobbs Hill
UV Treatment
Rome, NY
Unknown
UV Treatment
Unknown
UV Treatment
New York, NY
Hillview
Cover
Passaic Valley, NJ
Great Notch
Storage Tank
New Street
Storage Tank
Levine
Storage Tank
Newark, NJ
Cedar Grove
TBD
Trenton, NJ
Pennington
Floating Cover
Baltimore, MD
Ashburton
Covered Storage
Druid Lake
Covered Storage
Guilford
Covered Storage
Los Angeles, CA
Ivanhoe
Remove from Service
Elysian
Floating Cover
Los Angeles
UV Treatment
Upper Stone Canyon
Floating Cover
Rancho Estates, CA
Unknown
Cover
Unknown
Cover
Pauma Valley, CA
Upper
Consolidation with Yuima MWD.
Will be removed from service.
Lower
Consolidation with Yuima MWD.
Will be removed from service.
Portland, OR
Washington Park #3
Cover
Washington Park #4
Decommission
8.3 Summary of Information Supporting the LT2 Requirements
Bird and animal wastes are significant sources of contamination, potentially introducing human
pathogens such as Cryptosporidium and Giardia to the UCFWRs. In addition, runoff can enter
UCFWRs and may contain human pathogens, agricultural chemicals, automotive wastes, metals
and organic matter, as well as increase turbidity (LeChevallier et al., 1997; USEPA, 1999).
LeChevallier et.al. (1997) monitored for Cryptosporidium and Giardia in six UCFWRs and
found that the geometric mean concentration of Cryptosporidium oocysts increased from 1.2
oocysts/100 L in the inlet samples to 8.1 oocysts/100 L in the outlet samples. Similarly, the
researchers found that Giardia increased from 1.9 cysts/100 L to 6.1 cysts/100 L.
Graczyk et al. (2000) concluded that C. parvum can be transported by filth flies not only from
cattle sources but from any unhygienic source contaminated with C. parvum (i.e., toilets,
slaughterhouse, trash, carcasses and sewage).
Six-Year Review 3
Technical Support Document for LT2
8-2
December 2016

-------
8.4 Information Available Since the Promulgation of the LT2
Majewska et al. (2008) examined the epidemiologic link that aquatic birds present in water-
associated transmission cycles of Cryptosporidium and Giardia by performing a study to
determine the prevalence of free-ranging, captive and domestic birds of Western Poland that
shed C. parvum oocysts and G. lamblia cysts in their fecal droppings. Of the 499 samples from
308 free-ranging, 90 captive and 101 domestic birds, 26 tested positive for G. lamblia cysts and
19 tested positive for C. parvum oocysts. Furthermore, G. lamblia cysts and C. parvum oocysts
were found to occur considerably more in the feces of free-ranging aquatic birds than in birds not
typically associated with water (Majewska et al., 2008).
EPA collected information from seven PWSs on the strategies used to address potential public
health risks from UCFWRs. The information included permanent solutions, as well as temporary
solutions systems used while planning and implementing more permanent solutions. EPA
obtained the information from the City of Rochester (Rochester, 2013), the New York City
Department of Environmental Protection (NYCDEP) (NYCDEP, 2012), the Baltimore City
Department of Public Works (BCDPW) (BCDPW, 2012), the Fallbrook (California) Public
Utility District (FPUD) (FPUD, 2013), the Los Angeles Department of Water and Power
(LADWP) (LADWP, 2012), the Tacoma Public Utilities (TPU) (TPU, 2012) and Seattle Public
Utilities (SPU) (SPU, 2012). Where available, EPA also collected information on the
effectiveness of risk mitigation measures. This includes microbial monitoring data and public
health surveillance information.
On April 24, 2012, EPA hosted a public meeting to discuss information that could inform the
regulatory review of the LT2 UCFWR requirement. Discussion addressed the following topics:
• Status of Individual UCFWRs.
o New York, New York.
¦	The NYCDEP manages a protozoan assessment and risk mitigation
program for Hillview Reservoir.
¦	NYCDEP estimated $1.6 billion to cover the Hillview Reservoir
(NYCDEP, 2015)
¦	Cryptosporidium and Giardia data are collected from the inflow and
outflow of the Hillview Reservoir using EPA Method 1623.
¦	Risk mitigation approaches taken by NYCDEP at Hillview Reservoir,
include:
•	Wildlife management (e.g., bird harassment and deterrents,
mammal relocation),
•	security measures,
•	runoff control,
Six-Year Review 3
Technical Support Document for LT2
8-3
December 2016

-------
• public health surveillance,
•	microbial monitoring (e.g., Cryptosporidium, E. coli), and
•	Cryptosporidium and Giardia Action Plan.
¦	The New York City Department of Health and Mental Hygiene
(NYCDOHMH) has observed a declining trend in both giardiasis and
cryptosporidiosis since monitoring began in 1993 and 1994, respectively
(NYCDOHMH and NYCDEP, 2015).
o Seattle, Washington.
¦	At the time of the public meeting, two UCFWRs remained, holding 18
percent of the city's water. Both UCFWRs have since been
decommissioned.
o T acoma W ashington.
¦	Since 1999, Tacoma's reservoirs have been restructured, covered and
reduced in size, at a cost of $53 million to the city (TPU, 2012).
¦	Midge fly larvae elimination was a major concern.
¦	Previous significant losses of chlorine and pH in the UCFWRs have been
mitigated through the covering of the reservoirs.
•	Pathogen Risks Associated with UCFWRs.
o UCFWRs are vulnerable to fecal inputs and other pathogen risks, including birds,
rodents, feral cats and dogs, but additional treatment could address these risks.
Additionally, as part of the participant perspectives sessions at the April 2012 meeting, 16
members of the public provided comments. Also, during a panel discussion titled "Options and
Opportunities for Long Term 2 Revisions," five experts from water utilities, a state health
department, an academic institution and a non-profit organization discussed perspectives on the
information provided and the observations made throughout the meeting (USEPA, 2012d). These
observations included:
•	It is time to review the LT2 to consider new and innovative practices, such as monitoring
methods;
•	EPA should consider economics in the requirements, since some challenges are not
economically feasible. The appropriate response may not be covering or treating;
•	The risks from covered versus uncovered reservoirs are not equivalent. However, you
cannot generalize regarding the risks from all UCFWRs;
•	In no other case do we monitor the water to determine its safety. Instead, we monitor to
determine how the controls are working, and;
•	No justification for a rule change is apparent.
Six-Year Review 3
Technical Support Document for LT2
8-4
December 2016

-------
8.4.1 Permanent Solutions Taken
PWSs with UCFWRs have taken a variety of different approaches to permanently address
UCFWRs since the promulgation of the LT2 in 2006, including providing treatment for the
reservoir's effluent, covering the reservoir, replacing the reservoir with buried storage and
removing the reservoir from service. Based on information from the seven PWSs (14 UCFWRs)
described in Section 8.4, three of 14 UCFWRs were addressed by adding treatment (one UV
light and two microfiltration plants), two by adding floating covers to the reservoirs, seven by
replacing reservoirs with buried storage and two by removing them from service. Of the 24
UCFWRs that remain in the United States, UV disinfection is planned for five, covers are
planned for seven, covered storage is planned for seven, decommissioning or removal from
service is planned for two, consolidation with another system will address two, and the plans for
the remaining one had not been determined as of December 2015 (Exhibit 8.1).
8.4.2	Effectiveness of Permanent Solutions
To evaluate the effectiveness of permanent solutions, EPA attempted to collect and compare the
UCFWRs' microbial monitoring data before and after the PWSs provided their solutions. While
EPA was able to collect some microbial monitoring data, none of the PWSs had monitoring data
before and after covering the reservoirs or treating the reservoir outflow. Therefore, EPA was
unable to determine the effectiveness of covering or treating to address concerns regarding
microbial contamination of UCFWRs.
Covering UCFWRs is an effective strategy since covers would preclude most animals from
entering (and contaminating) the reservoirs, airborne contamination would be prevented, runoff
could not enter the reservoir, and security related concerns could be more adequately addressed.
Likewise, treating the water leaving the reservoir would inactivate or oxidize contaminants that
may enter the finished water in the reservoir. Given the many confounding factors (e.g.,
cryptosporidiosis and giardiasis from causes other than drinking water), a comparison of disease
rated before and after taking permanent solutions is very difficult to perform. A large change in
cryptosporidiosis or giardiasis rates would be necessary to be able to attribute the outcome to
covering or treating the reservoirs exclusively.
8.4.3	Temporary Solutions Taken
Some utilities have undertaken a range of different risk management approaches as temporary
measures while planning to cover/treat the reservoirs as a permanent measure. Some of the
measures prevent contaminants from entering UCFWRs. These include runoff control (e.g.,
raised berms, concrete walls), wildlife management (e.g., bird wires, animal capture and
relocation, bird shooting) and security measures (e.g., warning signs, fencing, cameras, guards).
Other measures include monitoring and surveillance activities, such as microbial monitoring
(e.g., Cryptosporidium, Giardia), animal counts, analysis of scat found near the reservoir and
public health surveillance. Additional measures can be categorized as treatment, and include
secondary disinfection, algae control and reservoir cleaning. Lastly, some risk management
activities relate to responses to elevated concerns based on the monitoring and surveillance
activities (e.g., microbial action plans, public notification, boil water alerts).
Six-Year Review 3
Technical Support Document for LT2
8-5
December 2016

-------
8.4.4 Effectiveness of Temporary Solutions
EPA collected reservoir microbial monitoring information and public health surveillance
information to determine the effectiveness of risk management approaches other than covering or
treating. Very limited data are available since very few utilities collect monitoring information
from reservoirs, and even fewer utilities collect enough information before and after the
implementation of a particular strategy to illustrate a change in microbial occurrence. The
effectiveness of individual risk management measures is also difficult to determine from
microbial monitoring or public health surveillance because individual measures (e.g., bird wires)
tend to be relatively small in scope and are typically provided in concert with other risk
management measures. As such, EPA has not been able to collect data that would indicate the
effectiveness of any of these measures. However, for New York City, EPA has qualitatively
observed a decrease over time in the incidence of cryptosporidiosis and giardiasis by comparing
New York City's waterborne disease surveillance reports across timeframes during which the
New York City Department of Environmental Protection has implemented various measures
(NYCDEP, 2013). The role each of these measures plays regarding the decrease in
cryptosporidiosis and giardiasis is unknown, since many other factors could have influenced this
decrease.
8.5 Implementation Issues Related to the LT2 Cover/Treat Requirements
Following the promulgation of the LT2, EPA identified some implementation issues related to
the cover/treat requirements, including costs; weather-related concerns; and local government,
utility and community receptiveness to the LT2 UCFWR requirements.
Systems have some options when making decisions regarding how to cover their UCFWRs,
including the use of a variety of materials, such as concrete, aluminum, fabric and floating covers
(e.g., polypropylene). Systems often make decisions about what type of cover to use based on
costs and the feasibility of cover types for their reservoirs. Concrete covers are typically the most
costly in terms of capital costs, while floating covers tend to be the least costly. According to
Lenz et al., (2008) the costs per square meter are $65 for floating covers, $430 for aluminum
covers, $480 for fabric covers and $2,200 for concrete covers. However, floating covers and
fabric covers tend to have higher operation and maintenance (O&M) costs. The O&M for
floating covers includes stormwater removal from the surface of the cover and minimizing water
surface fluctuations during icing conditions (Lenz et al., 2008). Systems need stormwater control
for some reservoir cover materials, such as concrete and aluminum. Concrete covers and
aluminum covers have relatively low O&M costs (Lenz et al., 2008). However, for aluminum
covers, expansion and contraction due to the weather can be a concern.
The useful life also varies greatly depending on the cover material, with concrete having a 100-
year useful life, aluminum having 30-50 years, floating having 20-25 years and fabric having
less than 20 years (Lenz et al., 2008). Some covers may be installed while the reservoir remains
in service (e.g., aluminum, and floating covers) while concrete covers, which can have a lengthy
construction schedule, require the reservoir to be taken out of service (Lenz et al., 2008).
The construction of reservoir covers can significantly change the appearance of the site for some
cover types. This has been a significant concern faced by some PWSs, where community groups
Six-Year Review 3
Technical Support Document for LT2
8-6
December 2016

-------
or local governments have voiced opposition. Some PWSs have overcome this opposition by
building parks on top of concrete covers, or involving the community in the aesthetic aspects of
the covers (LADWP, 2012).
Buried storage tanks are another option that many PWSs have implemented to address their
UCFWRs. In some cases, PWSs have decreased their available storage by replacing their
UCFWRs with buried storage, and as a result, have seen an improvement in water quality
(decreased water age), as well as less chemical usage (LADWP, 2012; SPU, 2012; TPU, 2012).
For some PWSs this may not be feasible, as they may need the additional capacity to meet high
demand. Some utilities, such as SPU, have also built parks over buried concrete storage tanks
(SPU, 2012).
Some systems have decided to install treatment to address their UCFWRs, with the most
common being UV disinfection and membrane filtration. Like covering reservoirs, adding
treatment is not without its challenges. LADWP installed a membrane filtration plant to treat
water from its Lower Stone Canyon reservoir (LADWP, 2012). The plant installation raised
community concerns related to noise and light pollution that LADWP had to abate before putting
the treatment into use. In the case of Pittsburgh's Highland 1 reservoir, the community wanted to
keep the reservoir open to the public, so the Pittsburgh Water and Sewer Authority (PWSA)
installed membrane filtration (PWSA, 2013). The FPUD, installed a UV system to treat water
from its Red Mountain reservoir in 2010 (FPUD, 2013) which required installing fencing to
prevent public access. PWSs have encountered community resistance to installing UV treatment
due to the local community objecting to having a treatment facility in the neighborhood or a city
park.
Six-Year Review 3
Technical Support Document for LT2
8-7
December 2016

-------
9 References
Abbaszadegan, M., T. Rauch-Williams, W.P. Johnson, and S.A. Hubbs. 2011. Methods to assess
GWUDI and bank filtration performance. Water Research Foundation.
Aboytes, R., G.D. Di Giovanni, F.A. Abrams, C. Rheinecker, W. McElro, N. Shaw, and M.W.
LeChevallier. 2004. Detection of infections Cryptosporidium filtered in drinking water.
American Water Works Association. 96(9): 88.
Alvarez, M., J. Rose, and B. Bellamy. 2010. Treating algal toxins using oxidation, adsorption,
and membrane technologies. Water Research Foundation.
American Water Works Association (AWW A). 1999. Water quality and treatment: A handbook
of community water supplies. McGraw-Hill, New York.
AWWA. 2000. The drinking water dictionary. American Water Works Association, Denver, CO.
AWWA. 2006. Manual M48. Waterborne pathogens. 2nd edition. American Water Works
Association, Denver, CO.
Amy, G. and C. Ha. 2012. Enhanced performance of ceramic membrane filtration for treating
impaired water quality sources in drinking water/reuse applications. King Abdullah University of
Science and Technology: Water Desalination and Reuse Center.
An, W., D. Zhang, S. Xiao, J. Yu, and M. Yang. 2011. Quantitative health risk assessment of
Cryptosporidium in rivers of southern China based on continuous monitoring. Environmental
Science and Technology. 45(11): 4951-4958.
Antenucci, J.P, J.D. Brookes, and M.R. Hipsey. 2005. A simple model for quantifying
Cryptosporidium transport, dilution, and potential risk in reservoirs. Journal of American Water
Works Association. 97(1): 86-93.
Aramini, J.J., C. Stephen, J.P. Dubey, C. Engelstoft, H. Schwantje, and C.S. Ribble. 1999.
Potential contamination of drinking water with Toxoplasma gondii oocysts. Epidemiology and
Infection. 122: 305-315.
Arora, H., M. LeChevallier, R. Aboytes, E. Bouwer, C. O'Melia, W. Ball, W. Weiss, and T.
Speth. 2000. Full-scale evaluation of riverbank filtration at three Midwest water treatment
plants. Proceedings of the Water Quality Technology Conference, Salt Lake City, Utah,
American Water Works Association, Denver, Colorado.
Astrom, J., S. Petterson, O. Bergstedt, T.J.R. Petterson, and T.A. Stenstrom. 2007. Evaluation of
the microbial risk reduction due to selective closure of the raw water intake before drinking
water treatment. Water and Health. 5(1): 81-97.
Astrom, J., T.J.R. Petterson, T.A. Stenstrom, and O. Bergstedt. 2009. Variability analysis of
pathogen and indicator loads from urban sewer systems along a river. Water Science &
Technology. 59(2): 203-212.
Six-Year Review 3
Technical Support Document for LT2
9-1
December 2016

-------
Baltimore City Department of Public Works (BCDPW). 2012. Personal communication.
Bandy, J. 2010. Building accurate and conservative engineering principles into microbe choice
for UV validation. Proceedings of the American Water Works Association - Water Quality
Technology Conference.
Becker, D.J., J. Oloya, and A.E. Ezeamama. 2015. Household socioeconomic and demographic
correlates of Cryptosporidium seropositivity in the United States. PLoS Neglected Tropical
Diseases. 9(9): e0004080.
Bellamy, W.D., G.P. Silverman, D.W. Hendricks, and G.S. Logsdon. 1985. Removing Giardia
cysts with slow sand filtration. Journal of American Water Works Association. 77(2): 52-60.
Berger, P. 2006. Email correspondence.
Bichai, F., Y. Dullemont, W. Hijnen, and B. Barbeau. 2014. Predation and transport of persistent
pathogens in GAC and slow sand filters: a threat to drinking water safety? Water Research. 64:
296-308.
Bohrerova, Z., and K.G. Linden. 2006. Assessment of DNA damage and repair in
Mycobacterium terrae after exposure to UV irradiation. Applied Microbiology. 101: 995-1001.
Bosch, A. 1998. Human enteric viruses in the water environment: a minireview. International
Microbiology. 1: 191-196.
Bosch, A., R.M. Pinto, and S. Guix. 2014. Human astroviruses. Clinical Microbiology Review.
27(4): 1048-1074.
Bottino, A., C. Capannelli, A. Del Borghi, M. Colombino, and O. Conio. 2001. Water treatment
for drinking purpose: ceramic microfiltration application. Desalination. 141: 75-79.
Bowie, W.R., A.S. King, D.H. Werker, J.L. Isaac-Renton, A. Bell, S.B. Eng, and S.A. Marion.
1997. Outbreak of toxoplasmosis associated with municipal drinking water. The BC Toxoplasma
Investigation Team. Lancet. 350(9072): 173-177.
Brehant, A., K. Glucina, I. Lemoigne, and J.M. Laine. 2008. Risk management approach for
monitoring UF membrane integrity and experimental validation using MS2-phages. Water
Science & Technology: Water Supply. 8(2): 239-244.
Brescia, C.C., S.M. Griffin, M.W. Ware, E.A. Varughese, A.I. Egorov and E.N. Villegas. 2009.
Cryptosporidium propidium monoazide-PCR, a molecular biology-based technique for
genotyping of viable Cryptosporidium oocysts. Applied and Environmental Microbiology.
75(21): 6856-6863.
Broman T., J. Waldenstrom, D. Dahlgren, I. Carlsson, I. Eliasson, and B. Olsen. 2004.
Diversities and similarities in PFGE profiles of Campylobacter jejuni isolated from migrating
birds and humans. Journal of Applied Microbiology. 96:834-843.
Six-Year Review 3
Technical Support Document for LT2
9-2
December 2016

-------
Brookes, J.D., J. Antenucci, M. Hipsey, M.D. Burch, N.J. Ashbolt, and C. Ferguson. 2004. Fate
and transport of pathogens in lakes and reservoirs. Environment International. 30: 741-759.
Brown, R.A. and D.A. Cornwell. 2007. Using spore removal to monitor plant performance for
Cryptosporidium removal. Journal of American Water Works Association. 99(3): 95-109.
Bukhari, Z., D.M. Holt, M.W. Ware, and F.W. Schaefer. 2007. Blind trials evaluating in vitro
infectivity of Cryptosporidium oocysts using cell culture immunofluorescence. Canadian Journal
of Microbiology. 53(5): 656-663.
Cama, V.A. and M.J. Arrowood. 2006. Molecular characterization of the Cryptosporidium
parvum Iowa isolate kept in different laboratories. Journal of Eukaryotic Microbiology. 53(S1):
40-42.
Centers for Disease Control and Prevention (CDC). 2001. HIV/AIDS Surveillance Report. 12(2):
7, 30.
CDC. 2002. HIV/AIDS Surveillance Report. 14: 1-50.
CDC. 2008. Surveillance for waterborne disease and outbreaks associated with drinking water
and water not intended for drinking - United States, 2005-2006. MMWR Surveillance
Summaries. 579: 39-62.
CDC. 2010. Cryptosporidiosis surveillance - United States, 2006-2008. MMWR Surveillance
Summary. 59(SS06);1-14.
CDC. 2012. Cryptosporidiosis surveillance - United States, 2009-2010. MMWR Surveillance
Summary. 61(SS05); 1-12.
CDC. 2011. Surveillance for waterborne disease and outbreaks associated with drinking water-
United States, 2007-2008. MMWR Surveillance Summary. 60(12): 38-68.
CDC. 2013. Surveillance for waterborne disease and outbreaks associated with drinking water
and other nonrecreational water - United States, 2009-2010. MMWR Surveillance Summary.
62(35): 714-720.
CDC, 2015a. Cryptosporidiosis surveillance - United States, 2011-2012. MMWR Surveillance
Summary. 64(SS03);1-14.
CDC. 2015b. Giardiasis surveillance - United States, 2011-2012. 64(SS03);15-25.
Chalmers, R.M., K. Elwin, S.J. Hadfield, and G. Robinson. 2011. Sporadic human
cryptosporidiosis caused by Cryptosporidium cuniculus, United Kingdom, 2007-2008. Emerging
Infectious Diseases. 17(3): 536-538.
Chappell, C.L., P.C. Okhuysen, C.R. Sterling, C. Wang, W. Jakubowski, andH.L. Dupont. 1999.
Infectivity of Cryptosporidium parvum in healthy adults with pre-existing anti-C. parvum serum
immunoglobulin G. American Journal of Tropical Medicine and Hygiene. 60(1): 157-164.
Six-Year Review 3
Technical Support Document for LT2
9-3
December 2016

-------
Chappell, C.L., P.C. Okhuysen, S. Tzipori, and G. Widmer. 2003. Final Report: Infectivity and
Virulence of Cryptosporidium Hon-parvum Species in Healthy Adult Volunteers. EPA Grant
Number R829180, 2003 Progress Report. Available at:
https://cfpub.epa.gov/ncer abstracts/index.cfm/fuseaction/displav.abstractDetail/abstract/1127/re
port/F
Chappell, C.L., P.C. Okhuysen, R.C. Langer-Curry, G. Widmer, D.E. Akiyoshi, S. Tanriverdi,
and S. Tzipori. 2006. Cryptosporidium hominis: Experimental challenge of healthy adults.
American Journal of Tropical Medicine and Hygiene. 75(5): 851-857.
Chappell, C.L., P.C. Okhuysen, R.C. Langer-Curry, D.E. Akiyoshi, G. Widmer, and S. Tzipori.
2011. Cryptosporidium meleagridis: Infectivity in healthy adult volunteers. American Journal of
Tropical Medicine and Hygiene. 85(2): 238-242.
Cheng, X., H. Shi, C.D. Adams, T. Timmons, and Y. Ma. 2009. Effects of oxidative and physical
treatments on inactivation of Cylindrospermopsis raciborskii and removal of
Cylindrospermopsin. Water Science & Technology. 60(3): 689-697.
Cinque, K. and N. Jayasuriya. 2010. Catchment process affecting drinking water quality,
including the significance of rainfall events, using factor analysis and event mean concentrations.
Journal of Water and Health. 08(4): 751-763.
Ciora, R.J., and K.T. Liu. 2003. Ceramic membranes for environmental related applications.
Fluid Particle Suspension. 15(1): 51-60.
City of Bend, Oregon. 2012. Letter to Oregon Health Authority. Long term 2 enhanced surface
water treatment rule compliance issues and bend system specific considerations. Available at:
http://www.bendoregon.gov/modules/showdocument.aspx? documentid=6787
Clancy, J.L., M.M. Marshall, T.M. Hargy, and D.G. Korich. 2004. Susceptibility of five strains
of Cryptosporidium parvum oocysts to UV light. Journal of the American Water Works
Association. 96(3): 84-93.
Clark, R., M. Sivaganesan, E. Rice, and J Chen. 2002a. Development of a Ct equation for the
inactivation of Cryptosporidium oocysts with ozone. Water Research. 36: 3141-3149.
Clark, R., M. Sivaganesan, E. Rice, and J Chen. 2002b. Development of a Ct equation for the
inactivation of Cryptosporidium parvum oocysts with chlorine dioxide. Water Research. 37(11):
2773-2783 (in press subsequently published in 2003).
Cleasby, J.L., D.J. Hilmoe, and C.J. Dimitracopoulos. 1984. Slow sand and direct in-line
filtration of a surface water. Journal of American Water Works Association. 76(12): 44.
Cornwell, D.A., N.E. McTigue, and S. Via. 2012. Lessons Learned from Use of the Toolbox: A
utility survey. Presented at USEPA Long-Term 2 Enhanced Surface Water Rule: Monitoring
Data Analysis, Occurrence Forecast, Binning, and the Microbial Toolbox Public Meeting.
Available at: http://vvvvvv.avvvva.Org/portals/0/files/legreg/documents/toolboxsurvev2012.pdf
Six-Year Review 3
Technical Support Document for LT2
9-4
December 2016

-------
Corona-Vasquez, B., J.L. Rennecker, A.M. Driedger, and B.J. Marinas. 2002. Sequential
inactivation of Cryptosporidium parvum oocysts with chlorine dioxide followed by free chlorine
or monochloramine. Water Research. 36(1): 178-188.
Corso, P.S., M.H. Kramer, K.A. Blair, D.G. Addiss, J.P. Davis, and A.C. Haddix. 2003. Cost of
illness in the 1993 waterborne Cryptosporidium outbreak, Milwaukee, Wisconsin. Emerging
Infectious Diseases. 9(4): 426-431.
Cotte, L., M. Rabodonirina, F. Chapuis, F. Bailly, F. Bissuel, C. Raynal, P. Gel as, F. Persat, MA
Pi ens and C. T repo. 1999. Waterborne outbreak of intestinal microsporidiosis in persons with
and without human immunodeficiency virus infection. Journal of Infectious Diseases. 180(6):
2003-2008.
Craun, G.F. 1988. Surface water supplies and health. Journal of American Water Works
Association. 80(2): 40-52.
Craun, G.F. and R. Calderon. 1996. Microbial Risks in Groundwater Systems: Epidemiology of
waterborne outbreaks. Under the Microscope: Examining Microbes in Groundwater. Proceedings
of the Groundwater Foundation's 12th Annual Fall Symposium.
Craun, M.F., G.F. Craun, R.L. Calderon, and M.J. Beach. 2006. Waterborne outbreaks reported
in the United States. Journal of Water and Health. 4 (Suppl 2): 19-30.
Craun, G.F., J.M. Brunkard, J.S. Yoder, V.A. Roberts, J. Carpenter, T. Wade, R.L. Calderon,
J.M. Roberts, M.J. Beach, and S.L. Roy. 2010. Causes of outbreaks associated with drinking
water in the United States from 1971 to 2006. Clinical Microbiology Reviews. 23(3): 507-528.
Cross, A.F. and J. Bunton. 2012. Microbial toolbox options: Two states' perspectives. Presented
at USEPA Long-Term 2 Enhanced Surface Water Rule: Monitoring Data Analysis, Occurrence
Forecast, Binning, and the Microbial Toolbox Public Meeting.
Cummins, E., R. Kennedy, and M. Cormican. 2010. Quantitative risk assessment of
Cryptosporidium in tap water in Ireland. Science of the Total Environment. 408(4): 740-753.
Daly, E.R., S.J. Roy, D.D. Blaney, J.S. Manning, V.R. Hill, L. Xiao, and J.W. Stull. 2010.
Outbreak of giardiasis associated with a community drinking-water source. Epidemiology and
Infection. 138(4): 491-500.
De Boutray, M.L. 2011. Development of lake ecosystem modeling approaches for responding to
cyanobacteria blooms in a drinking water source. Proceedings of the American Water Works
Association Water Quality Technology Conference.
DeLoyde, J.L, W.B. Anderson, S.A. Cleary, R.A. Ndiongue, M.G. LeCraw, and P.M. Huck.
2006. Recent progress in slow sand and alternative biofiltration processes. In: Removal of
Cryptosporidium Oocysts and Giardia Cysts by Pilot-Scale Multistage Slow Sand Filtration.
133-142.
Six-Year Review 3
Technical Support Document for LT2
9-5
December 2016

-------
DeSilva, M.B., S. Schafer, M.K. Scott, B. Robinson, A. Hills, G.L. Buser, K. Salis, J. Gargano,
J. Yoder, V. Hill, L. Xiao, D. Roellig, and K. Hedberg. 2015. Communitywide cryptosporidiosis
outbreak associated with a surface water-supplied municipal water system - Baker City, Oregon,
2013. Epidemiology and Infection. 144(2): 274-284.
Di Giovanni, G.D., F.H. Hashemi, N.J. Shaw, F.A. Abrams, M.W. LeChevallier, and M.
Abbaszadegan. 1999. Detection of infectious Cryptosporidium parvum oocysts in surface and
filter backwash water samples by immunomagnetic separation and integrated cell culture-PCR.
Applied Environmental Microbiology. 65(8): 3427-3432.
Di Giovanni, G.D., R.M. Hoffman, and G.D. Sturbaum. 2010. Cryptosporidium Genotyping
Method for Regulatory Microscope Slides. Water Research Foundation Report #4099.
Dixon, M., L. Ho, C. Chow, G. Newcombe, J.P. Croue, J. Cigana, H. Buisson, and R. Treuger.
2012. Evaluation of integrated membranes for taste and odor and toxin control. Water Research
Foundation and Water Quality Research Australia.
Dreelin, E.A., R.L. Ives, S. Molloy, and J.B. Rose. 2014. Cryptosporidium and Giardia in
surface water: A case study from Michigan, USA to inform management of rural water systems.
International Journal of Environmental Research on Public Health. 11(10): 10480-10503.
Dugan, N R., K.R. Fox, J.H. Owens, and R.J. Miltner. 2001. Controlling Cryptosporidium
oocysts using conventional treatment. Journal of American Water Works Association. 93( 12):
64-76.
Dullemont, Y.J., J.F. Schijven, W.A.M. Hijnen, M. Colin, A. Magic-Knezev, and W.A.
Oorthuizen. 2006. Recent progress in slow sand and alternative biofiltration processes. Removal
of Microorganisms by Slow Sand Filtration.
DuPont, H.L., C.L. Chappell, C.R. Sterling, P.C. Okhuysen, J.B. Rose, and W. Jakubowski.
1995. The infectivity of Cryptosporidium parvum in healthy volunteers. New England Journal of
Medicine. 332(13): 855-859.
Edzwald, J.K. and M.B. Kelley. 1998. Control of Cryptosporidium from reservoirs to clarifiers to
filters. Water Science and Technology. 37(2): 1-8.
Eischeid, A.C., J.N. Mayer, and K.G. Linden. 2009. UV disinfection of adenoviruses, molecular
indications of DNA damage efficiency. Applied and Environmental Microbiology. 75(1): 23-28.
Eischeid, A.C. and K.G. Linden. 2009. Protein damage in UV treated adenovirus. American
Water Works Association Water Quality Technology Conference.
Eischeid, A.C., J. A. Thurston, and K.G. Linden. 2011. UV disinfection of adenovirus: Present
state of the research and future directions. Critical Reviews in Environmental Science and
Technology. 41(15): 1375-1396.
Six-Year Review 3
Technical Support Document for LT2
9-6
December 2016

-------
Emelko, M., P. Huck, and R. Slawson. 1999. Design and operational strategies for optimizing
Cryptosporidium removal by filters. Proceedings of 1999 Water Quality Technology Conference,
American Water Works Association, Denver, CO.
Emelko, M.B., U. Silins, K.D. Bladon, andM. Stone. 2011. Implications of land disturbance on
drinking water treatability in a changing climate: Demonstrating the need for "source water
supply and protection" strategies. Water Research. 45(2): 461-472.
Enger, K.S. 2013. Cryptosporidiumparvum and Cryptosporidium hominis: Dose response
models. QMRA Wiki. Available at:
http://qmrawiki.canr.msu.edu/index.php/Crvptosporidium parvum and Cryptosporidium homi
nis: Dose Response Models
Englehardt, J.D. and J. Swartout. 2004. Predictive population dose-response assessment for
Cryptosporidium parvum: Infection endpoint. Journal of Toxicology and Environmental Health-
Part A-Current Issues. 67(8-10): 651-666.
Englehardt, J.D. and J. Swartout. 2006. Predictive Bayesian microbial dose-response assessment
based on suggested self-organization in primary illness response: Cryptosporidium parvum. Risk
Analysis. 26(2): 543-554.
Executive Order Number 13563, 76 FR 3821. January 21, 2011.
Fallbrook Public Utility District (FPUD). 2013. Personal Communication.
Fallon, K.S., T. M. Hargy, E.D. Mackey, H. B. Wright, and J. L. Clancy. 2007. Development and
characterization of nonpathogenic surrogates for UV reactor validation. Journal of the American
Waterworks Association. 99(3), 73-82.
Farkas, K., J. Plutzer, E. Moltchanova, A. Torok, M.J. Varro, K. Domokos, F. Frost, and P.R.
Hunter. 2015. Serological responses to Cryptosporidium antigens in inhabitants of Hungary
using conventionally filtered surface water and riverbank filtered drinking water. Epidemiology
and Infection. 143(13):2743-2747.
Faulkner, B.R., Y. Olivas, M.W. Ware, M.G. Roberts, J.F. Groves, K.S. Bates, and S.L.
McCarty. 2010. Removal efficiencies and attachment coefficients for Cryptosporidium in sandy
alluvial riverbank sediment. Water Research. 44(9): 2725-2734.
Feng, H., W. Nie, R. Bonilla, G. Widmer, A. Sheoran, and S. Tzipori. 2006. Quantitative
tracking of Cryptosporidium infection in cell culture with CFSE. Journal of Parasitology. 92(6):
1350-1354.
Feng, Y.Y., S.L. Ong, J.Y. Hu, L.F. Song, X.L. Tan, and W.J. Ng. 2003. Effect of particles on
the recovery of Cryptosporidium oocysts from source water samples of various turbidities.
Applied and Environmental Microbiology. 69(4): 1898-1903.
Six-Year Review 3
Technical Support Document for LT2
9-7
December 2016

-------
Feng, Y., X. Zhao, J. Chen, W. Jin, X. Zhou, N. Li, and L. Xiao. 2011. Occurrence, source, and
human infection potential of Cryptosporidium and Giardia spp. in source and tap water in
Shanghai, China. Applied and Environmental Microbiology. 77(11): 3609-3616.
Ferguson, C.M., B.F.W. Croke, P.J. Beatson, N.J. Ashbolt, and D.A. Deere. 2007. Development
of a process-based model to predict pathogen budgets for the Sydney drinking water catchment.
Journal of Water and Health. 5(2): 187-208.
Ferrer, O., S. Casas, C. Galvan, F. Lucena, A. Bosch, B. Galofre, J. Mesa, J. Jofre, and X.
Bernat. 2015. Direct ultrafiltration performance and membrane integrity monitoring by
microbiological analysis. Water Research. 83: 121-131.
Filtra Systems. 2009. Bag Filter Technologies.
Finn, M. 2012. Microbial Toolbox. USEPA long-term 2 enhanced surface water rule: Monitoring
data analysis, occurrence forecast, binning, and the microbial toolbox public meeting.
Fong, T.T. and E.K. Li pp. 2005. Enteric viruses of humans and animals in aquatic environments:
health risks, detection, and potential water quality assessment tools. Microbiology and Molecular
Biology Reviews. 69(2): 357-371.
Francy, D.S., O.D. Simmons III, M.W. Ware, E.J. Granger, M.D. Sobsey, and F.W. Schaefer III.
2004. Effects of seeding procedures and water quality on recovery of Cryptosporidium oocysts
from stream water by using U.S. environmental protection agency method 1623. Applied and
Environmental Microbiology. 70: 4118-4128.
Freeman, S. and H. Shorney-Darby. 2011. What's the buzz about ceramic membranes? Journal
of American Water Works Association. 103(12): 12-13.
Gall, A., M.A. Page, J.L. Shisler, and B.J. Marinas. 2010. Bacteriophages as surrogates for viral
pathogens in the assessment of suitable drinking water treatment technologies. Proceedings of
the American Water Works Association Water Quality Technology Conference.
Garvey, M., H. Farrell, M. Cormican, and N. Rowan. 2010. Investigations of the relationship
between use of in vitro cell culture-quantitative PCR and a mouse-based bioassay for evaluating
critical factors affecting the disinfection performance of pulsed uv light for treating
Cryptosporidium parvum oocysts in saline. Journal of Microbiology Methods. 80(3): 267-273.
Gates, D., G. Ziglio, and K. Ozekin (Eds.). 2009. State of the science of chlorine dioxide in
drinking water. Water Research Foundation.
Gaulinger, S. 2007. Coagulation pre-treatment for microfiltration with ceramic membranes.
TechneauD. 2.3: 2-1.
Gennaccaro, A.L., M.R. McLaughlin, W. Quintero-Betancourt, D.E. Huffman, and J.B. Rose.
2003. Infectious Cryptosporidium parvum oocysts in final reclaimed effluent. Applied
Environmental Microbiology. 69(8): 4983-4984.
Six-Year Review 3
Technical Support Document for LT2
9-8
December 2016

-------
Gollnitz, W.D., J.L. Clancy, J.B. McEwen, and S.C. Garner. 2005. Riverbank filtration for
IESWTR compliance. Journal of American Water Works Association. 97(12): 64-76.
Gollnitz, W.D., J.L. Clancy, M. Cunnane, and B. Beauchene. 2007. Riverbank filtration for
SWTR compliance. Proceedings of the 2007 Water Quality Technology Conference, American
Water Works Association.
Graczyk, T. K., R. Fayer, R. Knight, B. Mhangami-Ruwende, J. M. Trout, A. J. Da Silva, and N.
J. Pieniazek. 2000. Mechanical transport and transmission of Cryptosporidium parvum oocysts
by wild filth flies. American Journal of Tropical Medicine and Hygiene. 63(3, 4): 178-183.
Guo, H., X. Chu, and J. Hu. 2010. Effect of host cells on low- and medium-pressure UV
inactivation of adenoviruses. Applied and Environmental Microbiology. 76(21): 7068-7075.
Haas, C.N., C.S. Crockett, J.B. Rose, C.P. Gerba, and A.M. Fazil. 1996. Assessing the risk posed
by oocysts in drinking water. Journal of American Water Works Association. 88(9): 131-136.
Hall, T., J. Pressdee, and E. Carrington. 1994. Removal of Cryptosporidium oocysts by water
treatment processes. Report No. FR0457. Foundation for Water Research. Bucks, UK.
Hargy, T., T.T. Fong, and R. McCuin. 2011. Bacilluspumilus spores for bioassay validation of
UV reactors for virus disinfection credit. American Water Works Association Water Quality
Technology Conference.
Havelaar, A., M. van Olphen, and J. Schijven. 1995. Removal and inactivation of viruses by
drinking water treatment processes under full scale conditions. Water Science and Technology.
31:55062.
Hayes, S.L., M. Sivaganesan, K.M. White, and S.L. Pfaller. 2008. Assessing the effectiveness of
low-pressure ultraviolet light for inactivating Mycobacterium avium complex (MAC) micro-
organisms. Letters in Applied Microbiology. 47: 386-392.
Heath, M., H. Wright, and D. Gaithuma. 2009. New approaches for optimizing UV validation
testing. Proceedings of the American Water Works Association Water Quality Technology
Conference.
Hijnen, W.A.M., E.F. Beerendonk, and G.J. Medema. 2006. Inactivation credit of UV radiation
for viruses, bacteria and protozoan oocysts in water: A review. Water Research. 40: 3-22.
Hijnen, W.A.M., T.M.H. Suylen, J.A. Bahlman, A. Brouwer-Hanzens, and G. Medema. 2011a.
GAC adsorption filters as barriers for viruses, bacteria, and protozoan oocysts in water treatment.
In: Quantitative Methods to Assess Capacity of Water Treatment to Eliminate Micro-Organisms.
127-140.
Hijnen, W.A.M., Y.J. Dullemont, J.F. Schijven, A.J. Hanzens-Brouwer, M. Rosielle, and G.J.
Medema. 201 lb. Removal and fate of Cryptosporidium parvum, Clostridium perfringens and
small-sized centric diatoms {Stephanodiscus hantzschii) in slow sand filters. In: Quantitative
Methods to Assess Capacity of Water Treatment to Eliminate Micro-Organisms. 95-110.
Six-Year Review 3
Technical Support Document for LT2
9-9
December 2016

-------
Hijnen, W.A.M., L. Heijnen, and G. Medema. 2011c. Survival of Clostridium spores in river
water and in sand from a slow sand filter. In: Quantitative Methods to Assess Capacity of Water
Treatment to Eliminate Micro-Organisms. 141-148.
Hijnen, W.A.M., J.F. Schijven, P. Bonne, A. Visser, and G.J. Medema. 201 Id. Elimination of
viruses, bacteria and protozoan oocysts by slow sand filtration. In: Methods to Assess Capacity of
Water Treatment to Eliminate Micro-Organisms. 87-93.
Hijnen, W.A.M., A.J. van der Verr, J. van Beveren, and G.J. Medema. 201 le. Spores of sulphite-
reducing Clostridia (SSRC) as surrogate for verification of the inactivation capacity of full-scale
ozonation for Cryptosporidium. In: Quantitative Methods to Assess Capacity of Water Treatment
to Eliminate Micro-Organisms. 59-66.
Hinze-Selch, D., W. Daubener, L. Eggert, S. Erdaq, R. Stoltenberg, and S. Wilms. 2007. A
controlled prospective study of toxoplasma gondii infection in individuals with schizophrenia:
beyond seroprevalence. Schizophr Bull. 33(3):782-788.
Hipsey, M. R., J.P. Antenucci, J.D. Brooks, M.D. Burch, R.H. Regel, C. Davies, N.J. Ashbolt,
and C. Ferguson. 2005. Hydrodynamic distribution of pathogens in lakes and reservoirs.
American Water Works Association Research Foundation.
Hoeger S.J., D.R. Dietrich, and B.C. Hitzfeld. 2002. Effect of ozonation on the removal of
cyanobacterial toxins during drinking water treatment. Environmental Health Perspectives.
110(11): 1127-32.
Hoff. J.C. 1986. Inactivation of Microbial Agents by Chemical Disinfectants. Cincinnati, OH:
US Environmental Protection Agency. EPA-600-2-86-067. Available online at:
https://nepis.epa.gov/EPA/html/DLwait.htm?url=/Exe/ZyPDF.cgi/2000TL73.PDF?Dockev=200
0TL73.PDF.
Hoxie, N. J., J. P. Davis, J. M. Vergeront, R. D. Nashold, and K. A. Blair. 1997.
Cryptosporidiosis-associated mortality following a massive waterborne outbreak in Milwaukee,
Wisconsin. American Journal of Public Health. 87(12): 2032-2035.
Hu, X., S. Geng, X. Wang, and C. Hu. 2012. Inactivation and photorepair of enteric pathogenic
microorganisms with ultraviolet irradiation. Environmental Engineering Science. 29: 549-553.
Hubel, R.E. 2007. UV disinfection system downtime, effects and mitigation. Journal of
American Water Works Association. 99(4): 140-147.
Huffman, D.E., A.L. Gennaccaro, T.L. Berg, G. Batzer, G. Widmer, A.L. Gennaccaro, M.R.
McLaughlin, W. Quintero-Betancourt, D.E. Huffman, and J.B. Rose. 2006. Detection of
infectious parasites in reclaimed water. Water Environment Research. 78(12): 2297-2302.
Hunter, P.R. and Q. Syed. 2010. Community surveys of self-reported diarrhoea can dramatically
overestimate the size of outbreaks of waterborne cryptosporidiosis. Water Science and
Technology. 43(12): 27-30.
Six-Year Review 3
Technical Support Document for LT2
9-10
December 2016

-------
Insulander, M., C. Silverlas, M. Lebbad, L. Karlsson, J.G. Mattsson, and B. Svenungsson. 2012.
Molecular epidemiology and clinical manifestations of human cryptosporidiosis in Sweden.
Epidemiology and Infection. 1-12.
Inungu, J.N., A. A. Morse, and C. Gordon. 2000. Risk Factors, seasonality, and trends of
cryptosporidiosis among patients infected with human immunodeficiency virus. American
Journal of Tropical Medicine and Hygiene. 62(3): 384-387.
Jacangelo, J.G., N.L. Patania Brown, A. Madec, K. Schwab, D. Huffman, G. Amy, C. Mysore, J.
Leparc, and A. Prescott. 2006. Micro- and ultrafiltration performance specification based on
microbial removal. American Water Works Association Research Foundation.
Jiang, Y., J. Ren, Z. Yuan, A. Liu, H. Zhao, H. Liu, L. Chu, W. Pan, J. Cao, Y. Lin, and Y. Shen.
2014. Cryptosporidium andersoni as a novel predominant Cryptosporidium species in
outpatients with diarrhea in Jiangsu Province, China. BMC Infectious Diseases. 14: 555.
Johnson, A.M., P. A. Rochelle, and G.D. Di Giovanni. 2010. Detection of infectious
Cryptosporidium in conventionally treated drinking water, Project 3021. Water Research
Foundation.
Johnson, A.M., G.D. Giovanni, and P. A. Rochelle. 2012. Comparison of assays for sensitive and
reproducible detection of cell culture-infectious Cryptosporidium parvum and Cryptosporidium
hominis in drinking water. Applied Environmental Microbiology. 78(1): 156-162.
Kar, S., A. Daugschies, A. Cakmak, N. Yilmazer, K. Dittmar, andB. Bangoura. 2011.
Cryptosporidium parvum oocyst viability and behaviour of the residual body during the
excystation process. Parasitology Research. 109(6): 1719-1723.
Katona, P. and J. Katona-Apte. 2008. The interaction between nutrition and infection. Clinical
Infectious Diseases. 46: 1582-1588.
Keegan, A., D. Daminato, C.P. Saint, and P.T. Monis. 2008. Effect of water treatment processes
on Cryptosporidium infectivity. Water Research. 42(6-7): 1805-1811.
Keller, J.I. and W.G. Shriver. 2014. Prevalence of three Campylobacter species, C. jejuni, C.
coli, and C. lari, using multilocus sequence typing in wild birds of the Mid-Atlantic region,
USA. Journal of Wildlife Diseases. 50(1): 31-41.
Kelley, M., P. Warrier, J. Brokaw, K. Barrett, and S. Komisar. 1995. A study of two U.S. Army
installation drinking water sources and treatment systems for the removal of Giardia and
Cryptosporidium. Proceedings of the Annual Conference of the American Water Works
Association, Denver, CO.
Kramer, M.H., B.L. Herwaldt, G.F. Craun, R.L. Calderon, andD.D. Juranek. 1996. Waterborne
disease: 1993 and 1994. Journal of American Water Works Association. 88(3): 66-80.
Six-Year Review 3
Technical Support Document for LT2
9-11
December 2016

-------
Kreuger, W.S., E.D. Hilborn, R.R. Converse, and T.J. Wade. 2014. Drinking water source and
human Toxoplasma gondii infection in the United States: a cross-sectional analysis of NHANES
data. BMC Public Health. 14:711
Kuhn, R.C., C.M Rock, and K.H. Oshima. 2002. Effect of pH and magnetic material on
immunomagnetic separation of Cryptosporidium oocysts from concentrated water samples.
Applied and Environmental Microbiology. 68: 2066-2070.
Lalancette, C., G.D. Di Giovanni, and M. Prevost. 2010. Improved risk analysis by dual direct
detection of total and infectious Cryptosporidium oocysts on cell culture in combination with
immunofluorescence assay. Applied Environmental Microbiology. 76(2): 566-577.
Lanao, M., M.P. Ormad, C. Ibarz, N. Miguel, and J.L. Ovelleiro. 2008. Bactericidal effectiveness
of O3, O3/H2O2 and 03/Ti02 on Clostridiumperfringens. Ozone: Science and Engineering. 30(6):
431-438.
Lange, H., O.H. Johansen, L. Void, L.J. Robertson, I.L. Anthonisen, and L. Nygard. 2014.
Second outbreak of infection with a rare Cryptosporidium parvum genotype in schoolchildren
associated with contact with lambs/goat kids at a holiday farm in Norway. Epidemiology and
Infection. 142(10): 2105-2113.
Langlet, J., F. Gaboriaud, J.F.L. Duvald, and C. Gantzera. 2008. Aggregation and surface
properties of f-specific RNA phages, implication for membrane filtration processes. Water
Research. 42: 2769-2777.
LeChevallier, M., W. Norton, and T. Atherholt. 1997. Protozoa in open reservoirs. Journal of
American Water Works Association. 89(9): 84-96.
LeChevallier, M.W., G.D. DiGiovanni, J.L. Clancy, Z. Bukhari, S. Bukhari, J.S. Rosen, J.
Sobrinho, and M.M. Frey. 2003. Comparison of method 1623 and cell-culture PCR for detection
of Cryptosporidium spp. in source waters. Journal of Applied Environmental Microbiology.
69(2):971-979.
LeChevallier, M. and K. Au. 2004. Water treatment and pathogen control: Process efficiency in
achieving safe drinking water. World Health Organization and International Water Association
Publishing. Available at: http://apps.vvho.int/iris/bitstream/10665/42796/1/9241562552.pdf.
Lee, S.U., M. Joung, D.J. Yang, S.H. Park, S. Huh, W.Y. Park, and J.R. Yu. 2008. Pulsed-UV
light inactivation of Cryptosporidium parvum. Parasitology Research. 102: 1293-1299.
Lenz, M., M. Bell, and B. Carrico. 2008. Uncovered finished water reservoirs: Reflections,
regulations and resolutions. Journal of American Waterworks Association. 100(9): 90-101.
Li, D., S.A. Craik, D.W. Smith, and M. Belosevic. 2008. Survival of Giardia lamblia
trophozoites after exposure to UV light. Federation of European Microbiological Societies
Microbiology Letters. 278: 56-61.
Six-Year Review 3
Technical Support Document for LT2
9-12
December 2016

-------
Li, X., E.R. Atwill, L.A. Dunbar, and K.W. Tate. 2010a. Effect of daily temperature fluctuation
during the cool season on the infectivity of Cryptosporidium parvum. Applied Environmental
Microbiology. 76(4): 989-993.
Li, J., K. Hirota, H. Yumoto, T. Matsuo, Y. Miyake, and T. Ichikawa. 2010b. Enhanced
germicidal effects of pulsed UV-LED irradiation on biofilms. Applied Microbiology. 109: 2183—
2190.
Linden, K G., G.A. Shin, G. Faubert, W. Cairns, and M.D. Sobsey. 2002. UV disinfection of
giardia lamblia cysts in water. Environmental Science and Technology. 36(11): 2519-2522.
Linden, K.G., J. Thurston, R. Schaefer, and J.P. Malley. 2007. Enhanced UV inactivation of
adenoviruses under polychromatic UV lamps. Applied and Environmental Microbiology. 73(23):
7571-7574.
Linden, K.G., G. Shin, J. Lee, K. Scheible, C. Shen, and P. Posy. 2009. Demonstrating 4-Log
adenovirus inactivation in a medium-pressure UV disinfection reactor. Journal of American
Water Works Association. 101(3): 90-99.
Linden, K.G., K. Scheible, and P. Posy. 2011. Regulatory implications of new findings on UV
disinfection of adenovirus in drinking water. Proceedings of the American Water Works
Association Water Quality Technology Conference.
Logsdon, G.S., Frev, M M., Stefanich, T.D., Johnson, S.L., Feelv, D.E., Rose, J.B. and Sobsey,
M.D. 1994. Removal and disinfection efficiency of lime softening processes for Giardia and
viruses. American Water Works Association.
Los Angeles Department of Water and Power (LADWP). 2012. Personal communication.
Mach, R., J. Condon, and J. Johnson. 2003. Sioux City Riverbank Filtration Study. European
Geographical Society-American Geophysical Union-European Union of Geosciences Joint
Assembly. Nice, France, April 6-11.
MacKenzie, W.R., N.J. Hoxie, M.E. Proctor, M.S. Gradus, K.A. Blair, D.E. Peterson, J.J.
Kazmierczak, D.G. Addiss, K.R. Fox, J.B. Rose, and J.P. Davis. 1994. A massive outbreak in
Milwaukee of Cryptosporidium infection transmitted through the public water supply. New
England Journal of Medicine. 331: 161-167.
Mahmud, F., S.A. Craik, and M. Belosevic. 2006. The effect of upstream treatment processes on
the UV inactivation of Cryptosporidium parvum. Proceedings of the 2006 Water Quality
Technology Conference of American Water Works Association.
Majewska, A.C., T.K. Graczyk, A. Slodkowicz-Kowalska, L. Tamang, S. Jedrzejewski, P.
Zduniak, P. Solarczyk, A. Nowosad, and P. Nowosad. 2008. The role of free-ranging, captive,
and domestic birds of western Poland in environmental contamination with Cryptosporidium
parvum oocysts and Giardia lamblia cysts. Parasitology Research. 104: 1093-1099.
Six-Year Review 3
Technical Support Document for LT2
9-13
December 2016

-------
McTigue, N.E. and D.A. Cornwell. 2013. Large utilities' use of the microbial toolbox for
LT2ESWTR compliance. Journal of American Waterworks Association. 105(8): 47-48.
Mead, P.S., L. Slutsker, V. Dietz, L.F. McCaig, J.S. Bresee, C. Shapiro, P.M. Griffin, and R.V.
Tauxe. 1999. Food-related illness and death in the United States. Emerging Infectious Diseases.
5(5): 607-625.
M edema, G.J., M.H.A. Juhasz-Holterman, and J. A. Luijten. 2000. Removal of micro-organisms
by bank filtration in a gravel-sand soil. Proceedings of the International Riverbank Filtration
Conference.
Messner, M.J., C.L. Chappell, and P C. Okhuysen. 2001. Risk assessment for Cryptosporidium:
a hierarchical Bayesian analysis of human dose response data. Water Research. 35( 16): 3934-
3940.
Messner, M.J., P. Berger, and S.P. Nappier. 2014. Fractional poisson—a simple dose-response
model for human norovirus. Risk Analysis. 34( 10): 1820-1829.
Mitchell-Blackwood, J. 2010. Using Analytic models for risk-based responses to pathogenic
agents in the environment. Philadelphia, PA: Drexel University.
Moran, D. 2013. Personal Communication.
Muhammad, N., R. Sinha, R.E. Krishnan, H. Piao, C.L. Patterson, J. Cotruvo, S.L. Cumberland,
V.P. Nero, and C. Delandra. 2008. Evaluating surrogates for Cryptosporidium removal in point-
of-use systems. Journal of American Water Works Association. 100(12): 98-106.
Muhammad, N., R. Sinha, R.E. Krishnan, C.L. Patterson, R.C. Haught, H.H. Harms, and R.
Seville. 2010. Evaluating a composite cartridge for small system drinking water treatment. Water
and Health. 8(0): 212-223.
Murphy, H.M., S.J. Payne, and G.A. Gagnon. 2008. Sequential UV- and chlorine-based
disinfection to mitigate Escherichia coli in drinking water biofilms. Water Research. 42: 2083-
2092.
National Drinking Water Advisory Council (NDWAC). 2000. Working Group Meeting On
Contaminant Candidate List Regulatory Determinations and the 6-Year Review of Existing
Regulations. Available online at: https://www.epa.gov/sites/production/files/2015-
1 1/documents/march 1 2 2000 meeting on ccl and 6 year review.pdf.
National Research Council. 1997. Safe water from every tap: Improving Water service to small
communities. Washington, D.C.: National Academy Press.
Naumova, E.N, A.I. Egorov, R.D. Morris, and J.K. Griffiths. 2003. The elderly and waterborne
Cryptosporidium infection: Gastroenteritis hospitalizations before and during the 1993
Milwaukee Outbreak. Emerging Infectious Diseases. 9(4): 418-425.
Six-Year Review 3
Technical Support Document for LT2
9-14
December 2016

-------
Ndong, M., D. Bird, A. Zamyadi, M. Prevost, and S. Dorner. 2011. Cyanobacterial
contamination, hydrodynamic effect on the vulnerability of drinking water intake. Proceedings of
the 2011 Water Quality Technology Conference of American Water Works Association.
Newcombe, G. 2009. International guidance manual for the management of toxic cyanobacteria.
Global Water Research Coalition.
New York City Department of Environmental Protection (NYCDEP). 2012. Personal
Communication.
NYCDEP. 2013. New York City Department of Environmental Protection Waterborne Disease
Risk Assessment Program. Available at:
http://www.nyc.gov/html/dep/html/drinking water/wdrap.shtml
NYCDEP. 2015. The New York Municipal Water Finance Authority Fiscal Year 2015
Consulting Engineer's Report. Available at:
http://www.nvc.gov/html/nyw/pdf/fv2015 consulting engineers.pdf
The New York City Department of Health and Mental Hygiene (NYCDOHMH) and NYCDEP.
2015. Waterborne Disease Risk Assessment Program 2014 Annual Report. Available at:
http://www.nvc,gov/html/dep/pdf/reports/fad 8.1 waterborne disease risk assessment progra
m - 2014 annual report 03-15.pdf
Nicholson, E., A. Chen, D. Euler, A. Niblock, E. Kiefer, T. Elliott, and P. Swaim. 2011.
Operation, maintenance, and reporting activities for municipal drinking water UV disinfection
facilities. Proceedings of the 2011 Water Quality Technology Conference of American Water
Works Association.
Nwachuku, N. and C. P. Gerba. 2006. Health risks of enteric viral infections in children.
In Reviews of environmental contamination and toxicology. 1-56. Springer New York.
Okhuysen, P.C., C.L. Chappell, C.R. Sterling, W. Jakubowski, and H. L. DuPont. 1998.
Susceptibility and serologic response of healthy adults to reinfection with Cryptosporidium
parvum. Infection and Immunity. 66(2): 441-443.
Okhuysen, P.C., C.L. Chappell, J.H. Crabb, C.R. Sterling, and H.L. DuPont. 1999. Virulence of
three distinct Cryptosporidium parvum isolates for healthy adults. Journal of Infectious Diseases.
180(4): 1275-1281.
Okhuysen, P.C., S.M. Rich, C.L. Chappell, K.A. Grimes, G. Widmer, X. Feng, and S. Tzipori,
2002. Infectivity of a Cryptosporidium parvum isolate of cervine origin for healthy adults and
interferon-gamma knockout mice. Journal of Infectious Diseases. 185(9): 1320-1325.
Ongerth, J.E. 2013. The concentration of Cryptosporidium and Giardia in water - the role and
importance of recovery efficiency. Water Research. 47(7): 2479-2488.
Six-Year Review 3
Technical Support Document for LT2
9-15
December 2016

-------
Ou, H., N. Gao, Y. Deng, J. Qiao, and H. Wang. 2012. Immediate and long-term impacts of uv-c
irradiation on photosynthetic capacity, survival and microcystin-LR release risk of Microcystis
aeruginosa. Water Research. 46: 1241-1250.
Pang, L., M. Close, M. Goltz, M. Noonan, and L. Sinton. 2005. Filtration and transport of
Bacillus subtilis spores and the F-RNA phage MS2 in a course alluvial gravel aquifer:
Implications in the estimation of setback distances. Journal of Contaminant Hydrology. 77: 165-
194.
Pang, L., U. Nowostawska, L. Weaver, G. Hoffman, A. Karmacharya, A. Skinner, and N. Karki.
2012. Biotin- and glycoprotein-coated microspheres, potential surrogates for studying filtration
of Cryptosporidium parvum in porous media. Environmental Science and Technology. 46:
11779-11787.
Patania, N.L., J.G. Jacangelo, L. Cummings, A. Wilczak, K. Riley, and J. Oppenheimer. 1995.
Optimization of Filtration for Cyst Removal. American Water Works Association Research
Foundation. Denver, CO.
Payment, P. and E. Franco. 1993. Clostridiumperfringens and somatic coliphages as indicators
of the efficiency of drinking water treatment for viruses and protozoan cysts. Applied and
Environmental Microbiology. 59(8): 2418-2424.
Pereira, J.T., A.O. Costa, S.B. de Oliveira, M.B. Silva, W. Schuchard, S.C. Osaki, E.A. de
Castro, R.C. Paulino, and V.T. Soccol. 2008. Comparing the efficacy of chlorine, chlorine
dioxide, and ozone in the inactivation of Cryptosporidium parvum in water from Parana State,
Southern Brazil. Applied Biochemistry and Biotechnology. 151: 464-473.
Petri, B. and C. Odegaard. 2008. UV reactor challenges with a high resistance surrogate for
adenovirus credit. Proceedings of the 2008 Water Quality Technology Conference of American
Water Works Association.
Petri, B., S. Hayes, A. Festger, O.K. Schieble, C. Shen, P. Patil, C. Odegaard, and I.
Gobulukoglu. 2011. Use of a high-resistance challenge organism for validation of low pressure,
high output UV reactors for virus inactivation. Proceedings of the 2011 Water Quality
Technology Conference of American Waterworks Association.
Phillip, D.A.T., S.C. Rawlins, S. Baboolal, R. Gosein, C. Goddard, and G. Legall. 2008. Relative
importance of the various environmental sources of Cryptosporidium oocysts in three
watersheds. Journal of Water and Health. 61: 23-34.
Pittsburgh Water and Sewer Authority (PWSA). 2013. Personal communication with the
Pittsburgh Water and Sewer Authority.
Pope, M.L., B. Ellis, J.S. Rosen, K. Connell, J. Pulz, M. LeChevallier, C. Rodgers, S. Regli, and
D. Schmelling. 2002. Using /•]. coli to indicate source water susceptibility to high concentrations
of Cryptosporidium. In: Information Collection Rule Data Analysis. American Water Works
Association.
Six-Year Review 3
Technical Support Document for LT2
9-16
December 2016

-------
Pouillot, R., P. Beaudeau, J.B. Denis, and F. Derouin. 2004. A quantitative risk assessment of
waterborne cryptosporidiosis in France using second-order Monte Carlo simulation. Risk
Analysis. 241: 1-17.
Pozio, E., G. Rezza, A. Boschini, P. Pezzotti, A. Tamburrini, P. Rossi, M. Di Fine, C. Smacchia,
A. Schiesari, E. Gattei, R. Zucconi, and P. Ballarini. 1997. Clinical cryptosporidiosis and human
immunodeficiency virus (HlV)-induced immunosuppression findings from a longitudinal study of
HIV-positive and HIV-negative former injection drug users. Journal of Infectious Diseases. 176:
969-975.
Rajagopalan, N. 2001. Field evaluation of ceramic microfiltration membranes in drinking water
treatment. Small Systems Technical Assistance Center (TAC), Montana University System:
Water Center.
Rehn, M., A. Wallensten, M. Widerstrom, M. Lilja, M. Grunewald, S. Stenmark, M. Kark, and J.
Lindh. 2015. Post-infection symptoms following two large waterborne outbreaks of
Cryptosporidium hominis in northern Sweden, 2010-2011. BMC Public Health 15: 529.
Reynolds, K. A. 2013. One bad bug: The potential for waterborne spread of MRS A. The
Aquifer. Available at: http://carbonvvaters.oru/vvp-content/uploads/2013/02/The-Potential-for-
Waterborne-Spread-of-MRSA.pdf
Rhodes, E.R., L.F. Villegas, N.J. Shaw, C. Miller, and E.N. Villegas. 2012. A modified EPA
method 1623 that uses tangential flow hollow-fiber ultrafiltration and heat dissociation steps to
detect waterborne Cryptosporidium and Giardia spp. Journal of Visualized Experiments. 65:
4177.
Rice, E.W., K.R. Fox, R.J. Miltner, D.A. Lytle, and C.H. Johnson. 1996. Evaluating plant
performance with endospores. Journal of American Water Works Association. 88(9): 122.
Rochelle, P. A., D. M. Ferguson, T.J. Handojo, R. De Leon, M.H. Stewart, and R.L. Wolfe.
1997. An assay combining cell culture with reverse transcriptase PCR to detect and determine
the infectivity of waterborne Cryptosporidium parvum. Applied and Environmental
Microbiology. 635: 2029-2037.
Rochelle, P.A., D.M. Ferguson, A.M. Johnson, and R. De Leon. 2001. Quantitation of
Cryptosporidium parvum infection in cell culture using a colorimetric in situ hybridization assay.
Journal of Eukaryotic Microbiology. 485: 565-574.
Rochelle, P.A., E.R. Blatchley, III, P.S. Chan, O.K. Scheible, and C. Shen. 2010. Challenge
Organisms for Inactivation of Viruses by Ultraviolet Treatment. Water Research Foundation.
Rochelle, P.A., A.M. Johnson, R. De Leon, and G.D. Giovanni. 2012. Assessing the risk of
infectious Cryptosporidium in drinking water. Journal of American Water Works Association.
1045: E325-E336.
City of Rochester. 2013. Personal Communication.
Six-Year Review 3
Technical Support Document for LT2
9-17
December 2016

-------
Rodriguez, E., G.D. Onstad, T.P.J. Kull, J.S. Metcalf, J.L. Acero, and U. von Gunten. 2007.
Oxidative elimination of cyanotoxins, comparison of ozone, chlorine, chlorine dioxide and
permanganate. Water Research. 41: 3381-3393.
Rose, J.B. 1997. Environmental ecology of Cryptosporidium and public health implications.
Annual Review of Public Health. 18: 135-161.
Rosen, J.S., R. McCuin, T. Bartrand, J. Sobrinho, J.L. Clancy, Z. Rodriguez del Rey, A. Richter,
Y. Akagi, P. Westerhoff, C.A. Chiu, and J. Irving. 2014. Matrix effects on Cryptosporidium
oocyst recovery [Project #4348], Water Research Foundation.
Sakai, H., H. Katayama, K. Oguma, and S. Ohgaki. 2011. Effect of photoreactivation on
ultraviolet inactivation of Microcystis aeruginosa. Water Science and Technology. 636: 1224-
1229.
Sarkar, P. and C.P. Gerba. 2012. Inactivation of Naegleria fowleri by chlorine and ultraviolet
light. Journal of American Water Works Association. 1043: E173-E180.
Scheible, O. K., C. Shen, and E. R. Blatchley. 2008. Validation of UV reactors by Lagrangian
actinometry using dyed microspheres. In Proceedings of the Water Environment Federation,
WEFTEC. 3816-3829.
Schets, F.M., G.B. Engels, M. During, and A.M. de Roda Husman. 2005. Detection of infectious
Cryptosporidium Oocysts by cell culture immunofluorescence assay: Applicability to
environmental samples. Applied and Environmental Microbiology. 7111: 6793-6798.
Schuler, P.F. and M M. Ghosh. 1991. Slow sand filtration of cysts and other particulates.
Proceedings of the American Water Works Association Annual Conference. 235-252.
Seattle Public Utilities (SPU). 2012. Personal Communication.
Shaw, N.J., L.F. Villegas, B.J. Eldred, D.H. Gaynor, P.S. Warden, and B.V. Pepich. 2008.
Modification to EPA method 1623 to address a unique seasonal matrix effect encountered in
some U.S. source waters. Journal of Microbiology Methods. 75: 445-448.
Sheoran, A., A. Wiffin, G. Widmer, P. Singh, and S. Tzipori. 2012. Infection with
Cryptosporidium hominis provides incomplete protection of the host against Cryptosporidium
parvum. Journal of Infectious Diseases. 2056: 1019-1023.
Shin, G. A., J.K. Lee, and K.G. Linden. 2009a. Enhanced effectiveness of medium-pressure
ultraviolet lamps on human adenovirus 2 and its possible mechanism. Water Science and
Technology. 60(4): 851-857.
Shin, G.A., K.G. Linden, and G. Faubert. 2009b. Inactivation of Giardia lamblia cysts by
polychromatic UV. Letters in Applied Microbiology. 48: 790-792.
Sifuentes, L.Y. and G.D. Di Giovanni. 2007. Aged HCT-8 cell monolayers support
Cryptosporidium parvum infection. Applied and Environmental Microbiology. 7323: 7548-7551.
Six-Year Review 3
Technical Support Document for LT2
9-18
December 2016

-------
Signor, R.S., D.J. Roser, N.J. Ashbolt, and J.E. Ball. 2005. Quantifying the impact of runoff
events on microbiological contaminant concentration entering surface drinking source waters.
Journal of Water and Health. 34: 453-468.
Sivaganesan, M. and B.J. Marinas. 2005. Development of a Ct equation taking into consideration
the effect of lot variability on the inactivation of Cryptosporidium parvum oocysts with ozone.
Water Research. 39(11): 2429-2437.
Sivaganesan, M. and S. Sivaganesan. 2005. Effect of lot variability on ultraviolet radiation
inactivation kinetics of Cryptosporidium parvum oocysts. Environmental Science and
Technology. 39(11): 4166-4171.
Skotarczak, B. 2010. Progress in the molecular methods for the detection and genetic
characterization of Cryptosporidium in water samples. [Research Support, Non-U. S. Gov't
Review], Annals of Agricultural and Environmental Medicine. 171: 1-8.
Smeets, P.W.M.H., A.W.C. van der Helm, Y.J. Dullemont, L.C. Rietveld, J.C. van Dijk, and G.J.
Medema. 2006. Inactivation of Escherichia coli by ozone under bench-scale plug flow and full-
scale hydraulic conditions. Water Research. 40: 3239-3248.
Smith, H., R. Nichols, and A. Grimason. 2005 Cryptosporidium excystation and invasion:
Getting to the guts of the matter. Trends in Parasitology. 213: 133-142.
Smith, H., J. Cawson, G. Sheridan, and P. Lane. 2011. Desktop review - Impact of bushfires on
water quality. Australian Government Department of Sustainability, Environment, Water,
Population and Communities.
Staggs, S.E., E.M. Beckman, E.M. Beckman, S.P. Keely, R. Mackwan, M.W. Ware, A.P. Moyer,
J.A. Ferretti, A. Sayed, L. Xiao, and E.N. Villegas. 2013. The applicability of TaqMan-based
quantitative real-time PCR assays for detecting and enumerating Cryptosporidium spp. oocysts
in the environment. PLOS ONE. 86: e66562.
States, S., K. Stadterman, L. Ammon, P. Vogel, J. Baldizar, D. Wright, L. Conley and J. Sykora.
1997. Protozoa in river water: Sources, occurrence, and treatment. Journal of American Water
Works Association. 89:(9)74-83.
Sun, W. and W. Liu. 2009. A pilot-scale study on ultraviolet disinfection system for drinking
water. Journal of Water Supply: Research and Technology - AQUA. 58(5):364-353.
Sunnotel, O., W. J. Snelling, L. Xiao, K. Moule, J.E. Moore, B. Millar, and C.J. Lowery. 2006.
Rapid and sensitive detection of single Cryptosporidium oocysts from archived glass slides.
Journal of Clinical Microbiology. 449: 3285-3291.
Tacoma Public Utilities (TPU). 2010. Recommendation for Green River Water Treatment
Memorandum.
TPU. 2012. Personal Communication.
Six-Year Review 3
Technical Support Document for LT2
9-19
December 2016

-------
Talbot, C., M. Health, H. Wright, and D. Peters. 2011. 12-month UV fouling study on unfiltered
source water. Proceedings of the 2011 Water Quality Technology Conference of American
Water Works Association.
Teunis, P.F., C.L. Chappell, andP.C. Okhuysen. 2002a. Cryptosporidium dose-response studies:
Variation between isolates. Risk Analysis. 221: 175-183.
Teunis, P.F., C.L. Chappell, andP.C. Okhuysen. 2002b. Cryptosporidium dose-response studies:
Variation between hosts. Risk Analysis. 223: 475-485.
Theodos, C.M., J.K. Griffiths, J. D'Onfro, A. Fairfield, and S. Tzipori. 1998. Efficacy of
nitazoxanide against Cryptosporidium parvum in cell culture and in animal models.
Antimicrobial Agents and Chemotherapy. 428: 1959-1965.
Thurston-Enriquez, J. A., C. N. Haas, J. Jacangelo, and C. P. Gerba. 2005. Inactivation of enteric
adenovirus and feline calicivirus by chlorine dioxide. Applied and Environmental Microbiology.
716: 3100-3105.
Timms, S., J. S. Slade, and C. R. Flicker. 1995. Removal of Cryptosporidium by slow sand
filtration. Water Science and Technology 3 1(5-6): 81-84.
Unger, M. and R.M. Collins. 2008. Assessing Escherichia coli removal in the schmutzdecke of
slow-rate biofilters. Journal of American Water Works Association. 100(12): 60-73.
United States Census Bureau. 2001. Statistical abstract of the United States: 2001. Washington,
DC. Economics and Statistics Administration.
United States Environmental Protection Agency (USEPA). 1989a. National Primary Drinking
Water Regulations; Filtration, Disinfection; Turbidity, Giardia lamblia, Viruses, Legionella, and
Heterotrophic Bacteria; Final Rule. Part III. 54 FR 27486. June 29, 1989.
USEPA. 1989b. National Interim Primary Drinking Water Regulations; Total Coliform Rule;
Final Rule. Part III. Federal Register, 54 FR 27544. June 29, 1989.
USEPA. 1991. Guidance Manual for Compliance with the Filtration and Disinfection
Requirements for Public Water Systems Using Surface Water Sources. Available at:
https://www.epa.gov/sites/production/files/2015-
10/documents/guidance manual for compliance with the filtration and disinfection reciuirem
ents.pdf
USEPA. 1998. National Primary Drinking Water Regulations; Interim Enhanced Surface Water
Treatment Rule. 63 FR 69477. December 16, 1998.
USEPA. 1999. Uncovered Finished Water Reservoirs Guidance Manual. EPA 815-R-99-011.
USEPA. 2000. Stage 2 Microbial and Disinfection Byproducts Federal Advisory Committee
Agreement in Principle. 65 FR 83015. December 29, 2000. Available at:
https://www.gpo.gov/fdsvs/pkg/FR-2000-12-29/pdf/00-333Q6.pdf
Six-Year Review 3
Technical Support Document for LT2
9-20
December 2016

-------
USEPA. 2001a. Results of the Inter-laboratory Method Validation Study for Determination of
Cryptosporidium and Giardia Using USEPA Method 1623. EPA-821-R-01-028.
USEPA. 2001b. Low Pressure Membrane Filtration for Pathogen Removal: Application,
Implementation, and Regulatory Issues. EPA 815-C-01-001.
USEPA. 2002. National Primary Drinking Water Regulations: Long Term 1 Enhanced Surface
Water Treatment Rule; Final Rule. January 14, 2002. 67 FR 1812. EPA 815-Z-02-001.
USEPA. 2003a. National Primary Drinking Water Regulations; Announcement of Completion of
EPA's Review of Existing Drinking Water Standards; Notice. 68 FR 42908. July 18, 2003.
USEPA. 2003b. National Primary Drinking Water Regulations: Long Term 2 Enhanced Surface
Water Treatment Rule. 68 FR 47640.
USEPA. 2005a. Economic Analysis for the Final Long Term 2 Enhanced Surface Water
Treatment Rule. EPA 815-R-06-001. Available at:
http://nepis.epa.gov/Exe/ZyPDF.cgi/901 S0000.PDF?Dockev=901S0000.PDF.
USEPA. 2005b. Method 1622: Cryptosporidium in Water by Filtration/IMS/FA. EPA 815-R-05-
001.
USEPA. 2005c. Method 1623: Cryptosporidium and Giardia in Water by Filtration/IMS/FA.
EPA-815-R-05-002.
USEPA. 2005d. Occurrence and Exposure Assessment for the Final Long Term 2 Enhanced
Surface Water Treatment Rule. EPA-821-R-06-002.
USEPA. 2005e. Membrane Filtration Guidance Manual. EPA 815-R-06-009.
USEPA. 2006a. National Primary Drinking Water Regulations: Long Term 2 Enhanced Surface
Water Treatment Rule; Final Rule. 71 FR 654.
USEPA. 2006b. National Primary Drinking Water Regulations: Stage 2 Disinfectants and
Disinfection Byproducts Rule. 71 FR 388. January 4, 2006.
USEPA. 2006c. Ultraviolet Disinfection Guidance Manual for the Final Long Term 2 Enhanced
Surface Water Treatment Rule. EPA 815-R-06-007.
USEPA. 2007. Long Term 2 Enhanced Surface Water Treatment Rule Implementation
Guidance. EPA 816-R-07-006.
USEPA. 2009a. EPA Protocol for the Second Review of Existing National Primary Drinking
Water Regulations Updated. EPA 815-B-09-002. Available at:
https://vvvvvv.epa.gov/sites/production/files/2014-12/documents/815b09002.pdf
USEPA. 2009b. The Analysis of Regulated Contaminant Occurrence Data from Public Water
Systems in Support of the Second Six-Year Review of National Primary Drinking Water
Six-Year Review 3
Technical Support Document for LT2
9-21
December 2016

-------
Regulations. EPA 815-B-09-006. Available at: https://www.epa.gov/sites/production/files/2014-
12/docutnents/815b09006.pdf.
USEPA. 2010a. National Primary Drinking Water Regulations; Announcement of the Results of
EPA's Review of Existing Drinking Water Standards and Request for Public Comment and/or
Information on Related Issues. 75 FR 15500. March 29, 2010.
USEPA. 2010b. Memorandum: OGWDW Review of Small System Monitoring Requirements
Under the Long Term 2 Enhanced Surface Water Treatment Rule.
USEPA. 2010c. Long Term 2 Enhanced Surface Water Treatment Rule: Toolbox Guidance
Manual. EPA 815-R-09-016.
USEPA. 201 la. Improving Our Regulations: Final Plan for Periodic Retrospective Reviews of
Existing Regulations. August 2011. Available online at:
https://www.whitehouse.gov/sites/default/files/other/2011-regulatory-action-
plans/environmentalprotectionagencvregulatoryreformplanaugust2011 .pdf.
USEPA. 201 lb. Long Term 2 Enhanced Surface Water Treatment Rule: Cryptosporidium
Analytical Method Improvements and Update on Source Water Monitoring. Public Meeting.
Summary. December 7, 2011. Washington, D.C.
USEPA. 2012a. Method 1623.1: Cryptosporidium and Giardia in Water by Filtration/IMS/FA.
EPA 816-R-12-001.
USEPA. 2012b. Results of the Inter-laboratory Method Validation Study Using U.S.
Environmental Protection Agency Method 1623.1: Cryptosporidium and Giardia in Water by
Filtration/IMS/FA. EPA 816-R-12-002.
USEPA. 2012c. Expedited Approval of Alternative Test Procedures for Analysis of
Contaminants Under the Safe Drinking Water Act; Analysis and Sampling Procedures. 77 FR
125. June 28, 2012.
USEPA. 2012d. Long Term 2 Enhanced Surface Water Treatment Rule: Uncovered Finished
Water Reservoirs. Public Meeting. Summary. April 24, 2012. Washington, D.C.
USEPA. 2016a. Six-Year Review 3 Technical Support Document for Microbial Contaminant
Regulations. EPA-810-R16-010.
USEPA. 2016b. EPA Protocol for the Third Review of Existing National Primary Drinking
Water Regulations. EPA-810-R-16-007.
USEPA. 2016c. LT2ESWTRData Collection and Tracking System (DCTS). Available at:
www, epa. gov/d wsi \ vearrevi ew/revi ew-1 t2-rul e
United States National Library of Medicine. 2015. Campylobacter infection. Available at:
https://medlineplus.gov/encv/article/000224.htm
Six-Year Review 3
Technical Support Document for LT2
9-22
December 2016

-------
Van Beek, C.G.E.M., A H. De Zwart, M. Balemans, J.W. Kooiman, C. Van Rosmalen, H.
Timiner, J. Vandersluys, and P.J. Stuyfzand. 2010. Concentration and size distribution of
particles in abstracted groundwater. Water Research. 44(3): 868-878.
Vance, D.B. 2002. Particulate transport in groundwater part II - bacteria. Originally published in
1995 Edition of National Environmental Journal, 5(1): 25-26. Available at
http: // 2the4. net/p aartb act. htm.
Varughese, E.A., C.L. Bennett-Stamper, L.J. Wymer, and J.S. Yadav. 2014. A New in vitro
model using small intestinal epithelial cells to enhance infection of Cryptosporidium
parvum. Journal of Microbiological Methods. 106: 47-54.
Villegas, L. Fohl, C.C. Brescia, N.J. Shaw, C. Miller, B.V. Pepich, L. Xiao, and E.N. Villegas.
2010. Variations in oocyst recovery associated with the origin of Cryptosporidium isolates.
AWWA International Symposium on Waterborne Pathogens, Manhattan Beach, California (Oral
presentation).
Vonder Haar, T., M. Page, B. Marinas, J. Shisler, C. Boucherie, V. Heim, and F. David. 2010. A
comparative study of the inactivation of adenovirus coxsackievirus with ultraviolet light,
chlorine, and ozone. 2010 Water Quality Technology Conference of American Water Works
Association.
Vyas, A., S.K. Kim, N. Giacomini, J.C. Boothroyd, R.M. Sapolsky. 2007. Behavioral changes
induced by Toxoplasma infection of rodents are highly specific to aversion of cat odors.
Proceedings of the National Academy of Sciences. 104(15): 6442-6447.
Wang, J., R. Song, and S. Hubbs. 2001. Particle removal through riverbank filtration process. W.
Julich, and J. Schubert (eds). Proceedings of the International Riverbank Filtration Conference,
Dusseldorf, Germany, Internationale Arbeitsgemeinschaft der Wasserwork in
Rheineinzugsgebiet.
Ware, M.W., L. Wymer, H.A. Lindquist, and F.W. Schaefer. 2003. Evaluation of an alternative
IMS dissociation procedure for use with method 1622: Detection of Cryptosporidium in
Water. Journal of Microbiological Methods. 55(3): 575-583.
Ware, M.W., S.A.J. Augustine, D.O. Erisman, M.J. See, L. Wymer, S.L. Hayes, P. Dubey, and
E.N. Villegas. 2010. Determining UV inactivation of Toxoplasma gondii oocysts by using cell
culture and a mouse bioassay. Applied and Environmental Microbiology. 7615: 5140-5147.
Ware, M.W., S.P. Keely, and E.N. Villegas. 2013. Development and evaluation of an off-the-
slide genotyping technique for identifying Giardia cysts and Cryptosporidium oocysts directly
from US EPA method 1623 slides. Journal of Applied Microbiology. 115(1): 298-309.
Wilkes, G., N.J. Ruecker, N.F. Neumann, V.P.J. Gannon, C. Jokinen, M. Sunohara, E. Topp,
K.D.M. Pintar, T.A. Edge, and D.R. Lapen. 2013. Spatiotemporal analysis of Cryptosporidium
species/genotypes and relationships with other zoonotic pathogens in surface water from mixed-
use watersheds. Applied and Environmental Microbiology. 792: 434-448.
Six-Year Review 3
Technical Support Document for LT2
9-23
December 2016

-------
WHO. 2004. Guidelines for drinking-water quality: Recommendations (Vol. 1). World Health
Organization. WHO Geneva.
WHO. 2009. Risk Assessment of Cryptosporidium in Drinking-Water. WHO/HSE/WSH/09.04.
WHO Geneva.
WHO. 2011. Campylobacter. Fact Sheet No. 255.
http://vvvvvv.vvho.int/mediacentre/factsheets/fs255/en/
Wright, H., D. Gaithuma, C. Fonseca, J. Clancy, T. Hargy, K. Fallon, A. Cabaj, A.
Schmalwieser, A. Bierman, and C. Gribbin. 2007. Optimization oflJVdisinfection. AWWA
Research Foundation.
Wright, H., D. Gaithuma, T. Dzurny, C. Schulz, K. McCurdy, T. Bogan, A. Cabaj, A.
Schmalwieser, Y. Ohno, and T. Larason. 2009. Design and performance guidelines for UV
sensor systems. Water Research Foundation and Capital Regional District Water Department.
Wright, H., C. Odegaard, S. Hess, and K. Bircher. 2011. Impacts of reflection on UV dose
delivery-collimated beam apparatus and UV reactors. Proceedings of the Water Environment
Federation. 2011(3): 121-126.
Xiao, L. and U.M. Ryan. 2008. Molecular epidemiology. In: Cryptosporidium and
Cryptosporidiosis, 2nd Ed. CRC Press.
Yakub, G.P. andK.L. Stadterman-Knauer. 2004. Immunomagnetic separation of pathogenic
organisms from environmental matrices. Methods in Molecular Biology. 268: 189-197.
Yang, R., C. Murphy, Y. Song, J. Ng-Hublin, A. Estcourt, N. Hijjawi, R. Chalmers, S. Hadfield,
A. Bath, C. Gordon, and U. Ryan. 2013. Specific and quantitative detection and identification of
Cryptosporidium hominis and C. parvum in clinical and environmental samples. Experimental
Parasitology. 135: 142-147.
Yates, M.V., J. Malley, P. Rochelle, and R. Hoffman. 2006. Effect of adenovirus resistance on
UV disinfection requirements: A report on the state of adenovirus science. Journal of American
Waterworks Association. 986: 93-106.
Zahedi, A., A. Paparini, F. Jian, I. Robertson, and U. Ryan. 2015. Public health significance of
zoonotic Cryptosporidium species in wildlife: Critical insights into better drinking water
management. International Journal for Parasitology: Parasites and Wildlife. 5(1): 88-109.
Zimmer-Thomas, J. L., R.M. Slawson, and P.M. Huck. 2007. A comparison of DNA repair and
survival of Escherichia coli 0157:H7 following exposure to both low- and medium pressure UV
irradiation. Water and Health. 53: 407-415.
Six-Year Review 3
Technical Support Document for LT2
9-24
December 2016

-------
Appendix A
Data for Methods 1623 and 1623.1
Cryptosporidium Recoveries
Six-Year Review 3
Technical Support Document for LT2
December 2016

-------
Exhibit A.1 Single-Laboratory Comparison of Method 1623 and Method 1632.1 in
Reagent Water
Method 1623
Method 1623.1

Number
Std Dev


Number
Std Dev


of
of


of
of

Oocysts
Enumerated
Oocysts
Spiked
Spiked
Oocysts
Percent
Recovery
Oocysts
Enumerated
Oocysts
Spiked
Spiked
Oocysts
Percent
Recovery
132
149.04
1.93
88.6
127
149.04
1.93
85.2
80
119.50
1.74
66.9
91
119.50
1.74
76.2
135
178.83
2.13
75.5
123
178.83
2.13
68.8
130
178.83
2.13
72.7
133
178.83
2.13
74.4
141
178.83
2.13
78.8
141
178.83
2.13
78.8
101
139.75
1.54
72.3
102
160.15
1.46
63.7
115
160.15
1.46
71.8
120
160.15
1.46
74.9
113
160.15
1.46
70.6
117
160.15
1.46
73.1
127
150.33
2.02
84.5
108
130.58
1.08
82.7
Mean
72.7
Mean
75.3
Median
75.7
Median
74.9
Standard deviation
7.0
Standard deviation
6.6
Exhibit A.2 Observed Recovery from One Source Water and Three Artificial
Matrices, Single-Laboratory
Method 1623
Method 1623.1
Oocyst
Count
Spike #
% Recovery
Oocyst
Count
Spike #
%
Recovery
Diatomaceous Earth
48
100
48
68
100
68
48
100
48
52
100
52
50
100
50
63
100
63
36
100
36
49
100
49
74
100
74
76
100
76
58
100
58
66
100
66
68
100
68
65
100
65
61
100
61
73
100
73
Mean
55
Mean
64
Standard Deviation
12
Standard Deviation
9
Tennessee River Sediment
57
100
57
78
100
78
33
100
33
74
100
74
31
100
31
79
100
79
53
100
53
75
100
75
46
100
46
66
100
66
56
100
56
68
100
68
Six-Year Review 3
Technical Support Document for LT2
A-l
December 2016

-------
Method 1623
Method 1623.1
Oocyst
Count
Spike #
% Recovery
Oocyst
Count
Spike #
%
Recovery
52
100
52
74
100
74
41
100
41
65
100
65
Mean
46
Mean
72
Standard Deviation
10
Standard Deviation
5
Clay
0
101
0
47
101
47
0
101
0
44
101
44
1
101
1
32
101
32
0
101
0
39
101
39
0
100
0
52
100
52
2
100
2
44
100
44
1
100
1
55
100
55
1
100
1
47
100
47
-
-
-
38
100
38
-
-
-
53
100
53
Mean
1
Mean
45
Standard Deviation
1
Standard Deviation
7
Ohio River
41
100
41
66
100
66
60
100
60
77
100
77
46
100
46
68
100
68
54
100
54
66
100
66
48
100
48
53
100
53
60
100
60
79
100
79
63
100
63
72
100
72
57
100
57
78
100
78
Mean
54
Mean
70
Standard Deviation
8
Standard Deviation
9
Source: Chapter 5, Exhibit 5.1
Six-Year Review 3
Technical Support Document for LT2
A-2
December 2016

-------
Exhibit A.3 Observed Recovery from Nine Source Waters, Single-Laboratory
Method 1623
Method 1623.1
Oocyst
Count
Spike
#
Spike
Std Dev
%
Recovery
Oocyst
Count
Spike
#
Spike
Std Dev
%
Recovery
Ohio River; nephelometric turbidity units (NTU) = 10.6
41
100.00
1.15
41.0
66
100.00
1.15
66.0
60
100.00
1.15
60.0
77
100.00
1.15
77.0
46
100.00
1.15
46.0
68
100.00
1.15
68.0
54
100.00
1.15
54.0
66
100.00
1.15
66.0
48
100.00
1.15
48.0
53
100.00
1.15
53.0
60
100.00
1.15
60.0
79
100.00
1.15
79.0
63
100.00
1.15
63.0
72
100.00
1.15
72.0
57
100.00
1.15
57.0
78
100.00
1.15
78.0
Mean
53.6
Mean
69.9
Standard Deviation
7.8
Standard Deviation
8.6
Texas-1; NTU = 0.8
73
119.50
1.74
61.1
92
119.50
1.74
77.0
75
119.50
1.74
62.8
97
119.50
1.74
81.2
79
119.50
1.74
66.1
88
119.50
1.74
73.6
65
119.50
1.74
54.4
90
119.50
1.74
75.3
Mean
61.1
Mean
76.8
Standard Deviation
4.9
Standard Deviation
3.2
North Carolina; NTU = 7.4
45
178.83
2.13
25.2
110
178.83
2.13
61.5
51
178.83
2.13
28.5
111
178.83
2.13
62.1
26
178.83
2.13
14.5
103
178.83
2.13
57.6
22
178.83
2.13
12.3
131
178.83
2.13
73.3
Mean
20.1
Mean
63.6
Standard Deviation
7.9
Standard Deviation
6.7
Michigan; NTU = 0.1
5
159.74
1.60
3.1
81
159.74
1.60
50.7
14
159.74
1.60
8.8
81
159.74
1.60
50.7
20
159.74
1.60
12.5
70
159.74
1.60
43.8
23
159.74
1.60
14.4
94
159.74
1.60
58.8
Mean
9.7
Mean
51.0
Standard Deviation
5.0
Standard Deviation
6.1
Colorado; NTU = 6.8
101
139.75
1.54
72.3
114
139.75
1.54
81.6
92
139.75
1.54
65.8
114
139.75
1.54
81.6
64
139.75
1.54
45.8
115
139.75
1.54
82.3
114
139.75
1.54
81.6
90
139.75
1.54
64.4
Six-Year Review 3	A-3	December 2016
Technical Support Document for LT2

-------
Method 1623
Method 1623.1
Oocyst
Count
Spike
#
Spike
Std Dev
%
Recovery
Oocyst
Count
Spike
#
Spike
Std Dev
%
Recovery
Mean
66.4
Mean
77.5
Standard Deviation
15.2
Standard Deviation
8.7
Massachusetts; NTU = 1.9
38
169.13
2.31
22.5
134
169.13
2.31
79.2
32
169.13
2.31
18.9
108
169.13
2.31
63.9
43
169.13
2.31
25.4
116
169.13
2.31
68.6
45
169.13
2.31
26.6
103
169.13
2.31
60.9
Mean
23.4
Mean
68.1
Standard Deviation
3.4
Standard Deviation
8.0
Texas-2; NTU = 21
69
160.15
1.46
43.1
67
160.15
1.46
41.8
116
160.15
1.46
72.4
80
160.15
1.46
50.0
101
160.15
1.46
63.1
78
160.15
1.46
48.7
108
160.15
1.46
67.4
115
160.15
1.46
71.8
Mean
61.5
Mean
53.1
Standard Deviation
12.9
Standard Deviation
13.0
Montana; NTU = 0.6
141
189.19
1.99
74.5
159
189.19
1.99
84.0
149
189.19
1.99
78.8
158
189.19
1.99
83.5
150
189.19
1.99
79.3
154
189.19
1.99
81.4
172
189.19
1.99
90.9
166
189.19
1.99
87.7
Mean
80.9
Mean
84.2
Standard Deviation
7.0
Standard Deviation
2.6
Missouri; NTU = 251
84.2
150.33
2.02
56.0
86
150.33
2.02
57.2
77.9
150.33
2.02
51.8
59.9
150.33
2.02
39.8
104
150.33
2.02
69.2
72
150.33
2.02
47.9
99.9
150.33
2.02
66.5
117.9
150.33
2.02
78.4
Mean
60.9
Mean
55.8
Standard Deviation
8.3
Standard Deviation
16.6
Source: Chapter 5, Exhibit 5.2
Six-Year Review 3
Technical Support Document for LT2
A-4
December 2016

-------
Exhibit A.4 Observed Recovery from Three Source Waters, Four Laboratories
Method 1623
Method 1623.1
Oocyst
Count
Spike
#+
Spike
Std Dev
%
Recovery
Oocyst
Count
Spike
#
Spike
Std Dev
%
Recovery
Ohio
98
161.91
1.53
60.5
103
161.91
1.53
63.6
98
161.91
1.53
60.5
99
161.91
1.53
61.1
107
161.91
1.53
66.1
99
161.91
1.53
61.1
129
161.91
1.53
79.7
125
161.91
1.53
77.2
132
161.91
1.53
81.5
137
161.91
1.53
84.6
114
161.91
1.53
70.4
128
161.91
1.53
79.1
94
161.91
1.53
58.1
70
161.91
1.53
43.2
87
161.91
1.53
53.7
112
161.91
1.53
69.2
114
161.91
1.53
70.4
128
161.91
1.53
79.1
120
161.91
1.53
74.1
143
161.91
1.53
88.3
109
161.91
1.53
67.3
145
161.91
1.53
89.6
Mean
67.5
Mean
72.4
Standard Deviation
8.9
Standard Deviation
14.1
Montana
110
152.27
1.44
72.2
109
152.27
1.44
71.6
91
152.27
1.44
59.8
102
152.27
1.44
67.0
111
152.27
1.44
72.9
100
152.27
1.44
65.7
68
152.27
1.44
44.7
112
152.27
1.44
73.6
108
152.27
1.44
70.9
98
152.27
1.44
64.4
75
152.27
1.44
49.3
108
152.27
1.44
70.9
83
152.27
1.44
54.5
127
152.27
1.44
83.4
83
152.27
1.44
54.5
123
152.27
1.44
80.8
87
152.27
1.44
57.1
115
152.27
1.44
75.5
-
-
-
-
102
152.27
1.44
67.0
-
-
-
-
93
152.27
1.44
61.1
-
-
-
-
91
152.27
1.44
59.8
-
-
-
-
92
152.27
1.44
60.4
-
-
-
-
111
152.27
1.44
72.9
-
-
-
-
106
152.27
1.44
69.6
Mean
59.5
Mean
69.6
Standard Deviation
10.3
Standard Deviation
7.0
Six-Year Review 3
Technical Support Document for LT2
AS
December 2016

-------
Method 1623
Method 1623.1
Oocyst
Count
Spike
#+
Spike
Std Dev
%
Recovery
Oocyst
Count
Spike
#
Spike
Std Dev
%
Recovery
Michigan
0
169.27
1.46
0.0
27
169.27
1.46
16.0
0
169.27
1.46
0.0
40
169.27
1.46
23.6
0
169.27
1.46
0.0
47
169.27
1.46
27.8
4
169.27
1.46
2.4
65
169.27
1.46
38.4
0
169.27
1.46
0.0
39
169.27
1.46
23.0
0
169.27
1.46
0.0
35
169.27
1.46
20.7
2
169.27
1.46
1.2
71
169.27
1.46
41.9
32
169.27
1.46
18.9
47
169.27
1.46
27.8
34
169.27
1.46
20.1
56
169.27
1.46
33.1
2
169.27
1.46
1.2
58
169.27
1.46
34.3
1
169.27
1.46
0.6
37
169.27
1.46
21.9
0
169.27
1.46
0.0
17
169.27
1.46
10.0
Mean
3.7
Mean
26.5
Standard Deviation
7.4
Standard Deviation
9.3
Source: Chapter 5, Exhibit 5.3
Six-Year Review 3
Technical Support Document for LT2
A-6
December 2016

-------
Appendix B
Occurrence and Exposure
Six-Year Review 3
Technical Support Document for LT2
December 2016

-------
Exhibit B. 1 through Exhibit B.6 show the variable values for each of the six tested trigger values
for reservoir/lakes using the original cleaning procedure.
Exhibit B.1 Variable Values for an E. coli Trigger Value of 10 CFU/100 milliliters
(mL) for Reservoir/Lake Plants Using the Original Cleaning Procedures

Plants with avg.
Plants with avg.

Cryptosporidium
concentration <
Cryptosporidium
concentration >

0.075 oocysts/L
0.075 oocysts/L
Number and
298
14
percentage of
plants with avg. E.
coli concentration
42.75%
2.01%
>= 10


Number and
379
£
percentage of
0
plants with avg. E.
coli concentration
54.38%
0.86%
< 10


Exhibit B.2 Variable Values for an E. coli Trigger Value of 50 CFU/100 mLfor
Reservoir/Lake Plants Using the Original Cleaning Procedures

Plants with avg.
Cryptosporidium
concentration <
0.075 oocysts/L
Plants with avg.
Cryptosporidium
concentration >
0.075 oocysts/L
Number and
percentage of
plants with avg. E.
coli concentration
>= 50
118
9
16.93%
1.29%
Number and
percentage of
plants with avg. E.
coli concentration
<50
559
11
80.20%
1.58%
Six-Year Review 3
Technical Support Document for LT2
B-l
December 2016

-------
Exhibit B.3 Variable Values for an E. coli Trigger Value of 75 CFU/100 mLfor
Reservoir/Lake Plants Using the Original Cleaning Procedures

Plants with avg.
Plants with avg.

Cryptosporidium
concentration <
Cryptosporidium
concentration >

0.075 oocysts/L
0.075 oocysts/L
Number and
93

percentage of
y
plants with avg. E.
coli concentration
13.34%
1.29%
>= 75


Number and
584
11
percentage of
plants with avg. E.
coli concentration
83.79%
1.58%
<75


Exhibit B.4 Variable Values for an E. coli Trigger Value of 100 CFU/100 mL for
Reservoir/Lake Plants Using the Original Cleaning Procedures

Plants with avg.
Plants with avg.

Cryptosporidium
concentration <
Cryptosporidium
concentration >

0.075 oocysts/L
0.075 oocysts/L
Number and
65

percentage of
/
plants with avg. E.
coli concentration
9.33%
1.00%
>= 100


Number and
612
13
percentage of
plants with avg. E.
coli concentration
87.80%
1.87%
< 100


Six-Year Review 3
Technical Support Document for LT2
B-2
December 2016

-------
Exhibit B.5 Variable Values for an E. coli Trigger Value of 150 CFU/100 mL for
Reservoir/Lake Plants Using the Original Cleaning Procedures

Plants with avg.
Plants with avg.

Cryptosporidium
concentration <
Cryptosporidium
concentration >

0.075 oocysts/L
0.075 oocysts/L
Number and
36

percentage of
4
plants with avg. E.
coli concentration
5.16%
0.57%
>= 150


Number and
641
16
percentage of
plants with avg. E.
coli concentration
91.97%
2.30%
< 150


Exhibit B.6 Variable Values for an E. coli Trigger Value of 200 CFU/100 mL for
Reservoir/Lake Plants Using the Original Cleaning Procedures

Plants with avg.
Plants with avg.

Cryptosporidium
concentration <
Cryptosporidium
concentration >

0.075 oocysts/L
0.075 oocysts/L
Number and
22

percentage of
4
plants with avg. E.
coli concentration
3.16%
0.57%
>= 200


Number and
655
16
percentage of
plants with avg. E.
coli concentration
93.97%
2.30%
<200


Six-Year Review 3
Technical Support Document for LT2
B-3
December 2016

-------
Exhibit B.7 through Exhibit B. 12 give the variable values for flowing stream plants using the
original cleaning procedures.
Exhibit B.7 Variable Values for an E. coli Trigger Value of 10 CFU/100 mLfor
Flowing Stream Plants Using the Original Cleaning Procedures

Plants with avg.
Plants with avg.

Cryptosporidium
concentration <
Cryptosporidium
concentration >

0.075 oocysts/L
0.075 oocysts/L
Number and
413
73
percentage of
plants with avg. E.
coli concentration
66.83%
11.81%
>= 10


Number and
130
O
percentage of
Z
plants with avg. E.
coli concentration
21.04%
0.32%
< 10


Exhibit B.8 Variable Values for an E. coli Trigger Value of 50 CFU/100 mL for
Flowing Stream Plants Using the Original Cleaning Procedures

Plants with avg.
Plants with avg.

Cryptosporidium
concentration <
Cryptosporidium
concentration >

0.075 oocysts/L
0.075 oocysts/L
Number and
288
67
percentage of
plants with avg. E.
coli concentration
46.60%
10.84%
>= 50


Number and
255
8
percentage of
plants with avg. E.
coli concentration
41.26%
1.29%
<50


Six-Year Review 3
Technical Support Document for LT2
B-4
December 2016

-------
Exhibit B.9 Variable Values for an E. coli Trigger Value of 75 CFU/100 mLfor
Flowing Stream Plants Using the Original Cleaning Procedures

Plants with avg.
Plants with avg.

Cryptosporidium
concentration <
Cryptosporidium
concentration >

0.075 oocysts/L
0.075 oocysts/L
Number and
247
63
percentage of
plants with avg. E.
coli concentration
39.97%
10.19%
>= 75


Number and
296
12
percentage of
plants with avg. E.
coli concentration
47.90%
1.94%
<75


Exhibit B.10 Variable Values for an E. coli Trigger Value of 100 CFU/100 mL for
Flowing Stream Plants Using the Original Cleaning Procedures

Plants with avg.
Plants with avg.

Cryptosporidium
concentration <
Cryptosporidium
concentration >

0.075 oocysts/L
0.075 oocysts/L
Number and
206
61
percentage of
plants with avg. E.
coli concentration
33.33%
9.87%
>= 100


Number and
337
14
percentage of
plants with avg. E.
coli concentration
54.53%
2.27%
< 100


Six-Year Review 3
Technical Support Document for LT2
B-5
December 2016

-------
Exhibit B.11 Variable Values for an E. coli Trigger Value of 150 CFU/100 mL for
Flowing Stream Plants Using the Original Cleaning Procedures

Plants with avg.
Plants with avg.

Cryptosporidium
concentration <
Cryptosporidium
concentration >

0.075 oocysts/L
0.075 oocysts/L
Number and
173
53
percentage of
plants with avg. E.
coli concentration
27.99%
8.58%
>= 150


Number and
370
22
percentage of
plants with avg. E.
coli concentration
59.87%
3.56%
< 150


Exhibit B.12 Variable Values for an E. coli Trigger Value of 200 CFU/100 mL for
Flowing Stream Plants Using the Original Cleaning Procedures

Plants with avg.
Plants with avg.

Cryptosporidium
concentration <
Cryptosporidium
concentration >

0.075 oocysts/L
0.075 oocysts/L
Number and
135
50
percentage of
plants with avg. E.
coli concentration
21.84%
8.09%
>= 200


Number and
408
25
percentage of
plants with avg. E.
coli concentration
66.02%
4.05%
<200


Six-Year Review 3
Technical Support Document for LT2
B-6
December 2016

-------
Exhibit B. 13 through Exhibit B. 18 show the variables for all plants using the original cleaning
procedures.
Exhibit B.13 Variable Values for an E. coli Trigger Value of 10 CFU/100 mL for All
Plants Using the Original Cleaning Procedures

Plants with avg.
Plants with avg.

Cryptosporidium
concentration <
Cryptosporidium
concentration >

0.075 oocysts/L
0.075 oocysts/L
Number and
730
91
percentage of
plants with avg. E.
coli concentration
53.83%
6.71%
>= 10


Number and
527
8
percentage of
plants with avg. E.
coli concentration
38.86%
0.59%
< 10


Exhibit B.14 Variable Values for an E. coli Trigger Value of 50 CFU/100 mL for All
Plants Using the Original Cleaning Procedures

Plants with avg.
Plants with avg.

Cryptosporidium
concentration <
Cryptosporidium
concentration >

0.075 oocysts/L
0.075 oocysts/L
Number and
414
79
percentage of
plants with avg. E.
coli concentration
30.53%
5.83%
>= 50


Number and
843
20
percentage of
plants with avg. E.
coli concentration
62.17%
1.47%
<50


Six-Year Review 3
Technical Support Document for LT2
B-7
December 2016

-------
Exhibit B.15 Variable Values for an E. coli Trigger Value of 75 CFU/100 mL for All
Plants Using the Original Cleaning Procedures

Plants with avg.
Plants with avg.

Cryptosporidium
concentration <
Cryptosporidium
concentration >

0.075 oocysts/L
0.075 oocysts/L
Number and
347
75
percentage of
plants with avg. E.
coli concentration
25.59%
5.53%
>= 75


Number and
910
24
percentage of
plants with avg. E.
coli concentration
67.11%
1.77%
<75


Exhibit B.16 Variable Values for an E. coli Trigger Value of 100 CFU/100 mL for All
Plants Using the Original Cleaning Procedures

Plants with avg.
Plants with avg.

Cryptosporidium
concentration <
Cryptosporidium
concentration >

0.075 oocysts/L
0.075 oocysts/L
Number and
275
71
percentage of
plants with avg. E.
coli concentration
20.28%
5.24%
>= 100


Number and
982
28
percentage of
plants with avg. E.
coli concentration
72.42%
2.06%
< 100


Six-Year Review 3
Technical Support Document for LT2
B-8
December 2016

-------
Exhibit B.17 Variable Values for an E. coli Trigger Value of 150 CFU/100 mL for All
Plants Using the Original Cleaning Procedures

Plants with avg.
Plants with avg.

Cryptosporidium
concentration <
Cryptosporidium
concentration >

0.075 oocysts/L
0.075 oocysts/L
Number and
213
59
percentage of
plants with avg. E.
coli concentration
15.71%
4.35%
>= 150


Number and
1,044
40
percentage of
plants with avg. E.
coli concentration
76.99%
2.95%
< 150


Exhibit B.18 Variable Values for an E. coli Trigger Value of 200 CFU/100 mL for All
Plants Using the Original Cleaning Procedures

Plants with avg.
Plants with avg.

Cryptosporidium
concentration <
Cryptosporidium
concentration >

0.075 oocysts/L
0.075 oocysts/L
Number and
160
56
percentage of
plants with avg. E.
coli concentration
11.80%
4.13%
>= 200


Number and
1,097
43
percentage of
plants with avg. E.
coli concentration
80.90%
3.17%
<200


Six-Year Review 3
Technical Support Document for LT2
B-9
December 2016

-------
Exhibit B. 19 through Exhibit B.24 show the variables for reservoir/lake plants for the revised
cleaning procedure.
Exhibit B.19 Variable Values for an E. coli Trigger Value of 10 CFU/100 mL for
Reservoir/Lake Plants Using the Revised Cleaning Procedures

Plants with avg.
Plants with avg.

Cryptosporidium
concentration <
Cryptosporidium
concentration >

0.075 oocysts/L
0.075 oocysts/L
Number and
341
19
percentage of
plants with avg. E.
coli concentration
48.58%
2.71%
>= 10


Number and
340
O
percentage of
Z
plants with avg. E.
coli concentration
48.43%
0.28%
< 10


Exhibit B.20 Variable Values for an E. coli Trigger Value of 50 CFU/100 mL for
Reservoir/Lake Plants Using the Revised Cleaning Procedures

Plants with avg.
Plants with avg.

Cryptosporidium
concentration <
Cryptosporidium
concentration >

0.075 oocysts/L
0.075 oocysts/L
Number and
128
11
percentage of


plants with avg. E.
coli concentration
18.23%
1.57%
>= 50


Number and
553
10
percentage of


plants with avg. E.
coli concentration
78.77%
1.42%
<50


Six-Year Review 3
Technical Support Document for LT2
B-10
December 2016

-------
Exhibit B.21 Variable Values for an E. coli Trigger Value of 75 CFU/100 mL for
Reservoir/Lake Plants Using the Revised Cleaning Procedures

Plants with avg.
Cryptosporidium
concentration <
0.075 oocysts/L
Plants with avg.
Cryptosporidium
concentration >
0.075 oocysts/L
Number and
percentage of
plants with avg. E.
coli concentration
>= 75
92
10
13.11%
1.42%
Number and
percentage of
plants with avg. E.
coli concentration
<75
589
11
83.90%
1.57%
Exhibit B.22 Variable Values for an E. coli Trigger Value of 100 CFU/100 mL for
Reservoir/Lake Plants Using the Revised Cleaning Procedures

Plants with avg.
Plants with avg.

Cryptosporidium
concentration <
Cryptosporidium
concentration >

0.075 oocysts/L
0.075 oocysts/L
Number and
68
8
percentage of
plants with avg. E.
coli concentration
9.69%
1.14%
>= 100


Number and
613
13
percentage of
plants with avg. E.
coli concentration
87.32%
1.85%
< 100


Six-Year Review 3
Technical Support Document for LT2
B-ll
December 2016

-------
Exhibit B.23 Variable Values for an E. coli Trigger Value of 150 CFU/100 mL for
Reservoir/Lake Plants Using the Revised Cleaning Procedures

Plants with avg.
Plants with avg.

Cryptosporidium
concentration <
Cryptosporidium
concentration >

0.075 oocysts/L
0.075 oocysts/L
Number and
36

percentage of
4
plants with avg. E.
coli concentration
5.13%
0.57%
>= 150


Number and
645
17
percentage of
plants with avg. E.
coli concentration
91.88%
2.42%
< 150


Exhibit B.24 Variable Values for an E. coli Trigger Value of 200 CFU/100 mL for
Reservoir/Lake Plants Using the Revised Cleaning Procedures

Plants with avg.
Plants with avg.

Cryptosporidium
concentration <
Cryptosporidium
concentration >

0.075 oocysts/L
0.075 oocysts/L
Number and
19

percentage of
4
plants with avg. E.
coli concentration
2.71%
0.57%
>= 200


Number and
percentage of
662
17
plants with avg. E.
coli concentration
94.30%
2.42%
<200


Six-Year Review 3
Technical Support Document for LT2
B-12
December 2016

-------
Exhibit B.25 through Exhibit B.30 show the variables for flowing stream plants using the revised
cleaning procedures.
Exhibit B.25 Variable Values for an E. coli Trigger Value of 10 CFU/100 mL for
Flowing Stream Plants Using the Revised Cleaning Procedures

Plants with avg.
Plants with avg.

Cryptosporidium
concentration <
Cryptosporidium
concentration >

0.075 oocysts/L
0.075 oocysts/L
Number and
435
62
percentage of
plants with avg. E.
coli concentration
70.50%
10.05%
>= 10


Number and
119
1
percentage of
plants with avg. E.
coli concentration
19.29%
0.16%
< 10


Exhibit B.26 Variable Values for an E. coli Trigger Value of 10 CFU/100 mL for
Flowing Stream Plants Using the Revised Cleaning Procedures

Plants with avg.
Plants with avg.

Cryptosporidium
concentration <
Cryptosporidium
concentration >

0.075 oocysts/L
0.075 oocysts/L
Number and
290
55
percentage of
plants with avg. E.
coli concentration >=
47.00%
8.91%
50


Number and
264
8
percentage of
plants with avg. E.
coli concentration <
42.79%
1.30%
50


Six-Year Review 3
Technical Support Document for LT2
B-13
December 2016

-------
Exhibit B.27 Variable Values for an E. coli Trigger Value of 75 CFU/100 mL for
Flowing Stream Plants Using the Revised Cleaning Procedures

Plants with avg.
Cryptosporidium
concentration <
0.075 oocysts/L
Plants with avg.
Cryptosporidium
concentration >
0.075 oocysts/L
Number and
percentage of
plants with avg. E.
coli concentration
>= 75
251
52
40.68%
8.43%
Number and
percentage of
plants with avg. E.
coli concentration
<75
303
11
49.11%
1.78%
Exhibit B.28 Variable Values for an E. coli Trigger Value of 100 CFU/100 mL for
Flowing Stream Plants Using the Revised Cleaning Procedures

Plants with avg.
Plants with avg.

Cryptosporidium
concentration <
Cryptosporidium
concentration >

0.075 oocysts/L
0.075 oocysts/L
Number and
205
48
percentage of
plants with avg. E.
coli concentration
33.23%
7.78%
>= 100


Number and
349
15
percentage of
plants with avg. E.
coli concentration
56.56%
2.43%
< 100


Six-Year Review 3
Technical Support Document for LT2
B-14
December 2016

-------
Exhibit B.29 Variable Values for an E. coli Trigger Value of 150 CFU/100 mL for
Flowing Stream Plants Using the Revised Cleaning Procedures

Plants with avg.
Plants with avg.

Cryptosporidium
concentration <
Cryptosporidium
concentration >

0.075 oocysts/L
0.075 oocysts/L
Number and
169
40
percentage of
plants with avg. E.
coli concentration
27.39%
6.48%
>= 150


Number and
percentage of
385
23
plants with avg. E.
coli concentration
62.40%
3.73%
< 150


Exhibit B.30 Variable Values for an E. coli Trigger Value of 200 CFU/100 mL for
Flowing Stream Plants Using the Revised Cleaning Procedures

Plants with avg.
Plants with avg.

Cryptosporidium
concentration <
Cryptosporidium
concentration >

0.075 oocysts/L
0.075 oocysts/L
Number and
133
38
percentage of
plants with avg. E.
coli concentration
21.56%
6.16%
>= 200


Number and
421
25
percentage of
plants with avg. E.
coli concentration
68.23%
4.05%
<200


Six-Year Review 3
Technical Support Document for LT2
B-15
December 2016

-------
Exhibit B.31 through Exhibit B.36 show the variables for plants categorized as both
reservoir/lake and flowing stream using the revised cleaning procedures.
Exhibit B.31 Variable Values for an E. coli Trigger Value of 10 CFU/100 mL for
Plants Categorized as Both Using the Revised Cleaning Procedures

Plants with avg.
Cryptosporidium
concentration <
0.075 oocysts/L
Plants with avg.
Cryptosporidium
concentration >
0.075 oocysts/L
Number and
percentage of
plants with avg. E.
coli concentration
>= 10
30
5
56.60%
9.43%
Number and
percentage of
plants with avg. E.
coli concentration
< 10
17
1
32.08%
1.89%
Exhibit B.32 Variable Values for an E. coli Trigger Value of 50 CFU/100 mL for
Plants Categorized as Both Using the Revised Cleaning Procedures

Plants with avg.
Plants with avg.

Cryptosporidium
concentration <
Cryptosporidium
concentration >

0.075 oocysts/L
0.075 oocysts/L
Number and
12

percentage of
O
plants with avg. E.
coli concentration
22.64%
5.66%
>= 50


Number and
35

percentage of

plants with avg. E.
coli concentration
66.04%
5.66%
<50


Six-Year Review 3
Technical Support Document for LT2
B-16
December 2016

-------
Exhibit B.33 Variable Values for an E. coli Trigger Value of 75 CFU/100 mL for
Plants Categorized as Both Using the Revised Cleaning Procedures

Plants with avg.
Plants with avg.

Cryptosporidium
concentration <
Cryptosporidium
concentration >

0.075 oocysts/L
0.075 oocysts/L
Number and
8
O
percentage of
J
plants with avg. E.
coli concentration
15.09%
5.66%
>= 75


Number and
39
3
percentage of

plants with avg. E.
coli concentration
73.58%
5.66%
<75


Exhibit B.34 Variable Values for an E. coli Trigger Value of 100 CFU/100 mL for
Plants Categorized as Both Using the Revised Cleaning Procedures

Plants with avg.
Plants with avg.

Cryptosporidium
concentration <
Cryptosporidium
concentration >

0.075 oocysts/L
0.075 oocysts/L
Number and


percentage of
D
O
plants with avg. E.
coli concentration
11.32%
5.66%
>= 100


Number and
41
3
percentage of

plants with avg. E.
coli concentration
77.36%
5.66%
< 100


Six-Year Review 3
Technical Support Document for LT2
B-17
December 2016

-------
Exhibit B.35 Variable Values for an E. coli Trigger Value of 150 CFU/100 mL for
Plants Categorized as Both Using the Revised Cleaning Procedures

Plants with avg.
Plants with avg.

Cryptosporidium
concentration <
Cryptosporidium
concentration >

0.075 oocysts/L
0.075 oocysts/L
Number and
£
O
percentage of
0
J
plants with avg. E.
coli concentration
11.32%
5.66%
>= 150


Number and
41

percentage of
O
plants with avg. E.
coli concentration
77.36%
5.66%
< 150


Exhibit B.36 Variable Values for an E. coli Trigger Value of 200 CFU/100 mL for
Plants Categorized as Both Using the Revised Cleaning Procedures

Plants with avg.
Plants with avg.

Cryptosporidium
concentration <
Cryptosporidium
concentration >

0.075 oocysts/L
0.075 oocysts/L
Number and
4
2
percentage of


plants with avg. E.
coli concentration
7.55%
3.77%
>= 200


Number and
43
A
percentage of

plants with avg. E.
coli concentration
81.13%
7.55%
<200


Six-Year Review 3
Technical Support Document for LT2
B-18
December 2016

-------
Exhibit B.37 through Exhibit B.42 show the variable values for all plants using the revised
cleaning procedures.
Exhibit B.37 Variable Values for an E. coli Trigger Value of 10 CFU/100 mL for All
Plants Using the Revised Cleaning Procedures

Plants with avg.
Plants with avg.

Cryptosporidium
concentration <
Cryptosporidium
concentration >

0.075 oocysts/L
0.075 oocysts/L
Number and
806
86
percentage of
plants with avg. E.
coli concentration
58.75%
6.27%
>= 10


Number and
476

percentage of
4
plants with avg. E.
coli concentration
34.69%
0.29%
< 10


Exhibit B.38 Variable Values for an E. coli Trigger Value of 50 CFU/100 mL for All
Plants Using the Revised Cleaning Procedures

Plants with avg.
Plants with avg.

Cryptosporidium
concentration <
Cryptosporidium
concentration >

0.075 oocysts/L
0.075 oocysts/L
Number and
430
69
percentage of
plants with avg. E.
coli concentration
31.34%
5.03%
>= 50


Number and
852
21
percentage of
plants with avg. E.
coli concentration
62.10%
1.53%
<50


Six-Year Review 3
Technical Support Document for LT2
B-19
December 2016

-------
Exhibit B.39 Variable Values for an E. coli Trigger Value of 75 CFU/100 mL for All
Plants Using the Revised Cleaning Procedures

Plants with avg.
Plants with avg.

Cryptosporidium
concentration <
Cryptosporidium
concentration >

0.075 oocysts/L
0.075 oocysts/L
Number and
351
65
percentage of
plants with avg. E.
coli concentration
25.58%
4.74%
>= 75


Number and
931
25
percentage of
plants with avg. E.
coli concentration
67.86%
1.82%
<75


Exhibit B.40 Variable Values for an E. coli Trigger Value of 100 CFU/100 mL for All
Plants Using the Revised Cleaning Procedures

Plants with avg.
Plants with avg.

Cryptosporidium
concentration <
Cryptosporidium
concentration >

0.075 oocysts/L
0.075 oocysts/L
Number and
279
59
percentage of
plants with avg. E.
coli concentration
20.34%
4.30%
>= 100


Number and
1,003
31
percentage of
plants with avg. E.
coli concentration
73.10%
2.26%
< 100


Six-Year Review 3
Technical Support Document for LT2
B-20
December 2016

-------
Exhibit B.41 Variable Values for an E. coli Trigger Value of 150 CFU/100 mL for All
Plants Using the Revised Cleaning Procedures

Plants with avg.
Plants with avg.

Cryptosporidium
concentration <
Cryptosporidium
concentration >

0.075 oocysts/L
0.075 oocysts/L
Number and
211
47
percentage of
plants with avg. E.
coli concentration
15.38%
3.43%
>= 150


Number and
1,071
43
percentage of
plants with avg. E.
coli concentration
78.06%
3.13%
< 150


Exhibit B.42 Variable Values for an E. coli Trigger Value of 200 CFU/100 mL for All
Plants Using the Revised Cleaning Procedures

Plants with avg.
Plants with avg.

Cryptosporidium
concentration <
Cryptosporidium
concentration >

0.075 oocysts/L
0.075 oocysts/L
Number and
156
44
percentage of
plants with avg. E.
coli concentration
11.37%
3.21%
>= 200


Number and
1,126
46
percentage of
plants with avg. E.
coli concentration
82.07%
3.35%
<200


Six-Year Review 3
Technical Support Document for LT2
B-21
December 2016

-------
Appendix C
Toolbox Option Usage and Related
Implementation Issues
Six-Year Review 3
Technical Support Document for LT2
December 2016

-------
Exhibit C.l presents the LT2 toolbox options and the number of systems using each option as of
May 2013. Exhibit C.l includes systems that have been placed in Bin 2 or higher based on their
Cryptosporidium monitoring results, and are therefore required to install at least one toolbox
technology.
Data on the number of systems using each toolbox option was provided by EPA Regional offices
and the Association of State Drinking Water Administrators (ASDWA), findings from web
searches, and results of a detailed literature review. On some occasions during the gathering of
information, it was not absolutely clear whether systems were implementing toolbox
technologies to comply with the LT2 or whether they were installing the technologies for other
reasons.
Note that individual water systems have frequently opted to implement more than one toolbox
technology (e.g., membrane filtration combined with ultraviolet (UV) disinfection). Toolbox
technologies that have not been selected by the systems (for which EPA has collected
information) are not listed in Exhibit C.l. Exhibit C. 1 also provides brief summaries of
implementation challenges and concerns that are related to the particular toolbox option, as
discussed in the main body of this report.
Six-Year Review 3
Technical Support Document for LT2
C-l
December 2016

-------
Exhibit C.1 Systems in Bin 2 or Higher: Summary of the Number of Systems Applying Toolbox Options
and Related Implementation Challenges and Concerns
Toolbox Option
Credits Currently
Allowed
Number of
Systems Using
the Option1
% of Systems
Selecting
Toolbox Option
Implementation Challenges and Concerns
Source Protection and Management Toolbox Options
Watershed Control
Program (WCP)
0.5-log credit
11
6.6%
Few systems have chosen this option due to:
•	Tight time schedule and time-consuming efforts
needed for development and implementation,
•	Stringent elements required for the watershed
control plan, and
•	Relatively low removal credit available through
this option.
Alternative Source
/ Intake
Management
No prescribed credit
7
4.2%
Tight time schedule and time-consuming efforts
needed for development and implementation into
existing treatment processes
Pre-Filtration Toolbox Options
Pre-Sedimentation
Basin with
Coagulation
0.5-log credit:
2
1.2%
Time-consuming efforts needed for development and
implementation
Two-stage Lime
Softening
0.5-log credit
1
0.6%
Time-consuming efforts needed for development and
implementation
Bank Filtration
(BF)
0.5-log credit for 25-
foot setback;
1.0-log credit for 50-
foot setback;
Additional credit for
state-approved site
specific study
4
2.4%
The implementation issue identified is the
difficulty demonstrating more than 2-log
Cryptosporidium removal credit using bank filtration
Six-Year Review 3
Technical Support Document for LT2
C-2
December 2016

-------
Toolbox Option
Credits Currently
Allowed
Number of
Systems Using
the Option1
% of Systems
Selecting
Toolbox Option
Implementation Challenges and Concerns
Treatment Performance Toolbox Options
Combined Filter
Performance
0.5-log credit
43
25.9%
Concerns related to data and reporting integrity,
accuracy of the tool and the oversight necessary to
ensure the validity of the credit overtime
Individual Filter
Performance
0.5-log credit
39
23.5%
Concerns related to data and reporting integrity,
accuracy of the tool and the oversight necessary to
ensure the validity of the credit overtime
Demonstration of
Performance
(DOP)
Credit awarded to unit
process or treatment
train based on a
demonstration to the
state with a state-
approved protocol. [40
CFR 141.718 (c)]
3
1.8%
Issues raised include:
•	It is unclear if states have developed DOP
protocols and whether they are made available
to the public;
•	EPA's Toolbox Guidance Manual introduces
basic DOP concepts but, without state-
developed protocols issued in a timely manner,
a system would not be aware of all the required
elements to include in the DOP or length of
time needed for the study (e.g., pilot test), and;
•	Meeting the schedule for LT2 compliance may
have been a contributing factor for systems not
selecting this option.
Filtration2 (Option
Unknown)

12
7.2%

Additional Filtration Toolbox Options
Bag and Cartridge
Filters
Up to 2-log credit if
used singly;
Up to 2.5 log credit if in
series.
13
0.6%
Challenge testing determination may be a deterrent
Membrane
Filtration
Log credit equivalent to
removal efficiency
demonstrated in
challenge test for
device if supported by
direct integrity testing.
[40 CFR 141.719(b)]
18
10.8%
Issues raised include:
•	New technologies such as ceramic membranes
have entered the water treatment market, and
•	There is a lack of data and a consensus on
how to use the data in evaluating membrane
applications.
Six-Year Review 3
Technical Support Document for LT2
C-3
December 2016

-------
Toolbox Option
Credits Currently
Allowed
Number of
Systems Using
the Option1
% of Systems
Selecting
Toolbox Option
Implementation Challenges and Concerns
Second Stage
Filtration
0.5-log credit
2
1.2%
Implementation requires a large capital investment
Inactivation Toolbox Options
Chlorine Dioxide
Log credit based on
measured CT in
relation to CT table. [40
CFR 141.720 (b)]
1
0.6%
Implementation requirements of ongoing monitoring,
sampling and analysis
Ozone
Log credit based on
measured CT in
relation to CT table. [40
CFR 141.720 (b)]
3
1.8%
Documented operational issues include:
•	Operating costs,
•	Bromate formation may preclude the use of
ozone,and
•	Practicality of calculating CT is limited.
UV disinfection
Log credit based on
validated UV dose in
relation to UV dose
table; reactor validation
testing required to
establish UV does and
associated operating
conditions. [40 CFR
141.720 (d)]
19
11.4%
Documented operational issues include:
•	Monitoring of UV dose delivered,
•	Verifying inactivation for viruses,
•	How to verify pre-existing UV installations meet
Cryptosporidium inactivation requirements,
•	Maintenance of reactors, and
•	Operating costs.
Total Number of Applied Toolbox Options
166
100%

Notes:
1)	Systems may be counted more than once if they use multiple toolbox options. In addition, not every applied toolbox option is for LT2 compliance.
2)	Available data describes use of a filtration process only but does not describe which toolbox option (combined filter performance, individual filter
performance or DOP) the system used for credits.
3)	Available data indicates bag and/or cartridge filters only and does not provide information related to whether they are individual filters or filters operating in
series.
Six-Year Review 3
Technical Support Document for LT2
C-4
December 2016

-------