United States Office of Water April 1983
Environmental Protection Regulations and Standards (WH-553) EPA-440/4-85-021
Agency Washington DC 20460
Water
&EPA An Approach to
Assessing Exposure
to and Risk of
Environmental Pollutants
-------
DISCLAIMER
This is a contractor's final report, which has been reviewed by the Monitoring and Data Support
Division, U.S. EPA. The contents do not necessarily reflect the views and policies of the U.S.
Environmental Protection Agency, nor does mention of trade names or commercial products
constitute endorsement or recommendation for use.
-------
EPA-440/4-85-021
July 1980
(Revised April 1983)
AN APPROACH TO ASSESSING EXPOSURE TO AND
RISK OF ENVIRONMENTAL POLLUTANTS
by
Arthur D. Little, Inc.
Michael Slinak
Project Manager
U.S. Environmental Protection Agencv
U.S. EPA Contracts 68-01-3857
63-01-5949
Monitoring and Data Support Di'/ision (WH-553)
Office of Water Regulations and Standards
Washington, D.C. 20460
OFFICE OF WATER REGULATIONS AND STANDARDS
OFFICE OF WATER
"J.S. ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, D.C. 20460
-------
FOREWORD
Effective regulatory action for toxic pollutants requires an under-
standing of the ecosystem and human health risks associated with the
manufacture, use, and disposal of the substance. The process of assessing
these risks needs to develop information on the fluxes of the substance
through the technosphere and through the biosphere, and to couple this
with information on its biological effects. The analysis is thus in-
tended to allow an informed judgment about the likelihood of environmental
harm and to provide insight into the potential effectiveness of alternative
actions to control or reduce any unacceptable risks.
This document describes the exposure/risk assessment methodology
?™6 T aS ?a" °f 3 Pro«"» to addr^s 65 classes of chemicals (or
129 individual 'priority pollutants") named in the 1977 Clean Water Act.
The methodology is multi-media in scope, enabling all facets of environ-
mental risk to be viewed in perspective.
_ The methodology begins by identifying releases to the environment
during production, use, or disposal of the substance. It proceeds with
evaluating the fate of the substance in the environment and the resulting
ambient levels. It then predicts the human and aquatic life exposure to
the substance and, after interpreting the available data on toxicity
provides an assessment of risks.
The methodology has been applied to the nationwide assessment of
several dozen of the priority pollutants, and numerous examples taken
hlvl bt" fr\have befn P^ented. The analytical elements, however,
,have been found to apply readily to local, as well as nationwide studies.
Michael W. Slimak, Chief
Exposure Assessment Section
Monitoring & Data Support Division (WH-553)
Office of Water Regulations and Standards
iii
-------
TABLE OF CONTENTS
Page
LIST OF FIGURES
ix
LIST OF TABLES ±±
ACKNOWLEDGMENTS xvi
1.0 TECHNICAL SUMMARY
2.0 INTRODUCTION
2.1 Background
2.2 Types of Exposure and Risk Assessments l~]
2.3 Report Objectives and Content ?~^
3.0 EXPOSURE AND RISK ASSESSMENTS—AN OVERVIEW - ,
~ J—i
3.1 Overview
3.2 Initial Considerations in a Risk Assessment l~l
3.3 Materials Balance Environmental Loadin^ i_-
3.4- Monitoring Data 3
3.5 Environmental Pathways and Distribution \~J7
3.6 Exposure of Humans and Other Biota o
3.7 Health and Environmental Effects ,7.,
3.3 Risk Considerations , ;.
3.9 Presentation of Risk Assessments 3~^
4.0 MATERIALS BALANCE-SOURCE IDENTIFICATION AND LOADING
ESTIMATION
4-1
4.1 Introduction
4.2 Goals of a Materials Balance ^~\
4.3 Materials Balance Methods ,
4.4 Examples of Materials Balance Output 7~-
4.4.1 Introduction "
4.4.2 Chloroform ^~^
4.4.3 Copper 7~-
4.4.4 Pentachlorophenol 7~^i
4.5 Selected Examples from Materials Balances for Other '"'""
Pollutants
4.5.1 Releases during Transportation 4!^
4.5.2 Public-ly Ovnea Treatment Works ^q
4.:>.j Natural and Inadverzant Releases 7_]X
^.5.4 Releases to the Atmosphere 7~-5r,
References ' -+--50
4-32
v
-------
TABLE OF CONTENTS (Continued)
5.0 ENVIRONMENTAL PATHWAYS AND FATE ANALYSIS
7.0 HUMAN EXPOSURE AND EFFECTS
3-1
5.1 Introduction 5_]_
5.2 Goals of Environmental Pathway and Fate Analysis 5-4
5.3 Environmental Pathway and Fate Analysis Methods 5-5
5.3.1 Environmental Scenario/Case Example Method 5-6
5.3.2 Critical Pathway/Distribution Estimation Method 5-9
5.3.3 Modeling Approaches 5-12
5.4 Examples of Environmental Pathways and Fate Analysis 5-14
5.4.1 Environmental Scenario Method 5-1-
5.4.2 Critical Pathway/Distribution Estimation Method 5-13
5.4.3 Modeling Approaches 5-19
5.4.3.1 Phthalate Esters 5-19
5.4.3.2 Dichiorobenzenes 5-21
References 5-^5
6.0 MONITORING DATA AND ENVIRONMENTAL DISTRIBUTION 6-1
6.1 Introduction 6_]_
6.2 Goals and Objectives £_••>
6.3 Methods and Approaches 5-3
6.4 Examples of Monitoring Data 6-5
6.4.1 Copper and Silver 5-5
6.4.2 Pentachlorophenol 6-11
6.4.3 Dichloroethanes 6-13
References 6-16
7-1
7.1 Introduction 7_]_
7.2 Goals and Objectives 7.3
7.2.1 Human Exposure Analysis 7-3
7.2.2 Human Effects Analysis 7-4
7.3 Approaches and Methods 7-4
7.3.1 Exposure Analysis 7-4
7.3.1.1 General Approach 7-i
7.3.1.2 Sources of Exposure, Exposure Routes
and Subpopulation Groups 7-7
7.3.1.3 Exposure Levels from Major Exposure
Routes -_S
7.3.1.4 Summarizing Exposure 7-2S
7.3.2 Effects Analysis 7-23
7.3.2.1 General Approach 7-28
7.3.2.2 Details of Approaches and Examples 7-38
References 7-61
VI
-------
TABLE OF CONTENTS (Continued)
8.0 EXPOSURE AND EFFECTS—NON-HUMAN BIOTA
8.1 Introduction g,
8.2 Goals and Objectives Q_T
8.2.1 Exposure Analysis g_3
8.2.2 Effects Analysis g_3
8.3 Approaches and Methods g_4
8.3.1 Overview g_^
8.3.2 Effects Analysis g_5
8.3.2.1 Data Collection and Preliminary Data
Review g_r
8.3.2.2 Critical Data Review and Tabulation 8-7
8.3.2.3 Summary of Effects 3-3
8.3.3 Exposure Analysis 3-1?
8.3.3.1 Introduction 8-12
8.3.3.2 Identification of Sensitive Species 8-14
8.3.3.3 Identification of Areas with Expected
or Measured High Concentrations 8-14
8.3.3.4 Identification of Factors Modifying
Availability ' ° 8-lb
8.3.3.5 Identification of Locations in Which
Risk to Aquatic Organisms is Likely
to Occur 8-19
8.3.4 Terrestrial Effects and Exposure Analysis 8-19
References ' a T
9.0 RISK CONSIDERATIONS 9-1
9.1 Introduction „_,
9.2 Goals and Objectives o_0
9.3 Approaches and Methods o_2
9.3.1 General Considerations 9_3
9.3.1.1 Definitions of Risk 9.3
9.3.1.2 Overview of Evaluation Approaches 9-4
9.3.1.3 Approaches Described in the Literature 9-7
9.3.2 Evaluation of Risk to Human Health 9.3
9.3.2.1 Overview o_8
9.3.2.2 Qualitative Risk Analysis 9-9
9.3.2.3 Semi-Quantitative Risk Analysis 9-10
9.3.2.4 Quantitative Risk Analysis 9_n
9.3.3 Evaluation of Risk for Aquatic Species 9-70
9.3.4 Summary of Risk Considerations Q_O/,
References " ^_
VII
-------
TABLE OF CONTENTS (Continued)
Page
10.0 BIBLIOGRAPHY OF REFERENCE MATERIALS FOR USE IN EXPOSURE
AND RISK ASSESSMENTS 10-1
10.1 Introduction 10-1
10.2 Materials Balance 10-2
10.3 Fate and Pathways Analysis 10-6
10.4 Monitoring Data and Environmental Distribution 10-10
10.5 Human Exposure and Effects 10-12
10.6 Effects and Exposure—Non-Human Biota 10-15
10.7 Risk Estimation 10-16
APPENDIX A. MATHEMATICAL DETAILS OF RISK CALCULATIONS A-l
A.I Introduction A-l
A.2 Human Equivalent Doses ' A-2
A. 3 One-Hit Models A-5
A.4 Linear Extrapolation A-6
A.5 Log-Probit Extrapolation A-9
References A-14
viii
-------
Page
LIST OF FIGURES
Figure
No.
1-1 Major Components in Environmental Exposure and
Risk Assessment
3-1 Overview of Environmental Risk Assessment Process
3-2 Example of Materials Balance Checklist 3-6
3-3 Schematic Example of Fate and Pathways Analysis 3-9
3-4 Human Exposure Matrix 3-1?
4-1 Generalized Materials Balance Flow Diagram Showing
Typical Releases 4_9
4-2 Materials Balance Methodology Flow Chart 4-6
4-3 Example of Graphic Representation of Materials Balance
Output—Environmental Loading of Chloroform, 1978 ~ 4-16
4-4 Example of Graphic Representation of Materials Balance
Output—Environmental Loading of Copper, 1976 4-10
4-j Example of Geographic Distribution of Production 4-2.°
Sources—Locations of Pentachlorophenol Manufacturing
and Wood Treatment Plants
4-6 Example of Regional Distribution of Use Sources-
Regional Estimated Consumption of Pentachlorophenol
by Wood Preservation Plants ' £_?.
4-7 Example of End Use Data—Materials Treated with
Pentachloropehnol, 1978 /._-.,;
4-8 Example of Regional Consumption Data—U.S. Regional
Consumption of Wood Treated with Pentachlnroohenol*
1978 ' " , „„
5-1 Example of Environmental Pathways and Fate Analysis-
Major Environmental Pathways cf Trichloroethylane 5-2
5-2 Diagram of Environmental Scenario Approach to Pathways
and Fate Analysis ' " -_-,
5-3 Diagram of Critical Pathway/Distribution Estimation
Method for Environmental Pathways and Fate Analysis 5-10
5-4 Example of Environmental Scenario Identification—
Schematic Diagram of Major Pathways of Copper Released
to the Environment by Human Activities "' "-!'
ix
-------
LIST OF FIGURES (Continued)
Figure
No- Page
5-5
Example of Environmental Scenario Analysis—Typical 5-16-
Environmental Pathways of Copper ^-17
5-6 Example of Use of Simple Quantitative Model to Estimate
Environmental Distribution—Ground-Level Concentrations
of Pentachlorophenol in the Plume Downwind of a Cooling
Tower (Two Source Heights)
5-7 Example of Results of Modeling of Environmental Distri-
bution—Comparison of Calculated and Observed Levels
of Di(2-ethylhexyl)phthalate in Air, Sediment, Water,
and Fish 5-2?
6-1 Example of Surface Ivater Monitoring Data Distribution
by Concentration Ranges—Copper, 1970-1979 6-6
6-2 Example of Geographic Distribution of Monitoring Data
for Silver g_-
7-1 Example of Graphic Summary of Routes of Human Exposure
to Lead 7-29
7-2 Example of Graphic Summary of Estimated Exposures to
Lead for the General Adult Population 7-30
7-3 Example of Graphic Summary of Estimated Exposures to
Lead for a Specific Subpopulation (Children with Pica) 7-31
7-4 Flow Chart for Carcinogenic Risk Evaluation 7-42
7-5 Flow Chart for Mutagenicity Risk Evaluation 7-45
7-6 Flow Chart for ,,Teratogenicity Risk Evaluation 7-50
7-7 Flow Chart for General Evaluation of Chronic Functional
Disorders 7-55
7-8 Possible Protocol for Evaluation of Data on Chronic
Functional Disorders Resulting from Dermal Absorption 7-56
3-1 Flow Chart of Methodology for Effects and Exposure
Analysis for Non-Human Biota ' Q z
fj ~*O
3-2 Example of Graphic Presentation of Overlap Between
Observed Concentration and Water Oualitv Criteriou
for the Protection of Aquatic Life—Zinc $-17
O-1
Flow Chart for Developing Risk Considerations 9-6
-------
LIST OF FIGURES (Continued)
Figure
No. D
Page_
9-2 Example of Risk Considerations Summary for Aquatic 9.95
Biota—Arsenic Exposure and Toxicity to Aquatic
Organisms
A-l Cumulative Distribution Function (Px, t) A-3
A-2 Log-Probit Form for Dose-Response Curve A-10
xi
-------
LIST OF TABLES
Table
No.
Page
2-1 Parameters chat Characterize Risk Assessments 2-3
3-1 General Goals of Environmental Risk Assessment 3-2
4-1 Example of Source and Environmental Loading Matrix:
Phthalate Esters _
4-2
5-2
Source Identification for Materials Balance Matrix
4-3 Example of Materials Balance Output Involving Inadver-
tent Releases—Estimated Production and Use/Release of 4-14-
Chloroform, 1978 4_15
4-4 Example of Commercial Production and Use Data—Summary
of U.S. Copper Supply and Demand, 1976 " 4-17
4-5 Example of Materials Balance Output Involving Natural
Sources—Estimated Environmental Releases of Conner '-*-•> 8
1976
4-6 Example of Materials Balance for Publicly Owned Treat-
ment Works: Zinc ' 4-?]
5-1 Example of Results of Modeling of Environmental Dis-
tribution—Comparison of Results from MacXay's Equili-
brium Model and EXAMS for 1,2-Dichlorobenzene in a
Pond Svstem
Example of Results of Modeling of Environmental Dis-
tribution—EXAMS Output for 1,2-Dichlorobenzene
5-23
5-24
6-1 Example of Surface Water Monitoring Data Distribution
by Major River Basins—Copper g_g
6-2 Example of Sediment Monitoring Data Distribution by
Major River Basins—Copper " 5.9
6-3 Example of Surface Water Monitoring Data for Copper bv
Minor River Basins ' 6_1(
6-4 Example of Monitoring Data for Food and Feed—Penta-
chlorophenol g_--
6-5 Example cf Monitoring Data for Human Tissue and
Urine—Pancachloropher.ol A_T -
xii
-------
LIST OF TABLES (Continued;
Table
No.
6-6 Example of Ground Water Monitoring Data for Dichloro-
ethanes _ ,
b-j.4
6-7 Example of Monitoring Data for Dichloroethan-s in
Ambient Air , , .
Exposure Matrix
/ -o
7 2 Respiratory Volumes for Humans Engaged in Various
Activities
/ — 1J
7-3 Example of Estimated Inhalation Exposure to Trichloro-
ethylene
/ — LL
7-4 Per Capita Consumption of Fishery Products and Food
Fats and Oils in the U.S., 1976 7_15
/-5 Example of Estimated Ingestion Exposure to Di(2-
ethyl-hexyl)phthalate via selected Food Items ' 7-13
7-6 Example of Ingestion Exposure Estimates for Cooper
Based on Total Diet Studies ' 7,Q
7-7 Example of Ingestion Exposure Estimates of Mercury
tor a Specific Subpooulation - ^
i ~~ ±.\j
7-3 Example of Estimated Exposure to 1,2-Dichloroethane
via Drinking Water Including Population Size -_:>•>.
7-9 Example of Estimated Maximum and Typical Human Ex-
posures to Trihalomethanes via Drinking Wacer
7-10 Examples of Estimated Exposures to Pollutants bv
Absorption through the Skin
7-11 Example of Exposure Estimates of Lead for Adults and 7-3?
Children, including Estimated Absorbed Dose 7.34
7-12 Example of Laad Levels in Blood in Support of Exposure
Estimates
.' -JO
7-13 Matrix for Initially Organizing Analvsis of Human-
health Effects Information. - —
xiii
-------
LIST OF TABLES (Continued^
Table
No.
Pase
7-14 Example of Presentation of Mutagenicitv Data—Incidence
ot Chromosomal Aberrations in Spermatogonia of Phenol-
Treated Mice 7 , ,
/-4 /
7-l:> Example of Presentation of Teratogenesis Data—Effects
of Copper Salts in Hamsters " " 7.53
7-16 Tissue Growth Characteristics: Various Animals 7-59
8-1 Example of Acute Effects Data for Freshwater Fish—
Phthalate Esters 3_9
3-2 Example of Chronic/Sub lethal Effects Data for Fresh-
water Fish—Zinc i 1-iQ
8-3 Example of Lowest Reported Mercury Effects Data for
Aquatic Organisms
8-4 Example of Ranges in Effects Levels for Aquatic Biota—
Silver 8-15
8-5 Example of Data on Fish Kills—Phenol 8-16
8-6 Example of Consideration of Bioavailability Observed
in Concentrations of Zinc in Surface Water 8-13
9-1 Example of Risk Considerations by Use of Margins of
Safety—Petitachloropbenol 9-12
9-2 Example of Adverse Effects Summary—Adverse Effects of
Lead on Man 9-13
9-3 Example of Epidemiological Evidence of Human Exposure—
Lead Blood Levels in Man 9-14
9-4 Example of Carcinogenicity Data Used for Risk Extra-
polation of 1,2-Dichloroethane 9-18
9-5 Example of Estimation of Unit Carcinogenic Risk: Estimat-d
Number ot Excess Lifetime Cancers Per 1,000,000 Population
Exposed to Different Levels of 1,2-Dichloroethane ' 9-21
xiv
-------
LIST OF TABLES (Continued)
Page
9-6 Example of Estimation of Carcinogenic Risk due to
Environmental Exposures: Estimated Ranges of
Carcinogenic Risk to Humans due to 1,2-Dichloro-
ethane Exposure for Various Routes of Exposure 9-22
A-l Factors for Converting Doses Administered in
Laboratory Animal Studies to Human Equivalent
Doses " , ,
A-t
XV
-------
ACKNOWLEDGEMENT
This report: was prepared by Arthur D. Little, Inc. under Contracts
63-01-3857 and 68-01-5949. The Project Directors for this work were
Dr. George H. Harris and Dr. Alfred E. Wechsler. Principal contributors
to the development of the methodology described include: Rosaline C.
Anderson, Sam P. Battista, Marcos Bonazountas, Susan Coons, Charles B.
Cooper, Alan Q. Eschenroeder, Joseph Fiksel, Diane E. Gilbert, Muriel E.
Goyer, Judith C. Harris, Karl D. Loos, Warren J. Lyman, Pamela W.
McNamara, Joanne H. Perwak, Gerald R. Schimke, Kate M. Scow, Andrew
Sivak, and Richard G. Thomas.
The authors gratefully acknowledge the support and extensive con-
tributions of the staff of the exposure Assessment Section, Monitoring
and Data Support Division, of the Environmental Protection Agency, in°
particular, Michael W. Slimak, the Project Director; Charles Deles and
Michael Callahari were also helpful in guiding our efforts.
xvi
-------
an
1.0 TECHNICAL SUMMARY
There is a continuing need on the part of regulatory agencies
and_pnvate industry to describe and interpret the impacts of potentially
coxic_substances in the nation's environment. In this context, exposure
and_risk assessment methods are required to permit a quantitative charac-
terization of the sources, environmental pathways, and human or environ-
mental effects ot specific substances. In order to assist in the
determination of appropriate regulatory actions, the Monitoring and Data
rSf ^!isJ0n.(MDSD> of the Office of Water Regulations and Standards
IOWES), U.S. Environmental Protection Agency (U.S. EPA) has developed an
an^haTln^iT^10 ^T11 "" ?erforain§ exposure and risk assessments,
and has applied this approach to approximately sixty environmental ooilu-
tants of concern. Although the approach was developed using waterborne
pollutants its elements may be applied in a wide variety of situations
at varying levels of detail. This report describes the exposure and
ris* assessment methodology and provides selected examples of the use
or the methodology.
For purposes of this report, exposure is defined as the eacouncer
of a substance in the environment by human or animal populations, and
r^sk is detined as the probability of an exposed organism sufferin
adverse efrect as the result of such exposure.
The scope of an exposure or risk assessment mav be characterized
by a number of key features:
• Geographic scale, which may be global, national
regional, or local.
• Pollutant sources, which may include industrial,
residential, commercial, and non-point sources.
« Environmental media, which may include air, surface
water, soil, groundwater, biota, or any combination
tnereof.
• Pollutants addressed, which may be a specific substance
or a class of related substances.
• Receptor populations considered, which may include
humans, animals, plants, micro-organisms/or specific
sub-populations of the above that are exposed to
unusually high pollucant levels.
• Adverse effects considered, which may include acute
or cnronic health effects as well as environmental
efrects.
-------
• Time frame of the assessment, which may be retrospective,
current or prospective.
• Intended use of the assessment, which may be for regulatory,
scientific, or public information purposes.,
The methodology summarized and described in this report is sufficienr.lv
flexible so that it can be applied with respect to any of the above
definitions of scope.
An environmental exposure and risk assessment for a chemical substance
generally consists of a series of analytic components, or modules, each
addressing a particular set of relevant information about the substance.
These components are linked together as shown in Figure 1-1, culminating
in an evaluation of risks to humans and other biota due to the presence
of the substance in the environment. The essential aspects of each
component are as follows:
• Initial Considerations—the available information about
the substance and important environmental issues are
identified, the scope and focus of the detailed exposure
and risk assessment are established, and the subsequent
work effort is planned and organized.
" Materials Balance—the significant pollutant sources
are identified, and the locations and magnitudes of
environmental releases are characterized. This involves
a systematic examination of the various activities which
produce, transport, use, or consume the substance, and
often requires estimation of environmental loadings in
the absence of empirical knowledge.
• Monitoring Data—the concentrations of the pollutant
in all environmental media are investigated through
scanning of field data, and important temporal or
geographic variations are noted. The monitoring data
provide a means of confirming some of the materials
balance and environmental fate estimates.
» .Environmental Pathways and Distribution—the mechanism
of pollutant transport and transformation in the environ-
ment are investigated, leading to an assessment of the
substance's persistence and its likely partitioning
among the various environmental compartments. This may
involve the use of mathematical models to estimate the
distribution of the substance in specific media.
» Exposure of Humans and Other Biota—the potential exposure
of humans and other species is assessed through an in-
vestigation of the important environmental exposure
routes and the extent or frequency of exposure. For
-------
INITIAL CONSIDERATIONS
Monitoring
Data
Materials
Balance
Pathways
& Distribution
Health &
Environmenta"
Effects
Exposure of
Humans &
Other Biota
Principal
Information Flow
Selective Data Inputs
FIGURE 1-1
Risk
Considerations
.MAJOR COMPONENTS IN ENVIRONMENTAL EXPOSURE AND RISK ASSESSMENT
1-3
-------
humans this may address not only geographic differences,
but also the identification of specific subpopulations
that may have higher :han average exposure via ingestion,
inhalation, or dermal absorption.
• Health and Environmental Effects—the potential acute
and chronic effects of the substance are evaluated for
both humans and other species. Data on human effects
may come from either epidemiological or laborator}?-
studies, and will focus upon those effects most perti-
nent to the prevalent chemical forms and exposure routes
of the substance in the environment. To the extent
possible, metabolic information is also taken into
account.
* Risk Considerations—:he results of the previous components
are combined to yield an assessment of the potential
health risks to humans and other species due to the
presence of the substance in the environment. This
may involve simple comparisons of toxic levels with
environmental levels, or, as in the case of carcinogenic
effects, may require extrapolation of laboratory animal
dose-reponse data using mathematical models.
Each of the above components is treated in detail in separate
chapters of this report. A comprehensive discussion is given of the
means for collecting and interpreting relevant data, formulating and
applying analytic models or techniques and consolidating and presenting
the results. In addition, specific examples are provided of how these°
methodological components have been used for exposure and risk assess-
ments of selected priority pollutants.
An important issue that is addressed throughout the report is
data adequacy and the associated levels of confidence in che exposure
and risk assessment results. Depending on the accuracy and completeness
of the required data, the results can range from well-defined numerical
estimates to rough qualitative statements. Moreover, many of the tech-
niques utilized to analyze data, notably fate modelling and dose-response
extrapolation, involve a number of assumptions which may not be fully
verifiable. Therefore, it is crucial that the outputs of the exposure
and risk assessment are properly qualified in terms of model and'data
limitations. Despite such limitations, a well-organized and scientifically-
documented assessment can be an extremely useful instrument for understanding
pollutant impacts and guiding regulatory actions.
1-4
-------
2.0 INTRODUCTION
2.1 BACKGROUND
The growing concern regarding the nature, distribution, and poten-
tial effects of toxic and other hazardous chemicals in the nation's en-
vironment has been reflected in many federal government statutes rela-
tions, and rules. The Clean Water Act, the Clean Air Act, Resource °
Conservation and Recovery Act, Toxic Substances Control Act and Dela-
tions under these statutes have addressed these concerns, and several
regulatory agencies are charged with implementation of environmental
management responsibilities (for example, the Environmental Protection
Agencv. Consumer Product Safety Commission, and the Food and Drug
Aaministration). Industry organizations and independent research groups have
also investigated sources, pathways, and effects of potentially to—'c
materials and the exposure for humans and other species to these mortals
as part of a nationwide environmental program.
Throughout many of these efforts, there has been a focus OP the
analysis of risks associated with the presence of toxic and hazardous
chemicals in the environment. This analysis process, often referred to
as risrc assessment or "exposure assessment;1 encompasses many aspects
including in-depth toxicoiogical experimental investigations of health
effects using laboratory animals, environmental monitoring and ^easure-
ments, and extensive data collection and/or modeling efforts to determine
and predict tne concentrations and fate of toxic substances in the environ-
men.. This work is expected to continue at many levels, both oubliclv
and privately sponsored, throughout the foreseeable future.
The Monitoring and Data Support Division (MDSD) of the Office of
kater Regulations and Standards (OWES), U.S. Environmental Protection
Agency (J.S. EPA), is conducting a program to evaluate the exposure to
and risk or pollutants in the nation's environment. Part of this P£tor-
is a result ot the settlement agreement between the Natural Resources
Defense Council (NRDC) and the Environmental Protection A-enc- ru S
District Court, D.C., 1976)X. Under this agreement, the Monitor-in- and
Data Support Division is evaluating the exposure and risks to human and
non-human species resulting from the occurrence of 129 specific chemicals
in Cl,e water environment (hereafter referred to as the 129 orioritv
pollutants). On the basis of these evaluations, recommendations for
. Serlemen£
^
-------
regulatory actions are prepared to reduce the exposure to and risks of
priority pollutants in the environment. In order to provide a systematic
and comprehensive evaluation approach, an integrated risk assessment
methodology is being developed. This methodology is the subject of the
report.
7 7
TYPES OF EXPOSURE AND RISK ASSESSMENTS
Exposure and risk assessments vary with respect to their scope and
use. The scope of the assessment can be described by several parameters.
The scale of an assessment is defined by whether global or national,
regional or local exposure or risks are considered. An assessment can
also be characterized by which populations are considered (humans,
plants, animals, microorganisms, or all environmental species) and whether
average nationwide risks are evaluated or risks to specific subpopulations
in specific areas. The time frame of an assessment can be retrospective,
current, or prospective. Finally, an assessment can include evaluation
of one or many of the potential health or environmental effects associated
with the presence of a toxic substance in the environment.
An exposure assessment involves examination of all factors that
lead to an exposure for human and other species to the pollutant and a
quantification of that exposure. A risk assessment includes all the
elements of an exposure assessment and a qualitative or quantitative
estimation of the risk to a given population based upon the exposure
to and effects of a pollutant (e.g., the increased risk of carcinogenicitv
to the total U.S. population associated with the environmental presence
of a chemical). Throughout this report we will use the terms "risk
assessment" and "exposure assessment" interchangeably, recognizing that
risk assessments combine both analysis of exposure and analysis of
effects to yield an assessment of risk. (In the published literature.
these quantitative relationships between risk and exposure are often
called risk assessments.)
Risk assessments may have many different uses. Typical uses include:
development of regulatory approaches, requirements, or recommendations;
development of environmental standards and/or criteria; establishment
of information, monitoring, or research needs; providing public information,
education, etc. Table 2-1 characterizes risk assessments in accordance
with all of these general parameters. The methodology described herein
is called an integrated risk assessment methodology since, in principle,
the approaches used are applicable to all types of risk and
exposure assessment, independent of scope, depth or other characteristics:
specific portions of the methodology are suitable for independent
studies and assessments.
2.3 REPORT OBJECTIVES AXD GONTZNT
The objective of this report is to describe an integrated exposure
and risk assessment methodology. The methodology is intended to be used
by public and private organizations and individuals who seek guidance on
7-9
-------
TABLE 2-1. PARAMETERS THAT CHARACTERIZE RISK ASSESSMENTS
Scale
National, regional, local
Populations
Considered
Humans, plants, animals, micro-
organisms, all species
Time Frame
Retrospective, current, prospective
i Potential
Effects
Human Health—carcinogenicity,
chronic functional disorders, etc.
Ecological—habitat, foodchain,
reproductive, etc.
Intended
Use
Development of regulations, environ-
mental standards, criteria
Establishment of information or
research needs
Public information/education
-------
conducting exposure and risk assessments, and to provide a foundation
for further development of the methodology. Not all portions of the
metnodology have been considered in the same depth of'detail. For
certain aspects of the analysis, such as ecological modeling and
toxicological research, unique methodological approaches have already
been developed to a high degree of sophistication. Therefore, users'
may require additional detail in some areas depending upon the overall
purpose, level of effort, and intended use of the individual risk
assessment. This report presents basic approaches to risk assessment
so tnat users may select the most appropriate segments for each specific
application. The general approach is intended to guide the plannin* and
conduct of specific exposure or risk analyses rather than provide a°
detailed procedure. Since this risk assessment methodology was developed
for the EPA to address waterborne priority pollutants, it is focused
on assessment of exposure and risk where water contamination or
pollution is significant.
Chapter 3 of this report provides an overview of the risk assessment
process, including the goals and objectives of each major component,
the flow of information from one analysis area to another, and sorae of
the major assumptions and limitations of the process. The initial steps
of a risk or exposure assessment are then discussed. In Chapters 4
through 9, approaches to each component of the risk assessment process
are described in some detail. The organization of these chapters is
as follows:
Chapter 4—Materials Balance—Source Identification and Loading
Estimation
Chapter 5—Environmental Pathways and Fate Analysis
Chapter 6—Monitoring Data and Environmental Distribution
Chapter 7—Human Exposure and Effects
Chapter 8—Exposure and Effects—Non-Human Biota
Chapter 9—Risk Considerations
In each of these chapters, examples are drawn from actual exposure and
risk assessments performed for EPA. These examples are intended to
illustrate methods of data analysis or presentation and the reader is
referred to the full report for specific information regarding each
pollutant.
Chapter 10 provides a bibliography of source materials for the
conduct of exposure and risk assessment of environmental pollutants.
This bibliography is intended to give the investigator an initial means
of obtaining the numerous types of information needed to assess exposure
and risk.
The appendix discusses sorae of the mathematical details of quanti-
fying risk.
2-4
-------
3.0 EXPOSURE AND RISK ASSESSMENTS—AN OVERVIEW
3.1 OVERVIEW
The general goals of an environmental risk or exposure assessment
are snown in Table 3-1. Only some of these goals may actually be realized
in a specific risK assessment, depending upon the specific pollutant,
the resources available, and the time allowed for the assessment process.
_ Figure 3-1 shows the major components of the risk assessment procpss
ihe initial considerations component is intended to establish the scope
and focus of the risk assessment, assign priorities for investigation of
specific environmental pathways, exposures or effects, and provide the
initial basis upon which to proceed with the risk assessment. The mate-
rials balance component refers to a description and quantification of the
tlow of a pollutant from its generation through its initial release into
the environment. The environmental pathways and distribution component
rerers to analysis of the pathways traversed bv the pollutant in the en-
vironment, the intermedia and intramedia transfers that occur, and the
resultant environmental distribution, both spatial and temporal.
Monitoring data can provide a major input into the establishment
of the pollutant distribution. The exposure assessment component
attempts to characterize the type, size, location, and distribution of
populations and subpopulations-human and other biota-exposed to a pol^u-
- o a po
^ in tne environment and to establish actual and potential exposures
to tne pollutant in terms of extent, duration, level", etc. The health
and environmental effects component analyzes the known or anticipated "
acute, chronic, and other effects of pollutants on humans and oche/soecfes
ir possibxe, it provides a basis for extrapolation of the results of '
laboratory effects studies to real environmental situations and/or extrapo-
lation o, results of studies with laboratory animals to human population
S^!;..The riSk '"^derations component summarizes previouslv deviled
ooou^< e8tlmate; l^titativelv, if possible, the risks to various
population groups, and places the risks associated with pollutants sources
environmental pathways, exposure routes, and health effects in perspective
,,3l ^ fh°Wn in,Fi§ure 3-1, the major flew of information is from mate-
ant distributer1 mCnit°rin§ data Components to environmental pathvavs
walvie £- " components These data, combined with environmental fate
an? J ?:• ff analysis Cf exposure of humans and other biota. Exposure
and healtn effects analyses are combined to yield risk estimates. Mate
-ials balance has indirect inputs to health and environmental effects *rd
exposure components; similarly, environmental pathways and distributor
analysis have indirect inputs to health and environmental effects.
r remainder °f this chapter, each of the major components in
Ioa?-a^STnC Pr°C£SS- ^ dl3CUS3ed brief^> including focusing on
goal, and objectives or chese steps, some of the approaches used, ard
3-1
-------
TABLE 3-1. GENERAL GOALS OF ENVIRONMENTAL RISK ASSESSMENT
Establish pollutant sources, pathways and
distribution
Establish exposure to and effects of pollutants
Quantify the human health and biotic risks
Provide information base to derive approaches
to risk reduction
Identify data gaps and research needs
3-2
-------
INITIAL
CONSIDERATIONS
MATERIALS
BALANCE
T
i ___
HEALTH AND
ENVIRONMENTAL
EFFECTS
MONITORING
DATA
i
ENVIRONMENTAL
PATHWAYS AND
DISTRIBUTION
EXPOSURE OF
HUMANS AND
OTHER BIOTA
CONSIDERATIONS
Note: Solid lines indicate direct flows of
information; dotted lines indicate
selective duta inputs.
FlCUKIi 3-| OVliKVIIW OF KNV IKONMI'NTAL RISK ASSKSSMF.N'i
-------
the assumptions and limitations of the risk assessment approach. Subse-
quent chapters of this report discuss each of these components in more detail,
3.2 INITIAL CONSIDERATIONS IN A RISK ASSESSMENT
The initial considerations of a risk assessment include:
• Establishment of the scope and focus of the risk assessment.
» Identification of the subject material to be considered in
highest priority and with greatest detail.
• Development of a work plan and/or approach for completing
the risk assessment.
This component initiates the risk assessment process by providing
a basis for understanding the requirements of the risk assessment, and
an organization for the work conducted throughout the risk assessment
process. To avoid unnecessary effort and development of data on
topics of little significance, it is essential to carefully define
the desired goals and outcome of the specific risk assessment.
The scope should be established in terms of the parameters described
earlier—scale, populations considered, time frame, potential effects,
and intended use of the assessment.
Once the initial scope and focus of the risk assessment have been
defined, the next step is to determine in general terms the type and
availability of information for the risk assessment process. This
can be accomplished through brief literature reviews, consultation
with experts, analysis of recent reviews on particular chemicals,
etc. Next, priority areas of investigation are identified, for
example, a specific pathway, a specific set of health effects, an
industry of significance, etc. Priorities should be set according
to the overall requirements of the risk assessment and the expectations
of availability of information.
Following prioritization of areas of investigation, the final step
in the initial considerations is to develop a work plan for conducting
the risk assessment. The work plan should estimate the effort
devoted to each of the major components, indicate the major areas
of information flow and exchange, and estabiisn a timetable for tne con-
duct of the risk assessment.
After several risk assessments have been performed, the initial
considerations component will become a "natural process." Nevertheless, it
will still be important to identify the overall goals of the risk assess-
ment, establish priorities, and develop a work plan to increase the
potential for achieving those goals.
-------
3.3 MATERIALS BALANCE (ENVIRONMENTAL LOADING)
The objectives of the materials balance component are:
• To identify the important (if not all) pollutant sources
» To identify the chemical and physical form of the pollutant,
as it is released to the environment.
» To characterize the environmental loading of the pollutant-
quantities, geographic locations, rates, receiving environments.
• To identify uses and releases of the pollutant leading to dir2Ct
exposure.
• To achieve a balance between production and uses or releases.
• To establish the confidence or uncertainty of data on releases
of the pollutant.
The materials balance approach requires a systematic identificacion
or sources, estimates of environmental releases, and characterization of
tne receiving environment. A comprehensive analysis may be enhanced
through a checklist or ordered procedure for examinins all aspects of
the processes of generation and release of the pollutant. Figure 3-2
shows an example checklist, indicating a source matrix and an°environ-
mental input matrix. All types of manufacturing processes, transporta-
tion, storage, and disposal activities, as well as uses of the pollutant
or products containing the pollutant should be considered. Specific
processes, uses and releases, and the environmental compartments
receiving the release that can lead to direct exposure potential should
oe identified. As a check on the quantification of releases, the
degree of closure of the materials balance (the relationship of production
import, export, use, disposal, and environmental release data) is
estaolished. The ranges of uncertainty in environmental releases of
data for tne pollutant should also be established; several approaches
tor this task are discussed in Section 4.3.
The materials balance is often a difficult component of the risk
assessment to perform since there are many production processes, trans-
portacion and storage procedures, and use patterns that affect re-
leases to the environment. Processes may not be described, uses mav be
unknown, and quantitative data on releases may not be available Thus
in many cases, engineering estimates will have to be made in order to '
describe likely or expected environmental releases, The assumptions
and uncertainties associated with all environmental releases should be
aocumented wherever possible. Those releases that can result in ^ir-ct
exposure ot persons or other biota to the pollutant should be hi*hli*hfced
as this information will be directly used in the environmental pathwavs
and distribution analysis, as well as the exposure and health er>~e-ts
anaxvses
3-5
-------
SOURCE MATRIX
ENVIRONMENTAL INPUT MATRIX
I
G-\
EXTRACTION
Extraction
Method
drilling
strip inniinij
Type of
Release
routine
accidental
WATER
QUANTIFY
RATE
FORM
LAND
AIR
MANUFACTURING
Manufactuiincj
Processes
chemical
inactions
dujostion
Material
Class
contaminant
by products/
co-products
primary product
1 ype of
Huloase
routine
accidental
TRANSPORTATION
Disposal
I-11cnui' •)-•:> KXAMi-i.r. OK MATUKIAI.S HAI.ANCK
-------
3.4 MONITORING DATA
Goals of the monitoring data component of the risk assessment
are:
• To identify concentrations of the pollutant in all environ-
mental media.
• To determine the geographical and temporal distribution of
the pollutant in these media.
•
To identify geographical locations and other factors asso-
ciated with pollutant releases to the environment and to
proviae data on possible exposure of humans and other biota.
cure nM "mP°nent °ften be?ins with a comprehensive review of litera-
ture including access to available computerized environmental data bases
In? f?i r dlSCUSSed in Section 6'3- From these data, average amb^
and effluent concentrations in air and water nav be established- concen-
rations in soil, sludges, plants, animals, fish tissues, foods,' drink-
'.f^'-f0^ ^ determi*ed> evaluated, and summarized. It
nt to identify, wherever possible, the uncertainties in exper-
°f m°re monitor
indr ^ <« moni
mg data include: uncertainties in the chenical/analytica' orocedures
useu, conndence levels, and detection limits; uncertainties' in obtS-
-ng representative samples of the environmental media; the lack o* d-t-
on tne temporal variations in concentrations at different locations?
uncertainties in the chemical or physical forms of the pollutant? and
the lack of surriciently detailed and/or extensive data! Despite" tnte
limitations, monitoring data can provide an indication of the locations
of pollutant releases to the environment, a potential means f ^oca^on&
conf'rSf £-;rSUre °f ^T13 and °ther bi°ta' and a direct ™*™ of
fate InsiS? matSrialS balance and the environmental pathways and
3-5 ENVIRONMENTAL PATHWAYS AND DISTRIBUTION
If the environment and the pollutants were "static" and adequate monitor-
ing data were available, materials balance and monitorinz data, combined
witn information on receptor distribution could be used to estimate ex-
posure of tiumans and other biota to pollutants. However, the environ-
ment is not static-pollutants are transported, undergo transformation
accumulate and degrade— and the actual environmental distribution of a
pollutant is different from that associated directly with environmental
releases. The environmental fate and pathways component of a risk asse^-
ment is directed at estimating the actual distribution of the pollutant""
in the environment. Specific goals of environmental fate and oathwavs
component are numerous: .CL^WC..,;,
j-/
-------
• To define environmental compartments of importance.
• To identify important transport mechanisms; physical, biological,
and chemical transformation processes, and predominant
chemical forms of the pollutant.
• To summarize transfer and reaction rates, controlling
processes and lifetimes of the pollutant in the environment.
• To trace pollutant pathways from sources to sinks.
• To estimate pollutant concentrations in different environ-
mental media and their time dependence.
• To compare the results of the pathways and fate analysis
with monitoring data.
• To establish relationships between releases to the
environment and,exposure.
A variety of approaches may be used in environmental fate and path-
ways analysis; qualitative estimates may be based upon case examples
or environmental scenarios, simple analytical equilibrium or transport
models, or complex multi-media models. The materials balance component
provides inputs; evaluation of physical, chemical, and biological fata
processes defines che persistence of the pollutant in the environment;
and models are used to estimate environmental concentrations. Figure 3-3
shows one approach to pathway analysis. In utilizing environmental models,
it is important to assess average concentrations in environmental media of
broad geographical distribution, as well as environmental pathways
and resultant concentrations in specific localized areas. The output
of the fate and pathways analysis should yield pollucant concentration
distributions in sufficient spatial and temporal detail to allow
estimates of exposure of humans and other biota.
For many environmental situations, adequate models do not exist or
are just now under development. Furthermore, for new or uncommon chemicals
many of the physical, chemical, and biological properties needed to esti-
mate transformation rates, persistence, and distribution are not avail-
able. For example, few models exist to predict adequately the distribu-
tion of pollutants released from the landfill into ground water and surface
water. Models to estimate residual concentrations of pollutants in edible
foods resulting from the land disposal of sludges, contaminated irrigation
water, pesticide or nutrient application, dry deposition, are in very
early stages of development. Therefore, uncertainties and limitations of
the models should be identifiec., and estimates of pollutant distribution
based upon materials balance, fate and pathways analysis should be compared
with monitoring data.
-------
r
I
vo
REVIEW PHYSICAL. CHEMICAL
AND BIOLOGICAL CHARACTERISTICS
OF POLLUTANT
REVIEW FATE PROCESSES
CONDUCT SENSITIVITY ANALYSIS
TO DETERMINE THE IMPORTANCE
OF SOURCE PARAMETERS ON
ENVIRONMENTAL DISTRIBUTION
AGGREGATE LOADING FOR
DIFFERENT MEDIA
INITIAL ENVIRONMENTAL
PARTITIONING; ESTABLISH
CRITICAL PATHWAYS
USING SIMPLE MODELS TO
ESTIMATE RATES OF CHANGE
AND POLLUTANT CONCENTRATIONS
SUMMARIZE PATHWAYS AND
DISTRIBUTIONS, COMPARE
WITH MONITORING DATA
IDENTIFY EXPOSURE POTENTIAL
FIGURK 3-3 SCHEMATIC EXAMPLE OF FATE AND PATHWAYS ANALYSIS
-------
3.6 EXPOSURE OF HUMANS AND OTHER BIOTA
The analysis of exposure of humans and ocher biotic population groups
to pollutants is one of the most difficult and critical elements of the
risk assessment process. In general, there is no well established method-
ology or^body of literature on exposure. The combinations of exposure
routes, durations, extents, and the numbers and locations of persons or
organisms exposed may be large and difficult to identify or characterize.
Also, there is usually a wide range of estimated and actual exposures
corresponding to the nature and behavior of the subpopulation groups.
However, estimates of exposure are essential, since otherwise the risks
of the pollutant to various population groups cannot be ascertained.
The principal goals of human exposure analysis are:
•
To determine exposure of the general public to the pollutant in
terms of pollutant sources, exposure routes, exposure durations
and frequencies, exposure amounts or extents.
» To determine the exposure of the workplace population to the
pollutant in terms of occupations, types of " facilities or opera-
tions, the numbers of workers exposed and their characteristics,
the exposure routes, durations, frequencies, amounts or extents!
• To identify specific suspopulation groups in terms of geographic
location, size, occupation, age, sex, dietary or recreational
habits, with higher than typical exposures to the pollutant.
• Determine the exposure of individuals in these subpopulation
groups in terms of the aforementioned parameters.
Similarly, the general goals for exposure ana lv sis of biotic popula-
tions are:
• To identify the types, location, and number of biotic species
exposed to the pollutant.
• To determine the exposure routes, exposure durations and
frequencies, exposure amounts or extents for the exposed
species .
» To quantify as best possible the exposure of various species
to the pollutant under consideration.
Although no well established exposure methodologies exist, a
systematic approach to identifying and quantifying exposure is essential
to the process. All routes of exposure, i.e. ," ingestion, inhalation"
and ^ dermal absorption, must be included. Subpopulations should oe identi
fied with specific sources and exposure routes, for example, those
3-10
-------
populations drinking ground or surface water, urban and rural population
groups those with unique dietary patterns, using products containin-
in e« hl Wh° reSiQ2 near S°UrCeS °r Disposal operations.
In establishing exposure, it is important to identify the characteristics
of the population exposed; the duration and frequency of the exposure-
the range average and maximum, of actual or potential exposure for '
individuals in the population group that lead to estimates of dailv intake
for average individuals and those who belong to special population sub-
The inputs to the analysis of exposure come from the materials ba^a
monitoring data tor ambient air and water; concentrations in foods- re
ot patnways analysis; physiologic data such as respiratorv rate, average
drinking water intake, etc.; and the use of models or data which rellte
average daily intake by different exposure routes to total bodv bura^
of the pollutant It is frequently convenient to display exposure data
in a matrix form listing various routes of exposure, exposure parameters
and estimated or observed intakes for the general population and those
Throughout the analysis of exposure, it is important to iden-ifv
data gaps and uncertainties in data. For example, there will frequen
^^^^^^^
1 fata and toring da
e
^
groups v^h varying sources and quality of drinking water/ since this
route is one of tne most significant exposure routes. Similarly dat
data
:r wmaec rveys',ocher f°°d' fish ^^ «* w
1 <, nSeaed t0 SSSeSS exP°su^ through food in-estion
ingestion can oe considered a waterborne route of exposu^since
™ " ~-'
r.
or aquatic and otbe^- biota w'nor^ -i , -pec^es- Uata cm populations
potential concentrates o-: ??« n f^a *f CM then be associ"ed with
and .b-orptlc. assure rootas should be c
-------
5iMuso.ixr-i NVHIIH v-r 5ian:)i.-i
> prnrioossv
r-i
ri
«?
/iiojiL»|ndoj
-------
3 •? EFFECTS ON HUMANS AND OTHER BIOTA
The overall goal of a human health effects analysis is t
rLrTeriZe '^ adVSrSe health SffeCtS in h— that are r
ted to occur as a result of exposure to a pollutant. Specif
or numan health effects analysis are: ^pecit
. To evaluate acute and chronic health effects in humans resultin*
trom exposure to the pollutant based upon occupational or
accidental exposures and/or human epidemiological studies.
animals, test organisms, tissues, cell cultures, or other biota.
* excr-io116 f6 ^StribUCi°n' meCaboli3-> bioaccumulation and
excre.xon of pollutants in humans and laboratory animals to
response. ^ meChanl3mS and relationships between dose and
• To estimate dose-response relationships in humans based upon
epidemiological, accidental human data, extrapolation of labora-
tory animal data and to estimate "no effects" levels in humans
Similarly, the goals of analyzing effects on other biota are:
terrestrial organisms.
• To identify and evaluate the acute, chronic, and reproductive
eftects in various species as functions of exposure level.
To identify factors that influence the availability of a pollu-
ta ' F
tant to biota.
The general approach used to perform the human health effects
analysis includes literature search, analysis of epidemiological and
3-13
-------
between various exposure routes and health effects., and to establish if
animal models are suitable for extrapolation to humans. In undertaking
these analyses, it would be desirable if data were available on humans"
from epidemiologic studies or information on accidental exposures.
Frequently, however, these data are not available and reliance must be
placed on extrapolation of laboratory animal data. For many chemicals,
animal data are not available and only reports of in vitro data exist.
If no data are available on a pollutant, inferences' may be drawn from
data on related pollutants, using due caution. Examination of structure
activity relationships of various pollutants may provide information if
other data are not available.
Wherever possible, multiple studies using different species of
laboratory animals should be utilized. Several methods of extrapolating
dose-response in animals to dose-response in humans should be explored,
and the results of these extrapolations compared with any available epi-
demiologic or accident data. Throughout the analysis and extrapolation
procedures, uncertainties and assumptions used should be identified and
quantified. If possible, the end result of the human effects analvsis
should be the development of quantitative relationships between dose and
response of humans and a clear explanation of the data and rationale
leading to these relationships. In performing a risk assessment, it may
not be necessary to examine all types of health effects if only certain'
exposure routes are applicable. Thus, in the initial considerations of
a risk assessment, it will be important to identify major exposure rouces
so that the effects analysis can proceed in a direct and straightforward
manner.
An analysis of the effects in non-human species can be accomplished
through data collection and preliminary data review, followed by
critical data evaluation and summary reporting of the effects. Data
should be collected on both laboratory studies measuring the effects of
pollutants on various species arid field investigations or case studies
documenting actual effects of the pollutant in the environment. Informa-
tion on fish-kills, field reproduction studies, and ocher field data can
be especially important in verifying effects predicted from laboratory
studies. It is important to understand the experimental conditions of
laboratory tests of effects so that effects parameters such as LD5Q or
LC5Q can be extrapolated to potential field environmental conditions.
Following preliminary data collection, a critical review should be
accomplished. Lethal and sublethal, acute and chronic effects should be
examined for fish and aquatic invertebrates in fresh and salt water and
marine and estuarine species. Important parameters influencing the results
such as pH, temperature, water hardness, type of bioassay, exposure time,
etc., should be considered. The effects of different exposure routes
should be examined. Toxicity to terrestrial plant:; through root uptake
of pollutants in the soil and toxicity to animals through ingestion of
contaminated biota and water should be examined. The effects of the
pollutant on species in the human foodchain should also be evaluated.
After these data have been evaluated, they can be summarized to iJentifv
-------
sensitive aquatic or terrestrial species, "no effects" concentrations,
oTa nr±e"nf PS' *** C°nditi°ns which influence environmental
tor a numoer of important species.
_ The _ end result of the effects analysis for both human and non-hunan
species is a comprehensive summary of health effects data including un-
certainties and ranges in dose-response relationships, and applicability
ot tne efrects data to various potential routes and to real -nvircn- '
mental situations.
3.3 RISK CONSIDERATIONS
The overall goal of the risk considerations component is to develop
a qualitative and/or quantitative understanding of the nature extent
and severity of the risks imposed by a pollutant on humans, fish, wild-
fire and other biota. The specific goals are:
• To estimate the average health risks tc the general human
population based upon average- exposures and ranges of health effects
effects associated with the pollutant.
• To estimate the extent and severity of health risks associared
witn the pollutant in specific human subpopulations that sus'tain
greater than average risks .
t, HP- u C?! average risks to general populations of fish,
shellfish, wildlife, and other species based upon average ex-
posure ana the range of effects associated with the pollutant.
* I* *st;jmata the extent and severity of risks to auboopulations
of fish, snellfish, wildlife and other species that'sustain
nigher than average risks.
• To identify sources, pathways and causal factors associated
witn risks for human and other species in order to understand
possiole methods for risk reduction. uueraCc.na
As indicated earlier, the combination of exposure and health
streets are required to estimate risk to various species. In evaluat-
ing tne risks of an environmental pollutant, a single resu^ will U5uallv
not occur; ratner a risk assessment will describe a spectrum or risks
for subpopulations, characterized bv the tvpe of adverse ef'e-
to determine whether the acute or chronic
toxic errects and exposures are quantifiable, or whether qualitative
measures must oe used. Depending upon the degree of quantifi^on
outcomes or tne risk consideration include: (1) qualitative indications
-------
OL possible risks; (2) estimates of risk for hypothetical exposure
levels; (3) estimates of risks using conservative assumptions on health
effects, or (4) quantitative assessment of risk for subpopulations v'a
various exposure routes.
Several approaches include: a qualitative comparison of exposure
leveis with 'no effects" or "lowest effects" levels to indicat- the
general nature of risks to humans and other biota; a-semi-quantitative
analysis using safety factors and application of daily intake and health
errects data to result in a better defined range of risks for various
exposures of humans, a quantitative risk analvsis to predict, with clear
identification of the inherent assumptions, mortality or morbidity
resulting from exposure of general and subpopulation'groups to the
pollutant. For example, the output of the risk consideration component
may indicate that a margin of safety of 100 or 1000 exists between
typical average exposures of humans and known or extrapolated effects
levels. Another possible output would be an estimation of the ran^e
or numbers of tumors resulting from exposure of the general human popul-
ation to^a known carcinogenic pollutant. Similarly, in terms of biotic
risks, tne output of the risk consideration compbnent could include
either comparison of effects levels for various species with exposure
concentrations or, if possible, quantitative analvsis of mortal!tv of
various aspects as a result of exposure.
In summarizing risk considerations, the uncertainties present in
exposure and effects data should be addressed. The basis for the
effects and exposure data should be carefully examined and confidence
levels established, if possible. Risk quantification for chronic
health effects is often particularly difficult to express.
In order to assist the regulatory process, it is appropriate to
analyse the risks associated with various exposure routes or exposure
scenarios so that the benefits from environmental regulation or control
can be ascertained.
3.9 PRESENTATION OF RISK ASSESSMENTS
An important element of the risk assessment process is the clear.
thorough, and well-documented presentation of data and results in a
manner which can be understood by scientists, technical experts, regu-
lators, and the public. Although it is difficult to address the needs
of these varied audiences, attempts should be made to provide information
in the risk assessment report in various levels of detail, geared to
different readers. Frequently it will be necessary to use secondary
sources of information; but references should be clear and complete'.
It will be important to present information so that other investigators
can examine the validity of the assumptions, data, calculations, and
results for future studies. Only if the risk assessment report is pre-
pared in sufficient detail, with sufficient clarity, will it be most
useful for the purposes intended.
3-16
-------
4.0 MATERIALS BALANCE—SOURCE IDENTIFICATION AND LOADING ESTIMATION
4.1 INTRODUCTION
A thorough understanding of the distribution of a pollutant in the
environment is essential to determining the likelihood that humans and
other biota will be exposed to it and the magnitude of the exposure. In
principle, the distribution of a pollutant can be established by two
methods:
(1) review, analysis, and interpretation of available
environmental monitoring data; and
(2) development of estimates of sources and loadings
(discharges or inputs to the environment) of the
pollutant, coupled with analysis of environmental
pathways and fate of the pollutant.
For some well-studied pollutants, existing monitoring data may be suf-
ficient to provide a comprehensive view'of environmental distribution.
However, for most pollutants, and particularly for new chemicals or
recently identified pollutants, extensive monitoring data are not avail-
able and the environmental distribution must be estimated. Furthermore,
environmental monitoring data alone are not sufficient to establish the'
effects of alternative regulatory control strategies on the potential
risks associated with pollutants since monitoring data do not always provide
positive correlations between pollutant sources and environmental distri-
bution. Since certain chemical or product uses may lead to direct exposure,
assessment of the sources, uses, and environmental loadings of
a pollutant is an important first step in a comprehensive exposure analysis.
In the context of this exposure analysis methodology, the term
"materials balance" is defined as a description of the flow of a pollu-
tant from its generation through its initial release to one of the en-
vironmental compartments (air, water, land). Production and use, source
identification, and pollutant loading studies are often called materials
balances, depending upon the scope and nature of the work conducted. A
comprehensive materials balance analysis involves other evaluations as
well, including the pollutant's transport, storage, common and uncommon
uses, and eventual disposal.
The concept of a materials balance is illustrated in Figure 4-1
which depicts a pollutant (or product in which a pollutant is a contaminant)
at various stages in its life cycle. The pollutant (product) is first
extracted from the natural environment or synthesized, and after initial
transportation and/or storage, may be further manufactured, processed or
transported in many more stages than are shown in the figure. Other key
steps in the life cycle are stages cf use and disposal of the pollutant"
(product). The interior of the large box in the figure represents pro-
cesses generally conducced in the cultural or anthropogenic environment
-------
Ground water
Incineration or
Effluent
Spills/ -
Leaks
Spalls to
Land
Wastes from mining
operations to
water/air/land
ENVIRONMENT
I J>" = Transportation Process
FIGURE 4-J. OENERAMXED MATERIALS BALANCE FLOW DIACRAM SHOWING TYPICAL RELEASES
-------
At any step within the cultural environment, the pollutant may be released
to the natural environment (represented by the space outside of the lar*e
box), e.g., to air, land, water, biota, etc. The releases mav be planned
and controlled (e.g., a permitted discharge to the air or a receiving
waterbody), or accidental and uncontrolled (e.g., a suill from
a rail car or storage tank into the soil or water, or'runoff or leachin*
from abandoned mining sites). The materials balance is more complex when
natural sources of the pollutant already exist in the environment, inde-
pendent ot human activity. For completeness, significant natural sources
must be incorporated in the materials balance, although achieving closure
or oalance of sources, uses, and releases is far more difficult when a
large reservoir already exists in the environment.
Product use may result in a direct exposure of humans or other biota
ror example, lead in paint, cosmetics, cigarettes, etc. In addition
environmental Releases may also result in direct exposure of humans or
other biota, tor example, a release in the workplace. In other situations
the pollutant must be redistributed in the environment (e.g. into drink-'
ing water or biota such as fish) prior to exposure. Thus a complete
materials balance can provide insights into potential exposure as weU
as tne data necessary for estimation of environmental distribution.
4-2 GOALS OF A MATERIALS BALANCE
The overall goal of the materials balance portion of these exposure
analyses is to obtain a complete and quantitative description of the uses
and sources of a pollutant and a characterization of the form and mode of
should: P°lluta^ into the environment. A complete materials balance
(1) Describe the types of uses and use situations, especially fo-
consumers. J
(2) Identify all existing and potentially significant sources of
the pollutant.
(3) Identify the chemical and physical form of the pollutant PS ir
enters the environment.
(4) Characterize (qualitatively and, where possible, quantirat^el")
tne entry ot the pollutant into the environment (loadings) in ''
terms of: amounts, seasonally, geographic locations, rates
receiving environments.
(5) Identify uses and environmental releases that can lead to dire-
exposure or receptors. ^-^<^-<.
(6) Account for all material produced by achieving a balance between
the amount proauced naturally, inadvertently and by industry
anc the amount transformed, contained (unavailable for release)
stockpiled, and released to the environment. release),
4-3
-------
(7) Establish the confidence and/or uncertainties in the amounts
of pollutant releases by various sources to the environmental
compartments.
Ideally, a materials balance effort would address all potential, as
well as existing, pollutant sources. This may not be practicable because
of both data and resource limitations. Given these limitations, it is
tempting to focus first on the identification of major existing uses,
though sources deemed insignificant on a national scale may be very signi-
ficant in selected areas. Therefore, care must be taken in limiting the
scope of the analysis.
A systematic approach to source identification can aid this process;
many possible sources must be considered in order to determine which are
the most important by virtue of their national or local significance or
the opportunity for direct receptor exposure. The physical and chemical
form of the pollutant as it enters the environment are important because
these characteristics affect the significance of various environmental
pathways and the resultant distribution.
The spatial (geographic, source intensity) and temporal (rate and
frequency of release) characteristics of the environmental loading must
be considered. Total pollutant quantities involved and the character-
istics of the receiving medium are also important. This information will
ultimately be used to determined the environmental distribution. Depending
upon the scope of the risk analysis (e.g., local, regional or national).
quantification of releases may not be necessary. Both documented data
and engineering estimates may form the basis for quantification, when it
is desirable or possible.
Understanding and delineating the uncertainties in source and load-
ing estimates, i.e., determining confidence limits, increases their useful-
ness in the subsequent steps in environmental risk analysis, and any
regulatory control recommendations ultimately derived. Similarly, achiev-
ing closure of the materials balance—i.e., equating all of the production
or input of the pollutant with use, accumulation, destruction, or release
of the pollutant to the environment—makes the subsequent risk
analysis more comprehensive, substantive, and credible. The level to
which these goals and objectives; may be achieved will depend upon the a-
availability of data, the nature of the pollutant, and the effort that can
be devoted to this portion of the risk analvsis.
4.3 MATERIALS BALANCE METHODS
There are two major steps in performing a materials balance—the first
is a chorough identification of sources and the second is quantification
of loading/emission rates to the specific receiving compartments of the
environment. Environmental pathway analysis (see Chapter 5.0) can then
be used to establish transfers and reactions of the pollutant within ana
-------
among environmental compartments, which, in turn, determine environ-
mental concentrations and influence exposure. Therefore, the data
developed from the materials balance must be compatible with the require-
ments of environmental pathway and exposure methodologies. The major
challenges in developing a materials balance are the identification
of sources, assembly of data, and quantification of loadings for
pollutants that are not reported in the literature or are unknown or un-
quantified because of the lack of control technology.
The general approach for a materials balance is shown in the flow
chart in Figure 4-2. Key validation issues are: data completeness
and uncertainty, materials balance closure, and compatibility with the
needs of subsequent components of risk assessment. At many points in
the analysis, the need for better and/or additional data may arise.
Judgments regarding the value of higher quality information must be
made in order to determine whether engineering estimates or continued
literature review for direct measurements will be required.
The first step is to establish the goals and scope of the materials
balance effort, including the desired outputs of the work and criteria
for determining when this portion of the risk analysis has been completed
in sufficient detail. Next, the analysis should be focused on a quali-
tative description of the flow of the pollutant within the cultural en-
vironment and potential releases to the natural environment. This st-p
should highlight the unique character of the pollutant and indicate
areas requiring extensive data gathering and analysis. The description
should cover the complete range of industrial processes that involve
the pollutant: extraction, processing, storage, uses of the pollutant
or product containing the pollutant, and all potential disposal modes.
The greater the number of processes, the greater and more varied are
the potential opportunities for release to the environment and the more
complicated and potentially incomplete may be the analysis, due to
insufficient data. Knowledge of major uses of a product, product life-
times, and disposal processes may become important in establishing
total environmental releases; these data are likely to be difficult
to obtain for the entire range of possible products and uses.
A materials balance matrix, such as is shown in Table 4-1 for
phthalate esters (Perwak et al. 198la), provides a convenient method for
initially organizing data on pollutant sources and loadings. Ideally
such a matrix provides a logical framework for a thorough accounting'of
sources. The matrix consists of a source axis for identification of
points of release, and an environmental input axis for estimating and
organizing loading factors or rates for each source. The matrix includes
the processes relevant to establishing source identification,
e.g., extraction, refining, manufacturing, processing, transporta«-ion
storage, use, or disposal (see Table 4-2) and suggests materials
classes and types of releases to be considered. It also pro-
vides a framework for allocating and aggregating pollutant'releases to
environmental compartments of air, land, water ana biota and listine
data on quantities, rates and forms of release.
4-3
-------
Establish Materials
Balance Goals and Scope
Develop Qualitative Description of
Pollutant Flow-Outline Materials
Balance Checklist Framework
Literature
Research
and Data
Identify Pollutant I
Source and Forms
of Pollutant
Characterize Release of Pollutant:
to Environmental Compartments
(Spatial and Temporal)
Engineering
Estimates
Sunmarize Marerials Balance Data
in Format Compatible with Environ-
mental Pathways and Fate Analysis
Review Materials Balance
Completeness, Uncertainty
and Closure
Determine Critical
Environmental Loadings
for Pathway Analysis
Identify Major Direct
Exposure Routes
FIGURE 4-2. MATERIALS BALANCE METHODOLOGY FLOW CHART
-------
TAliLK 4-1. t'.XAMPI.K OF SOIIKCIi AND ENVIRONMENTAL LOADING MATRIX: I'HTllAI.ATIi JiSTKKS
1 (••!'>£
U S
S«H-l>
i^.i
iOJ 4
.
DNOP
2J
(At
. .
B.
B
I
(81
o
rvt i>u»[K tfti o 4
f !!• In cu.u jet
tow *i.rfi.v
cunt 4, i
uii.fi |n>ivM.eia . n
FVl f>ljai letter . 8i
tUhrr t-.)ly*c[b) .50
N»ll-pt44l tC til f 0i
...j-..,..
"
a
•d
C
C >.
k- «
(D
l o .
:. win* tug (100 1 U GI4
II «.»
Ujll Lt>V*
uiU-e I'Vt us*:, 1(111
1 U
1 U
1 n
1.0
1 0
1 o
!,„
i
to'*
„.
(u
c —
4 o
tt 4
1114 .»
.'1 1
a t
4 ;
t •>
it i
.' i
i .'
i .'
1 H
* V«porlz*(^ During »
" Compounding i? 3.CS r
0 I
1 1
1 1
" 1
t' 1
'"' '
'1 ?
•t 1
_.J
e
i
t
1 ;;
o i)/
U til
ii n;
n in
•i <> i
n u t
>i u;
n u •
— J
14-
X
<* r-
* a>
(i)
0 (.
,<.
n i.
" t
" I
1. J
n t
" J
.1
c
-t —
c
c
E ~,
tt t
"
*> ,
C l '1
; L
4lt /
il 4 >
t 1
1 'I
Ii *
it /
'.'l
- 1
at
«t
V
(K)
It L>U| (L» Jtl
,,m,2^,,.
il Illll i,, all
0 .* u lier
II (111 .1C
II III!.' .ill
^::;.(
c.
(1 1
0 J
0 1
O 1
11 1
11 1
II 1
0 mil
i) 1*114
It IlltJ
0 _'
U ',
~
- (^ears4
JO
7
10
I
It)
7
1
^
-
e
3
i
a i
1 V
tt t,
U 6
41 ft
It
t)
) <« t
J O
1 11
U „'
0 2
tl i
11 *»
*. kcfiiaining for Dtitosal
S 9*i tc Unc. 3^ burned
U 1'
VI 0
44,1.
/ «.
I't H
4 ^
U 1
tt 1
1 tl
II '1
U 1
0. t
/ fc
-
(J
3«
U 1
11 I
J 0
\_ I
4.4
0 1
1 )
o u->
TT^
U 1
It (.4
-
U 1
0 tlj1
"0~'fc
0 ^
JLI
n. t
u •»
a in
u si
Antttt.i
-
s
i:
a 2
U.2
a «
,„
l i.
4 t
O.I
U 2
U II
a t
T~T
O.I
a i
a (n
l> O<
0 (I
»•«
u 2
Jl.i.
0.1
O.'i
a i
4.i
. ._
/ It
I't 0
4 4
^n
a i
) i
0 1
i i
0 1
TT~
O.I
|>.I
0 2
2 1
Tl
b
*
«
| =
B •-••
U I
0 i
•«
1 6
TT
il. 2
1 2
0 1
IS
II 1
a (x»s
a l
11 (ii
JTK
.
li-i
0 2
U HI
1) 1
TTl
-
IWU
Hbl*
our
utr
ftKOI
^^^^
-------
TABLE 4-2. SOURCE IDENTIFICATION FOR MATERIALS BALANCE MATRIX
Extraction Method
drilling, dredging,
strip or pit mining
wastes (slags, to air,
water or other)
Refining Operations
washing, grinding,
extraction, distillation,
physical separation wastes
Manufacturing Process
chemical reaction, cracking,
digesting, forming wastes
(pollutants in form of primary
product, by product or co-
product, or containment)
Processing
extrusion, molding,
calendaring, drying,
pressing, cutting wastes
Transportation
loading/unloading, cleaning,
in transport (volatilization,
leaks) by truck, by rail, by
plane, contained in tank cars,
drums, disposal sacks
Consumptive Use
industrial, commerical,
household, institutional, or
agricultural uses in which
pollutant is confined/contained,
applied (e.g., swimming pool.
agriculture) or consumed in products
Naturally or Inadvertently Occurring
In minerals and soils, in aquatic
systems, in air, in biota,
volcanic activity, formation in
upper atmosphere, natural
combustion; inadvertent release
from urban runoff, or use of
pollutant-bearing products (e.g.,
fossil fuels, cement or other)
Disposal Methods
POTW, septic systems, solid
waste landfill, contained land-
fill, incineration, deep well
injection, discharge (treated or
untreated) to surface waters
including lakes, streams or
ocean, or deposition in "sealed"
drums
4-3
-------
The source and environmental input categories may be further sub-
divided depending upon the pollutant and the scope of the materials
balance. For example, the environmental input may be subdivided bv geo-
graphy (e.g., urban versus non-urban), by depth of environmental medium
(e.g., surface water versus groundwater), or by waterbody type (natural:
streams, lakes, estuarine or coastal waters; versus manmade: effluents,
reservoirs, or sewers). Deposition on soil may occur as the result of
aqueous discharges through leaching, adsorption, or sediment transport.
The relative contributions of these processes may need to be considered
and the matrix may have to be expanded accordingly.
Expansion of the matrix to include receptors as they interact
with receiving media would aid in classifying the relative importance
of receiving media. For example, the receptors may be people, fish,
game, livestock, crops or materials, most of which will have specific
interaction zones with various classes of emissions (e.g., people
exposed in the workplace or through product use at home). Consideration
of these interactions would be particularly useful in identifying
direct exposures to a pollutant.
After^possible sources, uses, and releases have been arrayed, the
next step is to develop and summarize data concerning sources" and load-
ings. The data abstracted from the open scientific literature and from
government publications or contractor reports may be supplemented by data
rrom industrial product literature, trade journals, or popular periodicals
and reports. A list of commonly available and reliable data sources is
presented in Chapter 10.
In general, one can expect considerable data gaps and, therefore
a high degree of uncertainty in the materials balance for ubiquitous
ana naturally occurring priority pollutants, for pollutants entering a
variety of media from numerous sources, and for new chemical pollutants
or newly recognized toxicants. Quantitative data may be lackin* for
various common source categories and estimation will have to be relied
upon to till the data gaps. Estimations made from chemical or industrial
engineering data will often be based on product levels, emission factors
anticipated spill frequencies, etc. Different sources of data and -s^'-'
mation procedures will probably be required for each process listed"in"
the materials balance checklist. Assumptions made in'the estimate should
be clearly stated so that uncertainties can be identified and checked at
a later time, if appropriate.
4-9
-------
The structure and approach to environmental fate analysis (for
example, whether or not computer models are used) will also help to
define the materials balance matrix. The appropriate scales for time
and geographical regions will become evident by the nature of available
data or as the available data are reviewed. Wherever possible, the
materials balance data should include identification of the physical/
chemical state (i.e., phases, chemical complexes, oxidation state.
etc.) of the substance in products and releases. In many risk analyses,
considerable interaction between the environmental pathways and the
materials balance components will be required with each continually
refined as a result of considerations arising from the other. Data describing
the geographical distribution of sources and loading race characteristics
are generated according to the information needs of air disper-
sion, stream flow, groundwater, or lake models used in pathway analysis.
After data have been summarized, it .is important to review
the results for completeness, close of balance, anc. uncertainty. Assess-
ment of completeness is subjective, since additional literature research
or investigation may lead to new data or estimates. Judgments will have
to be made as to whether more effort should be expended in developing
additional data on other sources or loadings. Examination of the degree
of closure of the materials balance may be useful in making these judg-
ments. This will involve comparing all of the pollutant generation steps
or inputs with pollutant releases or outputs over a. selected time frame.
If inputs agree well with outputs, greater certainty in the completeness
of the materials balance has generally been achieved.
There is some uncertainty associated with most, if not all, calcula-
tions and value determinations involved in developing a materials balance,
largely due to varying limitations on monitoring data, on assumptions
associated with approximations of pollutant release from different pro-
cesses or activities, and on simulation models. Sufficient monitoring
data, which is current and of high quality, will reduce the magnitude of
uncertainty associated with a materials balance. However, there are
many components involved in a materials balance and typically good
monitoring data are available only for a few, if any.
Three approaches are available for evaluating the uncertainty in
materials balance calculations in order to establish some degree of
confidence in the results. The first is a parametric approach based on
a mathematical model linking the input and output variables. In this
approach, ranges of input values (e.g., 100-200 kkg/year) are assigned
(through educated judgment or 'sensitivity analysis) and partial deriva-
tives are the principal tool in the parametric analysis. The second
approach is statistical in that parameters are statistically estimated
from the available data. Uncertainty associated with a parameter in
this approach is expressed in terms of a statistical confidence interval,
that is, the range within which the true value of the parameter is expected
to lie (e.g., 100 + 50 kks) with an assigned degree of confidence. As
there is uncertainty associated with each parameter or dependent
4-10
-------
parameter, the uncertainties can be combined by the method of error
propagation to obtain the confidence of the output variable.
These two approaches are described in more detail by Serth et al. (1978)
and are currently being studied by the Exposure Assessment GroujT of the
Office of Health and Environmental Assessment, U.S. EPA.
A third approach is a qualitative one in which uncertainty is
assigned to materials balance calculations on the basis of educated
judgment. The resulting uncertainty can be defined: in terms of expected
ranges; as the most likely of several values calculated by different
methods; or as a qualifying statement associated with one approximation
e.g., under x conditions, it is very likely that pollutant release from'
this source will be y.
The outputs of the materials balance are several: the source and
environmental input matrices, containing data on sources and loadings-
identification of large or critical environmental loadings and related
characteristics, which can serve as the basis for environmental pathway
scenarios/analyses; and identification of major routes for direct expo-
sure for use in human and nonhuman exposure analyses.
4.4 EXAMPLES OF MATERIALS BALANCE OUTPUT
4.4.1 Introduction
The examples in this section are presented in order to illustrate
the extent to which a materials balance for a particular pollutant can
be .ocused to characterize the predominant sources of the environmental
ourden. As previously mentioned, materials balances may quantify releases
of a specie cnemical that result from its commercial or inadvertent
production, its transportation or use, transportation or use of a
product in which it is a component, or natural sources. The three
materials balances discussed below - those for chloroform, copper
ana pentachlorophenol - are all different in their focus.
Examples drawn from the materials balance section of the chloroform
exposure and risk assessment (Perwak e£ al. 1980a) illustrate the
approach used to develop meaningful environmental release data for
inadvertent sources since the burden of chloroform in the aquatic
environment originates primarily from its production during chlorination
ratner than from commercial releases. The materials balance
a^°nS fr°m thS C°Pper exP°sure and risk assessments (Perwak et al .
address quantification of the environmental burden of an abundlnT
natural material. The chird example detailed in this section is ^aken
from the pentachlorophenol exposure and risk assessment (Scow et al
where the estimation of environmental releases associated w- th~7he~
ultimate use of materials containing pentachlorophenol was a major
challenge. J
4-11
-------
4.4 EXAMPLES OF MATERIALS BALANCE OUTPUT
•4.4.2 Chloroform
Two major reference documents provided the initial conceptual frame-
wor* and much of the data necessary to developing a materials balance for
chlorotorm (Perwak et al. 1980a); a large study the National Academv of
Sciences completed in 1978 and a study by a contractor for the U S '
Environmental Protection Agency, released in draft form in 1980 The
study by the National Academy of Sciences was completed soon aft-r the
health hazards of chloroform were initially recognized and hence did not
have the advantage of being able to draw upon the substantial bodv of
research tnat has been done subsequently. Therefore, some of the'conclu-
sions drawn by NAS researchers had to be Devaluated in light of the more
recent work. The very recent U.S. EPA contractor report treated the "
commercial processes that generate chloroform in great detail but did
not contain much information useful for purposes of developing a mat—ials
balance focused on the releases to the environment.
Unlike the situation with chemical substances that are exclusive^'
man-made, the literature concerning chloroform has noted several signifi-
cant sources of chloroform releases that originate outside of commercial
production or consumption of the chemical. Since 80-95% of commerciallv
produced chloroform is consumed by chemical reaction as feedstock to- '
cnlorodifluoromethanes and is not available for release to the env
-------
Detailed knowledge of industrial processes and the chemical industry
structure were required in order to generate estimates of source strength.
For example, calculations based on industry interviews (combined with
values reported in various symposia) resulted in an estimate of 12,500 HT
per year of chloroform produced by the pulp bleaching industry. The values
previously reported for this release ranged from 1 MT per year in -he U S
to 300,000 MT per year worldwide. Actually, part of the total release
occurs during the bleaching process inside the pulp plant and part
occurs during treatment of the plant effluent. Some data were available
concerning the chloroform release during effluent treatment stages, but
very few data were available for releases during other stages of the
bleaching process and, therefore, the best available approximation had to
be pieced together from interviews with industry experts.
The ultimate measure of success of a materials balance study must be
the degree of closure obtained between the sources and use/r«leases of
the chemical. The results obtained for chloroform are displayed in Table
4-3 and Figure 4-3. As indicated, the "unaccounted for" amount is equal
to more than D0% of the amount known to be released to the environment.
However, there are also uncertainties regarding the amount of chloroform
commercially produced and the amount devoted to the major consumotive use
(F-22 reedstock). Therefore, after laboratory use and stockpiles are
taken into account, nearly all of the "unaccounted for" amount could con-
ceivably result from the uncertainty in the production volumes.
4.4.3 Copper
Copper is one of the more abundant metals among the 129 priority
pollutants and as such has many sources of significant environmental' re-
leases . Data are generally available concerning releases from many of
these sources for incorporation into the materials balance fo- copoer
(Perwak, et al . 1980b) . A materials balance for copper is shown in
Tables 4-4 and 4-5 and Figure 4-4.
Copper is mined and milled in seven states, and effluent dischanzes
and solid waste disposal practices have been monitored in order to determine
the compliance with current environmental regulations. With these da-a
estimated f°r the known produc-'
e
-------
TABLE 4-3. EXAMPLE OF MATERIALS BALANCE OUTPUT INVOLVING
INADVERTENT RELEASES—ESTIMATED PRODUCTION AND
USE/RELEASES OF CHLOROFORM, 1978
Production (kkg)
Commercial Production
Methyl Chloride Process
Methane Process
Loss during Production
Imports
Production as Contaminant
Vinyl Chloride Monimer
CH3C1, CH2C12, and CC14
Chiorination of Water
Cooling Water
Potable Water
POTW*
Swimming Pools
Bleaching of Paper Pulp
Automobile Exhaust
Photodecomposition of Trichloroethylene
Marine Algae
122,500
36,000
500
159,000
,679
2,^60
912
91
3
7,670
2,733
3,466
12,500
965
^50
(unknown)
186,78^
-------
TABLE 4-3. EXAMPLE OF MATERIALS BALANCE OUTPUT INVOLVING
INADVERTENT RELEASES—ESTIMATED PRODUCT AND
USE RELEASES OF CHLOROFORM, 1978 (Continued)
Uses/Releases (kkg)
Feedstock for F-22 Production 142,700
Exports 7j900
Destroyed/Retained in Products/Storage 3,968
VCM Products 2,290
Pharmaceutical Production 1,610
F-ll/F-12 Production (and others) 47
CHC13 Production u
Pesticide Production u
Unaccounted for (including laboratory
use and stockpiles) \\ ^QQ
Air Water Land
Released to Environment 19.207 912 496 20,615
Pulp & Paper Bleaching 12,100 400~ ^~
Chlorination of Water 3,245 221
Pharmaceutical Extractions 1,525 275 290
Automobile Exhaust 965
CHC13 Production 370 14 6
Trichloroethylene
Decomposition 450 — —
VCM Production 187 2 200
Transportation & Storage Loss 177
F-22 Production 150
Pesticides 38
186,783
"'Publicly Owned Treatment Works
source: Perwak, J. et al. An exposure and risk assessment for trihalo-
mechanes. Final Draft Report. Contract EPA 68-01-3857.
Washington, DC: Monitoring and Data Support Division, Office
ot Water Regulations and Standards, U.S. Environmental Protection
Agency; 1980.
4-15
-------
Pioduclion
17.381 kke
CIlltNllldllllll
Ul Wdlll - 2(1%
Tiichliiioclliyleiu.- - 3%
Aulu Exhaust b'X.
Produced ti
ConUniiiuiil
2733 kkg
H«la«ed to
Eniliumncnl > 20,615 kkg
FTfillRK A-3 KXAMPLK OF GRAPHIC PRESliNTATlON OF MATERIALS MLANCK
OUTPUT—ENVIKONMKNTAL LOADTNC OF CHLOROFORM, 1978
k, eL aJ . An exposure and risk assessment, for tr ilia lomcthanea. Final Draft Report.
Contract KPA 68-01-3837. Washington, DC: Monitoringand Data Support Division, Off Lee of
Water Regulations and Standards, U.S. Environmental Protection Agency; 1980.
-------
TABLE 4-4. EXAMPLE OF COMMERCIAL PRODUCTION AND USE DATA--
SUMMARY OF U.S. COPPER SUPPLY AND DEMAND, 1976
Source/Consumer
Domestic mine production and
beneficiation
Refined Scrap
Unrefined Scrap
Imports (refined)
Imports (ores-concentrates)
Industry Stocks, 1 January 1976
Copper Wire iMills
Brass Production
Other
Industry Stocks, 31 December 1976
Total
Supply
(MT)
1,287,940
204,080
149,660-
235,810
99,770
419,940
Consumption
(MT)
2,397,200
1,349,288
567,092
39,110
441.710
2,397,200
Note: The above figures are for one year (1976). There is considerable
statistical variation from year to year; consequently, these do not
not reflect average values.
Source: Pervak, J. et al. An exposure and risk assessment for copper
Final Draft Report. Contract EPA 68-01-3357. Washington DC•
Monitoring and Data Support Division, Offica of Water~Regulations
and Standards, U.S. Environmental Protection Agencv; 1980
4-17
-------
TABLE 4-5,
Source
EXAMPLE OF MATERIALS BALANCE OUTPUT INVOLVING NATURAL SOURCES-
U.S. ESTIMATED ENVIRONMENTAL RELEASES OF COPPER, 1976
Release (MT/yr)
Primary Production
Smelting
Secondary Production
Metallic Ore Mining
& Related Activities
Copper Wire Mills
Brass Production
Iron 5, Steel Production
Coal Mining**
Pulp, Paper & Paperboard
Inorganic Chemicals
Steam Electric Industry
Machinery Mfgr.
Electroplating
Miscellaneous Sources
Area Sources:
Abandoned Metal Mines
Agricultural Applications
Urban Runoff
Suspended Sediment
Incineration/Refuse
POTW
Total
Air
^
2002
A
i
16^' ,2
31,2
171,2
u "
-
-
-
A
—
—
*
—
-
Direct
Aquatic POP,;
13. 4:
Unknown
0.33 73
343
134! 1.4841
15L1 2941
656
1811'2-
no3
43
1743
1511
4003 1,4003
723
9
314.
3,600= *
44 1 34
18,400
Land
1.078.2902
Unknown
&
Unknown
A
421'2
8961'2
_^
«.
_
_
^
9203
_
19.195'*'5'
*
_
1002
1.9002
9,6 8 O3
3,269"
1,110,923
Insignificant
*These emissions are directly applied to the category in which they are
reported; however, often during or shortly following release, they enter
other environmental media.
**Coal combustion is known to release some copper; insufficient data is
available to substantiate this quantity.
*Ihe total estimated POTW influent is 11,800 MT/yr (see Table 6). Thus,
only a portion of the sources have been identified.
^Versar, 1978.
Arthur D. Little Estimate.
Effluent Guidelines Monitoring Data, analyzed by Versar, EPA, 1979.
*U.S. Department of Agriculture, 1974.
5SRI, 1979.
6E?A, 1974.
7EPA, 1977.
3Table 6.
^Martin and Mills (1976).
Source: Perwak, et_ al. An exposure and risk assessment for copper.
Final Draft Report. Contract EPA 65-01-2857. Washington,~DC :
Monitoring and Data Support Division. Office of Water Regulations
and Standards,. U.S. Environmenral Protection Agencv; 1980
4-13
-------
ENVIRONMENTAL
COMPARTMENTS
INDUSTRY
Electroplating
(27.200 MT)
Brass Prod.
(567,092 MT)
CONSUMPTIVE
RELEASES
Primary Prndur"on
(1,281940 MT'
Copper Wire Mills
(1,349,288 MT)
Erosion-
Suspended
Sediment
18.400 MT
Urban Rur.ort
525 MT
rtefuse
2.000 MT
Land
1.110.923 MT
Air
484 MT
Water
26.909 MT
POTW
3.269 MT
Copper Use
Includes smelting
Industrial releases in which cooper exists as a trace element. Sources include iron and steel production,
coal mining, pulp and paperboard manufacture, steam electricity generation, other ore mining, and
abandoned mines.
^POTW effluent includes contributions from human and otner unknown sources.
Note: Boundaries between receiving meaium are often undefined and/or changing: Copper apparently
released to one compartment can result in another.
FIGURE 4-4 EXAMPLE OF GRAPHIC REPRESENTATION OF MATERIALS BALANCE
OUTPUT—ENVIRONMENTAL LOADING OF COPPER, 1976
Source: Perwak jjt_ a_l. An exposure and risk assessment for conoer. Final
Drart Report. Contract EPA c8-OI-3857. Washington, DC": Monitoring
and Data Support Div.. Office of Water Regulations and Standards,
>-' . o . cPA i 1980
4-19
-------
As with many of the metals, copper occurs in the natural environ-
ment in combination with other elements. The release associated with
the mining and milling of these related ores can be assessed from in-
formation on the nature and scale of production, the level of sophistica-
tion of recovery and waste treatment technology, the frequency with which
it is applied, and the availability of documentation and monitoring data
on all of the above. Copper is a significant by-product/co-product of
lead-zinc deposits, occurs in coals, and is released as a consequence of
iron and steel production. For these major associated production pro-
cesses, EPA documentation and other published research data were avail-
able to quantify the resulting copper release.
Copper is consumed in a variety of uses ranging from brass produc-
tion to electroplating to agricultural applications (as an algicide).
The Bureau of Mines annually publishes reports concerning the distribu-
tion of copper and other mined minerals in the U.S. economy. The U.S.
Bureau of the Census also publishes import-export data for all inorganic
chemicals. These sources provide a baseline for estimating releases
associated with the various stages of production and use of copper
(Table 4-4).
Frequently, information concerning treatment efficiencies or general
disposal practices can provide :he basis for estimating environmental re-
leases. For example in the case of brass production, effluent guidelines
data provided by the U.S. EPA were used to estimate aquatic discharges
of copper from this source; the computed total fell within the realm of
reasonable losses as a percentage of copper consumption in this applica-
tion. In the case of electroplating, however, estimated releases based
on the available Effluent Guidelines Data (or continuing data source of
the U.S. EPA) exceeded the amount of copper consumed by that industrv.
This overestimation can be attributed to some of the underlying
assumptions. Electroplacers do not always operate on a regularly scheduled
basis, nor is their volume of production consistent. Many are "captive'1
operations contained within a larger industry and produce only to meet
the needs of that parent industry. Independent electroplaters also use
copper somewhat intermittently since some materials are plated with
nickel, silver, zinc, or some combination. Therefore, a release estimate
for the materials balance based on "averaged" values from limited sampling
of electroplating effluents would be incorrect. Bureau of Mines staff
and industrialists familiar with the electroplating industry were consulted
in order to approximate the was^te recovery efficiency and the maximum
possible percentage of material loss.
A significant source of copper to the environment is through POTWs
which receive influent from urban runoff, industrial discharge, and domestic
and commercial units. A contractor study performed for the U.S. EPA
4-20
-------
provided removal efficiencies of metals for a "representative" sampling
of POTW influents and effluents for primary, secondary and advanced
treatment facilities. This information was combined with data concern-
ing total treated effluent from POTWs in the United States and outlying
territories and an assumed distribution of treatment levels (2% of the
flow from primary treatment plants, 64% from secondary and nearly 35%
from advanced).
A problem remained, however, in identifying the sources of copper
in POTWs. Average residential loading of 42 mg Cu/day/person was assumed
on the basis of samplings from sewer systems in St. Louis and Cincinnati.
Residential areas provided roughly one-third of the copper to POTWs.
Adequate data on commercial and industrial contributions were not avail-
able to permit a determination of the source of the remaining 66% of the
copper in POTW influent.
Whenever environmental releases are arrayed and totalled, care has
to be taken to avoid double counting. The best example from the materials
balance for copper is the case of urban runoff. The volume of copper
carried by urban runoff was determined from stream samples of urban runoff
to separate storm sewers, point sources, and unsewered areas. Possible
sources of copper released to the urban environment that would be re-
flected in runoff include exposed construction elements, transportation
vehicles, industrial applications (plumbing, tubing, valves, etc.) and
settled particulates from atmospheric releases (e.g., coal burning power
plants). This latter source has already been accounted for as an air
release. Clearly, though some of the atmospheric copper releases are
also included in the concentrations in runoff, separating these from the
other components of urban runoff is very difficult. Urban runoff itself
may flow to a POTW, so that there is opportunity for triple counting
the original releases to air. Often, as was the case in'the copper"mate-
rials balance, insufficient data are available to permit differentiating
the relative contributions of individual sources of a pollutant to urban
runoff and POTWs. Although instances of double counting cannot always be
avoided, they do result in uncertainties in the analysis and need to'be
identified.
4.4.4 Pentachlorophenol
Pentachlorophenol (PGP) is commonly used throughout the United States
as a wood preservative and its characteristics and applications are fairly
well known to this industry. As a result, much of the information in-
cluded in a materials balance for this pollutant came from specialists
in the timber and wood products industries, as well as from U.S. EPA
contractor reports and water quality programs. In comparison with mate-
rials balance analyses of others of the 129 prioritv pollutants, the PCP
materials balance (Scow et al. 1980) was quite simple and straightforward
due to the nature of both the manufacture and use of this compound.
At the present, PC? is manufactured at three locations in the U.c
as shown in Figure 4-5, by three different chemical companies". Each
I-?-!
-------
I
ro
I ->
v ^ K#2
:•• **i »T t *\ •
. ,.:i:. •> >:t..;4:-
«».n.,«.iN,MAl .; r v(Jv..L?3W..-
Notes: 1. Baitelte for EPA. 1975
2. Foieit Service, U.S. Department ot Agricultuie
FTCURIi 4-5 EXAMPLE OF GliOORAPHlC DISTRIBUTION OF PRODUCTION SOURCES—LOCATIONS OF
PENTCULOROPHENOL MANUFACTURING AND WOOD TREATMENT PLANTS
Source:
Scow el al. An exposure and risk ussessmen.1: for penLnclilorophenol . FJnal Draft Report.
Contract EPA 68-01-3857. Washington, DC: Monitoring and Data Support Division. Office
or Water Regulations and Standards, U.S. Environmental Protection Agency; 1980.
-------
facility produces POP by the same process during which phenol is
chlorinated in the presence of a catalyst. During production, releases
occur to the air and to water. Air emissions are limited by a scrubber
mechanism enabling PCP recovery; incentives for control of atmospheric
release are economic, as well as regulatory. At the time of the penta-
chlorophenol study (1980), industry production of pentachlorophenol
was not subjected to regulatory control by the U.S. EPA; sampling data
on aquatic discharges were not available. In addition, there were no
reliable data on the efficiency of the production process with respect
to aquatic discharge upon which gross annual discharge estimates could
be based. Most of the PCP that is domestically produced is consumed in
the U.S. and there are no identified imports.
The wood preserving segment of the timber industry consumes more
PCP than all other users combined. Though consumption varies from year
to year, wood preserving consumes roughly the same share of total produc-
tion each year: an estimated 78% (U.S. EPA timber industry) to as'high as
85% to 93% (chemical industry). When applied to wood, pentachlorophenol
enhances toughness, prevents discoloration, and prevents attack by wood-
destroying fungi and insects. The timber industry is a relatively mature
U.S. industry and as such is well established and fairly well known. For
this reason and because of the industry's dominating consumption of PC?,
78% to 93% of manufactured PCP can be fairly well traced through its life-
time of wood-associated uses.
The timber industry reports that 415 wood preserving plants operated
by 300 companies potentially use PCP. These plants are geographically
located on Figure 4-5, in a pattern consistent with the major timber re-
sources of the nation. Associated consumption of PCP by wood preserving
planes is shown in Figure 4-6 as it is distributed to six regions of the
U.S. At this stage, wood is impregnated with PCP and there are small
releases to POTWs and some to land. Most aquatic discharges from wood pre-
serving plants occurs during a wood conditioning process prior to applica-
tion of PCP.
Following PCP application, waste streams at 90% of the 415 wood
preserving plants are evaporated so that there is no aquatic discharge at all
ut tne remaining 10/4, most use a stream process cc apply PCP treatment and
treat their wastewaters with roughly 81% efficiency resulting in a dis-
charge of 5.1 MT of PCP to POTWs (Scow ei ai. 1980). The other plants use
tne Soulton process to treat wood and treat their wastewaters with a 44%
efficiency (Scow et_ al_. 1980) consequently discharging 0.2 MT of PCP to
POTWs. Only one small wood treatment plant discharges its waste stream
directly to surface waters. This facility operates on an intermittent
basis (less than 25 days per year) discharging less than 26.5 kg annuallv.
Most of the PCP released by wood preserving plants is contained in sludga,
and these releases were estimated to be 74.5 MT per year on the basis of
sludge practices and removal efficiencies (Scow et_ al. 1980).
4-23
-------
Nortneait
l Central
Southeast
Soutri Centra!
Mountain
Pacific
720
1080
4320
5940
2340
3600
J_
I
0 1000 2000 300C
Total: 18,000 Metric Tons
4000 5000
Metric Ton!
5000
FIGURE 4-6.
EXAMPLE OF REGIONAL DISTRIBUTION OF USE SOURCES
REGIONAL ESTIMATED CONSUMPTION OF PENTACHLORO-
PHENOL 3Y WOOD PRESERVATION PLANTS
Source:
Scow et al. An exposure and risk assessment for pentachlorophenc]
Final Draft Report. Contract EPA 68-01-3357. Washington, DC:
Monitoring and Data Support Division, Office of Water Regulations
and Standards, U.S. Environmental Protection Agency; 1980.
-------
Following the application of PCP to wood at wood preserving plants,
two important factors were identified and quantified: first, the end
use of materials treated with PCP as illustrated in Figure 4-7; and
second, the regional consumption of PCP-treated wood products as illus-
trated in Figure 4-8. Although it was not possible to combine these two
factors in a verifiable manner (e.g., identify the number of fence posts
in the North Central states), the end uses of an estimated 84.5% of
pentachlorophenol produced in 1978 were identified. From this end use,
it was estimated that 344 HT (or 1.9% of PCP used in preserving wood)
are volatilized to the atmosphere based on the known properties of treated
wood and PCP. It is also possible that PCP runs off the poles, fence
posts and railroad ties when exposed to rainfall and contributes at non-
point releases to groundwater, storm runoff basins, POTWs and surface
streams. There were insufficient data available, however, to quantify
this release.
Consumption of the remaining 15.5% of manufactured PCP is known, but
associated releases are generally not as well understood. Production
of sodium pentachlorophenol (NaPCP) is the second largest consumer, us-
ing 11.7% of annual PCP productipn. This is a relatively small segment
of the chemical industry whose waste streams are not yet subject to
Federal regulation. Consequently, insufficient data were available to
estimate releases resulting from production of sodium pentachlorophenol.
NaPCP is used to prevent bacteria growth in water towers, and in the
textile and tanning industries with small associated environmental re-
leases. It is also an additive to outdoor paints and is believed to be
used in some toy paints manufactured and applied outside of the U.S.
A major concern was identified with respect to NaPCP in paints in this
materials balance, namely that misuse of outdoor paints, (i.e., indoors) and
imported painted toys present a significant potential to human exposure.
However, it was not possible to estimate the magnitude of these exposures.
4.5 SELECTED EXAMPLES FROM MATERIALS BALANCES FOR OTHER POLLUTANTS
Several components of various materials balance analyses are pre-
sented here as samples of methods typically utilized to estimate re-
leases and as examples of special release situations. These selections
are discussed in brief and, where appropriate, are accompanied by figures
or tables.
4.5.1 Releases During Transportation
In the materials balance flow diagram (Figure 4-2) reference is
made to the transport processes between major points in the pollutant's
life cycle (extraction or synthesis, manufacture, storage, use and
disposal) and subsequent pollutant releases associated with transporta-
tion. Relatively few standardized data are collected or maintained
on potential environmental releases during transport largely due to wide
variations in transport methods and handling practices bv che carriers,
themselves. In fact, documentation, when available, is usually limited
-------
UumO«f
Fence ?osis
Otnet
0 1000 2000 3000
Totai: 18,000 Metre Tons
40CO
5000
tf 'C Tons
5000
7000 3000
9000
10.000
FIGURE 4-7 EXAMPLE DF END USE DATA—MATERIALS
TREATED WITH PENTACHLOROPHEXOL, 1978
Source:
Scow, ejt al. An exposure and risk assessment for pentachlorc-
pnenol. nnal Draft Report, Contract EPA 68-01-3857. Vash^
DC: Monitoring and Data Support Division, Office of Wate- =
S and Standards, U.S. Environmental Protection Agencv
1-26
-------
North
Central
5220 Metric Tons
south-
east
2880 Metric Tons
Northeast
4140 Metric Tons
South
Central
2880 Metric Tons
West
2880 Metric Tons
Total: 18,000 Metric Tons
FIGURE 4-3 EXAMPLE OF REGIONAL CONSUMPTION DATA—U.S.
REGIONAL CONSUMPTION OF WOOD TREATED WITH
PENTACHLOROPHENOL, 1978
Source:
Scow, et al. An exposure and risk assessment for pentachlorophenoi
Final Draft Report. Contract EPA 63-01-3857. Washington, DC•
Monitoring ana Data Support Division, Office of Water"Regulations
and Standards, U.S. Environmental Protection Agency; 1980.
-------
to reported occurrences of accidental spills or leaks. One estimation
method is based on knowledge of t,ne pollutant's principal transportation
mode and of the types of its secondary users (large or small, nature of
the operation).
In a materials balance for phthalate esters (Pervak et_ al. 1981a) ,
it was reported chat the chemicals were transported principally in
liquid form via unpressured rail tank car. motor tank car, and, to a
lesser extent, in small quantities (55 gallon drums).
The amount of loss associated with transportation (other than from
accidents) was assumed to be a function of the size of the shipping
container and the remaining amount after the container is "empty." Most
esters are shipped by rail tank cars or tank trucks to distribution
points and sites of major users. Some of the smaller operators among
the 8000 compounders of plastics probably receive the plasticizers in
55-gallon drums. Small operators using rotational molding, coating
processes, and small injection molding processes could possibly obtain a
major portion of their plasticizer in this manner. Products made with
these processes account for approximately 75,000 kkg of phthalate esters
per year.
If it is assumed that 80% of this production is accounted for by 20%
of the companies who are large enough to purchase in tank car lots, then
the remaining 20%, or 15,000 kkg, might be delivered from the manufac-
turer to the compounder in 55-gallon drums. If between one cup and one
quart of plasticizer remains in each empty drum, then between 0.11% and
0.46% or between 18 kkg and 68 kkg could be wasted and released to the
environment when the drum is reconditioned, destroyed, or stored in a
manner that allows the remainder to be released.
Though it is unknown what percentage of the phthalate transported
in tank cars or tank trucks remains after the material has been delivered
and the tank is "empty," an estimate of approximately one-tenth of one
percent remaining was considered reasonable. This amount will either be
cleaned from the tank prior to loading another commodity or will remain
in the tank if the vehicle is in dedicated service. Information on
numbers of tank cars that are dedicated is unavailable. For estimating
purposes, it was assumed that 0.1% of the material transported is cleaned
and flushed with water. The amount being transported in tank cars and
tank trucks would be approximately 97% of the total production. A
weighted average of the waste from the 3% delivered in 55-gallon drums
and the 97% delivered in tank cars is still approximately 0.1%. For
calculating a materials balance, it was assumed that 0.1% is lost be-
cause of transportation-related causes.
4.5.2 Publicly Owned Treatment Works
For some pollutants, discharge from publicly owned treatment works
(POTWs) constitutes one of the largest direct releases to surface
waters. Monitoring data on flow rates and pollutant concentrations for
4-23
-------
POTW influents and effluents and on plant efficiency levels have
-
been
(Fikslf ef af ai98enf %??* ?°" ^ WaS esti^ted "? three methods
Ui.Ksei et al. 1981), all based on data compiled from sampling and
analysis at 20 POTWs. One estimation was based on the averagf Sfluent
""I""?113 " 2° Plan" SUrVeyed "ere "Presantative of ail plants
across the country. An alternate approach used the total cyanide
S
Each ^ assue ha he oplar0t C7"11d'
tive of all U.S. POTWs (Ftksel « 2 198?)!
concsion »as f t 2 otho-.
to these standards and had suffici^t data oi ,n operated
to allow analvsis. The flo»-wei!hS ? U P3""41".' of Interest
was 8U for the 2' mean ° the rffi"°val a«
10 primary
used to characterize metals -pmn, ^ -*• 3 §overnment survey were
t«rclar7 treatmentZplStsa!S SS^^Sj"^.? W^f ? ««ond.ry and
from POTI^ undergoes primary treatmenc 39/secondarv ?«- ! tOt^ tl
secondarv, and 14% tert-farv ^1-. ^ secondaiy, ia/. advanced
tc remove 88? of zinc -hiL r % Advanced secondary is assumed
86%. nC' "nile tertla^ treatment is assumed to remove
4-29
-------
Table 4-6 summarizes the POTW zinc budget based on these assump-
tions and shows a total loading to POTWs of 22,083 MT, of which 7814 MT
of zinc is discharged by POTWs to the aquatic environment, while 14,259
MT is discharged to land (Perwak ejt al_. 1980c) .
4.5.3 Natural and Inadvertent Releases^
There are several sources of natural inadvertent releases of pollu-
tants that contribute potentially large but usually widely distributed
releases to the environment. The metals occur as natural constituents
of the earth's crust in soils and rock formations throughout the U.S.
As soils and rocks are weathered and eroded, the natural metals and min-
erals are released to surface streams. Nickel concentrations in soils
generally range from 5 mg/kg to 500 mg/kg; the concentration in U.S.
soils averages 30 mg/kg (McNamara _e_t a_l. 1981). Other sources indicate
that nickel is found at average concentrations of 50 mg/kg in sedimentary
rocks, shale, and carbonate rocks. The average annual total suspended load o
nickel in the United States is estimated to be 3.6 billion MT, 25% of which
enters the major streams. Assuming an average nickel concentration of 30 ma/'
in soil, approximately 27,000 MT of nickel is discharged to surface waters' vi,
this route (McNamara e_t_ al. 1981) .
Urban runoff also can provide a major contribution of pollutants
to POTWs and surface waters each year. Mercury has been found in urban
runoff at levels of about 0.2-35 ug/1. The mean value for a residential
area of 720 acres in Rochester, NY was found to be 18.1 ug/1, with the
median value for the same set of 10 data points in the range 4-5 ug/1.
A second study involving less intensive sampling of stormwater and com-
bined sewer runoff in 11 cities across the U.S. (including Rochester, NY)
revealed concentrations ranging from less than 0.2 ug/1 to 0.6 ug/1.
The mean and median values for this data set were both equal to 0.3 ug/1.
(The mercury concentration reported for Rochester in this study was
0.25 ug/1.) Lacking further information, a range of 0.2-20 ug/1 in
urban runoff was used to show the possible magnitude of the source.
Thus, for runoff volumes of 17.3 x 1012 1/yr and 3..6 x 1012 1/yr going
to surface waters and POTWs, respectively, 3.5-350 kkg goes to surface
waters and 0.8-30 kkg to POTWs each year (Perwak e_t al. 198lb) .
4.5.4 Releases to the Atmosphere
Atmospheric releases of a pollutant can be significant and are a
potential pathway to surface waters. Included among the releases to air
that are commonly evaluated in a materials balance are releases from
chemical production, processing or refining, releases as a result of
consumptive use and release as a byproduct of indirectly related processes.
Automobile exhausts provide a source of atmospheric emissions which is
typically considered as an area source of release and poses a serious
problem in areas with high traffic densities.
Cyanides are one group of pollutants that has been detected
automobile exhausts (7iksel e_t al. 1981) . The average rate of hydrogen
4-30
-------
TABLE 4-6. EXAMPLE OF MATERIALS BALANCE FOR PUBLICLY
OWNED TREATMENT WORKS: ZINC
Primary treatment
Secondary
Advanced secondary
Tertiary
Total
Treated Flow (MGD)(1'
7,525
10,137
4,731
3,812
26,205
Zinc Loading ._.
to POTW (MT)
6,341
8,543
3,987
3,212
22,083
Treatment
Removal
Efficiency
.17<3>
.81(3)
.88('()
.86(4)
POTW
Discharge
To Sludge
1,078
6,920
3,509
2,762
(MT)
To Water
5,263
1,623
478
450
.65
(overall)
14,269
7,814
) - 0.8427 x flow.
EI'A 1978 Needs Survey, FKD-2.
(2)L(MT/yr) = flow (MGD) x 610 (10~6 g/1) x 3.785 (1/gal) x 365 (day/yr) x 10~6
(3)
Flow-weighted mean value calculated from Sverdrup and Parcel Associates data, Fehruary 1977.
(4)
Assume advanced treatment removes Zn proportionately to TSS — estimated from tables 17, 27, 31 of
EPA 1978 Needs Survey, FKD-2.
Source:
I'ei-wak, J. et ill. An exposure and risk assessment for zinc. Final Draft Report. Contract
EPA 68-01-3857. Washington, DC: Monitoring and Data Support Division, Office of Water
Regulations and Standards, U.S. Environmental Protection Agency; 1980.
-------
cyanide emissions has been reported to be 12 mg/mile. A fleet
composite emission factor was estimated for hydrocarbons in automobile
exhaust: 8 g/mile in 1976. The resultant CN/HC emission ratio (1.5 x
10~3) multiplied by the total annual hydrocarbon emissions of
12 x 10b kkg/year yields an estimate of HCN emissions of 18,000 kkg/year.
Applying the CN/HC emission ratio to estimates of exhaust emissions
compiled by U.S. EPA, the largest cyanide emissions from automobile
exhausts would occur in areas of the highest traffic density, such as
California (1500 kkg CN/year) or the combined states of New York and
New Jersey (1500 kkg tons CN/year) (Fiksel et al. 1981).
' -i O
•4-jZ
-------
REFERENCES
Fiksel, J.; Cooper, C.; Eschenroeder, A.; Goyer, M.; Perwak, J.; Scow, K.;
Thomas, R.; Tucker, W.; Wood, M. An exposure and risk assessment for
cyanide. Final Draft Report. Contracts EPA 68-01-3857, 68-01-5949.
Washington, DC: Monitoring and Data Support Division, Office of Water
Regulations and Standards, U.S. Environmental Protection Agency; 1981.
McNamara, P.; Byrne, M.; Goodwin, B.; Scow. K.; Steber, W.; Thomas, R.;
Wood, M. : Wendt, S.; Cruse, P. An exposure and risk assessment for
nickel. Final Draft Report. Contracts EPA 68-01-5949 and 68-01-6017.
Washington, DC: Monitoring and Data Support Division, Office of Water
Regulations and Standards, U.S. Environmental Protection Agency; 1981.
Perwak, J.; Goyer, M.; Harris, J.; Schimke, G.; Scow, K.; Wallace, D.
An exposure and risk assessment for trihalomethanes. Contract EPA 68-01-
3857. Washington, DC: Monitoring and Data Support Division, Office of
Water Regulations and Standards, U.S. Environmental Protection Agency; 1980a
Perwak, J.; Bysshe, S.; Goyer, M.; Nelken, L.; Scow, K. ; Walker, P.;
Wallace, D. An exposure and risk assessment for copper. Final Draft
Report. Contract EPA 68-01-3857. Washington, DC: Monitoring and Data
Support Division, Office of Water Regulations and Standards, U.S.
Environmental Protection Agency; 1980b.
Perwak, J.; Goyer, M.; Nelken, L.; Schimke, G.; Scow, K.; Walker, P.;
Wallace, D. An exposure and risk assessment for zinc. Final Draft
Report. Contract EPA 68-01-3857. Washington, DC: Monitoring and Data
Support Division, Office of Water Regulations and Standards, U.S.
Environmental Protection Agency; 1980c.
Perwak, J.; Goyer, M.; Schimke, G.; Eschenroeder, A.; Fiksel, J.;
Scow, K.; Wallace, D. An exposure and risk assessment for phthalate
esters. Final Draft Report. Contracts EPA 68-01-3857, 68-01-5949.
Washington, DC: Monitoring and Data Support Division, Office of Water
Regulations and Standards, U.S. Environmental Protection Agency; 1981a.
Perwak, J.; Goyer, M.; Nelken, L.; Scow, K.; Wald, M.; Wallace. D. An
exposure and risk assessment for mercury. Final Draft Report. Contracts
EPA 68-01-3857, 68-01-5949. Washington, DC: Monitoring and Data Su"Pnort
Division, Office of Water Regulations and Standards, U.S. Environmental
Protection Agency; 1981b.
4-33
-------
Scow, K.; Cover, M.; Perwak, J.; Payne, E.; Thomas, R.; Wallace, D.;
Walker, P.; Wood, M. An exposure and risk assessment for pentachloro-
phenol. Final Draft Report. Contract EPA 63-01-3857. Washington, DC:
Monitoring and Data Support Division, Office of Water Regulations and
Standards, U.S. Environmental Protection Agency; 1980.
Serth, R.W.; Hughes, T.W.; Opferkuch, R. E.; Einautis, E.G. Source
assessment: Analysis of uncertainty principles and applications. EPA-
600/2-78-004u. U.S. Environmental Protection Agency; 1978.
a— j4
-------
5.0 ENVIRONMENTAL PATHWAYS AND FATE ANALYSIS
5.1 INTRODUCTION
The results of the materials balance analysis will normally provide
important information on pollutants that enter the environment; the
amounts, types, forms, rates, and locations (both in terms of region and
specific receiving medium) of environmental releases; and indications of
direct exposure routes associated with the release of pollutants to the
environment. If the environment and chemicals were static, materials
balances, combined with information on receptor distribution, could be
used to estimate exposure of humans and other biota to environmental
pollutants. In most cases, the environment is not static but is dynamic
in the sense that pollutants may be transported, undergo physical, biol-
ogical and chemical transformations, accumulate or disappear, resulting
in an environmental distribution quite different from that associated"'
with the initial environmental release. For example, Figure 5-1
summarizes the major environmental transformations and transfers of
trichloroethylene and illustrates the dynamic nature of the chemical's
behavior following its release into environmental media (Thomas ~t al
1981) Therefore the environmental pathways and fate Processes~f~he oollutant
must be considered before one can determine with a reasonable decree of '
confidence the pollutant's chemical form and environmental concentrations
to wnicn receptors might be exposed.
In analyzing environmental pathways and fate, the following types
of questions are important:
• Do the pollutants remain in the environmental media (air, water,
land, biota) in which they are initially released or are there
intermedia transfers?
• By what mechanisms are the environmental concentrations of the
pollutant decreased or increased, e.g., biodegradation or intra-
and intermedia transfer?
• What are the controlling external influences on intermedia
and intramedia transfer?
• What are the rates of these transfers or reaction mechanisms?
• Are there any potential degradation products of concern with
respect to environmental or health risks?
• Is a steady state pollutant concentration distribution in the
environment achieved? Is the total environmental load increas-
ing or decreasing? What are the environmental dynamics'
5-1
-------
Ui
I J
from
Sotuces
(bjckijtuund cuncuntrdiion
- Photochemical Duyiadation
,* '/iday - 2»laysl
Lejching to Deep Suilk
and Gruurtdwalar
(liont dump*,,
t'is, ulc )
Sorpuon/
Oesutpiion
Minor
Aquiler (loriQ residence lime)
Source:
FICURE 5-1 EXAMl't.E OF ENVIRONMENTAL PATHWAYS AND FATE
ANAI.YSIS— MAJOR PATHWAYS OF TRLCH1.0ROET11YLENR
wM f"r flchloroethylene. Final Draft Report.
at on and Standard- ul r "' MV»Icori"8 a"d »«ta Support Division, Office of' Water
at ions and .Standards, U.S. Environmental Protection Agency; 1981.
-------
• What is the anticipated spatial and temporal distribution of the
pollutant: in the environment, in different media, among different
types or forms of the pollutant, for different geographical areas,
for different time frames? Are these distributions confirmed bv
monitoring data?
Answers to these types of questions will provide information on the
exposure of different receptors that come in contact with various environ-
mental media, particularly if these receptors play a role in the trans-
port, reaction or distribution of the pollutant (e.g., biodegradation.
uptake by plants, etc.) or are associated with a particular medium that
accumulates the pollutant (e.g., persons who ingest contaminated fish
tissue).
Monitoring data have often been considered as a substitute for en-
vironmental pathways and fate analysis since monitoring data directlv
provide the environmental distribution of pollutants. Ideally, an en-
vironmental pathway and fate analysis should be conducted in addition to
review of monitoring data for several reasons:
(1) For many pollutants, particularly organic and new chemicals,
monitoring data are limited, sporadic, and/or of questionable
reliability.
(2) Analysis of available monitoring data does not enable the
estimation of environmental concentration distributions in
media or geographic locations for which monitoring data are
not available.
(3) Analysis of monitoring data does not provide information on
how and where physical, chemical and biological processes in-
fluence the environmental distribution of a pollutant.
(4) Analysis of the effects of different pollutant control options
requires some knowledge of the relationship between environ-
mental loadings and concentration distributions-, thi«
information is provided by fate models rather than by monitor-
ing data.
_ Monitoring data, when available, can be a direct source of informa-
tion for exposure analysis and can be used to calibrate (or e^tranolate
trom) models used to estimate environmental distributions. However in
most exposure analyses, it will be important to evaluate pathways and
fate data as wexl as aid in exposure determinations and in the d-velop-
ment of regulatory recommendations.
-------
5.2 GOALS OF ENVIRONMENTAL PATHWAY AND FATE ANALYSIS
The overall goal of environmental pathway and fate analysis is to
establish the distribution of pollutants—both spatially and temporally—
in all environmental media. This general goal can be divided into a
number of specific objectives as follows:
(1) Define environmental media or compartments of importance to
the environmental behavior of the pollutant, including sub-
compartments such as soil layers, aerobic or anaerobic zones,
where necessary.
(2) Identify important mechanisms for transport and physical,
biological and chemical change (pathways) of the pollutant
within and among environmental media.
(3) Summarize and/or develop data on the rates of these transfer
and reaction processes, determine the processes that control
environmental fate and distribution, and identify predominant
chemical forms or degradation products in various media.
(4) Estimate "lifetimes" or half-lives of pollutants in the environ-
ment
(5) Using materials balance/environmental loading estimates as in-
puts, trace the environmental pathways of pollutants from their
sources to their sinks or ultimate distribution in the environ-
ment.
(6) Estimate average or representative pollutant concentrations
and their time dependence in specific environmental media.
(7) Estimate concentrations and their time dependence in specific
geographical locations—e.g., river basins, streams, rain, air
sheds, etc.
(3) Use monitoring data ;o compare (and to improve) the results of
environmental pathways and fate analysis wherever possible.
(9) Using pathways and environmental fate analysis, develop informa-
tion for use in exposure analysis, and provide a basis for
estimating quantitative relationships between environmental
releases and exposure.
Ideally, a pathways and fate analysis traces all of the environmental
releases of pollutants from specific sources through the environmental
pathways that occur, and combines the resultant contributions of eacb
release and transfer Co obtain the spatial and temporal distri-
bution of the pollutant in the environment. A careful accounting of
pollutant inputs, inter- and intramedia transfers, -and transformation/
accumulation/degradation processes, should reveal the variation in
5-4
-------
distribution of the pollutant over time. This might be accomplished by
using a simple partitioning model for a rough estimate, a complex
environmental model or similar analytical techniques. Large-scale multi-
media models exist; however, they have been designed either for specific
pollutants or for specific environmental compartments and geographical
areas. These large models require extensive and elaborate calibration
procedures. Furthermore, the set of necessary input data—loading rates,
transport and transformation rates, and other characteristics of the
pollutant—is difficult to obtain or the data are not reliable enough
for the results to be credible. Resource allocations further constrain
such efforts. As a matter of necessity, environmental fate and pathway
analyses often are fragmentary and incomplete, focus on only a few major
pathways, and do not give complete distributions of concentrations of
the pollutant. To be useful in exposure analyses, environmental fate
and pathways analysis should at a minimum:
• distinguish key pathways from insignificant ones;
• focus on key pathways and on media where the amount or concentra-
tion of pollutant is large and where exposure of humans and
other biota is expected;
• estimate a probable range of pollutant concentrations in differ-
ent environmental media, with time and space resolution appro-
priate to pollutant sources and receptor exposures of concern;
• provide estimates of uncertainties that can be carried through
the exposure or risk analysis.
The outputs of the pathways and fate analysis will be greatly con-
strained fay the amount of information available concerning the physical,
chemical, and biological characteristics of the pollutant; the types,
nature, and location of the pollutant sources; the existence of models
or calculational approaches available to estimate' the concentrations; and
the resources committed to the analysis.
In keeping with the goals of exposure and risk assessments within the
Office of Water Regulations and Standards, the methodology focuses on
water-related pathways and fate analysis. Non-water-related pathways
should, of course, be considered because of the interrelations among
environmental media, and in order to obtain a perspective on total exposure
5.3 ENVIRONMENTAL PATHWAY AND FATE ANALYSIS METHODS
Three general methods useful for environmental pathway and fate
analysis are described in this section. The methods have similar goals
several common steps, and may use some of the same data and information.
Eacn one, however, has a different focus. The choice of approach wilj
depend upon the nature of the pollutant, the scope of the exposure
analysis, and the availability of data. In general, portions of mo-*
than one approach may be used and the results integrated in order to
-------
develop a more complete understanding of the environmental pathways and
fate of a pollutant.
5.3.1 Environmental Scenario/Case Example Method
This approach will provide a qualitative assessment of pathways and
fate mechanisms, supplemented by semi-quantitative information on pollu-
tant distribution where sufficient depth of analysis of specific case
examples is gained from literature sources. It begins with a brief
review of materials balance and environmental loading data to identify
relevant scenarios or case examples for a particular pollutant (see
Figure 5-2 for steps in process). Each major source category is iden-
tified and hypotheses are developed concerning pollutant fate, beginning
with the source and proceeding to an ultimate sink or environmental
distribution.
For example, considering agricultural application as a major source
of a substance used as an herbicide, one would indicate diagrammatically
the likely pathways and fate processes beginning with the application
and including: soil adsorption/desorption, chemical decomposition, bio-
degradation in the soil, volatilization, runoff and leaching into local
waterways, uptake by plants or animals and distribution along the food
chain (i.e., all major known fata mechanisms). Exposure routes suggested
by this scenario include ingestion through food and drinking water (both
humans and non-humans) and possioly skin absorption through contact with
polluted water. A scenario such as this describes the likely pathways to
be examined further or validated in subsequent steps.
The second step is to assemble and review available data to aid in
evaluation of the pathway and distribution hypotheses. Literature data
from both laboratory and field studies, as well as measured concentrations
in the environment would be reviewed for mechanisms and rates of transport
or chemical and biological transformation. Data on the physical, chemical,
and biological characteristics relevant to determining the pollutant's
fate in the environment would also be reviewed. Using herbicide applica-
tion again as an example, one would review laboratory and field data on
plant uptake, soil adsorption, concentrations found in soils and water,
the rates of transfer from soils to ground (through leaching) or surface
water (through runoff) or to air, as well as data on speciation, photolysis,
biodagradation, hydrolysis, and other transformation processes. These data
can provide quantification or at least support comparison of the processes
or major pathways involved.
When no data are available, one of several estimation techniques mav
be used to provide a rough idea of the significance of particular proper-
ties of a chemical in its environmental fate. For example, Lvman et al.
(1982) have compiled methods for estimating a number of physical, chemical
and biological properties of organic chemicals in one handbook; these
methods will be computerized in the near future.
-------
Develop scenarios
and hypotheses for
environmental path-
ways beginning
with materials
balance outputs
Review and analyze data
from "case examples" of
studies of pathways and
distribution of pollu-
tant
Review and analyze physical
chemical and biological
properties of the pollu-
tant, pertinent laboratory
or field tests on mobility
and stability in the en-
vironment
Draw general
and specific
conclusions on
major scenarios
and hypotheses
Associate environmental distribu-
tion vith sources, and extrapolate
if possible from loading estimates
to environmental distribution
FIGURE 5-2 DIAGRAM OF ENVIRONMENTAL SCENARIO APPROACH
TO PATHWAYS AND FATE ANALYSIS
5-7
-------
The third step is to draw both general and specific conclusions
concerning the key factors influencing the fate of the pollutant, how
the pollutant is partitioned in the environment, likely" concentrations
in each compartment, and so forth. General conclusions may indicate
that certain sources are associated with higher water concentrations in
a particular habitat than others, or that one transformation process is
more important than others in the ultimate fate of a substance. Specific
conclusions may be drawn from selected subjects described in the liter-
ature, and may yield either semi-quantitative relationships between
loading rates and environmental concentrations, or specific data relating
to the rates of typical processes.
The final step is to relate the specific sources and loadings given
in the materials balance to the pollutant's environmental distribution.
This step can indicate where likely exposure to the pollutant may occur
and give some idea of how exposure may change as a function of future
control strategies. In a sense, this final step involves reexamination
of the scenarios developed in the beginning, attempting to quantify them
and show the relationships between the sources and distribution of' the
pollutant and resultant exposure.
The environmental scenario method seems most appropriate in the
following situations:
(1) Sources of the pollutant are relatively well known and major
sources can be distinguished from minor ones in terms of the
quantity released to the environment and its relationship to
potential exposure.
(2) The pollutant is well-known in the sense that field and labora-
tory studies exist, data on the chemical, phvsical and biol-
ogical characteristics of the pollutant are'available, and the
behavior of the pollutant in the environment has, at least, been
addressed by others.
(3) Monitoring data for the pollutant are available, so that
measured data rather than estimates can be used in forecasting
exposure. The fate and pathway analysis in this case is more'
important for linking sources to environmental pathwavs and
spatial and temporal distribution rather than to estimate con-
centrations.
(4) Large-scale models describing the pollutant pathways are either
not available, not useful, overly complex, or too general or
gross in scale for application given the resources available.
Additionally, the availability of other data described above
may make the use of complex estimation techniques unnecessary.
5-6
-------
5'3'2 Critical Pathway/Distribution Estimation Method
This approach is focused on identifying the critical pollutant
pathways that are most influential in determining distribution of the
pollutant and the resultant concentrations in the environment. The
steps involved are shown diagrammatically in Figure 5-3 and described
briefly below.
This approach also begins with the results of the materials balance
analysis. However, rather than scenarios of environmental pathways re-
lated to particular production or use patterns, aggregate loadin*'rates
(temporal or spatial combinations) for different media are first°devel-
oped, eitner on a nationwide basis or for specific regions with identified
environmental loading patterns. For example, the total annual loading
from direct and indirect releases to air, water, soil, and occasionally
biota, etc., are determined from the materials balance results; estimates
are then made of the sizes of receiving media, e.g., volume of water, air
or soil receiving the total loading. If possible, these estimates mav be
made tor regional or other smaller scale geographical locations, to focus
on areas of concern. General environmental characteristics of the re-
ceiving media important to the pollutant's distribution are also estimated
e.g., pH, soil moisture content, etc.
_ Second, data on the physical, chemical and biological characteristics
or the pollutant are reviewed. Pertinent data may include molecular vei*h«-
aqueous solubility, vapor pressure, octanol/wacer partition coeffirient " "
biodegradacion rates, chemical reaction rates, etc. These data are uspd to
provide insight into important transformation processes that remove the
polxutant and/or convert it to other products or potential contaminants.
-roaacts identified may also be considered for toxicity and further ^rans-
rormation. (Time and resources generally prohibit full consideration of
these products in an exposure assessment.)
Third, pathways and processes that result in transfer of the pollutant
trom one medium to another are evaluated in order to identify the critical
transter pathways and estimate the relative rates of the transfer processes.
For example, for a pollutant initially released to water, vaporization and
sedimentation might be considered in order to determine whether and how
rapidly the pollutant is transferred to the air or soil media. Basicallv
this analysis is a simple partitioning study, to determine if the oollutant
tends to remain in the initial receiving medium or is transferred r0 others
xn some cases, all of the pollutant may be rapidly redistributed to other
media; wnile in others, a slowly established equilibrium distribution may
be indicated.
Following an initial determination of the partitioning, the next step
is a more detailed examination of the pollutant fate in the media tha«- are
ot most interest, e.g., those media or subcompartments into which the
poUutants are likely to be partitioned. Each major fate process is re-
viewed, using rate and equilibrium relationships and estimation techniques
.rom tne literature, pollutant specific data (e.g., rate constants) from
laboratory or rield studies, model ecosvstem results, etc. Tvt>ica> nro-
casses to oe examined include:
3-9
-------
Keviuw physical, chemical
and biological character-
istics of pollutant that
Influence late and pathways
Using materials
balance Lesult.s,
aggiegale load-
ing for dlflerent
im-dia, on national
and situ I IL- r bra le
where pos^ idle
O
tlxamlne and quan-
ti fy, where pos
slble, initial
parti tinning
among environ-
mental mediu;
establish critical
pathways for
t rans f«r
II-
cal
— -
Review tale pro-
cesses In each
media to estab-
lish key deler-
mlnanls of con-
centrations and
ills U 1 but I on
—
Using simple
mode 1 y and ra r e
01 tujui i ibr Linn
re lat ioitbhi pt» ,
et>l iiiule rules
of change, and
eijui 1 ibr Him
ul pol In tau C in
aacit mtdi a
-*
Using aggregate load-
ing escinutes. and
rates fiom previous
steps , estimat u concen-
tratJon ranges in c.u.li
mei.1 ia , both fi>i
gdderal and i>pecifit:
(1 1st 1 1 !»ut Jijut. ol soiiiVL-s.
Conduct "sen si t i vi Ly
analybis" to determine
va r i ous pa rauie te rs
--
Summarise cr 1 ti cat
pathways , concen-
trations , distri-
butions ; compare
with monitoring
data , and idenl i ty
expos ui e pi»tential
KICUKK *)-J
DIACRAM OK CRITICAL I'ATIHJAY/DISTRIBUTION F.ST1MATFON
MLTII01) FOR ENVIRONHI.NTAL PATHWAYS Atil) FATE ANALYSIS
-------
free radical oxidation, photolysis, adsorption to
aerosols, dry deposition, fallout, rainout,
scavenging
Water hydrolysis, photolysis, chemical oxidation, pre-
cipitation, adsorption/desorption with sediments,
volatilization, biodegradation
Soil leaching, hydrolysis, surface photolysis, chemical
oxidation, volatilization, uptake, adsorption,
complexation
Biota biodegradation, metabolism, bioaccumulation, bio-
magnification, etc.
Some of these processes may have been considered earlier, since they
are ones that result in intermedia transfer. By using very simple models
generalized rate equations, or measured values reported in the literature'
the rates or degradation or transfer within media can be estimated
Those processes are identified that are major determinants of the rate
levels "' decomposition' or ruction, and hence the ambient concentration
The next step is to utilize the aggregate loading rates, general
partitioning estimates, and the estimated transfer or reaction rates to
calculate likely ranges of concentration of the pollutant in the environ-
mental media. Frequently, these calculations will involve "single
compartment" models or equations along with physical/chemical properties
One would assume that the entire loading of the pollutant enters one medium
with a specific volume, and estimate the resultant concentrations of
pollutant, changes over time, and/or steady-state concentrations
Steaay-state values will be important for highly persistent pollutants; trans-
formation and transfer rates control the concentration distribution of
pollutants tnat are more mobile or shorter-lived in the environment
w?ll°h^areSK 36Verai enjironmental compartments (and/or decay processes)
will nave to be considered simultaneously, where both are important.
Feedback from one set of calculations to another is required In
general, however, these estimates are made to place boundaries on the
distribution and concentration of the pollutant and not to determine
absolute values. Thus, use of more complex, multi-compartment or sinale
compartment models may not be necessary or appropriate for this approach.
Estimates of environmental concentration ranges can then be mad* cor
smaller geograpnical locations, using specific source loadings and
parameters describing the associated receiving media. A number of simple
models are available for this approach, for example, Mackay's fugacUv
model (Mackay 1979) and simple computer models such as PLUME
(EPA 1979), among others. These models are described in grea^er
detail in Section 5.3.3. Sensitivity analvsis can be conducted to
determine which parameters influence the outcome to the greatest extent
-------
The final step in this approach is to summarize the results in terms
of the critical pathways, estimates of the environmental distribution and
pollutant concentration ranges., and comparison of these concentration
ranges with available monitoring data.
This overall method is appropriate for considerations in several
situations:
(1) when the pollutant is not well-known and few laboratory or
field environmental data are available;
(2) where there are only a few types of important pollutant
sources, and where distributions can be easily estimated,
or where data exist on major releases in specific areas;
(3) where monitoring data are sparse, or are widely distributed
or uncorrelated with release patterns;
(4) where insufficient data and resources are available to use
more complex environmental models, or where the quality of
input data does not justify their use;
(5) in specific situations identified by using the environmental
scenario method.
5.3.3 Modeling Approaches
As indicated earlier, simple calculations and modeling approaches
are an integral part of the environmental scenario and critical pathway/
distribution estimation methods. However, these stuple models usually'
do not account for intermedia transfers and equilibrium relationships'
between different media. Multimedia modeling may be warranted, resources
permitting, if sufficient data exist on the chemical, physical and
biological characteristics of the pollutant, if accurate source and
loading data are available, and. if appropriate models are available.
The results of such models can provide a more accurate estimated dis-
tribution of the pollutant in the environment and provide a useful
mechanism for estimating the effects of different regulatory approaches.
Once the models have been validated and calibrated, they can be'used in
many situations with modest resource commitments.
are:
The general steps to be followed in multimedia modeling approaches
(1) Identify, from materials balance results, the major pollutant
sources and geographical areas considered for modeling.
(2) Identify the most significant environmental pathway for the
pollutant under the conditions selected above.
5-12
-------
(3) Select the individual models or multimedia model applicable
to the situation, i.e., type of emitting source, pollutant,
receiving media.
(4) Compile the input data required by the model(s) selected,
e.g., source/loading data, pollutant characteristics and
properties, environmental characteristics, and time.
(5) Use the model(s) to estimate pollutant fate (transport,
transformation, concentration) in the different media.
(6) Compare the model results to the results of the two approaches
described above, and to monitoring data. Comparison with
monitoring data is frequently required for calibrating the
model, and should be carefully accomplished before the model
is used for predictions.
(7) Perform a sensitivity analysis of model parameters that are
uncertain, or vary significantly in different locations.
(8) Analyze modeling results for insight into exposure of various
species, effects of regulatory actions, impact of reduction
in loading rates on environmental levels, significance of
environmental process in determining pollutant fate, etc.
A number of computer models have been sponsored by the U.S.
Environmental Protection Agency and other agencies to aid in pollutant
environmental fate and exposure assessments. Information about specific
models is not reiterated in this report; instead, the reader is referred
directly to EPA's Environmental Modeling Catalogue (U.S. EPA 1979) for
detailed information on available models. Other model reviews include
Miller (1978), among others.
Some examples of available models considering a single environ-
mental medium include the EPA UNAMAP system (air), EXAMS (surface water)
EXPLORE (stream), ARM (watershed) and SESOIL (soil). Examples of multi-'
media models that link two or more environmental media includ<= UTM and
ALWAS (air to watershed/stream), CMRA (overland to stream), and TOHM
(air to watershed/water). Other models include Mackay's fugacity method
which estimates equilibrium partitioning of a pollutant between air
water, sediment and biota (Mackay 1979), and Neely's microcosm model
(Neely 1978).
Most pollutants released to the environment are likely to be trans-
ferred between media. A model can provide a fairly detailed approach
for tracing pollutant levels in different media when the substance is
subject to removal or transformation by competing processes. By account-
ing for tne net rate of pollutant transport or transformation, the model
5-13
-------
cnaracterizes che pollutant's mass distribution in the environment,
temporal and spatial. Sensitivity analysis enables determination of
the significance of a particular process or variable. With an
efficiently developed model or set of models, assuming input data are
available, the modeling approach can provide valuable information,
in a timely and cost-effective manner.
However, currently there are significant difficulties in applying
multimedia models to most pollutants. Although a considerable amount
of research has been done in this area, most of the multimedia models
are still under development or have not been fully verified. Adequate
data (either chemical or environment specific) for input to models
are also often lacking. When the input data used are uncertain or must
be estimated, the results lack precision. The unavailability of a
suitable, verified model and/or sufficient input data precludes the use
of multimedia models in many instances. When multimedia modeling is
performed, results must be interpreted with care, because a selected
model may exclude certain pathways related to the complete traverse of
the substance through the environment. Since the results are obtained
from environmental conditions, a scientist has to be concerned with
the extrapolation of the conditions simulatued to generalized results
and environments.
5'4 EXAMPLES OF ENVIRONMENTAL PATHWAYS AND FATE ANALYSIS
Several illustrative examples of the methods described earlier are
presented in this section. Only portions of the calculations, re=u^ts
or discussions are given to indicate the types and results of the approaches.
5•4•1 Environmental Scenario Method
In an exposure and risk assessment for copper (Perwak at al. 1930)
the environmental scenario method was used to link sources a^d~p~athwavs
with environmental distribution. This method was selected for this '
pollutant because sufficient monitoring data were availabl* for esf-
mating exposure, and field and laboratorv studies had documented the
physical, chemical and biological processes that determine the pollu-
tant s behavior in the environment. Figure 5-4 depicts the major environ-
mental pathways for copper released to the environment through human
activities. 5 '
Figure 5-5 shows in greater detail environmental scenarios for
several of the environmental pathways of copper. In the second scenario,
wastes from primary copper production, coal mining, and copper ore mining
and benefication are shown to enter the air and water environment by
several paths; runoff and leaching mechanisms carry wastes to surface water
or^ ground water, respectively. The surface water/sediment interaction and
other flows are also shown. In the fifth scenario, several uses of copper are
-------
/. Olhei Anthro-
pogenic Source*
of Copper
Copper Mining
and Piuiluclion
Other Oie Minuiy
Runutl .mil W«lc Wdtei
>i>l and Oiy fallout
Land
Surface Soils
I.iliiuji Piles
Laguoni
Landfillt
etc.
J Occam
Suiliicc Waters »rxl Scdimcntl
Gruuridwater
Noie: Quantities of coppei moving in each pathway are roughly piopottional to the thickness of each pathway shown.
Slow movement from gioundwateis to surface waters not shown.
FICURE 5-4 EXAMPLE OF ENVIRONMENTAL SCENARIO IDENTIFICATION-
SCHEMATIC DIAGRAM OF MAJOR PATHWAYS OF COPPER
RELEASED TO THE ENVIRONMENT BY HUMAN ACTIVITIES
L. An exposure and risk assessment for copper. Final Draft Report. Contract
. Washington, DC: Monitoring and Data Support Division, Office of Water
KeguLations and Standards, U.S. Environmental. Protection Agency 1980
-------
Ul
I
PATHWAY NO.
la.
2.
Atmospheric binissions
(Major Point Sources)
CuO. CuS. Cu(m)
Cu Production
Smell my
lion & Steel Production
Coal Combustion
Incineration
Atmospheric Emissions
(Non point Sources)
CuO (paniculate). Others
—•**» O''1'
\w»b
Chrome & Brass Coicusion
Oil & Lubricant Combustion
& Leakage
Solid Waste & Tailings,
Coal Piles & Open
Pit Minus
Prmiaiy Cu Production
Coal Minmy
Ore Mining and Beneficial ion
Pathway
Pavement & Local
I to.id Soils
Surlace Waters
Sediment
(Slow)
Groundwater
Dissolved Solids
Stisp. Sediment
FIGURE 5-5 EXAMPLE OF ENVIRONMENTAL SCENARIO ANALYSIS-TYPICAL ENVIRONMENTAL PATHWAYS OF COPPER
Source:
WMm'^7a-'u Aln.exP°tlll«,1und rlsk aaaessment for copper. Final Draft Report. Contract
MA 68-01-3857. Washington, DC: Monitoring and Data Support Division, Office of Water
Regulations and Standards, U.S. Environmental Protection Agency; 1980.
-------
> Discharges
Senefic laiion
Smelting
Cu Production
Elraii
Electroplating
1 t
4 POTW ^ Primary
Intluent Tre.ilini.-iit *"
i , , 1. — _
1
(iintui)ical
4
"
ElflllLMII
Sullaci; Waters A
Si.'climciili, •
Ocean DiirnpiiH) Ty
Incin
ei .iiuiii
Land "~
I'll b
Air
So.l
(Slow)
<
Giuundwaier
^" (X:i;.nii
KH.RE 5-5 KXAMl™ OF ENVIRONMENTAL SCliNARIO ANALYSIS-TY1TCAL ENV1RONMBNTAL PATHWAYS OF COPPER (Continued)
-------
shown — as an algicide (CuSO/^) and as an agricultural chemical. The path-
ways from soil to ground and surface water, to sediments, and ultimately
to the ocean are shown.
These general descriptions were the starting point for subsequent
literature review, quantification of flow rates fron selected sources,
and ultimately analysis of concentrations in the environment. For
example, in order co describe the first scenario, the nature of solid
wastes and tailings was reviewed, along with data from studies of acid
mine drainage, concentrations found downstream of mine drainage sites,
incidents of groundwater contamination, and leaching studies from a
variety of sites.
The analysis revealed that solid wastes, coal piles, and tailings
are major sources of copper disposed of on land. Copper exposed as a
result of mining practices is subject to greater translocation in the
environment than releases from the other two sources due to the acid
nature of the leachate. Surface streams draining mined areas have been
shown to have localized spikes in copper concentration, with the level
quickly decreasing as the stream recovers in pH and alkalinity values as
a function of distance. The major processes affecting this reduction in
copper concentration are dilution, sorption, and precipitation.
In municipal waste landfills the copper concentration in leachate is
typically between 0.04-0.4 mg/1. Copper is quickly attentuated by the
soil. Data on groundwater contamination were not available though' such
contamination is rare in a properly operated landfill. In old mined
areas, acid mine drainage, and porous tailings enhance the possibili
of grcundwater contamination.
For the fifth scenario involving the agricultural use of copper
sulfate, data describing the fa:e of copper in the soil, water and' sedi-
ments were analyzed; field data were examined for indications of the
roles of adsorption or sedimentation of copper and for mean and maximum
concentrations in water or sediment. Use of copper sulfate as an algicide
appears to be effective within a very short time frame, and field studies
indicate that concentrations of copper ion in the water column decrease
to background levels within a day following application. Copper is trans-
ferred from water to participates, algae and sediments through sorption.
Sediment core concentrations reflect the use of CuS04 over the years in-
dicating that sediment is a significant ultimate reservoir for copper in
aquatic systems.
5-4.2 Critical Pathway /Distribution Estimation Method
The Critical Pathway /Distribution Estimation Method was used in the
pathways and fate analysis for pentachlorophenol (PCP) (Scow et_ al. 1980)
which is characterized by limited monitoring data, adequate d~ata~on basic
chemical properties, and a good general understanding of its overall
materials balance. Little documentation was available on PCP emissions
from particular consumers or producers. The critical pathways approach
5-13
-------
was used in a risk assessment of this pollutant in order to provide PCP
concentration estimates in various media and to determine the critical
pathways influencing its distribution. Since PCP use is concentrated in
a few industry categories, the fate and pathways analysis was focused on
these operations.
In order to provide a better understanding of the fate, distribution
and potential for exposure to PCP following discharge from significant
sources, simple quantitative models were used. Four sources to air were
considered for their contribution to national atmospheric levels of PCP.
Three sources—cooling towers, wood preserver evaporation ponds, and
direct aquatic discharge as a general phenomenon—were considered for
their local impact and exposure potential. The type of input data required
included source characteristics (e.g., dimensions, emission or loading
rate, etc.), environmental characteristics for a representative set of
conditions (e.g., wind speed and direction) and chemical characteristics
(e.g., transformation rates). As an example of how the approach is
applied, the development of the equation describing local'pollutant con-
centrations from a cooling tower plume is described briefly.
First, the assumptions made in developing the equations were defined.
These assumptions included the values chosen for each variable, such as
plume buoyancy and height, wind speed, temperature, and others. Also,
variables not included in the equation but potentially influencing the
resulting concentrations were identified (e.g., rainout, large-scale
turbulence, chemical reactivity). Second, the fate of pentachlorophenol
during cooling tower evaporation was characterized by a Gaussian concentra-
tion distribution using a simple plume model. The output of the model
equation—PCP concentrations as a function of distance from the source
for two plume source heights—were plotted, as in Figure 5-6. The results
of the equation were used directly as concentrations"from which to estimate
human exposure in subpopulations residing within specified distances
of a cooling tower.
5.4.3 Modeling Approaches
Computerized and hand-calculator models have been used in the environ-
mental ..fate and pathways analysis of numerous priority pollutants. Three
approaches applied in exposure and risk assessments for phthalate esters
(Perwak et al. 1981) and dichlorobenzenes (Harris et_ _al_. 1931) are described
below. (No example is given of the application of a complex multi-media
model, though such models will undoubtedly prove to be useful in future
exposures and risk assessments.)
5.4.3.1 Phthalate Esters
Based on two existing environmental fate models for phthalate
esters: The Exposure Analysis Modeling System (EXAMS), developed by
the U.S. EPA (Wolfe et_ _al_. 1979), and Neely's partitioning model (Neely
1978), ambient concentrations of di(2-ethylhexyl) phthalate (DEHB) were
estimated in a simplified three compartment model for water (including
sediment and fish). Some of the model's assumptions were adapted in
this application to incorporate more current or relevant data develoned
-------
10
-3 _
Nonbuoyant Plume Source Height • 10m
C
8
cj
T)
5
5
"3
"
o
10
i
I
10'
10*
Buoyant Plume Source Height = 120m.
•4 _
10
km Downwind
FIGURE 5-6 EXAMPLE OF USE OF SIMPLE QUANTITATIVE MODEL TO
ESTIMATE ENVIRONMENTAL DISTRIBUTION—GROUND-LEVEL
CONCENTRATIONS OF PENTACHLOROPHENOL IN THE PLUME
DOWNWIND OF A COOLING TOWER (TWO SOURCE HEIGHTS)
Source: Scow, K". et al. An exposure and risk assessment for pentachloro-
phenol. Final Draft Report. EPA Contract 63-01-385?'. Washington,
DC: Monitoring and Data Support Division, Office of Water Regula-
tions and Standards, U.S. Environmental Froteccion Agency: 1980
5-20
-------
in the materials balance and in the review of physical and chemical
properties. Some of the data were scaled for a national approach from
a site-specific approach.
Concentrations in air in the U.S. were estimated using the equation:
dx Q ,
27 = I " kx
where: x = mass concentration
t = t ime
Q = area source strength
H = mixing height
k = rate constant for removal.
The concentrations predicted by the models for DEHP in water, sedi-
ment, fish and air were then compared with measured concentrations reported
in the literature. Figure 5-7 summarizes the predicted and measured levels,
The results of this analysis indicated that DEHP is usually presenc ac
extremely low concentrations in air, and at low levels in water; it is
subject to significant chemical transformation in air but virtually none
in water; and it is likely to accumulate in sediment and fish to levels
from two to three orders of magnitude greater than water column concentra-
tions.
5.4.3.2 Dichlorobenzenes
For the environmental distribution analysis of 1,2-dichlorobenzene.
two separate fate models were implemented and the results compared.
Mackay's Level I fugacity approach was used assuming all environmental
compartments—air, water, sediments, biota—were at equilibrium and
connected and there was no degradation or transport out of the selected
environment. The EXAMS model was run for three generalized, pre-compiled
aquatic systems—a pond, an oligotrophic lake and a river. Input data
to both models were based on materials balance information and physical/
cnemical properties of dichlorobenzene compiled from published literature.
For typical environmental loading rates, both models predicted a
high sediment to water ratio (two to three orders of magnitude) under
equilibrium conditions, and partitioning into biota. Table 5-1
summarizes the results. Table 5-2 gives more detailed results of the
EXAMS model. Volatilization was the primary means of disposition from
ponds and lakes, aquatic systems in which transport downstream is
minimal. The differences in the results of the' two models were due to
the fact that a greater proportion of dichlorobenzene partitioned to
the air compartment in the Mackay model and the fact that EXAMS
considered kinetic processes, as well as simple partitioning.
-------
Mostly Degraded
by Oil-reaction
t
Calc.
Obs.
.concentrat
AIR
Remote
0-.008
.0004
ions in ]i
Urban
.02-. 09
.3
8/m3)
4.3 x 103 kkg/yr
5.3 x JO3 kkg/yr
Ul
1
t-J
SEDIMENT
Calc. 5-50
Obs. 20-200
( i ' 1 1 n r P n r r ;\ t i .•» n Q in m o / U u \
—ml - -
1
WATFR
Calc. .006-. 06
Obs. .001-. 05
( coiicen tra t ions in nig/1)
1
r>»
FISH
Calc. 4-38(30-day exposure)
Obs. 0.3-3
(conceiitiatioiia in mg/kg)
Mostly Exported
l-'fCIJRK 5-7 EXAMPLE QV RESULTS OP MODELm\ OF ENVIRONMENTAL DISTRIBUTION—COMPARISON OF CALC'W ATFD
AND OBSERVED LEVELS OF DI (2-ETHYi,llEXYL) I'HTHALATE IN AIR, SEDIMENT, WATER AND FISH
Source: Perwak, .1. et al. An exposure and risk assessment fur phthalate esters. Final. Draft Report
Hl'A Contracts 68-01-3857, 5949. Washington, DC: Monitoring and Data Support Division
Oil ice of Water Regulations and Standards, U.S. EnvironmentaJ Protection Agency; 1981.'
-------
TABLE 5-1. EXAMPLE OF RESULTS OF MODELING OF ENVIRONMENTAL
DISTRIBUTION—COMPARISON OF RESULTS FROM MACKAY' S
EQUILIBRIUM MODEL AND EXAMS FOR 1,2-DICHLOROBENZENE
IN A POND SYSTEM
EXAMS Results
(Pond, 24 kg/day loading
370 kg steady state accumulation)
Maximum Concentrations
Water
Water Biota
Sediment Biota
Sediment
3.0 mg/1
630 mg/kg
610 mg/kg
460 rug/kg
Mackay Results
(370 kg in system)
Water
Water Biota
Sediment Biota
Sediment
Concentrations
0.0559 mg/1
18 mg/kg
12 mg/kg
35 mg/kg dry
weight
Accumulation
% in Water
% in Sediment
16.22
83.78
Percent of Chemical per Compartment
Z in Water3 0.30
% in Sediment 64.4
:Part of the initial aquatic load has been removed by volatilization.
Source: Harris, J. et al. An exposure and risk assessment for dichloro-
benzenes. Final Draft Report. Contracts EPA 68-01-5949, 6017.
Washington, DC: Monitoring and Data Support Division, Office of
water Regulations and Standards, U.S. Environmental Protection
Agency; 1981.
5-23
-------
TABLE 5-2. EXAMPLE OF RESULTS OF MODELING OF ENVIRONMENTAL DISTRIBUTION-
EXAMS OUTPUT FOR 1,2-DICHLOROBENZENES
a. Steady-State Concentrations of 1.2-dichlorobenzene in various generalized aquatic
systems resulting from continuous discharge at a rate of 1.0 kg/hour3
Maximum Concentrations
Oligotrophic
Lake
River
Loading
(kg/hr).
1.0
1.0
1.0
Water
Dissolved
(mg/1)
3.0
0.15
0.00099
Water
Total
3.0
0.15
O.C009
Maximum in
Sediment
Deposits
(rag/kg)
460
0.73
9 0.024
'Plankton
(ug/g)
630
30
0.21
Total
Steady-State
Benthos Accumulation
(Ug/g) (kg)
610 370
3.3 410
o.nis i i
Total
Daily
Load
(kg/dav
24
24
24
b. The fate of 1,2-dichlorobenzone in various generalized aquatic systems'
Percent Distribution Percent Lost by Various Processes
Oligotrophic
Lake
River
Transformed
Residing in Residing in Transformed by Lost Time for
Water at Sediment at by Chemical Biological by Other System Se!
Steady-State Steady-State Processes _ Processes Volatilized Processes^ Purificnri
16.22
75.52
83.73
1.39
24.48
0.0
0.0
0.0
0.05
0.0
0.0
91.91
94.64
1.44
3.05
5.36
98.36
282.3
13.19 d;
data simulated by the EXAMS (U.S. EPA-!,ERL, Athens, Ga.) model (see text for further Information).
Including loss through physical transport heyond system boundaries.
Estimate for removal of ca. 97= of the toxicant accumulated in system. Estimated from the results of
the half-Lives for the toxicant in bottom sediment and water columns, with overall cleansing ttmo
weighted according :o the pollutant's initial distribution.
Source: Harris, J. e_t a_l. An exposure inc. risk assessment for dichlorobonzenes. Final Draft Report.
Contracts EPA 68-01-5949, 6017. Washington, DC: Monitoring ,ind Data Support Division. O
of Water Regulations and Standards., U.S. Environmental Protection Agency; 1981.
-------
REFERENCES
Harris, J.; Coons, S.; Byrne, M.; Fiksel, J.; Goyer, M.; Wagner, J.;
Wood, M.; Moss, K. An exposure and risk assessment for dichlorobenzenes
Final Draft Report. Contracts EPA 68-01-5949 and EPA 68-01-6017
Washington, DC: Monitoring and Data Support Division, Office of'water
Regulations and Standards, U.S. Environmental Protection Agency; 1981.
Lyman, W.J.; Reehl, W.F.; Rosenblatt, D.H. (eds.) Handbook of chemical
property estimation methods. Environmental behavior of organic compounds.
Boston, MA: McGraw Hill; 1982.
1979 Flndin§ fuSacit? feasible. Environ. Sci. Technol. 13:1218-
Miller, C. Exposure assessment modeling, A. state of the art review
Contract _ EPA PB 600/3-78-065. Athens, GA: Athens Research Laboratorv
U.S. Environmental Protection Agency; 1978.
Neely, W.B. A preliminary assessment of the environmental exposure to
be expected from the addition of a chemical to a simulated aquatic
ecosystem. Intern. J. Environ. Studies 13:101-108; 1979.
J ; Bysshe, 3.; Goyer, M. ; Nelken, L. ; Scow, K. ; Walker, P.-
Wallace, D. An exposure and risk assessment for copper. Final Draft
Report. Contract EPA 68-01-3857. Washington, DC: Monitoring and Data
Support Division, Office of Water Regulations and Standards, U.S
Environmental Protection Agency; 1980.
Perwak, J.; Goyer, M. ; Schimke, G. ; Eschenroeder , A.; Fiksel, J. ;
bcow, K.; Wallace, D. An exposure and risk assessment for phthala^
esters. Final Drart Report. Contracts EPA 68-01-3857, 5949. Washington
Support Division, Office of Water
and Standards, U.S. Environmental Protection Agency; 1981.
Scow, K.^ Goyer, M.; Perwak, J.; Payne, E.; Thomas, R.; Wallace D •
? i ' *;! W?°d' M' An exP°sure and ris'
-------
U.S.Environmental Protection Agency (U.S. EPA). Environmental modeling
catalogue. Contract EPA 68-01-4723. Washington, DC: Management
Intormation and Data Systems Division, U.S. Environmental Protection
Agency; 1979.
Wolfe, N.L.; Steen, W.C.; Burns, L.A. Use of linear free energv rela-
tionships and an evaluation model for phthalate transport and fate esti-
U'S' Environmental
5-25
-------
6.0 MONITORING DATA AND ENVIRONMENTAL DISTRIBUTION
6.1 INTRODUCTION
Monitoring data, as used in the context of exposure analysis, can be
defined as data on the concentration of toxic pollutants in the environ-
ment. Ideally monitoring data indicate ambient concentrations over wide
geographic areas and different periods of time. They are sometimes supple-
mented by data measured in field studies, which can be used to indicate
local conditions or special situations.
Many different kinds of monitoring data may be useful in exposure
analysis. Both well-mixed equilibrium concentrations and unusually high,
temporary spill or discharge concentrations are of interest. The first
represents the long-term condition to which humans and other organisms
are typically exposed. The second, although a short-lived condition,
could potentially have acute adverse effects on exposed organisms. In
addition, for certain readily transformed or transferred chemicals, only
the second type of data will exist. Depending upon the environmental
loading and fate characteristics of the pollutant, emphasis may be placed
on finding monitoring data pertaining to either long-term or transient
conditions.
Traditionally, monitoring data have been considered as measured
concentrations reflecting ambienc concentrations in surface water,
sediment, foodstuffs, etc. However, for purposes of estimating ex-
posure to pollutants, information regarding concentrations in more
varied media are of interest.
Water—surface water (fresh and salt), ground water, raw and finished
drinking water, precipitation, POTW influent and effluent, landfill
leachate, industrial effluent, urban and rural runoff, etc.
Air—vapors, aerosols, and particulates in ambient air, industrial
emissions, automobile emissions, workplace air environment, etc.
Soils and sediments—dust; surface and subsurface soils; bedrock:
estuary, lake, river and stream sediments; etc.
Biota—soil and aquatic microorganisms, vertebrates, invertebrates,
mammals, birds, organisms in a foodchain; humans, including whole
body or organ tissues such as human milk, human adipose tissue, urine
and blood serum.
Food—milk, meat, dairy products, grain, vegetables, fish, animal
feed, etc. (both in a natural and/or prepared state).
o-l
-------
Other--treated items such as preserved wood, painted objects, food
packaging, clothing, or any other product or item that may contain
the compound of interest.
Again, the particular media of interest depend upon the environ-
mental loading and pollutant fate characteristics. For example, a highly
volatile product may never be found at significant concentrations in soil
and water but at high levels in air. A persistent compound with low
water solubility may be detectable only in sediments and soil. Use and
disposal characteristics of the pollutant will also determine which media
to consider; for instance, some chemicals may be found only in the air
of certain working environments. Therefore, flexibility is required in
the methodology for analyzing monitoring data in order to allow emphasis
on the important environmental reservoirs and sinks for various pollutants
with a wide range of fate characteristics.
6.2 GOALS AND OBJECTIVES
The primary goal of the monitoring data review within an exposure
analysis is to develop, analyze, and present comprehensive data on the
geographic distribution of pollutant concentrations in various environ-
mental media, indicating trends or changes over time if possible. Spe-
cific objectives include:
(1) In the initial definition and focusing of risk assessments, an
analysis of monitoring data is used to characterize the behavior
of a pollutant in the environment; to determine whether local,
regional, or national risks are important; to identify the geo-
graphical areas of concern; and when combined with effects data,
to reveal the significance of the potential risks.
(2) In some circumstances, when monitoring data are sufficiently
extensive to be representative of typical environmental concen-
trations, they provide a description of environmental distribution.
(3) In the analysis of monitoring data, baseline levels can sometimes
be established (ambient conditions) for comparison with concen-
trations in polluted environments. In such an analysis, for
example, background concentrations near ore deposits may be found
to be equal to or greater than those found in some industrial areas.
(4) Monitoring data can be used to confirm materials balance and en-
vironmental fate analyses, to provide a credible basis for extra-
polating materials balance and fate considerations, and to provide
input data for large-scale modeling of environmental fate.
(5) Monitoring data can suagest important routes of exposure for
humans as well as other species, provide direct inputs to esti-
mates of exposure (e.g., concentration in foods for estimating
human exposure via ingesticn), and to help define the risk to
regional and otrier suboooulations.
-------
Though not all of these objectives will be met in each particular analysis,
they provide a framework to be used to the degree possible, depending on
the available data.
6.3 METHODS AND APPROACHES
The approach to monitoring data analysis consists of three basic
steps:
(1) identification and systematic collection of data;
(2) evaluation, analysis and presentation of data; and
(3) interpretation and use of data in exposure analyses.
At the start of a monitoring data analysis, materials balance studies
should first be reviewed briefly to help identify likely media for emphasis
in the systematic search for monitoring data. It will also be important
to define the boundaries of the search for monitoring data—the geographic
focus, depth and breadth—so that the necessary effort is devoted to this
portion of the risk analysis. Also, a desired format for the presentation
of data should be developed early in the work. The initial focusing step
in a risk analysis will aid this process.
Collection of data should be based upon a systematic literature
search. The U.S. Environmental Protection Agency, through its air and
water quality programs, provides a comprehensive source of monitoring data.
Particularly important for risk analyses will be the STORET system, NASQAN
(National Stream Quality Accounting Network), SAROAD (Storage and Retrieval
of Aerometric Data), NOMS (the National Organics Monitoring Survey), the
Pesticide Monitoring Program, the Air Quality Monitoring Program, the
Human Tissue Monitoring Program, etc. (see Chapter 10 for a listing).
The STORET system, maintained by the Office of Water Regulations and
Standards of the U.S. Environmental Protection Agency is a centralized
system for storage and retrieval of water quality data. The largest file
in STORET is the Water Quality File which contains data concerning 40
million observations at more than 200,000 monitoring stations in the U.S.
These data are of graat use in considering the distribution of a chemical
in the environment. Another STORET file that often contains pertinent in-
formation is the Fish Kill File, which provides detailed and summary data
on major pollution-caused fish kills dating from 1960. Thus, in addition
to the traditional methods of literature search, the U.S. Environmental
Protection Agency centers responsible for these monitoring systems should
be contacted in order to obtain the most up-to-date data, provided the
scope of the exposure assessment warrants.
In addition to the U.S. EPA, computerized data bases and publications
of other federal agencies should be consulted such as the U.S. Geological
Survey, Department of Interior, Department of Energy, Department of Health,
Education and Welfare, Consumer Product Safety Commission, the National
6-3
-------
Aeronautics and Space Administration, National Occartographic and Atmos-
pheric Administration, and the Corps of Engineers. Many of these agencies
and sources have compilations o£ literature data which may be useful.
In evaluating monitoring data for use in exposure assessments, a
series of questions should be posed:
(1) How, where, and when were the monitoring data obtained?
(2) Was the sampling process adequate to represent the environmental
compartment or subcompartment being monitored?
(3) Were the analytical methods used appropriate to the monitoring
problem and to the pollutant being measured?
(4) What were the sensitivity, reproducibility, and confidence of
the analytical results?
(5) How were the data aggregated and reported?
i
The answers to these questions are not always available, and differ
for every study and every pollutant. Frequently, monitoring data are not
complete because results are reported for samples taken from only a few
geographical locations. As a result, it is often difficult to determine
whether the monitoring data are applicable only to specific geographical
locations or whether they are representative of general levels in the U.S.
The numerical concentration values presented in monitoring data must be
used with caution because detection limits, accuracy, and precision of
the measurements are frequently not reported. The analytical methods used,
potential interferences, and details of the measurement approach (for
example, whether the measurement: represents the total metal concentration,
specific ionic species, or whether a specific chemical or family of
chemicals, e.g., phenols, are included). Frequently little informa-
tion is given on the seasonality or other temporal variations in the
measurements. Another problem associated with monitoring data is that
other parameters useful in interpreting the data, such as suspended solids,
pH, or the presence of other chemical species, may not be given. Perhaps
the most frustrating aspect of monitoring data is the lack of additional
information that helps the investigator determine whether or not the
monitoring data are sufficiently representative to be used in an exposure
assessment. When reporting monitoring data, it is essential to provide
complete references and give any additional information that was reported
in the original reference. Because of these limitations, monitoring data
do not always provide a clear ar.d accurate verification of real environ-
mental concentrations of pollutants and may, in some cases, yield no better
information than estimates obtained from fate and pathways analysis.
However, monitoring data represent the only evidence of actual ex-
posure, since many aspects of materials balance development and fate
analysis are highly speculative. Therefore, although there may be some
-------
uncertainty about values used, monitoring data should be used whenever
possible in an exposure assessment.
The presentation of the available data is a difficult problem in
some cases, and to an extent depends upon how the data will be used. The
data may be summarized for presentation as maps, charts, graphs, tables,
overlays, etc. Attention to presentation style is important, since this
often effects the conclusions that are drawn from the data. Uncertainty
in the data should be indicated in the presentation. Ranges, average
and median values, and concentration frequency distributions should be
presented, along with detection limits, wherever possible. Illustration
of data trends (such as decreasing water concentrations over 10 years)
provides useful information on the anticipated impact of increased pro-
duction or tighter environmental regulation on pollutant concentrations.
6.4 EXAMPLES OF MONITORING DATA
6.4.1 Copper and Silver
A tremendous amount of monitoring data has been collected for copper
in all media (Perwak et al. 1980). Copper levels in aquatic ecosystems
(water, sediment, fish) are available from the EPA STORET data base.
Figure 6-1 shows the distribution of total copper observations for the
U.S. from 1970 to 1979. Obviously, this type of a figure cannot be used
directly in an exposure analysis, but it does give some indication of the
range of total copper concentrations that are found in the U.S.
The monitoring data can also be aggregated by major river basins,
which represent large areas of the country. They can be depicted geo-
graphically as is shoxro in Figure 6-2 for silver (Scow e_t al. 1981) , or
displayed in tabular form. The copper data for 1970-1979 aggregated for
major river basins are shown in Table 6-1 for water and Table 6-2 for
sediment. By use of this technique, certain areas of the country with
high copper levels can be identified. However, aggregation of data ever
such large areas can provide misleading results. Therefore, monitoring data
from minor river basins were also examined, including data only for 1978.
Table 6-3 shows that numerous minor river basins have mean concentrations
greater than 50 'i-g/1 total copper and at least 10% of the observations
greater than 120 ug/1. However, only a few locations had median levels
of total copper greater than 60 ug/1. In addition, some of these minor
river basins were identified as having soft water (<50 mg/1 CaCo^ , a
condition that increases toxicity. These results suggested that within
these minor river basins showing high mean levels, most observations were
less than 60 ug/1 and a few were greater than 120 ug/1.
Data from individual monitoring stations in four areas with high
average copper concentrations were examined and compared with information
on specific sources of copper and evidence of actual impact on aquatic
biota. This analysis showed that high average copper concentrations
6-5
-------
41
30%
20%
55
c
re
C/3
c
3
<£
10%
<\
-
P77!
i
44
i
y.
10
1.0
-001 -01 -1 1 10 100
Concentration
1,000 10,000 100,000
FIGURE 6-1 EXAMPLE OF SURFACE WATER MONITORING DATA
DISTRIBUTION 3Y CONCENTRATION RANGES—COPPER
1970-1979
Source: Perwak, J. et al. An exposure and risk assessment
for copper. Final Draft Report. Contract EPA 68-
01-3857.^ Washington. DC: Office of Water Regulations
and Standards, U.S. Environmental Procecticn A.E
1980.
6-6
-------
Hi r-"^,: :• E™ i.i.;;;
•*$ L-^\J__ i iV/4';
v4 I ' i l~ *-»- Vi P'
Source:
FJCUHE 6-2 EXAMPLE OF (.'EOGRAPHIC DISTRIBUTION OF MONITORINfi DATA FOR SILVER
'iA^'7' r " exl^ostire anj risk asseusment for silver.
J/, JJiJ, and hi'A 68-01-6017. Wash in
-------
TABLE 6-1. EXAMPLE OF SURFACE WATER MONITORING DATA
DISTRIBUTION BY MAJOR RIVER BASINS—COPPER
Region
New England
Mid Atlantic
Southeast
Great Lakes
Ohio
Tennessee
Upper Mississippi
Souris and Red of North
Missouri
Arkansas and Red
Western Gulf
Hawaii
Rio Grande and Pecos
Upper Colorado
Lower Colorado
Great Basin
Pacific Northwest
California
Alaska
United States
.100-1
ug/1
3
1
2
1
1
<1
<1
<1
<1
1
3
1
9
-------
TABLE 6-2. EXAMPLE OF SEDIMENT MONITORING DATA DISTRI-
BUTION BY MAJOR RIVER BASINS—COPPER
Percentage of Observations
New England
Mid Atlantic
Southeast
Great Lakes
Ohio
Tennessee
Upper Mississippi
Lower Mississippi
Souris and Red of North
Missouri
Arkansas and Red
Western Gulf
Hawaii
Rio Grande and Pecos
Upper Colorado
Lower Colorado
Great Basin
Pacific Northwest
California
Alaska
United States 30 60
1-10
mg/k g
33
31
41
14
24
20
23
24
24
54
57
37
<1
16
53
40
14
18
10-100
rag /kg
50
53
56
65
73
69
58
72
41
39
43
59
33
84
46
40
81
75
100-1,000 1,000-10,000
15 1
15
-------
TABLE 6-3. EXAMPLE OF SURFACE WATER MONITORING DATA
FOR COPPER 3Y MINOR RIVER BASINS
River Basin Mean Cu >50X of Cu >102 of Cu
Major /Minor Name >
2/3
2/5
2/6
2/7
2/3
3/7
3/8
3/9
3/13
3/311
3/32
3/43
4/3
4/7
4/8
5/9
5/lfl
5/21
6/4
7/2
7/13
9/12
10/11
10/16
10/19
10/20
10/21
11/4
12/1
12/2
13/2
13/3
14/41
14/51
14/9
15/7
Delaware R. - Zone 1
Delaware R. - Schuylkill
Delaware R. - Zone 2
Delaware R. - Zone 3
Delaware R. - Zone 4
Yadkin & Pee Dee Rivers
Catawba - Wateref, etc. Res.
Edisto - Comb aha f R.
Savannah R.
Apalachicola R.
Choctawhatchee R.
Pearl R.
French Broad R.
Duck R.
Tennessee R.
Big Sandy R.
East Fork, White R.
Ohio R.
I. Erie Shore, Maumee R. to
Sandusky R.
Hudson Bay, Rainy River
Chicago Calumet R. - Des Plaines R.
Lower Missouri R. from Niobrara R.
Lower Mississippi R. - Yazoo R.
Lower Red R_ — below Denison
Atchafalaya R.
Calcasieu R.
Lower Mississippi R.
Gila R.
Sabine R.
Heches R.
Clark Fork - Pend Oreille R.
Spokane R.
Central CA Coastal
Santa Clara R.
Sacramento R.
Great Salt Lake
>502 of Hardr.
50 ug/L >60 ug/L >120 ug/L Measurements <
*
*
*
* *
* *
* *
* *
*
*
*
*
* *
*
*
*
* *
*
*
*
*
*
* *
* * *
*
*
* *
* *
* * *
* *
*
*
* *
* *
*
*
*
*
*
*
it
*
*
*
*
*
*
*
*
Fewer than 10 measurements at this station.
Source: Perwak, J. at al. An exposure and risk assessment for copper.
Final Draft Report. Contract EPA 63-01-3857. Washington, DC:
Office of Water Regulations and Standards. U.S. Environmental
Protection Agency; 1980
6-10
-------
reported for some river basins were the result of a small number of very
high concentrations. The analysis of the Sacramento River showed that
the mean from 26 to 27 stations for 1978 was less than 30 ug/1. However,
data for one station showed a mean level of 4585 ug/1 for that year.
Furthermore, dilution volume and the nature of the receiving water
(particularly pH and hardness) had to be considered in conjunction with
monitoring data in analyzing the risks of copper exposure for aquatic
biota since sensitive species are known to exist in locations with high
levels of copper.
6.4.2 Pentachlorophenol
Monitoring data for pentachlorophenol (PC?) are sparse and exist for
scattered media and sampling sites (Scow jat_al. 1980). In 1980, the
total number of observations of PCP surfa~ce "water concentrations in
STORE! was 80. Additional surface water data were limited to scattered
observations of low levels in a small number of geographic areas. The
compound was reported to be present in influents to POTWs, but also
appeared to be removed effectively by treatment. PC? had been detected
(again at low levels) in a drinking water survey. No data were available
concerning levels in air or soil.
Despite the fact that PCP did not appear to be found at high levels
in aquatic media, the compound was reportedly present in some food products
(Table 6-4) and also found commonly in human tissue and urine (Table 6-5) ,
even in persons not occupationally exposed. Thus non-aquatic exposure
routes had to be considered. The use of PCP as a pesticide results in
numerous opportunities for human exposure, particularly via inhalation.
Since no data were available on ambient atmospheric levels, fate models
had to be used in the risk analysis to predict concentrations for the
most likely conditions under which the general population might be ex-
posed (e.g., in the vicinity of preservative-treated wood or open burning
of such wood and downwind of cooling towers or wood treatment wastewater
evaporation ponds).
6.4.3 Dichloroethanes
As is the case for many organic compounds, monitoring data for the
dichloroethanes are extremely limited (Perwak _et_ _al. 1982). Very few data
exist showing levels in surface waters. In fact, only 10 observations above
the detection limit were found for 1,2-dichloroethane in the STORET data
base in 1980. However, several reports of ground water contamination were
found, as is shown in Table 6-6. In addition, air concentrations have be°n
reported in heavily trafficked areas, as well as in highy industrialized
areas (Table 6-7) .
These limited results suggest that exposure occurs in specific areas,
but that exposure to the general population is generally low. Obviously
the limited sampling of ground water and air does not provide a representa-
tive sample of widespread conditions. In this case, generalizations about
exposure in other areas have to be made with caution due to the limited
sampling and the nature of the exposure route.
6-11
-------
TABLE 6-4. EXAMPLES OF MONITORING DATA FOR FOOD
AND FEED—PENTACHLOROPHENOL
Concentration (yg/1 or ug/kg)
Sample
Dairy
Grain and Cereal
Leaf Vegetables
Root Vegetables
Garden fruits
Fruits
Sugars
Peanut butter
Bovine milk
Mean
0.5
1
T
1
T
T
6
18
ND
Range
Reference5'
10 Johnson and Manske (1977)'
10 -13 Johnson and Manske (1977)
13 Johnson and Manske (1977)
10 Johnson and Manske (1977)
10 Johnson and Manske (1977)
11 Johnson and Manske (1977)
10 - 40 Johnson and Manske (1977)
1.8 - 62 Heikes (1979)2
Lamparski et al. (1978)
ND - Not Detected
T = average below detection linit. Samples collected through U.S. in
FDA's Market Basket Study.
2
Market Basket Study - U.S. population.
Michigan dairy herds, detection level = 10 ug/1.
*
See source indicated below for references.
Source: Scow, K. et al. An exposure and risk assessment for pentachloroohenol
Final Draft Report. Contract EPA 68-01-3857. Washington, DC:
Office of Water Regulations and Standards, U.S. Environmental
Protection Agencv; 1980.
-------
TABLE 6-5. EXAMPLE OF MONITORING DATA FOR HUMAN
TISSUE AND URINE—PENTACHLOROPHENOL
Population and Sample
Exposed workers - urine (Japan)
Non-exposed workers - urine (Japan)
General population - urine (Florida) 4.9
Occupational workers - urine (Florida) 119.9
General Population - adipose tissue 26.3
Occupational population - urine (Hawaii) 1302
Non-occupational population - urine (Hawaii) 40
Occupational/non-occupational population 217
- urine
Combination of the above three groups 537
(Hawaii)
Occupational worker exposure - urine -
by wood preserving methods (Oregon)
Dip 2330
Spray 980
Pressure 1240
U.S. General Population - urine 6.3
Concentration
(yg/kg or-ug/1)
Mean
Range
1100-5910
10-50
2.2-11.2
22.2-270
12-52
3-35700
ND-1840
3-38642
120-9680
130-2580
170-5570
ND-193
Reference"
Bevenue (1967a)
Bevenue (1967a)
Cranraer (1970)
Cranmer (1970)
Shafik (1973)1
Bevenue (1967b)~
Bevenue (1967b)
Bevenue (1967b)
ND-38642 Bevenue (1967b)
Arsenault (1976)
Kutz (1978)'
ND - Not Detected
1 Detection limit - 5 ug/kg.
Detection limit = 3 ug/1.
Detection limit - 5 ug/1.
*Detection limit « 5-30 yg/1.
"See source identified below for reference.
Source: Scow, K. ejt al. An exposure and risk assessment for penta-
chlorophenol. Final Dcaft Report. Contract EPA bS-Ol-jSS/,
Washington, DC: Office of Water Regulations and Standards,
U.S. Environmental Protection Agency; 1980.
6-13
-------
TABLE 6-6. EXAMPLE OP CROUND WATER MONITORING DATA FOR DIC1ILOROETHANES
% Positive
Compound _ No. Statea Tea ted No. Wei Is Tea led Samples Maximum
1,1-dlcliloroethane 9 785 18 11,'J30
1,2-dlchloroethane 12 1212 7 400
Source:
Perwalc, .). et a I. An exposure and risk assessment for dichloroethanes. Final Draft Report
Contract EPA 68-01-5949. Washington, DC: Office of Water Regulations and Standards,
U.S. Environmental Protect Jon Agency; 1982.
-------
I
!-•
Ol
TABUS 6-7. EXAMPLE OF MONITORING DATA FOR DlCHLOROETHANliS IN AMBIENT AIR
City No.
Niagara Falls, NY
Hahwjjy/Woodbridge,
Houndbrook, and
I'ussnic, NJ
fiat on Rouge, I.A
Houston, TX
aNot detected.
Trace.
o.
, , Concentiatit>»
ampled Ha,!aejtnfi/«3)
1,2-Dichloroethane
0/9
NI)a
2/8
Concentration
Range (ng/m3|
n\>
10/66
12/43
1/30
T-342
T-500
555
75/93
36/43
22/30
T-139,121
9-10,341
T-66,300
Source:
Perwak, J. et al. An exposure and risk assessment for dichloroethanes. Final Draft Report
Contact EPA 68-OL-5949. Washington, DC: Office of Water Regulations and Standards,
U.S. Environmental Protection Agency; 1982.
-------
REFERENCES
Perwak, J.; Bysshe, S.; Goyer, M, ; Nelken, L.; Scow, K.; Walker, ?.;
Wallace, D. An exposure and risk assessment for copper. Final Draft
Report. Contract EPA 68-01-3857. Washington, DC: Monitoring and Data
Support Division, Office of Water Regulations and Standards, U.S.
Environmental Protection Agency; 1980.
Perwak, J.; Byrne, M.; Goyer, M.; Lyman, W.; Nelken, L.; Scow, K.;
Wood, M.; Moss, K. An exposure and risk assessment for dichloroethanes.
Final Draft Report. Contracts EPA 68-01-5949 and EPA 68-01-6017.
Washington, DC: Monitoring and Data Support Division, Office of Water
Regulations and Standards, U.S. Environmental Protection Agency; 1982.
Scow, K.; Goyer, M.; Perwak, J.; Payne, E.; Thomas, R.; Wallace, D.;
Walker, P.; Wood, M. An exposure and risk assessment for pentachloro-
phenol. Final1Draft Report. Contract EPA 68-01-3857. Washington, DC:
Monitoring and Data Support Division, Office of Water Regulations and
Standards, U.S. Environmental Protection Agency; 1980.
Scow, K.; Goyer, M.; Nelken, L.; Payne, E.; Saterson; K.; Walker, P.;
Wood, M,; Cruse, P.; Moss, K. An exposure and risk assessmenc for silver.
Final Draft Report. Contracts EPA 68-01-3857, 5949 and EPA 68-01-6017.
Washington, DC: Monitoring and Data Support Division, Office of Water
Regulations and Standards, U.S. Environmental Protection Agency; 1981.
6-16
-------
.0 HUMAN EXPOSURE AND EFFECTS
7.1 INTRODUCTION
The consideration of human exposure to toxic pollutants and the re-
sultant effects is critical to 3. risk analysis or assessment. In the
past, risk assessments have commonly considered only human health effects,
often focusing on studies with laboratory animals and the extrapolation
of animal data to humans. As indicated earlier, the integrated
risk analysis approach described in this report considers human exposure
and effects a vital element, but not the sole element, of a comprehensive
risk analysis. In some respects, the human exposure and effects section
represents the culmination of the use of materials balance and environ-
mental fate analysis, since these efforts are often needed to estimate
exposure of humans to pollutants.
The exposure of humans to pollutants and the potential effects of
this exposure should be considered simultaneously. The rationale for this
is straightforward—unless an individual or groups of individuals are ex-
posed co a pollutant, they are not at risk of experiencing adverse effects,
even if the pollutant is thought to be capable of inducing serious effects.
Similarly, a pollutant known to produce no significant effects on humans
probably presents no substantial risk to humans even though there may be
widespread exposure. Thus, the risk to various populations and subpopula-
tions depends upon the combination of exposure of those populations to a
pollutant and the related effects of the pollutant.
As was the case with other parts of risk analysis, the comprehensive-
ness of the exposure and effects analysis is determined by the quantity
and quality of available data. In general, even for pollutants that have
long been recognized as toxicants, data on human effects, animal studies
or epiderniological studies, are expected to be more readily available
than are data on exposure. For recently identified toxicants, both effect?
and exposure data are likely to be unavailable, and extrapolations or esti-
mates based upon other pollutants may be necessary. Thus it is extremely
rare that both the potential for effects and exposure are thoroughly docu-
mented.
The general questions that need to be addressed in examining the
available data on exposure and effects are as follows:
(1) Is there evidence of actual exposure, i.e., monitoring data?
(2) Are the data available to estimate exposures of the general
population?
(3) Do the data indicate the existence of subpopulations receiving
higher exposures than the general population?
7-1
-------
(4) Are there documented haman effects (tests, accidental exposures,
occupational health studies) or must extrapolation from labora-
tory animal studies be made?
(5) Are there sufficient multi-species animal tests to permit
reliable extrapolation of the results to humans?
(6) Can evidence of exposure and adverse effects on humans be
validated from epidemiological studies and extrapolated to
other exposure situations?
(7) Are there significant differences in human effects for differ-
ent subpopulations?
In addressing the exposure of humans, one should bear in mind the
initial sources of the pollutant and the fate and transport mechanisms
that determine the magnitudes and routes of exposure,, Only when exposures
are related back to pollutant sources will it be possible to consider the
alternative actions—regulatory and control—that could reduce the poten-
tial or actual exposure. Therefore, all possible exposure pathways and
all of the environmental media responsible for the exposure should be
carefully delineated. Both occupational and general exposure should be
considered, bearing in mind exposure routes of inhalation, ingestion, and
dermal contact. Specific subpopulations with higher than average exposure
should be identified—these subpopulations may be delineated by geography,
age, sex, occupation, food consumption patterns, activity patterns, etc.
Identifying exposures in this manner requires heavy reliance on the mate-
rials balance and environmental fate portions of the risk analysis, since
these elements may be the basis for estimating environmental concentra-
tions at various locations where humans can be exposed, especially if no
monitoring data are available.
To some extent, the exposure analysis should reflect the nature of
effects. For example, if a pollutant is well studied and has been shown
to induce effects in laboratory animals at relatively high exposure
levels, worst case scenarios can be constructed for exposure in order to
differentiate the low degree of risk at more realistic exposure levels.
Thus, the efforts devoted to identifying and quantifying exposures of sub-
populations might be reduced.
In evaluating the effects cf pollutants on humans, one should con-
sider chronic functional disorders of various organ systems, as well as
the more often evaluated effects such as carcinogenicity, mutagenicity,
and teratogenicity. Chronic effects need to be emphasized since environ-
mental exposures for most chemicals (except perhaps those in the workplace
or resulting from accidental releases of chemicals) occur over a long
period of time, often at low exposure levels. To some extent, the human
effects portion of an integrated risk analysis can be performed indepen-
dently of other portions, since it depends upon the results of detailed
laboratory investigations or epidemiological studies rather than on
estimates of environmental loadings or pathways.
-------
7.2 GOALS AND OBJECTIVES
7.2.1 Human Exposure Analysis
The goal of human exposure analysis is to identify and quantify the
exposure of the general population and selected subpopulation groups to
a pollutant or family of pollutants. Ideally the specific objectives
include:
(1) Determination of the exposure of the general population to the
pollutant. The general population is meant to represent the
"typical" exposure, if such a population group can be defined
for a given pollutant. The following parameters must be identi-
fied :
• the source of the pollutant resulting in the exposure;
• the routes of exposure—e.g., ingestion inhalation and/or
dermal contact;
• the duration and frequency of exposure—e.g., continuous,
1 hour per week, 1 hour per day, etc.;
• the amount or extent of exposure—e.g., the consumption
as a function of respiratory flow, amount absorbed! etc.;
• the size of the population exposed.
(2) Determination of the exposure of the work force to the pollutant
in terms of:
• occupations in which exposure is encountered, the geo-
graphical locations and/or types of facilities and opera-
tions;
• the numbers of workers exposed and their characteristics
age. sex, etc. ;
• the source of the pollutant, the route of exposure, the
duration and frequency of exposure, and the dose or dose
rare as indicated above;
(3) Identification of specific subpopulation groups that experience
a higher exposure to the pollutant than the "typical" person.
These subpopulations may be identified by geographic location,
size, age, sex, dietary or activity patterns." The parameters'
of such exposure would be the same as those indicated above.
7-3
-------
7.2.2 Human Effects Analysis
The goal of human effects analysis is to identify and characterize
the health effects in humans that may occur as a result of exposure to
a pollutant. More specific objectives include:
(1) Examination of the distribution, metabolism, bioaccumulation,
and excretion of pollutants in humans and laboratory animals
in order to identify target organs or systems. In addition, it
is desirable to identify the underlying mechanisms responsible
for the effects of pollutants in humans and the relationships
between exposure level (dose) and response in various species.
(2) Determination of the acute and chronic health effects on humans
expected and/or observed to occur from occupational or accidental
exposures and the exposure pathways and levels that result in
these effects.
(3) Determination of the known acute and chronic health effects of
pollutants on humans on the basis of epidemiological studies
and the exposure pathways and levels that result in these
effects.
(4) Consideration of the acute and chronic health effects that may
be expected to occur from exposure to pollutants, based upon
review of laboratory animal studies, in vitro and in vivo
studies with mammals, test organisms, tissues, cell cultures,
or other biota. Extrapolation of the results to humans may be
possible in some cases.
(5) Estimation of the "no-effect" levels of the pollutant for various
exposure pathways, based on animal data or human data, when
available.
The information obtained in the effects analysis should ultimately be
presented in a form that can be combined with exposure analysis for tha
purposes of considering the risk to the general population or specific
subpopulations associated with the pollutant.
7.3 APPROACHES AND METHODS
7.3.1 Exposure Analysis
7.3.1.1 General Approach
Identifying and quantifying the exposure of the general population
and the subpopulation groups is a difficult task, complicated by un-
certainties and lack of data, ar.d requires numerous assumptions and new
and often unproven estimation techniques. Hoxvever, Ln order to estimate
the range of risks presented by a pollutant, some "informed" estimate
7-4
-------
of exposure must be made and this requires taking a systematic and compre-
hensive approach to analyzing the best available data.
The exposure analysis builds upon concepts and data from the materials
balance, monitoring data, and environmental fate analysis. The basic
steps in exposure analysis are as follows:
(1) Identify, as comprehensively as possible, all potential
sources of exposure of the human population to a chemical.
In this context, "sources" can signify environmental media,
human activities, or consumer products.
(2) For each source, identify the route of exposure associated
with the source, e.g., inhalation, dermal contact, ingestion.
(3) For each source and route, identify key subpopulations based
upon demographic/geographic characteristics that are expected
to affect exposures.
(4) For each specific population group (e.g., general population;
work force; specific subpopulation characterized by age, sex,
type of activity, location, etc.) and for all possible routes
and sources of exposure for each group, attempt to quantify
the exposure as an average daily uptake or some other parameter
that may be related to effects levels and the numbers of per-
sons exposed.
Arraying data and information on an exposure matrix such as the one
in Table 7-1 is a convenient way to organize this effort. Beginning at
the left-hand side with the general population's exposure through the
three main exposure routes, the matrix shows in the columns to the right
the steps taken to identify exposure routes and to characterize, first
qualitatively and then increasingly quantitatively, the exposure situa-
tion and the exposure level. When data permit, an attempt should be
made to estimate the amount of the pollutant intake actually absorbed
and to estimate as precisely as possible the size of each population or
subpopulation. Depending upon the chemical, occupational exposures may
need to be considered in the same manner. Special exposure situations
and scenarios may be identified in the materials balance, environmental
distribution, or fate analysis because of characteristics of sources or
environmental releases, geographic considerations arising from volume of
releases or intensity of sources, or unusual use situations. Often
these scenarios are a further refinement of the more generalized exposure
routes and need to be considered as separate exposure routes.
The subsequent discussion considers these steps in assessing ex-
posures, first those leading to the identification of exposure routes,
and then methods for estimating exposure levels for the general popula-
tion and subpopulations.
7-5
-------
TABU; i-\. EXPOSURE MATRIX
t'opul a t-i on
Central
Route
IngestIon
Inhalation
Uermal Absorption
Oceupat lonal Ingest Ion
Inhalation
Dermal Absorption
Special Situations or Scenarios:
spl 11s
use of special products
II vi! near disposal situ or oouicfc
Subpopulation/Associated
Source
Drinking water
typical
maximum
Food
typical
max Ituuin
Urban—
typical
maximum
Rural—
typical
maximum
Water—
typical
maximum
Concentration Exposure Exposure
in Medium
Constant
Adult —
2 liter per
day
Children—
1 liter per
day
child
4 m-Vday
adult
20 m /day
Exposure
Duration/ Calculated Absorbed Size of
Frequency Intake Dose Population
-------
7.3.1.2 Sources of Exposure, Exposure Routes, and Subpopulation Groups
Sources of exposure include media, products, or activities that
result in human exposure. This concept can best be explained by example.
Consider a chemical that is used in household detergent. Direct exposure
could result from contact with, or inhalation (perhaps even ingestion)
of, the product itself. Indirect exposure could result from contact
with the water solution in which the detergent is used, by contact with
the residual detergent on the clothing or material washed in the deter-
gent, by inhalation of vapor from the mixture, or by ingestion of the
residual detergent from food placed on dishes washed with the detergent.
Thus, direct and indirect use of the chemical or pollutant must be
considered along with the routes associated with exposure to the general
population. In identifying these exposures, several general exposure
sources—related to environmental media—must be considered: ambient
water, drinking water, ambient air, and food. Pollutants that nay exist
in these media may not be attributable to specific sources, but rather
to an aggregation of "sources," which yields a distribution of the chemi-
cal in the environment. The approaches and methods used in environmental
pathways and monitoring are, in fact, designed to describe or develop
this "ambient" media distribution. The source/exposure combinations in-
clude the background level of exposure for the general population in
addition to exposure of subpopulations in specific areas or engaged in
specific activities. For example, although an average general exposure
resulting from ingestion of food containing a pollutant might be develop-
ed from average diet considerations, exposure of special subpopulations
who eat large amounts of meat, freshwater fish, milk, etc., must be con-
sidered.
The identification of subpopulations should be approached in several
different ways in order to ensure a thorough examination of exposure. In
the discussion of the materials balance analysis, activities such as ex--
traction, refining, manufacture, transportation, distribution, storage,
use, and disposal were defined; each of these has the potential for re-
leasing the chemical to the environment or perhaps exposing persons direct-
ly. As an example, disposal operations, both of products and "in-plant"
materials, must be examined for the variety of exposures and routes.
Exposure might result from material disposed in a chemical waste facility,
perhaps indirectly through the ambient air environment of the site or
surrounding public water supplies, with possible exposure routes including
inhalation, contact, or ingestion. Another source may be the "municipal
dump" or transfer station at which exposure of the public could result
through contact with empty containers (with residual chemicals), or by
inhalation of dusts or particulates.
Thus, all of the steps in the life cycle of the pollutant should be
considered in order to determine the potential for human exposure. The
purpose of reviewing these steps is to tie exposures to specific sources
of the pollutant and to define subpopulations who sustain exposure
7-7
-------
levels greater than those of the general population. These sufapopula-
tions may be subject to occupational exposures, may live, work in, or
frequent areas of pollutant sources, or obtain drinking water from
supplies contaminated by pollutant sources.
Both the consideration of ambient pollutant levels to which the
general population may be exposed and the review of pollutant sources
are required for identifying subpopulations exposed. Identifying the
potential exposure of specific subpopulation groups with unusual and/or
narrowly defined characteristics is a difficult task and requires care-
ful consideration. Furthermore, characterizing the populations and ex-
posures quantitatively in subsequent steps of the exposure analysis is
often not possible because data are lacking on the size of the popula-
tion or the exposure level. Nevertheless, it is important to attempt to
identify these subpopulations and to estimate the range of possible
exposures, so that the range of risks (exposure combined with effects)
can be estimated. Furthermore,, differentiating the risk of exposure to
subpopulations from those of the "average" population may identify the
types of control strategies needed to reduce overall exposure and risk
associated with the pollutant.
The complexity of the sources, exposure routes, and subpopulation
groups and the effort devoted to identifying them will vary with each
exposure/risk assessment, and will depend upon the pollutant in question
and the purpose of the assessment. In some cases, it may be sufficient
to consider only the "workplace" exposure, the general population exposure,
or exposure of a single subpopulation group; in others, it may be
necessary to identify sources, routes, and groups to the fullest extent
possible. In determining what is reasonable and appropriate in each
case, one should bear in mind several key points:
(1) that the risk will be a function of both exposure and the
effects and, therefore, in-depth analysis of exposure may not
be warranted if human, health effects are not of concern;
(2) that the effort to quantify exposure with precision may be in
vain if health effects are not well established or the back-up
data for exposure lack precision; and
(3) that the effort should be focused on the combination of sources,
routes, and population groups that have the potential for highest
total exposure.
7-3.1.3 Exposure Levels From Major Exposure Routes
After exposure sources, exposure routes and population subgroups
have been identified, the next step is to characterize the exposures
quantitatively for each source-route-population group combination. For
/ -o
-------
simplicity, this process can best be described by consideration of the
major exposure routes—inhalation, ingestion, and dermal contact—as they
apply to the general population, workplace population, and special sub-
population groups.
Inhalation
Exposure of individuals by inhalation can often be estimated in a
straightforward way. Data are available on the respiratory rate and
volume for individuals as a function of activity level (see Table 7-2).
Once these parameters have been established, ambient concentrations need
to be established. Usually, monitoring data do not allow the computation
of a statistically meaningful mean or median that would describe average
exposure to the U.S. population. If data were adequate, however, such a
value could be used. Generally, it is more useful to consider the data
available, their geographical and source-related representation, and
choose a "typical value." Although this method requires judgment, it
can provide a more meaningful value for typical exposure.
In addition to the typical exposure, or exposure to the general pop-
ulation, a maximum exposure should be established. If a statistical treat-
ment is possible, the 95th percentile, or a similar value, may be chosen.
Otherwise, the data must be evaluated to determine what this value might
be.
If monitoring data do not exist, estimates based upon anticipated
release rates and simple air models will be required. Depending upon
the materials balance and fate studies, estimates of ambient concentra-
tions may be on a national, regional or more localized basis. Similarly,
OSHA, NIOSH. cr other agencies may have available monitoring data or
methods of estimating concentrations in the workplace.
Determination of atmospheric concentrations of a pollutant to which
special subpopulation groups are exposed may be difficult; however,
monitoring data may exist for selected materials and exposure situations—
for example, urban and rural environments, agricultural areas in which
pesticides have been applied, areas near production facilities, and other
industrial sources. Usually, however, air concentrations of pollutants
associated with special exposure sitautions will have to be estimated.
These estimates would normally be accomplished in environmental fate and
pathway analysis and would be based upon the specific process or activity,
quantity of pollutant, its chemical and physical characteristics, and
environmental factors such as wind, rain, temperature, etc. Persons
located near smelter operations, cooling towers, waste disposal sites,
or commercial cleaning facilities are examples of special groups for
which exposures may need to be evaluated.
Another source of inhalation exposure that may be important in some
situations is inhalation of water vapor or fog (mist, droplets), which
has evolved from a water stream containing a pollutant. For these
situations, estimation of the concentration of pollutant vaporized inco
7-9
-------
TABLE 7-2. RESPIRATORY VOLUMES FOR HUMANS ENGAGED
IN VARIOUS ACTIVITIES
Time Reference and
Activity
Per minute:
Resting
Light Activity
Per day:
8 hours of working
"light activity"
8 hours of nonoccupa-
tional activity
8 hours of resting
Total 24 hr
Air Volume (liters)
Adult
man
Adult
woman
Child
(10 v)
Infant
(1 v)
Newborn
7.5
20.0
6.0
19.0
4.8
13.0
1.5
4.2
0.5
1.5
9,600 9,100 6,240 2,500 90
(10 h) (1 h)
9,600 9,100 6,240
3,600 2,900 2,300 1,300 690
(14 h) (23 h)
2.3xl04 2.1xl04 l.SxlO4 0.38xl04 O.OSxlO4
Source: International Commission on Radiological Protection (ICRP).
Report of the Task Group on Reference Man. New York, NY:
Pergamon Press; adopted October 197^.
-------
the air space above the water source, or the pollutant concentration in
the mist or fog (suspended water droplets), must be estimated. Again
physical/chemical properties of the pollutant, concentrations of pollu-
tant in the original water stream, and environmental parameters will be
important in these estimates. Approaches to making these estimates have
been developed by the U.S. EPA (Adamson e_t. al_. 1979).
Once the air concentrations have been established to the extent
possible for each situation, this information can be combined with the
appropriate respiration rate to determine the estimated exposure level.
The general methods will yield estimates of the quantity and rate of
pollutant inhaled by various population groups, e.g., g/day, mg/hr, etc.
Consideration of the source, route and characteristics of the exposure
will determine whether the exposure is intermittent or continuous, short-
er long-term, and whether it is a one-time exposure or an average intake
over some time period.
The procedure described above considers potential exposure to a
pollutant. However, much of the pollutant inhaled may not be absorbed
into the blood stream. Therefore, before exposures can be compared with
effects levels and the risks presented by these exposures assessed, one
needs to know how much of the material that is inhaled is actually ab-
sorbed in humans as compared to laboratory animals. An evaluation of
rates of absorption and metabolic pathways is conducted in conjunction
with the human effects analysis (see Section 7.3). Often, though, the
available data are not sufficient to indicate what portion of the poten-
tial exposure is actually available to the body. In these cases, it is
necessary to assume, as the worst case, total absorption.
An example of inhalation exposure estimates is provided in Table
7-3, which gives ranges of exposure levels for trichloroethylene in
different environmental scenarios (Thomas et_ al. 1981). The atmospheric
concentrations are maximum reported values in the vicinity of the two
major sources of atmospheric releases (TCE manufacturing facilities and
degreasing sites) and reported ambient levels for other areas. A total
daily intake has been estimated for each of these exposure situations
on the basis of estimated durations of inhalation exposure and standard
respiratory volumes for humans (in Table 7-2).
Ingestion of Food and Drinking Water
The most widespread exposure to pollutants for the largest number
of people will probably occur through the ingestion of food and drink-
ing water. As a result, it will be important to consider each of these
ingestion routes carefully and assess exposure to the general population
ar.d specific subpopulation groups. In general, the exposure to the work-
place population from food and drinking water will be similar to that of
the general population so that this subpopulation does not need to be
considered separately for this exposure route.
7-11
-------
TABLE 7-3. EXAMPLE OF ESTMATED INHALATION EXPOSURE TO T1UCULOROETHYLENE
. ion
'
Maximum Observed
Concentration
H/m )
Weekday Dur-
ation of
(hrs/tlav)
Nea r fianufac! uri n; S i I .
Urban - Day (ne.;r i
- Nij->bl (Havoiin.-, N.l)
llnial - Day (luar I'l.iiiul a." t urei )
-- Nijihl (lal Jedej-a Nat. Forest)
Near De}« leas in)- Sites
I'rban - Day (Ai rural i Factory)
- Kij-.ht (llayi.inu-, N.I)
Kural - |i iy (Ail craft Factory)
- Nij-.lit 0'aUede.ya Nat. Forest)
Low Arnl.ieui - Rural (Talledcya Nat. Forest) or
Urban (East Coast)
Estimated
Total
J440
47
.1440
3
23.
47
23 S
3
3
8
16
8
J6
8
16
8
16
24
14
0.6
14
0.04
2.3
0.6
2. J
0.04
0.06
I 4 .04
.. (Ainl.ieiH Uackj-rouiuJ)
0. ()()()()
Concentration estimates are taken from Table 4-5 in source cited below.
Weekend exposures will be 24 hr/day at night time levels. Hence, these values provide an upper-
bound estimate daily on exposure levels.
m.u,/ BaaUd °n rca'llralil>n °f 1'2 ll)3/^- (awake), 0.4 ,n3/br (sJeeping), about 20 fl,3/day
(K.HP 1975). (See citation below.)
Source: Thomas, R. et al. An exposure and risk assessment for tricbloroethyleue. Final Draft Report.
Contract EPA 68-01-5(J49. Washington. Oil: M.m i I or i nt- ;md Ibir;i Sisnnnrl niui^i^,-. nfl !,... f^f
-------
Food
The ideal data for estimating human exposure through ingestion of
food are residues in various cooked and/or processed foods included in
the American diet, and food consumption patterns of the general popula-
tion and subpopulation groups. After the food consumption patterns
for various subpopulations have been identified, the food residue data
can be used to determine quantities of pollutants ingested.
The closest approximation to this ideal is probably the Total Diet
Studies conducted by FDA, Bureau of Foods (U.S. FDA 1977). In these sur-
veys, residues of certain chemicals in specified food groups are analyzed.
Composite samples are used for these food groups; residues in individual
foods are not available. The Total Diet Studies consider primarily the
diet of a 16-19-year old male for calculations of dietary intake. The
pollutants considered by FDA are primarily metals and pesticides, and.
comprehensive information for many organic chemicals is lacking. Thus,
these data are useful primarily for initial estimates of the total amount of
a pollutant ingested by average populations consuming standard food groups.
They do not generally identify specific foods with contamination problems,
or population groups with high consumption.
In the absence of data from Total Diet Studies, or perhaps in addi-
tion to it, data on specific cooked or processed foods are desirable.
Information on residues in all food groups is rarely available. Thus for
any given pollutant, two options are available; the first is to determine
if the available information is an adequate representation of what might
be expected in the whole diet; or second, to make assumptions about what
might be expected in other foods based on the fata of similar chemicals.
For example, residue data are often available for fish as the only food
item. Since bioaccumulation may have occurred, it may be reasonable in
these cases to assume that fish constituce a major dietary exposure route
for humans. For example, in developing ambient water quality criteria,
the EPA assumes an average daily consumption of 6.5 g of fish, utilizes
bioaccumulation data to estimate the amount of pollutant contained in that
amount of fish, and combines this dietary intake with drinking water in-
take to establish a total intake of water-related pollutant (U.S. EPA 1980).
One should, however, consider whether additional major food exposure routes
exist other than fish. For example, there may be some specific studies
on residues of pollutants in meat and poultry, where contamination problems
are expected to occur. Sources such as these should be reviewed to deter-
mine the existence of data potentially applicable to the pollutant in question.
Rarely, however, is information available on residues in cooked or
processed food; residue studies in raw foods are much more common. Pollu-
tant concentrations from raw food cannot easily be extrapolated to those
in cooked foods. The U.S. EPA has been grappling with this problem in
setting tolerances for pesticides in food and has not yet determined a
satisfactory way to extrapolate from data concerning residues in raw foods.
At present, tolerances are set on the basis of raw food.
7-13
-------
In the absence of specific residue data for food, materials balance
and fate considerations may be able to provide some insight into the
probability that a pollutant will occur in food. Data are scarce on
such things as pollutant concentrations in soil, uptake rate from soil,
bioconcentration by plants and animals, and it is unlikely that accurate
concentration levels in raw foods could be estimated., Furthermore, models
would still be required to determine the changes in pollutant concentra-
tions in foods during processing and preparation. Thus, except for
some pesticides and metals, only scattered data on isolated raw or
prepared food items are likely to be available, and in some cases no data
will be found. Unless the food items for which data are available repre-
sent the major source of dietary exposure, even these scattered data will
be of limited use in an exposure or risk assessment.
Once the data to be used for food contamination have been identified
or estimated, consumption patterns must be established. The Agricultural
Research Service of the USDA conducted extensive food consumption survevs
in 1965 and 1978 (USDA 1972, 1980). These surveys include average con-'
sumption patterns by age groups and geographic regions. Other surveys
conducted by the USDA contain some information pertaining to food consump-
tion; data on consumption of fishery products and food fats and oils from
a survey by the Economics, Statistics, and Cooperatives Service (USDA
1976) are shown in Table 7-4, as an example of the type of data that are
available. While data from USDA surveys are extremely useful for estimat-
ing ingesticn exposures, they do not provide information on variation in
consumption by different age group or populations in various geographic
regions. In addition, they may not provide consumption data for a specific
food item of interest, for example, peanut butter. Thus in many cases,
assumptions must be made about food consumption in order to estimate
typical or maximum intake of a specific food item.
Despite the numerous limitations described above, the following
process may be followed in attempting to determine the exposure to pollu-
tants through food ingestion.
(1) Examine the USDA/FDA Market Basket/Total Diet Studies to deter-
mine if the pollutant has been measured as part of a specific
or general study. Use values obtained as an indication of
general exposure through food ingestion.
(2) Review any specific studies related to the pollutant in terms
of residues or tolerances and, on the basis of food consumption
and diet information, determine the exposure through ingestion
of those specific foods. Project, if possible, the ingestion
of the pollutant from similar foods and/or the total diet.
(3) Through literature research, analysis of monitoring data, en-
vironmental fate considerations, analogies to other pollutants
and products, or simple models, determine the concentration
(range of concentrations) of the pollutant in the food or food
7-14
-------
TABLE 7-4. PER CAPITA CONSUMPTION OF FISHERY PRODUCTS
AND FOOD FATS AND OILS IN THE U.S., 1976
Fish
Item
Fresh and Frozen
Fish
Shellfish
Total
Per Capita
consumption
(pounds in 1976)
5.5
2.6
8.1
Food Fats and Oils
Item
Table Spreads
Butter
Margarine
Total
Per Capita
consumption
(pounds in 1976)
4.4
12.5
16.9
Canned
Salmon
Sardines
Tuna
Shellfish
Other
Total
0.
0.
2.8
0.4
0.4
4.3
Cooking Fats
Lard
Shortening
Total
2.8
18.2
21.0
Cured
TOTAL ALL FISH
0.5
12.9
Edible weight
Source:
U.S. Department of Agriculture (USDA). Food consumption.
prices, expenditures. Agricultural Economic Report No. 133
Supplement for 1976. Washington, DC: USDA; 1976.
7-15
-------
group under consideration. (Often the many uncertainties in-
volved will make this very difficult if not impossible.) If
feasible, combine these estimates to obtain daily intake of
the pollutant for the various subpopulation groups.
(4) Through consideration of the materials balance of the pollutant
and uses of the products that contain the pollutant, develop a
list of specific activities or scenarios that could result in
localized occurrence of the pollutant in food: e.g., use as a
pesticide on selected products, as a preservative, in a food
packaging material; discharge to a freshwater stream from which
people catch and eat fish; movement through the foodchain to
mother's milk; processing into a one-of-a-kind food product,
etc. For each scenario, attempt to identify the subpopulation
group by age, sex, location, habits, etc., that may be exposed.
(5) Consider changes in concentrations or residues of pollutants
in food processing and preparation. Although it is not possible
to evaluate these changes thoroughly, they need to be addressed
at least qualitatively in the estimation of exposure, especially
in cases where levels of a pollutant in food may actually be
increased during processing (e.g., the addition of lead to foods
from lead solder in cans).
(6) Combine, wherever possible, data on average daily intake and
exposure for the general population with data for specific sub-
populations to establish ranges of human exposure.
In performing the last few steps given above, there are a number of
scenarios (activities) that are more or less routine for each pollutant
or product in which the pollutant may be a contaminant and each scenario
may represent an exposure situation that should be analyzed as a separate
exposure route:
(1) use as a pesticide or fertilizer,
(2) use as a food preservative or additive,
(3) use in food processing or preparation activity, including
equipment,
(4) use in a food container or packaging material,
(5) release into soil or water from which food or food crops are
grown,
(6) release in the vicinity of grazing or rangeland,
(7) use in an animal food or feed or packaging thereof,
7-16
-------
(8) use in water system for food washing or preparation,
(9) release into water systems in which fish, shellfish or other
wildlife live or feed, and
(10) use as a supplement in a poultry or livestock feed.
Through systematic consideration of each opportunity for exposure
the total exposure of humans to pollutants through food ingestion
may be estimated. Clearly this is an area of risk assessment that needs
considerable research, development and evaluation.
Table 7-5, 7-6, and 7-7 give examples of the results of analyses of
exposure from food ingestion. The data have been developed and presented
in different ways. In Table 7-5, the ingestion of di(Z-ethylhexyl)
phthalate was calculated on the basis of data on concentrations found in
various food items, and the levels of consumption of these items (Perwak
e_t al. 1981a) . Although these foods obviously do not represent a total
diet, it was felt that they were a close approximation of total dietary
exposure since high fat items were sampled in which phthalate esters
might be expected.
Table 7-6 shows the dietary intakes of copper reported in the litera-
ture (Perwak et_ al. 1980a) . In this case, a separate analysis was not
conducted for two reasons. First, as is evident from Table 7-6 a con-
siderable amount of work has been done in this area, and the results are
in agreement. Second, the low order of copper toxicity to humans suggested
that dietary intake would not be a significant source of risk. Thus, a
great degree of accuracy in estimating dietary intake was not required.
Table 7-7 shows the estimated dietary intake of mercury by a select
subpopulation, fish eaters (Perwak et al. 198Ib). Again, intakes were
calculated through use of consumption data and general residue data for the
same fish species. The results show that an increased consumption can
substantially increase intake over that of the population average, which
in this case was about 3 ug/day attributable to seafood.
It is important to point out that a relatively large amount of data
were available for analysis of the pollutants in the examples given above.
Since this is often not the case, the possibility for detailed considera-
tion is reduced. In some cases, quantitative estimation of dietary intake
is not feasible, although intake can often be compared qualitatively with
other exposure routes.
Drinking Water
Exposure of humans to pollutants through drinking water can vary
widely, even within a very localized area, depending on the water supply.
The ideal information for estimating exposure through drinking water would
include a distribution of concentrations of the chemical in drinking water.
-------
TABLE 7-5. EXAMPLE OF ESTIMATED 1NGESTION EXPOSURE OF DI(2-
ETHYLHEXYL) PHTHALATE VIA SELECTED FOOD ITEMS
Average Daily Intake (mg/dav)
Food Consumption3 (g/day) Average
baked beans
corn meal
canned corn
white bread
eggs
cereal
meat
margarine b
processed American
cheese °
milkc
fish
Total
7.0
9.6
7.1
12.0
43.5
37
210
15.5
13.3
230
21.4
trace
0.002
trace
0.01
0.004
0.01
0.13
0.03
0.02
0.04
0.004
0.25
Maximum
0.01
0.02
0.001
0.14
0.03
0.13
0.63
0.69
0.12
0.14
0.15
2.1
Please note that some of the categories of foods for consumption volumes
do not exactly match the categories of sampled food items in all cases.
For example, consumption data are used for all meat, bread rolls, and
biscuits; however, only certain food items within these general categories
were sampled for DEHP. No estimate of consumption was found for baked
beans, so 7.0 g/day was assumed.
Consumption of chese foods has been corrected for fat conten"-
margarine, 30% fat; cheese, 25% fat; and milk, 2% fat.
Source:
Perwak, J., e_t al. An exposure and risk assessment for phthalate
esters. Final Draf-: Report. Contract EPA 68-01-3857.
Washington, DC: Monitoring and Data Support Division. Office
of Water Regulations and Standards, U.S. Environmental
Protection Agency; 1981.
7-13
-------
TABLE 7-6. EXAMPLE OF INGESTION EXPOSURE ESTIMATES
FOR COPPER BASED ON TOTAL DIET STUDIES
Intake
(mg/day)
0.34
0.91
1.0
1.04
1.2
1.5
1.8-2.1
1.9
2.4
3.8
7.6
Type of Diet
self-selected
(24-hr)
self-selected
self-selected
self-selected
Non-institutional
diets
diets (no liver)
balance study
institutional
diet
self-selected
diet composites
diets (with liver)
Number of
Subjects
Reference
4 female White (1969)
1 female Tipton et_ al_. (1966)
11 male, Holden et al. (1979)
11 female
36 female Tipton e_t al. (1966)
12 female Guthrie and
Robinson (1977)
12 female Guthrie and
Robinson (1977)
11 female Robinson e_t al. (1973)
12 female Guthrie and
Robinson (1977)
12 female Guthrie (1973)
1 male Zook and Lehman (1965)
11 female Guthrie and
Robinson (1977)
See source indicated below for references.
Source:
Perwak, j. et_ al. An exposure and risk assessment for copper.
Final Draft Report. Contract EPA 68-01-5949. Washington DC-'
ilomtonng and Data Support Division, Office of Water Reo-ula-"
tions and Standards, U.S. Environmental Protection Agency- 1980
7-19
-------
TAIJLE 7-7. EXAMPLES OF 1NCESTION EXPOSURE ESTIMATES
OF MERCURY FOR A SPECIFIC SUBPOPULAT10H
Mercury
Upper Limit
Daily Intake
Serving/
Assumes a 0.5 Ng/g mercury action limit.
Concentration (Pg)a - 95%
Confidence
Limi ts
51.37
Person Sjiecies g/Serving
Person L Pike
Uass
Perch (marine)
Not identified
i
hO
o
Person 2 Pike
Bass
Perch (marine)
206
167
144
150
253
218
181
Month
15
3
2
1
19
4
2
Av£. Max.
0.0]
0.75
0.13
0.01
0.75
0.13
1
2
0
1
2
0
.7
.0
.59
.7
.0
.59
79.46
Maximum Jntake
141
a
222
Source: Perwak, .1. et al. An exposure and risk assessment for mercury. Final Draft Report. Contract
EPA 68-OJ-5949. Washington, DC: Monitoring and Data Support Division, Office of Water
Regulations and Standards, U.S. Environmental Protection Agency; 1981.
-------
comprised of data from enough locations to be representative, and the
corresponding numbers of persons exposed to each concentration range.
Unfortunately, this ideal situation rarely occurs.
Monitoring data for drinking water are more generally available
than data on pollutant levels in air or food, but do not provide a com-
prehensive view of the many waterborne pollutants that may be found in
water supplies throughout the U.S. The most extensive monitoring of
drinking water was conducted in 1970; the survey sample in this study
included 6595 water supplies in the U.S., including well water, ground
water, surface water and tap water (U.S. DREW 1970). However, the only
parameters considered were those regulated by the 1962 USPHS standards.
More recently, in 1974, EPA conducted the National Organics Reconnaissance
Survey (NORS) (Symons et_ al_. 1975); this study sampled 80 water supplies
in the U.S. for halogenated organics. In 1976, EPA conducted the National
Organics Monitoring Survey, which looked at levels of a large number of
organics in 112 locations in the U.S. (U.S. EPA 1978a). These data com-
prise a partial basis for assessment of national exposure. Since 1976,
a number of additional studies'have been conducted, usually in specific
locations or for specific water supplies.
The monitoring studies described above can sometimes provide data
to estimate the distribution of the chemical in drinking water over the
U.S. Populations can be associated with the water supplies sampled and
with ground water and surface water in general, but the extrapolation of
Che distribution to the total U.S. population is not generally possible
with the data available.
If sufficient data on the pollutant concentration in drinking water
are not available on a national scale, localized data may be used in one
of two ways. If data are available for a location where high concentra-
tions would be expected on the basis of materials balance" and environ-
mental fate considerations, the subpopulation of residents in the loca-
tion can be identified and their exposure estimated. In this case, no
estimate of exposure to other suhpo-pulations can be made. If the data
are not from a local "hot spot," they might be used to validate the re-
sults of model(s), which would then be used to estimate maximum concentra-
tions in drinking water in other locations. Modelling of pollutant fate
in surface water is more highly developed than for groundwater. Hence
this approach is more likely to be useful for estimating exposure via
drinking water from surface water supplies.
In many cases and for ;nany locations, however, monitoring data for
raw or treated drinking.water are unavailable or inadequate for purposes
of exposure assessment. For a worst case consideration, ambient concentra-
tion data (measured or estimated from materials balance and pathways
analysis) may be used directly for chemicals that would not be formed
during water treatment or encountered in the water distribution system.
If a more precise analysis is needed, losses or additions during water
treatment must also be considered. (See chapters on Monitoring and Fate.)
7-21
-------
The steps involved in estimating pollutant exposure of general and
specific subpopulations from drinking water include:
(1) Review appropriate national surveys, STORE! data, and EPA
regional data to develop appropriate national average values
and concentration ranges of the chemical and number of persons
exposed to the chemical, if possible.
(2) Consider local data from appropriate municipal water districts,
surveys, etc., to determine local concentration levels and to
generalize to national average levels if practicable.
(3) From materials balance and environmental fate considerations,
identify any localized areas and pathways that might result
in contamination of drinking water supplies. Through modeling
efforts, and in comparison with available monitoring data, deter-
mine whether these sources have led to contamination and at what
levels. To the degree possible extrapolate these conditions
to other locations and exposure levels.
(4) If no (or limited) data are available on drinking water, con-
sider ambient water monitoring data for the pollutant (surface,
ground, etc.) and investigate to what extent treatment would
remove the contaminant from the water in the water supply/
treatment process.
(5) From materials balance considerations, examine other unconven-
tional routes of entry of a pollutant into drinking water; for
example, from chemicals used in treatment, pipes/valves used
in distribution systems, etc. From monitoring data or simple
models, evaluate the concentrations that may result in drink-
ing water.
Once the concentrations of a pollutant in drinking water have been
determined for various exposure subpopulations, they must be combined
with an appropriate exposure constant in order to estimate the pollutant
intake. Although consumptions of 2 liters per day for adults and 1 liter
per day for children are commonly assumed in exposure calculations, con-
siderable variation exists in consumption. In cases in which ingestion
via drinking water is a major exposure route, it may be appropriate r.o
consider a range of consumption values in estimating exposures. In addi-
tion, the rate of absorption of the pollutant in the gastro-intestinal
tract must be considered for pollutants in drinking water in the same
manner as for pollutants in food.
Tables 7-8 and 7-9 illustrate some results of analyses of drinking
water exposure. Table 7-3 shows the drinking water exposures for 1,2-
dichloroethane,with associated populations (Perwak e_t al. 1982a) . In
this case, as is common for many organic chemicals, the reported values
of the monitoring data are near the detection limit of the analytical
7-22
-------
TABLE 7-8. EXAMPLE OF ESTIMATED EXPOSURES TO 1,2-DICHLOROETHANE
VIA DRINKING WATER INCLUDING POPULATION SIZE
Population
General Population
Surface Water
Ground Water
Estimated
Population
Size
5 million
5 million
Assumption
2 ygA, 2£/day
0.3 'Mg/i, 2£/day
Calculated
Exposure
(ug/day)
4
0.6
Isolated Sub-
Populations
Surface Water
Ground Water
maximum level of
4.8 Ug/£, 21/day
maximum level of
400 ug/i, 2£/day
9.6
800
Source: Perwak et_ al. An exposure and risk assessment for dichloro-
ethanes. Final Draft Report. Contract EPA 68-01-5949.
Washington, DC: Office of Water Regulations and Standards,
U.S. Environmental Protection Agency; 1982.
7-23
-------
TABLE 7-9. EXAMPLE OF MAXIMUM AND TYPICAL ESTIMATED EX-
POSURES TO TRIHALOMETHANES VIA DRINKING WATER
Daily exposure (ag/day)
Assuming Maximum Adult Assuming Reference
Intake^ and Maximum Intakeb and Media
Trihalomethane Concentration (aig/1) Concentration in Water Concentration in Wa
Median Maximum ~
Chloroform 0-059 0.540 1.2 o 1
Bromoform 0.004 0.280 0.6 0.007
Dibromochloromethane 0.004 0.290 0.6 0.007
3romodichloromethane 0.014 0.180 0.4 Q.02
a2.1S liter per day
1.65 liter per day
Source: Perwak, et al. An exposure and risk assessment for trihalo-
methanes. Final Draft Report. Contract EPA 68-01-5949.
Washington, DC: Office of Water Regulations and Standards,
U.S. Environmental Protection Agency; 1980.
7-24
-------
procedures; hence there is considerable uncertainty attached to the values
shown and the calculated exposures. In such a case, it may be desirable
to be conservative, that is, to overestimate, rather than underestimate
typical exposure levels. In this example the population sizes were esti-
mated by extrapolating the percentage of water supplies in which the
compound was detected in the sample to the percentage of the total U.S.
population exposed. This extrapolation does not incorporate many compli-
cating factors, and the distribution in size of water supplies is assumed
to be the same in the sample as in the U.S. Though this assumption may
be valid for surface water, it is probably invalid for groundwater supplies
in which sampling has been very limited.
Table 7-9 shows human exposures to trihalomethanes via drinking water,
as estimated fay use of maximum and median observed concentrations and con-
sumption levels (Perwakjet al. 1980b). This table indicates (as does
Table 7-8) that a wide range of exposures can occur. In general it is not
possible to determine the population distribution of exposure levels. At
best, usually a median, mean, or "typical" and a maximum exposure can be
estimated.
Dermal Absorption
Dermal absorption of a pollutant from ambient or treated water and/or
directly from the use of the chemical or product contaminated by the chem-
ical should be examined in an exposure assessment. The process of esti-
mating^ the '^average daily intake" of a pollutant by the dermal route is
slightly different than for other exposure routes; the concentration of
pollutant in the water or solutions, the nature of the chemical contacted.
the time of contact, and the area, location and integrity of the skin
exposed can all affect the uptake.
The first step is to examine the types of human activities in which
direct contact exposure to the pollutant can occur. In addition to work-
place exposures or contact with pollutants during manufacture, the exposure
potential of use situations must be examined carefully, e.g., exposure
resulting from: mixing or application of pesticide formulations; pollutant
containment in paint, glue, stain, or similar materials; use of cosmetics,
gasoline or cleaning solvents; polymers, films or fibers in apparel or
other products, etc. Laboratory data on the rate of absorption through
the skin may be available for a few chemicals. In the absence of such
data, estimates might be made through use of octanol/water (or other) par-
tition coefficients, although these procedures are unvalidated. Thus,'in
many cases, the exposure analysis will be limited to establishing the
nature of the exposed population, its size, other characteristics affecting
the exposure (duration and frequency of exposure, extent and area of the
body exposed) and perhaps an extrapolation of rate based upon the rate of
absorption of similar chemicals. Data seem to be available concerning
chemicals used in pesticides and cosmetics; because of the variety of the
chemicals used and the apparent variation in rates of absorption,' these
data may not be very useful in general estimates of average daily intake.
7-25
-------
the second category of exposures that should be examined is the ex-
posure of the general population and specific subpopulations who are in
contact with ambient or treated water which may contain the pollutant
as a contaminant. In this case, three steps are required:
(1) defining the numbers of persons exposed and the characteristics
of the exposure;
(2) estimating the concentrations of pollutants in water to which
persons are exposed; and
(3) estimating the rate of transfer from the water to the person.
A considerable body of literature exists on the number of persons exposed
to various activities which involve water—swimming, boating, bathing,
fishing, dishwashing, etc. Data on seventeen exposure activities in
personal, recreational and household categories have been identified
and summarized by U.S. EPA (1979) including estimates of the populations
exposed, extent, frequency or duration of exposure. Estimates of the
concentrations of pollutant in the water used in these activities can
come from monitoring data or frora estimates generated in the materials
balance and environmental fate arid pathways analyses.
Estimating the rate of absorption through the skin is more difficult.
An analysis of this process (U.S. EPA 1979) indicates that the diffusion
rate of the pollutant through the stratum corneum layer of the skin may
be the controlling factor; this is dependent upon the permeability co-
efficient of the pollutant and the partition coefficient of the pollu-
tant between the human skin and the water. Some data exist upon which
to base estimates; laboratory investigations of the diffusion rate of
pollutants through skin are in progress. Through combination of analysis
of activities involving water, concentrations of pollutants in water, and
rate of absorption through the skin, order of magnitude estimates of the
actual exposure by dermal contact—in terms of an average daily intake
for various activities or subpopulations—can be made.
Table 7-10 gives some examples of the estimated dermal exposure to
pollutants based upon absorption through skin. The estimates for penta-
chlorophenol (PGP) were based on a permeability constant for phenol
(Scow e_t_ al. 1930). For the halomethanes, a permeability constant for
chloroform was used (Perwak e_t al. 1980b) . In most cases, dermal ex-
posure levels are small compared with those of other routes. However,
they can be large, as indicated by the home-use of PCP as a preservative.
Other Exposure Routes
Some other specific exposure routes should be considered for selected
pollutants. A major category is the use of medical products. For example,
food supplements for humans can greatly increase exposure (i.e., zinc,
copper, other trace nutrients). Intravenous solutions and other products—
plasma, blood, dextrose or saline solutions—can be a means of entrv of
7-26
-------
TABLE 7-10. EXAMPLES OF ESTIMATED EXPOSURES TO POLLU-
TANTS BY ABSORPTION THROUGH THE SKIN
ESTIMATED
EXPOSURE
POLLUTANT EXPOSURE (mg/dav)
Psntachlorophenol:
Persons bathing and dishwashing 0.003 - 0.03
with contaminated water
Home use of PGP as preservative ]_JQ
Handling of treated wood 0.5
Trihalomethanes:
Children swimming 1 hr/day in
freshwater pool containing
^160 ug/£ chloroform 0.2 (chloroform)
Children swimming 1 hr/day in
saltwater pool containing
n-6Q ug/i bromoform 0.7 (bromoform)
Sources: Scow, K. et_ al. An exposure and risk assessment for penta-
chlorophenol. Final Draft Report. Contract EPA 68-01-3857.
Washington, DC: Office of Water Regulations and Standards,
U.S. Environmental Protection Agency; 1980.
Perwak, et al. An exposure and risk assessment for trichloro-
methanes. Final Draft Report. Contract EPA 68-01-3857.
Washington, DC: Office of Water Regulations and Standards,
U.S. Environmental Protection Agency; 1980.
7-27
-------
a pollutant to humans both from contamination of the fluid or from the
packaging material or tubing. Similarly, dental materials and surgical
implants can be a source of exposure. The contributions of these sources
to total exposure levels are highly variable, but can be very significant
for some pollutants and subpopulations.
7.3.1.4 Summarizing Exposure
As a result of exposure analysis, exposure to a pollutant by various
routes can be summarized for the general population and for specific
sufapopulation groups. In this way, the activities that are most responsible
for human exposure can be identified, ranges of exposure levels can be
developed and used to help estimate the risk associated with exposure,
and pathways of exposure can be examined in order to determine the
potential effects of changes in control regulations.
The results of an exposure analysis for lead (Perwak, ejC al. 1982b) are
shown in Figures 7-1 through 7-3 and Tables 7-11 and 7-12. As shown in
Figure 7-1, the exposure routes for humans are numerous and complex.
The ingestion of paint chips is commonly thought to be the most pre-
valent lead exposure problem in the U.S. today, and this is borne out
by the high exposure levels shown in Figure 7-3. and Table 7-11. Intake
of lead in food is the primary pathway for adults not employed in lead-
related industries and among children without pica (Figure 7-2). Inhala-
tion exposures, shown in Table 7-11, are heavily influenced by proximity
to industrial sources.
Data were available to permit estimation of the rates of absorption
of lead, and exposure levels have been converted into absorbed doses
in Figures 7-2 and 7-11. The relative contributions of exposure routes
as absorbed doses to the exposure scenarios for adults and children with
pica are displayed in these figures.
Exposure may also be measured by other parameters, such as levels in
blood or tissue. In some cases, this information can be combined with
actual effects information from epidemiological studies to achieve an
estimate of risk. This was the case for lead, and this information
(shown in Table 7-12) in combination with the exposure estimates can
give a good basis for the identification of sources of'risk.
7.3.2 Effects Analysis
7.3.2.1 General Approach
In developing an approach to address the effects of toxic substances
in the environment on humans, a number of issues deserve attention.
First, one must determine what effects should be considered in the analysis.
7-28
-------
Source:
FIGURE 7-1 EXAMPLE OF GRAPHIC SUMMARY OF ROUTES
OF HUMAN EXPOSURE TO LEAD
Perwak, J. et_ al. An exposure and risk assessment for lead
Final Draft Report. Contract EPA 68-01-3857. Washington, DC:
Monitoring and Data Support Division. Office of Water Regulations,
U.S. Environmental Protection Asencv: 1982.
-------
Air - 1%
Rural Areas —
.
Dnnk.ng
Water - 1 %
(lead solder in
cans - 25%)
Urban Areas — 50 jug/day
(lead solder in
cans-31%)
Note:
Smelters, Lead Works, etc. - 160 .ug/day
Concentrations < 10 jug/2 in drinking water were assumed for these estimates, and no con-
sumption of wine or moonshine containing lead. In addition, these situations did not include
exposure from smoking.
FIGURE 7-2 EXAMPLE OF GRAPHIC SUMMARY OF ESTIMATED EXPOSURES
TO LEAD FOR THE GENERAL ADULT POPULATION
Source: Perwak, J. et al. An exposure and risk assessment for lead. Final
Draft Report"] Contract. EPA 68-01-3857. Washington, DC:
Monitoring and Data Support Division, Office of Water Regulations,
U.S. Environmental Protection Agency; 1982.
7-30
-------
Drinking Water - 1%
Drinking Water and
Food (4%) /Air - 1%
Paint and
Paint
Contaminated
Dirt
(90%)
Rural - 560 Mg/day
Drinking Water
and Air - 1%
Smelters. Lead Works, etc. - 1300 Kg/day
FIGURE 7-3
ource:
and
EXAMPLE OF GRAPHIC SUMMARY OF ESTIMATED EXPOSURES TO
LnAD FOR A SPECIFIC SU3POPULATION (CHILDREN WITH PICA)
•> J. &£ Al. An exposure and risk assessment for lead.
.- Report. Contract EPA 68-01-3857. Washington, DC •
.toring and Data Support Division, Office of Wate- Re
-------
TABLE 7-11
EXAMPLE OF EXPOSURE ESTIMATES OK LEAD FOR ADU]TS
AND CHILDREN INCLUDING ESTIMATED ABSORBED DOSE
Iocat ton
Kural
I'll..1.1
;lt Ing Area-.
Food
Prinking Water
InhalacIon
Drinking Water
Inhalation
l-ood
— —
1. it al liK-i
M, Him, hi II,.'
Wine
MOM SupplU...
UNUdiilnad.d
Highly Con t ami i ill 1
Suburban
'""" '•'
T,,tal lilct
M, Hill ,1,1 lll_
Wlnu
Mo:»l Sniip 1 li ;»
(•ontaulnatcU
Highly Coiiianiln.il 0,1
Hi L. in Air
1 ot a 1 t)i » t
W | rif
Asa ijm[>( 1 on
1 ">K/1 . 1 I/day
0.2 oig/1. 1 I/day
0.-' »t;/k. 5 I/day
' 10 uB/l. 2 I/day
• JO ,,g/l. 2 l/dav
- 1000 ,,K/|, . j/djy
See Table 5-lb
See Table 5-lb
1 "IS.M. 1 1/djy
" • * lft^J/1, 1 1 / J .1 y
0.2 iug/1 5 |/ddy
' 10 UK/1. 2 i/day
i« .'g/1. 2 1/Jay
1000 llt!/l. 2 I/day
See Table 5-lb
' ""«/». 1 I/day
°-2 «K/I . 1 I/day
".2 »B/1, 'j I/day
lii; .ike
(iig/duv)
100-200
10OO
200
1000
<20
• 1OII
> 2000
l.S-IS
0.4
-
100-200
1000
200
1000
< 20
• 10O
' 2000
li-62
-
1000
200
IO(/U
Ah^oibcd
TT.'gTjL
. .
100
11)0
v.
• .'Oil
•1. 1
1-5
lU-.'U
10 >
IUU
,
-
JOO
i-JI
1 -b
loo
fc 0
IOIJ
Do:,
-------
Population
CMldn-n
I
Lo
U)
TABLE 7-11. EXAMPLE OF EXPOSURE ESTIMATES OF LEAD FOR ADULTS AND
CHILDREN INCLUDING ESTIMATED ABSORBED DOSE (Continued)
Location
Rural
I'l i III 1.1,; W.I. I
Inhalation
Food
Drinking Ua
ar ton
Source-
MCI;, I Sup
(..mi .nulli
•Uglily C
Aablunt Air
Cigarettes
Total
Contanlnaled
Illglily Con tau Ina l
Aoibtent Air
'"' JL'i'J'1 '' '
•10 ,,„/!. I l/.,jy
' 'M UK/I . - i/Jjy
'1000 Mg/l. * |/.|.,y
10 us/".1. -'0 »,J/djy
10 Hi/1. I l/d.iv
50 ug/l. 1 i/j.,,,
See TabJe 5-l
I II C.I k 1-
' Idl)
2OMO
100
<10
<50
•' moo
0. 11,-l.t,
> IU
> 20(1
60
0.1-1
Urban
lead Paint
I'alnt or Othe
(uiil amlnaliij
Totjl
IX Uad. 1 mg chip
1000 ng/,- K-a.l |n
10 UK/Bum h ing, 10
0 l.g/1. 1 1/Uay
<50 l.g/1. 1 1/djy
> 1000 Mg/1. 1 |/Jjy
> 1OUO
II lead. 1 ag chip 1000
1000 jjg/g JeaJ in din, JUQ
10 BK/uouthlng. IU numtliln,;;./
day
10.000 ,lt/B |e.,d !„ dusi
10 UK diiit /imuillilllK, 10
nout h Ingtt/dav
|O00
<5
<25
> iOO
so
soo
-------
TABLE 7-J1. EXAMPLE OF EXPOSURE ESTIMATES OF LEAD FOR ADULTS AND
CHILDREN INCLUDING ESTIMATED ABSORBED DOSE (Continued)
i .111
li hi.1. Ale. i.,
Knute
Orluk I Hi; Water
l.il Ion
Intake
I' lea
tin.11 III el
Host SuiM'lli-'a
root.milnacc'il
Highly Cunt Jlaln.il
Ami. li-llt All
Lead Pdlnt
Dirt
AautimcJ tu he half ut
adult
• 10 PB/l. 1 I /day
'50 „,./), 1 |/JJV
• I'»00 i,B/|. 1 |/.|.JV
(lift/day)
500
< 10
-50
> 1000
t'B/*1. i w'/d.lv
IT U-.iJ, I i,.,
HI »n/u:.Hilliiii[;. |(| momli
J.iy
10.000 i,j../B U..,. ,,, j,,.,t
Hi mi; duat /ii..iul.inK. lo
>i.<.il I. liH'a,',! i,1
40
1 1 II ID
too
1000
(lig/djy)
< i
U
idl)
SO
500
J0% absorption of ingested lead is assumed for ad.Uia and SOX for children. U-oosltion
^Inhaled ledd is assumed to he JOJI and 100% if deposited lead is assumed to be -bsorLcd?
Table 5-1 in source cited below.
Sour,..!: Herwak. J. ej, .U. An exposure and risk assessment foi leao. Final Draft Report
Contract EPA 68-OJ-M49. Washington. DC: Monitoring and Data Support WvUion '
OfUce of Water Regulations and Standards. U.S. Fn»ir.-.n. ,1 „,„,'_.,_.. .
»009 * —' *" —•*•*••*-•• ••••• • *»>*-t.»_«.iuii /njency,
-------
TABLE 7-12. EXAMPLE OF LEAD LEVELS IN BLOOD IN
SUPPORT OF EXPOSURE ESTIMATES
LOCATION
Adults
Rural/Urban
Urban
Rural
Within 3.7 meters
of Highway
Living Near a Smelter
Children
Urban (primarily)
Within 30 meters
of Highway
Near Smelter—Kellogg,
ID—1974 (immediate
vicinity)
1975
1979
El Paso, TX
BLOOD LEVEL
(ug/100 ml)
9-24
Most
REFERENCE*
Ball et al. (1979)
16
18—mean (adjusted for age Tapper and Levin
and smoking)
Less than 5% >30
16—mean (adjusted for age
and smoking)
Less than 0.5% >30
23—mean
16% >40
(1972)
Daines et al. (1972)
Landrigan et al. (1975)
40,000 children detected
annually >30
•\ 20 yearly geometric mean Billick et al. (1980)
50% >40
99% >40
60% >60
Somewhat reduced'
Almost all >60a,
and most >40
70% >40
14%
Caprio et_ al. (1974)
Walter et al. (1980)
Anonymous (1979)
Landrigan et al. (1975)
Reduction as a result of reduced atmospheric emissions as well as increased
sanitary procedures for the workers who were apparently exposing their
children to lead through their clothing.
*See source indicated .below for references.
Source: Perwak, et al. An exposure and risk assessment for lead. Final
draft report. Contract EPA 68-01-3857. Washington, DC: Monitor-
ing and Data Support Division, Office of Water Regulations, U.S.
Environmental Protection Agency; 1982.
7-35
-------
Second, the spectrum and quality of health effects data available
for use in assessing risk to humans should be examined. For chemicals
that have long been recognized as toxicants, data may be available from
epidemiologic studies or from reports of effects on humans, as well as
data for laboratory animals; for other chemicals, especially newly
developed ones, only laboratory animal or in vitro data may be available,
if any at all. As a result, one must be prepared to adapt the scope of
the analysis in accordance with the available data.
Third, one must consider the format of the effects analysis.
Ideally, quantitative relationships based upon human data would be
developed between the exposure (expressed in terms of specific dose,
average daily intake, etc.) and the response of humans (death, morbidity,
changed reproductitive capacity, etc.). Generally, sufficient information
will not be available and, by necessity, data for Laboratory animals will be
extrapolated to man. Frequently there will be insufficient animal data
to develop dose/response relationships and risk will have to be assessed
in terms of no observed effect: levels and appropriate safety factors.
In cases in which no specific data associated with the chemical are
available, the only option remaining will be qualitative statements
based on structure/activity relationships or similarity with other chem-
icals .
Assessment of potential health hazards associated with exposure
to a particular chemical typically begins with a literature search.
There are many computerized and manual data bases for health effects
and toxicologic data (e.g., CANCERLINE, TOXLINE, MEDLINE, ENVIROLINE,
etc.) that can be used to obta.in citations concerning human safety as-
pects of the chemical in question (see Chapter 10 for a listing). The
number and scope of citations in these data bases are expanded regularly.
By careful selection of key words and structuring t:he search to include
the possible effects of the pollutant, a substantial amount of data can
be obtained.
When the literature has been obtained, all aporopriate and reliable
human and animal data should be evaluated. As an aid to the organiza-
tion and analysis of the information, a matrix of the types of data that
should be analyzed is presented in Table 7-13. As indicated earlier,
direct human data are more desirable, but generally not available.
Ideally, if one could fill in the human data columns of the matrix, then
data for the other columns would not necessarily have to be considered.
Not all types of health effects need to be thoroughly studied, in that
the data needs will be unique for each chemical. However, each area
should be examined briefly to determine if it is relevant to the chemical
in question, and if its inclusion in the risk assessment would be useful.
Once all available information has been thoroughly evaluated,
judgments should be made regarding the relevance of the mode of exposure
utilized in animal studies to that associated wich human exposure.
Interaction of agents that may result in synergiscic or antagonistic
effects should also be indicated, if known. On the basis of the kinds
-36
-------
TABLE 7-]3. MATRIX FOR INITIALLY ORGANIZING ANALYSTS OF HUMAN HEALTH EFFECTS INFORMATION
Data on Humans Mammalian In Vitro Inference/Extrapolation from
Health Effects Information Epidemiologies] Accident. Data '~"Da_ta~~ Other Related Chemicals
Metabolism, absorption,
accumulation, distribution,
excretion (pharmacokinetlcs
and mechanism of action)
Acute
Subchronic
Chronic
Ca r c ino gen t c 1 ty
MuCageiiicity
Teratogenicl ty
Fetotoxicity
Functional Disorders and
Effects
CNS
Reproductive
Hepatic
Renal
Cardiac
Gas trointes tinal
Uespi ratory
Di gestive
Circulatory, etc.
-------
of responses induced by the chemical, an assessment can be made of the
acute and long-term adverse effects that might result from exposure to
the chemical, in a form usable for risk analysis. The output of the
effects analysis should include:
(1) The type and nature of the effects of the pollutant expected in
humans;
(2) The levels of exposure (dose, intake, etc.) that produce these
effects in humans and/or experimental animals;
(3) The quantitative relationships, if any, that have been docu-
mented between effect and exposure in humans or experimental
animals;
(4) The variation in effects and exposure/effects relationships for
different human subpopulations (age, sex, diet, etc.).
(5) The levels of exposure at which no effects are observed; and
(6) The level of uncertainty in the available data.
That every chemical will induce negative health effects if adminis-
istered in sufficient quantity is axiomatic in toxicology. The challenge
is always to establish what exposure levels are probably non-threatening
and what exposure levels are associated with certain risk. Thus,
toxicology cannot avoid being a quantitative discipline.
Human biology, however, is very difficult to quantitate. Thus,
quantitative predictions must be developed for a highly heterogeneous,
poorly reproducible system. As a result, reported values are generally
thought of as representing some point (hopefully, the midpoint) of a
fairly broad range of effects rather than an exact number to be_taken
at face value. Any quantitative conclusions regarding health risks
must be reached with care and with recognition of reliability and/
or limitations of the data upon which they are based.
In the sections below, the types of data that are desired for human
effects analysis are briefly presented, along with an approach or hier-
archy for examining and analyzing these data. Examples of typical data
summaries from actual risk/exposure assessments are presented in the
discussion. [In Chapter 9.0, possible methods for extrapolating animal
data to humans (quantitative risk assessment) will be discussed.]
7.3.2.2 Details of Approaches and Examples
Absorption, Metabolism, Bioaccumulation and Excretion
In studying the effects of a pollutant on humans, it is important
to know the routes by which the pollutant can enter the body; the degree
of absorption, if any; the extent of metabolism; whether the pollutant
7-38
-------
is accumulated and in what tissues; and how it is excreted. These
factors are imporatnt for several reasons:
(1) They form the linkage between exposure of humans to concentra-
tions of pollutant in the environment and the possible effects.
(2) They establish the relative significance of the various
routes-
(3) They may indicate target organ systems-
(4) They can aid in the interpretation of data on concentrations
of the pollutant in human and animal tissues (monitoring data).
(5) They may provide a rational basis for estimating the effects
of the pollutant on humans based upon animal or in vitro
studies-
(6) They may suggest other related chemicals (metabolites or pre-
cursors of the pollutant) that need to be examined.
Recommending a generalized approach to seeking and evaluating these
types of data is difficult; nevertheless one would generally start with
human data if available and then proceed to experimental animal data.
For example, for established chemicals, pesticides, and most metals,
one can anticipate that epidemiologic studies, as well as animal data,
will be available. Important considerations needing investigation
include: degree of absorption; rate of clearance from blood or plasma;
principal routes of elimination; sites and amounts of residues or
accumulations in body tissues; the half-life in the body for the pollu-
tant and/or its metabolites.
For example, if biliary excretion was found to be a major route
of elimination, species differences in the rate of biliary excretion
of the compound into the bile might result in specias variation in the
biologic half-life of the compound and its toxicity. Another example
of the usefulness of these data is the use of tissue distribution
patterns in defining populations at risk. Toxicants are often con-
centrated in a specific tissue; some may be concentrated at their site
of toxic action, such as carbon monoxide, which has a high affinity for
hemoglobin. Other chemicals are sequestered harmlessly at storage
sites, but may be released at coxic levels on reniobilization of the store,
e.g., chlordane stored in body fat, can be remobilized under weight loss
conditions; lead stored in bone can be remobilized with increased calcium
demand, such as during pregnancy and/or lactation.
Another reason for reviewing metabolic and pharraacokinetic data is
that some substances in the environment are also essential elements or
nutrients in many specias, e.g., copper and zinc. Understanding the
pathways and uses of the element in the body can help co establish whs-he-
7-39
-------
the amounts obtained from environmental sources are excessive for normal
body function and whether large or small increments can lead to toxicity.
In the case of copper, for example, one finds that absorption of ingested
copper is very incomplete (Venugopal and Luckey 1978). Furthermore,
ionic copper has a strong emetic action. As a result, ingestion of
copper and its salts in small quantities does not usually present a high
risk. Inhalation of copper dusts, fumes or copper-containing products
may present a more serious risk (Perwak, et al. 1980a).
Other examples could easily be drawn from the literature, but these
should suffice to indicate the importance of metabolism, accumulation,
pharmacokinetics, and mechanism of action data in effects analysis.
Acute Effects
Although cases of acute human effects resultings from exposure to
environmental pollutants are not very prevalent, it is important to
examine acute human toxicity dz.ta for several reasons:
(1) acute accidental or occupational exposure to high concentra-
tions of pollutants may be the only human data available;
(2) acute effects may identify specific organ systems at risk to
chronic exposure; and
(3) the comparison of acute human effects with animal data combined
with metabolism and ether data, can support the use of chronic
animal data for extrapolation to humans.
The majority of acute human toxicity data that is most often avail-
able in the medical literature, "poison centers," and/or NEIS3, results
from suicidal or accidental exposure, often in children. Standard tests
on industrial safety and hygiene may also contain acute toxicity values
for inhalation, ingestion, and dermal absorption. Although large bodies
of data from humans are often not available, the types of acute toxicity,
symptoms, and effects, and in some cases, minimal lethal values for man
are generally available. The minimum lethal dose, however, only indicates
chat a single death due to the chemical has been recorded at that dose
which may be the results of a high dose accident or suicide attempt. In
fact, the minimum lethal dose may be equivalent to an 105 (the dose found
to be lethal to 57, of the exposed population) ; or it may be many times
higher than an LDgg (the dose found to be lethal to 90% of the exposed
population). Thus, quantitative conclusions on human risk must be reached
with care, according to the limitations of the data,,
Data may also be available concerning the acute toxic effects of a
chemical in laboratory animals, particularly rodents. Acute toxicity
studies provide information on the relative effects of different expo-
sure routes (inhalation, ingestion, skin contact), provide a measure of
comparison among many substances whose mechanism and sites of action may
7-40
-------
be markedly different, and are roughly indicative of the effects of
chronic exposure to small amounts of the chemical. Acute toxicity tests
are also frequently conducted to determine local effects of chemicals
when applied directly to the skin or eye. Thus, acute toxicity studies
place the overall acute toxicity of different pollutants in perspective.
In the development of new chemicals, for example, these acute tests are
often used as an initial screen to aid in the determination of whether
or not to continue to develop a chemical. Such tests are also required
by regulatory agencies for pesticides, drugs, food additives, etc.
There are not generally accepted standard data to search for, or
special means of data presentation. A clear understanding of the impli-
cations of the data is the important concern. For example, very low
exposure levels of cyanide are very acutely toxic, but are rapidly
cleared from the body (Williams 1959), while lead may present no acute
toxic response at low levels, but its accumulation in bone can result
in grave consequences to man (Mahaffey 1977).
Subchronic Effects
Subchronic testing involves the administration of the test chemical
on multiple occasions. Experiments are generally conducted for 90 days
with rats or mice, for 6 months to 1 year with dogs. Subchronic studies
are typically conducted at higher exposure concentrations than chronic
studies. Pathologic changes are thus more clearcut because they occur
more quickly with the higher doses and are not obscured by other chronic
changes such as aging. For example, focal myocarditis is'a common
spontaneous type of lesion found in high frequency in aging populations
of rats (Simms 1967). A marked increase in the incidence of this lesion
after 90 days' exposure would be noteworthy, but might be attributed to
aging or a small population remaining at termination of a chronic study.
Carcinogenicity
Cancer is characterized by an uncontrolled growth of abnormal cells:
a carcinogen is defined as any toxic substance which has been demonstrated
to cause tumors in mammalian species by induction of a tumor type not
usually observed, or by induction of an increased incidence of a tumor
type normally seen, or by its appearance at a time earlier than would
be otherwise expected (National Cancer Institute 1976).
As is the case with other effects, examination of carcinogenic risk
begins by consideration of human epidemiologic data, if available.
Figure 7-4 presents in flow chart form a procedure for evaluating data
on carcinogenicity. Ideally, one would follow the yes pathways to
develop the most reliable estimates of the carcinogenic effects of the
pollutant. Thus the chart is organized so that the items in the bottom
row appear from left to right in order of descending desirability and
reliability. If data are limited to in vitro data, data on related
compounds, or structure-activity relationships (SAR), the risk of carcino-
genicity can probably not be assessed reliably.
7-41
-------
-J
I
-p-
NJ
FIGURE 7-4 FLOW CHART FOR CARCINOGEN RISK EVALUATION
-------
The route shown on the left-hand side of Figure 7-4, based entirely
upon human data, is the ideal path in evaluating carcinogenic risks, but
will in reality very seldom be used because adequate data are lacking.
Epidemiologic data, even if available, most often do not represent
causal relationships, only correlations or associations, and must thus
be augmented by other types of data. Reports of occupational exposure
give a somewhat more direct indication of causality, but the dose-response
relationships may be difficult to define. Thus, in most cases, human
data alone will not provide a suitable risk estimate, although coupled
with experimental animal data, they may permit a more rigorous analysis.
If experimental animal data are available, there are four possible
routes to assessing risks depending on, first, the number of species
tested and, second, whether or not dose-response relationships are known.
In following any of these paths, careful attention must be paid to the
quality of the data, the incidence of spontaneous tumors in the control
population, consistency if more than one study is available, and
statistical validity. If the exposure route and experimental regimen
employed (e.g., intra-muscular injection) do not agree with the most
likely mode(s) of human exposure, the data must be interpreted
cautiously. Consideration should be given, to data on metabolism of
the compound by the animal species tested, as compared with metabolism
in humans if this information is known.
If only in vitro data are available, only qualitative estimates
may be possible because of uncertainties regarding the association
between in vitro results and human or animal effects; the availability
of associated pharmacokinetic data, however, may allow an approach to a
rough quantitative estimate. Even less reliability will be possible if
no experimental data are available and only SARs can be established
between the compound and related compounds for which data are
available.
In following the path indicated in Figure 7-4, information from the
National Cancer Institute's carcinogenicity programs should be examined,
as well as data from the medical literature. Discussions with qualified
toxicologists/oncologists should supplement a critical analysis of the
literature for chemicals for which the data are equivocal or conflicting.
For example, animal studies weakly support an association between exposure
to benzene and carcinogenicity (Snyder &t_ al. 1980), but the evidence that
benzene is a leukomogen for man is convincing (Askoy 1977; Infante at. al.
1977a,b; Ott jet aJL. 1978). Benzene may in fact play a co-leukomogenTc
role, which would explain the failure to induce leukemia in several
benzene-exposed animal species.
7-43
-------
Mutagenicity
A mutation can be defined as any heritable change in the genetic
material (DNA) of a cell or organism. Among the sequelae of a°mutation
are cell death, altered structure and/or function, and no overt immediate
effect, should the mutation be unexpressed fay virtue of its recessive
nature. The types of changes that occur in the genetic apparatus of a
cell can range from modifications in the individual base pairs, result-
ing in point mutations in a single gene, through major chromosomal
structural changes that may involve entire sets of genes, to disruption
of entire sets of chromosomes. Some examples of these kinds of occurrences
in humans are sickle cell anemia (an example of point mutation) and
Trisomy 21 (Down's syndrome, an example of a major chromosomal disorder).
Some relatively rapid and technically accessible bioassays for
mutagenicity are being used as predictive tools to identify not only
agents with possible mutagenic activity, but also those that may induce
cancer or cause a teratogenic response. A large amount of experimental
evidence indicates that many agents that are carcinogens also can damage
DNA, and the correlation between activity of chemicals in the mutation
screens and activity as carcinogens is high [e.g., the Ames test has been
found to be 80 to 90% accurate in detecting carcinogens as mutagens
(McCann et_ al. 1975)]. This and similar tests are widely used as an
initial testing mode to identify potential genotoxic agents, although
correlations between potency and response are often not good. Thus,
testing for mutagenicity has wider implications than merely determining
the potential for mutagenic ris
-------
I
f>
Ol
Are there
human
data?
Are there
experimental
data?
Structure
activity
relationships
Specific
locus or
heritable
translocaiion
Are there
direct
data?
Are there
indirect
data?
Evaluate assays
Evaluate assays
1. Epidemiology
2. Fetal wastage
1. Cytogenetics
2. Urinary mutagens
3. Sperm morphology
4. Testicular atrophy
Extrapolation to humans
Risk Estimates
Only very
qualitative
risk
assessment
statements
possible
FIGURE 7-5 FLOW CHART FOR MUTAGEIMICITY RISK EVALUATION
-------
and time required for the specific locus assay, there is not much likeli-
hood that this test will be used for materials except those that may have
a large medical and/or social importance. Thus, the heritable trans-
location assay appears to be the only readily available experimental tool
at this time for a direct measurement of transmissable genetic damage.
Its major disadvantage is that it measures only male-originated changes,
and it too is a relatively large and expensive bioassay.
The remaining battery of genetic tests, though useful, are only
indicative and the results must be evaluated with considerable care for
appropriate use in the risk analysis process. A number of criteria should
be examined in order to establish the usefulness of the experimental
data obtained from bioassays:
(1) heritable versus non-heritable changes,
(2) phylogenetic hierarchy,
(3) genetic endpoint,
(4) sensitivity of assay system, and
(5) validation of assay data and assay system.
These criteria are interdependent and complex,, so that a simple treat-
ment of these variables is not likely to be possible.
For example, a marked increase in chromosomal aberrations was noted
(see Table 7-14) in mouse spermatogonia following gavage administration of
doses of aqueous phenol solution far below levels associated with other
effects and at environmental exposure levels that a large fraction of the
human population may encounter (Bulsiewicz 1977). In addition, an apparent
trend toward increased aberrations within a single treatment group in each
of five successive generations was evident. In that (1) most mammalian
species, including man, handle phenol biologically in a similar manner,
(2) the treatment route of this study is the same to which man will
likely be exposed, and (3) in vivo cytogenetic analyses in mammals are
considered more relevant than similar tests in vitro or genetic tests
with lower organisms for predicting a mutagenic hazard for man, Bulsiewicz's
results were cause for concern. Unknown factors, however, such as tissue
and species made any discussion of the genetic implications of the re-
ported chromosomal aberrations for man more speculative than factual.
The means by which genetic assay data should be used in a risk assess-
ment has been considered by a number of prominent investigators in the field
(Freese 1973, Crow 1973, Bridges 1974, and Report of a Committee of the Euro-
pean Environmental Mutagen Society 1978). They generally agree that risk
analyses and subsequent action based on experimental tests other than the spe-
cific locus and heritable translocation assays would not be easily supportable
7-46
-------
7-14. EXAMPLE OF PRESENTATION OF MUTACENICITY DATA—INCIDENCE OF
CHROMOSOMAL ABERRATIONS IN SPERMATOCONIA OF PHENOL-TREATED MICE
Dosage
Generation (ug/kjjy
P 0
6.4
64
640
' 0
6.4
64
640
K-> o
6. 4
64
640
K3 °
6.4
64
640
F4 0
6.4
64
640
r5 0
6.4
64
640
Level Chromosome
'day) Breaks
0
1.7
5.8
9.2
0
3.3
10.8
12.5
0
9.2
15
19.2
0
5.0
10.8
10*
0
6.7
15.8
20
0
10
17.5
51.3*
Chroma t id
Breaks
0.8
3.3
5
7.5
0
10
15
14.2
1.7
8.3
15.8
17.5
0
5.8
14.2
10*
0.8
8.3
20
25
0
6.7
23
37.5*
Aneuploidy
0
1.7
5
10
2.5
11.7
15
17.5
0
9.2
17.5
19.2
0.8
13.3
22.5
36*
0.8
10
20.8
27.5
1.7
13.3
25.8
37.5*
Polyp loidy
0.8
3.3
10.8
13.3
2.5
1.7
15.8
19.2
0.8
5
14.2
22.5
1.7
8.3
15.8
32*
0.8
6.7
23.3
30.8
0
11.7
21.7
56.3*
A,
Associations
o
0 ft
2.5
1.7
o
3 3
5.8
7.5
o
5 n
7.5
5.8
0 1
3.3
9.2
8.0*
o
10
15.8
.17.5
0 2
6. 7
19.2
25*
Excludes 3 mice killed in moribund condition. Preparations made from the testes of these mice showed
absence of primary and secondary spermatocy tes . spermat ids, and spermatozoa.
Source:
Scow, et al . An exposure and risk assessment for phenol. Final Draft Renorr Pnnrr^
EPA-68-01-59A9 Washington, DC: Monitoring and DLa Support DlvSon, Office of Sater
Regulations and Standards, U.S. Environmental Protection Agency; 1981.
-------
Moreover, these investigators believe that a positive finding in the
heritable translocation assay should be given considerable weight in a
risk analysis.
No special data presentation methods need to be considered in muta-
genicity analysis; careful reporting of the interpretations of others
and correlation of one series of tests with another are, however,
important.
Teratogenicity
Teratology can be broadly defined as the study of malformations of
the newborn that occur as a result of an adverse effect(s) on the develop-
ing conceptus. In the past, the term malformation implied gross anatomic
malformations; but in recent times, the term malformation has been
broadened to include more subtle, functional abnormalities and even
postnatal behavioral and intellectual development.
A teratogen, the agent ex&rting an adverse effect on the developing
conceptus, exerts its effect in the time interval between conception and
the termination of morphogenetic development in the post-partam animal.
The picture is complicated somewhat in that certain morphogenic events
are terminated at widely varying times in different species. For instance,
re-opening of the eyelids, opening of the external acoustic meatus,
formation of the vaginal lumen, and descent of the testes occur pre-
natally in man, whereas these events take place postnatally in the mouse,
rat, and rabbit, the species mcst often used in experimental studies.
In addition, other factors such as metabolic differences, excretion rates.
placental variations, age of the dam, and nutritional status may all in-
fluence the potential teratogenicity of a chemical Ln a particular species.
Moreover, the dose, route, and time of gestation at which a conceptus is
exposed are critical in defining whether a particular chemical is tera-
togenic in a particular species .
The assay procedures presently available to test for the teratologic
potential of chemicals are empirical, largely because the detailed biological
mechanisms of teratogenesis are not well understood. Clearly, under-
standing the mechanisms of teratogenesis would be the most fruitful
approach to predicting the risk of exposure to new chemicals or even well
established ones. Unfortunately, the state of the art of understanding
the mechanisms of action of most known teratogens is quite primitive.
There is evidence to support about 8-10 mechanisms of action: mutation,
chromosomal aberrations, mitotic interference, altered nucleic acid
integrity or function, lack of precursors and substrates for bio-synthesis,
altered energy sources, enzyme inhibition and osmolar imbalance and
altered membrane characteristics (Wilson 1977). All of the above mech-
anisms are not mutually exclusive, i.e., both genetic and environmental
factors determine tertatogenic risk. In addition, nany of these mech-
ansisms are non-specific effects seen only at high doses (e.g., enzyme
inhibition, altered energy sources) and extrapolation of these findings
to man is of questionable value.
7-43
-------
This lack of information on the mechanisms of teratogenesis is high-
lighted by the findings that show chemicals to which humans are widelv
exposed such as aspirin, vitamin A and hydrocortisone, to be teratoeenic
in certain experimental systems. Although there are no data to eliminate
these chemicals from consideration as human teratogens, there is also no
evidence that their consumption by pregnant females or by males prior to
fertilization at doses normally employed has resulted in malformation in
their _ progeny. In contrast, methyl mercury and methotrexate, which have
been implicated as human teratogens, induce a teratogenic response in a
wide range of species, including the smaller rodents usually employed in
most experimental studies, i.e., mouse, rat, Syrian hamster.
One means of approaching a better understanding of the relationships
between teratogenic effects of chemicals in humans and in experimental '
animals is to examine those instances for which a chemical has been
identified as a human teratogen and has been tested later in experimental
SSdo* 'd ?°SK WSl1 kn°Wn "^ °f thiS C^e is the «« concerning
oflhe ±M. ".especially instructive, since it illustrates sLe
nfd ° US§ experimental animal data in a prospective manner.
Had
ma
been tested for teratogenicity according to the usual
protocols in rats or mice, there would have been little or no indication
of any problem In rabbits and subhuman primates, however, thalidomide
was demonstrated to be a potent teratogen. ^aomiae
fr0m ?^VhS Pr±maf7 difficulties in extrapolating experimental data
from laboratory animals to man is the high probability of differences
in metabolic fate of chemicals, especially their disposition in Jonadal
or placental tissues. Moreover, because of the compLx behaviorll
a
of
avor
l™CtiC*S°f humans> to contrast to the controlled regimens
^ Timal tSSt P0^'10"3' there -ay be wide individual
among humans witn respect to the ultimate metabolic fate of
chemicals taken into the body. Another source of variation in humans
with th SS1C S^eCiC individualit7 of each person in comparison
of thf " rajher. unJlf,OHB §enetic background of test animals, even chose
of the random bred" category. Since the susceptibility to teratogenic
stimuli appears to have a genetic component in humans, the presencfof
genetic diversity is a further complicating issue in the use of
Sck Wlf. terat°10gy data' f°r estlaation of such a risk in humans
Thus, although the methods of detection of potential teratogenic
agents have merit, at present there appears to be no clear correlation
between teratogenesis data for humans and that for experimental animals.
Accordingly, man has no alternative but to take a conservative approach
toward exposure to chemicals during pregnancy.
With the above precautions, the risks associated with teratogenesis
can be examined as shown in the general flow chart of Figure 7-6.
Following in depth examination of human data, if any, including
studies on related chemicals, in vivo and in vitro, laboratory
r-49
-------
I
Ul
o
Are there
human
data?
Are there
experimental
data?
Structure
activity
relationships
Exposure
to known
concentrations?
Are there
epidemiological
data?
Human
data on related
chemicals?
Assess
reliability
Evaluate studies
Evaluate assays
Extrapolation to humans
f possible
Only very
qualitative risk
assessment
statement
possible
Risk Estimates
FIGURE 7-6 FLOW CHART FOR TERATOGENICITY RISK EVALUATION
-------
experimental data can be examined and evaluated. High reliability and
certainty is given to animal studies in vivo, with caution to extrapola-
tion to humans as indicated above. In many cases, only qualitative
estimates of risk will be possible for teratogenesis.
Fetotoxicity and Other Reproductive Effects
The interrelationship among lethal action upon the embryo, maternal
toxicity, and teratogenic effect is complex and the distinction of one
type of effect from another is not always clear. Until recently,
reproductive hazards have not been considered in depth by scientists,
industry, and regulatory agencies. A major obstacle in resolving this
problem is the serious lack of clear scientific knowledge about toxic
agents that affect reproduction.
The reactions of an embryo to a particular chemical depend on a
number of factors: species differences involving absorption, metabolism,
excretion rates, distribution and concentration of a chemical in maternal
body tissues, transfer across the placenta, and the kinetics of a
chemical in the embryo-placental unit. In addition, maternal adaptation
to prolonged exposure and the adequate concentrations of the chemical
during organogenesis, contribute to the problem of predicting effects
in man on the basis of tests in other mammalian species.
Single generation studies include reproductive, teratologic, and
postnatal effects resulting from exposure to a particular chemical.
The study of fertility and general reproductive performance includes
effects on gonadal function in both sexes, mating behavior, estrous
cycle, and early stages of gestation.
In order to study the long-term effects of chronic exposure to a
chemical where concentration may be a factor, the single-generation
study may be extended for several generations into a multigeneration
study. The toxic responses are reported as a series of indices for
each generation. The fertility index or conception rate represents the
percentage of matings that result in pregnancy and is affected somewhat
by the fertility and libido of the male. The gestation index is an
indication of the number of litters that contain live pups. It is an
incomplete measure of fetal mortality unless the entire litter is
stillborn. The sex ratio gives an estimate of the relative fitness of
each sex and viability and weaning indices are used to measure the
ability of pups to survive.
Many of the problems cited in the teratogenesis section on extrapo-
lating experimental data from laboratory animals to man are also relevant
to the analysis of the potential embryotoxic and reproductive effects of
environmental agents on man. Epidemiologic data in humans are generally
unavailable since exposure is frequently unsuspected or difficult to
quantitate. A notable exception is the fetal alcohol syndrome.
7-51
-------
a
Various forms of the pollutant may have different effects, even in
single species of experimental animals. For example, Table 7-15 gives
results of a study of the effect of copper salts on pregnant golden
hamsters. Copper was given intravenously on day 8 of gestation. Copper
in chelated form (as citrate) was more embryopathic than uncomplexed
copper (as sulfate) although the embryocidal activities were similar.
One must be very cautious in attempting to relate fetotoxic and
other reproductive effects in inbred laboratory animals to similar effects
in a heterogeneous human population. However, positive findings in several
laboratory species would suggest the possibility of similar effects in man
particularly if the metabolic pathways of the chemical in humans and the
laboratory animals are similar.
Chronic Functional Disorders
Chronic functional disorders include irreversible changes resulting
from intermittent or continual exposure to low levels of a pollutant that
result in detectable detriments in functional capacity (pathological,
physiological, biochemical, behavioral), the ability of'the organism'to
maintain homeostasis, or to compensate for a treatment-induced enhanced
susceptibility to the deleterious effects of other environmental insults.
Although all significant toxic effects are of concern, a reversible
functional effect, although undesirable, would be of vastly less
consequence to man than the development of an irreversible functional
effect. In addition, most human exposures to environmental pollutants
are typically long'term exposures to low ambient concentrations, and,
therefore, chronic functional effects may be the most x^idespread
consequence of exposure to these compounds.
The ideal data for assessing the significance of a chemical as a
cause of chronic human disorders; would be the results of chronic
administration of measured amounts of pure chemical to human subjects
by the appropriate route. Since these data are not likely to be avail-
able, one must consider whatever human data are available, data from
laboratory animals, and, when there are no relevant data for the chemical
of interest itself, data for similar chemicals.
In all, six types of data may be used in assessing the risk of
chronic functional disorders:
(1) human - chronic exposure;
(2) human - subchronic exposure;
(3) animal - chronic exposure;
(4) animal - subchronic exposure;
(5) human or animal - acute exposure;
(6) extrapolation from other chemicals.
7-52
-------
TABLE 7-15. EXAMPLE OF PRESENTATION OF TERATOGENESIS DATA—EFFECTS OF COPPER SALTS IN HAMSTERS
I
Ul
Dose
Lc-ve 1
(mi-.Cu/kfO
as Copjn>r Sul fate
J. 13
4.25
7.5
10.0
as Copper Citrate
0.25-1.5
1 .8
1 .2
4.0
No.
Mothers
Treated
16
3
3
2
13
6
8
2
No.
Gestation
Sacs
210
49
30
maternicidal
172
81
99
maternicidal
No.
Living
Embryos (%)
155 (74)
7 (14)
0 (0)
-
143 (83)
48 (59)
65 (66)
__
No.
Resorp-
tions
55 (26)
42 (86)
22 (74)
-
29 (16)
33 (41)
34 (34)
_
No.
Abnormal
Embryos (%)
12 (6)
4 (8)
_
'
A (2)
14 (17)
35 (35)
f^lf!U'.°AS.^demlneralized waterj^
0.5-1.0
mJ/IOOg
10
125
115 (92)
10 (8)
0 (0)
Source: Perwak, J. et a1. An exposure and risk assessment for copper. Flnal Draft
Report. Contract EPA 68-01-3857. Washington, DC: Monitoring and Data
Support Division, Office of Water Regulations and Standards, U.S. Environmental
I'lotection Agency; 1980.
-------
These types of data are listed in order of the priority that should be
given to their evaluation. Figure 7-7 presents in flow chart form a
general procedure for incorporating the best available data into an
assessment of chronic functional effects. The least desirable pathway
is the one based upon physical or chemical properties and/or structure
activity relationships. Figure 7-8 is a schematic diagram showing in
greater detail the evaluation process that might be followed for a pollu-
tant to which humans are exposed by dermal contact.
In evaluating data on chronic effects, the questions one wishes to
answer are as follow:
(1) What is the dose?
(2) What are the localized effects specific to the route of entry?
(3) What are the specific: characteristics of route of entry, bind-
ing, absorption, distribution and elimination of this chemical?
(4) What are the chronic systemic health effects?
(5) What are the characteristics of systemic absorption, distri-
bution and elimination of this chemical?
Although human data are the most desirable and present the most
secure basis for answering these questions, these data are frequently
anecdotal descriptions of chronic diseases stemming from long exposure
to partly identified mixtures of chemicals. The symptoms are frequently
thought to be associated with particular chemicals, but careful analysis
may show that the association has not been substantiated. In addition,
complex behavioral and dietary practices and the intrinsic genetic
individuality of each person complicate estimates of risk associated with
a particular chemical. Very few pathological states are unique and
the pathognomic symptom is rare indeed. Thus, for example, demonstration
of cardiomyopathy among individuals exposed to a particular chemical
does not necessarily implicate that chemical in the initiation of the
effect.
In the event that sufficient information is not available from
chronic human exposures, one passes to the next most acceptable data
groupings, subchronic exposures and acute exposures.
Data from occupational exposures to a chemical are frequently
very valuable. The exposures tend to be chronic, and the chemical agent
may be well identified. Documentation of these factors is very valuable.
These data need to be carefully scrutinized since the occupational
history of any individual may include many different exposures and other
predisposing factors need to be evaluated.
7-54
-------
- J
I
Ul
Ul
Arc theie
human
data?
Are there
experimental
data?
Structure
activity
relationships
Are there
epideiniological or
occupational
data?
Quantifiable
exposure?
Human
data on related
chemicals?
Assess
reliability
Is extrapolation
appropriate?
Is extrapolation
appropriate?
Extrapolation to humans
if possible
Only very
qualitative
statement
possible
Risk Estimates
FIGURE 7-7 FLOW CHART FOR GENERAL EVALUATION OF CHRONIC FUNCTIONAL DISORDERS
-------
I
Ul
a.
mmAji DATA ANJMA1. DATA
-1 „ CHRONIC
1 1
SKIN
CIIKONfC
1~\ Is there .|uantitation , .,_ Ts there ...,.i,,. Ir.-.ri,.- L_
|^ Of dOSe?
ZYes
Ho
Are there Are there
data on systematic
local ,. absorption
hind ing and d.ila from
luciahnl i sm?
J J'"
(Distribution |
Are there ami eliiulna-
lucal | t n>n d.ua?
effects? 1
, I Systematic I
health
|_ effects? 1
/^
Combine with acute
health effects from
other exposure routes -I
"L"
Is> extrapolation
i 1
Yes
ASSti
*~ of dose? |
/ Yes
r -^— ,
Are there 1
data on j Are there
lucal ,J syGtenaric
binding and ahaorjn ion
metahol ism? data?
-j..; ZE:
Distribution
Are Uvre andelimiua-
lo(al l Inn ii.ua?
health ~T "
effecis? |
1 — I :>y.sl IMIUL li:
1 lii-.il Hi
1 t-flecth?
Combine with acute 1
1 health effects from
1 olhi'i exposure routes 1
1
1 1
la extrapolation 1
appropriate? 1
• V HO
Yes | * 1
Obtain
better 1
(data 1
1
>S RISK
Nl
H
n
MATIIKMA'riCAL MODEL
SKIN
CHRONIC
1.
Is there a mo. 1. 1 ..
of f-kin^ 1
absorption ? j
Yea
I Is the Nn
1 model validated?
Yes
Yes Is extrapolation | Nn
appropriate?
WORST CASE ASSUMPTION
SKIN
CHRONIC
1 Worst case 1
*^jdnse asbumptinn I
v Can dose assumption
1Ki- b(, juat|fled» i H«>
(obtain
I better
Lddla
Worst case
abtnrpt j.-.n
assumpt ion
r*"
1
1 Can absorption No
assuiupt ion be
YeJ |~ « .
| I Obtain 1
I better 1
data 1
7-8. ,,,,-s, „,.,; ,.«„„„:„,. rc.u r.v.M HATUW OF DATA ON CHRONIC FUNCHUNA,. DISOKDLRS Rt
c FROM DERMA,. ABSORPTION
-------
Human accidental exposure data are not necessarily predictive of the
potential for chronic functional disorders. These exposures are not
often quantitated and they are most often single, acute events. There
is no certainty that the target organs that show pathological changes
resulting from a single high dose exposure will be the target organs
that are most often affected by a repeated lower dose. Epidemiologic
studies, in general, serve to corroborate the findings of more specific
animal or human work, but only infrequently define cause and effect
relationships.
Frequently the available data for humans are very much less useful
than would be expected. They provide descriptive clues suggesting
critical organ systems but are insufficient to characterize quantitatively
the relationship between exposure and effect.
Thus, animal data are frequently the only source of information
available for assessing potential for chronic human impairment. Some
of the problems in the use of animal data are relatively easy to
predict and are the same as those mentioned previously for other health
effects: extrapolation from laboratory animals to man must be accomplished
by use of a scaling factor to compensate for body weight differences
or surface area differences. In some cases, differences in life span
present problems. Behavioral patterns may introduce difficulties in
generalization between animal and human reactions to a chemical, as will
subtle anatomical differences. For example, the structure of the rodent
respiratory system is such that nose breathing is obligatory—in humans
this is not the case. The result is a significant difference with
certain chemicals in the exposure producing toxic effects due to the
protection of the rodent lung by the extremely efficient nasal
filtering systems. Recognizing anatomically determined differences
between man and test animal requires biological sophistication and a
very cautious approach to extrapolation.
The other major problem of extrapolation of animal data to humans
stems from possible species differences in metabolic pathways. If it
can be demonstrated that a chemical is absorbed, stored, metabolized
and excreted by the same pathways in animals and man, one can expecc
similar toxic consequences. If the pathways or conversion rates or ex-
cretion patterns are very dissimilar, one should expect different and
usually unpredictable toxic consequences, in which case, the animal
studies would be an inappropriate basis for making predictions of the
effect in man.
A common problem with animal data is that fairly often a response
is species specific. A treatment-related response may be evident in
rats and monkeys, but not in dogs, rabbits, etc. If biochemical path-
way data are unavailable to explain the diversified response, the'con-
servative approach is ordinarily espoused.
7-57
-------
The final pathway shown in Figures 7-7 and 7-8 is evaluation of
data for similar compounds. In a homologous series of chemicals, toxic
effects are somewhat predictable for one member of the series based on
known effects of other members of that series. In addition, some work
is being pursued which would allow prediction of toxic effect by analysis
of chemical functional groups (Kramer and Ford 1968). At this time,
however, such models do not constitute a validated approach to predic-
tion of human chronic functional disorders.
After the best available information has been assembled from human
and animal studies, the potential for a pollutant to cause chronic effects
can be estimated. The prediction of human effects from human data is ob-
viously more reliable than thai: from animal data. Although the human dose
may be poorly defined, one can be fairly sure that the signs are charac-
teristic of man. Some parameters, however, are not: easy to interpret.
For example, changes in organ weight, change in liver enzymes in the
blood, increase or decrease in a particular antibody, have all been re-
ported at some time. It may not be possible, however, to determine
whether these changes are significant precursors to organ dysfunction
or whether they are meaningless, random deviations.
Other changes noted in chronic studies are reversible, that is,
they will disappear after the termination of the exposures. Two types
of effects tend to be reversible:
(1) exposures may temporarily modify cell function but fail to
cause significant cell death; or
(2) exposures may cause significant cell death in an organ
capable of regeneration.
Many cells in the body are essentially in final form (i.e., differ-
entiated cells that cannot divide and be replaced), and in limited
supply. Chemical exposure that destroys these cells, is of the greatest
seriousness. Perhaps the most well known example is the heart. The loss
of cardiac muscle cells is irreparable and presents serious consequences
that have been well documented. Other types of cells can be easily re-
placed. This includes the fairly well-known replacement of nonspecialized
epithelium, connective tissue, and blood cells. It also includes the
more specific liver parenchyma! cell. The result is that a healthy liver,
which is damaged even severely by chemical insult, has a very good'chance
of complete recovery. Thus, in evaluating toxicity data, damage to organs
that have no potential for regeneration is far more significant than
damage to organs that undergo continual replacement or have a capacity
to regenerate when appropriately stimulated. An assessment of the ability
of organs to regenerate is shown in Table 7-16.
Another difference that determines the seriousness of chemical-
induced organ damage is the degree of redundancy in that particular
organ. The kidneys have sufficient structural excess to give entirely
adequate function, even if 50% or more is lost. The conducting svstem
7-53
-------
TABLE 7-16. TISSUE GROWTH CHARACTERISTICS: VARIOUS ANIMALS
i
Ln
<£>
^;?s!^!Mt!*S^^^
-._ iji. ••** w c«
• Wl**. wK.mUlMtf •U«
• n« Minv^I* "»» U b«« -/.- — -1 Ml« rtMo.al J M».»ua m.«>LrMW i. A*,'. ..,-,*,„.
»4 ml* |Mi.nmli,iic utr.utl c*lc.a.i«| *. J«4 kr E * M t...d*r|u l.ittllrd
uiuwi .ryttHtH,,,.. ittSri'»c* U7g«iy i.. rr»ifc>ni«u7Siri«~s £;ri7^;
rrKSTIm •>*.. D • lVr7d r ,IF.Y.
,' MailOD Jw«i-u»rj»tON «l wuii/f u*mc. accvri M • *nHuib.r ul I., toi* all.rt tat. f,((angl>unic libx*1 nl cat »*«•»• (•!* M It-tl da,
\mttnin rr*t..rrJ i,, 44^t ____ _ _
fS*»fi(iltl* rfipmn-i*f (MMIH h« Inert*•« U •»!* of
CAi.rt"M>>iori%,r.. i»...brri^7ifis^.iniE.7rTi;—
....ten. MM* <.„,„„..„ ik-rMrt,... ,„„. ™.^ch/m,
t ill* HjriMilruptiji i*u**.l bj, iMrtM* «f *«rcB^tMm
hi r.tta. (oftlc iSn^r**;. 1» *u, ««pwt«tfalf/«rlK
Ciiir-*!.. ^ik ^r.al f^ifci ^^fflLl''"1'*^^'0' ft"U^|^ff.^"J^^^^^
! k4.ck «rv«r.Ht* al
iln>*. cum,,!.!*! ,» u j,t „,,., tlttlnttmr 0, oo,;y,,. „ -|w.
to rr-'-"'/10 f "^? '^^T...«*». .rifirs—} ^rbViTr.^^n.T^ori;. ».fc d. .k*.i-;
EjJ^p^^ *:;* r ±: r;:;^^;;j^rAt.y" ^^
«t.l« by lod
me.*** W.U U U a
pon*T of dl*i.lii* •«• lannalto* BOaMbl. IrttM
Source: W. S. Spector (ed). Handbook of Biological Data. W. B. Saunders Co., Philadelphia 1956.
-------
of the heart has no excess capacity and no alternative. Again, moderate.
damage to the heart conducting system is likely to have more serious
consequences than moderate kidney damage.
When a summary of data shown to be relevant has been assembled it
may be possible to draw conclusions concerning acceptable human exposure
levels. It may also be possible to draw tentative conclusions about the
serzousness or reversibility of the predicted disease state. In most
cases, these conclusions will be tentative and will be the result of com-
binations _of human anecdotal data and animal experimental data substantiated
oy epidemiological evidence. Predictions from chemical structure or cell
culture studies are not likely to give reliable information.
7-60
-------
REFERENCES
Adamson, F.; Gilbert, D.; Perwak, J.; Scow, K.; Wallace, D. Identifzca-
tion and evaluation of waterborne routes of human exposure through food
and drinking water. Draft Report. Contract EPA 68-01-3857, Task 4.
Washington, DC: Monitoring and Data Support Division, Office of Water
Regulations and Standards, U.S. Environmental Protection Agency; Jan. 1980.
Aksoy M. Testimony before Occupational Safety Health Administration.
U.S.'Department of Labor; July 1977. (As cited by U.S. EPA 1978b)
Bridges, B.A. Some general principles of mutagenicity screening and a
possible framework for testing procedures. In: Environmental Health
Perspectives, Experimental Issue No. 6, December , 1973: pp. 221-227.
Bulsiewicz, H. The influence of phenol on chromosomes of mice Mus_
musculus in the process of spermatogenesis. Folia Morpho. (Warsz.)
36(l):13-22; 1977.
Crow, J.F. Impact of various types of genetic damage and risk assessment.
In: Environmental Health Perspectives, Experimental Issue No. 6,
December, 1973: pp. 1-5.
deSerres, F.J., and W. Sheridan, eds. The evaluation of chemical muta-
genicity data in relation to population risk. In: Environmental Health
Perspectives, Experimental Issue No. 6, December, 1973.
Freese, E. Thresholds in toxic, teratogenic, mutagenic and carcinogenic
effects. In: Environmental Health Perspectives, Experimental Issues No.
6, December, 1973: pp. 171-178.
Infante, P.F.; Rinsky, R.A.; Wagoner, J.K.; Young, R.J. Leukemia in
benzene workers. Lancet 2:76-78; 1977a.
Infante, P.F.; Rinsky, R.A.; Wagoner, J.K.; Young, R.J. Benzene and
leukemia. Lancet 2:867: 1977b.
Leek, I. Correlations of malformation frequency with environmental and
genetic attributes in man. In: Handbook of Teratology, Volume 3,
Comparative Maternal and Epidemiologic Aspects.
Wilson, J.G.; Eraser, F.C.; eds. New York, NY: Plenum Press; 1977:
pp 117-157.
Mahaffey, K.R. Quantities of lead producing health effects in humans:
Sources and bioavailability. Environ. Health Perspect. 19:285-295; 1977.
McCann, J.; Choi, E. ; Yamasaki, E.; Ames, B.N. Detection of carcinogens
as mutagens in the salmonella/microsome test assay of 300 chemicals.
Proc. Nat'l Acad. Sci USA 72:5135-5139; 1975.
7-61
-------
National Cancer Institute (NCI). Guidelines for carcinogen bioassay in
small rodents. Carcinogenesis Technical Report Series No. 1 DHEW
Publication No (NIH) 76-801, Bethesda, MD: NCI; 1976: p. 46.
Ott, M.G. et al. Mortality among individuals occupationally exposed to
benzene. Arch. Environ. Health 33:3; 1978. (As cited by US EPA 1978b)
Perwak, J.; Bysshe, S.; Goyer, M.; Nelken, L.; Scow, K.; Walker, P.;
Wallace, D. An exposure and risk assessment for copper. Final Draft
Report. Contract EPA 68-01-3857. Washington, DC: Monitoring and Data
Support Division, Office of Water Regulations and Standards, U.S.
Environmental Protection Agency; 1980.a.
Perwak, J.; Goyer, M.; Harris, J.; Schimke, G., Scow, K.; Wallace, D.
An exposure and risk assessment for trihalomethanes. Final Draft Report.
Contract EPA 68-01-3857. Washington, DC: Monitoring and Data Support
Division, Office of Water Regulations and Standards, U.S. Environmental
Protection Agency; 1980b.
Perwak, J.; Goyer, M.; Schimke, G.; Eschenroeder, A.; Fiksel, J.; Scow,
K.; Wallace, D. An exposure and risk assessment for phthalate esters.
Final Draft Report. Contracts EPA 68-01-3857, 5949. Washington, DC:
Monitoring and Data Support Division, Office of Water Regulations and
Standards, U.S. Environmental Protection Agency; 1981a.
Perwak, J.; Goyer, M.; Nelken, L.; Scow, K.; Wald, M.; Wallace, D. An
exposure and risk assessment for mercury. Final Draft Report. Contracts
EPA 68-01-3857, 5949. Washington, DC: Monitoring and Data Support
Division, Office of Water Regulations and Standards, U.S. Environmental
Protection Agency; 1981b.
Perwak, J.; Byrne, M.; Goyer, M.; Lyman, W.; Nelken, L.; Scow, K. :
Wood, M.; Moss, K. An exposure and risk assessment for dichloroethanes.
Final Draft Report. Contracts EPA 68-01-5949 and EPA 68-01-6017.
Washington, DC: Monitoring and Data Support Division, Office of Water
Regulations and Standards, U.S. Environmental Protection Agency; 1982a.
Perwak, J.; Goyer, M. ; Nelken, L; Payne, E. ; Wallace,, D. An exposure and
risk assessment for lead. Final Draft Report. Washington, DC: Monitor-
ing and Data Support Division, Office of Water Regulations and Standards,
U.S. Environmental Protection Agency; 1982b.
Report of a Committee of the European Environmental Mutagen Society.
Mutagenicity screening: general principles and minimal criteria. Mutation
Res. _53_:361-367; 1978.
Scow, K. ,; Thomas, R.; Wallace, 'D.; Walker, P.; Wood, M. An exposure and risk
assessment for pentachlorophanol. Final Draft Report. Contract EPA 63-01-
3857. Washington, DC: Monitoring and Data Support Division, Office of Water
Regulations and Standards, U.S. Environmental Protection Agency; 1980.
7-62
-------
Scow, K.; Goyer, M.; Payne, E.; Perwak, J.; Thomas, R.; Wallace, D.;
Wood, M. An exposure and risk assessment for phenol. Final Draft
Report. Contract 68-01-3857. Washington, DC: Monitoring and Data
Support Division, Office of Water Regulations and Standards, U.S. En-
vironmental Protection Agency; 1981.
Simms, H.S. Pathology of Laboratory Rats and Mice. Oxford Publications-
1967: pp. 733-747.
Snyder, C.A.; Goldstein, B.D.; Sellakumar, A.R.; Bromberg, I.; Laskin,
S.; Albert, R.E. The inhalation toxicology of benzene: incidence of
hematopoietic neoplasms and hematoxicity in AkR/J and C57BL/6J mice.
Toxicol. Appl. Pharmacol. 54:223-331; 1980.
Symons, J.M.; Bellar, T.A.; Carswell, J.K. DeMarco, J.; Knopp, K.L.;
Robeck, G.G.; Seeger, D.R., Slocum, C.J.; Smith, B.L.; Stevens, A.A.
National organics reconnaissance survey for halogenated organics. J.
Am. Water Works Assn. 67:634-646; 1975.
Thomas, R.; Byrne, M.; Gilbert D.; Goyer, M. An exposure and risk
assessment for trichloroethylene. Final Draft Report. Contract EPA
68-01-5949. Washington, DC: Monitoring and Data Support Division, Office
of Water Regulations and Standards, U.S. Environmental Protection Agency
1981.
U.S. Department of Agriculture (U.S. DA). Food consumption of households
in the United States. Spring 1965. Report No. 11, Food and Nutrient
Intake of Individuals in the United States. Stock No. 0100-1599.
Washington, DC: U.S. Government Printing Office; 1972.
U.S. Department of Agriculture (U.S. DA). Food consumption, prices,
expenditures. Agricultural economic report no. 138, Supplement for
1976. Washington, DC: USDA; 1976.
U.S. Department of Agriculture (U.S. DA). Nationwide food consumption
survey 1977-1978. Preliminary Report No. 2. Food and nutrient intakes
of individuals in 1 day in the United States, Spring 1977. Washington,
DC: Science and Education Administration, U.S. DA; 1980.
U.S. Department of Health, Education and Welfare (U.S. HEW). Community
Water Supply Study, Public Health Service, Environmental Health Service.
Bureau of Water Hygiene; 1970.
U.S. Environmental Protection Agency (U.S. EPA). National Organics
Monitoring Survey. Unpubl. Washington, DC: Technical Support Division
Office of Water Supply, U.S. EPA; 1978a.
U.S. Environmental Protection Agency (U.S. EPA). Estimation of population
cancer risk from ambient benzene exposure. Washington, DC: Carcinogen
Assessment Group, U.S. Environmental Protection Agency; 1978b.
7-63
-------
U.S. Environmental Protection Agency (U.S. EPA). Water quality criteria
documents availability. Federal Register. 45(231):79318-79384-
November 28, 1980.
U.S. Food and Drug Administration (U.S. FDA). Compliance program evalu-
ation FY 1974. Total diet studies. Washington, DC: Bureau of Foods
U.S. FDA; 1977.
Venugopal, B. and T.D. Luckey. Metal ToKJcity in Mammals. 2. Chemical
toxicity of metals and metaloids. New York, NY: Plenum Press- 1978-
pp. 24-32.
Williams, R.T. Detoxification Mechanisms. 2nd ed. New York, NY:
John Wiley and Sons, Inc.; 1959: pp. 390-409.
Wilson, J.G. Current status of teratology. In: Handbook of Teratology.
Volume 1, General Principles and Etiology. Wilson, J.G.; Fraser F.C •
eds. New York, NY: Plenum Press; 1977: pp. 47-74.
7-64
-------
8.0 EXPOSURE AND EFFECTS—NON-HUMAN BIOTA
8.1 INTRODUCTION
Although the principal focus of the exposure and risk analyses per-
formed for the Office of Water Regulations and Standards has been on humans,
it is important to consider the exposure of fish, other aquatic organisms,
and wildlife to waterborne pollutants and the adverse effects of pollutants
on these species for several reasons:
(1) they may be part of the human food chain and/or of economic
importance to man;
(2) they may be threatened or endangered species;
(3) because of a pollutant's critical environmental pathways or
fate, its environmental impact may be on non-human rather than
human receptors;
(4) assessments and regulatory recommendations of others may over-
look significant environmental effects unless the hazards and
risks to non-human species are considered; and
(5) they may serve as warnings or indicators of an environmental
problem when media concentrations are low or non-detectable.
As in evaluating human risks, exposure and effects for other
species should be considered together . Since the risk to a species is a
function of both the exposure to a pollutant in a sufficient quantity or
for a sufficient duration to elicit adverse effects, the risk will be small
despite the potential toxicity of the pollutant. Similarly, exposure to
high concentrations or quantities of a pollutant will not result in
significant risk unless adverse effects can result from these exposure
levels. In general, effects data for aquatic organisms and wildlife
are more readily available than information on exposure. A large number
of laboratory investigations have been conducted, correlations have
been developed for factors such as bioaccumulation, and field or model
ecosystem studies have been conducted for many pollutants and species.
For most priority pollutants, the results of acute and chronic bioassays
on a limited number of species are available and published in the EPA
Criterion Documents.
In evaluating environmental effects, there are problems inherent
in extrapolating from laboroatory data to field conditions. Unlike
controlled laboratory systems, the natural environment is complex "and
multi-leveled, subject to both regular and irregular changes in its
physical and chemical make-up. Habitats even of the same tvpe (e.a.,
cold-water streams) may differ significantly in certain important °
variables.
3-1
-------
Many of these variables may significantly increase or decrease the con-
centration of a pollutant that triggers an adverse effect. For example,
differences in pH, hardness, temperature and other aspects of water
chemistry may cause different effects from those detected in a simple
laboratory test at the same "total" pollutant concentration.
Evaluation and quantification of exposure has its own difficulties
since one must know both the population distribution and habits of the
species of concern, in addition to the pollutant's environmental distri-
bution. Materials balance and environmental pathways data provide in-
formation on pollutant concentrations and distribution. The exposure por-
tion of the risk analysis should determine as best as possible whether
there is exposure of receptors at those locations where the pollutant
is present and, if so, the extent, duration, and frequency of exposure of
important subpopulations. There are only limited data in the literature
on the population distribution of fish, other aquatic organisms, and
wildlife; their potential exposure to polluted water (drinking rates,
migration patterns in and out of polluted areas) is even less well known.
Therefore, in many cases, estimates or ranges of exposure will have to be
first developed or postulated and then compared with scattered observa-
tions (such as fish kill reports;) in order to see if they are feasible
and realistic.
Although in principle all aquatic species and other biota that are
exposed to polluted water should be examined in risk analyses, the effort
and amount of data required generally prohibits such a detailed analysis.
Therefore, the exposure and effects analysis can be concentrated on:
(1) sensitive species representative of each species category;
(2) species known to inhabit geographical regions or habitats
where the pollutant is present;
(3) species for which adequate effects or distribution data exist;
(4) aquatic organisms, particularly fish.
Historical information on the long-term discharge patterns of the pollu-
tant is important in order to examine the adaptation of resistant strains
in the species present or shifts in the specres composition of the local
community. Information on wildlife—both exposure and effects data—is
usually less common than for fish and other aquatic organisms. Also,
livestock are not usually considered in this part of a risk analysis'be-
cause they are rarely exposed to lethal levels of a pollutant. Instead,
livestock are much more likely to concentrate pollutant levels in their
tissue and the potential exposure to this accumulation is a human problem.
In a similar manner, accumulation in edible aquatic species is addressed
in the human exposure section, drawing upon monitoring and biological fate
data. In cases where there is some understanding of the relationship be-
tween body burden and toxic effects levels, bioconcentration may be
addressed in the biotic effects and exposure chapter. Otherwise it is
discussed under biological fate in the environmental pathways chapter.
8-2
-------
8.2 GOALS AND OBJECTIVES
8.2.1 Exposure Analysis
The goals of analyzing the exposure of non-human species are to
determine or estimate the significant exposure routes and the extent to
which aquatic organisms are exposed to pollutant concentrations in water
sediment and other organisms, and the extent to which terrestrial organisms
are exposured to pollutant concentrations in soil, air, water and/or
other organisms. Exposure routes include ingestion, inhalation, and
dermal absorption. The extent can be defined in terms of the length
of time during which populations are exposed, the geographical area in
which exposure occurs, and the degree of exposure of an individual or
community. The degree of exposure may be expressed as the concentrations
to which the organisms are exposed or their daily intakes times
and absorption efficiency.
Ideally, the results of the exposure analysis for non-human biota
include:
• Identification of geographic areas with pollutant concentrations
in the water or other significant media (sediment, soil) high
enough to have deleterious effects on biota in order to identify,
geographically, subpopulations at risk. (Monitoring data may
reveal actual areas, and potential areas may be indicated by the
presence of sources of pollutant releases.)
• Identification of communities or particular species—size or
number, location (geographical or habitat-specific)—exposed to
the pollutant.
• Evaluation of behavior patterns (e.g., migratory, reproductive,
age-linked) of biota that may increase or decrease the potential
for exposure.
• Identification of time-dependent patterns of pollutant availa-
bility (persistence, seasonal fluctuations, etc.) and comparison
with species activity patterns.
• Evaluation of the existence of mitigating or exacerbating environ-
mental parameters that can affect pollutant toxicity and the likeli-
hood of their presence in areas or habitats in which environmentally
significant pollutant concentrations are known or estimated.
8.2.2. Effects Analysis
The objectives of the effects portion of risk analyses for non-human
biota are:
3-3
-------
(1) Identification of those concentrations or ranges of concentrations
at which a pollutant may have deleterious effects on aquatic and
terrestrial organisms.
(2) Identification and evaluation of these effects—acute, chronic,
reproductive—as a function of exposure levels, time, etc.
(3) Identification of factors that influence the availability and
degree of impact of the pollutant on biota.
The results of the effects analysis should be compatible with the
results of the exposure analysis in terms of how the levels are quantified
so that the risk to aquatic and terrestrial organisms can be ascertained.
8.3 APPROACHES AND METHODS
8.3.1 Overview
In undertaking an exposure and effects analysis for non-human species,
one could begin either with developing an understanding of exposure of
aquatic or terrestrial organisms and then consider the effects of such
exposure, or begin with effects and then consider exposure.
In the first approach, one would rely primarily on the results of the
materials balance, monitoring data, and the environmental pathways analysis
to identify the media and types of habitats in which exposure can occur.
their geographical distribution, and the concentrations and durations
associated with exposure, and then seek to establish the species or
communities in those areas most likely to be exposed. Effects analysis
would then focus on selected species or communities., evaluating the
potential acute or chronic effects resulting from the estimated exposure
levels. This approach has the advantage of limiting detailed consider-
ation of effects to those species and populations for which exposure is
anticipated or known, thereby limiting the scope and effort of the
effects and analysis. The disadvantage, of course, is that one may only
give detailed consideration to exposure of certain populations for which
significant effects of the pollutant are not likely to occur, or to
organisms for which the effects of the pollutant are unknown.
Alternatively, in beginning with effects analysis, one first identifies
toxic concentration levels by examining a number of laboratory studies
for a range of species and then seeks to determine the geographical areas
and real situations where the sensitive species or communities may be ex-
posed to levels sufficient to give harmful effects. This method also has
advantages and disadvantages similar to those described above.
For practical reasons, it would seem appropriate to use both
approaches concurrently, with the goal of quickly focusing on the exposure
conditions of significance and on sensitive organisms. However, as
3-4
-------
indicated in the introduction to this section, data on the harmful effects
of pollutants are more readily obtained by traditional methods of litera-
ture review and analysis, whereas exposure analysis may require a more
lengthy analysis and inputs from monitoring, fate and pathways, and
materials balance studies (tasks that may be proceeding concurrently).
Therefore, most exposure and effects analyses for non-human biota are
likely to begin with development of an understanding of potential effects
and then proceed to development of understanding of exposure situations.
Figure 8-1 gives a schematic representation of the methodology used
for this analysis. Note the close interaction with the other portions of
the risk analysis.
8.3.2 Effects Analysis
8.3.2.1 Data Collection and Preliminary Data Review
The first step in analyzing aquatic effects is to collect readily
available data on the pollutant under study. The amount and type of
data readily available depends upon the pollutant being examined. If
the pollutant is well known and effects have been documented, priority
can be given to review articles and data compilations [such as the EPA
Water Quality Criterion Documents (e.g., U.S. EPA 1980a, b)]; however,
this reliance on secondary sources must be complemented by review of
original publications to clear up errors and contradictions between
studies that may arise. If the effects of the pollutant have not been so
well studied and reviewed, then more effort must be devoted to search for
published data. In addition, persons currently conducting research on
the pollutant may be contacted.
Data should be collected from both laboratory studies measuring the
effects of the pollutant on various aquatic organisms and field investi-
gations or case studies documenting actual effects of the pollutant in
the environment. Several information sources can be used:
• EPA sources—materials in the MDSD priority pollutant file;
e.g., criterion documents, NRC reviews, fish kill data, EPA-
published reports from field laboratories, etc.
• Computerized literature search in conjunction with the human
effects studies using TOXLINE, Chemical Abstracts, Pollution
Abstracts, Bio Abstracts, etc.
• Formal literature search—this is a second stage search, which
involves retrieval of pertinent literature cited in the first
sources obtained, and hand search of selected journals, e.g.,
Pesticide Monitoring Journal, Environmental Contamination and
Toxicology, etc., which are likely to contain information on
environmental pollutants.
-------
EFFECTS ANALYSIS
EXPOSURE ANALYSIS
Input From I
Fate and I
£u t hways_ j
Oi
Data Collec-
tion and Pre-
liminary Data
Review
I
Identification
of Sensitive
Species and
Their Ranges
Critical Data
Review and
Tabulation
Input from!
Monitoring
Inputs From
Materials
Balance
Identification
of Areas Where
Concentrations
are Expected or
Measured t" be
High
Summary of
Effects
Concentra-
tions and
Influential
Parameters
I Input From
Fate and
Pathways
Analysis
Identification
of Factors
Modifying Avail-
ability of
a Pollutant at
a Measured
Concentration
j
Identification of Loca-
tions or Categories of
Locations Where Risk to
Aquatic Organisms is
Likely to Occur
Input
From
Monitor-
FIGURE 8-1 FLOW CHART OF METHODOLOGY FOR EFFECTS AND EXPOSURE ANALYSIS FOR NON-HUMAN BIOTA
-------
If information on effects of a pollutant is not available, the
effects analysis must be bypassed because, in general, no method exists
for estimating toxic effects. If information on a structurally similar
chemical is available, it should be examined but not considered to be sur-
rogate data for the risk analysis since the relationship between structure
and toxicity is relatively unknown, making tentative any extrapolation
from one chemical to another. In the absence of effects data, the re-
search required to assess the toxicity of the pollutant (using standard
aquatic species and testing procedures) should be recommended.
8.3.2.2 Critical Data Review and Tabulation
The second step of the effects methodology is to review critically
the data collected and tabulate effects concentrations, in addition to
consideration of the variables influencing these values.
Before the effects data are compiled, however, monitoring and fate
and pathways analysis results (or preliminary results) should be
reviewed. First, one should consider pollutant environmental concen-
tration ranges available from the monitoring section to determine
whether the pollutant is likely to have:
(1) no effect on aquatic organisms;
(2) some effects on certain sensitive species;
(3) effects on most species; or
(4) effects on all species, as a first approximation.
Second, a review of effects data and the initial results of fate and
pathways analysis can identify critical parameters influencing availa-
bility of the pollutant to biota (e.g. ,'pH, hardness, temperature).
These considerations will help limit the scope of and define the remain-
der of the effects analysis.
In the data review, pollutant concentrations that have been
reported to have lethal and sublethal effects on aquatic organisms
are examined. Either previously compiled data or the results from the
publications collected in the literature search, organized into tables
that present species and effects levels in order of increasing concen-
tration, are used. Each result must be reviewed for its scientific
validity and data with serious flaws (e.g., faulty control, death of
test subjects due to other causes, or lack of replication) rejected,
unless no other information is available; in which case, the weakness
should be highlighted.
The data are typically divided into several categories to facilitate
comparison: fish and aquatic invertebrate species, freshwater and
saltwater species, marine and estuarine species, lethal and sublethal
effects, and chronic and acute effects. Important parameters influencing
the effects at different concentrations are reported, when available,
3-7
-------
for each experiment. These parameters include pH, temperature, water
hardness, type of bioassay (static or flow-through), type of water and
time of exposure. Tables 8-1, 8-2, and 8-3 are examples of effects
data tabulated from risk assessments for phthalate esters, zinc, and
mercury (Perwak, _et_ al. 1981a, 1980, 1981b).
Exposure of aquatic organisms to the pollutant via gill absorption
is the major focus of the effects analysis for non-human biota. However,
depending on the availability of information and relevance to the pollu-
tant of concern, several other categories of effects must be considered
in this portion of the analysis:
• Toxicity of the pollutant to aquatic organisms through the
route of ingestion.
• Toxicity to terrestrial species, both (1) plants through root
uptake of pollutants in soil and aerial deposition and (2)
animals (usually avian and mammalian wildlife species) through
ingestion of contaminated biota, water or dermal contact with
soil.
_ Data on these subjects are developed and presented in a manner
similar to that for aquatic organisms.
8.3.2.3 Summary of Effects
The environmental factors that potentially affect uptake and
toxicity ot the pollutant are discussed either, through summarising
research indicating key factors (e.g., species groups, water hardness
duration) influencing the toxicity of the pollutant or, if that
is not possible, by compilation of results of a number'of separate studies
in which important factors have riot been controlled for. The importance
of these factors in the degree of impact of the pollutant on the environ-
ment and the likelihood of their existing at sites where significant
concentrations of the pollutant are found is discussed.
In many cases the information on the effects of the pollutant is
insufficient to prioritize available data relative to their relevance to
risk analysis. However, if possible, it is practical to consider effects
in the following general order. When data are available, chronic effects
(which usually occur at lower concentrations) have priority over acute
effects for persistent pollutants to which long-term exposure is likelv
For short-lived pollutants (e.g., highly volatile compounds) focus
should be placed on acute effects; however, if releases are on a continu-
ous basis, then chronic effects should also be considered. Effects on
fish and shellfish have priority over effects on other invertebrate
species because of their greater potential for ingestion bv humans.
8-8
-------
TABLE 8-1. EXAMPLE OF ACUTE EFFECTS DATA FOR FRESHWATER FISH--PHTHALATE ESTERS
Exposure Through Mater
Conctntrttlon
Comound (m/1 )
Di-o-butvi • 731
phchalate
1.2
1.3
2.91
6.47
10.0
Di(2~schylhexyl) -005
phchalace
.014
> 10.0
100.0
Butvlbeazyl *3-3
phchaiace
445.0
Diechyl 29-6
phchalata
96.2
DiaeciivL 49.5
phehaiatc
it 58.0
Species
_E_^M.S»
Blueglll
(lepc«rls aacrocM rui 1
•
Fathead NlnnoH
(Plaephales prom las)
Channel Catfish
(Ictalurus punctatus)
Rainbow Trout
(Salao oalrdnerl)
Blueglll
(LepQBTts. macrochlrusl
Rainbow Trout
(Sal mo galrdnerD
Fish1
Blueqlll
(Lepomli inacrochl rus )
Blueglll
(l»poiii
-------
TAWJi 8-2. EXAMPLE OF CIIRONIC/SUIU.ETHAL DATA FOR FRESHWATER FISH—ZINC
oo
Cone .
(ppb)
•>.f>
")]
106
180
187
260
640
852
Species Compound
Rainbow trout ZnSO,
(SujLjno galrd_neri)
Flagfish adults (females) ZnSO
(.lordanella florldac-) *
Fathead minnow Zn+f
(Plinephales prpmelas)
Fatliead minnow ZnSO
4
Chinook salmon Zn++
Kalnbow trout ZnSO
4
Hi'uuk trout Zn-H-
llardncss Tent
(mg/1) Duration Effects
13-15 20 mln. Threshold avoidance
level
44 100 days Crowth rt-diu-i-d
'•6 ? Effect on growth,
duct ion In 11 fe-
cycle test**
2 42 days Chronic bioassay
6. A mortality
3-1(> 6.9 mortal Ity
44 ? Effect on growth
survival, or repro-
duction In life-cycle
test**
Source***
Sprague (1968)
Speh.ir (1976)
lie-no it and
Hal combe*
Brunga (1969)
Chapman (1978)*
Sinley et al.
(1974)
,llolcooie et al.
(1978)*
**
The value represents the geometric mean of the levels at which there effects are observed. In the case
of embryo-larval tests the geometric mean la divided by 2 to obtain a value comparable to life-cycle studies.
***Sei- sourcu indicated below for references.
Source: I'erw.ik, .1. et ul . An exposure and risk assessment for zinc. Final Draft Report. Contract
T.l'A 68-01-3857. Washington, DC: Monitoring and Data Support Division, Office of Water
(Manning and Standards, U.S. Environmental Protection Agency; 1980.
-------
TABLE 8-3. EXAMPLE OF LOWEST REPORT MERCURY EFFECTS DATA FOR AQUATIC ORGANISMS
Form of
Mercury
Freshwater
Invertebrate
Lowest Reported Effect Level (ug/1)
Freshwater
Fish
Marine
Invertebrate
Marine
Fish
Inorganic 0.9a (Daphnia
magna)
(Salvelinus
fontinalis)
5 b (Pseudocalunus
mi nutus)
10 (Fundulus
heteroclitus)
oo
I
5 (Daphnia
raagna)
33.Ob (Salmo 3.6C (Mysidopsis
gairdneri) bahia)
200 (Fundulus
heteroclitus)
Organic O.la (Daphnia 0.04b (Salmo 1.2a (Mysidopsis
magna) gairdneri) bahia)
125a (Fundulus
heteroclitus)
5.1° (Salmo
gairdneri
150 (Gammarus
duebeni)
a
chronic value
Sublethal effect
c
Acute value
Source: Perwak, J., et al. An exposure and risk assessment for mercury. Contract 68-01-5949.
Washington, DC: Office of Water Regulations and Standards, U.S. Environmental Protection
Agency; 1981.
-------
Lethal effects and reproductive impairment effects have priority over
sublethal effects—e.g., avoidance behavior and physiological changes—
because of their known adverse impact on receptor populations.
To summarize the effects section, individual species, species groups,
and age groups that are most sensitive to the pollutant are distinguished,
along with the importance of other environmental parameters, in order to
identify specific subpopulations of aquatic organisms; that are likely to
be at higher risk.
8.3.3 Exposure Analysis
8.3.3.1 Introduction
Analysis of the exposure of aquatic organisms is generally more
qualitative than quantitative in nature. This is due to the difficulty
of reliably estimating biota populations and their distribution on a
national, or even regional, scale. Without this information, quanti-
tative exposure models are not useful. As described previously, exposure
analysis depends on the input of data from other portions of the risk
analysis.
Effects data point ouc the pollutant levels with significant bio-
logical impact and, therefore, better define the boundaries of the ex-
posure analysis for the pollutant. This can also aid in organizing
monitoring data retrieval from large data bases (e.g,, STORET). Effects
data are also useful in the initial part of exposure analysis in identi-
fying those species and their habitats on which to focus efforts, as
suggested by results indicating most sensitive species or identifying
environmental variables conducive to pollutant availability.
Monitoring data are very important to exposure analysis. Pollutant
concentrations in different environmental media or good estimates of
concentrations are required before effects data obtained in the labora-
tory can be evaluated for their relevance to natural conditions. Without
knowledge of a pollutant's environmental concentrations, the risk to
aquatic species cannot be estimated; effects data only satisfy one-half
of the data requirements. For this reason, monitoring data should be
collected with the sensitivity of non-human species in mind in order tc
facilitate exposure analysis. The focus can be on significant exposure
pathways (e.g., surface water) and pollutant concentrations (e.g., greater
than a minimum effects level).
Environmental fate and pathway information is significant in the
exposure analysis when used in conjunction with monitoring data. Under-
standing of the pollutant's behavior in the environment can indicate the
biological availability of the pollutant in environmental media to which
biota are exposed. For example, if the fate data, indicate a low free
fraction of a pollutant at high water hardness or a tendency for adsorp-
tion, this information can be used in the qualitative interpretation
8-12
-------
of pollutant concentrations reported in monitoring programs as "total"
levels, i.e., in assessing pollutant availability to and potential
effects on aquatic organisms. Areas that are likely to have the
environmental properties that might be conducive to pollutant availability
(e.g., areas with low water hardness, low concentrations of complexing
agents) can be identified through information from the fate analysis.
If more specific monitoring data (e.g., dissolved or available concen-
trations) are available, these can be used to corroborate inferences
about types of habitats with a high exposure potential.
In the absence of monitoring data, fate and materials balance in-
formation may permit estimation of ambient concentrations through the use
of fate models (e.g., EXAMS, fugacity models).
Materials balance information can also be used together with monitor-
ing data to identify geographic areas where exposure of aquatic organisms
is likely to exist due to the presence of releases. Because of its
general nature, it is more useful for indicating areas (regions, river
basins, etc.) likely to have high pollutant concentrations relative to
other comparably sized areas than for targeting specific potential
problem sites.
Since the focus of the risk analysis process is often national
rather than local, the monitoring data collected and used represent
large areas (usually no smaller than a minor river basin). Therefore,
the monitoring data are not likely to directly corroborate the
fate and materials balance analyses due to differences in scale or
rounding off. As a result, quantitative analysis (e.g., implementing
speciation models on total metal concentration) is not possible because
of requirements for site-specific input data.
Other types of information useful in analysis of non-human exposure
as a confirmation of monitoring, fate, and materials balance data, in-
clude fish kill data and site-specific investigations of the effects of
the pollutant in the environment. Ideally, fish kill data provide in-
formation on types of sources, locations, and temporal distribution of
pollutant concentrations actually observed to have lethal effects on
biota in the field. Information that would help interpret these results
is usually not available. Field studies are likely to be more detailed
and to measure parameters influencing the pollutant's behavior at the
site of investigation but, because of the specificity of each study and
the usual short time span of investigation, the generality of these
studies is limited. It is unlikely there will be studies on all eco-
systems or biotic communities of significance with respect to a particular
pollutant. Despite the inherent weaknesses in and specificity of these
data, they may serve to confirm or tie together independent pieces of
information from other sections of the risk analysis.
The following sections describe the steps of the exposure analysis
depicted in Figure 8-1.
3-13
-------
8.3.3.2 Identification of Sensitive Species
If sufficient data from the effects analysis are available, the
first step in exposure analysis is to identify those species most sensi-
tive to the pollutant (e.g., salmonoids) and to determine their range
(nationally distributed or locally found). In many cases, however
either data are not available for many species or the differences in
sensitivity among species is not great enough (perhaps due to a very
limited data base) to justify 'separate treatment of sensitive members.
When this is true, concentrations of the pollutant in water are identified
that are likely to cause deleterious effects on each group of aquatic
organism (e.g., marine fish, freshwater invertebrate). Table 8-4 is an
example of how these data may be organized (Scow et al. 1981a).
8•3•3•3 Identification of Areas with Expected or Measured High
Concentrations —— a_
This second step is approached in one of two ways, again depending
on availability of data. The first approach is to use monitorin* data
fior example, from the STORET data base, to determine the location°of areas
with concentrations equaling or exceeding the effects levels set in the
Preceding step. Specific locations (e.g., a particular minor river
basin) or larger areas of the U.S. (the Northeast) may be identified
according to the distribution pattern of the particular pollutant
Figure 8-2 (Perwak et al. 1980) is an example of one method by which
monitoring data may be organized for exposure analysis. If few monitor-
ing data are available, other information on the distribution of primary
sources and usage patterns (from materials balance or environmental path-
ways analysis) or fish kills (see Table 8-5, Scow et. al. 1981b) can be
used to identify locations where high pollutant concentrations mav be
found.
8-3.3.4 Identification of Factors Modifying Availability
_For certain pollutants, data from the environmental pathways analysis
can indicate qualitatively what fraction of pollutant concentration in
water is actually available to aquatic organisms. For example, for many
heavy metals, factors such as hardness and PH may significantly alter the
effective concentrations causing deleterious effects. Monitoring data
can be better interpreted through understanding these influential variables
even in a qualitative way. The STORET data base includes the distribu-
tion of water hardness and other chemical parameters on a national basis
and these characteristics can be combined with data on the pollutant
concentration distribution. As an example, Table 8-6 lists the major
river basins that meet zinc concentration and hardness criteria indicat-
ing a risk to aquatic biota (Perwak et al. 1980). Since the methods with
which data are aggregated regionally for each factor are not always
equivalent, quantification of fie relationship through techniques'such as
regression analysis has noc been possible up to this time.
8-14
-------
TABLE 8-4,
Silver
Concentration
EXAMPLE OF RANGES IN EFFECTS LEVELS
FOR AQUATIC BIOTA—SILVER
Effect
<0.1 ug/1
0.1-1.0 ug/1
1.2 ug/1
2.3 ug/1 (maximum at
any time)
1-10 ug/1
13 ug/1
10-100 ug/1
100-1000 ug/1
>1000 ug/1
No effects reported for any species
Chronic effects on most sensitive freshwater
fish (mortality of trout in soft water) and
invertebrates (mayfly LCso). Acute effects on
most sensitive marine invertebrates. (Sea
urchin egg development.)
EPA criterion to protect freshwater aquatic
life at water hardness of 50 mg/1 as CaC03.
EPA criterion to protect saltwater aquatic
life from acute toxicity.
Acute effects on most sensitive freshwater
vertebrates (guppy) and invertebrates (daphnia),
The typical concentration range for chronic
effects on freshwater vertebrates and inver-
tebrates. Chronic effects on most sensitive
and typical marine invertebrates.
MA criterion to protect freshwater aquatic
life at water hardness of 200 mg/1 as CaC03.
Most reported effects levels for freshwater
vertebrates and invertebrates fell within this
range. Chronic effects (growth retardation)
on freshwater algae. Typical range for acute
effects on marine invertebrates.
Includes the highest concentration reported to
cause acute and chronic effects on marine
invertebrates. (Shrimp LCso at 262 ug/1 and
no spawning at 103 ug/1). Sublethal effects
noted for marine algae in 4 days.
Includes the maximum reported concentration
causing acute effects on freshwater invertebrates.
(1400 ug/1. reported for rotifer LCso) and
chronic effects on algae (freshwater) (toxic
at 2000 ug/1).
Source: Scow, K. et al_. An exposure and risk assessment for silver.
Contracts EPA 68-01-5949, 6017. Washington, DC: Office of Water
Regulations and Standards, U.S. Environmental Protection
Agency; 1981.
3-15
-------
TABLE 8-5. EXAMPLE OF DATA ON FISH KILLS—PHENOL
Dale
Water Body
Location
Number
Killed
Source
Co
I
cr>
5-25-71 Roaring Brook
6-8-71 - Casey Fork Cr.
8-6-71 Tunungwant Cr.
8-6-71 Tunugwant Cr.
8-6-71 Allegheny R.
1971 Ohio R.
1971 Milwaukee R.
1972 Severn Run (Branch)
1973 Kingsland Cr.
5-18-74 llardisty Pond
5-22-74 Banmers Pond
6-18-74 Red Clay Cr.
6-19-74 New Haven Harbor
7-29-74 Black Warrior R.
6-17-76 Black Rock Harbor
6-22-76 Bridgeport Harbor
11-17-76 Great Miami R.
r>/6 Bear Cr.
5-10-77 Hebble Cr.
6-1-77 Sanders Br.ineh
B-2-77 Beaverdam Cr.
Glastonbury, CT
Mt. Vernon, IL 6.000
Bradford, PA 53,000
NY, near Bradford, PA 45,000
Irvine Mills, NY 62,000
New Martinsville, WV 5,000
Gratton, Wl 1,500
Odenton, MD 100
Lyndhurst, NY 5,000
Southbury, CT 550
Naugatuck, CT 010
Newcastle, DE 2,000
New Haven, CT 20,000
Tuscaloosa, AL 10.700
Bridgeport, CT 25,000
Bridgeport, CT 20,000
Ohio 0.848
Fail-view. PA 28.000
Greene Co., OH 1.000
Hampton, SC Tot ;i)
Damascus, VA 150
High phenol, Zn, Cu in fish tissues
No toxics measured in water
Wood preservation
Discharge for chemical industry
in area
From Bradford, PA
From Bradford, PA
Phenols from nearby chemical industry
Phenols, oil from storm sewer (?)
Phenols from plastics industry
Phenolic discharge from chemical industry
Mixed solvents, heavy oil, and phenol
Asphalt and phenol
Haveg Industry phenol spill
High phenol, Al, pll, BOD, and collforin
17,000-21,500 Ib phenol spill by
Reichhold Chemical
Chemical, textile, metal industries, and
POTW nearby: high phenol, Cu,and Zn
in fish tissues
Discharges from POTW, power plant
Metal and cyanide production
Phono I;;, rvanidrs from agric. operations
"Government operations"
Railway phenol spill
Discharge by American Cyanamid
Source: Scow, K. et al. An exposure and risk assessment for phenol. Final Draft Report. Contract
68-01-5949. Washington, DC: Office of Water Regulations and Standards, U.S. Environmental
Protection Agency; 1981.
-------
GO
Source:
STORE! SYSTEM
2IHC: OBSERVATION / ACUTE CRITERIA
75TH PERCENT HC5
ft . ™ .0 ;«::
ft j ox: ?3 j 03;;
M ' i 033:
iiO'f.t -1 .00;;;: ?!• ;;o 03 r:iil'ln(1
, J. et al. An exposure and risk assessment for zinc. Final Draft Report
of
Contract
FKJURE 8-2 EXAMPLE OF CRAl'HIC PRESENTATION OF OVERLAP BETWEEN OBSERVED
CONCENTRATION AND WATER QUALITY CRITERION FOR THE PROTECTION
OF AQUATIC LIr-'E-- ZINC
-------
TABLE 8-6. EXAMPLE OF CONSrnERATION OF UNAVAILABILITY OF
OBSERVED CONCENTRATION OF ZINC IN SURFACE WATER
CO
I
GO
River Basin
Major/Minor Name
1/9 Merrlmack R.
1/14 Presumpcot R. 4, Casco Bay
1/24 Lake Champlaln
2/8 Delaware R. - Zone 4
2/15 Rappahannock & York Rivers
5/2 Monongahela R.
5/3 Beaver R.
5/7 Kanawlta R
5/13 Miami R.
5/21 Ohio R., main stem & trihs
6/3 Cuyahoga R.
6/13 Detroit
7/2 Hudson Bay, Rainy River (23/02)
7/3 Upper portion, upper Mississippi R.
7/6 Lower portion, upper Mississippi R.
7/12 Mississippi, Salt Rivers
7/16 Fox R.
7/19 Meramec R.
8/3 Menominee
8/24 Green Bay, W. Shore
8/49 Calumet-Burns Ditch Complex
9/i4 S. Central Missouri R.
9/7 Big Sioux R.
9/12 Lower Missouri R.
Zinc
N
126
24
10
305
296
331
25
338
86
257
21
9
5
135
189
9
24
1 f\
H£
50
42
42
70
Mean Zn
>120 ppb
A
*
*
A
*
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
^50% of ppb
observations
>120 ppb Zn
A
A
A
A
A
of ppb
observations
>300 ppb Zn
^50% of hardness
Measurements <5Q ppm
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
37
Source:
aml »sk assessment f°r zinc. Final Draft Report. Contract
-------
8 • 3.3.5 Identification of Locations in Which Risk to Aquatic Organisms
is Likely to Occur
The final step in the biotic effects and exposure analysis is to
summarize the results of the two sections to indicate areas where a
significant exposure potential exists. Areas may be regional (e.g., the
Northeast) or categorical (e.g., at the mouths of major rivers) depend-
ing on the data base and other factors discussed previously. In addi-
tion, exposure may vary temporarily if discharge patterns are seasonal,
if certain age-groups of a species are more sensitive (these also vary
seasonally), or if certain seasonal environmental processes (e.g., spring
rains) increase the availability of a pollutant.
8.3.4 Terrestrial Effects and Exposure Analysis
Although most of the information described here is concerned with
waterbome routes of exposure to priority pollutants, terrestrial systems
should also be considered for those situations in which a pollutant" is
directly applied to plants (e.g., as a herbicide, seed fungicide) or for
pollutants that are likely to be distributed on or be disposed of in the
soil and expose plants through root uptake. Effects on plants, as well
as those on higher members of terrestrial food chains (e.g., pheasants)
should be considered when data are available. The effects and exposure'
analyses for these terrestrial species are usually very brief, at most
indicating sites (e.g., vicinity of manufacturing plants, landfills)
where exposure of terrestrial biota may occur and the range of possible
effects. Discussion of environmental factors determining exposure
levels (e.g., leachability of pollutant, soil pH) should be included when
applicable.
8-19
-------
REFERENCES
Perwak, J; Goyer, M.; Nelken, L.; Schimke, G.; Scow, K.; Walker, P.;
Wallace, D. An exposure and risk assessment for zinc. Final Draft
Report. Contract EPA 68-01-3857. Washington, DC: Monitoring and Data
Support Division, Office of Water Regulations and Standards, U.S.
Environmental Protection Agency; 1980.
Perwak, J.; Goyer, M.; Schimke, G. ; Eschenroeder, A.;, Fiksel, J.;
Scow K.; Wallace, D. An exposure and risk assessment for phthalate
esters. Final Draft Report. Contracts EPA 68-01-3857, 5949. Washington.
DC: Monitoring and Data Support Division, Office of Water Regulations
and Standards, U.S. Environmental Protection Agency; 1981a.
Perwak, J.; Goyer, M.; Nelken, L.; Scow, K.; Wald, M,.; Wallace, D.
An exposure and risk assessment for mercury. Final Draft Report.
Contracts EPA 68-01-3857, 5949. Washington, DC: Monitoring and Data
Support Division, Office of Water Regulations and Standards, U.S.
Environmental Protection Agency; 1981b.
Scow, K.; Goyer, M.; Nelken, L.; Payne, E.; Saterson, K., Walker, P.;
Wood M.; Cruse, P.; Moss, K. An exposure and risk assessment for
silver. Final Draft Report. Contracts EPA 68-01-3857, 5949 and EPA
68-01-6017. Washington, DC: Monitoring and Data Support Division,
Office of Water Regulations and Standards, U.S. Environmental Protection
Agency; 1981a.
Scow, K.; Goyer, M.; Payne, E. ; Perwak, J.; Thomas, El.; Wallace, D.;
Wood M. An exposure and risk assessment for phenol. Final Draft
Report. Contract EPA 68-01-3857. Washington, DC: Monitoring and Data
Support Division, Office of Water Regulations and Standards, U.S.
Environmental Protection Agency; 1981b.
U.S. Environmental Protection Agency (USEPA). Ambient water quality
criteria for chlorinated benzenes. Report No. EPA 440/5-30-028.
Washington, DC: Criteria and Standards Division, Office of Water
Regulations and Standards, U.S. Environmental Protection Agency; 1980a.
U.S. Environmental Protection Agency (USEPA). Ambient water quality
criteria for chlorinated ethanes. Report No. EPA-440/5-80-029.
Washington, DC: Criteria and Standards Division, Office of Water
Regulations and Standards, U.S. Environmental Protection Agency; 1980b.
8-20
-------
9.0 RISK CONSIDERATIONS
9.1 INTRODUCTION
In the previous sections of this report, goals and objectives, methods
and approaches have been presented for evaluating the characteristics of
a pollutant—the sources of its release to the environment, its pathways
and distribution in the environment, and its exposure and effects on humans
and other biota. Each of these components is important in its own right;
yet for the regulatory agencies, as well as the public, it is essential
to integrate them in order to establish, as best possible, the current or
potential impact of the pollutant on man and his environment. Thus one
needs to establish a "bottom line" for the analysis—how much of a problem
is the pollutant, what are the risks associated with each increment in ex-
posure to the pollutant, and how do the risks compare with those of other
pollutants? The answers to these questions place risks and problems asso-
ciated with the pollutant in perspective, so that they can be evaluated
and acted upon, if necessary, by the parties involved.
In other research on the risks of environmental pollutants, the term
"risk assessment" has been given two general interpretations. First, it
has been used to connote a broad assessment of the overall risk associated
with a pollutant, including risk to humans, fish and wildlife. Second, it has
had a narrower meaning, namely the quantitative human health risks asso-
ciated with a pollutant, often as a result of documented or estimated
carcinogenicity or mutagenicity (e.g., the extrapolation of laboratory
animal data on carcinogenicity to humans). In this report, the term "risk
considerations" is used to signify the evaluation and integration of the
information on the pollutant for the purpose of yielding an understanding
of the nature and extent of risks to humans and other biota associated
with the pollutant.
More specifically, the "risk considerations" portion of the risk
assessment should answer the following types of questions:
• Does the pollutant cause a significant increased health risk to
the general human population?
• Does the pollutant cause significant increased risks to general
populations of fish, shellfish, wildlife and other aquatic species?
• What is the nature of the increased risks? Can the risks be
quantified? What are the risks to the general population groups?
• Are there identifiable subpopulations based on geography, age,
sex, lifestyles, etc., for whom the risks are higher than those
of the general human population? What is the range of risks for
different subpopulations?
-------
• What are the key components of, or contributors to, increased
risk for both general and specific subpopulations of humans and
other biota?
• Are there environmental or other factors that can mitigate the
extent, severity, or consequences of the risks attributed to the
pollutant?
• What are the sources and environmental pathways to which signifi-
cant widespread risks to humans and other biota can be attributed?
Quantitative answers to all of these questions would be desirable.
Practically, this may not be possible because of the lack of data on ex-
posure or effects of a pollutan:, the uncertainties in existing data and
the lack of agreement on methods to define and quantify risk. Thus only
in the very best of circumstances will there be data of sufficient quantity
and quality to specify the actual and potential risks associated with a
specific pollutant. More likely, ranges of estimated risks will have to
suffice. However, formal analysis of risk can indicate areas for addi-
tional data development, identify the areas of the greatest uncertainties,
and point the direction for possible measures to reduce risk, if needed.
9.2 GOALS AND OBJECTIVES
The overall goal of this portion of a risk assessment is to develop
a qualitative and/or quantitative understanding of the nature, extent,
and severity of the risks imposed by a pollutant on humans, fish, wildlife,
and other biota. A subsidiary goal is to establish the sources, pathways,
or causal factors associated with these risks so that control actions
for risk reduction can be identified and evaluated, when such are required.
For a given pollutant (or family of pollutants) specific objectives
for this work include:
(1) Estimating the average health risks to the general human popula-
tion, based upon average exposure ana the range of health effects
associated with the pollutant.
(2) Identifying those human subpopulations—on the basis of age,
sex, geographic location, occupation, lifestyle, or other
descriptors—that sustain greater than average risks, and
estimating the extent and severity of the health risks asso-
ciated with the pollutant.
(3) Estimating the average risks to general populations of fish,
shellfish, other aquatic species and wildlife based upon
average exposure and the range of effects associated with
the oollutant.
9-2
-------
(4) Identifying the subpopulations of fish, shellfish, other aquatic
species and wildlife—by geographic location, species, habits
and other descriptors—that sustain higher than average risk,
and estimating the extent and severity of the risks to these
sub-populations associated with the pollutant.
(5) Identifying the sources, pathways, and causal factors asso-
ciated with risks for human and other species in order to
allow investigation of possible methods for risk control or
reduction.
(6) Presenting the information on risks in a manner that is informa-
tive and understandable to technical and non-technical audiences.
9.3 APPROACHES AND METHODS
9.3.1 General Considerations
9.3.1.1 Definitions of Risk
Risk may be defined as the potential for negative consequences of
an event or activity. In the context of assessment of risk from environ-
mental pollutants, the event or activity is the release of a pollutant
into and its subsequent traverse through the environment such that humans
and other biota are exposed, and the negative consequences are any ad-
verse effects on the exposed populations. Thus, if a pollutant is be-
lieved to be harmful and if it is present in the environment, there is
certainly a potential for exposure and subsequent harm; that is, some
risk exists. The purpose of the risk considerations portion of risk
assessments is to go beyond such a qualitative statement of potential
risk, by estimating or measuring this potential.
Although the nature of adverse effects may be well understood, the
key difficulty in risk estimation lies in determining the probability
that adverse effects will occur. The probability is comprised of two
factors:
• The likelihood that groups of organisms will be exposed to various
levels of the pollutants.
• The likelihood that exposed organisms will experience adverse
effects.
These two factors correspond to the two major branches of investigation
described in previous sections—exposure and effects.
Analyzing the probability of adverse effects of different pollutants
will present different types of problems, depending upon pollutant proper-
ties and effects. For a highly persistent substance that is present in the
9-3
-------
human diet and known to have long-term effects, the main challenge lies
in estimating the likelihood of adverse effects based upon observed
exposure levels. On the other hand, for a substance that is degraded
rapidly and appears only in scattered locations, but is known to be an
acute toxicant, the focus should be on estimating the likelihood of
exposure. Therefore, the risk estimation methodology must be flexible
enough to encompass these and a multitude of other situations.
For a population of susceptible organisms, risk may be expressed
in several ways. One can state the probabilities that certain fractions
of the population will be adversely affected (e.g., 5% chance that 9/10
will be affected, 20% chance that 1/3 will be affected). This sort of
quantitative estimate is usually difficult to achieve. Alternatively,
one can state the expected number that may be affected, allowing a cer-
tain margin for error to reflect uncertainties in the underlying data
(e.g., 200,000 + 50,000). Finally one can give an order-of-magnitude
estimate that has no real measure of confidence attached to it (e.g.,
at most 5% will be affected). Each of these ways of expressing the
degree of risk can be more detailed in terms of types of effects, e.g.,
the chance of a specific disease, premature death, extent of disability,
etc.
Hence, risk estimates may be classified into three types, correspond-
ing to decreasing level of precision with which the population at risk
and the degree of risk can be characterized.
• probability distribution,
» numerical interval, and
• o rder of magnitude.
The level of precision of a risk estimate cannot exceed the precision
of the exposure and effects data from which it is obtained. In cases
where probabilistic risk estimates cannot be obtained, it may be possible
to develop a range or numerical interval of risks. In other cases, lack
of data may preclude any process ether than the mosst general or compara-
tive estimate of risk.
9.3.1.2 Overview of Evaluation Approaches
An evaluation of the risk.3 associated with an environmental pollutant
will usually consist of more than one result; it will describe the spec-
trum of risks identified in a variety of different cases characterized
by features such as:
• nature of the adverse effect,
• subpopulations affected, and
• temporal aspects (e.g., frequency).
9-.1
-------
Often different receptor populations will be exposed in different
ways over differing periods of time, and will experience different
effects as a result. The spectrum of such risks must, therefore, be
described to the extent permitted by the available data on exposure and
effects, developed according to the methods of the preceding sections.
For some pollutants, these data may not be sufficient for quantitative
estimates, and consequently the risk assessment may be only qualitative.
However, even with incomplete data, it is often possible to make meaning-
ful statements about risk.
An overview of an approach for guiding the risk estimation process
is shown in Figure 9-1. As shown, effects and exposure are first con-
sidered in parallel. Then, depending upon the level of precision with
which effects and exposure can be quantified, the results are combined
into one of four possible outputs.
For considering health effects, the first task is to review effects
data for the pollutant in order to ascertain whether toxic levels can be
quantified for specific toxic effects. The methods for dealing with
chronic or acute effects are substantially different; they have been
discussed in Section 7.0, and will be explored further below. The level
of precision of the toxicity estimates will determine the attainable
level of precision for the resulting risk estimates and will likely be
different for each category of toxic effect.
Exposure data are also reviewed in order to ascertain whether exposure
can be quantified, and to select a suitable level of precision for combin-
ing exposure with effects data.
There are four distinct possible outcomes of this procedure:
• Neither effects nor exposure are quantifiable. A qualitative
indication of risk may be given if the nature of the effects,
the predominant exposure routes, and populations at risk can
be identified (output 4 on Figure 9-1).
• Only exposure is quantifiable. By making conservative assump-
tions about effects levels, a hypothetical discussion of potential
risks is possible. Thus, if a risk indeed exists, one can at
least identify the subpopulations that would be most severely
affected (output 3 on Figure 9-1).
• Only effects are quantifiable. In this case, by postulating
realistic exposure levels, one can discuss the risk that would be
present under various exposure scenarios (output 1 on Figure 9-1).
• Both effects and exposure are quantifiable. This is the only
output for which a detailed and quantitative assessment of risk
would be possible. By combining estimates of exposure and
toxicity with information about the size and distribution of the
9-5
-------
cute
Is
exposure
quantifiable
la
toxlclty
uuautlf table?
Compute dose-
response re-
lationships
. .^
Ill'tl'l IN ill.:
levul ..f
ptOCl !• JOI1
k
//
Discuss riak lor
liyjx:tlietlc.il
1
/
/
\
Coir.blae
, 1— _ effects an it
exposure
est Imntes
1
Assess rJ.-ilc for
sub pcpuliit Ions
exposed by
varjoiiti routes
•>
\
\
-— —
Dlscusa risk under
conservative
assuirpc Ions
about effects
(Hvo qunlltat tve
indication of
possible risk
FIGURE 9-1. FLOW CHART FOR DEVELOPING RISK CONSIDERATIONS
-------
populations at risk, one can express numerically the risks to
different receptor categories, including the specification of
exposure routes, geographic extent, and frequency that character-
ize these risks (output 2 on Figure 9-1).
Although the discussion above suggests a straightforward approach
to estimation of risk according to four possible schemes, practical con-
siderations complicate the actual risk estimation. First, some health
effects may not be quantifiable in sufficient detail for numerical analysis,
whereas others may be. Discrepancies may exist among data on effects or
there may be a widespread distribution of effects among different sub-
populations. Second, exposure may not be quantifiable in sufficient
detail for numerical analysis. It may be possible to quantify exposure
for certain subpopulations and not for others; or the size of various
subpopulations may not be known. Thus, for any pollutant, one can expect
that there are many exposure/effects combinations (potential risks) that
can be considered only qualitatively, though some quantitative expression
of risk may be possible for some exposure/effects combinations.
The risk estimates obtained through these procedures should be quali-
fied by two important types of information: the assumptions incorporated
into each estimate, and the degree of confidence attached to numerical
estimates. Furthermore, any risk analysis should indicate what additional
data are required either to improve accuracy or precision or to confirm
certain assumptions.
9.3.1.3 Approaches Described in the Literature
Within the past few years, there has been considerable interest by
regulators, regulated industries, and the public in methods for esti-
mating risks, particularly the risks to human health. The setting of
tolerances for pesticides in food or feed, instituted in 1947, re-
quired consideration of exposure and effects to develop safe levels of
pesticide residues. The Delaney Amendment to the Pure Food and Drug Act
required a different approach, by setting a zero tolerance for food ad-
ditives that contained substances carcinogenic to experimental animals.
In 1976, the federal government suggested procedures and guidelines for
health risk assessment of suspected carcinogens (U.S. EPA 1976).
Albert _et_ al. (1977) suggested a rationale for assessment of carcinogenic
risk developed by the Environmental Protection Agency. In 1977, a com-
mittee of the National Academy of Sciences dealing with safe drinking
water discussed evaluation of risk of carcinogenicity and recommended
the linear extrapolation approach to low doses (NAS 1977).
The Environmental Protection Agency published its approach to the de-
velopment of water quality criteria, which considers quantitative and qual-
itative examination of both human health and environmental effects data
9-7
-------
(U.S. EPA 1979). The Interagency Regulatory Liaison Group further dis-
cussed methods of analysis and extrapolation of health data from labora-
tory animals to humans (U.S. EPA 1979b). Similarly, federal agencies
have suggested methods for evaluating the risks for air pollutants and
hazardous waste materials (U.S. EPA 1978; U.S. EPA 1979c).
A number of health specialists have criticized the inflexible quan-
titative approach developed by regulatory agencies, and/or suggested
other approaches to the evaluation of risk from pollutants (Kensler,
1979; Gori, 1980; Peto, 1980; Whittemore, 1980). Clearly there is a
highly volatile controversy over the most desirable and appropriate
approach to evaluate the health risks to man and other biota. In this
methodology, several alternative approaches are recommended for consider-
ation. Depending upon the specific nature of the pollutant and the data
available on exposure and effects, one or more suitable methods may be
chosen for use in each risk analysis. These methods should always be
selected with a clear understanding of the associated uncertainties and
assumptions. The remainder of this section discusses in greater detail
the possible qualitative and quantitative approaches to risk estimation.
The reader is referred to the citations given above and to the appendix of
this report for additional details of quantitative risk assessment pro-
cedures.
9.3.2 Evaluation of Risk to Human Health
9.3.2.1 Overview
Earlier in this section, the goals were presented for identifying
and evaluating the human health risks in a qualitative and quantitative
manner for both the general and special population groups. The pro-
cedures used to evaluate risks are the same for both the general popu-
lation and subpopulations; however, the exposures may be different and
the resulting risk estimates may differ.
The first step in considering exposure should be to summarize the
exposure of the general population and the exposure of specific subpop-
ulations. The exposure can be summarized in terms of an average daily
intake or dose for each of several different exposure routes, or the
total cumulative exposure from all routes in the form of a daily intake
or dose, for the average individual. Alternatively, the additional ex-
posure to specific defined subpopulations can be presented separately as
average daily intakes for each of the various exposure routes. These data
will have been developed from the methods and approaches described in
Section 7.0. in addition, the numbers of persons in each of the various
subpopulations often can be estimated. The summary will normally include the
mean or range of daily intake for the typical person, regardless of geo-
graphic location, and may include a range based upon age or sex. Maximum
values or ranges should also be given for selected subpopulations whose
characteristics result in greater than average intakes.
9-8
-------
In a parallel process, human health effects of the pollutant are
summarized as indicated by the available data from humans, experimental
animals, and other test systems concerning the range of possible adverse
effects. Finally, data are considered from supporting studies that may
confirm health effects such as the results of mechanism of action or
pharmacokinetic studies. To the extent that the available data are
sufficient, no-observed-effect levels (NOEL) or lowest-observed-effect
levels (LOEL) for different types of health effects are presented; gaps
in the data are so identified. In these summaries, it is important to
identify the exposure routes, the dose levels for the different responses,
and the species for which the observations were made.
9.3.2.2 Qualitative Risk Analysis
The most general type of risk analysis that can be accomplished is
a simple comparison of the various exposure levels with the NOEL or LOEL
levels. From such a comparison, a qualitative indication can be obtained
of the nature and types of risk that persons may incur. For example, if
the lowest acute toxicity level for a particular functional disorder is
X mg per kg body weight and the average exposure is much less than X mg
per kg body weight, the risk of large-scale acute effects is low. If on
the other hand, a certain subpopulation is, or can be, exposed to levels
approaching or greater than X, the members of this subpopulation may be
at significant risk of the acute effect. This type of qualitative compari-
son places the overall risks in perspective, and indicates the areas
(effects and exposures) requiring additional studies or evaluations.
Since one is not attempting to obtain quantitative estimates of risk with
this approach, "no effects" levels in laboratory animals can be compared
with human exposure levels in order to identify the potential (not probable)
risks to humans.
The degree of certainty with which these comparisons can be made
depends upon the precision with which effects and exposures can be
characterized. For example, the effects can be better defined if they
are based on human data, or experimental animal data substantiated with
appropriate pharmacokinetics and mechanism of action data. In vitro data
and data from cellular studies may also be useful in these simple compari-
sons in determining qualitatively whether human health effects can be
anticipated.
The qualitative approach to risk assessment, relying on general
comparisons of effects and exposure levels, is used when either both
exposure and effects cannot be quantified (output 4 on Figure 9-1) or
exposure can be quantified but not effects (output 3). If exposure
cannot be quantified, some hypothetical exposure values can be developed
based upon plausible scenarios, and these exposures can be compared with
"no effect" or "lowest effect levels," as described above.
9-9
-------
9.3.2.3 Semi-Quantitative Risk Analysis
When exposure to the general and specific subpopulations is known,
and effects data are available with some precision (at least for animal
systems), the analysis given above can be extended by consideration of
margins of safety. In the development of pesticide tolerances and water
quality criteria, the U.S. EPA considers the use of margins of safety to
develop tolerances and criteria for chemicals that are not carcinogenic.
The same approach can be used in risk analysis to obtain a relative rank-
ing of the risks of various subpopulations to specific effects. The
procedure is simply to match as best possible the NOEL or LOEL with the
exposure levels (by specific route) and to develop a margin of safety by
dividing the exposure level into the effects level. Ranges in exposures
and ranges in effects levels can also be used in order to determine ranges
in margins of safety.
The advantage of this semi-quantitative approach is that some
order of prioritization can be made for risks to different subpopulations
and risks of different adverse effects. For example, a margin of safety
for adults for a specific effect may be 1000, but for children only 10,
if the pollutant exposure is primarily a result of contamination of milk
and the effects levels are based on total body burden or weight. Alter-
natively, the margins of safety for those who work in a particular industry
may be many times less than the margins of safety for persons living
near or far from the industry. Analysis of margins of safety for differ-
ent exposure routes by which the same population group can be exposed
may help to suggest the type of environmental controls that could reduce
the exposure. In evaluating the significance of margins of safety, one
must bear in mind uncertainties in the underlying data and assumptions,
the accuracy and precision of the exposure levels used, and the relevance
of the available effects data to the possible human exposure and effects.
Another term frequently used to establish the magnitude of risk
associated with ingestion of an agent is the ADI (acceptance daily intake).
The ADI is an empirically derived value that reflects a particular com-
bination of knowledge and uncertainty concerning the relative safety of
a chemical. The uncertainty factors (U.F., also called safety factors) used
to calculate ADI values (NOEL/U.F. = ADI) represent the level of confidence
that can be justified on the basis of the available toxicological data.
Generally established guidelines for uncertainty factors are 10, 100 or
1000. When the quality and quantity of data are high, the uncertainty
factor is low and when the data are inadequate or equivocal, the un-
certainty factor must be larger.
In development of regulations, safety factors from less than 10
to over 10,000 are used in an attempt to reduce risks to negligible or
acceptable values or to balance risks with costs. In the risk evaluation
process described above, the regulating agencies ara assigned the task of
defining what constitutes acceptable levels of risk since the analysis
attempts only to rank these risks semi-quantitatively.
9-10
-------
Table 9-1 illustrates the expression of risk considerations by
use of margins of safety (Scow et. al. 1980), as an example of a semi-
quantitative approach to risk assessment. The worst case scenario in
the table would result in an exposure to 3.2 mg/kg per day, providing
a margin of safety of 1.7 over the lowest reported no-effect level
(6 mg/kg), which is for fetotoxicity. Most of the exposure in this
scenario can be attributed to a spill of PGP. When the spill is excluded,
the exposure would be 0.4 mg/kg per day with a margin of safety of 15.
Other non-occupational exposure routes appear to have a margin of safety
of at least 200.
A variant of the semi-quantitative approach may be taken for certain
chemicals for which there is a significant amount of reliable epidemio-
logical data or monitoring data and the effects in humans are well
understood. For some pollutants, there may be a direct relationship
between the level of the pollutant in body tissues such as adipose tissue,
and acute health effects or chronic functional impairment. Also, epi-
demiology studies may provide information to relate exposure or daily
intake of a pollutant to observed levels in body tissues. This type of
information, when combined with average intakes for the general population
or specific subpopulations, can show the potential risk levels in sub-
population groups. Thus the approach combines exposure information
with epidemiology, bioaccumulation, and health effects studies, or moni-
toring data, to predict risk levels for exposed populations.
An example of applying this approach is taken from an assessment
of risks associated with lead in the environment (Perwak et al., 1982a).
Tables 9-2 and 9-3 present selected results of a considerable~amount of
research that has been done on the epidemiology of lead exposure and
effects in humans. The exposure levels in human blood for various sub-
populations and pathways can be compared directly with lowest reported
effects levels and no effects levels for humans." As indicated by this
comparison, humans appear to be at significant risk of incurring'adverse
effects of lead exposure, especially children exposed through ingestion
of paint or inhalation and ingestion of contaminated dirt and dust and
urban populations or those near industrial areas of highways with heavy
vehicular traffic.
9.3.2.4 Quantitative Risk Analysis
The most quantitative and precise estimation of risk can be obtained
by use of human health data from epidemiologic studies. This is often net
possible because of the lack of quantified exposure data, the uncertainty
in or lack of human effects data, or the confounding influence of a
number of other variables affecting exposure and/or health effects.
Because of the lack of human health effects data for many chemicals,
the availability of data for laboratory animals, and the continued desire
to set specific levels of ambient concentrations of pollutants for the
protection of human health, a great deal of emphasis has been placed in
9-11
-------
TABLE 9-]. EXAMPLE OF RISK CONSIDERATIONS BY USE OF MARGINS OF SAFETY—PENTACHLOROPHENOL
I
M
l-o
Exposure SituatIon/Pathway
Max 1 muin Ex posmrti
Food
Drinking Water
Inhalation - Ambient
Dermal - Home Use - Spill
TOTAL
Exposure of Typical Person
Food
Drinking Water
1 "ilia 1 u t j oil Amii 1 en t
Dermal
TOTAL
Exposure of Person Living
Near Cooling Tower
Food
Drinking Water
Inhalation
Dermal
TOTAL
Exposure
(mg/day)
24
0.024
0.003
170
194.0
1.5
0.00002
0.003
0.003
1.5
1.5
0.00002
2
0.003
3.5
(mg/kg/day)
0.4
0.0004
2.8
3.2
0.025
_
_
0.025
0.025
0.03
0.55
Estimated
Margin of
Safety
15
15000
20
2.14
1.7
240
240
240
200
109
Ratio of lowest reported no effect level (6 mg/kg for fetotoxicity) to exposure level.
It is not known how "typical" these exposures are; the levels in drinking water are known to he low
and numerous locations have been sampled. No monitoring data are available for air. Limited data
are available for food, and the detection of PCP was not widespread.
Source: Adapted from Scow, K. et al. An exposure and risk assessment for pentachlorophenol Final
Draft Report. Contract EPA 68-01-3857. Washington, DC: Monitoring and Data Support Division
Office of Water Planning and Standards, U.S. Environmental Protection Agency; 1980.
-------
TABLE 9-2. EXAMPLE OF ADVERSE EFFECTS SUMMARY-
ADVERSE EFFECTS OF LEAD ON MAN
Adverse Effect
Carcinogenesis
Mutagenesis
Impaired Spermatogenesis
Fetotoxicity
Encephalopathy
Noticeable Brain
Dysfunction
Peripheral Neuropathy
Nephropathy
Reversible
Anemia
Elevated free erythrocyte
protoporphyrin
Lowest Reported Effect
Level
(US Pb/100 ml)
50
30-40
80—children
100—adults
50-60—children
50-60
40—children
50—adults
50-60—adults
15-20—children
and women
Urinary 5-aminolevulinic
acid
5-aminolevulinate de'nydratase
25-30—men
40
No-detected-effeet-Level
(ug Pb/100 ml)'
> 40 occupational
40-120 occupational
23-41
60—children
> SO—adults
i
50—children
40
10
40—children
50—adults
20—children and
women
25—men.
< 40
< 10
Source: Perwak, J. £t a^. An exposure and risk assessment for lead.
Final Draft Report. Contracts EPA 68-01-3857 and 68-01-5949.
Washington, DC: Monitoring and Data Support Division, Office of
Water Regulations and Standards, U.S. Environmental Protection
Agency; 1982a.
9-13
-------
TABLE 9-3.
EXAMPLE OF EPIDEMIOLOGICAL EVIDENCE OF
HUMAN EXPOSURE—LEAD BLOOD LEVELS IN MAN
Location.
Adults
Rural/Urban
Urban
Rural
Within 3.7 meters of
Highway
Living Near a Smelter
Children
Urban (primarily)
Within 30 meters of
Highway
Near Smelter—Kellogg,
ID—1974 (immediate
vicinity)
1975
1979
El Paso, TX
Blood Level
(ug/ 100 ml)
9-24
Most ^
Reference-
Bell _ec_ al. (1979)
Tepper and Levin (1972)
18 — mean (adjusted
for age and smoking)
Less than 57, > 30
16 — mean (adjusted
for age: and smoking)
Less than 0.5% > 30
23 — mean Daines et al. (1972)
167, >40
40,000 children de-
tected annually
> 30
"" 20 yearly geo-
metric :nean
507, > 40
99% > 40
60% > 60
Somewha:: reduced'
Almost all < 60f
and raosc < 40
70% > 40
14% > 60
Landrig:an e_c_ al. (1975)
Billick et_ al. (1980)
Caprio et_ al. (1974)
Walter e_t_ al. (1980)
Anonymous (1979)
Landrigan e_t_ al. (1975)
Reduction as a result of reduced atmospheric emissions as we'l as
increased sanitary procedures for r:he workers who were apoarently
exposing their children to lead through their clothing.
*See source indicated below for references.
Source: Perwak, J. et_ al_. An exposure and risk assessment for lead.
Final Draft Report. Contracts EPA 68-01-3857 and 68-01-5949.
Washington, DC: Monitoring and Data Support Division, Office
of Water Regulations and Standards, U.S. Environmental Protec-
tion Agency; 1982a.
-------
recent years on extrapolating laboratory animal test data to estimate
health effects in humans and assigning environmental criteria or standards
based upon this quantitative approach. As mentioned earlier, there has
been a great deal of controversy over the type of laboratory animal data
that should be considered, methods of extrapolation, the validity of the
results, and the use of these extrapolation procedures for the develop-
ment of regulatory standards. A complete discussion of these issues is
beyond the scope of this risk analysis methodology document. However,
since this type of quantitative analysis can be conducted where data are
available, it deserves some discussion, if only to indicate how the
methods can be used, and to stress precautions in their use. (Mathematical
details of the application of the methods are discussed in Appendix A.)
The toxicity of a substance in a particular species can often be
expressed in terms of a dose-response curve, which quantifies the like-
lihood or degree of a specific harmful effect occurring at various dose
levels. In some cases, acute toxic doses for humans may have been iden-
tified. However, in order to obtain the dose-response relationships for
sub-acute or chronic effects in humans, controlled laboratory experiments
must be performed with a species of laboratory animal presumably having
similar sensitivity to the substance. When data in humans are lacking,
acute effects data for laboratory animals are generally easy to obtain.
However, evaluation of chronic exposure at low-dose levels, corresponding
to typical ambient concentrations of pollutants in the environment, re-
quires an enormous number of experimental animals to demonstrate a statis-
tically meaningful response frequency. Instead, a practice has evolved to
perform such experiments with a moderate number of animals at high dose
levels (maximum tolerated dose), and then to extrapolate the results of
lower doses. The extrapolation procedure raises a number of questions.
One point of controversy is the existence of a threshold for carcino-
genic and mutagenic response to a pollutant. Some argue that an organism
is able to cope with low doses of a substance through metabolic processes
or repair mechanisms, so that harmful effects do not appear until a
certain minimum threshold , or "safe dose," has been surpassed. There
is evidence to suggest that for many types of chemicals different meta-
bolic processes occur at high dose levels than at low dose levels, and
this raises questions about the validity of linear extrapolation models.
Others contend that a toxic substance must be considered potentially
harmful at any dose and that a "zero tolerance" level should be assumed.
This issue has often been circumvented by the approach of selecting an
"acceptable" risk level and determining the corresponding acceptable
dose. From a practical point of view, the behavior of the dose-response
curve at low doses may be an academic question, since there is unavoid-
able background response due to a multitude of naturally occurring toxic
agents, as well as the genetic heterogeneity of human populations! Hence,
for a specific substance the real issue is whether the human response to
the substance significantly emerges from this general background "noise."
9-15
-------
Another important issue is the applicability to humans of experi-
mental data on animals. The derivation of a human dose-response curve
from animal data is predicated on the assumption that a substance with
demonstrated toxicity in certain laboratory animals has a probable
analogous effect on man. However, the toxic effects of many substances
appear to be species dependent, as a result of different metabolic patterns
Toxic effects of a chemical may differ even among strains of the same
species,_or; for different sexes and ages. Ideally, the toxicity for man
should be verified through epidemiologic studies in situations where
the substance was known to be present. Even if the substance is indeed
toxic to man, the issue remains of how to estimate the relative potency
of the substance in man as compared with animals. The common practice
is to use body weight or some power thereof to normalize the dose levels
between different species. However, there remain the questions of the
similarity of metabolism, bioaccumulation, and excretion of the pollutant
and its pharmacokinetics within the laboratory animal and man. One must
also reconcile the life span of the animal and its stages of development
relative to those of man.
Finally, the issue arises of what shape to ascribe to the dose-
response curve, when extrapolating from high to low doses. The simplest
assumption that can be made is that the dose-response relationship is
linear throughout the entire dose range. This follows from the so-called
"one-hit" hypothesis, which holds that each molecule of the substance
contributes equally to the likelihood of toxic effect, and hence that
there is no threshold. A rival hypothesis is offered by the Mantel-Bryan
method, which uses an S-shaped dose-response curve that generally yields
a much lower risk when extrapolated to low aoses. Other methods that
consider multi-hit or multi-stage response, time to response, and repair
mechanisms have also been discussed in the literature. (These methods are
described in the Appendix.) In practice, the linear "one-hit" model is
the easiest to apply, although it tends to give conservative results,
which may overestimate toxicity. At present none of these models has
been verified for specific health effects, and the use of any of them is
still controversial. For extrapolation of cancer risks, the multi-stage
models appear to agree best with known biological phenomena and are
presently recommended bl the EPA.
If these models are used in the attempt to quantify the relationship
between animal and human effects and effects levels, explicit mention
should be made of the assumptions in the process. These might include:
• Comparative susceptibility of humans and experimental animals.
• Interpretation of observed effects in animals.
• Method of dose administration.
• Computational procedure for dose conversion.
• Model selected for extrapolation co low doses.
9-IS
-------
Given the present state of the art, the uncertainty associated with such
assumptions cannot be quantified, except perhaps by subjective evaluation.
However, it is possible to derive, in some cases, statistical confidence
bounds on the dose-response estimates, based upon the size of the experi-
ment and the number of responses. Thus, at least a part of the overall
uncertainty may be expressed numerically.
When it is possible to attempt a quantitative analysis of health
effects by extrapolating from laboratory animals to humans, the approach
may be summarized as follows. From a careful review of the laboratory
animal studies, the one or ones are selected that most closely represent
the human health effects considered in terms of animal species, dose
levels, exposure routes, biological and metabolic processes, confidence
of data, and other variables and assumptions made earlier. Wherever
possible, a variety of models—one-hit, log-probit, multi-hit, etc.—
should be used to extrapolate from the high doses of the animal experi-
ments to the low doses of the anticipated exposure levels for the general
and specific human population groups. Using these methods, one can then
estimate the range of risks to humans associated with exposure to environ-
mental concentrations, using confidence levels if possible.
A quantitative analysis of the carcinogenic risks of 1,2-dichloro-
ethane exemplifies this approach (Perwak _et_ al. 1982b). No data were
found directly relating doses of 1,2-dichloroethane to responses in
humans. Because of apparent species specificity of responses and incon-
clusiveness of the results, studies of mutagenicity and'many other toxic
effects in laboratory animals could not be extrapolated to humans. The
data selected for extrapolation were the NCI data that demonstrated in-
creased alveolar/bronchial adenomas in male mice and increased mammary
adenocarcinomas in female rats (NCI 1978). These data are listed in "
Table 9-4. Other types of carcinomas were observed in both species such
as hemangiosarcoma, but the implied dose-response relationships were not
as severe.
The experimental results in Table'9-4 for both mice and rats show
three animal groups: the vehicle controls (zero dose), the low-dose
group, and the high-dose group. In both species the low-dose results
were not statistically significant, so that the high-dose results alon-
were used for extrapolation to humans. The first step in this extrapola-
tion was to calculate the equivalent human dose rate corresponding to the
experimental treatment. The approach recommended by the EPA was followed
which accounts for the duration of exposure relative to the animal life- '
span and normalizes the dose rate according to body surface area (U.S.
EPA^1979d). This approach is conservative, in that it results in a'lower
equivalent human dose than would be obtained from simple multiplication
of animal dose rate (mg/kg/day) by human body weight.
_ Whether surface area or body weight is a more appropriate normaliza-
tion factor is still open to debate. The former method yields a dose
rate about 6 times lower for rats, and about 14 times lower for m^'ce Thus
'
-------
TABLE 9-4. EXAMPLE OK CARCINOGENICITY DATA USED FOR RISK EXTRAPOLATION OF 1,2-DICHLOROETHANE
I
M
00
Species
Tested
Male mice
Average
Body
Weight
(kg)
0.025
Time-Weighted
Average Dose
(mg/kg/day)
195
Observed
Response
(%)
15/48 (31%)
Observed
- Effects
alveolar/
bronchial
adenomas
Duration of
Exposure
(week)
78
Animal
Life span
(week)
90
Female rats
0.32
97
0
1/47 (2%)
(vehicle controls)0/20
95
47
18/50 (36%)
1/50 (2%)
mammary
adeno«-
cinomas
78
110
(vehicle controls)0/20
Source:
Perwak, J. et a_l. An exposure and risk assessment for dichloroethanes. Final Draft Report.
Contracts EPA 68-01-5949 and 68-01-6017. Washington, DC: Monitoring and Data Support Div.,
Office, of Water Regulations and Standards, U.S. Environmental Protection Agency; 1982.
-------
The actual calculation of equivalent human dose was performed as
follows, assuming an average human weight of 70 kg:
!_ duration of
Human dose = 70 kg X animal dose X (anim*l weight 3 x (5.} (exposure
human weight 7 animal '
lifespan
The correction factor for body surface area is the cube root of the
ratio of animal to human weight, as shown by the U.S. EPA (1979d) . A
correction factor of 5/7 was also included since the animals were treated
only on five days per week. As a result, it was concluded that:
« the dose of 195 mg/kg/day, which produced a 31% effect in male
mice, was equivalent to a human dose of approximately 600 mg/day;
and
• the dose of 95 mg/kg/day, which produced a 36% effect in female
rats, was equivalent to a human dose of approximately 560 mg/day.
These results are roughly the same, with slightly greater potency implied
by the rat experiment. Therefore, only the rat data were used in subse-
quent risk estimation.
Three separate extrapolation models were applied—the linear,log-probit
and multi-stage models (see the appendix) using the data for female rats
(i.e., 36% response at a human equivalent of 560 mg/day). The "one-hit"
extrapolation is performed by simply assuming a constant increase in
probability of tumor induction for each increment of dose. This leads
to a gradually rising dose-response curve, which is nearly linear at
sufficiently low doses. The log-probit model assumes that carcinogenic
doses are log-normally distributed, resulting in an S-shaped dose-response
curve with a threshold-like effect. These two models, generally speak-
ing, tend to bound the range of risk estimates that could be obtained
from other dose-response models. The one-hit model is conservative, in
that it probably over-estimates the true response at low doses, whereas
the log-probit model usually results in much lower risk estimates for
typical human exposure levels. The multi-stage model was applied to the
combined rat and mouse data. The multi-stage model generally *ives dose-
response estimates intermediate to the on-hit and log-probit'models.
It must be noted that interpretation of the results from these three
extrapolation models for assessment of human risk due to exposure to
1,2-dichloroethane is subject to a number of important qualifications and
assumptions:
9-19
-------
• Although positive carcinogenic findings exist, there have been
contradictory negative findings in tests with the same species
using different routes of exposure. No adequate explanation has
been found for these disparate results.
• Assuming that the positive findings indeed provide a basis for
extrapolation to humans, the estimation of equivalent human
doses involves considerable uncertainty.
• Occurrences of human exposure to 1,2-dichloroethane are assumed
to be numerous.
» The effect on rodents of chronic exposure at: low doses, such as
those possibly encountered in human exposure, may be deduced by
extrapolating from higher gavage doses used in the NCI experiments.
o Due to inadequate understanding of the mechanisms of carcinogenesis ,
there is no scientific basis for selecting among several alternate
dose-response models, which yield differing results.
In Table 9-5 the estimated risks of exposure to 1,2-dichloroethane
obtained from these models are summarized. The expected number of cancers
per million exposed population is shown for daily exposures to 1,2,-
dichloroethane ranging from 1 ug to 1 mg. The gap between the estimates
is large in the low-dose region; only at doses above 10 ug/day does the
log-probit dose/response curve begin to rise more steeply. The dose
corresponding to a per capita risk of 10~- is about 100 yg/day according
to the log-probit model, which is about eight times greater than the
level obtained from the linear model. The multi-stage model predicts
a risk intermediate between these two levels in the range of 1 ug/day to
100 ug/day.
In Table 9-6 the results from the three extrapolation models are
applied to estimated average lifetime exposure of the general population
and several subpopulations to 1,2-dichloroethane via ingestion and in-
halation. As shown, the subpopulation drinking highly contaminated ground-
water appears to be the group at highest possible risk due to waterborne
exposure. Because of limited monitoring of levels in groundwater, the
size of this subpopulation cannot be estimated reliably. Other uncertain-
ties result from the availability of only limited data on residues in
spices and other foods and on atmospheric concentrations in urban areas.
Thus there is a substantial range of uncertainty concerning the
actual exposure levels and carcinogenic effects of 1,2-dichloroethane.
However, present scientific methods and limited data availability do
not permit a more definitive assessment of risk to humans resulting
from environmental exposure to this compound.
9.3.3 Evaluation of Risk for Aquatic Species
Although much of the focus of the risk considerations section of an
overall risk assessment is devoted to evaluating human health risks,
risks to fish, other aquatic species, and wildlife should also be con-
9-20
-------
TABLE 9-5. EXAMPLE OF ESTIMATION OF UNIT CARCINOGENIC RISK: ESTIMATED
NUMBER OF EXCESS LIFETIME CANCERS PER 1,000,000 POPULATION
EXPOSED TO DIFFERENT LEVELS OF 1,2-DICHLOROETHANE
Niamber of Excess Lifetime Cancers Per 10°
Extrapolation Method Population at Exposure Level*
1 ug/day 10 ug/day 100 yg/day 1000 ^g/day
One-hit extrapolation 0.8 8 80 800
Log-probit extrapolation negligible 0.1 13 690
Multi-stage model 0.5 5 '50 500
Carcinogen Assessment Group Q.5 5 50 son
Estimated excess lifetime cancers are given based on three different
dose-response extrapolation models. The lifetime excess incidence
per 1,000,000 population exposed represents the increase over the
normal background incidence, assuming that an individual is con-
tinuously exposed to 1,2-dichloroethane at the indicated daily intake
over their lifetime. There is considerable variation in the estimated
risk due to uncertainty introduced by the use of laboratory animal
data, by the conversion to equivalent human dosage, and by the appli-
cation of hypothetical dose-response curves. In view of several
conservative assumptions that were utilized, it is likely that these
predictions overestimate the actual risk to humans.
Source: Perwak, J., et al. An exposure and risk assessment for
dichloroethanes. Final Draft Report. Contracts EPA
68-01-5949 and 68-01-6017. Washington, DC: Monitoring and
Data Support Division, Office of Water Regulations and
Standards, U.S. Environmental Protection Agency; 3982.
9-21
-------
TABLE 9-6. EXAMPLE OF ESTIMATION OF CARCINOGENIC RISK DUE TO
ENVIRONMENTAL EXPOSURES: ESTIMATED RANGES OF CARCINO-
GENIC RISK TO HUMANS DUE TO 1,2-DICHLOROETHANE EXPOSURE
FOR VARIOUS ROUTES OF EXPOSURE
Route
Drinking water
Food
Inhalation
Estimated
Average Lifetime
Exposure- (ug/day)
NO. Excess
Estimated Lifetime Cancers
(per million exposed)*3
<2
-5
One hit
1.6
4
Probit
CAG
1
3
rural <0.4
urban <0.8
industrial 32-120
in the vicinity of
production facilities 0.8-80
Isolated subpopulations
groundwater (maximum) 800
inhalation in industrial 1300
area
0.3
0.6
30-100
0.6-60
600
1000
1-20
< 0.1-10
500
1000
0.2
0.4
20-60
0.4-40
400
700
uata taken from Table 7-3 of the source cited below.
Estimated excess lifetime cancers are given based on three different
dose-response extrapolation models. The lifetime excess incidence of
cancer represents the increase over the normal background incidence
assuming that an individual is continuously exposed to 1,2-dichloro-
ethane at the indicated daily intake over their lifetime. There is
considerable variation in the estimated risk due to uncertainty
introduced by the use of laboratory rodent data, by the conversion
to equivalent human dosage, and by the application of hypothetical
dose-response curves. In view of several conservative assumptions
that were utilized, it is likely that these predictions over- :a
estimate the actual risk to humans.
Source: Perwak, J., et al. An exposure and risk assessment for
dichloroethanes. Final Draft Report. Contracts EPA 68-01-
5949 and 68-01-6017. Washington, DC: Monitoring and Data
Support Division, Office of Water Regulations and Standards,
U.S. Environmental Protection Agency; 1982.
9-22
-------
sidered. In general, qualitative approaches seem to be more practicable
than quantitative approaches. This is a consequence of the large number
of species that might be considered, and the general lack of detailed
exposure data for these individual species. Case studies or exposure
scenarios seem to be a useful approach for characterizing the range of
risks.
The first step in the evaluation of risk to aquatic species is to
summarize information available on the exposure of different species,
the locations of that exposure, and the environmental conditions that
affect exposure. Some of this information may be general, in the sense
that actual exposure of fish and wildlife to observed concentrations of
pollutants may or may not occur; however, the situations may be described
as potential exposure conditions, where it is known that both the en-
vironmental concentrations exist in the water, and fish and/or other
aquatic species are known or suspected to inhabit the area in which these
concentrations are found. Environmental conditions that affect exposure
include factors such as rainfall, hardness of water, pH, seasonality of
pollutant concentrations, salinity, etc.
The second step is to summarize data on fish and wildlife effects
in terms of the most sensitive species, types of effects observed, LC 's,
and/or other indicators of toxic effects, and parameters that may in-50
fluence toxic effects (e.g., other pollutants, water hardness). As was
the case in the human effects analysis, it should be possible to summarize
no-effect level data for different species, or lowest reported effects
data, as well as more commonly available values such as LCcn's. In both
the exposure and effects summaries, ranges of values for exposure and
effects should be presented, if available, as well as the relative degree
of confidence in the data.
The next step requires a qualitative comparison of the exposure and
effects data. From this comparison, one can determine the species and
locations in which significant adverse effects might be expected to occur.
For example, if it were established that the LC- 's for a particular
species had values that were in the same range as environmental concen-
trations of pollutant in a particular location, and it were expected that
species might inhabit that location, then there would be a possibility
of significant risk to that species in that location. Thus the analysis
becomes one of establishing the "key intersections" between exposure
and effects data with regard to specific species and geographic locations.
In this regard, risk considerations for fish and wildlife often tend to
be more specific with respect to geographic locations than do risk
considerations for human health.
The end result of this process can be a listing or summary of
species, locations, exposures, and effects levels, which indicate the
combinations most likely to result in high risks. In some cases, it is
possible to attempt some quantitative comparisons, i.e., to determine"
the extent of a specific health effect on fish and wildlife by utilizing
values such as LC5Q's and concentrations in the ambient water. As ex-
plained in the section on aquatic effects, caution must be exercised in
developing these mathematical relationships because of the differences'
between results under laboratory conditions and fiald conditions. For
this reason, it is important to identify those environmental factors or
conditions that can influence the adverse effects.
9-23
-------
In order to develop the risk analyses further, case studies of in-
dividual exposure/effects situations can be investigated. For example,
one can pick several of the "key intersections" of exposure and effects
data and determine through field interviews, discussions with local
experts, examination of evidence of fishkills, or actual field sampling
programs, whether there is real evidence of exposure and/or damage to
the fish or wildlife population. Specific sources located in or near
the area can be investigated, appropriate mathematical models of pollu-
tant dispersion can be used to estimate concentrations of the pollutant
in the water and comparisons with ambient monitoring data can be made
in order to help define the potential risk for those sensitive species
that inhabit the area. Environmental factors should be considered that
could affect toxicity and are specific to these geographic areas. The
main purpose of these case studies is to confirm the existence of sig-
nificant risk to fish or wildlife species in areas in which exposure
can occur and effects are anticipated. The number of case studies con-
ducted depends upon the scope of the risk analysis and the numbers and
types of locations in which exposure is expected to occur, and the nature
and magnitude of the potential adverse effects.
An example of the information that can be obtained in the case study
approach is given below. Analysis of the aquatic risks associated with
copper in the environment, showed that LCso's for a number of sensitive
species were below 100 ug/1 in the laboratory (Perwak e_t al. 1980). River
basin summaries (STORET) revealed that the mean levels of copper reach or
exceed this level in numerous locations in the U.S. and this suggests
that the potential risk to fish and invertebrates is widespread. Further
examination of detailed data from individual monitoring stations in
several of these river basins indicated that the mean concentrations were
not representative of ambient conditions, but resulted from very elevated
concentrations in a few locations. Consideration of the form of copper
involved, factors favoring complexation and adsorption, and actual reports
of fish kills indicated that risk exists to organisms in specific loca-
tions, but that the risk is neither as severe nor as widespread as would
have been predicted from laboratory data.
The end result of risk considerations for fish and wildlife will
generally be a series of summary statements indicating the locations in
which adverse effects are likely to occur, the species that are likely
to be affected, the environmental conditions that influence whether or
not the potential effects actually occur, and the results of case
studies to confirm or establish the magnitude of the potential problems.
In addition, areas for further investigation should 'be identified.
9.3.4 Summary of Risk Considerations
The approaches described above can provide specific information on
the nature and extent of the risks to both general and specific human
subpopulations and to fish and other aquatic biota. Depending upon
the type and level of data available concerning exposure and effects,
9-24
-------
specific conclusions may be drawn giving the ranges of risks and the
degree of precision associated with these risks. In some cases, only
qualitative aspects of risks can be presented; in others, quantitative
information may be appropriate.
In either case, the overall risk posed by the pollutant should be
portrayed so that regulators and the public can visualize whether or not
significant problems are expected to occur. A number of methods of pre-
senting this overall summary of risks are possible: tables or charts
showing specific risks to humans and other species, charts or graphs
showing the relative risk associated with different effects, etc. An
approach that has the potential for effectively summarizing the results
of risk considerations is to prepare a graphic presentation of exposure
and effects in terms of the same variable and to indicate the areas in
which the combinations of exposure situations and effects levels can
present significant risk. This approach could be followed for both human
health effects and effects on fish and aquatic species and could be
specific to a particular type of effect, type of exposure, or other
characteristics.
One method of presentation is to plot the relative frequency of
exposure in terms of concentration or average daily intake, the frequency
being defined in general terms such as "usual," "frequently," "occasional,
"rare exposure." On the same graph, one could plot the likelihood of
various toxic effects at levels such as LC5Q for various species. The
intersection of the exposure and effects curves, or more precisely the
area bounded by the intersection, would indicate the areas of significant
potential risk.
As an example of this type of presentation, Figure 9-3 shows a dia-
grammatic plot of the relative frequencies (ordinate) of both aquatic
exposure levels and reported effects levels in terms of the surface
water concentrations of arsenic (abscissa) (Scow _e_t al. 1982). -Surface
water concentrations of 1 mg/1 are rarely observed. The shading indi-
cates the approximate degree of uncertainty associated with the data
points used. The area under the intersection of the curves represents
the region of potential risk where the observed water concentrations have
values that exceed reported adverse effects levels.
Though such a plot of exposure and effects frequencies can be a
useful conceptual tool, its interpretation must be made in light of the
representativeness of the monitoring data base, the validity of gener-
alizing from the available toxicological data (number of species tested,
chronic versus acute effects), and other factors such as bioavailability.
If sufficient data are available concerning effects and exposure
levels for humans, a similar plot could be used to summarize the like-
lihood of significant potential risk.
9-25
-------
Possibility
of Occurrence
>£>
I
Usual
Frequent
Occasional
Rare
Aquatic
Exposure
Levels
Adverse
Effects
Levels
0.0001 0.001
Concentration
Surfac
Water
(mg/l)
0.01 f 0.1 f 1.0 10 100
Lowest Freshwater
Reported Criterion for
Acute Effects Protection
Level of Aquatic Life
FIGURE 9-2 EXAMPLE OF RISK CONSIDERATIONS SUMMARY FOR AQUATIC BIOTA—ARSENIC EXPOSURE
AND TOXICITY TO AQUATIC ORGANISMS
Source: Scow, K. , e± a±. An exposure and risk assessment for arsenic. Final Draft Report,
Contract 68-01- 6160, 6017. Washington DC: Monitoring and Data Support Division,
Office of Water Regulations and Standards, U.S. Environmental Protection Agency,
1982.
-------
REFERENCES
Albert, R.E.; Train, R.E.; Anderson, E. Rationale developed by the
Environmental Protection Agency for the assessment of carcinogenic risk.
J. Nat. Cancer Inst. 58:1537-1544; 1977.
Cornfield, J. Carcinogenic risk assessment. Science 198:693-699; 1977.
Gori, G.B. The regulation of carcinogenic hazards. Science 208:256-
261; 1980.
Kensler, C.J. Experimental procedures in the evaluation of chemical
carcinogens. Coulston, F. ed. Regulatory aspects of carcinogens and
food additives: The Delaney Clause. New York, N.Y.: Academic Press;
1979: 239-260.
National Academy of Sciences (NAS). Drinking water and health.
Washington, DC: National Academy of Sciences; 1977.
National Cancer Institute (NCI). Bioassay of 1,2-dichloroethane for
possible carcinogenicity. Tech. Report NCI-CG-TR-44. Washington, DC:
National Cancer Institute; 1978.
Perwak, J.; Bysshe, S.; Delos, C.; Goyer, M.; Nelken, L.; Schimke, G.;
Scow, K.; Walker, P.; Wallace, D. An exposure and risk assessment for
copper. Final Draft Report. Contract EPA 68-01-3857. Washington, DC:
Monitoring and Data Support Division, Office of Water Planning and
Standards, U.S. Environmental Protection Agency; 1980.
Perwak, J.; Goyer, M.; Nelken, L.; Payne, E.; Scow, K.; Wallace, D.;
Wood, M. An exposure and risk assessment for lead. Final Draft Report.
Contracts EPA 68-01-3857, 68-01-5949. Washington, DC: Monitoring and
Data Support Division, Office of Water Planning and Standards, U.S.
Environmental Protection Agency; 1982a.
Perwak, J. ; Byrne, M.; Goyer, M. ; Lyman, W,. ; Nelken, L. ; Scow, K. ;
Wood, M.; Moss, K. An exposure and risk assessment for dichloroethanes.
Final Draft Report. Contracts EPA 68-01-5949, 68-01-6017. Washington,
DC: Monitoring and Data Support Division, Office of Water Regulations
and Standards, U.S. Environmental Protection Agency; 1982b.
Peto, R. Distorting the epidemiology of cancer: the need for a more
balanced overview. Nature 284:297-300; 1980.
Scow, K.; Gilbert, D.; Goyer, M.; Perwak, J.; Payne, E.; Thomas, R.;
Walker, P.; Wallace, D.; Wechsler, A.; Wood, M.'; Woodruff, C. An
exposure and risk assessment for pentachlorophenol. Final Draft Report.
Contract EPA 68-01-3857. Washington, DC: Monitoring and Data Support
Division, Office of Water Planning and Standards, U.S. Environmental
Protection Agency; 1980.
9-27
-------
U.S. Environmental Protection Agency (U S EPA} Tni-^-
water
connnents on report. Federal Register 44:39858-3987 197%
is.™ ••
9-28
-------
10. BIBLIOGRAPHY OF REFERENCE MATERIALS FOR
USE IN EXPOSURE AND RISK ASSESSMENTS
10.1 INTRODUCTION
The following bibliography is intended to provide an initial means
of identifying reference materials for use in conducting exposure and
risk assessments for environmental pollutants. The bibliography is
organized according to the following six topical areas of investigation:
• Materials Balance
• Environmental Pathways and Fate
• Monitoring Data and Environmental Distribution
• Human Exposure and Effects
• Exposure and Effects—Non-Human Biota
• Risk Estimation
10-1
-------
10.2 MATERIALS BALANCE
Abstracts and Searches
Applied Science and Technology Index
Bibliographies from U.S. Bureau of Mines
Chemical Abstracts
Engineering Index
Environmental Abstracts
Metals Abstracts
NTISearch
Pollution Abstracts
Industry and Consumer Associations
American Chemical Society
American Institute of Chemical Engineers
American Institute of Industrial Engineers
American Institute of Mining, Metallurgical and Petroleum
Engineers
American Iron and Steel Institute
American Paper Institute
American Petroleum Institute
American Public Works Association
American Society of Sanitary Engineers
Association of Home Appliance Manufacturers
Chemical Manufacturers Association
Environmental Defense Fund
Environmental Information Center
Gas Appliance Manufacturers Association
Glass Packing Institute
National Agricultural Chemicals Association
National Ash Association
National Association of Recycling Industries
National Family Option
National Association of Manufacturers
National Lime Association
National Solid Waste Management Association
Natural Resources Defense Council
Society of Manufacturing Engineers
Society of Mining Engineers
Society of Plastic Engineers
Synthetic Organic Chemicals Manufacturers Association
The Fertilizer Institute
Water and Wastewater Equipment Manufacturers Association
Zinc Institute
10-2
-------
Periodicals
Agricultural Chemicals
AIChE Journal
American Dyestuff Reporter
American Paint Journal
American Paper Industry
Automotibe Engineering
Beverage Industry
Chemical and Engineering News
Chemical and Metallurgical Engineering
Chemical and Petroleum Engineering
Chemical Engineering (Chemical Engineering Equipment Buyers Guide)
Chemical Engineering Progress
Chemical Marketing Reporter
Chemical Processing
Chemical Week
Chemtech
Coal Age
Coal Mining and Processing
Engineering and Mining Journal
Food Engineering
Food Industry
Industrial Wastes
Journal of the American Water Works Association
Journal of the Air Pollution Control Association
Journal of the Water Pollution Control Federation
Machine Design
Mining Engineer
Modern Packaging
Modern Plastics
Oil and Gas Journal
Packaging Design
Pit and Quarry
Plant Engineering
Plastics Engineering
Plastics Technology
Plastics World
Process Engineering
Pulp and Paper
Rubber Age
Solid Waste Report
Solid Wastes Management
TAPPI
Textile Worlds
Water and Sewage Works
Water and Wastes Digest
Water and Wastes Engineering
10-3
-------
Books
Census of Manufacturers, U.S. Dept. of Commerce, Washington, DC
U.S. Geological Survey Yearbook, U.S.G.S., Washington, DC
Mineral Facts and Problems, U.S. Bureau of Mines, Washington, DC
SME Mining Engineering Handbook, I.A. Given (ed.), AIME, NY
Chemical Process Industry, R. Shreve and J. Brink, McGraw-Hill, NY
Unit Operations of Chemical Engineering, W. McCabe and J. Smith,
McGraw-Hill, NY
Metal Bulletin Handbook, R. Packard (ed.), Metal Bulletin Ltd.,
London
Metal Statistics, P. Cere (ed.), Fairchild Publications, NY
Minerals Yearbook, U.S. Depc. of Interior, Bureau of Mines,
Washington, DC
Mining Engineer's Handbook, R. Peele (ed.), John Wiley and Sons,
NY
Oil and Gas International Yearbook, P. Jenkins (ed.), Business
Enterprises, London
Petroleum Processing Handbook, W.F. Bland and R.L. Davidson
(eds.), McGraw-Hill, NY
Chemical Regulation Reporter, The Bureau of National Affairs,
Washington, DC
Chemical Engineering Practice, H.W. Cremer and T. Davis (eds.)
Butterworths Scientific Publications, London
Moody's Industrial Manual, Moody's Investors Service, NY
Encylcopedia of Chemical Technology, R.E. Kirk and D.F. Ohmer,
Interscience Encyclopedia, Indc. NY
The Encyclopedia of Chemistry, C.A. Hampel and G.G. Hawley
(eds.), Van Nostrand Reinhold Co., NY
Directory of Chemical Producers, U.S.A., Chemical Information
Services, Stanford Research Institute, Menlo Park, CA
Chemical Sources, U.S.A., Directories Publishing Co.,
Fleming, NJ
Manufacturing Processes Dictionary, H.R. Clause:: (ed.)
Technomic Publishing Co., Westport, CN
Industrial Product Directory, Cahness Publications, Stamford
SOCMA Handbook, American Chemical Society, NY
Rare Metals Extraction by Chemical Engineering Techniques,
W.D. Jamrock, Pergamon, NY
Jane's World Mining, Jane's Yearbooks, London
Mines Register, New York
Handbook of Non-Ferrous Metallurgy, D.M. Liddele (ed.) McGraw-
Hill, New York
Clarke, R.K., J.T. Foley, W.F. Hartman, D.W. Larson. 1976.
Severities of Transportation Accidents, Sandia Laboratories
Report, SLA-74-001.
Annual Reviews, U.S. Air Carrier Accidents, National Transporta-
tion and Safety Board, Bureau of Aviation Safety
Preliminary Analyses of Aircraft Accident Data. (Annual) U.S.
Civil Aviation NTS3 Reports-
10-4
-------
Accidents of Large Motor Carriers of Property. (Annual)
Bureau of Motor Carrier Safety, Federal Highway Administra-
tion, U.S. Department of Transportation
Accident Bulletins and annual Summary and Analyses of Accidents
on Railroads in the United States. U.S. Department of
Transportation, Federal Railroad Administration, Bureau of
Railroad Safety
AAR-RPI Railroad Tank Car Safety Research and Test Project.
A series of Reports 1972-1976
Solomon, K.A., M. Rubin, and D. Okrent, 1976. On Risks from
the Storage of Hazardous Chemicals, University of California,
University of California, Los Angeles Report UCLA - ENG -
76125, December 1976
Major EPA Studies
Development Documents for Effluent Limitations Guidelines and
Standards (by point source category). Washington, DC:
Effluent Guidelines Division, Office of Water and Waste
Management, U.S. Environmental Protection Agency
Fate of Priority Pollutants in Publicly Owned Treatment Works.
EPA-440/1-80-301. Washington, DC: Effluent Guidelines
Division, Office of Water and Waste Management, U.S.
Environmental Protection Agency; 1980.
Water-Related Fate of 129 Priority Pollutants. Volumes I and
II. EPA 440/4-79-029a, b. Washington, DC: Office of
Water Planning and Standards, U.S. Environmental Protection
Agency; 1979.
10-5
-------
10.3 FATE AND PATHWAYS ANALYSIS
Periodicals
Advances in Agronomy
Advances in Applied Microbiology
Advances in Chemistry Series
American Chemical Society Symposium Series
American Journal of Botany
Analytical Chemistry
Applied and Environmental Microbiology
Archives of Environmental Contamination and Toxicology
Atmospheric Environment
Biochemistry
Bulletin of Environmental Contamination and Toxicology
Canadian Journal of Chemistry
Canadian Journal of Microbiology
Chenosphere
Chemical Engineering News
Endeavor
Environmental Health and Pollution Control
Environmental Health Perspectives
Environmental Pollution
Environmental Science and Technology
Estuarine and Coastal Marine Science
Journal Agricultural Food Chemistry
Journal of Air Pollution Control Association
Journal of Association Off. Anal. Chem.
Journal of Environmental Quality
Journal of the American Chemical Society
Journal of Water Pollution Control Federation
Marine Chemistry
Marine Pollution Bulletin
National Academy of Science (NAS) documents
Nature
Pesticide Monitoring Journal
Proceedings of the American Society of Horticultural Science
Proceedings of the Industrial Waste Conference (Purdue University
Engineering Bulletin)
Proceedings of the Royal Society of London
Science
Soil Biology and Biochemistry
Soil Science
Soil Science Society American Proceedings
The Science of the Total Environment
Water, Air and Soil Pollution
Water Pollution Abstracts
Water Research
Water Resource Bulletin
Water Waste Treatment
Weed Science
World Health Organization (WHO) documents
10-6
-------
Books, Articles, and Reports
Altshuller, A. P.; Bellar, T.A. Photochemical aspects of air pollution:
a review. Environ. Sci. Tech. 5.39; 1971.
Callahan, M.A. ; Slimak, M.W. ; Gabel, N.W.; et_ al. Water-related environ-
mental fate of 129 priority pollutants. EPA-44014-79-0292. Washington,
DC: U.S. Environmental Protection Agency; 1979.
Chion, C.T.; Peters, L.J.; Freed, V.H. A physical concept of soil water
equilibria for non-toxic compounds. Science 206:831-832; 1972.
Coniglio, W.A.; Miller, K. ; MacKeenen, D. Briefing on the occurrence of
volatile organics in drinking water. Washington, DC: U.S. Environmental
Protection Agency; 1980.
Cox, R.A.: Derwent, R.G. ; Eggleton, A.E.J.; Lovelick, J.E. Photochemical
oxidation of halocarbons in the troposphere. Atmos. Environ. W -305-308 •
1976.
Billing, W.L.; Teferhiller, N.B.; Kallos, G.J. Evaporation of methylene
chloride, chloroform, 1,1,1-trichloroethane, trichloroethylen, tetra-
chloroethylene, and other chlorinated compounds in dilute' aqueous solutions
Sci. Technol. 9(9) :833-839 ; 1975.
Fiksel, J.; Bonazountas , M. ; Ojha, H.; Scow, K. ; Freed, R. ; Adkins, L.
An integrated geographic approach to developing toxic substance control
strategies. Final Draft Report. Contracts EPA 68-01-6160 and 68-01-6271.
Washington, DC: Office of Policy and Resource Management, U.S. Environ-
mental Protection Agency; 1981.
Jungclaus, G. ; Lopez-Avila; Hitres , R.H. Organic compounds in an industrial
wastewater: a case study of environmental impact. Environ. Sci. Tech
12(l):88-96; 1978.
Karickhoff, S.W.; Brown, D.S.; Scott, T.K. Sorption of hydrophobia
pollutants on natural sediments. Water Res. 13:241-248; 1979.
Kirk-Othmer Encyclopedia of Chemical Technology. 3rd ed . , New York,
NY: Interscience Publishers; 1976.
Lillian D.; Smith, H.B.; Appleby, A.; Lobban, L. ; Arntz, R. ; Gumpert, R.-
Hague, J.; Toomey, J.; Kazazis, J.; Antel, M. ; Hansen, D.; Scott B
°f halogenated compounds. Environ. Sci. Technol. 9:1042-
Lyman, W.J.^Reehl, W.F.; Rosenblatt, D.H. (eds.). Handbook of chemical
^havior of organic compounds.
10-7
-------
Mabey, W. ; Mill, T. Critical review of hydrolysis of organic compounds
in water under environmental conditions. J. Phys. Chem. Ref Data
7:383-415; 1978.
Mackay, D. Finding fugacity possible. Environ. Sci. Technol. 13:1218-
x^»— j j j» 7 / 7 •
Mackay, D.; Yuen, T.K. Volatilization of organic contaminants from
rivers. Proc. 14th Canadian Symp., 1979. Water Pollut. Res. Can.; no
date.
Mackay, D.; Leinonen, P.J. Rate of evaporation of low solubility contaminants
from water bodies to atmosphere. Env. Sci. Technol. 9:: 1128-1180; 1975.
McConnell, G.; Ferguson, D.M. ; Pearson, L.R. Chlorinated hydrocarbons
and the environment. Endeavor 34(12) : 13-18; 1975.
Miller, C. Exposure assessment modelling, A. state of the art review
Contract EPA PB 600/3-78-065. Athens, GA: Athens Research Laboratory,'
U.S. Environmental Protection Agency; 1978.
Morrison, R.J.; Boyd, R.N. Organic chemistry. Boston, MA: Allan and
Bacon; 1973.
Neeley, W.B. A preliminary assessment of the environmental exposure to
be expected from the addition of a chemical to a simulated aquatic eco-
system. Intern. J. Environ. Studies 13:101-108; 1979.
Pellizzari, E.D.; Erickson, M.C.; Aweidinger, R.A. Formulation of a
preliminary assessment of halogenated organic compounds in man and environ-
mental media. Washington, DC: U.S. Environmental Protection Agency: 1979.
Smith, J.H.; Bomberger, D.C. Prediction of volatilization rates of high
volatility chemicals from natural water bodies. Env. Sci. Technol
14(11) .-1332-1337; 1980.
Southworth, G.R. The role of volatilization in removing polycyclic aromatic
hydrocarbons from aquatic environments. Bull. Environ. Contain'. Toxic ol
21:507-514; 1979.
Stanford Research Institute (SRI) . Estimates of physical-chemical
properties of organic priority pollutants. Preliminary draft. Washington.
DC: Monitoring and Data Support Division, U.S. Environmental Protection
Agency; 1980.
Tabak, H.H.; Quaves , A.; Mashni, C.I.; Barth, E.F. Biodegradability
studies with priority pollutant organic compounds. Cincinnati, OH:'
Environmental Research Laboratory, U.S. Environmental Protection Agency;
1980 .
10-8
-------
U.S. Environmental Protection Agency (U.S. EPA). Environmental
modeling catalogue. Contract EPA 68-01-4723. Washington, DC:
Management Information and Data Systems Division, U.S. Environmental
Protection Agency; 1979.
U.S. Environmental Protection Agency (U.S. EPA). Exposure analysis
modeling system AETOX 1. Athens, GA: Environmental Systems Branch,
Environmental Research Laboratory, Office of Research and Development,
U.S. Environmental Protection Agency; 1980.
Weast, R. ed. Handbook of chemistry and physics. 55th ed. Cleveland,
OH: Chemical Rubber Company; 1974.
Wilson, J.T.; Enfield, C.G.;Dunlap, W.J.; Cosby, R.L.; Foster, D.K.;
Baskin, L.B. Transport and fate of selected organic pollutants in a
sandy soil. Ada, OK: Robert S. Kerr, U.S. Environmental Protection
Agency; 1980.
10-9
-------
10.4 MONITORING DATA AND ENVIRONMENTAL DISTRIBUTION
Data Bases
Environmental Contaminant Monitoring Program - U.S.D.I.
Pesticide Soils Monitoring Program - U.S. EPA
NASQAN - National Stream Quality Accounting Network, USGS
SAROAD - Storage and Retrieval of Aerometric Data, U.S. EPA
Research Triangle Park
STORET - Storage and Retrieval of Waste Quality File, Water
Quality, File, U.S. EPA
Literature
Bellar, T.A.; Lichtenberg, J.J.; Kroner, R.C. The occurrence
of organohalides in chlorinated drinking waters. Report No.
670/4-74-008, U.S. Environmental Protection Agency;
November 1979.
Bozzelli, J.W.; Kebblekus, B.B. Analysis of selected volatile
organic substances in ambient air. Trenton, NJ. New Jersey
Department of Environmental Protection; 1979. Available from:
NTIS, Springfield, VA; PB 80 14469 4.
Coniglio, W.A. ; Miller, K. ; MacKeever, D,, The occurrence of
volatile organics in drinking water. Washington, DC:
Criteria and Standards Division, U.S. Environmental Protection
Agency; 1980.
Ewing, B.B.; Chian, E.K. Monitoring to detect previously
unrecognized pollutants in surface waters. Report No. EPA
560/6-77-015a. Washington, DC: Office of Toxic Substances,
U.S. Environmental Protection Agency; 1979.
McConnell, G.; Ferguson, D.M.; Pearson, C.R. Chlorinated
Hydrocarbons and the Environment. Endeavor 34(121):13-18; 1975,
National Academy of Science (NAS). Drinking water and health.
Washington, DC: U.S. Environmental Protection Agency; 1977.
Available from NTIS, Springfield, VA; PB 269519
Pearson, C.R.; McConnell, G. Chlorinated. C and C? hydrocarbons
in the marine environment. Proc. R. Soc. London B 189:305-
322; 1975.
10-10
-------
Pellizzari, E.D.; Erickson, M.C.; Zweidinger, R.A. Formula-
tion of a preliminary assessment of halogenated organic com-
pounds in man and environmental media. Washington, DC: U.S.
Environmental Protection Agency; 1979. Available from NTIS,
Springfield, VA; PB 80 112170.
Symons, J.M.; Bellar, T.A.; Carswell, J.K.; DeMarco, J.;
Kropp, K.L.; Robeck, G.G.; Seeger, D.R.; Slocum, C.J.;
Smith, B.L.; Stevens, A.A. National organics reconnaissance
survey for halogenated organics. J. Am. Water Works Assn.
67:634-646; 1975.
U.S. Dept. of Health, Education and Welfare (HEW), 1970.
Community Water Supply Study, Public Health Service, Environ-
mental Health Service, Bureau of Water Hygiene.
U.S. Environmental Protection Agency (U.S. EPA) National
Organizations Monitoring Survey. Unpubl. Washington, DC:
Technical Support Division, Office of Water Supply, U.S.
EPA; 1978.
U.S. Food and Drug Administration. 1974. Total Diet Studies.
Compliance Program Evaluation.
10-11
-------
10.5 HUMAN EXPOSURE AND EFFECTS
Computer-Based lexicological Information Services
Database
CANCERLIT
CANCER PROJ
EMIC
ETIC
EXCERPTA MEDICA
MEDLINE
NTIS
RTECS
TDB
TOXLINE/TOX3ACK
Literature
Accessible Through Database Content
National Library of Medicine All asoects of cancer
National Library of Medicine Ongoing cancer research
projects for most recent
3 years
Department of Energy
Department of Energy
Lockeed's DIALOG Service
Chemical mutagenesis
Chemical teratogenesis
Human medicine and related
disciplines
National Library of Medicine International biomedical
literature
Lockheed's DIALOG Service Government-sponsored re-
search plus analyses pre-
pared by federal agencies
or their contractors/
grantees
National Library of Medicine Acute toxicity, eye/skin
irritation, recommended
exposure levels
National Library of Medicine Chemical, pharmacologic,
and toxicological data
extracted from 80 standard
reference textbooks, mono-
graphs
National Library of Medicine Human and animal toxicity,
effects of environmental
chemical pollutants published
within or prior to the last
5 vears
Adamson, F.; Gilbert, D.; Pervak, J.; Scow, K.; Wallace, D.
Identification and evaluation of waterborne routes of human
exposure through food and drinking water. Dra^t Report
Contract EPA 63-01-3857, Task 4. Washington, DC: Monitorin-
and Data Support Division, Office of Water Regulations and
Standards, U.S. Environmental Protection Agencv; Jan. 1980.
10-12
-------
Casarett, L.J. Casarett and Doull's Toxicology: The Basic
Science of Poisons, 2nd ed. New York: Macmillan Publishing
Co., Inc.; 1980.
Clayton, G.D.; Clayton, F.E. Patty's Industrial Hygiene
and Toxicology, 3rd revised edition. New York: John Wiley
and Sons; 1981
Gosselin, R.E.; Hodge, H.C.; Smith, R.P.; Gleason, M.N.
Clinical Toxicology of Commercial Products, 4th ed.
Baltimore: The Williams & Wilkins Co.; 1976.
International Agency for Research on Cancer (IARC). Monographs
on the Evaluation of Carcinogenic Risk of Chemicals to Man
(series). Lyon, France.
International Commission on Radiological Protection (ICRP).
Report of the Task Group on Reference Man. New York, NY:
Pergamon Press; adapted October, 1974.
National Academy of Sciences (NAS). Drinking water and health.
Washington, DC: NAS; 1977.
National Cancer Institute (NCI). Bioassays of compounds for
possible carcinogenicity (series). Washington, DC: NCI
National Institute for Occupational Safety and Health (NIOSH).
The registry of toxic effects of chemical substances. HEW
Public. No. (NIOSH) 76-191. Washington, DC: U.S.DHEW; 1976.
Sax, N.I. Dangerous properties of industrial materials,
New York: Reinhold; 1979.
Shepard, T.H., ed. Catalog of teratogenic agents, 3rd edition,
Baltimore: John Hopkins University Press; 1980.
Survey of Compounds Which Have Been Tested for Carcinogenic
Activity (series). Washington, DC: HEW GPO
U.S. Department of Agriculture (U.S.DA). Food consumption
of households in the United States, spring 1965. Report No.
11, Food and Nutrient Intake of Individuals in the United
States. Stock No. 0100-1599. Washington, DC: U.S.
Government Printing Office; 1972.
U.S. Department of Agriculture (U.S. DA). Nationwide
Food Consumption Survey 1977-1978. Preliminary Report No.
2. Food and nutrient intakes of individuals in 1 day in
the United States. Spring 1977. Washington, DC: Science
and Education Administration, U.S. DA; 1980.
1O TO
J-vJ— IJ
-------
U.S. Food and Drug Administration (U.S. FDA). Compliance
program evaluation FY 1974. Total diet studies. Washington
DC: Bureau of Foods, U.S. FDA; 1977.
U.S. Environmental Protection Agency (U.S. EPA). Guidelines
and methodology used in the preparation of health effect
assessment chapter of the consent decree water criteria
documents. Federal Register 44(52): 15641; 1979.
U.S. Environmental Protection Agency (U.S. EPA). Water
quality criteria documents availability. Federal Register
45(231):79318-79384; November 28, 1980.
10-14
-------
10.6 EFFECTS AND EXPOSURE— NON-HUMAN BIOTA
Periodicals
Archives of Environmental Contamination and Toxicology
Biological Bulletin
Bulletin of Environmental Contamination and Toxicology
Bulletin of the Japanese Society of Scientific Fisheries
Comparative Biochemistry and Physiology
Ecology
Environmental Pollution
Fishery Bulletin
Hydrobiologia
Journal of Experimental Marine Biology and Ecology
Journal of Fish Biology
Journal of Fisheries Research Board of Canada
Journal of Great Lakes Research
Journal of Invertebrate Pathology
Journal of Marine Biological Association
Journal of Zoology
Marine Biology
Marine Fisheries Review
Pesticides Monitoring Journal
Proceeding of the National Academy of Sciences
Progressive Fish Culturist
Transactions of the American Fisheries Society
Water Pollution Control Federation
Water Research
Water Resources Bulletin
Other Sources
National Academy of Sciences, Washington, DC: studies of
pollutants .
U.S. Atomic Energy Commission (U.S. AEC) . Toxicity of power
WASH~1249' VC-11'' Washington,
U.S. Environmental Protection Agency (U.S. EPA). Water qualit"
criteria documents availability. Federal Register 45P31) • '
/9318-79384; Nov. 28, 1980. '
U.S. Environmental Protection Agency (U.S. EPA). Reported fish
kills. Washington, DC: Monitoring and Data Support Division "
Office of Water Regulations and Standards, U.S. EPA, constant!-/
updated tile.
10-ij
-------
10.7 RISK ESTIMATION
Albert, R.E.; Train, R.E.,; Anderson, E. Rationale developed by the
Environmental Protection Agency for the assessment of carcinogenic
risk. J. Nat. Cancer Inst. 58:1537-1544; 1977.
Armitage, P.; Doll, R. Stochastic models for carcinogenesis.
In Lecam and Neyman (eds.) Proceedings of the Fourth Berkeley
Symposium on Mathematical Statistics and Probability, No. 4.'
Bliss, C.I. The method of probits. Science 79:38; 1934.
Brewen, J.G. Human genetic risk assessment for chemical substances.
Regulatory Toxicology and Pharmacology, 1:78-83; 1981.
Brown, C.C. The statistical analysis of dose-effect relationships.
In Butter, G.C. (eds.) Principles of Co-Toxicology. New York N Y •
Wiley & Sons; 1978. ' "
Brown, S.M. The use of epidemiologic data in the assessment of
cancer risk. J. Environ. Path. Toxicol. 4:573-580; 1980.
Carlborg, F.W. Dose-response functions in carcinogenesis and the
Weilbull Model. Fd. Cosmet Toxicol. 19:255-263; 1981.
Carter, L.J. Dispute over cancer risk quantification. Science
203:1324-1325; 1979.
Chand, N.; Hoel, D.G. A comparison of models for determining safe
levels of environmental agents. In Proschan, F.; Serfling, R.J.
(eds.) Reliability and biometry statistical analysis of life-
length. Philadelphia, PA: SIAM; 1974.
Cornfield, J. Carcinogenic risk assessment. Science 198:693-699;
Cramer, G.M. ; Ford, R.A. ; Hall, R.L. Estimation of toxic hazard—
a decision tree approach. Cosmet. Toxicol. 16:255-276; 1978.
Crump, K.S.; Hoel, D.G.; Langley, C.H.; Petro, R. Fundamental
carcinogenic processes and their implications for low-dose risk
assessment. Cancer Res. 36:2973; 1976.
Eschenroeder, A. A summary of quantitative risk assessment
approaches for chemicals :Ln the environment. Washington, DC:
Edison Electric Institute:; 1980.
Gehring, P.J.; Watanabe, P.G.; Park, C.N. Resolution of dose-
response toxicity data for chemicals requiring metabolic activation:
example - vinyl chloride. Toxicol. Appl. Pharnacol. 44-581-=:91-
1978.
10-16
-------
Gillette, J.R. Application of pharmacokinetic principles in the
extrapolation of animal data to humans. Clinical Toxicology
• j •!. .7 / O •
ooA reSulation of carcinogenic hazards. Science 208:
; 1980.
Guess, H.; Crump, K. ; Peto, R. Uncertainty estimates for low-dose
37:3475-3^3 1977S °f animal CarcinoSenicit^ d*ta. Cancer Res.
Hartley, H.O.; Sielken, R.L. , Jr. Estimation of "safe doses" and
carcinogenic experiments. Biometrics 33:1; 1977.
Higginson, J.; Muir, C.S. Environmental carcinogenesis: miscon-
Hoel, D.G.; Gaylor, D.W.; Kirschstein, R.L. Saffiotti U •
Schneiderman, M.A. Estimation of risks of irreversible, delaved
toxicity. J. Toxicol. Environ. Hlth. 1:133-151; 1975.
Kensler, C.J. Experimental procedures in the evaluation of chemical
aspects of
Leape J.P Quantitative risk assessments in regulation of environ
mental carcinogens. Harvard Environ. Law Rev. 4:86; 1980.
Mantel, N.; Bohidar, N.R,; Brown, D.C.; Ciminera, J.L.; Tukev J W
An improved "Mantel-Bryan" procedure for "safety testin," of
carcinogens. Cancer Res. 35:759; 1971.
of
nlr' HHC' RelationshlPs of bioaasay data on chemicals to their
for
Raabe, O.G.; Book, S.A.; Parks, SI. Bone cancer from radium:
explains data £or mi
10-17
-------
Truhaut, R. An overview of the problem of thresholds for chemical
carcinogens. In Davis, W.; Rosenfeld, C. (eds.) Carcinogenic risks/
strategies for intervention. Lyon, Fr.:IARC Scientific Publica-
tions No. 25; 1979: 191-202.
Whittemore, A.S. Mathematical models of cancer and their use in
risk assessment. J. Environ. Pathol. Toxicol. 3:353-362; 1980.
10-13
-------
APPENDIX A. MATHEMATICAL DETAILS OF RISK CALCULATIONS
A.I INTRODUCTION
In rare instances, epidemiological studies provide direct informa-
tion about the adverse effects of a substance to humans. However, even
in these instances, the actual exposure intensity and duration that was
responsible for these effects may be difficult to quantify. Consequently,
most risk estimates for humans are based upon laboratory studies with ex-
perimental animals. It is preferable to base an estimate upon several
studies with different species, in which case a range of potency may be
established for the pollutant in question. Wherever possible, confidence
limits and measures of statistical significance should be introduced to
qualify these results. Even in a single experiment, different interpre-
tations of the data may lead to uncertainty about the response. Further,
the extrapolation from responses in laboratory animals administered high
doses to the low doses characteristics of most human exposure to environ-
ment pollutants is necessarily an uncertain process. Hence several
different methods should be used to establish the range of potential risk.
A dose-response curve can be defined as a relationship between the
amount or rate of the chemical administered and the probability of the
subject experiencing an adverse effect at that dose. Hence, the curve is
a cumulative probability distribution function and should increase from
zero to one, assuming that higher doses are increasingly more toxic.
The estimation of risk on the basis of experimental data involves
the selection of a hypothetical dose-response curve, and the fitting of
the parameters of this curve to the data. Although ideally algorithms
would exist for calculations of risk due to all types of health hazards,
at the present time the only effect for which a substantial body of theory
exists is carcinogenesis. Three different types of dose-response models
for assessing cancer risk are described in the following sections.
linear non-threshold model (Chand and Hoel 1974, Cornfield
1977, NAS 1977) is based upon the "one-hit" principle, which
asserts that each molecule of the substance has an equal probabilitv
of producing a specific effect. The resulting dose-response curve
is exponential, and is approximated by a linear curve in the low-
dose regions. This method tends to produce "upper bound" risk
estimates, which probably exceed the actual risk to humans.
Tne Mantel-Bryan (log-probit) model (Bliss 1934, Mantel et_ al.
1971, Mantel and Bryan 1961) assumes that the susceptibiTity"of
receptor organisms is normally distributed with respect to the
log of the dose. Hence the cumulative distribution of response
is the integral of a log-normal density function. The resulting
dose-response curve has an S-shape, and usually yields lower
estimates of risk because of the implied threshold effect.
A-l
-------
• The multi-stage model has been independently developed by several
investigators (Armitage and Doll 1961, Crump e_t al. 1976, Crump
et. al. 1977, Hartley ar.d Sielken 1977) and generalizes the linear
one-hit model to allow polynomial functions of dose and time. It
often reduces to a linear model in the low-dose region, but pro-
vides a better fit when, the data indicate that there is a threshold
effect.
These three methods and their application are discussed in greater detail
below.
A.2 HUMAN EQUIVALENT DOSES
The calculation of the equivalent human dose for extrapolation of
animal data can usually be accomplished by simply multiplying the human
weight (70 kg) by the animal dose expressed in mg/kg per day. To cover
situations in which doses are expressed in different units, a procedure
was developed to compute equivalent human doses for any experimental
situation. The following formulae require knowledge of the weights,
dietary intakes, and respiration rates of the experimental animals.
Let D , D denote human and animal doses respectively,
W , W denote human and animal weights (kg),
I ,1. denote dietary intakes (kg/day), and
n, A.
•^
R ,R denote respiration rates (m /day)
n A
Four cases are addressed corresponding to four different units of measure
for the animal dose D^. In each case, the equivalent human dose is com-
puted by normalizing the intake relative to body weight. For skin absorp-
tion, however, DA refers not to an estimated intake, but simply to a
concentration in water.
(i) D expressed in ug/day:
W
DH (ug/day) = DA • Ji-
(ii) D expressed as ppb in diet:
DH (ppb) = D • -£ • -A
H A WA IR
W
D (Ug/day) = D I • -£
n A A W,
A
A-2
-------
(i;Li) D expressed as yg/1 for skin contact:
DH (ug/day) = 0.002 DA
(this assumes human absorption of 2 ml/day of
water containing a pollutant)
(iv) DA expressed as ng/m3 for inhalation
3 W R
DR (ng/m ) . DA . Ji ,
A
w
(yg/day) = D, R • — • 10
A A W,
A
In order to facilitate the application of these formulae, the conversion
cnart in Table A-l shows numerical conversion factors for experiments
with mice and rats. The approach for converting acute doses is entirely
analogous, except that the units of intake are ug rather than ug/day. '
In the case of carcinogenic effects, it is sometimes assumed that
equivalent doses are proportional to body area. The U.S. EPA recommends
me m^0?n™f desi^ating equivalent doses of carcinogenic substances
(U.S. SPA 1979). Presumably this method reflects a view that carcinogenesis
is related to the area of some physiological membrane. It is also the
logical choice when the route of exposure is skin absorption. Thus:
Weight of substance to which human is exposed
Body surface area of human
is equivalent to
Weight of substance to which animal is exposed
Body surface area of animal
Because the surface areas of similar solids are proportional to the squares
of corresponding linear dimensions and volumes or weights are proportional
to the cubes of corresponding linear dimensions, one can approximate sur-
face area by weight raised to the two-thirds power. Thus
(Body surface area) is proportional to (Body Weight) 2/3
A-3
-------
TABLE A-l. FACTORS FOR CONVERTING DOSES FROM LABORATORY
ANIMAL STUDIES TO HUMAN EQUIVALENT DOSES
ASSUMPTIONS
a
Weight
Species (kg)
Human 70
Mouse 0.025
P-
4>
Uat 0.3
Rate of
Ingeation
(kg/day)
4
0.003
0.015
Rate of
Respira-
tion
(mVday)
]0.7
0.033
0.14
Dose
1 ug/day
1 ppb
1 ue/1
1 ng/m3
1 ug/day
1 ppb
1 Ug/l3
1 ng/m
HUMAN EQUIVALENT DOSE
IJg/day
Total Food
2800
8.4 4.2
0 . 00?
0.09
233
3.5 1.75
0.002
0.03
ug/1
Contact
1
1
ng/m
Breathing
8.6
3
Equivalent human doses are assumed proportional to weight and rate of intake. Rates of respiration
are based on minute volume while resting (Spector 1956).
Note: Adult mice and rats are usually heavier than 25 and 300 g. Thus the numbers above must be
used with caution. If surface area is used as a normalizing factor rather than body weight,
then human equivalent doses should be reduced by a factor of 14 for mice and a factor of
6 for rats.
-------
This has the effect of reducing equivalent human doses by about 14 in
the case of mice, and by about 6 in the case of rats (Table A-l).
In many experiments no explicit account is taken of time. However
time is an explicit variable in the multi-stage extrapolation described'
below, and it is an implicit variable in all other extrapolations. Cancer
occurrence increases with age in both humans and experimental animals
and is a multi-step, time-dependent process. For mice and rats the inter-
val of the usual life span of 2-3 years is the normal exposure interval
for a valid carcinogenicity study. The most plausible assumption is that
the normal lifetime of an experimental animal is equivalent to the normal
lifetime of a human.
Procedures for estimating human equivalent doses often account for
time factors in several ways:
• If the substance was administered to experimental animals on an
intermittent basis, e.g., 5 days per week, then the assumed dose
is reduced by a corresponding factor, e.g., 5/7.
• If the duration of exposure was shorter than the experimental
animal lifetime, then the assumed dose is reduced by the ratio
of those times, i.e., duration/lifespan.
It should be noted that such procedures do not take into account the
relevant phannacokinetics, and are merely a device for reducing the dose
and hence obtaining a more conservative risk estimate. They generally
do not affect the results by more than a factor of 50%, which is small
compared with the total range of uncertainty. Uncertainties due to dose
estimation and techniques of extrapolation can often span several orders
of magnitude.
A. 3 ONE-HIT MODELS
The "one-hit" models describe a mechanism of carcinogenesis in which
a single event triggers the cancer. The event may be a molecule of the
carcinogenic substance reaching a suitable receptor site or it may be
some more complex but undescribed happening. The basic supposition is
that the probability of this event in a short time interval dt (short
compared witn the time of observation, which is usually the major part
of a lifetime) is proportional to the duration of the time interval The
factor of proportionality, called the "hazard function," may be a function
of time and or the dose level. Thus the hazard function is written
h(x,t) where x is the dose level and t is time, measured in lifetimes.
to dost:^ Unear extrapolation- this function is taken to be proportiona
h(x,t) = bx _
A-5
-------
where b is a factor of proportionality and x is the dose level. The
probability of a "hit," that is, initiation of cancer, is proportional
to the duration of a time interval:
P (initiation of cancer in the interval (t,t = dt)}
= bx dt
In this form one can see the reason for calling the extrapolation linear—
doubling the dose, for example, doubles the probability of initiating
cancer.
The multi-stage model generalizes this approach, replacing the
product bx by the product of two polynomials, one a polynomial in t, the
other a polynomial in x.
h(x,t) = IV t1 £b.x:i (A-2)
This approach allows a great deal of flexibility, though as a practical
matter the polynomials cannot be too large or one loses track of their
significance.
Let Q(x,t) be the probability that no cancer has been initiated in
the interval (0,t) when the animal has been subjected continually since
time 0 to a dose level of x. The second underlying assumption is that
the probability of initiating cancer at any particular time is independent
of whether or not cancer has previously been initiated; this assumption
allows us to multiply probabilities and
Q(x,t + dt) = Q (x,t) (1 - h(x,t)dt). (A-3)
Rearranging terms
|£+h(x,t)Q = 0
ot
Q(x,t) = exp [- /Qh(x,t)dt] (A-4)
A-6
-------
Let P(x,t) be the probability that cancer is initiated in the interval
(0,t)
P(x,t) - 1 - Q(x,t)
= 1 - exp [-/Ti(x,t)dt] (A-5)
P(x,t) can also be identified as the cumulative distribution function
for the random variable "time until cancer is initiated" — that is, the
probability that the time until cancer is initiated is less than t is
P(x,t). The derivative dP/dt is the probability density function f(x,t)
for these intervals
«"'> ' I
= h(x,t) exp [-/Ch(x,t)dt]
o
= h(x,t) Q(x,t) (A-6)
The function f(x,t)dt has the interpretation of the probability that
cancer is initiated in the interval (t,t+dt). Finally, we note that
h(x,t)dt = f (x,t)dt/Q(x,t) can be interpreted as the probability that
cancer is initiated in the interval (t,t+dt) given that it has not been
initiated up until time t.
A. 4 LINEAR EXTRAPOLATION
In the linear extrapolation h(x,t) is given by Equation (A-l) .
Q(x,t) = e~bxt (A_7)
P(x,t) = 1 - e' (A_8)
If there is a "background" or "spontaneous" cancer incidence present
even when the dose rate of the toxic substance is zero, then one can use
either of two smaller approaches. One can assume that h(x,t) has the
form
h(x,t) = bx + c
(A-9)
-------
in which case
Q(x,t) = e~Ct e~
P(x,t) - 1 - e~Ct e (A-ll)
At Che conclusion of the experiment one can identify the fraction of
animals in the control group which contracted cancer in spite of the fact
they were not intentionally subjected to a carcinogen as the probability
9 of "spontaneous" cancer. Then it is easy to see from Equations
(A-10) and (A-ll) that
Q(x,t) = (1 - 9)e
P(x,t) = 1 - (1 - 6)e~bxt (A-13)
and hence
9=1- e~Ct (A-14)
Having measured 9 by observing response in the control group, one could
infer the value of c, but this is usually not done.
Alternatively, one can determine both parameters b and c by a data-
fitting procedure involving the maximum likelihood estimators. This is
the multi-stage procedure when two non-zero parameters in the polynomials
are permitted. It is detailed below.
We have noted that ?(x,t) is the cumulative distribution function
for the random variable time-to-initiation-of-cancer. P(x,t) as a func-
tion of t has the form shown in Figure A-l.
y t or x
FIGURE A-l. CUMULATIVE DISTRIBUTION FUNCTION ?(x,t)
A-8
-------
Obviously, from Equation (A-13), P(x,t) considered as a function of x
has the same form as that when it is considered a function of t. For
this reason it is common to regard P(x,t) as the cumulative distribution
function of a random variable called susceptibility. Thus, the fraction
of experimental animals in whom cancer is initiated at dose x is regarded
as a group of animals in whom tolerance is less than or equal to x. This
logical transition from a determinate parameter of the experiment, the
dose x, to a random variable, tolerance, is plausible but not necessarily
correct. P(x,t) is measured by giving a group of animals the dose x and
noting the fraction in whom cancer is initiated. It is conceivable that
P(x,t) as a function of x could increase as the dose increases and then
decrease if, for example, high doses brought into play a new mechanism,
such as vomiting, not encountered at low doses. It is important to recog-
nize that consideration of the dose-response curve as the cumulative distri-
bution function of a random variable, tolerance, requires the supposition
that P(x,t) as a function of x increases monotonically.
The most rigorous mathematical procedure for estimating the para-
meters b and c (or b and 6) is maximum likelihood estimation. More
commonly, experimenters simply plot the fractions of animals in which
cancer is found at the conclusion of the experiment at time T and use
these fractions as estimates of the parameter P(.x,t). In the linear
extrapolation, it is customary to expand P(x,T) about zero.
?(x,T) - 1 - (1 - 0) (1 - bxT)
- 9 + bxT (A-14)
and to fit a straight line to the data points P' (xi,T) =
P(x-j_,T) - 3 = bxj_T. Because the expansion in Equation (A-14) is
valid only for small x, one should use data where x is as small as
possible.
It should be noted that the crucial supposition in the linear model
is the selection of the form for h(x,t) given in Equation (A-l). The
consequence of this selection is that the curve for P(x,t) in the
vicinity of x = 0 will have a positive slope. Because the data are
gathered for relatively large values of x, the experiment does not test
this supposition and therefore belief in the linear model amounts to an
assumption. It is often argued that the real situation is unlikely to
be worse than this and that therefore the linear model provides a con-
servative estimate of risk.
A.5 LOG-PROBIT EXTRAPOLATION
Log-probit extrapolation (also called Mantel-Bryan extrapolation
after the original proponents of the method) assumes a different form
for the dose-response curve. Specifically, it assumes that the logarithm
A-9
-------
of tolerance is normally distributed. There is little theoretical
underpinning for this assumption, though a number of physiological
variables (such as heights of humans) seem to follow this log-normal
distribution.
The mathematical form of the dose-response curve is
P(x,T)
a + b log1Qx
- 1/25
(A-15)
On log-normal graph paper this function is a straight line with a slope
of b probits (standard deviations) per decade of change in the dose x
and a y-intercept of a probits, as shown in Figure (A-2).
The steeper (or the larger) the slope of this line the smaller the
values of P(x,t) found by extrapolations to small values of x (large
negative values of log x). One could plot the data and fit a straight
line, either by eye or by some analytical procedure:, thus finding values
for the parameters a and b. However, the more common procedure is to
set b = 1 and to find the best fit to the data of a line with this slope
It is argued that b = 1 is the shallowest slope observed over a wide
variety of carcinogens already studied and that selecting a line with
unit slope is therefore conservative.
-a 0 log x
FIGURE A-2. LOG-PROBIT FORM FOR DOSE-RESPONSE CURVE
A-10
-------
The supposition that the slope b - 1 is, of course, somewhat arbitrary
r^ J significant than the equally arbitrary supposition that
the dose-response curve should be described by Equation (A-15).
in^ r that f°r large ne§ative values of the variables of
integration 5 the integrand in Equation (A-15) is very small. In fact
the function exp (-52) and all its derivatives vanish' as t- - -» For '
this reason the log-profait extrapolation generally yields extremely low
actuallvV eXtrap°lated risks at low dose levels. For the numbers
on tS otherOUtnfd1ffn experjments on one h^ and in the environment,
™1£ ! ? u ? dlfference between extrapolated risk by the log-probit
6 llnar raethod ca» ^e several
several orders of magnitude. he
lo nrnr . magntue. e
If extrapolation is therefore regarded as a probable low estimate
of
r smat
risk and the two methods are often taken as a way of bracketin*
the true risk. It is important to recognize that both method, produce
results strongly dependent on the presumed form of the dose-response '
It should also be noted that this method can deal only with the
aifference between response at a dose x and response when x = 0 (the
spontaneous" rate), since log 0 - -~. Thus, there is no Possibilitv
of titting a line or a curve through the datum when x - 0.
Multi-Stage Extrapolation
exv extraP°lation ^ a one-hit model in which more
thus nt 1? f ^ the f°rm °f the hazard function h(x,t). It
addition Sf ^ Presuppositions about the dose-response curve. In
' f T & mOre rig°r°US "«hematical procedure for
estimat * f ca proceure fo
estimating the value of the unknown parameters, and it takes exnlic-'r
the Ci— -^^tion-of-cancer whoever theL dat^are
In order to avoid confusing generality we shall adopt a soecific
h(x,t) - ax+bx+c
to be £ound from
P(x,t) = 1 - exp [-(ax2 + bx + c)t] (A_17)
S'c 'Ythef^sl "^ doae-rMP°°" C— « any time. We note that
c ^ u tnere is a spontaneous" incidence of cancer; if b * 0, the
A-ll
-------
2
extrapolation to low doses is linear; by including the term ax we allow
for the possibility that response at low doses may not be linear, in
which case we would expect to find a £ 0, b = 0.
Most available data record only the number of animals which had
cancer at the conclusion of the experiment. For such experiments we
have a sample of P(x,T) for various values of x. Sometimes the number
of animals with cancer at some intermediate time is reported in which
case we have samples of P(x,T-j) for other values of T and various values
of x. Occasionally when the cancer can be detected without killing
the animal (as in the case of skin cancers) one has the time-to-initiation-
of-cancer as well as dose for each animal. The multi-stage method
accommodates all these possibilities and produces the values of a, b,
and c which best fit the totality of the data.
Since none of the data available in these studies records time-to-
time initiation-of -cancer for each animal, we shall not carry the terms
necessary to accommodate these data in the mathematical development
which follows. The likelihood function L is the product of terms of
the form P(x,t) given by Equation (A-17) and its complement Q(x,t) =
1 - P(x,t).
N n. N. - n
L = n P(x.,t ) X Q(x.,t ) 1 ± (A-18)
where i is the index for the S experimental data points , each of which
has a value for x^ and t^ associated with it; and n^_ is the number of
positive responses, Nj_ is the total number of animals and N^ - n^ is
the number of negative responses
log L =2 n. P(x.,t.) + (N. - n.) Q(x.,t.) (A-19)
111 ii -LI
The standard procedure is to differentiate Equation (A-19) with respect
to the unknown parameters and. to set the derivatives equal to zero in
order to find the maximum of log L. However, it is more convenient
simply to seek the maximum of log L by a hill-climbing method. One
selects initial values of a, b, and c, and then evaluates log L with
small changes, first of a, then of b, then of c, then of a again, etc.,
continuing this procedure until log L begins to decrease. In this way
one can find the values of a., b, and c that maximize log L or L. These
are the maximum likelihood estimators.
If it should turn out that a = 0, then the result obtained is
equivalent to the linear extrapolation found earlier. In the former
case, however, we imposed the form from the beginning; here we have
allowed the data to produce the form. Note, too, that this method
A-12
-------
automatically introduces variations in the duration of an experiment
and that it automatically weights data where many animals are involved
more heavily than data where only a few are involved. For these reasons
the multi-stage method seems superior to either the linear extrapolation'
or the log-probit extrapolation.
A-13
-------
REFERENCES
Armitage, P.; Doll, R. Stochastic models for carcinogenesis. In
Lecam and Neyman (eds), Proceedings of the Fourth Berkeley Symposium
on Mathematical Statistics and Probability, No. 4.
Chand, N.; Hoel, D.G. A comparison of models for determining safe levels
of environmental agents. In Proschan, F. ; Serfling, R.J. (eds),
Reliability and Biometry Statistical Analysis of Lifelength. Philadelphia,
PA: SIAM; 1974.
Cornfield, J. Carcinogenic risk assessment. Science 198:693-699; 1977.
Crump, K.S.; Hoel, D.G.; Langley, C.H.; Peto, R. Fundamental carcinogenic
processes andtheir implications for low-dose risk assessment. Cancer
Res. 36:2973; 1976.
Crump, K.S.; Guess, H.A.; Dial, K.L. Confidence intervals and tests of
hypotheses concerning dose response relations inferred from animal
carcinogenicity data. Biometrics 33:437; 1977.
Hartley, H.O.; Sielken, R.L., Jr. Estimation of "safe doses" and
carcinogenic experiments. Biometrics 33:1; 1977.
Mantel, N. Bohidar, N.R. ; Brown, D.C.; Ciminera, J.L,,; Tukey, J.W. An
improved "Mantel-Bryan" procedure for "safety testing" of carcinogens.
Cancer Res. 35:759; 1971.
Mantel, N.; Bryan, W.R. "Safety" and testing of carcinogenic agents.
J. Nat. Cancer Inst. 27:455; 1961.
National Academy of Sciences (NAS). Drinking water and health.
Washington, DC: NAS; 1977.
Spector, W.S. (ed.) Handbook cf biological data. Philadelphia. PA:
WB Saunders Co.; 1956.
U.S. Environmental Protection Agency (U.S. EPA). Water quality criteria.
Federal Register 44:1526-1594; 1979*.
A-14
------- |