EPA/600/Ar93/256
Engineering Aspects of Waterborne Disease
Outbreak Investigations
Kim R. Fox
Drinking Water Research Division
U.S. Environmental Protection Agency
Cincinnati, Ohio 45268
Paper presented for the Sunday Seminar on
"Waterborne Disease Surveillance and
Investigation Without Going Undercover"
INTRODUCTION
Two recent headline causing events have reinforced the
concern about the spread of diseases through waterborne routes.
These two events include the current cholera epidemic in the
western hemisphere that has caused over 750,000 reported cases of
cholera through April, 19931, and the more recent Cryptosporidium
outbreak in Milwaukee, Wisconsin where between 200,000 and 400,000
people had diarrhea during the time frame of concern. Although
many of the disease cases can be contributed to other than the
waterborne mode of transmission, the rapid spread and large number
of illnesses are frequently the result of waterborne transmission.
In both of these events, a breakdown in proper water treatment
allowed the etiological agent to survive in and be transported by
the drinking water, which in turn allowed for a rapid spread of
the agent over large areas of population. Whenever an enteric
disease outbreak occurs, or even when a small outbreak happens and
drinking water is implicated, an investigation should take place
to help determine if drinking water was a contributing factor.
This investigation will attempt to overlay the epidemiological
data collected by local, state, and federal health agencies, and
any and all data available from the drinking water systems. These
data will not be limited to water quality data, but will also
include information about system operation, abnormal conditions
(i.e. power losses or pipe breaks), and almost any other available
information about the water system. In most cases, by the time
drinking water is implicated, the disease outbreak is over or at
least on a decline, therefore, the available data will depend on
data records. This paper will discuss the types of information
gathered in three cases studies from past suspected waterborne
outbreaks and how these data were used to implicate water as a
mode of transmission. This information should help those involved
with future outbreaks with guidance to determine what types of
information are useful in waterborne disease investigations
(Figure 1).
l

-------
DRINKING WATER IMPLICATED
In a suspected waterborne disease outbreak, drinking water
will have been implicated as a carrier of the disease causing
organisms. The implication that the water was suspect generally
comes from the epidemiological data gathered by local, state, and
federal agencies. A waterborne outbreak is normally defined as an
acute illness affecting two or more persons with similar symptoms
that is epidemiologically associated with ingesting of water or
some other exposure to water intended for drinking.1 By the time
water is implicated in an outbreak, the conditions in water
quality may have changed, and in most of the previous outbreaks,
the etiologic agent was not isolated from the drinking water. The
majority of waterborne outbreaks are not recognized, investigated,
or even reported, and In only half of the reported outbreaks is
the causative agent identified. Investigating waterborne disease
outbreaks has been likened to "fire fighting--when an outbreak
occurs, investigators rush to the scene, assess the damage, find
its cause, correct the problem, and return the system to its
normal state".2 The investigators are normally trying to find the
engineering solution to a breakdown in the system that normally
protects us from waterborne pathogens (i.e. water treatment plants
and distribution systems).
In fighting the fire (or conducting the investigations), the
investigators must be prepared to work with any and all
individuals, organizations, groups, and officials involved with
the current outbreak. These contacts may include (but not limited
to) local, state, and federal health agencies, water utility
workers, local administration officials, state and federal
regulatory agencies, the public, and even the press. In some
cases, the investigating team may spend more time dealing with the
many groups than actually doing the investigations. The
investigators do need to remember that the water utility will
probably need to be producing water during the investigation and
the first order of business is to provide a safe water to their
customers. The second concern is to determine what happened so
that measures can be taken to prevent a future occurrence.
Another important factor in fighting the fires is to keep an
open mind about what events may have occurred to cause the
suspected waterborne disease outbreak. In the next section, three
case studies are presented to highlight three different waterborne
outbreaks investigations. Each of the outbreaks resulted from
suspected contamination of the drinking water, but the method of
contamination was significantly different in each case. The
ability to keep an open mind allows the investigating team to look
at many scenarios before attempting to make a judgment about what
might have happened.
2

-------
CASE STUDIES
Each of the three case studies presented in this paper will
focus on a different method of contamination of the drinking water
that resulted in diarrheal illnesses and in some instances deaths.
The first case described is a contamination in a single facility,
the second case describes contamination of a distribution system,
and the third case discusses contamination that passed through a
water treatment plant.
In Case Study I, 39 people in a residence hall for a
hospital became 111 over a three day period in July. The
etiological agent was a blue-green algae, and drinking water was a
suspected carrier.The drinking water supply for this building is
connected to the munitipal water supply. The water entering the
hall passed through three pressure filter units plumbed in
parallel and each one was approximately 5-6 ft in diameter and
about 7 ft tall. When water was flowing through the filters, each
filter had about 5 psi pressure loss across the filter indicating
that media was present in each filter. The hall's service
personnel indicated that the filters had not been backwashed for
at least 10 years (that's how long the present service personnel
had been there). The service personnel had been told that the
filters were not in service any more.
After the water passed through the filters, it flowed into a
surge tank that fed the main pump. The surge tank was not covered
and the water was exposed to the basement area. The main pump
then transferred the water to two roof storage tanks that fed the
hall's distribution system by gravity (Figure 2). The two roof
storage tanks (each approximately 5000 gallons) were housed in a
penthouse area and were uncovered. A tarp was available for
covering the tanks, but the tarps did not always completely cover
the tanks. The water line from the basement pump was split into
two lines and each line served one of the storage tanks. The
influent to each storage tank was at the top of the tank and water
free fell into the tank. The effluent lines from the tanks were
located approximately 8-10 inches above the bottom of the tanks
and the effluent from the two tanks were combined again before
serving the hall's distribution system. The water depth in each
storage tank normally varied from a low of 4-5 ft to a high of 7-8
ft.
Early in the morning (-1:00 am) on the day of the first
diarrheal case, the water pump in the basement that pumps water to
the roof storage stopped and roof storage tanks were drained by
normal use. The pump was repaired and restarted at about 7:00 am.
Because of the timing of the pump breakdown (just prior to the
onset of the outbreak) the water supply was immediately suspect as
a probable cause. The onset of the illnesses began on July 5 and
continued through July 7. By the time the cases were reported and
the pump problem noted, it was July 10 and clean water had been
pumped through the building system for over five days. Fecal
3

-------
specimens taken from patients Indicated that blue-green algae
caused the illnesses. Water samples taken from kitchen taps and
drinking water fountains were all negative for blue-green algae.
The penthouse area (where the storage tanks were housed) was
not sealed from the outside and several windows were broken and no
screens were available. There was evidence of birds in the
penthouse with bird feces on the storage tank brim, on the pipes
located above the tanks, and on the tarp that partially covered
the one tank that still contained water.
Drinking water was the suspected source of contamination and
the point where the contamination entered the water was likely the
roof storage tanks. When the pump failure occurred, the water
level in the tanks drained down to the bottom of the effluent line
(approximately 8-10 Inches above the bottom of the tanks). The
bottom 8 inches of water in those tanks was a stagnant zone where
normally the water moved either very little or not at all. This
zone would not normally mix with the water above this zone and
chlorine concentration in this zone would be very low. Because
this zone would not have normal levels of chlorine and would not
be flowing in normal conditions, this was a likely area for
biological growth. When the pump was turned back on, the new
water coming into the storage tank would mix with this stagnant
water into the new water and thus distribute the stagnant water
throughout the building. The bird fecal material and the birds
using the tanks for drinking or swimming were a possible source of
initial contamination.
The new water flowing into the storage tanks both flushed
the tanks out and added fresh chlorine into the lower areas of the
storage tank. The flushing action would be enough to reduce the
algal concentration in the tank so that the later sampling for
blue-green algae was negative.
In this example, the epidemiological data collected by the
Centers for Disease Control could not completely rule out one
other mode of disease transmission (a food route was suggested),
but the scenario described above is considered the most likely
route of exposure.
In Case Study II, a small town with a population of 2090
experienced an outbreak where 243 cases of diarrhea (85 cases were
bloody diarrhea) were reported, and four deaths were recorded over
a 35-day period. The epidemiological study implicated the
drinking water system when it was observed that individuals living
within the limits of the water distribution system (or those that
frequently came into the area and consumed water) were 18 times
more likely to become infected than those that lived just outside
the limits of the distribution system, or those that drank bottled
water.5 The etiological agent identified in stool samples was
Escherichia coli serotype 0157:H7.. ..This E; coli had been
4

-------
identified in other outbreaks but had always been limited to
hamburger or milk contaminations.
In this case, the engineering investigation did not begin
until four weeks after the main impact of the outbreak. Prior to
the engineering investigation, a boll water order had been issued
for this community and the water utility began chlorinating the
previous untreated well water before sending it out into the
distribution system. An engineering investigation was requested
to determine if a sequence of events could have taken place that
would have caused the spread of the E. coli through the drinking
water distribution system.
Since the investigation of the water system was conducted
four weeks after the outbreak, the strategy for investigating the
possible involvement af the water supply focused on the study of
long term water quality data from the municipal wells and
distribution system. A computer model was used to investigate the
movement of water through the system. In addition, a general
inspection of the drinking water supply system and operating
practices was also conducted. In the course of the investigation,
two major pipe breaks and numerous curbside meter replacements
were done just prior to the onset of the outbreak and during the
outbreak. The investigation also showed that the wells used by
the community were 1n protected aquifers and the well heads
appeared sound. This indicated that if contamination occurred, it
had to have happened within the distribution system. The timing
of the pipe breaks and meter replacements also helped to implicate
drinking water in the outbreak. The municipal sewage and storm
water collection systems were under designed for the capacity that
was being transmitted and there were indications that the sewage
system was prone to Infiltration from storm water run off. The
sewage collection system routinely overflowed during rain events
and sewage products were visible around many manhole covers. One
of the known sewage overflow areas was in the proximity of one of
the major water supply pipe breaks and several of the meter
replacements. Water utility personnel noted that in many of the
meter replacements, water had to dipped out of the meter box
before replacing the water meter.
Circumstantial evidence strongly suggested that a break in
the public health barrier concept did occur between sewage, storm
water, and water supply. For example, six cases of bloody
diarrhea were identified as having occurred prior to the first
water main break but after 43 meter replacements on the system.
Seven other cases were reported between the two water main breaks
that occurred 3 days apart, with the remaining 72 cases identified
within a week after the second break. This situation points to
the possibility that E. coli 0157:H7 was prevalent for several
weeks in the water supply.
A dynamic analysis of the*movement; of water under normal and
pipe break.conditions was simulated with EPA's Dynamic Water
Quality Model (DWQH).2-3 The model was applied to predict the
5

-------
movement of water flow and contaminant dispersion in the system
under normal operating conditions prior to the breaks being
repaired and hydraulic situations simulating recovery following
the two repairs, or meter replacement. To simulate the break
conditions, a conservative contamination level of 10J organisms
per ml in a 0.6 L/sec flow for a period of 4-hours after break
repair or meter replacement was used to simulate sewage
contamination of the distribution system. Combining the patterns
of organisms spread from both breaks provided an overlay of
contaminated water (10-100 organisms per liter) that covered 85%
of all household case locations (Figures 3,4). The model showed
how rapidly the organisms would spread and how wide an area would
be affected by the contamination occurring at the pipe breaks.
(The pipe break areas were not disinfected after being repaired).
If sewage or surface water drainage was the origin for this
pathogen, E. coli 0157:H7. then the question would arise as to why
the organism and other coliforms were not detected in the
contaminated water supply. It is important to note that no
official monitoring of the public water supply was done during the
35 day outbreak period. One water sample was taken by a nonwater
sampler and a certified laboratory analysis showed 22 total
coliforms per 100 mL (no tests were done for fecal coliforms or E^.
colj). A follow up water sample was taken from the same location
after chlorination was implemented and that sample was negative
for coliforms.
The evidence strongly suggests that fecal contamination of
the water system occurred in the distribution system and the
movement of the water caused the rapid spread of the etiological
agent. Water samples taken by EPA at the extremities of the
distribution system (4 weeks after the outbreak) also showed signs
of fecal contamination but the pathogen E. coli was not detected
in the water system. Because of the elapsed time between the
outbreak and the engineering investigation, the investigating team
did not believe that water supply samples taken during the
investigation would show E. coli 0157:H7.
In Case Study III, another small town (16,000 population)
experienced several thousand cases of cryptosporidiosis and the
drinking water system was implicated.4 In this case, the
engineering investigation concentrated on the conventional
coagulation, sedimentation, and filtration surface water treatment
plant. Earlier EPA laboratory and pilot plant research5 had
indicated that turbidity breakthrough, or passage of particulates,
could be accompanied by protozoan cyst breakthrough. At this
plant, turbidity was not routinely measured on each of the ten
filter effluents, (only clearwell measurements were required) but
were on a few occasions during the investigation. Analysis of the
filtered water turbidity of each filter's effluent suggested that
the practice of stopping and restarting filters (Figure 5) without
backwashing resulted(jn-,lv1.gh^	water passing
through the filters '(Tablev;ThM^passage of turbidity would
have also allowed passage of Cryptosporidium oocysts if the

-------
oocysts were present In the raw water. The water leaving the
treatment plant had a turbidity of less than 1.0 NTU (and below
0.1 NTU on many occasions), and always had a chlorine residual
present. Further investigation of the plant showed that the
flocculation system was not performing optimumly, and
modifications were suggested to improve flocculation which in turn
would improve sedimentation and lessen the load on the filters.
Each of the three case studies described have shown a
different source of contamination of the drinking water. Each
outbreak required the investigative team to evaluate a different
part of the water supply system as to what was the causing factor
related to the drinking water system. Each of these cases also
show that if the pathogen is known before the investigation
begins, a probable path of contamination may be suggested. These
three case studies ar6 not meant to be all inclusive of the
problems in investigating waterborne outbreaks, but are intended
to be representative of waterborne outbreaks.
OTHER AREAS OF CONCERN
In addition to the events investigated in the case studies
presented, there are many other areas that should be considered in
an outbreak situation. Some of these areas include making a
complete inspection the treatment (or nontreatment) facilities
based on visual observation and review of historical and current
data records. This evaluation would include reviewing the source
water quality, intake structures (and locations), the entire
treatment train (operation and equipment), effluent water quality,
and distribution system. This investigation should also include
looking at abnormal conditions that include severe weather events,
power loses, pipe breaks, fire demands, and even illegal dumping
of contaminants. In the future, better reporting and new
analytical techniques may help investigators to do a better job in
tracking the causes of waterborne disease outbreaks, and to make
recommendations that may help prevent future outbreaks.
There are two publications that are very good handbooks to
help in waterborne disease investigations. These books are:
1. Methods for the Investigation and Prevention of
Waterborne Disease Outbreaks. Edited by Gunther F.
Craun, U.S. Environmental Protection Agency, Office of
Research and Development, EPA/600/l-90/005a, September
1990. (Note: All participants of this Sunday Seminar
will receive a copy of the above book).
2. Basic Need-to-Know on How to Conduct a Sanitary Survey
of Small Water Systems. Learner's Guide for the
Training Course, U.S. Environmental Protection Agency,
Office of Ground-Wate'r and Drinking Water. January
1992 1

-------
Both of these books have chapters that deal with specifics
on conducting treatment plant evaluations, watershed protection
surveys, and distribution system analysis, and would be good
guides to have on an investigator's desk.
8

-------
REFERENCES
1.	Cholera Epidemic in the Americas - Update 193-2, Diarrheal
Diseases Section, Centers for Disease Control.
2.	Methods for the Investigation and Prevention of Waterborne
Disease Outbreaks, U.S. Environmental Protection Agency,
EPA/600/1-90/0Q5a, September 1990.
3.	Geldreich, E. E., et al. Searching for a Mater Supply
Connection in the Cabool, Missouri Disease Outbreak of
Escherichia Coli 0157:H7, Water Research, 26:8:1127 (1992).
4.	togsdon, G. S., et al. Trouble Shooting an Existing Plant,
Presented at AWWA Annual Conference June 1988
t
5.	Fox, K. R., Removal of Cryptosporidium in laboratory and
Pilot Plant Studies, in Advances in Filtration and
Separation Technology, Vol 5, American Filtration Society,
1992.
9

-------
TABLE 1. CASE STUDY III-- FILTER WATER TURBIDITY (NTU)
(February 3)
Clean Filters*	Dirty Filters
n #5 #7 #9	#3 #8 #10
6:19 pm 0.11 0.10 -- --	0.20
6:39 -- -- 0.07 0.08	-- 0.41 1.6
6:59 -- -- -- --	0.54 8.8 2.2
7:14 (Settled water at Filter #8 is	8.8 NTU)
7:33	0.74 2.8 2.5
9:08	0.86 0.54 3.2
9:15 (Settled water at Filter	#8 is 8.8 NTU)
9:20	0.18 0.10 0.84
9:35 pm	0.10
The clean filter samples were taken from filters that had been
recently backwashed and put on line. The dirty filters were
filters that had been used for a short period of time, shut down,
and then restarted without backwashing.
10

-------
Untreated or inadequately
treated surface water
49%
Untreated
or
inadequately
treated
groundwater
Miscellaneous
11%
Distribution or
16% storage
deficiencies
Figure 1.
Causes of 502 waterborne disease outbreaks, 1971-1985.

-------
Penthouse
Storage Tanks
Building
Distribution
System
i
Municipal Water Line
Basement
Filters
Figure 2. Building Schematic (Case I)
12

-------
Figure 3. Dynamic Simulation of Contaminant Introduced at First Break

-------
Rf ure 4. Dynamic Simulation of Contamination Introduced at Second Break
IB. IB - Valla
T50. TOOa TT - Tknks
A - MaUr
a*
D - Dklry
blut of
CwitomlMtkB
m

-------
U1
CD
Q
a>
o_
CO
CD
£•
Q
o
CO
n
eg
to
CD
cr
16
14
12
10
8
Outbreak Period
10	20	30	40
Days Surrounding Crypotosporidiosis Outbreak
Figure 5. Filter operation surrounding the outbreak (Case III)

-------
TFCHNICAL REPORT DATA
Pf'jcse read innmcccus on the reverse dciore cuinp
1 REPORT NO. • 2-
EPA/600/A-93/256 ' \
4. TI7.E AMD SUBTITLE
ENGINEERING ASPECTS OF WATERBORNE DISEASE OUTBREAK
INVESTIGATIONS
5 REPORT DATE
6. PERFORMING ORGANIZATION CODE
7. AUTHORIS]
KIM R. FOX
8. PERFORMING ORGANIZATION REPORT NO
9. PERFORMING ORGANIZATION NAME AND ADDRESS
USEPA DWRD/RREL/ORD
26 W MARTIN LUTHER KING DR
CINCINNATI, OH 45268
10. program element no.
11 CONTRACT/GRANT no
12. SPONSORING AGENCY NAME AND ADDRESS
uS M^ff/J$gineering Laboratory Cincinnati, OH
26 W MARTIN LUTHER KING DR
CINCINNATI, OH 45268
13. type of report and period covered
Published Paper
14. sponsoring agency code
EPA/600/14
15. SUPPLEMENTARY NOTES
TO BE PUBLISHED IN PROCEEDINGS AWWA NATIONAL CONFERENCE, SAN ANTONIO, TX JUNE 1993
Project Officer = Kim Fox (513) 569-7820
16. ABSTRACT -- . .
ABSTRACT
—<= Two recent headline causing events have reinforced the concern about the spread of
diseases through waterborne routes. These events include the current cholera
epidemic in the Western Hemisphere that has caused more than 750,000 reported case -
of cholera through April, 1993, and the more recent cryptosporidiosis outbreak in
Milwaukee, Wisconsin where 370,000 people had watery diarrhea during the time frame
of concern.--Whenever an enteric disease outbreak occurs, or even a small outbreak
happens and drinking water is implicated, an investigation should take place to help
determine >f'drinking water was a contributing factor. The investigation should
.attempt to overly the epidemiological data collected by local, state, and federal
vvShealth agencies, and any data available from the drinking water from the systems.
This paper discusses the types of information gathered in three case studies from
past suspected waterborne disease outbreaks and how the data were used to implicate
water as a mode of transmission. ^T-h'iTlnformation should help those involved with
future outbreaks with guidance to determine what types of information are useful in
waterborne disease investigations.
17. KEY WORDS AND DOCUMENT ANALYSIS
a. descriptors
b.IDENTIFIERS/OPEN ENDED TERMS
c. cosati Field/Group
Waterborne disease outbreaks
Cyrpotospori d i os i s
E. Coli
Multiple barrier concept
Epidemiology
Drinking Water
Diarrhea


IB. mSTR^TT
-------