wEPA
          United States
          Environmental Protection
          Agency
            Office of Research and
            Development
            Cincinnati, Ohio 45263
EPA/600/R-99/025
March 1999
Review Draft
Risk Characterization
Materials for Peer
Review — March, 1999

Review Draft
(Do Not Cite or Quote)

-------

-------
                     Draft Risk Characterization Materials
                          Prepared for Peer Review
                      Scheduled for March 24-25, 1999
                                 Contents
Risk Characterization Handbook Materials
            Charge for the EPA Risk Characterization Handbook

            Draft EPA Risk Characterization Handbook
Risk Characterization Case Study Materials

            Charge for Generic Ketone Risk Characterization
            Draft Generic Ketone Risk Characterization Case Study

•           Charge for Waquoit Bay Watershed Ecological Risk - Planning and
            Scoping                         ,
            Draft Waquoit Bay Watershed Ecological Risk Case Study

•           Charge for Mitec Risk Characterization
            Draft Mitec Risk Characterization Case Study

•      .     Charge for Midlothian Risk Characterization
            Draft Midlothian Risk Characterization Case Study

-------

-------
                 Charge for the Risk Characterization Handbook

        The Risk Characterization Handbook offers a single, centralized form of risk
 characterization implementation guidance for Agency risk assessors and managers. The goal of
 risk characterization is to clearly communicate the strengths and limitations of the risk
 assessment so its use in decision making can be put into context with ,the other information
 critical to evaluating options for rules, regulations and negotiated agreements (e.g., economics,
 social values, public perception, policies, etc.).                ''

        On March 21,1995, an Agency-wide Risk Characterization Policy was issued that called
 for all risk assessments performed at EPA to include a risk characterization to insure that the risk
 assessment process is transparent and that risk assessments are clear, reasonable and consistent
 with other risk assessments of similar scope prepared by programs across the Agency. The
 Policy and Handbook do not provide guidance on the conduct of risk assessments -- guidance
 that is provided in EPA's various risk assessment guidelines.

        To implement the Policy, an Agency-wide Risk Characterization Guidance document and
 Program Office- and Region-specific Risk Characterization Implementation Plans (RCIPs) were
 written. Many meetings (colloquia for risk assessors and roundtables for risk managers) were
 held to identify and resolve risk characterization issues.  Based on the experiences of those
 attending the meetings and using these various documents to help characterize risk across the
 Agency since then, ,we found that a single Agency-wide document is needed.  This draft
 Handbook is the product of that effort.

 Points for Peer Reviewers to Address:

 1)    Please comment on the adequacy of the Handbook guidance for helping risk assessors
       express results clearly, articulate major assumptions and uncertainties, identify reasonable
       alternative interpretations, and separate scientific conclusions from policy judgments.

 2)    Is this guidance comparably useful for health and for ecological risk assessments? If not,
       please identify any special guidance needed for either of these general areas.

 3)    Given the Risk Characterization Policy's requirements for Transparency, Clarity,
       Consistency, and Reasonableness (TCCR), please comment on the adequacy and utility
       of the Handbook guidance for helping decision-makers and other risk managers
              a)      understand risk assessment findings, and
              b)      use these findings in decision makingi

 4)    Please comment on the utility of the Handbook guidance for helping assessors and
       managers ensure that usually "unseen" aspects of the assessment are brought forward to
       make clear to decision-makers and the public choices made during the risk assessment
       and why they were made.

• •  '.  '     - "• ' •           .,                1              '•.''."':'•

-------
                                                                              Handbook
5)     Please comment on the utility of the Handbook to help those outside EPA understand the
       Agency's risk characterization goals and to help evaluate EPA's progress toward
       achieving those goals.

6)     Please comment on the overall utility and adequacy of the criteria for judging
       transparency, clarity, consistency and reasonableness (i.e., TCCR).  Your review of the
       case study(s) assigned to you will assist your assessment of these criteria.

7)     While the primary focus of this peer review is on the risk characterization guidance in the
       draft Handbook, please comment on the usefulness of Appendices B, C, and D for risk
       assessors and risk managers.

8)     EPA welcomes any additional comments or suggestions.

-------
 1
 2
 3
 4
 5
 7

 8
 9

10

11

12

13
14
15
16
17
18
19
20
•21
22
23
24
25
December 8,1998
C:\DOS\RCHDBK4A.DFT.wpd

  EPA 100-B-99-OOX
     December 1998
                              DRAFT

 U.S. Environmental Protection Agency

      RISK CHARACTERIZATION
                  HANDBOOK
           Prepared for the U.S. Environmental Protection Agency
      by members of the Risk Characterization Implementation Core Team,
             •  a group of EPA's Science Policy Council
                  Science Policy Council
            U.S. Environmental Protection Agency
                  Washington, DC 20460

-------
      Page ii
                     DRAFT Risk Characterization Handbook
1
2
3
4
5
6
7
8
                                DISCLAIMER

      [This document has been reviewed in accordance with U.S. Environmental Protection
Agency policy and approved for publication and distribution. Mention of trade names or
commercial products does not constitute endorsement or recommendation for use.]

NOTE: THIS DOCUMENT HAS NOT YET BEEN PEER REVIEWED FOR DISTRIBUTION

-------
                             DRAFT Risk Characterization Handbook
                                                                           Page iii
.  1
  2
  3
  4
  5 '
  6
  7
  8
  9
10
11
12
13
14
15
16
17
18
19
20
21 .
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
                           TABLE OF CONTENTS

FOREWORD : .		....:. Page vii

SUMMARY OF THIS RISK CHARACTERIZATION HANDBOOK
       	.,..;......;.		. Pagel
      TCCR Points to Consider About Risk Characterization	 Page 2

RISK CHARACTERIZATION GUIDE	,..;...,.	 Page 4

1,  THE NEEDJOR RISK CHARACTERIZATION ......'.,.............,....... Page 5
      1.1     Overview Statement		 Page 5
      1.2     Understanding Risk Characterization	'	 Page 6
           ,1.2,1   What is Risk Characterization? 	...	......... v.... Page 6
             1.2.2   How Do Characterizations Written As Part of Ecological
                   Risk Assessments Differ from Those Written As Part of
                   Human Health Risk Assessments? ..........,...........>... Page 7
             1.2.3   Are Risk Assessment and Risk Characterization the Same? ...... Page 8 '
             1.2.4   What does the Risk Characterization Policy Tell Risk Assessors? . Page 8
             1.2.5   How will Risk Characterization Help the Risk Assessor?	 Page 8
             1.2.6   How will Risk Characterization Help the Risk Manager?	Page 9
             1.2.7   What Is the Difference Between Risk Characterization and
                   Risk Communication? .......... ^ .	 Page 9
             1.2.8   What is the Value of Risk Characterization in the Regulatory
                   Development Process?	 Page 10
             1.2.9   What Role does Risk Characterization have in Regulatory
                   Negotiations?	Page 10.
             1.2.10  Do Risk Characterizations Need Peer Review? .;............. Page 10
      1.3    What the Agency Has Done Since the Policy Was Issued 	....... Page 11
             1.3.1   What Were the Major Messages from the Colloquia
          .         and Roundtables?			 Page 11
      1.4    Need for Risk Characterization Criteria .............	 Page 12
      1.5    The Roles of People and Organizations in Risk Characterization  ....... Page 12 ,
             1.5.1   Who is Ultimately Accountable for-Risk Characterization?  ..... Page 12
             1.5,2   Who Are the Agency Staff Involved in Risk Characterization?  .. Page 13
          ,   1.5.3   Am I a Risk Assessor and What Are My Responsibilities? ...... Page 13
             1.5.4   Ami a Risk Manager and What Are My Responsibilities? ...... Page 14
            1.5.5   Wftat is the Roleof the Science Policy Council (SPCJ?	 .Page 16

-------
       Page iv
                      DRAFT Risk Characterization Handbook
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
             1.5.6  Which Office/Region or Other Agency is Responsible
                   for Writing the Risk Characterization?		Page 16
             1.5.7  What is the Responsibility of Organizations that Submi
                    Risk Assessments to EPA?	Page 16

2.  RISK CHARACTERIZATION IN THE RISK ASSESSMENT PROCESS ........  Page 17
      2.1    Overview Statement		Page 17
      2.2    Criteria for Achieving TCCR	 1	  Page 17
             2.2.1 What are the Criteria for Transparency?	Page 17
             2.2.2  What are the Criteria for Clarity?		Page 18
             2.2.3  What are the Criteria for Consistency?	  Page 18
             2.2.4  What are the Criteria for Reasonableness?	  Page 19
      2.3    Elements of a Risk Characterization That Meet the Criteria for TCCR ..  Page 20
             2.3.1 , What Are the Major Conclusions to Carry Forward?  	Page 20
             2.3.2  What Information Needs to Be Identified During the
                   Risk Assessment Process to Prepare for Risk Characterization? ..  Page 21
             2.3.3  What Are the Key Issues and Their Impact on the Conclusions? .  Page 23
             2.3.4  How Do I Put this Risk Assessment into a Context with
                   Other Similar Risks?	Page 23
             2.4.1  What Populations Must You Specifically Characterize
                   hi Your Risk Assessment?	  Page 24

3.  RISK CHARACTERIZATION PRODUCTS ...........'.	Page 26
      3.1    Overview Statement	Page 26
      3.2    Forms of Risk Characterization			Page 26
             3.2.1  Can Simple "Bright "Lines" or Numbers be the Essential
                   Information in Every Risk Characterization Product?	Page 27
      3.3    Audiences for Risk Characterization Products	Page 27
             3.3.1  Who Are the Audiences for Risk Characterization Products?	Page 27
             3.3.2  Can You Use the Same Risk Characterization Product
                   for All Audiences?	Page 28
             3.3.3  How Do I Ensure that the Irreducible Set of Risk
                   Characterization Information is Carried Forward in All
                   Risk Characterization Products?	  Page 28
      3.4    Risk Characterization Length   	Page 29
             3.4.1  What is the Appropriate Length for a Risk Characterization? ....  Page 29
      3.5    ResearchNeeds	  Page29
             3.5.1  Should Identification of Research Needs Be Partof

-------
                             DRAFT Risk Characterization Handbook
                                                                           Pagev
 1
 2
 ,3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
                   Risk Characterization Products?	Page 29
             3.5.2  Should Decisions Be Delayed Until Research is Completed?	Page 29

 4. ADMINISTRATIVE ISSUES	......:...	.Page 30
       4.1    Overview Statement	Page 30
       4.2    What is the Risk Characterization Record?	Page 30
             4.2.1  How Can the Risk Characterization Record Improve the Risk
                   Characterization Process?	Page 30
             4.2.2  Where Should the Risk Characterization Record be Kept
 .        •  •-        and For How Long?  .			.;	Page31
       4.3    Budget Planning		..		 Page 31
       4.4    Legal Considerations	... Page31
             4.4.1 Are There Legal Ramifications from the Risk: Characterization
:                   Policy? ...I....................-.:.............	Page31
             4.4.2  Is Legal Advice Needed?	 ...PageSl

 References Concerning Risk Characterization	..	Page 32

 APPENDIX A	.-..-..'.	;	................:..	 Page A-l
      "Policy for Risk Characterization ...-.'	 i. Page A-2

 APPENDIXB	.,-.,;	.......:...;	 PageB-1
       ELEMENTS TO CONSIDER WHEN DRAFTING EPA RISK
             CHARACTERIZATIONS	...'.-	...:;. Page B-2

 APPENDIX C. USING RISK CHARACTERIZATION PRINCIPLES TO PLAN AND SCOPE
       A RISK ASSESSMENT .................,.:....................... Page C-l
       1.     Overview Statement	 Page C-l
       2.     Planning and Scoping	 Page C-l
             2.2.1  What Should You Discuss During Planning and Scoping? ..... Page C-2
             2.2.2  Wojuld the Table in Section 2.3.2 Be Helpful During
                   Planning and Scoping?	 PageC-3
             2.2.3  Should the Planning and Scoping Discussion Focus on
                   What the Risk Assessment Results Should Be?  	......... Page C-3
             2.2.4  What Are the Benefits of Planning and Scoping?	 Page C-4
             2.2.6  When Does the Risk Assessor/Risk Manager Dialog-End? ..... Page C-5

 APPENDIX D. ROLE OF SCIENCE IN DECISION MAKING	Page D-l
       1.     Overview Statement	'........	Page D-l

-------
      Page vi
                DRAFT Risk Characterization Handbook
1
2
3
4
5
6
2.    Is the Scientific Risk Assessment with its Characterization the
      Driving Force Behind Decision Making?	Page D-l
3.    What Are the Major Factors that Affect Decision Making?  	Page D-l
4.    Are the Economic and Other Non-Risk Assessments Subject to Risk
      Characterization?	.....	Page D-4

-------
DRAFT Risk Characterization Handbook
Page vii
1 ;
2 : ••'
3 ;
4 Placeholder
5 	 ' 	 ' 	 ;; 	
•'.•'-.•'•• • ' .•'"•••
FOREWORD
. ••••>. •
• • ' • • . •'..•'-'• '
' • • . ' - • • ._ r . • -

-------

-------
                             DRAFT Risk Characterization Handbook
                                                                            Pagel
 1
 2
 3
 4,
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
                  U.S. Environmental Protection Agency

        SUMMARY OF THIS RISK CHARACTERIZATION
                                 HANDBOOK
      This Risk Characterization Handbook is divided in two. The first part is a Risk
Characterization Guide designed to give Risk Assessors, Risk Managers and other Decision-
Makers an understanding of the'goals and principles of risk characterization. The Risk
Characterization Policy calls for a transparent process and products that are clear, consistent
and reasonable. All risk assessments have a product, but effective characterization depends on
transparency, clarity, consistency and reasonableness (TCCR). TCCR is the key to a successful
risk characterization.
            Effective characterization depends on transparency, clarity,
                   consistency and reasonableness (TCCR)
      The appendices make up the second part.  They contain the risk characterization policy,
an updated elements document, and discussions of planning and scoping and risk management as
they relate to risk assessment. In addition, case studies are presented which contain examples of
risk characterizations from risk assessments that apply the principles described in the risk
characterization guide.   --------

      The folio whig two pages give a concise description of maj or points to consider when
preparing the risk characterization component of a risk assessment. They can be used as a stand
alone desk tool or "pocket guide."

-------
        Page 2
                       DRAFT Risk Characterization Handbook
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
             TCCR Points to Consider About Risk Characterization

1. Transparency - This refers to transparency in the risk assessment process. Making the
       process open and frank helps make the default and policy choices known and helps
       achieve full disclosure in terms of:
       a)     the assessment approach employed
       b)     the use of assumptions and their impact on the assessment
       c)     the use of extrapolations and their impact on the assessment
       d)     the use of models vs. measurements and their impact on the assessment
       e)     plausible alternatives and the choices made among those alternatives
       f)      the impacts of one choice vs. another on the assessment
       g)     significant data gaps and their implications for the assessment
       h)     the scientific conclusions identified separately from default assumptions and.
              policy calls
       i)      the major risk conclusions and the assessor's confidence in them
       j)      the relative strength of each risk assessment component and its impact on the
              overall assessment (e.g., is the case for the agent posing a hazard is strong, but the
              overall assessment of risk is weak because the case for exposure is: weak)

2. Clarity - Clarity refers to the risk assessment product. Making the product clear makes the
assessment complete and understandable. Clarity is achieved by:
       a)     brevity
       b)     avoiding jargon
       c)     using plain language so it's understandable to EPA risk managers and the
              informed lay person      -
       d)     avoiding the use of technical terms and, if used, defining technical terms simply
       e)     describing any quantitative estimations of risk simply
       f)      using simple tables and graphics (avoid equations) to present the technical data
3. Consistency - Consistency provides a context for the reader and refers to the presentation of
       the material hi the risk assessment (i.e. ensuring the assessment takes a similar approach
       to other EPA assessments or describing why the approach differs).  Consistency also
       provides a context by describing how the risk assessment findings agree or don't agree
       with other assessments done by EPA and others. Consistency is achieved by:
       a)      following statutory requirements and program precedents
       b)      following appropriate Agency-wide assessment guidelines

-------
                              DRAFT Risk Characterization Handbook
Page 3
 1            c)     using Agency-wide information, where appropriate, from systems, such as the
 2                   Integrated Risk Information System (IRIS),                             ,
 3            d)     putting the risk assessment in context with other similar risks         .
 4                   1)     how does it compare to other EPA assessments of similar agents or sites
 5                   2)     how likely it is that this assessment will be accepted by the scientific and
 6            '"••','••   regulatory community (e.g., other federal and state agencies, by other
 7                        ; countries and/or by various interest groups)
 8                         i)      how do the conclusions drawn by others differ from EPA's
 9                                assessment
10                         ii)     what .are the strengths and limitations compared to EPA's
11               ,                 assessment
12            e)     defming and explaining the purpose of the risk assessment (e.g. regulatory
13                   purpose, or policy analysis, or priority setting, etc.)
14            f)     defining the level of effort (e.g. quickscreen> extensive characterization) put into
15      .          ^  the assessment and the reason(s) why this level of effort was selected
16            g)     following established Agency peer review procedures
17   "'      ' '   .           ' .       .         -      •.'•'.•-.•    ••     .-         "-",'..
18     4. Reasonableness - Reasonableness refers to the findings of the risk assessment in the context
19            of the state-of-the science, the default assumptions and the policy choices made.Has a
20            picture of risk been provided that seems reasonable when considered in light of these
21            factors? Reasonableness is achieved when:
22            a)     the risk characterization is determined to be sound by the scientific community,
23                   EPA risk managers, and the lay public, because the components of the risk
24                   characterization are well integrated into an overall conclusion of risk which is
25                   complete, informative, well balanced and useful for decision-making
26            b)     the characterization is based on the best available scientific information
27            c)     the policy judgments required to carry out the risk analyses use common sense
28                   given the statutory requirements and Agency guidance
29            d)     the assessment uses generally accepted scientific knowledge   ,
30            e)     plausible alternative estimates of risk under various candidate risk management
31                   alternatives are identified arid explained
32
                  Transparency is the principle value from among the
                  Administrator's four TCCR values, because, when followed, it
                  leads to clarity, consistency and reasonableness

-------

-------
    Page 4
       DRAFT Risk Characterization Handbook
1

2

3

4
5
6
U.S. Environmental Protection Agency
RISK CHARACTERIZATION GUIDE

-------

-------
                              DRAFT Risk Characterization Handbook
                                                                               PageS
 1
 ,2
 3
 4
 5
 6
 7
 8
 9
 10
 11
 12
 13
 14
 15
 16
 17
 18
 19
20
21
22
23
24
25
26
27
28
29
30
31
              1. THE NEED FOR RISK CHARACTERIZATION
1.1    Overview Statement

       The Risk Characterization Handbook is created as a single, centralized form of risk
characterization implementation guidance for Agency risk assessors and managers. Effective
risk characterization is achieved through TCCR or transparency in the risk assessment process
and clarity, consistency, and reasonableness of the risk assessment product.

       Risk characterization at the U.S. Environmental Protection Agency (EPA) takes many
different forms depending on the nature of the risk assessment being characterized. The level of
information contained in each risk characterization varies according to the type of assessment for
which the characterization is written and the audience for which the characterization is intended.
The goal of risk characterization is to clearly communicate the strengths and limitations of the
risk assessment so its use in decision making can be put into context with the other information
critical to evaluating options for rules, regulations and negotiated agreements (e.g., economics,
social values, public perception, policies, etc.).
          A  balanced discussion of reliable conclusions and related uncertainties
                enhances, rather than detracts, from the overall credibility of each
                assessment.
       To address concerns about the public's perception of and confidence in EPA's risk
assessments, former Deputy Administrator Henry Habicht issued an Agency-wide policy for risk
characterization on February 26,1992.  He noted that "... scientific uncertainty is a fact of life
(and) ... a balanced discussion of reliable conclusions and related uncertainties enhances, rather
than detracts, from the overall credibility of each assessment..." Administrator Carol Browner
reaffirmed the central role of risk characterization for the Agency on March 21,1995 when she
issued the Agency-wide Risk Characterization Policy (Appendix A). The Policy called for all  .
risk assessments performed at EPA to include a risk characterization to insure that the risk
assessment process is transparent and that the risk assessments are clear, reasonable and
consistent with other risk assessments of similar scope prepared by programs across the Agency.

-------
       Page 6
                       DRAFT Risk Characterization Handbook
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
                     Scientific uncertainty is a fact of life
J
                                                                           To
implement the Policy, an Agency-wide Risk Characterization Guidance document and Program
Office- and Region-specific Risk Characterization Implementation Plans (RCIPs) were written.
Many meetings (colloquia for risk assessors and roundtables for risk managers) were held to
identify and resolve risk characterization issues. Based on the experiences of those attending the
meetings and using these various documents to help characterize risk across the Agency since
then, we found that a single Agency-wide document is needed. While the Handbook supersedes
the original guidance and RCIPs, Agency offices and regions may wish to prepare brief, tailored
guidance that meets their individual needs to supplement .the information in this Handbook.

1.2    Understanding Risk Characterization

       1.2.1  What is Risk Characterization?

       Risk characterization is the final step of the risk assessment process for both ecological
and health effects. For example, the National Research Council (NRC) of the National Academy
of Sciences (NAS) described the health risk assessment process in 4 key steps (NAS,  1983:
popularly known as the "Red Book").

       a)     Hazard identification — the determination of whether a particular chemical is or is
             not causally linked to particular health effects

       b)     Dose-Response Assessment — the determination of the relation between the
             magnitude of exposure and the probability of occurrence of the health effects in
             question

       c)     Exposure Assessment — the determination of the extent of human exposure before
             or after application of regulatory controls

       d)     Risk Characterization ~ the description of the nature and often the magnitude of
             human risk, including attendant uncertainty

       EPA's Guidelines for Ecological Risk Assessment describes the process for conducting a
risk assessment for ecological risks (US EPA 1996).

       a)     Problem formulation — the evaluation of goals, selection of assessment endpomts,

-------
                                DRAFT Risk Characterization Handbook
                                                                                 Page?
  1
  2
  3
  4
  5
  6
 7
  8
  9
 10
 11
 12
 13
 14
 15
 16
 17
 18
 19
 20
 21
 22
.23
 24*
 25
 26
 27
 28
 29
 30
 31
 32
 33
              preparation of the conceptual model, and development of an analysis plan

       b)     Analysis — the evaluation of exposure to stressors and identification of the
              relationship between stressor levels and ecological effects

       c)     Risk Gharacterization — the estimation of ecological risks, discussion of overall
              degree of confidence in the risk estimates, citation of evidence.supportmg risk
              estimates, and interpretation of the adversity of ecological  risks

       In order to write an overall risk characterization, each risk assessment section needs to
have its own characterization. For health risk characterizations, separate characterizations
accompany the hazard identification, dose-response assessment and exposure assessment
sections. For ecological risk separate characterizations accompany the analysis plan, the
stressor-response profile and the exposure profile. These component characterizations carry
forward the key findings, assumptions, strengths and limitations, etc. for each section and
provide a fundamental set of information that must be conveyed in the final overall risk
characterization.      .             /      ,

       The overall risk characterization lets the manager, and others, know why EPA assessed
the risk the way it did hi terms of the available data and its analysis, uncertainties, alternative
analyses, arid the choices made.  A good risk characterization will express results clearly;,
articulate major assumptions and uncertainties, identify reasonable alternative interpretations,
and separate scientific conclusions from policy judgments. The Risk Gharacterization Policy
calls for the explanation of the choices made to be highly visible.
                  f'                   "               • , "        •
       Finally, remember that risk characterization is not just about science.  It is also about
making clear that science doesn't tell us certain things and that policy choices must be made. It
is about explaining why the risk assessment result is the way it is given all the choices made
during the course of the risk assessment process. And when others have also assessed the agent
in a biologically plausible fashion, even if their assessment does not agree with EPA's
assessment, it is about making clear that EPA has assessed the agent this way  but that others ,
have assessed it differently.                       -...        .,
                  Risk characterization is not only about science — it is also about
                  making clear that science doesn't tell us certain things and that policy
                  choices must be made           .
34

-------
        PageS
                       DRAFT Risk Characterization Handbook
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
       1.2.2  How Do Characterizations Written As Part of Ecological Risk Assessments
             Differ from Those Written As Part of Human Health Risk Assessments?

       They don't. Although, there are different elements in an ecological risk assessment
compared to a health risk assessment, the principles for writing the risk characterization section
of an assessment apply equally to both.

       1.2.3  Are Risk Assessment and Risk Characterization the Same?

       No, they're not the same.  Risk assessment is a process comprised of several steps (see
section 1.2.1 and NAS, 1983). Risk characterization is the final step of the risk assessment
process. It communicates the key strengths and weaknesses of the assessment through a
conscious and deliberate effort to bring all the important considerations about risk into an
integrated picture by being transparent, clear, consistent and reasonable. Every risk assessment
has a product, but unless you really characterize the assessment in the final step, the risk
assessment is not complete. Remember that risk characterization is an integral component of
every risk assessment. Just giving the quantitative risk estimate ("the number") is not a risk
characterization.

       1.2.4  What does the Risk Characterization Policy Tell Risk Assessors?

       The policy tells risk assessors to include the following in the risk characterization:

       a)     Carry forward the key information from hazard identification, dose-response, and
             exposure assessment, using a combination of qualitative information, quantitative
             information, and information about uncertainties

       b)     Discuss uncertainty and variability appropriate for the level of analysis

       c)     Present risk conclusions and information regarding the strengths and limitations
             of the assessment at the level appropriate for the risk assessment (e.g., if it is a
             screening assessment the risk characterization portion of the risk assessment
             should be brief)
35
                Risk characterization communicates the key strengths and weaknesses
                of the assessment through a conscious and deliberate effort to bring all
                the important considerations about risk into an integrated picture

-------
                               DRAFT Risk Characterization Handbook
                                                                                Page 9
  1
  2
  3   '.
  4
  5
  6
  7
  8
  9
10
11
12'
13   '
14
15
16
17
18
19
20
21'
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
       1.2.5  How will Risk Characterization Help the Risk Assessor?

       Risk Characterization makes the whole risk assessment story clearer and easier to
communicate. If you characterize risk, your risk assessment is easier to explain, justify, and
defend.  .
       1.2.6  How will Risk Characterization Help the Risk Manager?

       Risk characterization allows you to understand and better communicate risk assessment
findings. You can better convey information up the management decision-making chain and to
the public.  Transparency is a powerful tool. You can use it to ensure clarity, consistency and
reasonableness to achieve a better-informed decision. Risk managers have made the following
comments about risk characterization:                   '                      •
       a)
              Being transparent .helps me make better-informed decisions

              1)     .It brings out the usually unseen parts of the assessment

              2)     If I require transparency, I can incorporate clarity, consistency and
                    reasonableness to achieve a better-informed decision
       b)
      c)
             Communication helps and it has two parts

             1)     When"I ask questions, getting to TCCR is facilitated

             2) ,    I need to ask early and to check progress often

             Transparency

             1)     Helps me understand the scientific basis of my decisions

             2)     Helps me build trust and credibility with staff, public and stakeholders

       1.2.7  What Is the Difference Between Risk Characterization and Risk
             Communication?
'''..''        ':    '        '          '   .    •    '       '     ^
       Risk characterization is an integral part of a risk assessment that summarizes the  ,
strengths and limitations of the risk assessment for risk managers and others. While it provides

-------
        Page 10
                       DRAFT Risk Characterization Handbook
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
information that may be used to inform the public, it, is not the same thing as risk
communication. Risk communication is the exchange of information with the public, or some
sector of the public, about levels of health or environmental risks for:

       a)     Information and education

       b)     Behavioral change and protective action

       c)     Disaster warnings and emergency information

       d)     Joint problem solving and conflict resolution

       1.2.8  What is the Value of Risk Characterization in the Regulatory Development
             Process?

       The risk characterization section of risk assessments that support rulemaking actions is an
important, fundamental step informing the policy setting process. Risk characterization plus peer
review provide a mechanism to help the Agency achieve scientific credibility. TCCR is
essential, because new rules, and the work products supporting them, must often withstand
intense scrutiny by the general public and the stakeholders affected by EPA's decisions.
Although no rule or regulation itself is subject to the Risk Characterization Policy, the risk
assessments that help inform the final rules and regulations are subject to the policy, and they
should include risk characterization prior to use in any rule.

       1.2.9  What Role does Risk Characterization have in Regulatory Negotiations?

       Regulatory negotiations are not risk assessments; however, to ensure final decisions are
based on sound and credible science, any risk assessments used during the regulatory negotiation
need to be properly characterized before the negotiation takes place.

       1.2.10 Do Risk Characterizations Need Peer Review?

       The principle underlying the Peer Review Policy is that all major scientific and/or
technical work products used in Agency decision making will be peer reviewed. A risk
assessment can be a candidate for peer review provided it is considered a major scientific or
technical work product.  Use the criteria in the Peer Review Handbook to determine which
assessments need to be peer  reviewed (USEPA, 1998).

       The risk characterization is an intrinsic part of the risk assessment to be peer reviewed,

-------
                               DRAFT Risk Characterization Handbook
                                                                              Page 11
  1
  2
  3
  4
  5
  6
  7
  8
  9
 10
 11
 12
 13
 14
 15
 16
 17-
 18
 19
 20
 21
'22'
 23
 24
 25
 26
 27
 28
 29
 30
 31
 32
.33
 34
but it is not the risk characterization per 5e that is peer reviewed.  Rather the risk assessment,
with its risk characterization section, is peer reviewed. In reviewing the risk characterization
section of the risk assessment you need Jo make sure that the TCCR criteria (see section 2.2) are
addressed.
                Peer review is an important element to ensure the scientific
                soundness of a risk characterization                         .
                                                                                I
1.3    What the Agency Has Done Since the Policy Was Issued

       A series of colloquia for risk assessors and roundtables for risk managers was held to
test the Implementation Plans- against case studies, and to work out just what it takes to
characterize risk adequately. More than 200 EPA employees participated in these events. > They
shared with other EPA staff their office's culture and their own experiences and perspectives
about risk characterization.  They found that other office's cultures., experiences, and perspectives
were often different from their own. These meetings resulted in the major conclusion to bring
these experiences, perspectives, and lessons learned from testing the Implementation Plans into a
single, comprehensive document for Agency-wide use, i.e., the Risk Characterization Handbook.

       1.3.1  What Were the Major Messages from the Colloquia and Roundtables?

      Major messages from the colloquia and roundtables were:

      a)     Risk Characterization is an integral part of risk assessment

             1)     It is the philosophy of TCCR
    .              i)     Transparency in the risk assessment process
                   ii)     Clarity of the presentation of the risk assessment results
                   iii)    Consistency in the approaches used to assess risk as well as the
                          presentation of the risk assessment         %
                   iv)    Reasonableness in the assumptions and approaches used for risk
                          assessment
                  '   .'         '                   •      ~                     ' i'
             2)     It is a part of the risk assessment process that has not yet been adequately
                   done  at EPA                             ,    ,

-------
        Page 12
                       DRAFT Risk Characterization Handbook
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
       b)
3)     EPA risk assessment products must include adequate risk characterization

A cultural change is needed at EPA if the Agency is to implement successfully the
Risk Characterization Policy
       c)     A risk management framework for decision making is needed to effectively
             integrate science with the other factors that go into a decision (see Appendix D)

1.4    Need for Risk Characterization Criteria

       Because risk characterization, as called for in the March 1995 policy, clarifies EPA's way
of doing business

       a)     risk assessors need some criteria to know what is asked of them as they prepare
             risk characterizations

       b)     risk managers need some criteria to know what to look for as they read risk
             characterizations

       c)     the public needs some criteria to help them judge EPA's success in characterizing
             risk

        Section 2 describes risk characterization in the risk assessment process and contains the
criteria for TCCR and a discussion of how to achieve TCCR. Before launching each risk
assessment the criteria listed in section 2.2 should be kept in mind to help ensure that the risk
assessment is well-characterized. After the assessment is complete, the criteria can be used to
measure how well the assessment was characterized.

1.5    The Roles of People and Organizations in Risk Characterization

       1.5.1  Who is Ultimately Accountable for Risk Characterization?

       Under the March 21,1995, Risk Characterization Policy, the Administrator has
designated the Assistant Administrators, Associate Administrators, Regional Administrators
(AAs and RAs), the General Counsel, and the Inspector General to be accountable for
implementing the Policy in their respective organizations. In her memo to Senior Agency
Management (Appendix A), the Administrator noted that:

-------
                                DRAFT Risk Characterization Handbook
                                                                                 Page 13
  1
  2
 ,3
  4
  5
  6
  7
  8
  9
10
11
12
13
14
15
16
17
18
19
20
21.
22'
23
24
25
26
27,
28
29
30
31
32
33
34
35
36
37
38
39
       "These core values of transparency, clarity, consistency, and reasonableness need to guide each
       of us in our day-to-day work; from the toxicologist reviewing the individual cancer study, to the
       exposure and risk assessors, to the risk manager, and through to the ultimate decision maken I
       recognize that issuing this memo will not by itself result in any change. You need to believe in
       the importance of this change and convey your beliefs to your managers and staff through your
       words and actions in order for the change to occur. You also need to play an integral role hi
       developing the implementing policies and procedures for your programs."    „

       While the above-persons are ultimately accountable for ensuring health and ecological
risk assessments from their organizations have proper risk characterizations, it is recognized
much of the responsibility to ensure that the risk assessments include risk characterizations that
are done well according to the principles of TCCR is delegated to their Risk Managers (see
Section 1.5.4 for full responsibilities of Risk Managers).          •

       1.5.2  Who Are the Agency Staff Involved in Risk Characterization?

       The principal Agency staff are risk assessors and risk managers. Risk assessors are the
scientific and technical staff who actually perform the various components of the risk
assessment.

       1.5.3  Am I a Risk Assessor and What Are My Responsibilities?

       People who perform the risk assessment, in whole or in part, are the risk assessors. Your
major responsibility as a risk assessors is to communicate your major risk conclusions and your
confidence in them in the risk characterization section of your assessment.

       Your specific responsibilities are to:  •

       a)      Explain what is the risk, what individuals, populations or systems are affected and
              by what route of exposure                                     .   ,

       b)      Describe your level of comfort with the conclusions and what degree of certainty
              you place in them   . .                         ,             r            /

              1)     Summarize and identifythe key pieces of information critical to your
                    evaluation

              2)     Let your manager know whether the key data used for me assessment are
                    considered experimental, state-of-the art or generally accepted scientific

-------
Page 14
DRAFT Risk Characterization Handbook
1
2
3
' 4
5
6
7
S
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39


c)


d)



e)



f)


g)










h)



i)


j)'


k)

knowledge

Describe quantitative risk estimates in plain English; the use of tables and
graphics may be helpful as a supplement

Refer the reader to an Agency risk assessment guideline or other easily obtainable
reference that explains terminology (e.g., how a Qj* was developed)


Put this risk assessment into a context with other similar risks that are available to
• you and describe how the risk estimated for this stressor, agent or site compares to
others regulated by EPA

Describe how the strengths and weaknesses of EPA's assessment compare with
other assessments prepared by EPA hi the past

Describe the rationale and bases for the conclusions drawn by those outside EPA
about this agent, stressor or site

1) If their conclusions differ from yours, let the manager know whether theirs
is a reasonable alternative

2) Can their conclusions reasonably be derived from the data set

3) Inform the manager of the pros and cons of their evaluations compared to
yours

If you have developed specific assessments for one or more risk management
alternatives, let the risk manager know what changes in risk would occur under
these various candidate risk management alternatives

Highlight areas hi the assessment which might be overlooked or misinterpreted by
the risk manager

Keep the decision maker informed of the status of your risk assessment and risk
characterization

Organize, conduct, and complete the risk characterization following Agency
procedures

-------
                                DRAFT Risk Characterization Handbook
                                                                                 Page 15
  1
  2
  3
  4
;  5
  6
  7
  8
  9
 10
 11
 12
 13
 14
 15
 16
 17
 18
 19
 20
 21
 22
 23
 24
 25
 26
 27
 28
,29
 30
 31
 32
 33
 34
 35
 36
 37
 38
 39
        1)     Archive the risk characterization record in a manner consistent with your
               organization's archiving procedures

        1.5.4  Am I a Risk Manager and What Are My Responsibilities?

        Risk Managers are generally the decision makers in their organization. TheAA/RA,is
 the ultimate decision maker for his/her organization and is accountable for both the risk
 characterization process and products in his/her office. The AA/RA may designate Office
 Directors, Division Directors, and/or Branch Chiefs (or other appropriate level line^managers) as
, the front-line decision makers.  Generally, the decision makers commit the resources needed to
 ensure a proper risk assessment which includes a complete risk characterization. As a risk
 manager and/or decision maker, you are responsible for ensuring that risk assessments,
 containing risk characterizations, are properly performed and documented. You are also
, responsible for ensuring that the key information from each risk characterization is honestly and
 clearly elevated,up the managment chain and communicated to senior management.

        Your specific responsibilities are:

        a)     Promote a culture supportive of preparing risk characterizations and ensure that
               all risk assessment work products produced by or submitted to your organization
               are well characterized

        b)     Provide advice,  guidance, and support for the preparation, conduct, and
          1'.  .  completion of a full risk characterization for each assessment

        c)     Ensure that sufficient funds are designated in the office's budget request to
               conduct a risk characterization for each risk assessment

        d)     Establish a realistic risk assessment schedule that includes risk characterization

        e)   ,  Designate the stage(s) of product development where risk characterization is
               appropriate

        f)     Ensure  that the characterizations prepared by individual risk assessors for their
               portion of each risk assessment document are integrated into a complete risk    ;
               characterization for each risk assessment

        g)     Provide proper risk assessment training for your staff including how to write risk
               characterizations

-------
        Page 16
                       DRAFT Risk Characterization Handbook
  1
  2
  3
  4
  5
  6
  7
  8
  9
 10
 11
'12
 13
 14
 15
 16
 17
 18
 19
 20
 21
 22
 23
 24
 25
 26
 27
 28
 29
 30
 31
 32
 33
 34
 35
 36
 37
 38
 39
       h)     Establish systems to maintain records of the risk assessments, including risk
             characterizations, prepared by risk assessors under your supervision

       i)     Ensure that the key points from the risk characterization are carried forward in all
             deliberations or considerations for decision making
       1.5.5  What is the Role of the Science Policy Council (SPC)?

       The Science Policy Council (SPC) will consult with each Program Office and each
Region as they implement risk characterization. The SPC will also periodically evaluate the
Agency's experience with risk characterization and as necessary will revise this Handbook. The
implementation of the Risk Characterization Policy is the responsibility of management within
each Office or Region.

       1.5.6  Which Office/Region or Other Agency is Responsible for Writing the Risk
             Characterization?

       The organization preparing the risk assessment is normally responsible for preparing the
risk characterization. If more than one Agency office or region or other agencies is involved,
each is responsible for characterizing that component of the assessment they prepared.  The
responsibility for preparing the overall risk characterization is usually accepted by the office
making the decision, but this can be negotiated.

       1.5.7  What is the Responsibility of Organizations that Submit Risk Assessments to
             EPA?

       Just as the Agency is expected to follow its own guidance for characterizing risk in every
risk assessment, the Agency expects that any risk assessment done by any organization for EPA
consideration and possible use will include a proper risk characterization that is transparent,
clear, consistent and reasonable.  The Agency reserves the right to determine the acceptability of
the submitted risk assessment and its characterization and will evaluate each submission in line
with the principles hi this Handbook.

       If the submitting party has questions about any aspect of the risk assessment, it may want
to contact the agency office or region that is associated with and ultimate  recipient of the
assessment. Care should be taken to make it clear that while  the Agency is glad to comment on

-------
                             DRAFT Risk Characterization Handbook
Page 17
1   -,  questions presented about the assessment and risk characterization, it will not provide any
2     approval or conlmitments prior to its evaluation of the actual final submission..

-------

-------
       Page 18
                       DRAFT Risk Characterization Handbook
 1
 2
 3
 4 •
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
 2.  RISK CHARACTERIZATION IN THE RISK ASSESSMENT PROCESS

2.1     Overview Statement
                                             v      ,.  . :
       The risk characterization is one of the four steps in risk assessment. It is the final
component of the risk assessment process. The communication of the risk characterization will
take different written.and oral forms to meet the needs of the intended audiences (e.g., risk
managers, the public and other risk assessors). Thus, the risk assessment process produces
different risk characterization products for different audiences at different times.  This chapter
deals with risk characterization in .the risk assessment process; the next chapter deals with risk
characterization products and their audiences.

2.2     Criteria for Achieving TCCR

       Risk Characterizationls judged by the extent to which a risk assessment achieves
Transparency, Clarity, Consistency, and Reasonableness (TCCR). This section describes the
criteria for these four principles.
               Criteria are needed to folly implement the TCCR Values and to
               evaluate success
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
      2.2.1 What are the Criteria for Transparency?

      a)     The assessment approach is described

             1)    The assumptions are described and their impact on the assessment is
                   discussed

             2)    The extrapolations made are described and their impact on the assessment
                   is discussed

             3)    Whether the result is based on a model outputs or actual measurements is
                   described and the impact on the assessment is discussed

      b)     Plausible alternative assumptions, extrapolations, and models that would make a
             significant difference in the assessment, but not used are described and a
             discussion of the choices made is provided, including the impacts of a particular

-------
DRAFT Risk Characterization Handbook
Page 19
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39


c)


d)




e)


f)




2.2.2

a)

b)

c)



d)



e)



f)

2.2.3
choice vs. its plausible alternatives on the assessment

Significant data gaps are identified and their implications for the assessment
discussed

The scientific conclusions are identified separately from default assumptions and
policy calls; policy calls are clearly identified and described (e.g., to use a
linearized rather than a nonlinearized cancer risk model, to err on the side of
safety, to protect children, etc.)

The assessor's confidence in the major risk conclusions are clearly stated and
discussed

The relative strength of each risk assessment component and its impact on the
overall assessment is described (e.g., the case for the agent posing a hazard is
strong, but the overall assessment of risk is weak because the case for exposure is
weak)

What are the Criteria for Clarity?

The major risk conclusions are clearly stated

Brevity is achieved and jargon is avoided

Plain English is used and the risk characterization is understandable to the
scientific community, to EPA risk managers and other decision makers, and to the
informed lay person

The strengths and limitations of the assessment can be understood without
needing to understand the technical details of the assessment; technical terms are
avoided but if used they are defined simply

Tables and graphics (not equations) that are understandable to the scientific
community, EPA risk managers, and the informed layperson are used to present
technical data

Quantitative estimations of risk are described clearly

What are the Criteria for Consistency?

-------
Page 20
DRAFT Risk Characterization Handbook

I
2
3
4
5 . ,'• ' "'
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25 .
26
27
28
29
30
31
32
33
34
35
36
37
38
39


a)

b)

c)


d>
'













e)


f) .



"§)

2.2.4

a)




i ' • • • . • -. ',.-... . '•..'.•'

Statutory requirements and program precedents are followed
' .. ' ' . . - ' •
Appropriate Agency-wide assessment guidelines are followed ' :

Agencyrwide information is used, where appropriate, from systems such as the
Integrated Risk Information System (IRIS)

The risk assessment is put into context with other similar risks

1) How does it compare to other assessments of similar stressors, agents or ;
sites within the Office and across EPA

2) How likely it is that this assessment will be accepted by the scientific and
regulatory community (e.g., other federal and state agencies, by other
countries and/or by various interest groups)

i) How dp the conclusions drawn by others differ from EPA's
. assessment . .
, - - ' • ••.(''
ii) What are the strengths and limitations compared to EPA's
assessment >
.' '> ' . " '
The purpose of the risk assessment is defined and explained (e.g. regulatory
purpose, or policy analysis, or priority setting, etc.)

The level of effort for the assessment is described (e.g. quick screen, extensive
characterization) and the reason(s) why this level of effort was selected is also
described J
' • ' ' ••-••'" -
Peer review procedures are followed

What are the Criteria for Reasonableness?

The risk characterization is determuied to, be sound by the scientific community,
by EPA risk managers, and by the lay public, because the components of the risk
characterization are well integrated into an overall conclusion of risk which is
complete, informative, well balanced and useful for decision-making
• - ' •-..-• . . _ -

-------
                               DRAFT Risk Characterization Handbook
                                                                               Page 21
  1
  2
  3
  4
  5
  6
  7
  8
  9
 10
 11
 12
 13
 14
 15
 16
 17
 18
 19
 20
 21
 22
 23
 24
 25
 26
 27
 28
 29
 30
 31
 32
 33
 34
35
36
37
38
39
       b)     The characterization is based on the best available scientific information

       c)     The policy judgments required to carry out the risk analyses use common sense
              given the statutory requirements and Agency guidance

       d)     The assessment uses generally accepted scientific knowledge

       e)     Plausible alternative estimates of risk under various candidate risk management
              alternatives are identified and explained

23    Elements of a Risk Characterization That Meet the Criteria for TCCR

       TCCR are the guiding principles for writing a risk assessment and its risk
characterization. Elements are presented in the original Risk Characterization Policy package
and are discussed hi this section in the form of specific questions to answer to help achieve
TCCR. The elements provide a means to meet the criteria for TCCR when preparing the risk
characterization section of a risk assessment. The approach can apply to both health and
ecological risk assessments.  Specific questions to aid human health risk characterization are
found in the "Elements to Consider When Drafting EPA Risk Characterizations" in Appendix B.
This document is an revision of the original Risk Characterization Guidance package issued in
March 1995 which has been updated to include information about special populations (e.g.,
children and infants).

       The first portion of the elements document provides questions about the hazard
identification, dose-response and exposure assessment steps in the risk assessment process as an
aid to bring forward the major conclusions from these portions of the risk assessment. The
second part asks questions to help draw the major conclusions together in the risk
characterization portion of the risk assessment.

       Ecological risk assessments do not exactly follow the hazard identification, dose-response
assessment and exposure'assessment paradigm developed for human health risk assessment.
However, in characterizing ecological risks, the major conclusions of each step in the risk
assessment process are carried forward and integrated in the risk characterization portion of the
ecological risk assessment.
                                                                           »
       2.3.1  What Are the Major Conclusions to Carry Forward?
                                                         /
       Each component of the risk assessment (e.g., hazard identification, exposure assessment,
etc.) contains its own summary characterization. When integrated, these identify the

-------
       Page 22
                       DRAFT Risk Characterization Handbook
 "1.
 2
 3
 4
 5
 6
 7'
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
fundamental, irreducible set of key points that must be communicated to characterize adequately
any risk assessment. Because every risk assessment has many uncertainties, and involves many
assumptions, the challenge in characterizing risk for decision makers, whose time is limited and
who are not risk experts, is to convey that small subset of key strengths and limitations that
really makes a difference in the assessment outcome.

       a)     Bring out those key strengths and weaknesses in plain English consistent with
             TCCR

       b)     Provide a brief bottom line statement about the risks, including your confidence in
             any estimate(s) of risk and in your conclusions

       c)     Help the reader quickly grasp what is known about the nature, likelihood and
             magnitude of any risk                                              -
       The goal of Risk Characterization is not to repeat the entire
       assessment, just the key issues and conclusions

       The idea is to relay to the risk manager in frank and open terms the strengths and
limitations of the assessment. An example of possible strengths of an assessment would be that
the overall weight of evidence of the data indicates that the quality and quantity of data
supporting the hazard and/or exposure is high.  There might also be general consensus within the
scientific community on certain points used to build the hazard/exposure case.

       If you know of information mat would yield changes in me risk estimates under various
candidate risk management alternatives, let the manager know. For instance, if a feasibility
study has been performed that evaluates the risk associated with different treatment technologies
or remedial alternatives, discuss the range of possible outcomes and the implications of each.

       2.3.2  What Information Needs to Be Identified During the Risk Assessment
             Process to Prepare for Risk Characterization?

       The following table provides a tool that you may wish to use to help identify risk
characterization information from the risk assessment.  At each stage of the assessment for
human health or ecological risks, the assessor identifies:

       a)      The studies available and how robust they are (e.g., have the findings been

-------
                               DRAFT Risk Characterization Handbook
                                                                                Page 23
  1
  2
  3
  4
  5
  6
  7
  8
  9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
             repeated in an independent lab)

       b)    The assumptions made during that stage of the assessment, and the extrapolations
             made, the residual uncertainties
       c)     Use of defaults, policy choices and any risk management decisions made (e.g.,
             refer the reader to an Agency risk assessment guideline or other easily obtainable
             reference source that explains the meaning of terminology)

       d)     Whether the key data used for the assessment are considered experimental, state-
             of-the art or generally accepted scientific knowledge

       e)     The meaning of quantitative data in an easily understandable form — the use of
             tables and graphics may be helpful

       The contents of the following tables, the first for human health and the second for
ecological risk,  evolve as the risk assessment proceeds and culminates in a risk characterization.
Health Elements
Process
Hazard
Characterization
Dose-Response
Characterization
Exposure
Characterization
Risk
Characterization
Studies
Complete




Incomplete




Uncertainties



Assumption
s




Extrapolations




Variability




Policy Choices
Science




Decisio
n




Key
Data
Needs




Ecological
Elements
Process
Analysis Plan
Studies
Complete

Incomplete

Uncertainties

Assumption
s
•
Extrapolations

Variability

Policy Choices
Science

Decisio
n

Key
Data
.Needs


-------
Page 24
DRAFT Risk Characterization Handbook
1
2
3 .
4
5
6
7
8
9
10
11

Stressor-
Response Profile
Exposure Profile
Risk
Characterization


. ••

• * -. .
„•/-••
- • • • ' ' ' • , . "



2.3.3 What Are the Key Issues and Their Impact on the Conclusions?
„/
i -" -
By the time you have completed your assessment you should have Identified the universe
12 of policy choices, management decisions, and uncertainties as well as the conclusions of your
13 risk assessment.
(As noted earlier you may find the table in section 2.3.2 a helpful tool for this
14 purpose.) The point of risk characterization is not to repeat the entire risk assessment, but rather .
15 to describe the key issues and conclusions (i.e.. not all the issues and conclusions, only the key
16 ones) from each
17 issues differ for
step of the human health or ecological assessment paradigm. Because key
each assessment it is not possible to define exactly what the key issues are
18 generically. Professional judgment is necessary to define them. You will want to alert the risk
19 ' manager to the major issues that affect the conclusions: , - ,.
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34


' ' t
a) Identify those uncertainties, policy choices, and management decisions that if
changed would make a real impact on the risk assessment

-
b) Identify major imbalances among the components of the assessment (e.g., is the
case for the stressor or agent posing a hazard is strong, while the overall
assessment of risk is weak because there are no data about whether there is-
exposure to the stressor, or agent)


c) Highlight areas hi the assessment which might be overlooked or misinterpreted by
the risk manager


2.3.4 How Do I Put this Risk Assessment into a Context with Other Similar Risks?


Discussions about how the estimated risk from this agent or site compares to other risks is
35 important for risk managers to know. Two types of comparisons should be considered. The first
36 is to compare this risk assessment with previous Agency decisions to provide a feel for the
37 comfort level, weight of evidence, and likely problems the Agency will have with, this
38 assessment when comparing it to past Agency assessments. The second is to provide a sense of

-------
                               DRAFT Risk Characterization Handbook
                                                                               Page 25
  1
  2
  3
  4
  5
  6
  7
  8
  9
 10
 11
 12
 13
 14
 15
 16
 17
 18
 19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
 how generally the assessment is accepted by the scientific and regulatory community at large by
 comparing the results of EPA's assessment on this agent or site with available assessments made
 on the same agent or site by other federal and state agencies, by other countries and/or by various
 interest groups:

       a)     Comparisons to Agency assessments
                                                                 ,»
              1)     Let the risk manager know what other risk assessments have been
                    performed on this agent or site or similar agents and sites

              2)     Describe how the strengths and weaknesses of EPA's assessment compare
                    with other assessments prepared by EPA in the past

       b)     Comparisons to assessments done by others

              1)     Describe the rationale and bases for the conclusions drawn by others about
                    this agent if they differ from EPA's assessment

              2)     If their assessment differs from EPA's, is it a reasonable alternative (i.e.,
                    can their conclusions reasonably be derived from the data set)

              3)     What are the pros and cons of their evaluations compared to EPA's
                    assessment
        Discussions about how the likely risk from this stressor, agent or
        site compares to others regulated by EPA can provide a valuable
        tool to risk managers
2.4    Sensitive Populations, Ecosystems & Species

       In its risk assessments and risk characterizations, the EPA attempts to identify the
universe of people that may be affected, including sensitive populations (e.g., children, ethnic
groups, gender, age, nutritional status, other genetic predisposition) ecosystems or ecological
entities (e.g., endangered species) and those that are highly exposed (e.g., human, wildlife, etc.).
In the planning and scoping phase of the risk assessment process, the potential for exposures or
for unique adverse effects to sensitive populations should be noted. Any sensitive populations
that are identified should be evaluated in the risk assessment, and the assessment should be

-------
        Page 26
                       DRAFT Risk Characterization Handbook
  1
 2,
 3
 4
' 5
 6
 7
 8
 9
10
11
12
!3
14
15
16
17
18
19
20
21
appropriately characterized. Its not necessary to do a quantitative risk assessment on each one;
For instance, where there are many sensitive population groups for a given pollutant, it may be
sufficient to estimate risks for the most sensitive group and as long as they are protected other
groups may be protected adequately.
                                                                 ,           _        ^
       2.4.1  What Populations Must You Specifically Characterize in Your Risk
             Assessment?                     °                               ,

       While all sensitive populations need to be considered, Executive Order 13045 entitled
"Protection of Children from Environmental Health Risks and .Safety Risks" (April, 1997), and
the Administrator's "Policy on Evaluating Health Risks to Children" (October, 1995),
specifically require that all EPA risk assessments, risk characterizations, and environmental and
public health standards will always characterize health risks to infants and children.  Appendix B
includes questions that may be illustrative of the information that can be valuable in human
health risk characterization for children.

       In addition, the Agency has issued specific guidance for rule writers about how to address
children's risk pursuant to Executive Order 13045. This is found in the "EPA Rule Writer's
Guide to Executive Order 13045" issued as interim final guidance in April 1998 (USEPA, 1998).
All of the key points included in the interim guidance have been incorporated into the broader
risk characterization guidance included in the Appendix.

-------

-------
                               DRAFT Risk Characterization Handbook
                                                                              Page 27
  1
  2
  3
  4
  5
  6
  7
  8
  9
 10
 11
 12
 13
 14
 15
 16
 17
 18
 19
20
21
22
23
24
25
26
27
28
29
30
31
32-
33
34
35
36
                 3. RISK CHARACTERIZATION PRODUCTS

3.1    Overview Statement                                .

 '  ,    Remember, risk characterization is the final component of the risk assessment process,
and communication of the risk characterization will take different written and oral forms to meet
the needs of the intended audiences. Thus, the risk assessment process produces different risk
characterization products for different audiences at different times. The last chapter dealt with
risk characterization in the risk assessment process; this one deals with risk characterization
products and their audiences,

       The level of information contained in' each risk characterization product will vary
according to the detail of the risk assessment for which the characterization is written. In
addition, it will often vary hi format or detailin order to effectively communicate with the
intended audience. Use good judgment and common sense.  A one hundred page risk
characterization would not be appropriate as part of the report for a premanufacture notification
screening risk assessment prepared under Section 5 of the Toxic Substances Control Act.
However, a one hundred pag^risk characterization may be appropriate for a comprehensive risk
assessment prepared during a National Ambient Air Quality Standard review for any of the six
criteria air pollutants under the Clean Air Act.

3.2   , Forms of Risk Characterization

        Because there is more than one audience for each assessment, there will probably be
more than one risk characterization written or spoken about the risk assessment. There are also
different types of risk assessment that vary in length and degree of detail. Each risk
characterization is as simple or complex as the assessment from which it derives and the
audience for which it is prepared. The purpose of a risk characterization is full disclosure, but
that does hot mean that you have to be wordy.

       Further, each office and region produces different types of risk assessments, often
producing more than one type at any given tune. The differences are due to the requirements of
enabling legislation, the types of decisions to be made, the culture, of the office and to other
factors.  Examples of case studies start in Appendix E-
               Risk characterization is an abstracting process that takes different
               forms for different audiences and purposes

-------
Page 28
DRAFT Risk Characterization Handbook
1
2
3
4
5
6
7
8
9
10
11
12
13
14
*~
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
• '
«
Risk Characterization as an Abstracting






Process
Abstracting Process

O jc If ' ' • ' . : ' ..... . . . i . . . . _
Assessment 	 ^ Integrated . — -»- S u m m arv HesV 	 ^.Communication
(HI.DR.Exp) Analysis '•'-."••'• *>um m °ry ues; Pieces




^




r • 1




' ., .•''.''•'• • •'•+. •• . ' • 'T
Peer Reviewers Peer Reviewers Peer Reviewers Peer Reviewers
Risk Assessors R isk Assessors Risk Managers Public





3.2.1





Can Simple "Bright





"Lines" or Numbers be the Essential Information in
Every Risk Characterization Product?



No. Despite the form the risk characterization product takes, don't just give the
"number." The goal is to give an understandable, rich description of the strengths and
weaknesses of the assessment and to
avoid a "bright line" discussion. Every risk characterization
has a fundamental, irreducible set of information consisting of the key points that must be
conveyed to every audience to adequately characterize the risk, but it is more than just a number.



3.3 Audiences for Risk Characterization Products

3.3.1



Who Are the Audiences for Risk Characterization Products?


While not specifically defined, they run the gamut from risk assessors through line
managers to the decision makers, the
Administrator, the scientific community, and the general
public. The risk characterization product needs to be tailored to each of these specific audiences
in terms of depth and detail.







,
j




-------
                              DRAFT Risk Characterization Handbook
                                                                               Page 29
  1
  2
  3
  4
  5
  6
  7
  8
  9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
3.3
34
35
3'6
37
38
39
       3.3.2   Can You Use the Same Risk Characterization Product for All Audiences?
                                                       /             '
       While the full risk characterization is written for the type of assessment conducted, as you
present it to various audiences the characterization product needs to be tailored to that audience.
For fellow risk assessors, and other scientists, use the complete characterization. As you present
it to your non-technical colleagues and to those whose time is limited (e.g., managers), shorten
and focus the characterization product, but always include the fundamental, irreducible set of key
points that must be communicated to characterize adequately the essence of any risk assessment.
For these nontechnical audiences, and especially when you present the characterization to the
general public, write in plain English.

       3.3.3   How Do I Ensure that the Irreducible Set of Risk Characterization
              Information is Carried Forward in All Risk Characterization Products?

       Risk characterization is a synthesizing and abstracting process that yields a range of
products that might be written at different tunes by different people for different audiences
(Figure 2). To ensure that the key messages are carried forward, peer review is an important
component of the risk characterization portion of the risk assessment process, because it helps
ensure the scientific integrity of the risk characterization, especially as it is distilled and
simplified. At these points  in time, there is a need to ensure that the key points are faithfully
passed on and interpreted. Formal peer review may not be practical for small quick risk
assessments, or as the risk characterization products are turned into briefings. However, each
office needs to have procedures in place to ensure that as this is done, the major points of the
characterization are faithfully captured.

[NOTE: INSERT RISK CHARACTERIZATION AS AN ABSTRACTING PROCESS FIGURE
2 HERE]                    ."'..'

3.4    Risk Characterization Length

       3:4.1  What is the  Appropriate Length for a Risk Characterization?

       Common sense should be used; each risk characterization should reflect the length, depth
and breadth of the corresponding risk assessment and the audience for which it is intended. The
length of a risk characterization for a screening assessment will not likely change much when
adapted for different audiences because of the limited information usually available. The length
of a risk characterization for an intermediate or comprehensive risk assessment will be longer
when presented to fellow assessors and shorter for managers, but don't forget to always, include
that irreducible set of key points that really makes a difference in the assessment outcome.

-------
       Page 30
                       DRAFT Risk Characterization Handbook
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
3.5    Research Needs

       3.5.1  Should Identification of Research Needs Be Part of Risk Characterization
             Products?

       Yes. While many data needs and methodology gaps are identified when assessing risk,
only the key ones that really make a difference in the risk assessment outcome are highlighted
in the risk characterization portion of the risk assessment. A systematic capturing of such needs
identified during risk characterization may provide an effective way to identify high priority
scientific support needs and a mechanism to reduce the tension within EPA between the need for
immediate technical support for today's regulations and the need to improve test methods and
risk assessment models to more realistically estimate risk from environmental exposures.

       3.5.2  Should Decisions Be Delayed Until Research is Completed?

       No. Research is never certain and it often raises additional questions. The main benefit
of risk characterization is that it provides context for available information for use in decision
making and for strengthening the scientific underpinnings of the Agency's decisions.

-------
                               DRAFT Risk Characterization Handbook
                                                                              Page 31
  1
  2
  3
  4
  5
  6
  7
  8
  9
10
11
12
13
14
15
16
1.7
18
19,
20
21
22
23
24
25
26
27
28
29
30
31
32.
33
34
35
36
37
38
39
                         4.  ADMINISTRATIVE ISSUES

4.1    Overview Statement
                                 f-                    .                 '        ' .
       Risk characterization, as a component of risk assessment, is done by many people over
time. It is often iterative in nature. Thus, risk characterizations should be memorialized in
writing, by the regions and offices as part of each risk assessment. If the risk assessment is done
piecemeal, each risk assessment section should be accompanied by a written risk characterization
for that section of the assessment to be stitched together with the risk characterizations for each
of the other sections as they are completed later to prepare a risk characterization of the overall
risk assessment.  Similarly, when sections of the risk assessment are updated, the risk
characterization for that section should be updated too, in writing.

       Decision makers are responsible for ensuring that a risk characterization is written for
each risk assessment and that a risk assessment/risk characterization record is maintained.
        Risk characterizations must be placed in writing
4.2    What is the Risk Characterization Record?

       In essence, it is the written risk characterization. The record should include the planning
and scoping materials (see Appendix C), a record of the risk assessors/risk managers decisions,
all parts of the risk assessment, including their individual characterizations and the final risk
characterization, with any updates.  It needs to be maintained in accordance with the
organization's archiving procedures.

       4.2.1  How Can the Risk Characterization Record Improve the Risk
             Characterization Process?

       A good risk characterization record allows future reference to the key conclusions and
strengths and weaknesses of the assessment. It can be studied by the risk manager to help better
inform nun/her about the facts at hand. In addition, a good record helps ensure that EPA's Risk
Characterization Policy is followed.

-------
        Page 32
                       DRAFT Risk Characterization Handbook
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
       4.2.2   Where Should the Risk Characterization Record be Kept and For How
              Long?

       During the active conduct of the risk characterization, it is likely that each risk assessor
maintains the risk characterization record until his/her portion of the risk characterization is
completed. Establishment and maintenance of an archive where the risk characterization records
ultimately reside are an organization's responsibility. The risk characterization record is part of
the risk assessment record.

4.3    Budget Planning

       As soon as it is known in the planning and scoping process that a risk assessment will be
done, the resources needed to conduct the risk assessment and its characterization need to be
designated. It is the risk manager's/decision maker's responsibility to ensure that the necessary
resources are requested as part of the usual Agency budgetary processes.  Risk characterization
needs to be considered as a normal part of doing business, just as peer review should be. Risk
assessment/risk characterization resource considerations should also be addressed in the analytic
blueprint for Agency rule-making actions.

4.4    Legal Considerations

       4.4.1 Are There Legal Ramifications from the Risk Characterization Policy?

       The Risk Characterization Policy does not establish or affect legal rights or obligations.
Rather, it confirms the importance of risk characterization where appropriate, outlines relevant
principles, and identifies factors Agency staff should consider hi implementing the policy.
Except where provided otherwise by law, risk characterization is not a formal part of or
substitute for notice and comment on rulemaking or adjudicative procedures. EPA's decision to
characterize risk as part of the risk assessment in any particular case is wholly within the
Agency's discretion.        •

       4-4.2   Is Legal Advice Needed?

       With respect to the Risk Characterization product, it is highly unlikely  that legal advice
will ever be needed.  However, as part of the Risk Characterization process, legal counsel, as
appropriate, should be included in the "team" supporting the decision maker and work with risk
assessors, economists, and others, from planning and scoping through to the final decision.

-------
                                DRAFT Risk Characterization Handbook
                                                                                 Page 33
  1
  2
'  3.
  4
  5
  6
  7
  8
  9
 1.0
 11
 12
 13
 14
 15
 16
 17
 18
 19
20
21
22
23
24
25'
26
27
28
29
30
31
32
33
34
35
36
37
38
39
               References Concerning Risk Characterization

Browner, C. (1995) Risk Characterization Memorandum issued March 21, 1995 (Note: Attached as
       Appendix A to this Handbook)

Commission on Risk Assessment and Risk Management (1997) Framework/or Environmental Health
       Risk Management, Final Report Volume 1; Washington, DC.      .

Commission on Risk Assessment and Risk Management (1997) Risk Assessment and Risk Management
       in Regulatory Decision-Making, Final Report Volume 2, Washington, DC.

Habicht, F.H. (1992) Guidance on Risk Characterization for Risk Managers and Risk Assessment
      , Memorandum, Washington, DC

National Research Council (1983) Risk Assessment in the Federal Government: Managing the Process,
       Washington, DC: National Academy Press, March 1983.       .

National Research Council (1994) Science and Judgment in Risk Assessment, Washington, DC: National
       Academy Press.      ,         .',••-•

National Research Council (1996) Understanding Risk: Informing Decisions in a Democratic Society,
       eds. Paul C.  Stern and Harvey V. Fmeberg, Washington, DC: National Academy Press.    -'
       ,                       '        -              ^       '__•;•           .           -   |-
U.S. Environmental Protection Agency (1984) Risk Assessment and Management: Framework for
       Decision Making, EPA 600/9-85-002, Washington, DC: U.S. Environmental Protection Agency,
       December 1984.

U.S. Environmental Protection Agency (1997), Guidance "on Cumulative Risk Assessment. Part L
       Planning and Scoping, Science Policy Council, Washington, DC, July 1997.

U.S. Environmental Protection Agency (1998) Science Policy Council Handbook: Peer Review, EPA
       100-B98-001, Washington, DC: U.S. Environmental ProtectioaAgency, January 1998.

U.S. Environmental Protection Agency (1998) EPA's Rule Writer's Guide to Executive Order 13045:
       Guidance for ConsideringRisksto Children During the Establishment of'Public Health-Related .
       Standard, Interim Final Guidance, Washington, DC.

U.S. EPA. 1998. Guidelines for Ecological Risk Assessment. Federal Register 63(93): 26846-26924. May
       14,1998.                                      "-                         _  ,.

-------

-------
                      DRAFT Risk Characterization Handbook
                                           Page A-l
 1
 2
 3
 4
'5
 6
 7
 8
 9
10
           APPENDIX A
U.S. Environmental Protection Agency
             /
   Policy for Risk Characterization
            March 1995

-------
        Page A-2
                       DRAFT Risk Characterization Handbook
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
                       Policy for Risk Characterization
INTRODUCTION
Many EPA policy decisions are based in part on the results of risk assessment, an analysis of
scientific information on existing and projected risks to human health and the environment.  As
practiced at EPA, risk assessment makes use of many different kinds of scientific concepts and
data (e.g., exposure, toxicity,  epidemiology, ecology), all of which are used to "characterize" the
expected risk associated with a particular agent or action in a particular environmental context.
Informed use of reliable scientific information from many different sources is a central feature of
the risk assessment process.

Reliable information may or may not be available for many aspects of a risk assessment.
Scientific uncertainty is a fact of life for the risk assessment process, and agency managers
almost always must make decisions using assessments that are not as definitive in all important
areas as would be desirable. They therefore need to understand the strengths and the limitations
of each assessment, and to communicate this information to all participants and the public.

This policy reaffirms the principles and guidance found in the Agency's 1992 policy (Guidance
on Risk Characterization for Risk Managers and Risk Assessors, February 26,1992).  That   ,
guidance was based on EPA's risk assessment guidelines, which are products of peer review and
public comment. The 1994 National Research Council (NRC) report, "Science and Judgment in
Risk Assessment," addressed the Agency's approach to risk assessment, including the 1992 risk
characterization policy.  The NRC statement accompanying the report stated, "... EPA's overall
approach to assessing risks is fundamentally sound despite often-heard criticisms, but the
Agency must more clearly establish the scientific and policy basis for risk estimates and better
describe the uncertainties hi its estimates of risk."

This policy statement and associated guidance for risk characterization is designed to ensure that
critical information from each stage of a risk assessment is used in forming conclusions about
risk and that this information  is communicated from risk assessors to risk managers (policy
makers), from middle to upper management, and from the Agency to the public. Additionally,
the policy will provide a basis for greater clarity, transparency, reasonableness, and consistency
in risk assessments across Agency programs. While most of the discussion and examples in this
policy are drawn from health  risk assessment, these values also apply to ecological risk
assessment. A parallel effort  by the Risk Assessment Forum to develop EPA ecological risk
assessment guidelines will include guidance specific to ecological risk characterization.

-------
                               DRAFT Risk Characterization Handbook
                                                                              Page A-3
  1
  2
  3
  4
  5
  6
  7
  8
  9
 10
 11
 12
 13
 14
-15
 16
 17
 18
 19
 20
 21
 22
 23
 24
 25
 26
 27
 28
 29
 30
 31
 32
-33
 34
 35
 36
 37
 38
 39
Policy Statement

       Each risk assessment prepared in support of decision-making at EPA should include a  •
risk characterization that follows the principles and reflects the values outlined in this policy. A
risk characterization should be prepared in a manner that is clear, transparent, reasonable and
consistent with other risk characterizations of similar scope prepared across programs in the
Agency.  Further, discussion of risk in all EPA reports, presentations, decision packages, and
other documents.should be substantively consistent with the risk characterization.  The nature of
the risk characterization will depend upon the information available, the regulatory application of
the risk information, and the resources (including time) available.  In all cases, however, the
assessment should identify and discuss all the major issues associated with determining the
nature: and extent of the risk and provide commentary on any constraints limiting fuller
exposition.
Key Aspects of Risk Characterization                      ,                 .

       Bridging risk assessment and risk management. As the interface between risk
assessment and risk management, risk characterizations should be clearly presented, and separate
from any risk management considerations. Risk management options should be developed using
the risk characterization and should be based on consideration of all relevant factors, scientific
and nonscientific.             '                                              •

       Discussing confidence and uncertainties. Key scientific concepts, data and methods
(e.g., use of animal or human data for extrapolating from high to low doses, use of
pharmacokinetics data, exposure pathways, sampling methods, availability of chemical-specific
information, quality of data) should be discussed. To ensure transparency, risk characterizations
should include a statement of confidence in the assessment that identifies all major uncertainties
along with comment on their influence on the assessment, consistent with the Guidance on Risk
Characterization (attached).

       Presenting several types of risk information. Information should be presented on the
range of exposures derived from exposure scenarios and on the use of multiple risk  descriptors
(e.g., central tendency i high end of individual risk, population risk, important subgroups, if
known) consistent with terminology in the Guidance on Risk Characterization, Agency risk
assessment guidelines, and program-specific guidance. In decision-making, risk managers
should use risk information appropriate to their program legislation.

-------
        Page A-4
                       DRAFT Risk Characterization Handbook
 1
 2
 3
 4
,5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
       EPA conducts many types of risk assessments, including screening-level assessments of
new chemicals, in-depth assessments of pollutants such as dioxin and environmental tobacco
smoke, and site-specific assessments for hazardous waste sites.  An iterative approach to risk
assessment, beginning with screening techniques, may be used to determine if a more
comprehensive assessment is necessary. The degree to which confidence and uncertainty are
addressed hi a risk characterization depends largely on the scope of the assessment.  In general,
the scope of the risk characterization should reflect the information presented in the risk
assessment and program-specific guidance. When special circumstances (e.g., lack of data,
extremely complex situations, resource limitations, statutory deadlines) preclude a full
assessment, such circumstances should be explained and their impact on the risk assessment
discussed.
Risk Characterization in Context

       Risk assessment is based on a series of questions that the assessor asks about scientific
information that is relevant to human and/or environmental risk. Each question calls for analysis
and interpretation of the available studies, selection of the concepts and data that are most
scientifically reliable and most relevant to the problem at hand, and scientific conclusions
regarding the question presented. For example, health risk assessments involve the following
questions:

       Hazard Identification — What is known about the capacity of an environmental agent for
       causing cancer or other adverse health effects in humans, laboratory animals, or wildlife
       species? What are the related uncertainties and science policy choices?

       Dose-Response Assessment — What is known about the  biological mechanisms and
       dose-response relationships underlying any effects observed in the laboratory or
       epidemiology studies providing data for the assessment? What are the related
       uncertainties and science policy choices?

       Exposure Assessment — What is known about the principal paths, patterns, and
       magnitudes of human or wildlife exposure and numbers  of persons or wildlife species
       likely to be exposed? What are the related uncertainties  and science policy choices?

Corresponding principles and questions for ecological risk assessment are being discussed as part
of the effort to develop ecological risk guidelines.

       Risk characterization is the summarizing step of risk assessment. The risk

-------
                                DRAFT Risk Characterization Handbook
                                                                               Page A-5
  1
  2
'  3
  4
  5 *
  6
  7
  8
  9
.10
 11
 12
 13
 14
 15
 16
 17
 18
 19
20
21
22
23
24
25
26
27
28
29'
30
31
32
33
34
35
36
37
38
39
 characterization integrates information from the preceding components of,the risk assessment
 and synthesizes an overall conclusion about risk that is complete, informative and useful for
 decision makers.

       Risk characterizations should clearly highlight both the confidence and the uncertainty
 associated with the risk assessment. For example, numerical risk estimates should always be
 accompanied by descriptive information carefully selected to ensure an objective and balanced
 characterization of risk in risk assessment reports and regulatory documents. In essence, a risk
 characterization conveys the assessor's judgment as to the nature and existence of (or lack of)
 human health or ecological risks. Even though a risk characterization describes limitations in an
 assessment, a balanced discussion of reasonable conclusions and related uncertainties, enhances,
 rather than detracts, from the overall credibility of each assessment.      ,

       "Risk characterization" is not synonymous with "risk communication." This risk
 characterization policy addresses the interface between risk assessment and risk management.
 Risk communication, in contrast, emphasizes the process of exchanging information and opinion
 with the public — including individuals, groups, and other institutions.  The development of a risk
 assessment may involve risk communication.  For example, in the case of site-specific
 assessments for hazardous waste sites, discussions with the public may influence the exposure
 pathways included in the risk assessment.  While the final risk assessment document (including
 the risk characterization) is available to the public, the risk communication process may be better
.served by separate risk information documents designed for particular audiences.
Promoting Clarity, Comparability and Consistency

       There are several reasons that the Agency should strive for greater clarity, consistency
and comparability in risk assessments.  One reason is to niinimize confusion. For example,
many people have not understood that a risk estimate of one in a million for an "average"
individual is not comparable to another one in a million risk estimate for the-"most exposed
individual." Use of such apparently similar estimates without further, explanation leads to
misunderstandings about the relative significance of risks and the protectiveness of risk reduction
actions.               -  .

       EPA's Exposure Assessment Guidelines provide standard descriptors of exposure and
risk. Use of these terms in all Agency risk assessments will promote consistency and
comparability. Use of several descriptors, rather than a single descriptor, will enable EPA to
present a fuller picture of risk that corresponds to the range of different exposure conditions
encountered by various individuals and populations exposed to most environmental chemicals.

-------
        Page A-6
                        DRAFT Risk Characterization Handbook
  1
  2
  3
  4
  5
  6
  7
  8
  9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
Legal Effect

       This policy statement and associated guidance on risk characterization do not establish or
affect legal rights or obligations.  Rather, they confirm the importance of risk characterization as
a component of risk assessment, outline relevant principles, and identify factors Agency staff
should consider hi implementing the policy.

       The policy and associated guidance do not stand alone; nor do they establish a binding
norm that is finally determinative of the issues addressed. Except where otherwise provided by
law, the Agency's decision on conducting a risk assessment in any particular case is within the
Agency's discretion. Variations in the application of the policy and associated guidance,
therefore, are not a legitimate basis for delaying or complicating action on Agency decisions.
Applicability

       Except where otherwise provided by law and subject to the limitations on the policy's
legal effect discussed above, this policy applies to risk assessments prepared by EPA and to risk
assessments prepared by others that are used in support of EPA decisions.

       EPA will'consider the principles in this policy hi evaluating assessments submitted to
EPA to complement or challenge Agency assessments. Adherence to this Agency-wide policy
will improve understanding of Agency risk assessments, lead to more informed decisions, and
heighten the credibility of both assessments and decisions.
Implementation

       Assistant Administrators and Regional Administrators are responsible for implementation
of this policy within their organizational units.  The Science Policy Council (SPC) is organizing
Agency-wide implementation activities. Its responsibilities include promoting consistent
interpretation, assessing Agency-wide progress, working with external groups on risk
characterization issues and methods, and developing recommendations for revisions of the policy
and guidance, as necessary.

-------
                              DRAFT Risk Characterization Handbook
                                                                             Page A-7
 1
\2
 3
 4
 5
 6
 7
 3
 9
10
11
12
13
14
15
16
17
18
19
       Each Program and Regional office will develop office-specific policies and procedures
for risk characterization that are consistent with this policy and the associated guidance. Each
Program and Regional office will designate a risk manager or risk assessor as the office
representative to the Agency-wide Implementation Team, which will coordinate development of
office-specific policies and procedures and other implementation activities.  The SPC will also
designate a small cross-Agency Advisory Group that will serve as-the liaison between the SPC
and the Implementation Team.1                                          .

       In ensuring coordination and consistency among EPA offices, the Implementation Team
will take into account statutory and court deadlines, resource implications, and existing Agency
and program-specific guidance on risk assessment.  The group will work closely with staff
throughout Headquarters and Regional offices to promote development of risk characterizations
that present a full and complete picture of risk that meets the needs of the risk managers.
                           HI
APPROVED:  .
               Carol M. Browner, Administrator
DATE:    MAR 21 1995

-------

-------
                      DRAFT Risk Characterization Handbook
                                            Page B-l
1
2
3
4
5
6
7
8
9
           APPENDIX B
U.S. Environmental Protection Agency
  Elements for Risk Characterization

-------
Page B-2
DRAFT Risk Characterization Handbook
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39



ELEMENTS TO CONSIDER WHEN DRAFTING EPA RISK
CHARACTERIZATIONS

1. Background — Risk Characterization Principles

There
Consistency,

a)




b)





c)


d)






e)




are a number of points to consider about how to achieve Transparency, Clarity,
and Reasonableness (TCCR) for risk characterization:

Risk assessments should be transparent, in that the conclusions drawn from the
science are identified separately from policy judgements, and the use of default
values or methods and the use of assumptions in the risk assessment are clearly
articulated.

Risk characterizations should include a summary of the key issues and
conclusions of each of the other components of the risk assessment, as well as
describe the likelihood of harm. The summary should include a description of the
overall strengths and the limitations (including unbertainties) of the assessment
and conclusions. .

Risk characterizations should be consistent in general format, but recognize the
unique characteristics of each specific situation.

Risk characterizations should include, at least in a qualitative sense, a discussion
of how a specific risk and its context compares with other similar risks. This may
be accomplished by comparisons with other chemicals or situations in which the
Agency has decided to act, or with other situations which the public may be
familiar with. The discussion should highlight the limitations of such
comparisons.

Risk characterization is a key component of risk communication, which is an
interactive process involving exchange of information and expert opinion among
individuals, groups and institutions.

2. Conceptual Guide for Developing Chemical-Specific Risk Characterizations


The following outline is a guide and formatting aid for developing risk characterizations
for chemical risk assessments. Similar outlines will be developed for other types of risk
characterizations, including site-specific assessments and ecological risk assessments. A
common format will assist risk managers in evaluating and using risk characterization.

-------
                               DRAFT Risk Characterization Handbook
                                                                               Page B-3
 1
 2
 3
 4
 5.
 6
 7
 8
 9
10
11
12
13
.14'
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33 '
34
35
36
37
38
39
       The outline has two parts.  The first part tracks the risk assessment to bring forward its
 major conclusions. The second part draws all of the information together to characterize risk]
 The outline represents the expected findings for a typical complete chemical assessment for a
 single chemical. However, exceptions for the circumstances of individual assessments are
 allowable and should be explained as part of the risk characterization. For example, particular
 statutory requirements, court-ordered deadlines, resource limitations, and other specific factors
 may be described to explain why certain elements are incomplete.

       This outline does not establish or affect legal rights or obligations. Rather, it confirms the
 importance of risk characterization, outlines relevant principles, and identifies factors Agency
 Staff should consider in implementing the policy.  On a continuing'basis, Agency management is
 expected to evaluate the policy as well as the results of its application throughout the Agency and
 undertake revisions as necessary. Therefore, the policy does not stand alone; nor does it
 establish a binding norm that is finally determinative of the issues addressed. Minor variations in
 its application from one instance to another are appropriate and expected; they thus are not a
 legitimate basis for delaying or complicating action on otherwise satisfactory scientific,
 technical, and regulatory products.                            '......••,....
  \         t                       ,          '"',',',•
 3.     Special Consideration Should Be Given to the Evaluation of Risks to Infants and
       Children

       The following points are illustrative of the information that can be valuable in the
 assessment and characterization of children's risk.  As risk assessors conduct their risk
 assessment, they should consider these factors about children's risks in their risk
 characterization. The first two points to consider should be part of the fundamental, irreducible
.set of information carried forward hi the risk characterization:

       a)     Are children more susceptible to the toxic effects of chemicals than other
              populations (i.e. are they a sensitive subpopulation)?

       b)     Based on the exposure data available, are children disproportionately exposed to
              environmental chemicals?

 4.     PART ONE:  SUMMARIZING MAJOR CONCLUSIONS INRISK
       CHARACTERIZATION

       4.1    Characterization of Hazard Identification

       a)    . Does the chemical belong to a class of compounds where exposure during

-------
Page B-4
DRAFT Risk Characterization Handbook
, 1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
development leads to adverse effects that may be manifest at any time during life:

1)
4

2)

3)

4)


b) Are

effects on potential parents' ability to reproduce and raise their young (i.e.,
parental exposure led to the ability to conceive but the fetus died inutero)?

effects induced by in utero exposure?

effects induced by extra utero exposure during development?

does the chemical have a greater potential to produce an adverse effect
because it persists in the environment or body, or bioaccumulate?

there studies or reports of hazards to infants and children or to developing
animals?


c) Is there information bearing on

1)


2)

3)

4)

5)


6)

d) Are

1)

2)

3)


effects on the embryo and fetus (developmental toxicity) including
malformations, growth retardation and death?

effects on postnatal development (e.g., sexual and mental maturation)?

reproductive effects?

endocrine disruption (hereditary factors)?

target organ and. system effects (e.g., nervous and immunological 'Systems,
lung, skin)?

modulation of existing disease (e.g., exacerbate asthma)?

there effects unique to infants and children?

critical exposure periods?

exposure route/magnitude?

effects throughout life? v


-------
DRAFT Risk Characterization Handbook
Page B-5
1
2
3
4
5
6
7
8
9
10 .
11
12
13
14
15
16,
17
18
19,
20
21
22
23
24
25
26
27
28
29
30
31
32 .
33 •
34
35
36
37
38
39
e) What
is the key toxicological study (or studies) that provides the basis for health
concerns? ; • .

1)

2)


3)




4> .

• . ,
' 5)

6)


how good is the key study?
• - - _ f
are the data from laboratory or field studies? In single species or multiple
species?

if the hazard is carcinogenicity, comment on issues such as: observation
of single or multiple tumor sites; occurrence of benign or malignant
tumors; certain tumor types not linked to carcinogenicity; use of the
maximum tolerated dose (MTD). '

if the hazard is other than carcinogenicity, what endpoints were observed,
and what is the basis; for the critical effect?

Describe other studies that support this finding.

Discuss any valid studies which conflict with this finding.
•''.•• - -' ' •'. '
f) Besides the health effect observed in the key study, are there other health
endpoints of concern? For these health endpoints: -,

1)

2)

• ' ' . • . ' , '
What are the significant data gaps that make this a secondary study?

What are the implications of this endpoint to the assessment?

g) Discuss available epidemiological or clinical data. For epidemiological studies:

1) '

2)

3)


4)


What types of studies were used, i.e., ecologic, case-control, cohort?

Describe the degree to which exposures were adequately described.

Describe the degree to which confounding factors were adequately
accounted for.
- • • , -
Describe the degree to which other causal factors were excluded.

h) How much is known about how (i.e. through what biological mechanism) the

-------
        Page B-6
                        DRAFT Risk Characterization Handbook
  1
  2
  3
  4
  5
  6
  7
  8
  9
 10
 11
 12
 13
 14
 15
 16
 17
 18
 19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
       i)
chemical produces adverse effects?

1)     Discuss relevant studies of mechanisms of action or metabolism.

2)     Does this information aid in the interpretation of the toxicity data?

3)     What are the implications for potential health effects?

Comment on any non-positive data in animals or people, and whether these data
were considered in the hazard identification.
       j)      If adverse health affects have been observed in wildlife species, characterize such,
              effects by discussing the relevant issues as in a through i above.

       k)     Summarize the hazard identification and discuss the significance of each of the
              following:

              1)     confidence in conclusions;

              2)     alternative conclusions that are also supported by the data;

              3)     significant data gaps; and

              4)     highlights of major assumptions.

       4.2     Characterization of Dose-Response

       The current approaches to dose-response assessment include the cancer slope factor, RfD
and RfC, which constitute components  of the risk estimate for cancer and noncancer
assessments. For the dose part, either exposure or some internal dose might be the starting point.
The following issues/questions include considerations for children as compared to solely an adult
approach.

       a)     Dosimetry issues:

              1)     Are the chemical's pharmacokinetics different in children than in adults?

             2)    , If yes to 1, are any pharmacokinetic data considered to compare children
                    to adults?

-------
DRAFT Risk Characterization Handbook
PageB-7
1 ,
2
' 3
4
5
6
7
8
9
10
11
12 .
13
14
15
16
17
•18
19
20
21
22
23 .
24
25
26
27
28
-29 - ,
30
31
32 .
33 '
34
35
36
37
38
39
3)




4)


5)


6)

, 7)

If a toxic metabolite has been identified, and an internal dose is used in
the dose-response, have specific children's physiological parameters been
used in the derivation of the internal dosimetry, like inhalation rate, water
consumption, and food consumption?

Any adjustment for children's physiological parameters in the RfC
calculation? ,

Has there been any consideration of the developmental or reproductive
"critical windows" of exposure?

What kind of exposure and effects were observed?

Is there any information available on susceptible developmental stages?

b) Health effects response issues:

1)

2> .-



. 3)
f

'4)



c) What

1)




>


Is there any quantitative influence of age on the dose/response?

If the mode of action of agents is known, are there any specific childhood
diseases that might need to be considered? (site concordance is usually
not required for cancer)

Is there any information concerning the relative susceptibility of infants
and children versus adults for the identified hazards?

Are there data suggesting different No Observed Adverse Effect Levels
(NOAELs) or Lowest Observed Adverse Effect Levels (LOAELs) for
effects in children versus adults? In offspring versus parental?
. .-
data were used to develop the dose-response curve?

If animal data were used:

I) Which species was/were used? Most sensitive, average of all
species, or other? ••....
.
ii) Were any studies excluded? Why? Would the result have been •
significantly different if based on a different data set?

-------
        Page B-8
                 DRAFT Risk Characterization Handbook
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
       2)     If epidemiological data were used:

              i)     Which studies were used (only positive studies, all studies, or
                    some other combination)?
                                                                               /
              ii)     Were any studies excluded? Why?

              in)    Was a meta-analysis performed to combine the epidemiological
                    studies? If so, what approach was used?  Were studies excluded?
                    Why?

d)     What model was used to develop the dose-response curve? What rationale
       supports this choice? Is chemical-specific information available to support this
       approach?

       1)     For non-carcinogenic hazards:

              I)     For the RfD or RfC (or equivalent), especially if obtained from
                    reproductive or developmental neurotoxicology studies, has there
                    been any special consideration of post natal exposure and effects in
                    children?

              ii)     How was the RfD/RfC(or equivalent) (or the acceptable range)
                    calculated?

              iii)    What assumptions or uncertainty factors were used?

              iv)    What is the confidence in the estimates?

       2)     For carcinogenic hazards:

              I)     What dose-response model was used?  LMS or other linear-at-low-
                    dose model, a biologically-based model based on metabolism data,
                    or data about possible mechanisms of action?

              ii)     What is the basis for the selection of the particular dose-response
                    model used? Are there other models that could have been used
                    with equal plausibility and scientific validity? What is the basis
                    for selection of the model used in this instance?

-------
                                DRAFT Risk Characterization Handbook
                                                                        Page B-9
  1
  2
  3
  4
  5
  6
  7
  8
  9
 10
 11
 12
 13
 14
 15
-16
'\1
 18
 19
 20
 21
 22
 23
 24
 25
 26
 27
 28
 29
 30
 31
 32
 33
 34
 35
 36
 37
 38
 39
              iii)    If the dose-response for cancer is obtained from animal studies, has
       .             there been any consideration of exposure in the early life of the
                    animals, or consideration of window of exposure susceptibility
                    which could lead to early tumor formation in childhood?

              iv)    For dose-response relationships obtained from adult animal
                    studies, has there been any consideration given to extrapolating
                    cancer slope factors to children using a different coefficient value
                    for the power of the body weight? Has there been any
                    consideration given to determining if this coefficient for
                    extrapolating between young animals and infants and children is
                    the same as that for adults?

              v)     If the dose-response for cancer is obtained from occupational
                    epidemiologic studies, has there been any consideration for   -
                    children?

e)     Discuss the route and level of exposure observed, as compared to expected human
       exposures
                                                /             '.     -'   '     ., •
       1)      Are the available data from the same route of exposure as the expected
              human exposures? If not, are pharmacokinetic data available to
              extrapolate across route of exposure?

       2)      How far does one need to extrapolate from the observed data to
              environmental exposures (one to two orders of magnitude? multiple orders
              of magnitude)? What is the impact of such an extrapolation?

f)     If adverse health affects have been observed in wildlife species, characterize dose-
       response iriforination using the process outlined in a-c

4.3    Characterization of Exposure

a)     What are the most significant sources of environmental exposure?.

       1),      Are there data on sources of exposure, from different media?

       2)      What is the relative contribution of different sources of exposure?

-------
Page B-10
1 b)
2
3
4
5 c)
6
7 d)
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22 e)
23
24
25
26
27
28
29
30
31
32 f)
33
34
35
36
37
38 -
39
DRAFT Risk Characterization Handbook
What are the most significant environmental pathways for exposure?

1) What conceptual models were used to assess the exposures?

Were all routes of exposure assessed? If not, why were some excluded?

Describe the populations that were assessed, including the general population,
highly exposed groups, and highly susceptible groups

1) What population groups do you need to consider in the exposure
assessment for children (fetus, infants, child, women of childbearing age,
pregnant women, fathers)?

2) Are there differences across racial and ethnic subgroups?

3) What specific age groups need to be considered?

4) Are there gender differences?

5) Is there information on exposures to the groups of interest?

What are the anatomical and physiological characteristics of children that may
lead to unique exposures? For example, does the assessment account for a
child's: .

1) Inhalation rate?

2) Body weight?

3) Skin surface area?

What are the behavioral characteristics of children that may lead to unique
exposures? For example:
1) Playing on the floor and the ground,

2) Crawling,

3) Consumption rates for food and water, and/or preferential consumption of
certain food

-------
DRAFT Risk Characterization Handbook
Page B-ll
1
2
3
4,
•5 ' •
6
7
8
9
10
11.
12
13 g)
14
15
16
17 • -. ',
18
19
20
21
22
23
24
25
26
27
28
29
30
31 '••
32
33
34 -
35 h)
36
37
38
39
4)

5)

6)

?) '

8)

9)

There
above

1)


2)

3)

4)

"•
5)

6)

1,
7)'

8)

What

1)
,
'-
Consumption of breast milk (infants),

Spending a larger fraction of time indoors at home, '

Pica/soil ingestion,
- , . - . ^ . . i
Mouthing activities (e.g. , hand to mouth, obj ect to mouth, etc.)

Being carried by. adults

Personal hygiene ,

are several indirect factors that may associate higher exposures with the
behaviors. Examples include:

Contact with parents' clothing that has been contaminated at work (e.g.,
farm worker clothing contaminated with pesticides)

Environmental tobacco smoke in the home

Paint flaking from home, surfaces

Other characteristics of the child's environment (e.g., cleanliness of day
care facilities) *

Interactions with other children and adults that may lead to exposure

Carpets in the home serving as a reservoir for such pollutants as: lead,
mercury, bacteria, asbestos, fungi, insect parts, paint dusts, and solvents

Geographic location of the residence;
: • • -• '. , • „. '-.-...
Occupational exposures to legally and illegally employed children

chemical characteristics may indicate unique exposures? For example:

What is the concentration of the chemical in the child's environment (e.g.
home, lawn, garden)? , '
. •- • '-"••-• , • •• • --••""

-------
        Page B-12
                 DRAFT Risk Characterization Handbook
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
       2)     What are the fate and transport properties of the chemical (e.g., dissipation
              rates)?

       3)     Does the chemical cross the placenta?

       4)     Is the chemical excreted in breast milk?

       5)     Is the chemical used on or otherwise present in foods?

       6)     Is bioavailability of the chemical via the various routes of exposure higher
              in children than in adults?

i)     What routes (i.e., dermal, inhalation, and/or ingestion) of exposure need to be
       considered for children? For example, is dermal uptake of the chemical
       significant          compared to other routes of exposure?

       1)     For information on inhalation and ingestion rates, see the Exposure
              Factors Handbook

       2)     For mformation on dermal exposures, see the Exposure Factors Handbook
              and the Superfund Dermal Guidance

j)     Describe the basis for the exposure assessment, including any monitoring,
       modeling, or other analyses of exposure distributions such as Monte-Carlo or
       krieging

k)     What are the key descriptors of exposure?

       1)     Describe the (range of) exposures to: "average" individuals, "high end"
              individuals, general population, high exposure group(s), children,
              susceptible populations.

       2)     How was the central tendency estimate developed?

       3)     What factors and/or methods were used in developing this estimate?

       4)     How was the high-end estimate developed?

       5)     Is there information on highly-exposed subgroups?

-------
DRAFT Risk Characterization Handbook
Page B-13
1
2
3
4
5
6
7 ' '..
8
9
10 •
11
12 \
13
14
15
16
17
18
19
20
21
22
23 , . -
24
25 '
26
27
28
29
30 5.
31
32
33 •
34
35
36
37
38
39
, I) Who are they?

ii) What are their levels of exposure?
• " ' • ' i ' ' ' - . •• • • - •
iii) How are they accounted for in the assessment?

1) Is there reason to be concerned about cumulative or multiple exposures because of
ethnic, racial, or socioeconomic reasons?
. . • - ., -
m) Have children's exposures been aggregated across pathways (e.g., food, air, water,
surfaces, etc.) and routes (e.g., dermal, inhalation, and ingestion)?

n) Is the child exposure pattern continuous or episodic, chronic or acute, and are
there any critical time windows for exposure that need to be considered? Do
exposures vary with age?
•.-.''•'
o) If adverse health affects have been observed in wildlife species, characterize
wildlife exposure by discussing the relevant issues as in a through e above.

, p) Summarize exposure conclusions and discuss the following:

1) Results of different approaches, i.e. modeling, monitoring, probability
distributions .
• - \ • , . . • .
2) Limitations of each, and the range of most reasonable values

3) Confidence in the results obtained, arid the limitations to the results


PART TWO: RISK CONCLUSIONS AND COMPARISONS
" ' " . , :
5.1 Risk Conclusions

a) What is the overall picture of risk, based on the hazard identification, dose-
response and exposure characterizations?

1) Is there indication that children have sensitivities different than adults?
Qualitative differences? Quantitative differences?
.••-.•' - '

-------
Page B-14
DRAFT Risk Characterization Handbook
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20 -
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39






b).



c)

d)





5.2

a)



b)

c)










d)

2) What are the exposures to children? Do they differ significantly from
those of adults?

3) What are the expected risks to children? Relative to adults? Relative to
other subpopulations?

What are the major conclusions arid strengths of the assessment in each of the
three main analyses (i.e., hazard identification, dose-response, and exposure
. assessment)?
,
What are the major limitations and uncertainties in the three main analyses?

What were the science policy options used in each of the three major analyses?

1) What are Hie alternative approaches evaluated?

2) What are the reasons for the choices made?
•
Risk Context

What are the qualitative characteristics of the hazard (e.g., voluntary vs.
involuntary, technological vs. natural, etc.)? Comment on findings, if any, from
studies of risk perception that relate to this hazard or similar hazards.

What are the alternatives to this hazard? How do the risks compare?

How does this risk compare to other risks?

1) How does this risk compare to other risks in this regulatory program, or
other similar risks that the EPA has made decisions about?

2) Where appropriate, can this risk be compared with past Agency decisions,
decisions by other federal or state agencies, or common risks with which
people may be familiar?

3) Describe the limitations of making these comparisons.

Comment on significant community concerns which influence public perception
of risk, if known

-------
                              DRAFT Risk Characterization Handbook
                                                                            Page B-15
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
       5.3    Existing Risk Information

       Comment on other risk assessments that have been done on this chemical by EPA, other
federal agencies, or other organizations. Are there significantly different conclusions that merit
discussion?

       5.4    Other Information

       Is there other information that would be useful to the risk manager or the public in this
situation that has not been described above?

-------

-------
                              DRAFT Risk Characterization Handbook
                                                                            Paged
 1
 2
 3
• 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
  APPENDIX C; USING Risk CHARACTERIZATION PRINCIPLES TO
                  PLAN AND SCOPE A RISK ASSESSMENT

1.     Overview Statement                                           .
                                                                   *
  - s
       The risk characterization principles described in section 2 focus on the end result of risk
assessment. However, the participants in the colloquia and roundtables noted that the risk
characterization principles also offer a powerful tool to help plan and scope a risk assessment
before it is begun.  Therefore, these principles should be considered by risk assessors, risk
managers and others 'as they launch each new assessment.  Planning and scoping is an important
first step to insure that each risk assessment has a clear purpose, has a defined scope, and is well
thought out. These provide a sound foundation for judging'the success of the risk assessment
and for an effective risk characterization.
                  If you begin the overall risk assessment process with planning and
                  scoping you are positioned from the start to assure a good risk
                 . characterization
 2.    Planning and Scoping

      Based on EPA's experience with the four step NAS risk assessment paradigm (NAS,
1983) it has become clear that the additional step of planning and scoping is needed at the front
end of the risk assessment process to ensure that a risk assessment product is well done and is
well characterized. Planning and scoping can be viewed as a lens that defines the purpose and
scope of a risk assessment and focuses the issues involved [NOTE INSERT FIGURE C-l]. The
risk characterization portion of the risk assessment, in turn, is a second lens that focuses the
products of the risk assessment into a coherent picture.  At the end of the risk assessment, a
comparison of the risk assessment product, including the risk characterization, with the goals and
objectives defined during planning and scoping can provide a useful measure of success.

      Clarity in the planning and scoping stage also helps to avoid raising false expectations
about what the risk assessment will achieve. By stating up front just what the assessment will
cover and likely achieve, the application of risk characterization principles to planning and
scoping can make clear that decisions are not informed by science alone, but that economics,
values and other social concerns are important too (see  Appendix D). Policy choices must be
made in order to reach any decision.            '..'

-------
6
 CD
                                .2
                                .a
                                • ^
                             M  0>
                                IS
                                 g
                                U
I
CO
0>
CO
CO
                                                                      1/5

-------
Page C2
DRAFT Risk Characterization Handbook
1
2
3
4
' 5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39

2.2.1

During
risk managers

a)

b)

c)

d>

' e)
•
' . lf)

• «P.










h)

- •. .•'

What Should You Discuss During Planning and Scoping?

the planning and scoping phase of the risk assessment process risk assessors and
should engage in a dialog to identify:

Management goals and policies

Context of the risk

Scope and coverage of the effort , . *
• ' • -
Current knowledge .

What are the available data

Are there key data gaps for which research is needed?

An agreement about how to conduct the assessment including identifying

1) Resources available to do the assessment

2) Participants in the process ,

3) Plans for coordinating across offices, with other agencies and with
stakeholders

4) Schedule
''.'•"-.'
Plans for how the results will be communicated to senior managers and to the
public

Planning and scoping provides the opportunity for the risk manager, working with the
risk assessor(s) and other members of the "team", to define what is expected to be covered in the
risk assessment and to explain the purposes for which the risk assessment information will be
. used. Specific

a)

questions you probably should ask:
• . • -
What is motivating the need for the risk assessment? Is it public concern?
Scientific findings? Other factors? ,

-------
                               DRAFT Risk Characterization Handbook
                                                                               Page C3
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
       b)     What level of resources is available?
                                                             'V
       c)     What is the time frame?

       d)     Will other people conducting legal, economic, or social analyses need scientific
              data from the risk assessor to support their analyses? If so, what data are needed,
              and when?

       e)     What other management issues will be considered?

       Another outcome of the Planning and Scoping process is the identification of key data
gaps and thoughts about how to fill the information needs in the near-term using existing
information, the mid-term by conducting tests with currently available test methods to provide
data on the agent(s) of interest, and over the long-term to develop better, more realistic
understandings of exposure and effects and better more realistic test methods to evaluate agents
of concern. In keeping with TCCR care must be taken not to set science up for failure by
delaying environmental decisions until more research is done. Recognition must be given to the
fact that such decisions must always be based in part on policy and other non-science
considerations (e.g., economics — see Appendix D).  Thus, planning and scoping discussions
about filling information gaps should include:

       a)     When will the results be available

       b)     Will the results likely  make a real difference in the assessment

       c) .    To what extent will a policy call have to be made

       d)     Given the current data what is your ballpark assessment of the risks

       2.2.2   Would the Table in Section 2.3.2 Be Helpful During Planning and Scoping?

       Yes.  The table was specifically designed as a tool to identify the key issues and
conclusions to carry forward to risk characterization. This begins with planning and scoping and
continues throughout the risk assessment process. The contents of the table evolve throughout
the planning and scoping and risk assessment  processes, culminating in the risk characterization.
                                           i

       2.2.3   Should the Planning and Scoping Discussion Focus on What the Risk
              Assessment Results Should Be?

-------
        Page C4
                       DRAFT Risk Characterization Handbook
 .1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
       No! While Agency risk managers should meet often with their risk assessors and other
team members to discuss the need for, and the context of, the risk assessment, the discussions
should definitely not touch upon what the risk assessment result should be. The purpose of these
discussions is to insure that the needs for the assessment are well understood by those conducting
it, that the assessment is properly scoped, and that the results will be timely and useful for the
intended purpose.

       2.2.4   What Are the Benefits of Planning and Scoping?

       The planning and scoping process helps risk assessors understand how their risk, .
assessment and its characterization fits into the overall environmental decision-making process.
Preliminary information on the various inputs to decision-making, the possible roles and
participation of stakeholders, and how the analyses will be peer reviewed are considered at the
planning and scoping step.  Management concerns about funding, human resources, timing etc.
are also discussed.  This is important to a risk assessor, because TCCR requires that the risk
characterization portion of the risk assessment include a clear statement of the assessment's scope
and the reasonableness of its conclusions.

       Planning and scoping promotes:

       a)      Initial planning to save time and resources, and buy-in by stakeholders or
              interested parties by setting realistic expectations

       b)      Better-informed decisions, and less controversy (e.g., fewer court cases, criticism)

       c)      Participation by those from many disciplines (e.g., economists, lawyers) to help in
              the process thereby ensuring that each risk assessment and characterization is
              useful for the intended audience(s), and is of the scope and degree of complexity
              needed to inform the decision at hand in conjunction with other analyses, for
              instance, economics.

       2.2.5   Who Does Planning and Scoping?

       The planning and scoping process involves relevant risk managers, risk assessors and
other members of the "team" working on the decision that needs to be made. The "team"
includes the economists, lawyers, engineers, policy makers, etc. working on the issue at hand.
To insure that risk assessment meets the Agency's needs, and that those who will use the results
are fully informed, the communication within the team begun during the planning and scoping
phase should continue throughout the risk assessment process until the final risk characterization

-------
                              DRAFT Risk Characterization Handbook
                                                                              Page C5
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
is communicated to the decision maker(s).
       2.2.6  When Does the Risk Assessor/Risk Manager Dialog End?

       Once the risk management decision the dialog usually ends. Risk assessors work with
risk managers and others as a team. Ongoing dialog before and during the is essential for
successful completion of the risk assessment.

-------
                               DRAFT Risk Characterization Handbook
                                                                              Page Dl
  1
  2
  3
  4
  5
  6
  7
  8
  9
 10
 11
 12
 13
 14
 15
 16
 17
 18
 19
 20
 21
 22
 23
 24
 25
 26
 27
 28
 29'
 30
 31
•32
 33
 34
 35
 36
 37
 38
 39
1.
        APPENDIX D.  ROLE OF SCIENCE IN DECISION MAKING
Overview Statement
       A series of colloquia and roundtables were held to implement the Risk Characterization
Policy. They involved hundreds of managers and risk assessors from across the Agency. Their
major conclusion was that science is important to inform risk managers, but that are other factors
that also drive decision making (i.e., science informs/ policy decides).     "
  1                                           j                 •                   •    •
2.     Is the Scientific Risk Assessment with its Characterization the Driving Force Behind
       Decision Making?

       While scientific factors have  ostensibly been the primary basis for most regulatory and
risk management decisions, it is increasingly apparent that factors in addition to scientific risk
assessment (and economic analyses) play an important role in decision making.  It has been
recognized by outside parties as well (e.g., NAS (1994) and the Presidential/Congressional
Commission on Risk Assessment and Risk Management (1997)) that many other factors are
important in environmental decision making.
                                               . '    ;
3.     What Are the Major Factors that Affect Decision Making?

       Most risk management decisions are informed by at least six other factors in addition to
science (Figure D-l). Risks cannot be characterized and discussed without putting all these
factors into context.

       a)      Scientific factors provide the basis for the risk assessment, including information
              drawn from toxicology, chemistry, epidemiology, ecology, mathematics, etc.

       b)      Economic factors inform the manager on the cost of risks and the benefits of
              reducing them, the costs of risk mitigation or remediation options and the
              distributional effects

       c)      Laws and legal decisions are factors that define the basis for the Agency's risk
              assessments, management decisions, and, in some instances, the schedule, level or
              methods for risk reduction

       d)      Social factors, such as income level, ethnic background, community values, land
              use, zoning, availability of health care, life style, and psychological condition of
              the aifected populations, may affect the susceptibility of an individual or a

-------
        Page D2
                 DRAFT Risk Characterization Handbook
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
       definable group to risks from a particular stressor

e)     Technological factors include the feasibility, impacts, and range of risk
       management options

f)     Political factors are based on the interactions among branches of the Federal
       government, with other Federal, state, and local government entities, and even
       with foreign governments; these may range from practices defined by Agency
       policy and political administrations through inquiries from members of Congress,
       special interest groups, or concerned citizens

g)     Public values reflect the broad attitudes of society about environmental risks and
       risk management

-------
       Figure D-1.
   Risk Management
     Decision Steps
          " • ' .  i   .  • -*-
       Planning and Scoping
          Analysis
        Characterization
           Synthesis
                     Public] Values
Decisio
gal Factors

-------
       Page D4
                        DRAFT Risk Characterization Handbook
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
 [Figure D-l Caption]
 Risk Management Decision Framework. At least seven factors (represented by the arrows)
 affect and inform risk management decisions. Each factor passes through four analytical steps to
 integrate the information for a risk management decision (Figure 2 details1 scientific facter
-arrow):—                                                   .
 4.      Are the Economic and Other Non-Risk Assessments Subject to Risk
        Characterization?

        The risk characterization policy applies only to clarifying the risk assessment inputs to
 the decision making process.  However, the goal of risk characterization is to openly
 communicate the full range of scientific considerations surrounding a risk assessment. If
 managers wish to apply the values of transparency, clarity, consistency, and reasonableness to
 economic assessments, and to the other inputs besides risk that inform their decisions, the risk
 characterization criteria and points to consider can be readily adapted.  A risk manager who is
 informed by comprehensive information, analysis and characterization, can weigh all factors to
 make the decision, and help the public better understand the basis for his/her risk management
 decision.

-------

-------

-------
        Charge for the Generic Ketone Risk Characterization Case Study

Risk Characterization Issue for Peer Review

       The adequacy of any risk characterization depends in the first instance on the reliability
and credibility of the scientific and technical data and analyses in the risk assessment.  At the
same time, it is equally important that other risk assessors, decision-makers and risk managers,
and the public fully understand and appreciate the strengths and limitations of the assessment;
that is, the overall scientific "character" of the results. To test this aspect of risk characterization,
EPA's'Risk Characterization Policy and the draft Handbook set forth requirements for
transparency, clarity, consistency, and reasonableness (TCCR), and offer guidance on developing
and evaluating this aspect of the characterization of risk.

       This peer review does not focus on the underlying scientific analyses, but on whether the
presentation of the data and analyses adequately represents the risk assessment results.

Case Study Context and Background

       Under the reporting requirements of Section 313 of the Emergency Planning and
Community Right-To-Know Act of 1986 (EPCRA), facilities that use greater than 10,000
pounds or manufacture or process greater than 25,000 pounds of any chemical on the Toxic
Release Inventory (TRI) are required to report the total annual emissions to EPA and the states.
The criteria that are used to determine whether a chemical should be on the TRI include
consideration of both human health and ecological effects. It is possible for outside parties to
petition the EPA to list chemicals that are not currently included on the TRI or to delist
chemicals from the TRI. When a petition is submitted, the Agency has 180 days to respond.
Given this timeline, the Agency conducts'a "screening level" hazard^ and, if necessary, risk
assessment of the chemical in question.  The EPA recently received a petition to remove a
chemical, referred to in this assessment as "generic ketone", from the TRI. The purpose of this
assessment is to provide a screening level risk characterization of the potential risks associated
with exposure to generic ketone at the "fence line" of facilities that use, manufacture or process
generic ketone to determine whether the chemical should be included on' the TRI.

       This risk characterization is divided into three sections. Section 1 provides a background
to the case study and section 2 provides a description of the scope  of the assessment. Section 3
provides an overview of the risk assessment results.  This section is divided into four
subsections: the first provides a summary of the human health hazards associated with generic
ketone, the second provides a summary of the exposure information for facilities that use,
manufacture or process generic ketone, the third provides a summary  of the margin of exposure
approach used to estimate the "risk" at the "fence line", and the fourth provides an overview of
the conclusions of the analysis. Details on the hazard and exposure assessments are provided in
Appendix 1. The intended audience for this risk characterization is other risk assessors in the
TRI program and risk managers who must make the decision whether to retain or delete generic
ketone from the TRI.                           :            •

-------
                                                                           Generic Ketone
Points for Peer Reviewers to Address

1)     Using the criteria set forth in the draft Handbook, please evaluate the genetic ketone case
       in terms of the principles of transparency, clarity, consistency, and reasonableness.

2)     EPA is interested in your comments on the extent to which the characterization conveys a
       sense of transparency and clarity with regard to the usefulness of this "screening level"
       presentation for making a decision to determine whether a chemical should be included
       ontheTRI.

3)     Please comment of the topics listed below, which preliminary EPA reviews have
       identified as several areas of special interest:

              a)     Overall summary of the human health hazards and the uncertainties in the
                    context of EPCRA.
              b)     Overall summary of the exposure models and the uncertainties in the
                    context of their use for EPCRA.
              c)     Language describing the margin of exposure approach.
              d)     Discussion of the uncertainties and their impact on the margin of exposure
                    analyses.

4)     EPA welcomes any additional comments or suggestions.

-------
               RISK CHARACTERIZATION OF GENERIC KETONE
 1. BACKGROUND

        Under the reporting requirements of Section 313 of the Emergency Planning and
 Community Right-To-Know Act of 1986 (EPCRA), facilities that use greater than 10,000
 pounds or manufacture or process greater than 25,000 pounds of any chemical on the Toxic
 Release Inventory (TRJ) are required to report the total annual emissions to EPA and the states.
 The criteria that are used to determine whether a chemical should be on the TRI include
 consideration of both human health and ecological effects as follows:

        1) The chemical is known to cause or can reasonably be anticipated to cause significant
 adverse acute human health effects at concentration levels that are reasonably likely to exist
 beyond facility site boundaries as a result of continuous, or frequently recurring, releases;

        2) The chemical is known to cause or can reasonably be anticipated to cause in humans
 cancer or teratogenic effects, or serious or irreversible effects including reproductive
 dysfunctions, neurological disorders, heritable gene mutations^ or other chronic health effects.

        3) The chemical is known to cause or can reasonably be anticipated to cause, because of
 its toxicity, its toxicity and persistence in the environment, or its'toxicity and tendency to
 bioaccumulate in the environment, a significant adverse effect on the environment.

        In accordance with Agency science policy, traditionally cancer (and heritable gene
 mutations) have been viewed as non-threshold effects, whereas non-cancer effects have been
, viewed as threshold effects. Accordingly, the analyses required to include a chemical on the TRI
 differ for heritable gene mutations and cancer versus non-cancer effects. Hazard information that
 provides evidence that the chemical causes or can be reasonably anticipated to cause heritable
 gene mutations or cancer hi humans is  sufficient for a chemical to be included on the TRI. In
 contrast, for non-cancer effects, the hazard data must first be evaluated and determined to be
 sufficient to provide evidence that the chemical can reasonably be anticipated to pose a hazard to
 humans. If the hazard case is determined to be strong enough, then a risk assessment is
 subsequently conducted to .demonstrate that under the specific exposure conditions, the chemical
 can be reasonably  anticipated to cause the effect in humans.
                                                                  •   •     '  *
        It is possible for outside parties to petition the EPA to list chemicals that are not currently
 included on the TRI or to delist chemicals from the TRI.  When a petition is submitted, the
 Agency has 180 days to respond. Given this timeline, the Agency conducts a "screening level"
 hazard, and, if necessary, risk assessment of the chemical in question.

-------
 2. SCOPE

       The EPA recently received a petition to remove a chemical, referred to in this assessment
 as "generic ketone", from the TRI.  Originally, EPA had included generic ketone on the TRI due
 to concerns for developmental toxicity, neurotoxicity, hepatic toxicity, and renal toxicity; there
 were no concerns for ecological effects. The petitioner stated that new information indicate that
 the toxicity profile for generic ketone does not meet the criteria for listing on the TRI. In
 addition, the petitioner stated that the exposure estimates for facilities with the highest reported
 releases did not support a concern for risk to human health.

       The scope of this risk characterization is to provide a screening level risk assessment of
 generic ketone to EPA senior managers in order to assist in addressing the delisting petition All
 available epidemiology and animal toxicology studies of generic ketone were reviewed as to
 whether they met the above EPCRA criteria for listing.  Potential ecological effects were not
 addressed since this had not been the basis for the original listing of generic ketone and no new
 information is available that would impact the original assessment. Agency risk assessment
 guidelines were followed for the hazard assessment. The hazard summary of the potential human
 health effects is provided in the Appendix.

       For human health effects, two exposure scenarios are considered when assessing the
 exposure to a chemical that is listed on the TRI. The first scenario is the ambient air
 concentration at the fenceline of a particular facility, and the second is the concentration in the
 surface water that feeds into a drinking water facility. Estimates of ambient air concentrations of
 generic ketone at the fenceline of specific facilities were derived by use of the Industrial Source
 Complex Short Term model. Estimates of generic ketone in drinking water were derived by the
 ReachScan model. Agency exposure guidelines were followed for the exposure assessment. The*
 exposure analysis is also provided in the Appendix.

       The risk characterization for generic ketone is presented below. It was concluded from
the risk assessment that there is low concern for human health effects resulting from exposure to
ambient air concentrations of generic ketone at the fence line or from surface water releases of
generic ketone.

3. RISK CHARACTERIZATION

3.1 Hazard Characterization                                          ,

      The epidemiology and animal toxicology studies were evaluated to determine the overall
toxicological profile of generic ketone and to determine whether  sufficient evidence  exists to
demonstrate that generic ketone can cause or reasonably be anticipated to cause severe or
irreversible health effects hi humans. In general, there are very limited data available concerning
the potential toxicity of generic ketone. Generic ketone is an eye and respiratory irritant in
humans at concentrations of 100-500 ppm, but it has not been associated with significant

-------
 neurobehavioral effects.                               :                   •

       In animal studies, generic ketone has low acute toxicity by the oral, dermal and inhalation
 routes. For example, in acute oral toxicity studies, the LD50 in rats, mice, and guinea pigs, ranges
 from 1.9-4.6 g/kg. A dermal LDSO in the rabbit of >16 g/kg has been reported. In acute
 inhalation toxicity studies hi rats, the LC50 ranges from 2000 to greater than 4000 ppm.

       Subchronic animal studies have provided-equivocal evidence of neurotoxicity, hepatic
 toxicity, and renal toxicity. However, chronic studies were not available so it was not possible to
 support or refute the findings from the short-term studies. Similarly, there are no data available
 regarding the potential reproductive toxicity or carcinogenicity of generic ketone. The results of
 mutagenic assays indicate that generic ketone has little mutagenic activity.    ,

       Inhalation prenatal developmental  toxicity studies have been conducted in rats and mice.
 Developmental toxicity was observed in mice and rats, but maternal toxicity was not observed in
 either species. In mice, exposure to generic ketone was associated with an increased incidence of
 dead fetuses, reductions in fetal body weight, and delayed ossification. In rats, exposure to
 generic ketone was associated with reduced fetal body weight and delayed ossification.  For both
 species, the LOAEL was 3000 ppm and the NOAEL was 1000 ppm.

       The only toxicological studies that provide sufficient evidence that generic ketone can be
 reasonably.anticipated to cause serious or irreversible he,alth effects are the  developmental
 toxicity studies.  According to the EPA guidelines for developmental toxicity risk assessment
 (1991), evidence of developmental toxicity in a single animal study is sufficient to assume a
 potential hazard to humans. Since this is a non-cancer endpoint, it is necessary to conduct a risk
 assessment to determine whether there is a potential hazard to humans as specified under EPCRA
 313.                   '                           •."•'"-.'..-

  }     There are several sources of uncertainty associated with the hazard assessment. The
 major uncertainty is due to the paucity of human health effects and toxicological information on
 generic ketone. Sufficient evidence exists to support a potential concern for developmental
 effects. However, there is a great deal of uncertainty regarding the potential for other health
 effects due to a lack of information.

 3.2 Exposure Characterization                         :

       For human health effects, two exposure scenarios are considered when assessing the
 exposure to  a chemical that is listed on the TRI. The first scenario is  the ambient air
 concentration at the fenceline of a particular facility, and the second is the concentration hi the
 surface water that feeds hito a drinking water facility. Facilities that meet the reporting
requirements of EPCRA 313 must report the total  annual emissions of the chemical listed on the
TRI to the EPA.  The information is supplied simply as the total annual emission and may be
based on actual measurements of the emissions or on estimates of the emissions.  No information

-------
is provided regarding the pattern of the emissions throughout the ye'ar. Therefore, for this
exposure assessment it was necessary to estimate daily air concentrations'and water
concentrations of generic ketone from a single estimate of the total amount released during the
year by each facility.                   .
                                                        .                       -j
       Releases reported for generic ketone during 1994 were retrieved from the Toxic Release
Inventory System (TRIS) data base.  According to'TRIS, more than 25,500,000 pounds of
generic ketone were released in 1994 from 1,031 sources nationwide. Of this amount, 27 percent
was from fugitive or nonpoint source emissions and 72 percent originated from stack or point
source emissions to the atmosphere.  In addition, lesser amounts of generic ketone (less than 1
percent) were released to surface waters, underground injection of wastes, and the land.

3.2.1. Estimates of Ambient Air Concentrations

       The Industrial Source Complex Short Term (ISCST3) model was used to derive estimates
of the ambient air concentration of generic ketone at the fenceline. For this assessment,
modeling was conducted for the three facilities that reported the highest releases of generic
ketone hi 1994 to air (stack and fugitive). The ISCST3 model was used to calculate estimates of
the ambient air concentration for three scenarios, the single worst day (highest concentration) of
the year, the 50th worst day of the year and the average day. Each of these estimates for the three
facilities is shown in Table 1.

                                        Table 1
Estimated Ambient Air Concentrations of Generic Ketone For Three Facilities
Facility
A
B
C
Type of Day Modeled
Highest Concentration Day
50th Highest Concentration Day
Average Day
Highest Concentration Day
50th Highest Concentration Day
Average Day
Highest Concentration Day
50th Highest Concentration Day
Average Day
Estimated Ambient Air
Concentration (ppm)
4.8
1.9
0.5
2.3
1.3
0.3
2.0
1.2
0.25

-------
       There are many uncertainties associated with the estimates shown in Table 1. As noted
 above, the only site-specific information available was some meteorological information based
 on the zip code of the facility. All other parameters used in the model were default values which
,for the most part are based on conservative assumptions. The largest source of uncertainty is
 associated with the pattern and duration of the release of generic ketone. The values shown in
 Table 1 were derived based on the assumption that releases of generic ketone take place
 continuously over 365 days per year.  This assumption is necessary since the only site-specific   -
 data available are a single estimate of total annual release; therefore the actual number of days
 where releases occur is unknown.  However, if releases actually occur over shorter periods, the
 model would estimate higher ambient air concentrations. The impact of different assumptions
 regarding the pattern of releases of generic ketone on the estimates of the ambient air
 concentration is shown in Table 2.

                                  Table 2

       Estimated Ambient Air Concentrations of Generic Ketone -
                    Impact of Pattern of Release   ••-'..•
If Releases Occurred:
  Air Concentrations
•  Would Increase By:
Over 24 hours/day,     ,
but only on weekdays

Over 24 hours/day every day,
but only 6 months/year

Over 365 days, but only
one 8-hour shift per day

Over 24 hours/day every day,
but only 1 month/year  .

Over 24 hours/day every day,
but only 1 week/year
  A factor of 1.5
  A factor of 2
  A factor of 3
  A factor of 12
  A factor of 52

-------
3.2.2. Estimates of Surface Water Concentrations and Drinking Water Consumption
       The ReachScan model was used to derive estimates of the surface water concentration of
generic ketone resulting from reported releases of generic ketone to surface water. The potential
exposure of the population to releases of generic ketone to surface water would be through
consumption of drinking water. Drinking water consumption can be calculated on a daily basis
or it can be calculated as a lifetime average. The former estimate is generally referred to as the
acute potential dose rate (APDR) and the latter is generally referred to as the lifetime average
daily dose (LADD). The APDR is calculated when the toxic effect of a chemical is thought to be
the result of a short-term acute exposure, whereas the LADD is calculated when the toxic effect
is thought to be the result of long-term chronic exposure. For generic ketone, the primary
concern is for potential developmental toxicity. A central assumption in developmental toxicity
risk assessment is that developmental effects can result from a single exposure to the chemical.
Therefore, for this assessment, it is most appropriate to use estimates of the APDR.

       The three facilities with the highest annual releases to surface water were modeled by
ReachScan for this assessment. The estimated surface water concentrations and the associated
drinking water APDRs are presented in Table 3. The concentration in the water ranged from 4.4
to 47 ug/L at the three highest drinking water utilities. The drinking water APDRs range from
1.4 x 10"4 to 1.4 x 10'3 mg/kg/day.

                                        Table3

              Acute Exposures Resulting From Surface Water Releases
Facility
1
2
3
Estimated Surface Water
Concentration (ug/L)
47
9
4.4
Acute Potential Dose Rate
(mg/kg/day)
0.0014
0.00028
0.00014
       There are several sources of uncertainty associated with the estimates of surface water
concentration and the resultant estimates of APDR.  The greatest source of uncertainty pertains
to the assumption that generic ketone is released continuously over 365 days per year. However,
if releases actually occur over shorter periods, the model would estimate higher surface water
concentrations. For example, if the releases occurred over 10 days per year the value calculated
for the surface  water concentration would increase by a factor of 37.  The value would increase

-------
by a factor of 12 or 1.5 if releases occurred over 30 days per year or 250 days per year,
respectively. The resultant APDRs would increase by the same amount

3.3 Risk Assessment

       This assessment focuses on the potential risk of developmental toxicity associated with
exposure to generic ketone.  Developmental effects are considered in the analysis because this
was the only endpoint for which the hazard data were consistent with the criteria specified by
EPCRA 313. Other types of health effects are not considered either because the available data do
not support a concern that is consistent with the criteria, or the data are lacking. With regard to
potential human exposures, two exposure scenarios are considered: ambient air exposures at or
beyond the facility site boundary, and drinking water exposures due to releases to the surface
water. For each exposure scenario, exposure estimates were derived for the three facilities with
the highest releases of generic ketone. The estimates were derived through the use of 1994
annual release information submitted under TRI and standard modeling techniques.
                                                      /
       A margin of exposure (MOE) approach is used in this assessment to describe the potential
for developmental toxicity associated with exposure to generic ketone.  The MOE is calculated
as the ratio of the NO AEL for developmental toxicity to the estimated exposure level. The MOE
does not provide an estimate of population risk, but,simply  describes the relative distance
between the exposure level and the NO AEL. The value of the MOE that is associated with a
concern for toxic effects is generally expressed as the product of the applicable uncertainty and
modifying factors; uncertainty factors that the Agency considers for non-cancer effects are
described  in IRIS (1998). For consideration of developmental toxicity, the applicable uncertainty
factors are described in the developmental toxicity risk assessment guidelines (1991).  These
include two uncertainty factors, one for consideration of intraspecies variation and another for
interspecies variation.  In accordance with Agency science policy, each of these uncertainty
factors is given a value of 10.  Thus, for developmental effects, an MOE greater than 100 would
generally indicate a low level of concern, whereas a value less than 100 is judged to be of
concern.                                                             •  > . '

       As described previously, inhalation developmental toxicity studies of generic ketone have
been conducted in mice and rats. No maternal toxicity was noted in either study. Similar
developmental effects were noted in both species, and included reduced fetal body weight,
delayed ossification, and increased fetal death (mice only)., For both species, the LOAEL was
3000 ppm and the NO AEL was 1000 ppm. This NOAEL of 1000 ppm was used in the
derivation of the MOE. Separate anaylses were conducted for-the two exposure scenarios.  Each
is presented below.

3.3.1   MOE Calculations for Ambient Air Concentrations at the Fence Line

       To determine the MOEs for exposure at the fenceline of the three facilities with the
highest releases of generic ketone, the NOAEL of 1000 ppm was divided by the estimated

-------
ambient air concentrations (Table 1). MOEs were calculated for three exposure scenarios at each
facility: the day with the highest ambient air concentration of generic ketone, the 50th highest
day of the year and the average day of the year. These values are shown below in Table 4. The
MOEs are greater than 100 under all three exposure conditions at each facility, which in
accordance with Agency science policy would indicate a low level of concern for developmental
toxicity resulting from exposure to generic ketone at the fenceline for these facilities.
                                  Table 4
       MOEs for the Average to Worst Case Day of the Year
Facffity
A
B
C
Type of Day Modeled
Highest Concentration Day
50th Highest Concentration Day
Average Day
Highest Concentration Day
50th Highest Concentration Day
Average Day
Highest Concentration Day
50th Highest Concentration Day
Average 'Day
MOE
209
530
1944
430
785
3710
510
833
4082
       There are several sources of uncertainty associated with the hazard/dose response
assessment and the exposure assessment that impact the specific MOEs that are calculated. The
hazard assessment was conducted in accordance with the criteria used to assess whether a
chemical should be on the TRI.  Thus, the hazard data were assessed within a framework of
whether the data are sufficient to determine with reasonable certainty that serious or irreversible
effects are likely to occur in humans. There was some equivocal evidence of neurotoxicity,
hepatic toxicity and renal toxicity from short term animal studies; however, chronic toxicity
studies were not available so it was not possible to provide evidence to support or refute the
findings from the short term studies.  Similarly, there were no data available regarding the
potential reproductive toxicity or carcinogenicity of generic ketone. Thus, while the toxicologic
data for effects other than developmental effects were not strong enough for the purposes of
EPCRA, there is uncertainty regarding the potential for other types of effects resulting from
                                            8

-------
 exposure to generic ketone.

        A second source of uncertainty in the hazard/dose response assessment is the use of a
 NOAEL from a rodent study in the calculation of the MOE. The Agency has developed
 guidance for dosimetric conversions of animal inhalation concentrations to human equivalent
 concentrations. Unfortunately, for a compound such as generic ketone, such conversions require
 knowledge of the blood-to-air partition coefficient. This was not available for generic ketone and
 therefore derivation of a human equivalent concentration was not feasible. The impact of this on
 the MOE is not known.

        There are also many uncertainties associated with the exposure assessment.  As noted
 previously, the only available information on actual releases of generic ketone are total annual
 releases from the facilities. Therefore, ambient air concentrations are derived through the use of
 the ISCST3 model. This model requires the input of various parameters, and in the absence of
 site specific information default values are used.  In this case, the only site specific information
 that was available was some meteorological data obtained from the zip codes of the facilities.
 For the default values, the largest source of uncertainty exists for the number of hours per day   .
 and number of days per year that generic ketone was actually released from the facility.  A
 default value of 24 hours per day, 365 days per year was used to calculate the ambient air.
 concentrations. These estimates can increase up to a factor of 52 when releases are estimated
 over shorter durations (Table 2). If it was assumed that releases occurred only 8 hours per day
 (one work shift), 365 days per year, the estimates of ambient air concentrations would increase
 by a factor of 3.  This would result in estimates ranging from 14.4 ppm (facility A, highest
 concentration day) to 0.75 ppm (facility G, average day). The resulting MOEs would then range
 from 70 (facility A, highest concentration day) to 1333 (facility C, average day). This.would not
 change the level of concern for facilities B or C, but would indicate a higher level of concern for
 facility A for the worst case day. If it was assumed that generic ketone was released 24 hours per
 day, but only for 1 week per year the estimates of ambient air concentrations would increase by a
 factor of 52.  The estimates would then range from 13 ppm (facility G, average day) to 250 ppm
 (facility A, highest concentration day). Increasing the ambient air concentrations by a factor of
 52 would result in MOEs ranging from 4 (facility A, highest concentration day) to 77 (facility C,
 average day). In accordance with Agency policy, MOEs in this range would suggest that there is
 a relatively high concern for potential developmental effects.        .                    .

3.3.2. MOECalculations for Releases to Surface Water

       The potential exposure of the population to releases of generic ketone to surface water
would be through consumption of drinking water. Ideally, A NOAEL from an oral study would
be used for the derivation of MOEs in this exposure since this is the route .of concern. -However,
there are no oral developmental toxicity data available for generic ketone. Therefore, in order to
determine the MOEs for exposure from drinking water it was necessary to assume that the
inhalation developmental toxicity data was relevant for oral exposures. Accordingly, the

-------
NOAEL of 1000 ppm was converted to units of mg/kg/day1.
       MOEs were then derived by dividing the NOAEL in mg/kg/day by the estimated acute
potential dose rates (APDRs) (shown in Table 3); these are presented in Table 5. The APDR
estimates resulting from surface water releases for the top three discharging facilities range from
2.8 x 10'5 - 1.4 x 10'3 mg/kg/day. Using the rat NOAEL, the MOE values for these estimates
range from 3.3 x 107- 3.3  x 107. Using the mouse NOAEL, the MOE values for these estimates
range from 2.3 x 107- 4.6  x 106.

                                          TableS

                            MOEs for Drinking Water Consumption
Facility
1
2
3
MOE
(using rat NOAEL)
3.3 X 106
1.6X107
3.3 X 107
MOE
(using mouse NOAEL)
4.6 X106
2.3 X 107
4.6 X 107
       There are uncertainties associated with the assessment that could influence the calculated
MOEs. The largest source of uncertainty in the exposure assessment is the default value used
for the number of hours per day and number of days per year that generic ketone was actually
released from the facility. A default value of 24 hours per day, 365 days per year was used to
calculate the surface water concentrations. These estimates can increase up to a factor of 37
when releases are estimated over shorter durations. However, because the MOEs that are
calculated for drinking water are so large, increasing the APDR by a factor of 37 would not alter
the level of concern; the ADPR would have to increase by close to a factor of 1000 to have any
appreciable effect on the level of concern for developmental effects. Therefore, even though
              [ppm X (molecular weight/24.5) X rodent ventilation rate (mVday)]/ rodent body weight = mg/kg/day

For the rat, this becomes:

       [1000 ppm X (100/24.5) X 0.14 mVday]/ 0.124 kg = 4608 mg/kg/day

For the mouse, this becomes:

       [1000 ppm X (100/24.5) X 0.04 mVday]/ 0.025 kg = 6531 mg/kg/day
                                            10

-------
there is a great deal of uncertainty associated with the drinking water exposure, the MOE is so
large that there is a high level of confidence that there is no appreciable concern for
developmental effects resulting from exposure to generic ketone due to releases to surface water.

3.4  Conclusions

       Overall, the assessment supports a low concern for potential developmental effects
resulting from releases of generic ketone to air (stack or fugitive) or surface water.  There are
substantial uncertainties associated with the exposure assessments that could result in increases
in the estimates of ambient air concentrations by a factor of 52 and increase estimates of surface
water concentrations by a factor of 37.. Such an increase would not affect the level of concern for
releases to surface water since these estimates would have to increase by a factor of 1000 to
change the MOE enough to affect the level of concern.  Increasing the estimates of the ambient
air concentration by a factor of 52 may increase the level of concern for developmental toxicity.
                                           11

-------

-------
                    APPENDIX
HAZARD AND EXPOSURE ASSESSMENTS OF GENERIC KETONE
                       12

-------
 1. HAZARD SUMMARY

 1.1  Absorption and Metabolism                        ,

          Absorption and metabolism studies in animals suggest that generic ketone is well-
 absorbed from the lung, GI tract, and skin, well distributed, and rapidly metabolized. Although
 no metabolism studies have been conducted in humans, short-term exposure to generic ketone is
 associated with eye and respiratory irritation and clinical signs of reversible CNS effects; this is
 consistent witii the animal studies that demonstrate that generic ketone is rapidly absorbed.

 1.2  Acute Toxicity

       Available data indicate that generic ketone is associated with low toxicity in humans and
 animals following acute exposures. In humans, short-term inhalation exposures up to 30 minutes
 each day to concentrations as high as 500 ppm produced irritation of the eyes and upper and
 lower respiratory system, effects characteristic of solvent exposure. In some studies, reversible
 CNS and irritant effects were seen after 8 hour exposures to 100 ppm, while in other studies, 100
 ppm produced no effects (additional studies are described in section 1.8)

       In acute inhalation toxicity studies, rats were able to tolerate concentrations of 2000-4000
 ppm for periods up to 6 hours, while concentrations > 20,000 ppm produced death in all animals
 within an hour.  The 4- and 6-hour LC50 in rats were estimated as being 3000 and > 4000 ppm,
 respectively.  Although no mortality was reported in mice exposed to concentrations as high as
 900 ppm generic ketone, a decrease in the duration of immobility in a behavioral despair
 swimming test and a reduction in the respiratory rate were observed. In acute oral toxicity tests,
 the LDSO ranged from 1900-3000 mg/kg in the mouse to 3000-4600 mg/kg in the rat. The dermal
 LD50 in rabbits has been reported as being greater than 16 g/kg.

 13  Mutagenicity
                                                                               * •
       In general, generic ketone does not appear to be associated with genotoxicity in vitro or in
 vivo .  Generic ketone is not a gene mutagen in Salmonella typhimurium strains TA98, TA100,
 TA1535 and TA1538 either without or with metabolic activation.  It is weakly positive in
 L5178Y TK+/" mouse lymphoma cells in vitro without, but not with activation. It is not a
 chromosome mutagen in vitro in Chinese hamster ovary (CHO) and rat RL4 cells, nor does it
 induce micronuclei in vivo in the mouse micronueleus assay by the intraperitoneal injection
route.  Generic ketone does not induce DNA effects in the Saccharomyces cereviside
homozygosis and recombination assay, and it is equivocal in the unscheduled DNA synthesis
(UDS) assay in rat hepatocytes in vitro. It induces morphological cell transformation in BALB/c
3T3  cells in culture without and possibly with metabolic activation.
                                          13

-------
1.4  Carcinogenicity

            There are no human or animal carcinogenicity data on generic ketohe.

1.5  Systemic Toxicity from Repeated Doses

          Only one epidemiological worker study and a follow-up study are available on the
potential effects of generic ketone in humans. However, no information was provided
concerning the exposure of the individuals and potential confounders were not accounted for. As
a result, no definitive conclusions could be made.

       Limited animal data are available regarding the potential systemic toxicity of generic
ketone. A 90-day inhalation toxicity study in rats and mice has been conducted. In that study,
14 male and 14 female Fischer 344 rats and B6C3F1 mice per group were exposed to 0, 50, 250,
and 1000 ppm generic ketone by vapor inhalation 6 hr/day, 5 days/week for 14 weeks.
Parameters assessed for toxicity included clinical observations, body and organ weight data,
water consumption, urinalysis, serum chemistry, hematology, gross pathology, and histology.
No treatment-related effects were noted in the mouse study. In the rat study there was evidence
of hepatic toxicity as demonstrated by a dose-related increase in serum cholesterol levels in male
rats exposed to 250 and 1000 ppm (23% and 35% higher than controls, respectively).  In
addition, there was evidence of renal toxicity; there were statistically significant dose-related
increases in urine glucose excretion in male rats (55%, 37% and 23% for the 1000, 250 and 50
ppm levels, respectively) and hi female rats at 1000 ppm (28% above control values). In
addition, increases in total urinary protein were also observed. The authors of the study
suggested that the urinary glucose and protein excretion may be due to functional impairment of
normal reabsorption in the renal proximal convoluted tubules. Increases in renal hyaline droplets
were also noted in mid- and high-dose male rats. Although the presence of hyaline droplets in
the renal proximal tubules may be considered male-rat specific and could explain the functional
impairment of glucose and protein absorption in the kidney tubules, increased urinary glucose
was observed at the high dose in both sexes. Significant alterations in other parameters of renal
function did not occur.  •'-'.,,                        ,

       Another group of investigators conducted a 90-day inhalation toxicity study in dogs, rats
and monkeys.  In that study, 8 beagle dogs, 100  Wistar rats, and 2 monkeys were continuously
exposed for 90 days under space cabin conditions (reduced atmospheric pressure) to 100 ppm
generic ketone. Dogs and monkeys did not appear to have any toxic responses to the exposure.
Special staining of dog kidney sections did not reveal treatment-related effects. Since only 2
monkeys were used per group, the etiology of the chronic inflammation of the kidney hi one of
the exposed monkeys is uncertain.  Renal toxicity was clearly present in the rats as demonstrated
by hyaline droplet nephrosis. The lesions developed within two weeks of exposure and were
reversible after 90 days of exposure.                      .            '
                                           14

-------
        In addition, a subchronic oral toxicity study in rats has been conducted. In that study,
 four groups of an unspecified number of Sprague-Dawley rats were given 0, 50,250, and 1000
 mg/kg/day generic ketone by oral gavage daily for 90 days.  Doses of 250 mg/kg/day produced
 increased kidney weights and urinary ketones in both sexes and epithelial cells in males.  There
 were no treatment-related effects at 50 mg/kg/day.

       In summary, the major target organs in both the 90-day rat subchronic inhalation and oral
 toxicity studies were the liver and kidney. Oral gavage doses produced more severe reactions as
 demonstrated by the changes in clinical chemistry parameters indicative of hepatic toxicity,
 urinalyses, and histopathological changes in the male rat kidney. The main effects in the
 inhalation study appeared to be due to functional changes hi the liver and kidney; increased liver
 weight and increased serum cholesterol, and impaired renal absorption of protein and glucose in
 male rats. However, the elevations in serum chemistry parameters were slight and the liver and
 kidney effects in the inhalation study were considered to be relatively minor with no major signs
 of histopathological lesions with the exception of increases in renal hyaline droplets in mid-: and
 high-dose males.

 1.6  Developmental Toxicity

       Inhalation prenatal toxicity studies have been conducted in rats and mice.  In that study,
 30 pregnant CD-I mice and 35 pregnant Fischer 344 rats per group were exposed to 0, 300,
 1000, and 3000 ppm generic ketone by vapor inhalation 6 hr/day on gestation days (GD) 6-15.
 In mice, there was no evidence of maternal toxicity.  There was evidence of developmental
 toxicity as demonstrated by an increased incidence of dead fetuses, reductions hi fetal body
 weights per Utter, and delayed ossification were observed at the high- dose of 3000 ppm. No
 effects were noted at 1000 or 300. ppm.

       In rats, there was no evidence.of maternal toxicity. Exposure to 3000 ppm resulted in a
 reduction in fetal body weight per litter and delayed skeletal  ossification. Additionally, at 300,
 but not at 1000 ppm, there was evidence of reduced fetal body weights per litter and an increase
 hi delayed ossification. The authors of the study reported that historical control data from their
 laboratory for Fischer 344 rats indicate an inverse relationship between litter size and fetal body
 weight and offered this as an explanation for the decreases in fetal body weight observed at the
 low dose group of 300 ppm. Fetal body weight per litter was examined by dose and by litter size
 and evaluated statistically. Their analysis indicated that fetal body weights differed significantly
 from controls for both small and large litters at 3000 ppm, indicating a treatment-related effect.
 Conversely, fetal body weights at the 1000 ppm dose group were comparable to controls for both
 large and small litters. At 300 ppm, fetal body weights for small, but not large, litters were
 significantly reduced compared to controls. The authors contend that the significant reduction in
fetal body weight per litter seen in small Utters at 300 ppm was actually an artifact of
exceptionally heavy fetuses in two small litters in the control group and therefore not treatment-
related. Furthermore, the authors argued that since the control group overall had more smaller
litters as opposed to 300 ppm, the evidence of minimal delayed ossification at 300 ppm was

                                           15

-------
, consistent with larger litter sizes with concomitant lower fetal body weights and reductions in
 ossification and likewise is not considered to be treatment-related.

        In conclusion, adverse developmental effects were noted in the mouse and rat studies;
 maternal toxicity was not observed in either species. For both species, the NOAEL for maternal
 toxicity was 3000 ppm; the LOAEL for developmental toxicity was 3000 ppm and the NOAEL
 was 1000 ppm.

 1.7 Reproductive Toxicity

        No reproductive/fertility studies have been conducted with generic ketone. The only
 information available is from the 90 day inhalation toxicity study in mice and rats described
 above. In that study, organ weight and histological data in high-dose rats and mice were
 comparable to controls for the ovaries, uterus, oviducts, vagina, cervix, testis, epididymis,
 prostate, and seminal vesicles. However, this is not sufficient information to characterize the
 potential for reproductive toxicity of generic ketone.

 1.8 Neurotoxicity   .

           Several human studies have examined the neurotoxicity of generic ketone. Although
 the data are limited to studies with small numbers of subjects, the results are fairly consistent.
 One group of investigators  tested neurobehavioral performance following a 4-hour exposure to
 100 ppm generic ketone. Five different psychomotor tests, chemical measurements, and reports
 of sensory and irritant effects were measured. No marked neurobehavioral effects were reported,
 but sensory and irritant effects (i.e., odor, headache, nausea, throat irritation, tearing) were
 reported by 20-30% of the subjects exposed to generic ketone. In another study, 8 subjects were
 exposed to 50 ppm generic ketone for shorter exposure periods.  There were no significant
 effects on simple reaction time or mental arithmetic tasks. However, irritation of the nose,
 throat, headache, and vertigo were reported by up to a third of the subjects (based on a
 questionnaire). Subjects reported an increase in degree of irritative and CNS symptoms for
 exposures of 24 ppm and 48 ppm, as compared to 2.4 ppm. Similar results were also reported for
 2-hour exposures to 2.4 and 48 ppm generic ketone where subjects reported fatigue and irritation
 to arrways, but no reduction of performance on reaction time  or arithmetic tests.

       Numerous studies have-been conducted in animals to  assess the neurotoxic potential of
 generic ketone and generally have found no evidence of permanent impairment of neurological
 function.  For example, histological examination of the nervous system including the brain,
 spinal cord, and peripheral nerves, was unremarkable and no  clinical signs of neurotoxicity were
 observed in rats or mice exposed to doses as high as 1000 ppm generic ketone by vapor
 inhalation for 90 days (study described above in section 3.5).

     •  A 90-day inhalation study of schedule-controlled operant behavior (SCOB) in rats has
 also been conducted. In this study, male rats were subjected to exposures of 0,250, 750 and

 ','••'"•••             ,             16

-------
1500 ppm generic ketone for 6 hr/day. The results showed that rats exposed to 1500 ppm
exhibited reduced activity and sialorrhea (excessive salivation) following one hour of exposure.
This effect was transient, and did not persist after ten weeks of exposure. The same effect was
seen in animals exposed to 750 ppm, following two hours of exposure, but was only seen during
weeks 1 through 8. No effects were seen during exposure to 250 ppm.  No significant
behavioral effects were detected following exposure at any of the doses tested, using the
schedule-controlled operant behavior test.  These data indicate that generic ketone may cause
transient neurologic effects.

       In summary, the available human data are consistent with data previously summarized
that show that exposure to generic ketone is associated with eye and respiratory irritation at high
concentrations. However, there are no human or animal data that demonstrate an association
between exposure to generic ketone and serious and irreversible neurological effects.

1.9 Hazard Characterization

       Human studies have reported irritation of the eyes and mucous membranes as well as
symptoms such as headache, nausea, and vertigo (effects characteristic of solvent exposure) due
to inhalation of generic ketone at concentrations ranging from 100 to 500 ppm. However, no
significant neurobehavioral effects have been reported at these concentrations.

       In animal studies, generic ketone has low acute toxicity by the oral, dermal and inhalation
routes. For example, in acute oral toxicity studies, the LDSO in rats, mice, and guinea pigs, ranges
from 1.9-4.6 g/kg. A dermal LD50 in the rabbit of >16 g/kg has been reported. In acute
inhalation toxicity studies in rats, the LC50 ranges from 2000 to greater than 4000 ppm.
                                                                            \
       The results of mutagenic assays indicate that generic ketone has little mutagenic activity.
There is neither-human nor animal data on the potential carcinogenicity of generic ketone.
Subchronic inhalation studies in rats, mice, dogs, and monkeys exposed to concentrations
ranging from 50-2000 ppm generic ketone indicate liver and kidney toxicity. However,  in the
absence of appropriate chronic data, the data were considered inadequate to support a concern
for serious or irreversible effects. Neurotoxicity studies in rats on generic ketone indicate that
transient CNS depression can occur at high exposure levels? however, there is no evidence to
support a concern for serious or irreversible neurological effects.  No conclusions regarding the
potential for reproductive toxicity upon exposure to generic ketone can be made since no
reproductive/fertility studies have been conducted.

       Inhalation developmental toxicity studies in rats and mice demonstrate that generic
ketone is not associated with maternal toxicity. Developmental  toxicity was observed in mice
and rats. In mice, exposure to generic ketone was associated with an increased incidence of dead
fetuses, reductions in fetal body weight, and delayed ossification.  In rats, exposure to generic
ketone was associated with reduced fetal body weight and delayed ossification.  For both species,
the LOAEL was 3000 ppm and the NOAEL was 1000 ppm. According to the EPA guidelines

                                           17

-------
for developmental toxicity risk assessment (1991), evidence of developmental toxicity in a single
animal study is sufficient to assume a potential hazard to humans.

2. EXPOSURE SUMMARY

       For human health effects, two exposure scenarios are considered when assessing the
exposure to a chemical that is listed on the TRI.  The first scenario is the ambient air
concentration at the fenceline of a particular facility, and the second is the concentration in the
surface water that feeds into a drinking water facility.  As noted previously, facilities that meet
the reporting requirements of EPCRA 313 must report the total annual emissions of the chemical
listed on the TRI to the EPA. The information is supplied simply as the total annual emission
and may be based on actual measurements Of the emissions or on estimates of the emissions.  No
information is provided regarding the pattern of the emissions throughout the year. Therefore,
for this exposure assessment it was necessary to  estimate daily air concentrations and water
concentrations of generic ketone from a single estimate of the total amount released during the
year by each facility.

       Releases reported for generic ketone during 1994 were retrieved from the Toxic Release
Inventory System (TRIS) data base. According to TRIS, more than 25,500,000 pounds of
generic ketone were released in 1994 from 1,031 sources nationwide. Of this amount, 27 percent
was from fugitive or nonpoint source emissions and 72 percent originated from stack or point
source emissions to the atmosphere.  In addition, lesser amounts of generic ketone (less than 1
percent) were released to surface waters, underground injection of wastes, and the land.

       The Industrial Source Complex Short Term (ISCST3) model was used to -derive estimates
of the ambient air concentration of generic ketone at the fenceline. For this assessment,
modeling was conducted for the three facilities that reported the highest releases of generic
ketone in 1994 to air (stack and fugitive). The ReachScan model was used to. derive estimates of
the surface water concentration of generic ketone.  This information was then used to calculate
general population  exposures resulting from surface water releases to drinking water sources. A
description of the ISCST3 and the ReachScan models is provided below.

2.1 Modeling Ambient Air Concentrations of Generic Ketone

       The ISCST3 model was used to estimate short-term ambient air concentrations. The
model requires the input of certain information such as: pollutant emission rate; stack height (for
point sources); release height (for area sources); stack gas temperature; stack diameter; stack gas
exit velocity; location of the point of emission with respect to surrounding topography, and the
character of that topography; a detailed description of all structures hi the vicinity of the stack in
question; and, similar information from other significant sources in the vicinity of the subject
source stack height. Ideally, the  input for these parameters would be based on site-specific data.
However, in the absence of site-specific information ^generic default values are used.
                                          18

-------
       For this case, the only site-specific information available was some meteorological data.
By using the zip code or the latitude/longitude of the release site, the ISCST3 model can access
meteorological data (e.g., wind speed and direction)from the nearest weather station. If several
stations are nearby, the user selects the one that he/she believes adequately portrays the release
site. The following generic parameters were used for the three facilities:
 STACK PARAMETERS
 Duration of releases:
 Release height:
 Inner stack diameter:
 Exit gas temperature:
 Exit gas velocity:
 Distance to fence line:
 Site layout:
 Generic ketone half life:
 Other modeling options:

 FUGITIVE (AREA) PARAMETERS
 Duration of releases:
 Release height:
 Exit gas temperature:
 Area source size:
 Exit gas velocity:
 Distance to fence line:
 Site layout:
 Generic ketone half life:
 Other modeling options:
 ASSUMPTIONS
 24 hours
 10 meters
 0.01 meters
 293°K
 0.01 meters/sec
 100 meters
 flat, rural
 164,160 seconds (1.9 days)
 default  ,

 ASSUMPTIONS
 24 hours
 3 meters
 293°K
 10m by 10m
0.01 meters/sec
 100 meters
 flat, rural
 164,160 seconds (1.9 days)
default
       The ISCST3 model was used to calculate estimates of the ambient air concentration for
three scenarios, the single worst day (highest concentration) of the year, the 50th worst day of the
year and the average day. Each of these estimates for the three facilities is shown in Table 1.
                                           19

-------
                           Table 1
Estimated Ambient Air Concentrations of Generic Ketone For Three Facilities
Facility
A
B
C
Type of Day Modeled
Highest Concentration Day
50th Highest Concentration Day
Average Day
Highest Concentration Day
50th Highest Concentration Day
Average Day
Highest Concentration Day
50th Highest Concentration. Day
Average Day
Estimated Ambient Air
Concentration (ppm)
4.8
1.9
0.5
,2.3
1.3
0.3
2.0
1.2'
0.25
       There are many uncertainties associated with the estimates shown in Table 1.  As noted
above, the only site-specific information available was some meteorological information based
on the zip code of the facility. All other parameters used in the model were default values which
for the most part are based on conservative assumptions. The largest source of uncertainty is
associated with the pattern and duration of the release of generic ketone. It is necessary to use an
assumption since the only site-specific data available are a single estimate of total annual release;
therefore the actual number of days where releases occur is unknown. The values shown in
Table 1 were derived based on the assumption that releases of generic ketone  take place
continuously over 3 65 days per year. However, if releases actually occur, over shorter periods,
the model would esthnate higher ambient air concentrations. The impact of different ,
assumptions regarding the pattern of releases of generic ketone on the estimates of the ambient
air concentration is shown in Table 2.
                                          20

-------
                                 Table 2

       Estimated Ambient Air Concentrations of Generic Ketone
                    Impact of Pattern of Release
IF RELEASES OCCURRED:
                    AIR CONCENTRATIONS
                    WOULD INCREASE BY:
Over 24 hours/day,
but only on weekdays '

Over 24 hours/day every day,
but only 6 months/year

Over 365 days, but only
one 8-hour shift per day

Over 24 hours/day every day,
but only 1 month/year

Over 24 hours/day every day,
but only 1 week/year
                          A factor of 1.5
                          A factor of 2
                          A factor of 3
                          A factor of 12
                          A factor of 52
2.2 Modeling of Surface Water Concentration of Generic Ketone

       The ReachScan model was used to estimate the surface water concentrations resulting
from reported annual releases of generic ketone to'surface water.  ReachScan is a simple dilution
model used to estimate steady-state chemical concentration in surface water bodies (mainly river
reaches) due to a continuous loading from a single discharging facility. Several default
parameters are used for the modeling including:
 Duration of releases:
 Distance of search:
 Direction of search:
 Type of search:
 Flow type:
Constant over 365 days per year
200km
Downstream
Search for utilities
Harmonic mean
      The model provides stream concentration estimates at the reach where the releasing

                                        "21

-------
 facility is located and at the reach where the drinking water utilities are located. In addition to
 the default assumptions listed above, it was also assumed that generic ketone was not removed
 in-stream (e.g., volatilization) or at the drinking water facility.  Thus, the estimate of the stream
 concentration at the drinking water utility is assumed to be the same as the concentration in the
 drinking water.

       As noted above, the greatest source of uncertainty pertains to the assumption that generic
 ketone is released continuously over 365 days per year. However, if releases actually occur over
 shorter periods, the model would estimate higher surface water concentrations. For example, if
 the releases occurred over 10 days per year the value calculated for the surface water
 concentration would increase by a factor of 37.  The value would increase by a factor of 12 or 1.5
 if releases occurred over 30 days per year or 250 days per year, respectively.

 2.3    Estimation of Acute Potential Dose Rates Via Drinking Water Consumption

       The potential exposure of the population to releases of generic ketone to surface water
 would be through consumption of drinking water. Drinking water consumption can be calculated
 on a daily basis or it can be calculated as a lifetime average. The former estimate is generally
 referred to as the acute potential dose rate (APDR) and the later is generally referred to as the
 lifetime average daily dose (LADD). The APDR is calculated when the toxic effect of a
 chemical is thought to be the result of a short-term acute exposure, whereas the LADD is
 calculated when the toxic effect is thought to be the result of long-term chronic exposure. For
 generic ketone, the primary concern is for potential developmental toxicity. A central
 assumption in developmental toxicity risk assessment is that developmental effects  can result
 from a single exposure to the chemical.  Therefore, for this assessment, it is most appropriate  to
 use estimates of the APDR.

       In accordance with Agency guidelines for exposure asse'ssment, the APDR was calculated
 using the following equation:

 APDR = CxIRxCFl
            BW

where:

 C      =     surface water concentration (ug/1)
IR     =     water intake rate (I/day)
CF1   =     conversion factor from ug to mg  (0.001)                                      "
BW   =      body weight (kg)

The following assumptions were made:                              '

C  , •   =     Calculated using a simple dilution water model executed in ReachScan. The

 '  '  '•          .  .      . .     -      -     '  22"    '       '      "'   •

-------
IR
CF    =
BW   =
concentration in the water varies from facility to facility. Used the highest stream
concentration estimated by ReachScan and assumed no removal
2 I/day, this value represents the high-end value for water consumption
Conversion from ug to mg (1 .Oe"3)
65 kg, this value represents adult females
            The three facilities with the highest annual releases to surface water were modeled
by ReachScan for this assessment. The estimated surface water concentrations and the
associated drinking water APDRs are presented in Table 3. The concentration in the water ranged
from 4-4 to 47 ug/L at the three highest drinking water utilities. The drinking water APDRs
range from 1.4 x 10"4 to 1.4 x 10'3 mg/kg/day.

                                          TableS
                *      '      •                    •   ' ' '' '•'-.:
                     Acute Exposures Resulting From Surface Water Releases
Facility
1
2
3
Estimated Surface Water
Concentration (ug/L)1
47
9
4.4
Acute Potential Dose Rate
(mg/kg/day)1
0.0014
0.00028
0.00014
 Assumes that the amount reported released to water in 1994 was released over 365 .days/year.
If the number of release days were changed to 10, 30, or 250, the resulting surface water concentrations and APDR would
increase by a factor of 37,12 and 1.5, respectively.

2.4 EXPOSURE CHARACTERIZATION

       Ninety-nine percent of generic ketone released to the environment is through stack (point)
and fugitive (area) emissions into the atmosphere. The remaining one percent of releases go to
surface waters, landfill and deep well injections.  For this assessment, ambient air concentrations
and surface water concentrations were estimated for the three facilities with the highest releases
of generic ketone; these values were estimated through the use of two models, the ISCST3 and
ReachScan models.

       In the absence of site-specific information, each model requires the use of various default
assumptions which introduce uncertainties in the analysis. The greatest uncertainty is due to the
assumption regarding the number of days during the year that the facility releases generic ketone.
This assumption arises because of the fact that facilities only report total annual releases and do
                                            23

-------
not provide information on the pattern of the releases. For this assessment, it was assumed that
the air and water releases of generic ketone occurred evenly over 365. days per year. However,
the values calculated could be substantially higher if releases occurred for less than 24 hours per
day (such as only during a 8 hour work shift) or occurred for less than  365 days per year.
                                           24

-------

-------

-------

-------
  Charge for Waquoit Bay Watershed Ecological Risk Assessment Case Study
        Problem Formulation Summary and Proposed Risk Characterization

 Risk Characterization Issue for Peer Review

       The adequacy of any risk characterization depends in the first instance on the reliability
 and credibility pf the scientific and technical data and analyses in the risk assessment.  At the
 same time, it is equally important that other risk assessors, decision-makers and risk managers,
 and the public fully understand and appreciate the strengths and limitations pf the assessment,
 that is, the overall scientific "character" of the results. To test this aspect of risk characterization,
 EPA's Risk Characterization Policy and the draft Handbook set forth requirements for
 transparency, clarity, consistency, and reasonableness (TCCR), and offer guidance on developing
 and evaluating this aspect of the characterization of risk.

       This peer review does not focus on the underlying scientific analyses, but on whether the
 presentation of the data and analyses adequately represents the risk assessment results.

 Case Study Context and Background
                         /                ,             -           '               -
       Waquoit Bay is a small estuary on the south coast of Cape Cod, Massachusetts.  Initially
 valued for hunting, farming, and fishing, the Waquoit Bay watershed now primarily provides
 aesthetic and recreational opportunities. Population growth and development in the Cape Cod
 area have placed increasing pressures on the natural resources pf Waqupit Bay.  Problems such
 as fish, kills and the decline of key fish and shellfish populations have lead to a number of
 initiatives, including an ecplpgical risk assessment which is now underway and is the subject of
 this case study.                                   .                 •

       This case study evaluates the risks to valued ecological resources in Waquoit Bay and
 was initiated to provide management options for local decision-makers.  The Waquoit Bay
 system is complex, and many possible stressors and ecological values of concern have been
 identified. In this assessment, risk assessors are focusing on one aspect of the system ~ estuarine
 eelgrass habitat abundance and distribution — because it was judged to be a key to the recovery
 of many other elements of the system. Assessprs also chpse to fpcus pn nutrient Ipading as the
 key stresspr, and develppedmpdels tp link nutrient spurces, transpprt, and fate in Waqupit Bay
 with pptential effects pn eelgrass. Since the assessment is presently in the analysis phase, a
 cpmpleted risk characterizatipn cannpt be presented at this tune.

       This document begins with background information on the Waquoit Bay estuary and the
 initiation of the risk assessment (section 1).  Section 2 includes elements of planning and
problem formulation. Problem formulation provides the foundation for an ecolpgical risk
assessment. The pbjectives and scope for the risk assessment are. refined, the nature of the
problem is evaluated and a plan for analyzing data and characterizing risk is developed.

-------
                                                                              Waquoit Bay
Elements of planning and problem formulation include establishing management goals and
objectives, defining assessment endpoints, preparing, a conceptual model, and developing an
analysis phase plan.  The proposed risk characterization is discussed in section 3.

Points for Peer Reviewers to Address

1)     This case study differs from the others in that it is an ongoing assessment without a
       completed risk characterization. Nevertheless, using the criteria set forth in the draft
       Handbook, please evaluate the Waquoit Bay case elements in terms of the principles of
       transparency, clarity, consistency, and reasonableness.

2)     EPA is interested in how well this case study provides a clear and transparent linkage
       between the proposed risk characterization and the elements of planning and problem
       formulation (management goals and'objectives, assessment endpoints, a conceptual
       model, and an analysis plan).

3)     Section 1. Please comment on how well this section summarizes background information
       needed to understand the ecological and management context for the Waquoit Bay
       assessment.

4)     Section 2. Please comment on the following areas identified hi preliminary reviews:

       a)      Rationale for focusing the assessment on one stressor (nutrients) and one
              assessment endpoint (eelgrass).
       b)      Effectiveness of the conceptual model in clearly expressing relationships between
              sources, stressors, the ecological system, and assessment endpoints.
       c)      Identification of critical assumptions and uncertainties for the exposure  and
              effects models.

5)     Section 3.  Please comment on the following issues:

       a)     Ability of the proposed risk characterization to address keyjnanagement
          ;   concerns:
             i)     Sources of nutrients and their relative contributions.
             ii)     Effects of different degrees of nutrient loadings.
       b)     Should the risk characterization attempt to address the comparability of nutrient
             loading situations hi other estuarine sites, e.g., rates of decline or recovery of
             eelgrass hi other areas?
       c)     Consideration of recovery of the eelgrass community.
       d)     Significance of possible ecological changes — how best could the linkage between
             recovery of eelgrass and recovery of other estuarine organisms be addressed?

-------
                                            3'  .--                               WaquoitBay

       e)     The case study acknowledges the existence of a range of stressors in Waquoit Bay
              but focuses on nutrients as the primary stressor of concern.
              i)      How should the case study reflect the likelihood of eelgrass return if
                     nutrient reduction is achieved but other  stressors are still present?
              ii)     How should the case study consider the likelihood of return of other
                     valued ecological resources given that eelgrass returns but other stressors
                     are still present?                                          .

6)     EPA welcomes any additional comments or suggestions.

-------
r

-------
The Waquoit Bay Watershed Ecological Risk Assessment Case Study:
Problem Formulation Summary and Proposed Risk Characterization
1. Introduction
This document includes summary information from the planning and problem formulation report
produced for the Waquoit Bay ecological risk assessment case study and a description of the
planned risk characterization component of the risk assessment.
         '  '    •                            »  • ,->                               :        •
                                                                                        ,1
The Waquoit Bay watershed ecological risk assessment was conducted to evaluate the danger to
valued water resources from stressors caused by human activities, and to provide resource
managers with viable options to protect the resources. A qualitative risk analysis identified
nitrogen-loading as a primary stressor in estuarine habitats of the watershed and eelgrass habitat
as the most important assessment endpoint. Because of these findings and due to constraints of
limited data to assess other endpoints, the risk assessment focused on the risk to eelgrass habitat
from nitrogen loading from the adjacent watershed.

The goal of the Waquoit Bay ecological risk assessment is to provide managers with answers to
key questions.

       What are the sources of nutrients and their relative contributions?
       What will be the effects of different degrees of nutrient reduction?
 - •  .    i          .    -      *             -     r               -            .         • '
1.1 The Watershed
Waquoit Bay is a small estuary on the south coast of Cape Cod, Massachusetts. Its watershed
covers about 53 square kilometers (21 square miles) of freshwater streams and ponds, salt ponds
and marshes, pine and oak forest, barrier beaches, and open estuarine waters. The land and water
are home, spawning ground, and nursery for plant and animal life including piping plovers, least
terns (endangered birds), the sandplain gerardia (an endangered plant), alewife, winter flounder,
blue crab, scallops and clams, and other fish species that migrate through the estuary. Initially

-------
valued for hunting, farming, and fishing, Waquoit Bay now primarily provides aesthetic and
recreational opportunities, demands that have generated residential development and business for
local marine-dependent industries.

Cape Cod's economic viability is largely dependent on tourists who are drawn to the sandy
beaches, seafood restaurants, boating opportunities and water recreation areas. Thus the
economy on Cape Cod and the environment on Cape Cod are mutually inter-dependent. The
once rural surroundings have become increasingly suburbanized as bedroom and retirement
communities have sprung up. Barnstable County, where the Waquoit Bay watershed is located,
is the fastest growing county hi Massachusetts. As the population grows, so does pressure on the
valuable natural resources that have attracted people to the area.

Living in bottom sediments of shallow embayments of the northwestern Atlantic is a flowering
plant known as eelgrass (Zostera marina). Numerous studies have shown that eelgrass meadows
provide a very good habitat for many commercially and recreatiohally important fish and
shellfish. Eelgrass needs a  lot of light to grow. In Waquoit Bay, increased phytoplankton
(microscopic one-celled organisms) and seaweed populations, fueled by the addition of nitrogen
from coastal development, have decreased the amount of light penetrating the water. In 1951
eelgrass meadows covered most of Waquoit Bay proper and its adjoining coastal ponds and
rivers.  Today, eelgrass is absent from the Bay proper and has declined significantly in the ,
adjoining tributaries and ponds. Species dependent on eelgrass, particularly scallops, have
likewise decreased. In 1987,1988, and 1990, fish kills occurred in Waquoit Bay, and the
northern beach was covered with thousands of dead winter flounder, shrimp, blue crabs, and
other estuarine species.

In Ashumet and Johns Ponds, blooms of phytoplankton have changed the  color of the water and
depleted oxygen levels in the bottom waters  of the pond. Fish kills occurred in Ashumet Pond in
1985andl986.

-------
.The Massachusetts Military Reservation, a Superfund site within the watershed of Waquoit Bay,
is the source of several plumes of toxic chemicals that threaten drinking water supplies.
                                      \            "                •            "      !  -
As with many coastal areas where marine recreation is important, the number of boats and
request, for permits to build docks have increased in the Waquoit Bay area. Resuspended
sediments from boating activities, toxic chemicals from pressure treated wood in docks, propeller
scarring from boat motors, and shading of eelgrass beds from docks are all potential spurces of
stress to valuable marine resources.

Concern about the effects of development on Cape Cod have led to several initiatives. Among
these have been the creation of a regional planning agency, the Cape Cod Commission, that has
authority over developments of regional impact; the work of the Association for the Preservation
of Cape Cod, that has contributed to the protection of the Cape's drinking water supply, among
other issues; the efforts of the Waquoit Bay Land Margin Ecosystem Research Project, a multi-
institutional, interdisciplinary program that has contributed to our knowledge of the problem of
nitrogen overloading; the designation of a U.S. Fish and Wildlife Refuge in parts of the Waquoit
Bay watershed, which will remove many areas from development, the designation of the
Waquoit Bay area as an Area of Critical Environmental Concern, a Massachusetts designation
that provides for special scrutiny to any alterations that might impact natural resources, and the
designation of the Waquoit Bay National Estuarine Research Reserve, that also serves to protect
the resources of the Bay and its adjacent lands.

1.2 The Watershed Case Study Team
The EPA-sponsored ecological risk assessment underway in the Waquoit Bay watershed builds
on the above efforts by creating a mechanism to integrate the results of various research and
planning efforts into management options for local coastal decision-makers. The Waquoit Bay
watershed was selected as one of several EPA-sponsored ecological risk assessment case studies
because of interest by local, state, and federal organizations in the watershed, the type of
watershed (estuarine), the diversity of stressors (e.g., nutrients, toxic chemicals, obstructions,

-------
altered flow), a substantial existing database, and willingness by the Waquoit Bay National
Estuarine Research Reserve (WBNERR) and EPA Region I to lead the risk assessment team.

The prior activities and current and planned work of the risk assessment are described in the
following sections that emphasize the major elements of planning and problem formulation
(management goal development, selecting assessment endpoints, preparing a conceptual model,
and producing  an analysis plan) and a proposed risk characterization.

2. Problem Formulation
 2.1 Planning and the Selection of Management Goals and Objectives

The management goal was developed through a multistep planning process initiated and
completed by the team: a public meeting to initiate the process, evaluation of goals by interested
organizations in the watershed, and a meeting of members of these organizations to review and
approve the management goal and team-derived objectives. The management goal is a qualitative
statement that captures essential interests expressed by different management organizations and
the public hi the Waquoit Bay watershed. The goal developed for the Waquoit Bay watershed .
risk assessment through community involvement is:
 Reestablish and maintain -water quality and habitat conditions in Waquoit Bay and associated
wetlands, freshwater rivers, and ponds to (1) support diverse, self-sustaining commercial,
recreational, and native fish and shellfish populations and (2) reverse ongoing degradation of
ecological resources in the watershed.
In order for the management goal to support an ecological risk assessment, the goal was
evaluated by the team and interpreted as 10 management objectives believed to be required to
achieve the goal (see Table 1). These objectives were intended to state explicitly what kinds of
management results were implied in the general goal statement. By performing this kind of
evaluation, the team provided feedback to the managers on the ecological characteristics of the
goal, developed a systematic process for identifying assessment endpoints that could be directly

-------
linked to the management goal, and provided a way to measure achievement of the goal for risk
managers.
Table 1 is partitioned into three categories. The "Estuarine and Freshwater" category includes
three objectives that are common to both surface water types. Four objectives under the
"Estuarine" category and three objectives under the "Freshwater" category are unique to those
waters. The 10 objectives are stated as goal for specific aspects of exposure, stressors, and
valued ecological resources. Assessment endpoints were selected and justified based on these
objectives.  Although the goal was developed by the risk managers, the specific management
objectives were generated by the team based on available information on watershed resources.
The objectives were then provided to the risk managers for their consideration and approval.

-------
Table 1.  The Waquoit Bay Watershed Management Goal, Interpreted as 10 Management
Objectives.
Affected Area
Estuarine and
Freshwater
Estuarine
Freshwater
Number
1
2
3
4
5
6
7
8
9
10
Component Management Objective
Reduce or eliminate hypoxic or anoxic events
Prevent toxic levels of contamination in water, sediments, and biota
Restore and maintain self-sustaining native fish populations and their habitat
Reestablish viable eelgrass beds and associated aquatic communities in the
bay , .
Reestablish a self-sustaining scallop population in the bay that can support a
viable sport fishery
Protect shellfish beds from bacterial contamination that results in closure
Reduce or eliminate nuisance macroalgal growth
Prevent eutrophication of rivers and ponds
Maintain diversity of native biotic communities
Maintain diversity of water-dependent wildlife
2.2 Assessment Endpoints
Following the assessment of available information for the watershed, assessment endpoints were
selected that directly link management goals to measurable ecological values in the watershed.
Assessment endpoints are measurable attributes of valued resources identified by the
stakeholders that represent ecologically important components of the ecosystems. Assessment
endpoints include both an entity (e.g., eelgrass) and a measurable attribute (e.g., distribution),
and they provide direction for the assessment as well as a basis for the development of questions,
predictions, models, and analyses. The team selected eight assessment endpoints; the first seven
below represent ecological concerns about estuarine and freshwater components of the
ecosystem.
• Estuarine eelgrass habitat abundance ^tid distribution
• Resident and juvenile nursery estuarine finfish species diversity and abundance
• Estuarine benthic invertebrate diversity, abundance, and distribution
• Migratory (stream) fish reproduction
• Freshwater stream assemblages diversity  and abundance
• Freshwater pond trophic status

-------
• Wetlands habitat                                       .
• Barrier beach habitat          '

2.3 Conceptual Model                                                    *

Devised by the ecological risk assessment team with input from stakeholders, the general
watershed conceptual model (Fig. 1) is a broad representation of relationships among human
activities in the watershed (sources), the stressors believed to occur as a result of those sources,
and ecological effects likely to occur in each of the assessment endpoints. The pathways, from
sources of stressors to valued resources, are actually risk hypotheses that can be analyzed during
the ecological risk assessment process..

Because eelgrass is the foundation for the estuarine community and because its presence
indicates good water quality, it was targeted as a high priority assessment endpoint in this
ecological risk assessment (Fig.  1).

2.4    Analysis

Problem formulation concludes with the development of an analysis plan. For the Waquoit Bay
ecological risk assessment, the risk assessment team first conducted a comparative risk analysis
to help prioritize which stressors, assessment endpoints, and relationships should be examined
further. Once a focus for the assessment was selected, appropriate exposure and effects measures
and models were determined and the approaches to characterizing risks were described.

2.4.1 Comparative Risk Analysis

To help focus the risk assessment, the risk assessment team ranked stressors in terms of their
potential risk to all resources in the watershed, using a "fuzzy set" decision analysis method
based on best professional judgment (Harris et al., 1994). The analysis ranks the  stressors in
order of greatest overall contribution of risk to the endpoints, based on an ordinal effect of a

-------
stressor on that endpoint, ranging from no effect to severe effect.  For example, in Table 2, the
effect of nutrients on eelgrass habitat is given a 3 (severe indirect effect), but the effect of
physical alteration on eelgrass habitat is given a 1 (slight effect).
The results of the comparative analysis ranked nutrients as the primary stressor in the watershed,
followed by physical alteration of habitat, flow alteration, harvest pressure, resuspended
particulates, and toxic chemicals (Tables 2, 3).

-------
Table 2. Impact Matrix for the Waquoit Bay Watershed. Each cell represents the estimated
effect of a srressor on an endpoint, on an ordinal scale from O (no effect) to 3 (severe effect).
Stressors
Toxic
Chemicals
Altered Flow
Resuspended
Particulates '
Nutrients
Physical
Alteration
Harvest
Pressure
Assessment Endpoints
Migratory
Fish
.1
-3.
1
1
.1
2 '
Fresh-
water
Biota
, 1
2 .
1
1
1
1
Wetland
Habitat
,1
2
1
1
1
0
Pond
Trophic
Status
0
0
' 0 ,
3 .
0
0
Eelgrass
Habitat
0
0
1
- . '3
2
0
Estuarine
Inverte-
brates
. 1
0
1
2
1
2
Estuarine
Fish
1
1
1
• 2
1
2
Barrier
Beaches
0
0
0
0
2
0
Table 3. Stressor Rankings Based on Overall Effects on All Assessment Endpoints.
Stressors
Nutrients
Physical Alteration
of Habitat
Altered Flow
Toxic Chemicals
Harvest Pressure
Resuspended
Particulates
Unweighted
1
2 " '
3
4
- 5
6
Weighted for
Persistence
1
. • •- 2 .
3
4
• ' 5 "
6
Weighted for
Persistence and
Interaction
1
.2
3
4
5
6
The comparative analysis established that nutrients affected three assessment endpoints in the
estuarine system to different degrees: eelgrass habitat (severe effect), estuarine invertebrates

(moderate effect) and estuarine fish (moderate effect).  These assessment endpoints are

-------
interrelated because eelgrass meadows provide habitat to both estuarine fish and invertebrate
species. Therefore, protecting eelgrass will protect fish and invertebrate species.

The comparative analysis ranked other stressors to eelgrass in addition to nutrients: resuspended
particulates (minor effect) and physical alteration of habitat (moderate effect). The team
concluded that these stressors were not as important for reasons discussed below.

Although rivers enter Waquoit Bay, these do not carry a sediment load because rivers on Cape
Cod are fed by groundwater and are really drains for the aquifer; the particle size and
composition of the Cape's sandy glacial soils are such that any suspended particles sediment out;
and the sandy soils quickly absorb precipitation so there is very little surface runoff.

The resuspended particles in waters of Waquoit Bay are organic matter from decaying algae,
plants and other estuarine organisms. Studies of particle settling following passage of boats
whose motors disrupt the bottom show that the particles very quickly settle out. Although there
are many boats on the bay and adjacent tributaries and ponds on weekend, there is little boat
traffic during weekdays. Docks and marinas, where heavy boat use is expected, comprise only a
very small part of the surface area of the Waquoit Bay estuarine complex.

Physical alteration of habitat, due to activities such as shellfish harvesting, motor boat operation,
and construction of docks can fragment or eliminate eelgrass habitat. The number, frequency
and placement of these activities are such that deleterious effects would be restricted to a small
area of the overall estuarine complex.
2.4.2 Focus of Analysis Plan
The team concluded that reducing nutrient loads to restore water quality to conditions that would
support eelgrass growth was the most important stressor-endpoint relationship to evaluate and
that less critical stressors, such as resuspended particulates and physical alteration of habitat,
would be important to monitor and assess once water quality was unproved via reducing the
nutrient load.

-------
 Therefore the risk assessment team decided to focus its efforts on one stressor (nitrogen) and one
 assessment endpoint (eelgrass) based on the results of the comparative analysis and also on
 limitations of data and funding. Many other valued resources in the estuarine waters utilize
 eelgrass beds. For example, juvenile scallops attach to eelgrass blades, reducing their risk from
 predators. Winter flounder spawn in eelgrass meadows. The team believed that focusing on
 eelgrass distribution would encompass risks to other valued resources.

 Although it has been known for some time that nitrogen loading contributes to estuarine
 eutrophication and loss of submerged aquatic vegetation in Waquoit Bay and other estuaries of
 Cape Cod, predictive relationships between nitrogen sources and loading and the biological
 response of the estuary have not been developed for estuaries such as Waquoit Bay. The
 objective of this analysis is to develop a link between modeled estimates of nitrogen loading and
 predicted ecological effects in the estuary.

 The analysis plan to evaluate risk from nitrogen loading to eelgrass habitat involves 1) estimating
 the loading of nitrogen to the watershed and estuary (measures of exposure), and 2) evaluating
 how a given load of nitrogen directly or indirectly impacts eelgrass habitat (measures of effects).
 These analyses are performed on sub watersheds and their  adjacent estuaries that have
 experienced different degrees of development resulting in different amounts of nitrogen entering
 the estuaries.  Information about past and present land use is employed to forecast future changes
 in the estuary in response to future loads of nitrogen.

 2.4.2.1  Measures of Exposure

 2.4.2.1.1 Estimating the load of nitrogen from the watershed and subwatersheds

 The hypothesis underlying this part of analysis is that development on coastal watersheds
increases the amount of nitrogen entering coastal waters. On Cape Cod, the number of houses
has been positively related to the median amount of nitrate measured hi groundwater (Persky

-------
1986). Nitrogen, in ground-water eventually travels to receiving waters of the Waquoit Bay
estuarine complex.
                           f                     »             .'
The analysis reh'es on a nitrogen-loading model to .estimate the amount of nitrogen that arrives at
the edge of the estuary (Valiela et al. 1997).  The model sums all nutrient loads, subtracts losses
during transport and yields a value for nitrogen arriving at the edge of the estuary (or salt marsh).
The nitrogen loading model includes more than 50 input terms (e.g., number of houses, area in
agriculture, amount of nitrogen fertilizer applied to lawns, per capita contribution of nitrogen to
septic systems, percent loss of nitrogen in septic systems, percent loss of fertilizer nitrogen, etc).
                                                              '•                '
Many of the parameters in the nitrogen loading calculation are very uncertain.  For example, the
amount of nitrogen lost'in septic systems on sandy soils like those on Cape Cod ranges from 10-
90% (Valiela et al. 1997). Estimates of the contribution of dry deposition and of dissolved
organic nitrogen to the total amount of atmospheric nitrogen are also highly uncertain due to
limited sampling and analyses. Estimates of uncertainty surrounding model inputs and outputs
have been calculated (Collins et al. submitted) and will be applied to the final nitrogen loading
values.

Because groundwater travels approximately 100 meters per year in the watershed, there is a lag
between the time of development and the time that nitrogen arrives at the estuary. The nitrogen
loading model can also be run in  dynamic mode to determine the actual load of nitrogen arriving
at the estuary at any given tune (Fig. 2).
The nitrogen-loading model can also be run in static or dynamic mode using historic land use
information to hindcast nitrogen loading, and under a variety of future build-out scenarios to
predict future loading and effects.

Within the Waquoit Bay watershed are several subwatersheds that can, in turn, be divided into
recharge areas. The load of nitrogen can be estimated for the entire watershed or its component
parts.                                                    _

-------
The nitrogen loading model shows that atmospheric deposition, septic systems and fertilizer use
are the three major contributors to nitrogen overloading  (Table 4). Although more nitrogen is
delivered to the watershed from the atmosphere, much of that nitrogen is taken up by vegetation,
in soils, and in the aquifer during travel to the estuary. Septic systems are the largest source of
nitrogen to the estuary (Valiela etal. 1997). The relative contribution of-these three sources are
important to local coastal decision-makers since the source of most of the atmospheric nitrogen
is far outside the watershed.
Table 4. Estimates of Percent Nitrogen Loading from Atmosphere, Fertilizer and
Wastewater to Waquoit Bay.
Source
Atmospheric deposition
Septic system, ..
Fertilizers
Upper ponds
Percent to Watershed
56
,27
14
2
Percent to Estuary
30
48
15'
8
2.4.2.1.2. Validating the nitrogen loading model

Modeled predictions of the load of nitrogen to the edge of the estuary were validated in two
ways. First the model predictions were verified against actual measurements of nitrogen in
groundwater about to enter estuaries. As with model predictions, there is uncertainty associated
with the groundwater measures. Second, model predictions of wastewater nitrogen were
compared to stable isotopic ratios of nitrogen in groundwater. The predictions of nitrogen
coming from wastewater agreed with the valued derived from stable isotopes.
The nitrogen loading model predicts a concentration of nitrogen arriving at the edge of a salt
marsh (if present) or at the edge of the water, but a correction is necessary to estimate the amount
of nitrogen actually available to primary producers in the water. A biological process
(denitrification) that occurs within salt marshes can reduce the amount of nitrogen that finally

-------
enters the bay. Salt marsh areas and data on denitrification in Waquoit Bay sediments are used to
estimate the potential interception of land-derived nitrogen. These terms are applied as correction
terms to the model predictions. Water column nutrient and salinity data from different estuarine
reaches are used to estimate losses or gains of nitrogen in excess of dilution during down-estuary
transport.

The validated estimates of nitrogen from the watershed minus losses in marshes, sediments and
during travel down the river yield an amount of nitrogen available to the primary producers in the
estuary(ies).

2.4.2.3 Measures of Effects: Evaluating the Effects of Nitrogen on the Eelgrass Assessment
Endpoint                                       ,

As is shown in the conceptual model (Fig. 1), the deleterious effect of excess nitrogen on
eelgrass in shallow coastal bays is primarily an indirect one. Nitrogen stimulates the rapid
growth of phytoplankton (microscopic one-celled plants) and seaweed. Phytoplankton shade the
water column and seaweed grow over, shade and displace the eelgrass.  Eelgrass requires a lot of
light; thus the shading byphytoplankton and seaweed is the direct mechanism that reduces
eelgrass growth. Therefore, to analyze effects of nitrogen on eelgrass requires first estimating the
effects on algae growth and other intermediates. To these are added physical and temporal factors
of the estuarine system that affect nutrient availability and other aspects of plant growth.

The analysis utilizes an estuarine model to simulate the effects of nitrogen inputs,  water
residence time, mixing in the water, and seasonal changes in light and temperature on the  system
metabolism of phytoplankton, seaweed and  eelgrass. The model compares responses (especially
eelgrass decline) to different nitrogen loading rates across a variety of subestuaries. The
influence of any one subestuary on another,  or on the whole Waquoit Bay system, is assessed.
Model output is validated with data from estuaries not used in development of the model.

-------
 Increases in nitrogen change the mix of primary producers (plants such as eelgrass, seaweed and
 phytoplankton) in receiving waters.  The estuarine simulation model predicts the response of
 different producers to increasing nitrogen loads (Fig. 3).

 The nitrogen loading model and estuarine system model can be performed under a variety of
 nitrogen loading scenarios (e.g., build-out) in an attempt to hindcast and forecast loading and
 response. As new information becomes available during the analysis phase of the risk
 assessment, the models can be updated.

 3  Risk Characterization

 Research in Waquoit Bay and elsewhere suggests that development in coastal watersheds
 increases the amount of nitrogen entering coastal watersheds and their adjacent waters.  On Cape
 Cod, the nitrate concentration in groundwater is higher below developed landscapes than below
 naturally vegetated areas (Persky 1986). The nitrogen in groundwater travels to coastal bays
 where it fertilizes vegetation. Research shows that once in coastal bays nitrogen is rapidly taken
 up by some species of algae (phytoplankton and seaweed) increasing their growtti rates. These
 algae shade the water column so less light reaches the bottom.                    :

 Thus, increased loads of nitrogen from coastal development leads to overgrowth of opportunistic
 species of algae that alter the functioning of the estuarine system. These alterations include
 changes in water chemistry (e.g., dissolved oxygen concentration), habitat loss and changes in
 abundance of some species.

 3.1 Risk Characterization Elements
To define the stressor-respbnse relationship, nitrogen loading rates provided by the static and
dynamic loading models will be plotted against measures of ecological effects. Achieving low
nitrogen loading to Waquoit Bay will require nitrogen source control, as well as a sufficient tune
lag to allow nitrogen currently in the groundwater to be flushed out. The travel times of

-------
groundwater vary across the watershed, thus, nitrogen loading to the estuary is not a function of
land use at any one point in time. Using information about the groundwater travel tune and
location of houses in the watershed, the time to remove different percentages of nitrogen, and the
time it will take for the remaining nitrogen to travel to the estuary can be determined.  The two
sets of models and their estimated uncertainties can be used to predict the effects of different
                              •      '      '     '    '  '   , -                •
nutrient management scenarios for Waquoit Bay.

Septic systems and fertilizers are two local sources of nitrogen, but atmospheric deposition can
originate hundreds of miles from the Waquoit Bay watershed. Viable options to reduce nitrogen
to the extent necessary to improve water quality will depend on the relative contributions of
different nitrogen sources, the amount of nitrogen that needs to be  eliminated and the uncertainty
surrounding that estimate of the amount to be removed.

3.2 Integrating the exposure to nitrogen with the response by eelgrass.

Risk characterization will include integration of the output from the nitrogen loading models
with the predictions of the ecological response model (see section 2.4.2 for description of
individual models). The response of eelgrass to effects of increased nitrogen can be depicted as
inFig.4.

3.3 Predicting eelgrass recovery under different nitrogen loading scenarios

 Ecosystems are highly complex and variable systems that can and do over time change in
species composition, distribution and abundance.  Scientists working hi the waters of  Waquoit
Bay agree that nitrogen overloading is the major stressor on eelgrass and that decreasing the load
of nitrogen to the Bay may result in water quality  conditions that could support eelgrass, but
there is no certainty that eelgrass will reestablish itself or maintain itself if replanted.

 Predicting changes in water quality over tune that may result from a decrease in nitrogen loading
requires incorporating the travel tune of nitrogen hi groundwater.  With knowledge of the

-------
 location of houses and groundwater travel times, it is possible to .estimate how much nitrogen
 will be removed under different management scenarios and how much longer the rest of the
 nitrogen will remain in the aquifer traveling to the Bay. But that information alone is not
 sufficient to predict the time when water quality conditions would support eelgrass.  The
 contribution of benthic processes-and sediment conditions also must be factored into predictions.
 These parameters increase the uncertainty surrounding the .ability to estimate time for recovery.

 3.3.1 Output for examining model results and attendant management options

 A target load of nitrogen that will lead,to water quality conditions that support eelgrass can be
 identified for specific subembayments or the whole system. Figure 5 illustrates how such a
 relationship might be portrayed.  The probability that eelgrass might cover 10% or less of
 available habitat is plotted against nitrogen loading levels to the estuary. If the uncertainty  levels
 in these estimates are high, a curve with a shallow slope results.  If uncertainty is low, there is a
 closer relationship between eelgrass cover nitrogen loading exists and the slope of the curve will
 be steep. For a 25% probability that eelgrass habitat will be 10% or less of available habitat (or a
 75% chance of recovery), required nitrogen loading would be estimated to be much lower under
 conditions of high uncertainty.
                              ,                *
 3.4 Additional effects of other stressors on eelgrass

 If nitrogen is reduced, and eelgrass reestablishes itself or is replanted, other stressors may
 become important. For example, remaining small stands of eelgrass may be further impacted by
 natural events (e.g., in 1991, Hurricane Bob overwashed a spit on Washburn Island, burying an
 eelgrass bed on the inside of Eel Pond). Building docks over eelgrass beds, dredging activities,
 propeller scour from passing boats or mooring scars are all  stressors to eelgrass.

 4.  Other stressors affect valued resources in the Waquoit Bay Watershed

As funding permits, relationships among other stressors and valued resources will .be evaluated.

-------
Physical Alteration of Habitat. The loss of barrier beaches due to insufficient sand transport is
caused by armoring of the coastline well outside of the watershed and cannot be addressed in this
assessment.  Wetlands are an important habitat in the watershed. To date little work has been
done on wetlands loss; as results become available, they will be incorporated into the  final risk
assessment document if possible.

Altered Flow. The sandy soils of the Waquoif Bay watershed hold copious amounts  of water.
Future development could affect water quantity as. in many places on Cape Cod, the quality of
groundwater has been degraded due to development, so there are not as many possible well sites.
Other potential problems include loss of wetland function and loss of trout spawning habitat and
loss of alewife spawning habitat due to changes in flow. It is hoped that, as more research is
conducted, these issues can be addressed.

Toxic Chemicals.  The problem of toxic plumes emanating from the Superfund site are being
                           i    '           -•         '      ,   i       -     .        ~ *
evaluated by a large contingent of scientists and policy-makers. These stressors may  affect the
quality of Johns and Ashumet ponds, as well as freshwater and saltwater bodies downgradient.
As results become available, the risk assessment team will include their findings in the final
Waquoit Bay Ecological Risk Assessment product if possible.

Harvest pressure. Harvesting offish mainly occurs offshore and is beyond the scope of this
assessment.  Increased stress on valuable finfish populations comes from degraded estuarine
habitats where many of these offshore fish come to spawn.
                                    t                       '
Resuspended Participates. There is concern that boating, dredging and shellfishing activities
                                /                           ••*.,''
may resuspend sediments causing harm to valued resources.  These issues are under study at
Waquoit Bay by the National Estuarine Research Reserve Research Coordinator, Dr.  Richard
Crawford. Pertinent results will again be added to the risk assessment document if possible.
5. Literature Citations

-------
Collins, G., J. Kremer, and I. Valiela. submitted. Asssessing uncertainty in estimates of nitrogen
       loading to estuaries.
Harris, H. J., R. B. Wenger, V. A. Harris, and D. S, Devault. 1994. A,method for assessing
       environmental risk: a case study of Green Bay, lake Michigan, USA. Environmental
       Management 18(2) :295-3 06.                                                .
Persky, J. H.  1986. The relation of ground-water quality to housing density, Cape Cod,
       Massachusetts. Water Resources Investigation Report 86-4093. U.S Geological Survey.
       Marlborough, MA.
Valiela, I., G. Collins, J. Kremer, K. Lajtha, M. Geist, B. Seely, J. Brawley, and C. H. Sham.
       1997. Nitrogen loading from coastal watersheds to receiving estuaries: new method and
       application. Ecological Applications 7(2): 358-380.

List of Figures

            Figure 1.     Conceptual model of the Waquoit Bay watershed ecological risk
                         assessment.                                .
Figure 2.      Differences in arrival time of nitrogen between the static and dynamic nitrogen
              loading models. An example from the Jehu Pond subwatershed of the Waquoit
     .  ,       Bay watershed
Figure 3.      Changes in nitrogen loading alter the relative contribution of primary producers to
              total production in shallow estuaries.
Figure 4.      Hypothetical response of eelgrass to increases in nitrogen load.
Figure 5.      Hypothetical relationship between the probable extent of eelgrass habitat and
              nitrogen loading under high and low uncertainty.

List of Tables
Table 1.      The Waquoit Bay Watershed Management Goal, Interpreted as 10 Management
             Objectives.
Table 2.      Impact Matrix for the Waquoit Bay Watershed.

-------
Table 3.      Stressor Rankings Based on Overall Effects on All Assessment Endpoints.
Table 4.      Estimates of % Nitrogen Loading from Atmosphere, Fertilizer and Wastewater to
             WaquoitBay.             '

-------
The Waquoit Bay Watershed Ecological Risk Assessment Case Study:
Problem Formulation Summary and Proposed Risk Characterization
1. Introduction        .                .
This document includes summary information from the planning and problem formulation report
produced for the Waquoit Bay ecological risk assessment case study and a description of the
planned risk characterization component of the risk assessment.

The Waquoit Bay watershed ecological risk assessment was conducted to evaluate the danger to
valued water resources from stressors caused by human activities, and to provide resource
managers with viable options to protect the resources. A qualitative risk analysis identified
nitrogen-loading as a primary stressor in estuarine habitats of the watershed and eelgrass habitat
as the most important assessment endpoint. Because of these findings and due to constraints of
limited data to assess other endpoints, the risk assessment focused on the risk to eelgrass habitat
from nitrogen loading from the adjacent watershed.       ,

The goal of the Waquoit Bay ecological risk assessment is to provide managers with answers to
key questions.

       What are the sources of nutrients and then- relative contributions?
       What will be the effects of different degrees of nutrient reduction?

1.1 The Watershed

Waquoit Bay is a small estuary on the south coast of Cape Cod, Massachusetts. Its watershed
covers about 53 square kilometers (21 square miles) of freshwater streams and ponds, salt ponds
and marshes, pine and oak forest, barrier beaches, and open estuarine waters. The land and water
are home, spawning ground, and nursery for plant and animal life including piping plovers, least
terns (endangered birds), the sandplain gerardia (an endangered plant), alewife, winter flounder,
blue crab, scallops and clams, and other fish species that migrate through the estuary. Initially

-------
valued for hunting, farming, and fishing, Waquoit Bay now primarily provides aesthetic and
recreational opportunities, demands that have generated residential development and business for
local marine-dependent industries.
Cape Cod's economic viability is largely dependent on tourists who are drawn to the sandy
beaches, seafood restaurants, boating opportunities and water recreation areas. Thus the
economy on Cape Cod and the environment on Cape Cod are mutually inter-dependent. The
once rural surroundings have become increasingly suburbanized as bedroom and retirement
communities have sprung up. Barnstable County, where the Waquoit Bay watershed is located,
is the fastest growing county in Massachusetts. As the population grows, so does pressure on the
valuable natural resources that have attracted people to the area.
Living in bottom sediments of shallow embayments of the northwestern Atlantic is a flowering
plant known as eelgrass (Zostera marina). Numerous studies have shown that eelgrass meadows
provide a very good habitat for many commercially and recreationally important fish and
shellfish. Eelgrass needs a lot of light to grow.  In Waquoit Bay, increased phytoplankton
(microscopic one-celled organisms) and seaweed populations, fueled by the addition of nitrogen
from coastal development, have decreased the amount of light penetrating the water. In 1951
eelgrass meadows covered most of Waquoit Bay proper and its adjoining coastal ponds and
rivers.  Today, eelgrass is absent from the Bay proper and has declined significantly in the
adjoining tributaries and ponds. Species dependent on eelgrass, particularly scallops, have
likewise decreased. In 1987,1988, and 1990, fish kills occurred in Waquoit Bay, and  the
northern beach was covered with thousands of dead winter flounder, shrimp, blue crabs, and
other estuarine species.
In Ashumet and Johns Ponds, blooms of phytoplankton have changed the color of the water and
depleted oxygen levels in the bottom waters of the pond. Fish kills occurred in Ashumet Pond in
1985 and 1986.

-------
The Massachusetts Military Reservation, a Superftmd site within the watershed of Waquoit Bay,
'is the source of several plumes of toxic chemicals that threaten drinking water supplies.

As with many coastal areas where marine recreation is important, the number of boats and
request for permits to build docks have increased in the Waquoit Bay area. Resuspended
sediments from boating activities, toxic chemicals from pressure treated wood in docks, propeller
scarring from boat motors, and shading of eelgrass beds from docks are all potential sources of
stress to valuable marine resources.

Concern about the effects of development on Cape Cod have led to several initiatives. Among
these have been the creation of a regional planning agency, the Cape Cod Commission, that has
authority over developments of regional impact; the work of me Association for the Preservation
of Cape Cod, that has contributed to the protection of the Cape's drinking water supply, among
other issues; the efforts of the Waquoit Bay L,and Margin Ecosystem Research Project, a multi-
institutional, interdisciplinary program that has contributed to our knowledge of the problem of
nitrogen overloading; the designation of a U.S. Fish and Wildlife Refuge in parts of the Waquoit
Bay watershed, which will remove many areas from development, the designation of the
Waquoit Bay area as an Area of Critical Environmental Concern, a Massachusetts designation
that provides for  special scrutiny to any alterations that might impact natural resources, and the
designation of the Waquoit Bay National Estuarine Research Reserve, that also serves to protect
the resources of the Bay and its adjacent lands.

1.2 The Watershed Case Study Team

The EPA-sponsbred ecological risk assessment underway in the Waquoit Bay watershed builds
on the above efforts by creating a mechanism to integrate the results of various research and
planning efforts into management options for local coastal decision-makers.  The Waquoit Bay
watershed was selected as one  of several EPA-sponsored ecological risk assessment case studies
because of interest by local, state, and federal organizations in the watershed, the type of
watershed (estuarine), the diversity of stressors (e.g., nutrients, toxic chemicals, obstructions,

-------
altered flow), a substantial existing database, and willingness by the Waquoit Bay National
Estuarine Research Reserve (WBNERR) and EPA Region I to lead the risk assessment team.

The prior activities and current and planned work of the risk assessment are described in the
following sections that emphasize the major elements of planning and problem formulation
(management goal development, selecting assessment endpoints, preparing a conceptual model,
and producing  an analysis plan) and a proposed risk characterization.

2. Problem Formulation
 2.1 Planning and the Selection of Management Goals and Objectives

The management goal was developed through a multistep planning process initiated and
completed by the team: a public meeting to initiate the process, evaluation of goals by interested
organizations in the watershed, and a meeting of members of these organizations to review and
approve the management goal and team-derived objectives. The management goal is a qualitative
statement that captures essential interests expressed by different management organizations and
the public in the Waquoit Bay watershed. The goal developed for the Waquoit Bay watershed
risk assessment through community involvement is:
 Reestablish and maintain -water quality and habitat conditions in Waquoit Bay and associated
•wetlands, freshwater rivers, and ponds to (1) support diverse, self-sustaining commercial,
recreational, and native fish and shellfish populations and (2) reverse ongoing degradation of
ecological resources in the -watershed.
In order for the management goal to support an ecological risk assessment, the goal was
evaluated by the team and interpreted as 10 management objectives believed to be required to
achieve the goal (see Table 1). These objectives were intended to state explicitly what kinds of
management results were implied in the general goal statement. By performing this kind of
evaluation, the team provided feedback to the managers on the ecological characteristics of the
goal, developed a systematic process for identifying assessment endpoints that could be directly

-------
linked to the management goal, and provided a way to measure achievement of the goal for risk
managers.
Table 1 is partitioned into three categories. The "Estuarine and Freshwater" category includes
three objectives that are common to both surface water types. Four objectives under the
"Estuarine" category and three objectives under the "Freshwater" category are unique to those
waters. The 10 objectives are stated as goal for specific aspects of exposure, stressors, and
valued ecological resources. Assessment endpoints were selected and justified based on these
objectives. Although the goal was developed by the risk managers, the specific management
objectives were generated by the team based on available information on watershed resources.
The objectives were then provided to the risk managers for their consideration and approval.

-------
Table 1. The Waquoit Bay Watershed Management Goal, Interpreted as 10 Management
Objectives.
Affected Area
Estuarine and
Freshwater
Estuarine
Freshwater
Number
1
2
3
4
5
6
7
8
9
10
Component Management Objective
Reduce or eliminate hypoxic or anoxic events
Prevent toxic levels of contamination in water, sediments, and biota
Restore and maintain self-sustaining native fish populations and .their habitat
Reestablish viable eelgrass beds and associated aquatic communities in the
bay
Reestablish a self-sustaining scallop population hi the bay that can support a
viable sport fishery
Protect shellfish beds from bacterial contamination that results hi closure
Reduce or eliminate nuisance macroalgal growth
Prevent eutrophication of rivers and ponds
Maintain diversity of native biotic communities
Maintain diversity of water-dependent wildlife
2.2 Assessment Endpoints
Following the assessment of available information for the watershed, assessment endpoints were
selected that directly link management goals to measurable ecological values in the watershed.
Assessment endpoints are measurable attributes of valued resources identified by the
stakeholders that represent ecologically important components of the ecosystems.  Assessment
endpoints include both an entity (e.g., eelgrass) and a measurable attribute (e.g.,  distribution),
and they provide direction for the assessment as well as a basis for the development of questions,
predictions, models, and analyses. The team selected eight assessment endpoints; the first seven
below represent ecological concerns about estuarine and freshwater components of the
ecosystem.
• Estuarine eelgrass habitat abundance and distribution
• Resident and juvenile nursery estuarine finfish species diversity and abundance
• Estuarine benthic invertebrate diversity, abundance, and distribution
• Migratory (stream) fish reproduction
• Freshwater stream assemblages diversity and abundance
• Freshwater pond trophic status

-------
• Wetlands habitat
; Barrier beach habitat

2.3 Conceptual Model         x

Devised by the ecological, risk assessment team with input from stakeholders, the general
watershed conceptual model (Fig. 1) is a broad representation of relationships among human
activities in the watershed (sources), the stressors believed to occur as a result of those sources,
and ecological effects likely to occur in each of the assessment endpoints. The pathways, from
sources of stressors to valued resources, are actually risk hypotheses that can be analyzed during
the ecological risk assessment process.
                                         \     - '                                       ,
Because eelgrass is the foundation for the estuarine community and because its presence
indicates good water quality, it was targeted as a high priority assessment endpoint in this
ecological risk assessment (Fig. 1).

2.4    Analysis

Problem formulation concludes with the development of an analysis plan. For the Waquoit Bay
ecological risk assessment, the risk assessment team first conducted a comparative risk analysis
to help prioritize which stressors, assessment endpoints, and relationships should be examined
further. Once a focus for the assessment was selected, appropriate exposure and effects measures
and models were determined and the approaches to characterizing risks were described.

2.4.1 Comparative Risk Analysis
To help focus the risk assessment, the risk assessment team ranked stressors in terms of their
potential risk to all resources in the watershed, using a "fuzzy set" decision analysis method
based on best professional judgment (Harris et al., 1994). The analysis ranks the stressors in
order of greatest overall contribution of risk to the endpoints, based on an ordinal effect of a

-------
stressor on that endpoint, ranging from no effect to severe effect. For example, in Table 2, the
effect of nutrients on eelgrass habitat is given a 3 (severe indirect effect), but the effect of
physical alteration on eelgrass habitat is given a 1 (slight effect).
The results of the comparative analysis ranked nutrients as the primary stressor in the watershed,
followed by physical alteration of habitat, flow alteration, harvest pressure, resuspended
particulates, and toxic chemicals (Tables 2,3).

-------
Table 2. Impact Matrix for the Waquoit Bay Watershed.  Each cell represents the estimated
effect of a stressor on an endpoint, on an ordinal scale from O (no effect) to 3 (severe effect).
Stressors
Toxic
Chemicals
Altered Flow
Resuspended
Particulates
Nutrients
Physical
Alteration
Harvest
Pressure
Assessment Endpoints
Migratory
Fish
1
3
1
1
1
2
Fresh-
water
Biota
1 '
2
1
1
1
1
Wetland
Habitat
1
2
1
1 '
1
0
Pond
Trophic
Status
0
0
0
3
o
0
Eelgrass
Habitat
0
0
1
3 '
2
0
Estuarine
Inverte-
brates
1
0
1
2
1
2
Estuarine
Fish
1
1
1
2
1
. 2
Barrier
Beaches
0
0
0
0
2
0
Table 3. Stressor Rankings Based on Overall Effects on All Assessment Endpoints.
Stressors
« ' "
Nutrients
Physical Alteration
of Habitat
Altered Flow
Toxic Chemicals
Harvest Pressure
Resuspended
Particulates
Unweighted
1-
2
3
4
5
6
- • .
Weighted for
Persistence
1
2
- 3
4
• 5
6
Weighted for
Persistence and
Interaction
1
2
3
4
'.. • 5
6
The comparative analysis established that nutrients affected three assessment endpoints in the

estuarine system to different degrees: eelgrass habitat (severe effect), estuarine invertebrates

(moderate effect) and estuarine fish (moderate effect). These assessment endpoints are

-------
 interrelated because eelgrass meadows provide-habitat to both estuarine fish and invertebrate
 species. Therefore, protecting eelgrass will protect fish and invertebrate species.

 The comparative analysis ranked other stressors to eelgrass in addition to nutrients: resuspended
 particulates (minor effect) and physical alteration of habitat (moderate effect). The team
 concluded that these stressors were not as important for reasons discussed below.

 Although rivers enter Waquoit Bay, these do not carry a sediment load because rivers on Cape
 Cod are fed by groundwater and are really drains for the aquifer; the particle size and
 composition of the Cape's sandy glacial soils are such that any suspended particles sediment out;
 and the sandy soils quickly absorb precipitation so there is very little surface runoff.

 The resuspended particles in waters of Waquoit Bay are organic matter from decaying algae,
 plants and other estuarine organisms.  Studies of particle settling following passage of boats
 whose motors disrupt the bottom show that the particles very quickly settle out.  Although there
 are many boats on the bay and adjacent tributaries and ponds on weekend, there is little boat
 traffic during weekdays. Docks and marinas, where heavy boat use is expected, comprise only a
 very small part of the surface area of the Waquoit Bay estuarine complex.

 Physical alteration of habitat, due to activities such as shellfish harvesting, motor boat operation,
 and construction of docks can fragment or eliminate eelgrass habitat. The number, frequency
 and placement of these activities are such that deleterious effects would be restricted to a small
 area of the overall estuarine complex.
2.4.2 Focus of Analysis Plan
The team concluded that reducing nutrient loads to restore water quality to conditions that would
support eelgrass growth was the most important stressor-endpoint relationship to evaluate and
that less critical stressors, such as resuspended particulates and physical alteration of habitat,
would be important to monitor and assess once water quality was improved via reducing the
nutrient load.

-------
Therefore the risk assessment team decided to focus its efforts on one stressor (nitrogen) and one
assessment endpoint (eelgrass) based on the results of the comparative analysis and also on
limitations of data and funding. Many other valued resources in the estuarine waters utilize
eelgrass beds.  For example, juvenile scallops attach to eelgrass blades, reducing their risk from
predators. Winter flounder spawn in eelgrass meadows.  The team believed that focusing on
eelgrass distribution would encompass risks to other valued resources.

Although it has been known for some time that nitrogen loading contributes to estuarine
eutrophication and loss of submerged  aquatic vegetation in Waquoit Bay and other estuaries of
Cape Cod, predictive relationships between nitrogen sources and loading and the biological
response of the estuary have not been  developed for estuaries such as Waquoit Bay. The
objective of this analysis is to develop a link between modeled estimates of nitrogen loading and
predicted ecological effects in the estuary.

The analysis plan to evaluate risk from nitrogen loading to eelgrass habitat involves 1) estimating
the loading of nitrogen to the watershed and estuary (measures of exposure), and 2) evaluating
how a given load of nitrogen directly or indirectly impacts eelgrass habitat (measures of effects).
These analyses are performed on subwatersheds arid their adjacent estuaries that have
experienced different degrees of development resulting in different amounts of nitrogen entering
the estuaries.  Information about past arid present land use is employed to forecast future changes
in the estuary in response to future loads of nitrogen.

2.4.2.1 Measures of Exposure

2.4.2.1.1 Estimating the load of nitrogen from the watershed and subwatersheds
The hypothesis underlying this part of analysis is that development on coastal watersheds
increases the amount of nitrogen entering coastal waters. On Cape Cod, the number of houses
has been positively related to the median amount of nitrate measured in groundwater (Persky

-------
1986). Nitrogen in ground-water eventually travels to receiving waters of the Waquoit Bay
estuarine complex.

The analysis relies on a nitrogen-loading model to estimate the amount of nitrogen that arrives at
the edge of the estuary (Valiela et al. 1997). The model sums all nutrient loads, subtracts losses
during transport and yields a value for nitrogen arriving at the edge of the estuary (or salt marsh).
The nitrogen loading model includes more than 50 input terms (e.g., number of houses, area in
agriculture, amount of nitrogen fertilizer applied to lawns, per capita contribution of nitrogen to
septic systems, percent loss of nitrogen in septic systems, percent loss of fertilizer nitrogen, etc).

Many of the parameters in the nitrogen loading calculation are very uncertain.  For example, the
amount of nitrogen lost in septic systems on sandy soils like those on  Cape Cod ranges from 10-
90% (Valiela etal. 1997). Estimates of the contribution of dry deposition and of dissolved
organic nitrogen to the total amount of atmospheric nitrogen are also highly uncertain due to
limited sampling and analyses. Estimates of uncertainty surrouriding  model inputs and outputs
have been calculated (Collins et al.  submitted) and will be applied to the final nitrogen loading
values.

Because groundwater travels approximately 100 meters per year in the watershed, there is a lag
between the time of development and the time that nitrogen arrives at the estuary. The nitrogen
loading model can also be run in dynamic mode to determine the actual load of nitrogen arriving
at the estuary at any given tune (Fig. 2).

The nitrogen-loading model can also be run in static or dynamic mode using historic land use
information to hindcast nitrogen loading, and under a variety of future build-out scenarios to
predict future loading and effects.
Within the Waquoit Bay watershed are several subwatersheds that can, in turn, be divided into
recharge areas. The load of nitrogen can be estimated for the entire watershed or its component
parts.

-------
The nitrogen loading model shows that atmospheric deposition, septic systems and fertilizer use
are the three major contributors to nitrogen overloading (Table 4). Although more nitrogen is
delivered to the watershed from the atmosphere, much of that nitrogen is taken up by vegetation,
in soils, and in the aquifer during travel to the estuary. Septic systems are the largest source of
nitrogen to the estuary (Valiela et al. 1997). The relative contribution of these three sources are
important to local coastal decision-makers since the source of most of the atmospheric nitrogen
is far outside the watershed.
Table 4. Estimates of Percent Nitrogen Loading from Atmosphere, Fertilizer and
Wastewater to Waquoit Bay.
Source
Atmospheric deposition
Septic system
Fertilizers
Upper ponds
Percent to Watershed
56
27 ,
14 ,
2
Percent to Estuary
30
48
15
8
2.4.2.1.2. Validating the nitrogen loading model

Modeled predictions of the load of nitrogen to the edge of the estuary were validated in two
ways. First the model predictions were verified against actual measurements of nitrogen in
groundwater about to enter estuaries. As with model predictions, there is uncertainty associated
with the groundwater measures. Second, model predictions of wastewater nitrogen were
compared to stable isotopic ratios of nitrogen in groundwater. The predictions of nitrogen
coming from wastewater agreed with the valued derived from stable isotopes.
The nitrogen loading model predicts a concentration of nitrogen arriving at the edge of a salt
marsh (if present) or at the edge of the water, but a correction is necessary to estimate the amount
of nitrogen actually available to primary producers in the water. A biological process
(denitrification) that occurs within salt marshes can reduce the amount of nitrogen that finally

-------
enters the bay. Salt marsh areas and data on denitriflcation in Waquoit Bay sediments are used to
estimate the potential interception of land-derived nitrogen. These terms are applied as correction
terms to the model predictions. Water column nutrient and salinity data from different estuarine
reaches are used to estimate losses or gains of nitrogen in excess of dilution during down-estuary
transport.

The validated estimates of nitrogen from the watershed minus losses in marshes, sediments and
during travel down the river yield an amount of nitrogen available to the primary producers in the
estuary(ies).

2.4.2.3 Measures of Effects: Evaluating the Effects of Nitrogen on the Eelgrass Assessment
Endpoint

As is shown in the conceptual model (Fig. 1), the deleterious effect of excess nitrogen on
eelgrass in shallow coastal bays is primarily an indirect one.  Nitrogen stimulates the rapid
growth of phytoplankton (microscopic one-celled plants) and seaweed. Phytoplankton shade the
water column and seaweed grow over, shade and displace the eelgrass.  Eelgrass requires a lot of
light; thus the shading by phytoplankton and seaweed is the direct mechanism that reduces
eelgrass growth. Therefore, to analyze effects of nitrogen on eelgrass requires first estimating the
effects on algae growth and other intermediates. To these are added physical and temporal factors
of the estuarine system that affect nutrient availability and other aspects of plant growth.
The analysis utilizes an estuarine model to simulate the effects of nitrogen inputs, water
residence time, mixing in the water, and seasonal changes in light and temperature on the system
metabolism of phytoplankton, seaweed and eelgrass. The model compares responses (especially
eelgrass decline) to different nitrogen loading rates across a variety of subestuaries. The
influence of any one subestuary on another, or on the whole Waquoit Bay system, is assessed.
Model output is validated with data from estuaries not used in development of the model.

-------
 Increases in nitrogen change the mix of primary producers (plants such as eelgrass, seaweed and
 phytoplankton) hi receiving waters. The estuarine simulation model predicts the response of
 different producers to increasing nitrogen loads (Fig. 3).       ,

 The nitrogen loading model and estuarine system model can be performed under a variety of
 nitrogen loading scenarios (e.g., build-out) in an attempt to hindcast and forecast loading and
 response. As new information becomes available during the analysis phase of the risk
 assessment, the models can be updated.
*                               r               '
                                                                             j          i
 3 Risk Characterization

 Research in Waquoit Bay arid elsewhere suggests that development in coastal watersheds
 increases the amount of nitrogen entering coastal watersheds and their adjacent waters.  On Cape
 Cod, the nitrate concentration hi groundwater is higher below developed landscapes than below
 naturally vegetated areas (Persky 1986). The nitrogen in groundwater travels to coastal bays
 where it fertilizes vegetation. Research shows that once in coastal bays nitrogen is rapidly taken
 up by some species of algae (phytoplankton and seaweed) increasing then- growth rates.  These
 algae shade the water column so less light reaches the bottom.

Thus, increased loads of nitrogen from coastal development leads to overgrowth of opportunistic
 species of algae that alter the functioning of the estuarine system. These alterations include
 changes  in water chemistry (e.g., dissolved oxygen concentration), habitat loss and changes hi
 abundance of some species.
                                                                            x
 3.1 Risk Characterization Elements
To define the stressor-response relationship, nitrogen loading rates provided by the static and
dynamic loading models will be plotted against measures of ecological effects. Achieving low
nitrogen loading to Waquoit Bay will require nitrogen source control, as well as a sufficient time
lag to allow nitrogen currently hi the groundwater to be flushed out.  The travel times of

-------
groundwater vary across the watershed, thus, nitrogen loading to the estuary is riot a function of
land use at any one point in time. Using information about the groundwater travel time and
location of houses in the watershed, the time to remove different percentages of nitrogen, and the
time it will take for the remaining nitrogen to travel to the estuary can be determined.  The two
sets of models and their estimated uncertainties can be used to predict the effects of different
nutrient management scenarios for Waquoit Bay.

Septic systems and fertilizers are two local sources of nitrogen, but atmospheric deposition can
originate hundreds of miles from the Waquoit Bay watershed. Viable options to reduce nitrogen
to the extent necessary to improve water quality will depend on the relative contributions of
different nitrogen sources, the amount of nitrogen that needs to be eliminated and the uncertainty
surrounding that estimate of the amount to be removed.
                               v'                         -                              •: ,
                      i
3.2 Integrating the exposure to nitrogen with the response by eelgrass.

Risk characterization will include integration of the output from the nitrogen loading models
with the predictions of the ecological response model (see section 2.4.2 for description of
individual models). The response of eelgrass to effects of increased nitrogen can be depicted as
inFig.4.

3.3 Predicting eelgrass recovery under different nitrogen loading scenarios
                              r            •    .               •            •>
                              i

 Ecosystems are highly complex and variable systems that can and do over time change in
species composition, distribution and abundance.  Scientists working hi the waters of Waquoit
Bay agree that nitrogen overloading is the major stressor on eelgrass and that decreasing the load
of nitrogen to the Bay may result in water quality conditions that could support eelgrass, but
there is no certainty that eelgrass will reestablish itself or maintain itself if replanted.
 Predicting changes hi water quality over tune that may result from a decrease in nitrogen loading
requires incorporating the travel time of nitrogen in groundwater. With knowledge of the

-------
location of houses and groundwater travel times,  it is possible to estimate how much nitrogen
will be removed under different management scenarios and how much longer the rest of the
nitrogen will remain in the aquifer traveling to the Bay.  But that information alone is not
sufficient to predict the time when water quality conditions would support eelgrass. The
contribution of benthic processes and sediment conditions also must be factored into predictions.
These parameters increase the uncertainty surrounding the ability to estimate time for recovery.

3.3.1  Output for examining model results and attendant management options

A target load of nitrogen that will lead to water quality conditions that support eelgrass can be
identified for specific subembayments or the whole system. Figure 5 illustrates how such a
relationship might be portrayed. The probability that eelgrass might cover 10% or  less of
available habitat is plotted against nitrogen loading levels to the estuary.  If the uncertainty levels
hi these estimates are high, a curve with a shallow slope results.  If uncertainty is low, there is a
closer relationship between eelgrass cover nitrogen loading exists and the slope of the curve will
be steep. For a 25% probability that eelgrass habitat will be 10% or less of available habitat (or a
75% chance of recovery),,required nitrogen loading would be estimated to be much lower under
conditions of high uncertainty.                       ,

3.4 Additional effects of other stressors  on eelgrass
           '                       •         '  ,                      ' '        .
If nitrogen is reduced, and eelgrass reestablishes itself or is replanted, other stressors may
become important. For example, remaining small stands of eelgrass may be further impacted by
natural events (e.g., in 1991, Hurricane Bob overwashed a spit on Washburn Island, burying an
eelgrass bed on the inside of Eel Pond). Building docks over eelgrass beds, dredging activities,
propeller scour from passing boats or mooring scars are all stressors to eelgrass.

4. Other stressors affect valued resources in the Waquoit Bay Watershed
As funding permits, relationships among other stressors and valued resources will be evaluated.

-------
Physical Alteration of Habitat. The loss of barrier beaches due to insufficient sand transport is
caused by armoring of the coastline well outside of the watershed and cannot be addressed in this
assessment.  Wetlands are an important habitat in the watershed. To date little work has been
done on wetlands loss; as results become available, they will be incorporated into the final risk
assessment document if possible.

Altered Flow. The sandy soils of the Waquoit Bay watershed hold copious amounts of water.
Future development could affect water quantity as in many places on Cape Cod, the quality of
groundwater has been degraded due to development, so there are not as many possible well sites.
Other potential problems include loss of wetland function and loss of trout spawning habitat and
loss of alewife spawning habitat due to changes in flow. It is hoped that, as more research is
conducted, these issues can be addressed.

Toxic Chemicals.  The problem of toxic plumes emanating from the Superfund site are being
evaluated by a large contingent of scientists and policy-makers. These stressors may affect the
quality of Johns and Ashumet ponds, as well as freshwater and saltwater bodies downgradient.
As results become available, the risk assessment team will include their  findings hi the final
Waquoit Bay Ecological Risk Assessment productif possible.

Harvest pressure. Harvesting offish mainly occurs offshore and is beyond the scope of this
assessment.  Increased stress on valuable finfish populations comes from degraded estuarine
habitats where many of these offshore fish come to spawn.

Resuspended Participates. There is concern that boating, dredging and shellfishing activities
may resuspend sediments causing harm to valued resources.  These issues are under study at
Waquoit Bay by the National Estuarine Research Reserve Research Coordinator, Dr. Richard
Crawford. Pertinent results will again be added to the risk assessment document if possible.
5. Literature Citations

-------
 Collins, G,, J. Kremer, and I. Valiela. submitted. Asssessing uncertainty in estimates of nitrogen
       loading to estuaries.                :
         I                         .'              •               ':               •        -
 Harris, H. J., R. B. Wenger, V. A. Harris, and D. S. Devault. 1994. A method for assessing
       environmental risk: a case study of Green Bay, lake Michigan, USA. Environmental
       Management 18(2):295-306.
 Persky, J. H. 1986. The relation of ground-water quality to housing density, Cape Cod,
       Massachusetts. Water Resources Investigation Report 86-4093. U.S Geological Survey.
       Marlborough, MA.                             ,
 Valiela, I, G. Collins, J. Kremer, K. Lajtha, M. Geist, B. Seely, J. Brawley, and C. H. Sham.
       1997. Nitrogen loading from coastal watersheds to receiving estuaries: new method and
       application.  Ecological Applications'7(2): 358-380.

 List of Figures             .
        . "1          .'                     !                 ' .             .
             Figure 1.     Conceptual model of the Waquoit Bay watershed ecological risk
       •  ' ,   •             assessment.
 Figure 2.     Differences in arrival time of nitrogen between the static and dynamic nitrogen
             loading models. An example from the Jehu Pond subwatershed of the Waquoit
             Bay watershed                                                      ,
 Figure 3.     Changes in nitrogen loading alter the relative contribution of primary producers to
             total production in shallow estuaries.
 Figure 4. '   Hypothetical response of eelgrass to increases in nitrogen load.
 Figure 5.     Hypothetical relationship between the probable extent of eelgrass habitat and '
             nitrogen loading under high and low uncertainty.

List of Tables        »
Table 1.      The Waquoit Bay Watershed Management Goal, Interpreted as 10 Management
             Objectives.
Table 2,      Impact Matrix for the Waquoit Bay Watershed.

-------
Table 3.      Stressor Rankings Based on Overall Effects on All Assessment Endpoints.
Table 4.      Estimates of % Nitrogen Loading from Atmosphere, Fertilizer and Wastewater to
             WaquoitBay.

-------
T3




 co "
 1
 c
 o
 I
 s
 as
 CO
•a

•a
 co




 I


 £

CQ
 o

 I
 o
.S3
 O

O
     E
     O
     co
     CO

     g

    ts

-------
 I
 1
 en

 1

 I
 §
 1
E. e

-------
6000 _
5000 i
4000 _

                                                                            Total
2000 _
1000 -r
  . 0 -
   1850  1860  1870 1880 1890 1900 1910.1920  1930  1940  1950  1960  1970  1980  1990  ,2000 2010 2020 2030 2040 2050
                    .       .".•'    .    "    •  '    •    .   Year .                   ...•''              '
         Figure 2. Differences in arrival time of nitrogen between the static and dynamic nitrogen loading
         models. An example from the Jehu Pond subwatershed of the Waquoit Bay watershed.

-------
           o>


           o

          U
          .SP
          "3

          «

          v
          -u
          o
          PH
                                    Nitrogen Load
Figure 4. Hypotiietical response of eelgrass to increases in nitrogen load.

-------
     .js ^   ipo%-
     si   '


   o
.!
— ; O
   7
   V
       75%-
            50%-
            25%
Target load
for 75% chance,
high uncertainty
                           Low Uncertainty
                                                             High Uncertainty
                                                         1000    Hypothetical
                   Target load
                   for 75% chance,
                   low uncertainly
                                       Nitrogen Load
                                                                 Current Load
Figure 5.  Hypothetical relationship between uie probable extent of eelgrass habitat and nitrogen
loading under high and low uncertainty (J. Gerritsen, Pers. Com.)- See text for explanation.

-------

-------

-------

-------
             Charge for the Mitec Risk Characterization Case Study
                                              )'•-'"
 Risk Characterization Issue for Peer Review

       The adequacy of any risk characterization depends in the first instance on the reliability
 and credibility of the scientific and technical data and analyses in the risk assessment. At the
 Same time, it is equally important that other risk assessors, decision-makers and risk managers,
 and the public fully understand and appreciate the strengths and limitations of the assessment;
 that is, the overall scientific "character" of the results.  To test this aspect of risk characterization,
 EPA's Risk Characterization Policy and the draft Handbook set forth requirements for
 transparency, clarity, consistency, and reasonableness (TCCR), and offer guidance on developing
 and evaluating this aspect of the characterization of risk.

       This peer review does not focus on the underlying scientific analyses, but on whether the
 presentation of the data and analyses adequately represents the risk assessment results.

 Case Study Context and Background

       Materials for the Mitec case study include both a question and answer document and
 preliminary risk estimates for exposure to a food use pesticide used on field, fruit, nut, and
 vegetable crops. The question and answer (Q & A) document presents to the general public the
 Agency's perspective on its decision to register the pesticide. .

       The preliminary risk characterization document is divided into 5 sections. Section 1
 provides background and an introduction to the case study.  Section 2 provides the planning and
 scoping which identifies the exposure scenarios.  Section 3 presents the hazard and dose response
 assessment of Mitec.  Section 4 presents the multi-pathway exposure assessments.  Section 5
 presents the risk characterization, which is partitioned into sections of dietary and occupational
 risks with a description of the limitations of the analysis and its conclusions.

       Apples, grapes, oranges, almonds, walnuts, cotton, field corn, and mint comprise 80% of
 Mitec usage in the US. In all risk analyses there are a range of mixes of choices that can be
 made. Of those that we consider reasonable for Mitec, occupational exposures give both the
 highest and lowest estimates of risk to workers.  The preliminary document presents OPP's
judgement for the best mix of choices made in this assessment of the estimated risk for dietary
 cancer among children which was a value judgement based upon the risk estimates calculated,
 policy considerations for occupational risks including its voluntary nature, and differences in
 data quality between dietary and worker exposure data.  The total dietary risk from all published
 agricultural uses of Mitec is 1.6 X 10'5 with exposure to children from apples, peaches and
 grapes (2.5 x  10"7 to 9.1 x 10'6) resulting in the major source of risk concern. Occupational
 cancer risks (skin exposure) ranged from 10'7 to 10"4 with high risks estimated for wettable
 powders on grapes (open and closed cabs) and commercial mixer/loaders (open mix system).
 Occupational cancer risks in the range of lO^4 to 10"6 are acceptable based upon the OPP Cancer

-------
                                           2                                       Mitec

Worker Risk Policy. Noncancer risks (Margin of Exposures, MOEs) were of low concern based
upon limited exposure and minimal systemic toxicity.  Confidence in the basis for a dietary
cancer risk and for exposure is high. Occupational exposure estimates are considered
preliminary.

Points for Peer Reviewers to Address

1)     This case offers two different risk characterization products to meet the expected needs of
       two different audiences. One is a document written for the Office of Pesticide Programs
       risk managers and the other is written as questions and answers to address concerns of the
       general public. Using the criteria set forth in the draft Handbook, please evaluate the
       Mitec case in terms of the principles of transparency, clarity, consistency, and
       reasonableness.                          .

2)     Please comment on the relative suitability of the two different forms of risk
       characterization for the two different audiences.

3)     Please comment on whether the risk characterization illustrates a transparent, clear
       analysis of uncertainty and variability for the OPP risk manager. That is, EPA is
       interested in your comments on the extent to which the characterization conveys a sense
       of transparency and clarity  with regard to the discussions on the uncertainties and
       variabilities which surround the estimations of cancer and noncancer risk.

4)     Please comment on the way the choices in the assessment are presented and the way the
       risk .estimates are arrived.

5)     Please comment on the discussion of how the most significant parameters affect the risk
       assessment outcome in the preliminary document.

6)     Please comment on the characterization of the alternative approaches to the assessment
       and if there are other plausible ways to characterize them.

7)     EPA welcomes any additional comments or suggestions.

-------
  RISK CHARACTERIZATION CASE STUDY
                                     DO NOT QUOTE OR CITE

                                               MITEC
Risk Conclusions and Comparisons: This assessment of Mitec is the product of many choices in the assumptions used
to deal with the uncertainties.  In all risk analyses there are a range of mixes of choices that can be made. Of those that
we consider reasonable for Mitec, occupational exposures give both the largest and least estimates of risk to workers.
Our judgement for the best mix of choices made in this assessment of the estimated risk is for dietary cancer among
children, which was a value judgement based upon the risk estimates calculated, policy considerations for occupational
risks including its voluntary nature and differences in data quality between dietary and worker exposure data.   The
total dietary risk from all published agricultural uses of MITEC is 1.6 X  10~5 with exposure to children from apples,
peaches and grapes (2.5 x 10'7 to 9.1 x. 10"6) resulting in the major source of risk concern. Occupational cancer risks
(skin exposure) ranged from 10"7 to 10"4 with high risks estimated for wettable powders on grapes (open and closed
cabs) and commercial mixer/loaders (open mix system). Occupational cancer risks in the range of 10"4 to 10"6  are
acceptable based upon the OPP Cancer Worker Risk Policy.  Noncancer risks (Margin of Exposures, MOEs) were of
low concern based upon  limited exposure and minimal systemic toxicity. Confidence in the basis for a dietary cancer
risk and for exposure is high. Occupational exposure estimates are considered preliminary. Alternate (lower) cancer
risks estimated by the registrant are not considered as scientifically valid since they do not account for time-to-tumor
information.  Special attention should be given to the uncertainties outlined below regarding the cancer risk models and
the exposure assessment methodologies.                                             ,

1. BACKGROUND

Mitec is ai miticide used on field, fruit, nut, and vegetable crops. Apples, grapes^ oranges, almonds, walnuts,
cotton, field corn, and mint comprise 80% of Mitec usage in the US.  BigCorp is the sole manufacturer and
owner of Mitec. Wettable powder and emulsifiable concentrate are the most widely used products. Mitec is
one of the last remaining miticides on the market. One of its major benefits is that it can be used in Integrated
Pest Management (IPM) because it does not preferentially kill beneficial mites.  There are few registered
alternative miticides; none target specific mite populations.

2. PLANNING AND  SCOPING

During the planning and scoping for the Mitec case, dietary and occupational (mixer/loader, applicator,
growers, commercial applicators) risks were the primary risks identified. There were no household or yard
uses identified. Risk to children was of particular concern since preliminary dietary estimates show that
exposure to fresh fruit commodities comprise between 75 to 89% of the total dietary exposure of Mitec to
adults, infants (non-nursing) less than one year of age and children one to six years of age. For  infants,
exposure to processed fruit commodities contributes greater than 82% of their total exposure to  fruits.  EPA's
Office of Pesticide Programs (OPP) decided to perform a  full risk assessment and characterization of Mitec
based on these concerns; even in the face of time constraints (see below) - resources were reallocated to ensure
the full assessment.

The whole toxicological database was reviewed and appropriate endpoints characterized for the dietary and
occupational risks by the OPP Reference Dose (RfD) Committee and the Less Than Lifetime Committee .  The
Committees'  functions were to validate the toxicity conclusions and choose appropriate endpoints for risk

-------
  RISK CHARACTERIZATION CASE STUDY
                                     DO NOT QUOTE OR CITE

assessment.  The RfD Committee recommended further evaluation of the database for the carcinogenicity
potential by the OPP Cancer Peer Review Committee.

Potential exposures in the diet will be estimated with data from BigCorp's market basket survey, monitoring
data from the U.S. Department of Agriculture's (USDA's) Pesticide Data Program (PDF), and field trials and
processing studies. USDA's 1977-1978 Food Consumption Database1 will be used to determine dietary food
consumption of commodities containing Mitec residues. Potential exposures to mixers, loaders, and
applicators as well as private growers and aerial applicators using typical Mitec products will be estimated,
with much being based on the Pesticide Handlers Exposure Database (PHED).

During the planning and scoping phase, several considerations for a possible decision(s) were discussed for
which the risk characterization will be used:

  • Adopt risk mitigation measures on any/all uses of Mitec to ensure conformance with its dietary and
    worker exposure standards for acceptable risk; if already meet acceptable risk standards, then little or no
    action.
  • Possible immediate or emergency suspension of Mitec use on certain crops that may most impact
    children's potential risk.
  • The next growing season for Mitec use begins in six weeks; if you have not taken action to suspend Mitec
    use on certain crops that may impact children's risk the, most by that time, EnviroGroup will sue the
    Agency for inaction and begin a public campaign targeting Mitec and the Agency's inaction.
    EnviroGroup's campaign may result in a  food scare and significant loss of revenue for farmers and
    supermarkets selling fruits and vegetables treated with Mitec.

The remainder of this Mitec risk characterization is presented in sections based on the risk assessment
paradigm: a combined hazard identification and dose response assessment, exposure assessment, and risk
characterization.

3. HAZARD IDENTIFICATION/DOSE RESPONSE ASSESSMENT

From the reviewed toxicological database for Mitec, cancer and systemic toxiciry (i.e., decreased body weight
gain) are determined to be the critical endpoints for risk assessment. A developmental toxicity concern was
raised, but was dropped as additional data were submitted alleviating the concern. Workers may have an acute
risk issue with severe dermatitis; Mitec's corrosivity is a concern (see Section 3.3).

3.1 Cancer
      1 Data from the survey were translated into food forms, which are the individual components of various foods. For example,
      apple pie would be translated into the food forms of apples, sugar, wheat flour, spices, and fat. Survey data are available for
      the entire US population and 22 subgroups, including non-nursing infants, and children. Dietary exposure for Mitec was
      calculated by combining information on anticipated residues for various foods with food consumption to derive an average
      lifetime exposure estimate for Mitec residues consumed on food.

-------
  RISK CHARACTERIZATION CASE STUDY                           '
                                    DO NOT QUOTE OR CITE

Mitec is classified according to the EPA's Guidelines for Carcinogen Risk Assessment as a Group B2 <
carcinogen by OPP's Health Effects Division's (HED's) Carcinogen Peer Review Committee2 (CPRC) based
on two rat bioassays showing undifferentiated jejunal sarcoma.  This is a malignant arid extremely rare tumor
type. In the first bioassay with these tumors there was an issue with use of propylene oxide, a known
carcinogen and mutagen, as the stabilizer. Tumors were confirmed with the second bioassay using epoxidized
soybean oil stabilizer. Mutagenicity data on Mitec contributed to the weight-of-evidence for carcinogenicity -
there is evidence for induction of gene mutations in cultured mammalian cells and chromosome breaking
ability in exposed mice. A cancer bioassay in mice did not demonstrate tumors, but this study is not used as it
is deemed limited due to inadequate1 dose levels and the use ,of older animals at the initiation of dosing.

BigCorp proposed that Mitec is carcinogenic via a non-linear mechanism, where cell proliferation in the small
intestinal (jejunum) is followed by tumor formation. BigCorp has submitted a short term cell proliferation
study to support their claim. HED has reviewed the data and found that it is not sufficient to support the
mechanistic claim. An additional 65-week cell proliferation study is planned; HED has reviewed the protocol
and provided comments to BigCorp.

Outside of the cell proliferation data, there were no other data submitted to OPP that addressed a possible
mode of action for Mitec tumor induction.  Due to the lack of a plausible mode of action other than mutagenic
activity, the rarity of the tumor type, the malignancy, and no evidence that this tumor induction is not relevant
to humans, a low dose extrapolation model for projected human exposures was selected as most appropriate.   .
The cancer mode of action has important risk assessment implications. If Mitec is carcinogenic by a non-
linear mode of action and tumors are only found above a certain critical dose, then the Agency would use a
different model, non-linear model to evaluate risk at human exposures. Risk numbers would be significantly
lower (see table below).

The data from the first rat bioassay are deemed adequate for dose-response quantification. The Mitec unit risk
for cancer, or Qj*, is 1.71 X 10"1 (mg/kg/day)"1 for males; the Qj* is in human equivalents.  The Qj* is based on
male rat fatal jejunum sarcoma and is estimated using the Time-to-Tumor Multi-Stage Model and a body
weight374 interspecies scaling factor (the quantification did not include interim sacrificed animals).  A time-to-
tumor model fatal tumor analysis was used to account for dose related mortality and high incidence of fatal
sarcomas. The Q* represents the slope of the 95% upper bound of the dose response curve for jejunum
sarcomas. Cancer risk estimates based on the Qj*3 are an estimate of the upper bound risk as per Agency
policy; the true value of the cancer risk is unknown. The use of this extrapolation model assumes a linear dose
response relationship at lower doses.

BigCorp has calculated a separate Qj* of 3.2 X  10~2 (mg/kg/day)"1 using a slightly different statistical model;
this represents the geometric mean of the Qj* values for males and females from the same tumor data set. The
Quanta! Multistage Model used by BigCorp does not account for differential mortality. OPP believes that this
      2 This is a standard internal peer review conducted within OPP. The conclusions of the CPRC have not been evaluated by an
      external peer review body, which normally is the OPP Scientific Advisory Panel.

      3 This upper bound estimate of risk is generally thought to cover the range of human variability and is considered by the EPA
      to be inherently conservative of public health.  It may also be an overestimate of the risk.               ,

                                                  3

-------
  RISK CHARACTERIZATION CASE STUDY                                                     '
                                   DO NOT QUOTE OR CITE

statistical model is inappropriate because it is necessary to account for differences in mortality noted between
dose groups because death was from malignant, fatal tumors. The choice of statistical model makes a
difference between the cancer risk estimates by a factor of about ten.
3.1.1. Impact of Proposed Revised Cancer Guidelines

Based on the direction of the proposed revisions to the Carcinogen Risk Assessment Guidelines, HED has
evaluated risk using another linear extrapolation method.  HED has calculated an LED 10 (lower bound of the
10% effective dose or ED 10) for Mitec using the time to tumor model. The LED 10 is used as the point of
departure from the observed data to draw a straight line to the origin for a low dose extrapolation. The unit
risk based on this line is very close to the Q,*. Thus use of the LED10 does not significantly impact the cancer
quantification or the projected risk estimates (see table below). The LED10 versus ED10 is the current
Agency consensus for the point of departure stemming from the proposed Guidelines.

BigCorp submitted their own LED 10 for Mitec (not presented). However,  the company used the Quanta!
Multistage Model to fit the data, which OPP believes is inappropriate for the reason stated above.

3.2. Systemic Toxicitv

The tpxicological database for Mitec was evaluated by the OPP Less than Lifetime peer review committee to
select an appropriate endpoint for assessing the non-cancer risks associated with short (1-7 days) and
intermediate (1 week-several months) term exposure tp agricultural workers.  Toxicity studies of relatively
short duration were reviewed as workers are only exposed to Mitec for a few days to a few weeks per year.
The endpoint selected for short and intermediate term occupational exposures is decreased maternal body
weight gain from a rabbit developmental toxicity study. The No Observed Adverse Effect Level (NOAEL) is
6 mg/kg/day; this NOAEL is supported by another rabbit developmental study, and a rat and a dog chronic
feeding studies with NOAELs in the range of 6-8 mg/kg/day. However, this endpoint may not be of concern
with the current occupational exposure scenarios.

3.3. Acute Toxicitv

Animal studies were evaluated for acute toxicity, irritation, and dermal sensitization.  Developmental toxicity
data were also considered in evaluating acute toxicity because developmental toxicity can result from a single
exposure to a chemical. Mitec has low acute toxicity with most LD50 values >5 g/kg. Mitec is a reported skin
and eye irritant hi animals; there are also reports of severe dermatitis in farm workers reentering fields in
California treated with Mitec. Further evaluation of this issue is underway  (see Section 5.2, data gaps). HED
                                                4

-------
  RISK CHARACTERIZATION CASE STUDY                              /
                                      DO NOT QUOTE OR CITE

determined that the weight-pf-evidence, including the developmental toxicity data4, is insufficient to support
an acute risk assessment.                                                '
3.4 . Route-to Route Extrapolation

The most appropriate toxicological studies for risk assessment are by the oral route; however, dermal exposure
is the primary route for mixer/loader and applicator exposure (inhalation is expected to be a minor pathway
and therefore a minor contributor to risk). Therefore, an absorption factor is estimated for route to route
extrapolation. Dermal absorption of Mitec was estimated by comparing oral and dermal toxicity because there
is no adequate dermal absorption study. The NOAEL from a rabbit 21 -day dermal toxicity study was
compared to the NOAEL for reduced maternal body weight gain from the rabbit developmental toxicity study.

The dermal absorption is estimated as follows:                                        •
% dermal absorption ~   systemic NOAEL rabbit oral developmental tox study
                     systemic NOAEL rabbit 21 -day dermal tox study
                                                                              X  100
The rabbit 21-day dermal toxicity study ori Mitec showed hematologic -changes at the high dose of 100
mg/kg/day, but this effect was attributed to secondary inflammation resulting from dermal irritation. More
definitive signs of systemic toxicity, such as decreased body weight gain, were not observed at this dose.
Therefore, 100 mg/kg/day was considered to be the NOAEL for the rabbit 21 -day dermal toxicity study.

Therefore, the dermal absorption factor was calculated as:      ,

    % dermal absorption  ~   6 mg/kg/dav  X 100 = 6%                                          •
                          100 mg/kg/day

HED considers the dermal absorption factor of 6% to be a conservative estimate.  The route to route
extrapolation used to estimate dermal absorption adds conservatism due to differences in dosing and
pharmacokinetics by the different routes.
3.5. FOPA Concerns
      4 The Peer Review Committee for Reproductive and Developmental Toxicity (1991) concluded that Mitec does not induce
      developmental toxicity in two species of animals in the absence of maternal toxicity, i.e., maternal toxicity was observed at
      dose levels either below (rabbits) or equal (rats) to those in which developmental toxicity was seen (in rodents prenatal
      developmental toxicity was not observed at all, postnatal developmental toxicity was seen only at maternally toxic doses).
      Based upon the'type of developmental findings observed (fused sternebrae in rabbits, increased pup mortality in rats on '
      postnatal day 7) and its correlation with maternal toxicity, it was decided that an acute toxicity risk assessment was not
      appropriate.

-------
  RISK CHARACTERIZATION CASE STyDY         .
                                     DO NOT QUOTE OR CITE

This risk characterization was developed prior to the enactment of the Food Quality Protection Act of 1996.
Therefore, it did not address in a more comprehensive way the issues of the special susceptibility of infants
and children to developmental (including neurological) toxicity, the cumulative effects of pesticides and other
substances that have a common mode of action, the necessity of adding an additional tenfold uncertainty
factor to account for potential pre- and postnatal toxicity or the completeness of the toxicity data, available
information concerning the variability of the sensitivities of major identifiable subgroups of consumers, and
whether the pesticide may have an effect in humans through endocrine disruption.
4. E2gQSURE ASSESSMENT5

4.1. Dietary Exposure

OPP requires submission of residue chemistry data to support all food use pesticides. A typical data set
includes studies on 1) the nature of the pesticide residue in plants and animals, 2) the magnitude of the residue
in treated crops, processed food and feed, and livestock commodities, such as meat, milk, and eggs 3)
reduction of residue, 4) analytical methods for detecting the residue(s), and 5) chemical identity.  These data
are used to determine the residue(s) of concern, to establish tolerance levels for enforcement, and to estimate
"anticipated" residues of the pesticide in foods.

Anticipated residues were calculated for Mitec to provide a realistic picture of pesticide residues in foods and
subsequent dietary exposure.  The data used in this assessment was based on actual monitoring data at the
point of distribution, i.e., market basket survey data and subsequent monitoring data from USDA's Pesticide
Data Program (PDP) data. Initially, BigCorp conducted a market basket survey (MBS) for major food crops
treated with Mitec.  This MBS measured residues collected at the supermarket. Later, the TJSDA PDP
collected residue data at regional food distribution food centers explicitly for dietary risk assessment. The
USDA PDP data confirmed the results of the registrant's Market Basket Survey.  Anticipated residues are
typically based on data from field trials, processing studies, FDA and USDA monitoring data, and market
basket data, when available.  Therefore, the residue data for Mitec was the highest tiered data typically
available for risk assessment purposes. Listed below are the commodities showing the highest number of
detects from the Market Basket Survey:
    Commodity
    Apples
    Infant Apple Sauce
    Raisins
    Peaches
% Detects       Detects/Total Samples
  31%               62/200
  52%               103/200
  99%               197/200
  30%               44/146"
     5 More detailed information regarding the basis for the dietary and occupational exposure assessments is available in the OPP
     Internal Review Document for Mitec (May, 1997) and the supporting reviews. This includes information on the range of
     concentrations of MITEC noted in the detects in the commodity samples; use parameters for mixer/loaders and applicators of
     MITEC and unit exposure estimates for dermal, hand, total dermal and inhalation exposures for these workers based on the
     PHED database.

-------
  RISK CHARACTERIZATION CASE STUDY
                                   DO NOT QUOTE OR CITE
    Infant Peaches
40%
79/199
 Food consumption data from USDA's. 1977-IB Nationwide Food Consumption Survey
 were used to estimate dietary exposure.. Data from the survey were translated into food forms, which are the
 individual components of various foods. For example, apple pie would be translated into the food forms of
 apples, sugar, wheat flour,  spices, and fat.  Survey data are available for the entire US population and 22
 subgroups, including non-nursing infants, and children.

 Dietary exposure for Mitec was calculated by combining information  on anticipated residues for various
 foods with food consumption to derive an average lifetime exposure estimate for Mitec residues consumed on
 food.

 OPP has extremely high confidence in the anticipated residue values used in the dietary risk assessment
 because Market Basket and/or PDF data were used for most commodities, particularly those that contributed a
 large portion of total dietary exposure. In addition, OPP and BigCorp agree on the residue values used in the
 risk assessment. Although this is considered one of the most complete  dietary exposure assessments ever
 performed by the Agency, subgroups such as vegetarians were not included due to the limitations in the .
 available food consumption survey data (see Section 5.4.2).

 4.2. Occupational Exposure
                                                              ...       •  •   v

 Exposure estimates for mixer/loaders and applicators are estimated from the following uses of Mitec: grapes,.
 almonds, apples, oranges, walnuts, corn, and cotton. Exposure estimates were provided only for major use
 crops The crops selected comprised high volume uses of Mitec. Application methods used in the analysis
 include airblast (for fruit and nut crops) and ground boom and aerial application for field crops (com and
 cotton); these are the major application methods for the selected commodities. Occupational exposure
 estimates were based on generic worker monitoring data available in the Pesticide Handlers Exposure
 Database (PHED) and usage data for Mitec in OPP possession. PHED is used hi lieu of adequate chemical
 specific monitoring data. PHED may also be used in conjunction with adequate chemical specific data to
 obtain a larger sample pool of monitoring replicates. Although BigCorp submitted worker exposure
 monitoring studies for Mitec, these studies were all of poor quality. When OPP reviewed the worker
 monitoring studies for Mitec, it became apparent that the monitoring studies did not meet basic guideline
 requirements under subdivision U of the Pesticide Assessment guidelines;

 Re-entry exposure monitoring studies for Mitec were also conducted by BigCorp, but these studies were poor
 quality. These studies suggest that longer re-entry intervals may be necessary in some cases (citrus, cotton,
 apples).
                               >                     ,
 4.3 FOPA Concerns

 This risk characterization was developed prior to the enactment of the Food Quality Protection Act of 1996.
Nevertheless, it did attempt to address in a more comprehensive way the issues of the consumption patterns
 among infants and children that are likely to result in disproportionately high consumption of foods in

-------
  RISK CHARACTERIZATION CASE STUDY
                                    DO NOT QUOTE OR CITE

comparison to the general population. The issue of aggregate exposures, which includes dietary exposures and
all other exposures for which there may be reliable information, was not addressed.

5. RISK CHARACTERIZATION

5.1.  Dietary Risk

OPP estimated dietary cancer risks for Mitec using the low-dose extrapolation model. Cancer risk estimates
are given for typical application rates for Mitec. A table comparing linear and non-linear model options is
presented below. Dietary risk from all published uses is 1.6 X 10~5. This total dietary cancer risk is the
accumulation of all cancer risks calculated for each commodity using the following equation:
    Extra cancer risk = Qj* X Anticipated Residue Contribution (ARC)

        where Q," = 1.71 X 10"' (mg/kg/day)"1 based on jejunum sarcoma in male rats
        and ARC is calculated for each commodity based on anticipated residues and food consumption

These commodities are of particular concern:
    Apples
        fresh
        cooked/canned

    Peaches

    Grapes
        wine/sherry
        raisins
9.1 X 10 6   high risk contributor, high use, kids
6.1 X 10-6
2.4 X10-6

2.5 X 10~6   high risk contributor, kids

1.0 X 10"6   high risk contributor, kids
4.4 X10-7
2.5 X ID'7
The refined total dietary cancer risk for Mitec to the US population was estimated to be 1.6 X 10-5. OPP has
very high confidence in this estimate because it is based on highly refined anticipated residues and a Ql *
based on a rare tumor type and a scientifically defensible dose-response extrapolation.

5.2 Risk to Children
                                                              - i
Because high detects of MITEC residues are found hi apples, raisins, and other commodities consumed by
children in large quantities, OPP was particularly concerned about potential risks to children. However, the
Agency does not have adequate methodology to address cancer risks to workers at this time.  A crude
assessment was performed to give OPP risk managers information about potential cancer risks to children from
short term exposure to Mitec. In this crude assessment, OPP estimated cancer risk to children from 1 year's
exposure to MITEC (risk to children was estimated by this approach: annual risk was calculated according to
exposure for each subgroup; this risk was then divided by 70 years to estimate annual risk).  These risk
                                                 8

-------
 TUSK CHARACTERIZATION CASE STUDY                               .
                                    DO NOT QUOTE OR CITE

estimates were calculated using food consumption values for nursing and non-nursing infants and children of
various ages. Annual dietary risk for each subgroup was amortized over a 70 year lifetime, which assumed
that childrens' food consumption patterns remained the same over a lifetime.  A similar approach was used to
calculate dietary cancer risk to children from alar.

Children's cancer risk from 1-year's exposure to Mitec was estimated to be in the 10~6 range.

5.3. Occupational Risks

The most appropriate toxicological studies for risk assessment are by the oral route; however, dermal exposure
is the primary route for mixer/loader and applicator exposure (inhalation is expected to be a minor pathway
and therefore a minor contributor to risk). Therefore, an absorption factor is estimated for route to route
extrapolation. Dermal absorption of MITEC was estimated by comparing oral and dermal toxicity because
there is no adequate dermal absorption study. The NOAEL from a rabbit 21-day dermal toxicity study was
compared to the NOAEL for reduced maternal body weight gain from the rabbit developmental toxicity study.
          ...                      "                       -'                  *       '             ^   ,   •
The dermal absorption is estimated as follows:
    % dermal absorption ~   systemic NOAEL rabbit oral developmental tox study
                        systemic NOAEL rabbit 21-day dermal tox study
                                                                 X  100
The rabbit 21-day dermal toxicity study on MITEC showed hematologic changes at the high dose of 100
mg/kg/day, but this effect was attributed to secondary inflammation resulting from dermal irritation. More
definitive signs of systemic toxicity, such as decreased body weight gain, were not observed at this dose.
Therefore, 100 mg/kg/day was considered to be the NOAEL for the rabbit 21-day dermal toxicity study.

Therefore, the dermal absorption factor was calculated as:                                      •
     .       -'        >  \               I''.'                . ,   •   •
    % dermal absorption ~   6 mg/kg/dav  X 100 = 6%
           ,          ,   100 mg/kg/day

HED considers the dermal absorption factor of 6% to be a conservative estimate. The route to route
extrapolation used to estimate dermal absorption adds conservatism due to differences in dosing and
pharmacokinetics by the different routes.

5.3.1. Cancer Risk Estimates
Since cancer risk was decided to be based on a linear low-dose extrapolation model, the occupational cancer
risk estimates using the Qj* value were calculated as follows:

    Extra cancer risk = Qi* X LADD
        s               ^L
where Qi*
and LADD
                        1.71 X 10"' (mg/kg/day)"' based on jejunum sarcoma in male rats (oral bioassay)
                        Lifetime average daily dose, worker exposure at typical application rate, amortized over a 70 year
                        lifetime (with a 35 year work life) and adjusted for 6% dermal exposure.

-------
  RISK CHARACTERIZATION CASE STUDY
                                    DO NOT QUOTE OR CITE

 Occupational cancer risk estimates range from 10~7 to 10~4 for both growers and commercial applicators (aerial
 sprayers). High risk (10~4) scenarios include growers using the wettable powder on grapes (open and closed
 cab scenarios), and commercial mixer/loaders supporting aerial sprayers (open mix system).

 5.3.2. Margins of Exposure for Mixers/Loaders and Applicators

 For noncancer effects, margins of exposure (MOEs) are calculated as a measure of how close the estimated
 exposures come to a no-effect level in an animal study. For MITEC, MOEs are calculated for systemic toxicity
 (decreased weight gain) based on a NOAEL of 6 mg/kg/day for maternal toxicity from a rabbit developmental
 toxicity study. This effect was also noted in other oral studies in other species.  MOEs are given for
 mixer/loaders and applicators using maximum (i.e., label) application rates. Separate MOEs are given for
 inhalation and dermal exposure.
MOEs are calculated using the following equation:
    MOE
NOAEL fmg/kg/dav)
            exposure (mg/kg/day) X % absorption
100% absorption is assumed for inhalation and 6% absorption is assumed for dermal exposure.

MOEs for MITEC ranged from 33 to 30,000 for total exposure. Uses with low MOEs (<100) for
mixers/loaders or applicators included almonds and walnuts.  Mixer/loaders using MITEC in a wettable
powder formulation had MOEs (for total exposure) of 75 for almonds and 58 for walnuts. Growers
(mixer/loader/applicators) had low MOEs (<100) with almonds and walnuts under the following exposure
scenarios: wettable powder/open cab and wettable powder/closed cab.  Label restrictions would preclude
MITEC application by airblast in open cabs at the maximum rates for almonds and walnuts; this is the scenario
that would otherwise be associated with MOEs <50 for growers.

HED has low overall concern for the noncancer risk based on the best available exposure data, despite MOEs
of less than 100.  Mixer/loaders and applicators have systemic toxicity MOEs less than 100 with two crops:
almonds and walnuts. For these crops, MITEC exposure is likely to be of short duration based on average
farm size and average acreage treated per day. For MITEC, HED does not support the systemic toxicity
endpoint for short duration or intermittent exposures.  The severity of effect and route to route extrapolation
further lessen HED's concern for potential systemic effects from Mitec exposure following treatment of
almonds and walnuts.
                                                10

-------
  RISK CHARACTERIZATION CASE STUDY %       ,
                                    DO NOT QUOTE OR CITE
LINEAR
RESPONSE AT
LOW DOSE
"one-hit" single
dose could
trigger sequence
of events
leading to
cancer
NONLINEAR
tumors expected
only above a
critical dose
MODEL
Time-to-Tumor
Multistage/Ql*
Quantal
Multistage/Ql*
Time-to-Tumor
Multistage
LED 10
Margin of
Exposure
(MOE)
UNIT RISK
ESTIMATE
0.171
0.032
0.16Q
NOAEL = 4 •
mg/kg/day
TOTAL
DIETARY
RISK
ESTIMATE
1,6 XIO-5
2.9 XIO'6
1.5 XIO-5
MOE>40,000
(not of concern)
GROUP
SUPPORTING
USE OF
MODEL
USEPA
BigCorp
USEPA, 1996
Proposed Cancer
Guidelines
BigCorp
5.4.  Strengths and Uncertainties of Risk Assessment

The risk assessment for Mitec contains strengths and uncertainties based on the existing toxicological and
exposure data, data gaps, and gaps in scientific knowledge. This assessment uses standard assumptions
regarding human body weight, work life, interspecies and intraspecies uncertainty (safety) factors and other
exposure parameters; interspecies extrapolation; and exposure prorated over a lifetime to estimate cancer risks.
Additional assumptions were made regarding route to route extrapolation.

5.4.1. Hazard Identification and Dose Response

The existing evidence for the carcinogenicity endpoint is strong. Carcinogenicity was confirmed by two rat
bioassays; the tumor type, undifferentiated sarcoma of the jejunum, is rare and fatal.  Mitec is classified as a
Group B2, probable human carcinogen. HED also has high confidence in the statistical model used for cancer
dose-response  (Q/ and EDI0).  The Q* was derived using the tune to tumor statistical model because it
incorporates the biological observations of dose-related mortality issues in the rodent bioassay. HED believes
this model is the most scientifically appropriate model for this particular case. Although BigCorp proposed a
potential non-linear mode of action for the carcinogenicity of Mitec, the database does not support this
argument. CALEPA and HEALTH CANADA agree with HED's interpretation of the data.  The Ql * is the
95th percent confidence limit of the dose response curve, which is the default Agency policy. The actual
cancer potency may be considerably less than estimated using this value.
                                                11

-------
  RISK CHARACTERIZATION CASE STUDY
                                   DO NOT QUOTE OR CITE

The evidence for the systemic toxicity endpoint, decreased weight gain, is also strong.  This effect was noted
in three species: the rat, rabbit, and dog.  For the chronic rat and dog feeding studies and the rabbit
developmental toxicity study (gavage), the NOAEL was within the range of 4-6 mg/kg/day with Lowest
Observed Adverse Effect Levels (LOAELs) for all species based on decreased weight gain.
                                                          '   '       '       £               ,,'•'"!'
There is uncertainty regarding the relevance of the systemic toxicity endpoint, decreased weight gain, to the
anticipated exposure scenarios for MITEC.  HED believes this endpoint is not relevant to single day,
intermittent exposures.  HED believes that repeated exposure would be required to produce this effect at levels
to which humans are exposed. Growers who apply MITEC at the rate specified on the  label would be exposed
from 1 to 5 days per treatment and from 2 to 12 days per season. Exposure  duration is  likely to vary with farm
size and the severity of the pest problem.  Commercial applicators would be expected to be exposed
throughout the treatment window for a particular crop and pest.

An additional area of uncertainty pertaining to the systemic toxicity endpoint is route to route extrapolation.
Most of the toxicology data used to support the systemic endpoint are from  studies with oral gavage or dietary
administration. The rabbit developmental study used for risk assessment was conducted by oral gavage.
Mixer/loaders and applicators are exposed to MITEC by the dermal and inhalation routes. There are likely to
be significant differences in absorption and pharmacokinetics between routes due to the bolus effect associated
with oral gavage, local irritation seen with dermal dosing, and other factors. Another issue for consideration is
severity of effect. The systemic endpoint, decreased weight gain, is a relatively minor effect.
5.4.2. Dietary Exposure Estimates

HED has extremely high confidence hi these dietary risk estimates.  Two highly reliable sources of residue
data (the market basket survey and USDA's PDF) are in close agreement. Residue values are extremely
refined, after numerous internal discussions as well as discussions with BigCorp. Food consumption data from
the 1978-79 USDA survey were used. While it is acknowledged that this is relatively "old" food consumption
data (1977-78 survey), the data are used consistently in OPP. A later food consumption survey was deemed
not releasable by the Inspector General. The latest food consumption survey data are not yet available.
Overall, the MITEC case is one of the most refined dietary risk assessments ever performed in the Agency.
OPP's analyses were based on a well designed Market Basket Survey as recommended by the 1993 NAS
report on Pesticides in the Diets of Infants and Children.

5.4.3. Occupational Exposure Estimates

The occupational exposure estimates provided herein should be considered preliminary for a number of
reasons. HED limited the occupational exposure assessment to the major crops and application methods. For
many of the scenarios evaluated, PHED exposure estimates were based on data sets with either low number of
replicates or poor data quality. Therefore, HED has low confidence in many of the exposure estimates.
Exposure scenarios with medium or high confidence include open mixing for liquid formulations, airblast
application with closed cabs, and aerial spraying.
                                                 12

-------
 BISK CHARACTERIZATION CASE STUDY
                                    DO NOT QUOTE OR CITE

Exposure estimates given for commercial applicators (for aerial spraying) are based on the maximum number
of acres that could be treated in a given day. No information was available on the average acreage treated per
year by commercial applicators. The best available information on pest control practices indicates that MITEC
may be used up to 3 months per growing season.  Therefore, to estimate cancer risk, the lifetime average daily
dose for cancer risk assessment was estimated for two scenarios: application of MITEC either 1-3 days/year or
5 days/week for 3 months for corn and cotton.  The first scenario is a likely an underestimate of lifetime
exposure, the second may be an overestimate.

The major limitation of the exposure assessment was the quality of data available in the PHED database. OPP
has low overall confidence in many of the exposure scenarios due to a low number of replicates and /or poor
data quality.

5.5. BigCorp Risk Assessment

In 1993, BigCorp submitted a dietary and occupational risk assessment, which showed dietary risk to the total
US population of 1.53 X 10"6 to 9,25 X 1CT10 and occupational risks in the range of mid 10'5 to 10"9.
However, BigCorp made significantly different assumptions from HED, including use of a different Qj*,
different residue values, and a different method of dietary analysis (Tolerance Assessment System or TAS). In
addition, BigCorp has rebutted many of the assumptions made in the current dietary risk assessment.
However, OPP has discussed these differences in assumptions at great length internally and OPP scientists
believe that the available data strongly support the Agency position on dietary cancer risks of Mitec.
5.6.  Cumulative Risk Assessment

This risk characterization does not address this recent guidance document. The risk characterization does
however, present a planning and scoping phase, which is one of the essential elements in the cumulative risk
assessment policy.  In addition, as recommended by the Cumulative Risk Policy, stakeholders (grower groups,
environmental and public interest groups) were involved. However, the stakeholders felt that they had not
been involved early enough in the process. The aspect of the cumulative, integrated nature of the risk
assessment was not addressed nor the development of a conceptual model6 done. A conceptual model would
of necessity been of an abbreviated nature due to the time constraints on the analysis.

5.7.  Recommendations Regarding Data Gaps

    o BigCorp claims that Mitec is carcinogenic via a non-linear mechanism, where cell proliferation in the
    small intestinal (jejunum) is followed by tumor formation. A 65-week cell proliferation study is planned
    since short term data was determined to be inadequate. Determination of a threshold dose for cell
    proliferation would indicate a non-linear mechanism for tumors formation.  It is likely the Agency would
    then use a different model to calculate risk, and significantly modify our risk concerns.    .
      6 The conceptual model is defined as "...a description or diagram, of the relationship between the predicted responses of a
      population (or entity of concern) and its stressors laying out the environmental pathways and routes of exposure in the context
      of the assessment."

                                     .13

-------
RISK CHARACTERIZATION CASE STUDY
                                  DO NOT QUOTE OR CITE

  o Pharmacokinetic data is needed to verify the assumption of 6% dermal absorption as well as to validate
  the general use of ratios of oral and dermal NOAELs as a scientifically acceptable method of estimating
  dermal penetration.

  o High quality occupational exposure data is needed for the major crops and application methods.

  o More recent food consumption survey data is needed to validate the assumptions made regarding dietary
  consumption, particularly with regard to infants, children and other possible subsets of the populations
  with possible increased susceptibility to risk.

  o Additional investigation of the exposures being experienced by workers reentering fields treated with
  Mitec, including adequacy of protective equipment is needed based on the issue of it's corrosivity.
                                              14

-------
                 United States
                 Environmental Protection
                 Agency
Prevention, Pesticides
And Toxic Substances
(7506C)
                 Questions  &  Answers
 Release of Mitec Risk Assessment for PublJc Comment
  1.   WHATISMitec?

      Mitec is an acaricide (miticide) used to control a variety of mites on ornamental plants and a number
of agricultural commodities, including: apples, peaches, pears, plums/prunes, apricots, figs, cranberries,
grapes (wine and table), strawberries, green beans, other succulent beans, lima beans, dry beans, mint, hops,
oranges, nectarines, grapefruit, lemons, field corn, sweet corn, popcorn, cotton, almonds, walnuts, and -
peanuts.

      In the U.S., Mitec is manufactured by BigCorp, Inc. of Washington, DC. BigCorp is the sole
producer of Mitec. Mitec has been registered since 1977 and approximately 3 to 5 million pounds are used in
the U.S. annually. The extent of use varies annually and geographically with the extent of mite infestations.
Generally, hot dry weather conditions trigger more severe mite outbreaks.
  2.   WHAT ACTION IS THE AGENCY ANNOUNCING?

      EPA's Office of Pesticide Programs (OPP) is seeking public comment on the preliminary risk
assessment it has conducted for Mitec. The risk assessment that EPA is making available is preliminary and
may be refined in the future if additional health and environmental effects data, use data, or other relevant
information on Mitec become available. A public docket for the preliminary risk assessment was established
April 1.  Comments will be accepted for the next 60 days. Through this opportunity for notice and comment,
the Agency is seeking to strengthen stakeholder involvement, and advance the openness and scientific
soundness underpinning its review of Mitec and other pesticides. This process is designed to assure that
America continues to enjoy a safe and abundant food supply. To view a copy of the risk assessment and
related documents, visit the OPP Pesticide Docket, Public reformation and Records Integrity Branch, Room
119, Crystal Mall2,1921 Jefferson Davis Highway, Arlington, VA.                 .   .

      Following EPA's assessment of the cancer studies in laboratory animals and residue data on foods, the
Agency has concluded that long-term exposure from current uses of Mitec poses an unacceptable dietary
cancer risk to consumers. Short term exposure is not expected to be harmful to consumers. Cancer risks are
only exceeded following repeated, long term exposures, such as dietary exposures over a lifetime. Due to
Mitec's high benefits to farmers, the Agency is seeking public comment before making any regulatory
decisions. Most of me occupational (worker) cancer risks estimates are within the range considered

-------
acceptable. The occupational risk estimates should be considered to be preliminary and will be finalized in
the near term.
 3.   WILL THE PUBLIC HAVE AN OPPORTUNITY TO HAVE INPUT ON THIS ACTION?

      Yes. EPA has formally requested public comment by issuing a Federal Register Notice which has a
60-day comment period to provide input to the Agency.
  4.   WOULD IMPLEMENTATION OF EPA'S AGENCY'S PROPOSED NEW CANCER RISK
      ASSESSMENT GUIDELINES CHANGE EPA'S RISK ASSESSMENT FOR Mitec?

      EPA considered the potential impact of the Agency's proposed Cancer Risk Assessment Guidelines
and concluded that, based on existing data, these guidelines would have no significant impact on the risk
assessment completed for Mitec. When additional studies proposed by BigCorp are submitted to the Agency,
a formal re-evaluation of Mitec's cancer classification and risk assessment methodology would be completed
atthattime.
  5.   IS IT SAFE FOR CONSUMERS, INCLUDING INFANTS AND CHILDREN, TO CONTINUE
      EATING FRUITS AND VEGETABLES WHICH MAY HAVE BEEN TREATED WITH
      Mitec?

      Yes. Fresh fruits and vegetables are an essential part of a healthful diet. Any small increase in health
risks from Mitec residues in foods that have" already been treated witii this pesticide would be greatly
outweighed by the benefits to good health from greater fruit and vegetable consumption.

      It is important to remember that EPA's dietary risk assessment of Mitec is based on a lifetime of
exposure to this pesticide and to these crops. Also, in estimating the potential risk of eating foods that contain
residues  of Mitec, EPA has taken into account the fact that infants and children eat proportionately greater
amounts of certain foods than adults do. Exposure to children by eating apples, peaches, and grapes is
considered to be the major source of risk.  However, short term exposures are not expected to pose health
concerns to children.
  6.   WHY ISN'T EPA PURSUE A SUSPENSION OR EMERGENCY SUSPENSION ACTION TO
      IMMEDIATELY REMOVE THESE Mitec USES?                    ,

      EPA considers a suspension or emergency suspension to be a drastic action which is rarely necessary.
EPA has not pursued a suspension or emergency suspension because short-term exposures (i.e., less than a
lifetime) are not of concern.  Also, farmers rely heavily on Mitec and an emergency suspension could result in
crop losses, financial difficulties for farmers, or even a food scare.  Although EPA is concerned about the
risks, we want to gather information from stakeholders, especially farmers, before making a decision to take
drastic regulatory action, such as a suspension.

-------
  7.    WHY DOESN'T EPA TAKE ALL Mitec FOOD USES OFF THE MARKET
       IMMEDIATELY?

       This voluntary action by the manufacturer will dramatically reduce risk immediately by reducing
consumers' potential exposure to Mitec residues on the affected crops. EPA believes that the actions being
taken at this time will reduce exposure to adults and children by more than 85%, and to infants and children "
by more than 98%.

       The potential risks posed by different Mitec crop Uses vary because of differences in agricultural
practices among crops, variations in the handling, processing and preparation of different foods, and varying
contributions of foods to the American diet The remaining uses also provide significant benefits to growers,
in controlling crop damage from mites. Benefits also vary according to crop and the availability of alternative
pest control techniques.
  8.    EVEN D7 CERTAIN USES OF Mitec ARE SAFE ENOUGH TO REMAIN, IS THERE A
       SIGNIFICANT RISK IN EATING/DRINKING A DIET THAT INCLUDES CONSUMPTION
       OF MANY OR ALL OF THE RETAINED USES?

       In evaluating food crop uses of Mitec, EPA has considered the exposure from each individual food
crop as well as the combined exposure from all the treated food crops. This analysis includes an assessment
of the types and quantities of food consumed throughout a lifetime, including differences in consumption
based on age, sex, and other significant factors, such as geographical region. EPA looks at a number of
important population subgroups, including infants, young children, and ethnic groups.  The Agency believes
that, over a lifetime, it would be prudent for consumers to eat/drink fewer Mitec-treated foods than they do
now. Therefore, EPA is'taking action to eliminate Mitec uses that, based on our current assessments, appear
to pose the greatest risks. In this way, EPA is acting to ensure that consumers, especially infants and children,
are adequately protected.
  9.    DO IMPORTED FOODS ALSO CONTAIN Mitec RESIDUES? HOW IS THE
       GOVERNMENT ENSURING THE SAFETY OF IMPORTED FOODS CONTAINING Mitec
       RESIDUES?

       Imported foods currently may contain Mitec residues, provided that the levels do not exceed the
tolerances (maximum legally permissible residue levels) established by EPA. EPA's tolerance regulations
apply equally to domestically-produced and imported foods. EPA plans to revoke the tolerances for.the
canceled uses of Mitec after allowing some time for legally-treated crops to pass through the food distribution
"pipeline." Once the tolerances are revoked, it will be unlawful to ship foods to the U.S. if they contain Mitec
residues that are no longer covered by a tolerance.  USD A and FDA have monitoring programs in place to
measure residues of Mitec on imported foods.

-------
 10.   ARE THERE BENEFITS TO Mitec USE?

       While there are benefits to Mitec use, EPA believes that the potential risks outweigh the benefits for
the uses that will be canceled. The availability of alternative means of pest control will also help alleviate
negative impacts.

       In the past, Mitec has played a role in programs to manage mite resistance. Mitec also is not very
toxic to beneficial insects. This characteristic has led to extensive use in Integrated Pest Management (IPM)
programs that rely on scouting and maintaining healthy populations of predators.  In some cases, alternative
chemicals may provide less effective controls that could result in some yield losses. In other cases, alternative
control strategies may be more costly. Shifting IPM strategies to systems that do not use Mitec may be
difficult and costly for some growers.

       However, the Agency believes that the potential impacts of removing these uses of Mitec can be
mitigated through use of the alternatives, the registration of new alternatives, through changes in cultural
practices, and, when necessary in emergency situations, through emergency use exemptions (i.e., section 18's)
for states for other compounds likely to be less dangerous under section 18 of the Federal Insecticide,
Fungicide and Rodenticide Act (FIFRA).
 11.   ARE THERE ALTERNATIVES TO Mitec USE?

       There are registered chemical alternatives to Mitec.  However, mites have a history of developing
resistance to chemical controls, and few growers rely on a single chemical for effective mite control. Other
chemicals have restrictions and limitations that may make them less effective for controlling mites. Many
growers have adopted IPM systems for insect pest control and substituting alternative chemicals may require
significant adjustments in these overall systems. For some sites, there may be no registered alternatives.
 12.   WHAT IS THE STATUS OF Mitec IN RE-REGISTRATION?
           i    i                                 -               .

       Mitec is currently being evaluated through the re-registration process. Mitec's eligibility for re-
registration will be determined by the year 2000.

-------

-------

-------
              Charge for Midlothian Risk Characterization Case Study

  Risk Characterization Issue for Peer Review

         The adequacy of any risk characterization depends in the first instance on the reliability
  and credibility of the scientific and technical data and analyses in the risk assessment.  At the
  same time, it is equally important that other risk assessors, decision-makers and risk managers,
  and the public fully understand and appreciate the strengths and limitations of the assessment;'
  that is, the overall scientific "character" of the results. To test this aspect of risk characterization,
  EPA's Risk Characterization Policy and the draft Handbook set forth requirements for
  transparency, clarity, consistency, and reasonableness (TCCR), and offer guidance on developing
  and evaluating this aspect of the characterization of risk.

        This peer review does not focus on the underlying scientific analyses, but on whether the
  presentation of the data and analyses adequately represents the risk assessment results.

  Case Study Context and Background

        This analysis presents screening level risk estimates for direct and indirect exposures
  attributable to emissions from three cement companies and a steel mill located in Midlothian,
  Texas. Region 6 developed the estimates by following the procedures outlined in the U.S.
  Environmental Protection Agency's (EPA) draft Guidance for Performing Screening Level Risk
  Analyses at Combustion Facilities Burning Hazardous Wastes. Releases associated with
  combustion sources were modeled using facility-specific emissions rates, stack characteristics,
.  and estimated representative receptor locations around the facility.  The four human exposure'
  scenarios that were considered include an adult and child resident, a subsistence farmer, and a
  subsistence fisher.

        This document is divided into 4 sections. Section 1 provides background and an
  introduction to the case study. Section 2 provides a site characterization and description of
  exposure scenarios and pathways.  Section 3 provides an overview of the risk assessment results.
  Section 4 presents the risk characterization, which is partitioned into sections describing the
 limitations of the analysis and its conclusions. Results of the screening assessment indicate that  -
 the most significant cancer risk identified in the study was to a subsistence fisherman at a level of
  1E-4. Pathways contributing to this risk include ingestion offish, ingestion of drinking water,
 incidental ingestion of soil, ingestion of vegetables, and inhalation. The intended audience for
 this risk characterization is other risk assessors, midmanagers (who have some knowledge of risk
 assessment), the Regional Administrator and external affairs people (with a lesser degree of
 knowledge).                                       :

-------
                                                                                Midlothian
Points for Peer Reviewers to Address
1)     Using the criteria set forth in the draft Handbook, please evaluate the Midlothian case in
       terms of the principles of transparency, clarity, consistency, and reasonableness.

2)     EPA is interested in your comments on the extent to which the characterization conveys a
       sense that the conclusions about risk are reasonable in light of the available data, the fact
     •  that this is a screening level analysis, and current limitations as to information and
       methods regarding evaluating cumulative risks.

3)     Please comment on the topics listed below, which preliminary EPA reviews have
       identified as several areas of special interest:

       a)     Use of the ISCSTDFT dispersion model for industrial combustion sources only.
       b)     Basis for differences in alternative analyses and opinions.
       c)     Results of other air modeling analyses for Midlothian and other geographic areas.
       d)     Alternative risk estimates.
       e)     Discussion on general acceptance of risk screening methods, air dispersion
              models, and EPA's risk based regulatory approach.
       f)      Research implications of data gaps and assumptions.

4)     EPA welcomes any additional comments or suggestions.

-------
   MIDLOTHIAN CUMULATIVE RISK ASSESSMENT
           RISK CHARACTERIZATION

Multimedia .Planning and Permitting Division
    U.S.  Environmental  Protection  Agency
                  Region 6
              1445 Ross Avenue
              Dallas, TX 75202

               December,1998

-------
                        TABLE  OF CONTENTS

1.   INTRODUCTION	Page  1

2.   STUDY AREA AND EXPOSURE  PARAMETER OVERVIEW	.Page  2

     2.1  Characteristics of Study Area	Page  2
     2 .2  Scenarios and Pathways	 Page  3
                         (  '            •    ••'•''        .    •
3 .   SUMMARY OF RESULTS. .	; . . . Page  7

4 .   RISK CHARACTERIZATION. . . ... .	,	 . Page  12

     4.1  Limitations	.	Page '12
     4 .2  Uncertainty	.- • • •	 Page  12
          4.2.1 Emission Rates	Page  12
          4.2.2 Exposure Parameter Uncertainty.	Page  14
          4.2.3 Limitations  of  ISCSTDFT Air Modeling...Page  15
          4.2.4 Uncertainty.Associate with Scenarios ... Page  15

5 .   CONCLUSIONS.'.	. .Page  15

6.   MAPS AND TABLES

     Map 1:   Points of Maximum Combined Deposition,
              Exposure Locations,  and Air
              Concentration	Page  6
     Table 1: Overall Direct and Indirect Cancer Risk
              Across All Carcinogenic Chemicals	Page  11
     Table 2: Comparison of Modeled and Measured
              Concentrations	Page  12
     Table 3: Comparison of Unit  Deposition Rates and
              Air Concentrations	Page  19
     Table 4: Comparison of Emission  Rates	Page  21
     Table 5: Example of Direct  and Indirect Pathway
              Chemicals	;•	Page  22
7.  REFERENCES	% . .	Page  23

-------
 1.  INTRODUCTION
       his .risk assessment was initiated during the RCRA
permitting process,  in response  to  citizens  concerns  over the
permitted burning of hazardous waste  by Texas  Industries  (TXI)  in
Midlothian,  TX.   Results  of  the  final report were  communicated to
concerned citizens groups, mayors and industry representatives.
The  report was. submitted  to  the  Texas Natural  Resources and
Conservation Commission (TNRCC)  to  support their development  of
TXI's hazardous waste burning permit.

     This document presents  an excerpt overview of screening
level risk lestimates for  direct  and indirect exposures.
attributable to emissions from three  cement  companies and a steel
mill located in Midlothian,  Texas.  The risk screening document
is Midlothian Cumulative  Risk Assessment, written by assessors•in
the Region's Multimedia Planning and  Permitting Division.   Region
6 developed  the estimates by following the procedures outlined  in
the U.S.  Environmental  Protection Agency's (EPA) draft Guidance
for Performing Screening  Level Risk Analyses at Combustion
Facilities Burning Hazardous Wastes.  The risk  estimates presented
in this document  are limited by  the uncertainties inherent  in the
models and the data  upon  which the  analysis  is  based.

     Region  6 attempted to minimize uncertainties by

           evaluating and  incorporating area  data collected'by-the.
           Texas Natural Resource Conservation Commission,
     -     requesting air  emission rate  information directly from
           each of  the facilities,
           developing emission rates" based on tests conducted  at
           similar  facilities when no specific data were provided,
          ' and                    '
           analyzing  the data provided  for the facilities against
           data from  other sources to evaluate its overall
           reasonableness.                  '

     Releases associated with combustion sources were modeled
using facility-specific emissions rates, stack characteristics,
and representative receptor(exposed individuals)locations around
the facility!  The four human exposure scenarios that were

-------
 considered include an adult and child resident, a subsistence
 farmer,  and a subsistence fisher (see further discussion under
 Section 2.2 on Scenarios and Pathways).
      This document is divided into 4 sections.   Section 2
 provides a site characterization and description of exposure
 scenarios and pathways.  Section 3 provides an overview of the
 risk assessment results.   Section 4 presents the risk
 characterization which is partitioned into sections describing
 the limitations of the analysis and its conclusions.

 2. STUDY AREA AND EXPOSURE PARAMETER OVERVIEW
                                         .'.•.'.         f    "'
 2.1 Characteristics of Study Area

      The area subject to  this study is located approximately 30
 miles south of the Dallas-Ft. Worth metropolitan area.   From
 Texas Industries (TXI; see Map 1), the. study area extends 8 miles
 north to Joe Pool Lake, 3 miles south,  3 miles  east,  and 6 miles
 west.  The area is characterized by small hills and valleys with
 elevations generally ranging from approximately 800 feet mean sea
 level south of TXI to 500 feet mean sea level at Joe Pool Lake.
 Predominant wind direction is from the south.
      Chaparral  Steel Corporation (CSC)  and TXI are the two
 southern most facilities.   CSC is located 0.7 miles southwest of
 TXI.   North Texas  Cement Company (NTCC)  and Holnam Cement Company
 are located approximately 4 and 5 miles northeast of TXI,
.respectively.

      With the exception of the city of  Midlothian (approximate
 population of 5100)  which is located approximately 3 miles
 northeast of TXI,  the land use of the study area is predominately
 agricultural with  some industrial development.  The area is home
 to  several small cattle operations and  rural residential
 developments.   Gardens were sighted at  many homes in the area
 during several  site  visits.

      In addition to  Joe Pool Lake (surface area approximately
 7600  acres),  the area also contains two privately owned lakes
 known as Soil Conservation Service (SCS)  Lakes 9 and 10 (combined

-------
 surface area of approximately 84 acres).  SCS .Lakes 9 & 10 are
 located approximately 2 to 3 miles .northwest and north,-
 respectively,  of the CSC/TXI complex very near residential
 developments.            .-.    .                 ,

 2.2 Scenarios  and Pathways

      The four  human scenarios that'were- considered in this
 screening level risk'assessment are the subsistence, farmer, the
 adult and child resident, and the subsistence fisher.   The
 scenarios are  those suggested in the cited 1994 EPA combustion
 guidance.  The individuals included in each of these scenarios
 were assumed to be exposed to contaminates from the emission
 sources via/ingestion of above-ground vegetables,  and incidental
 ingestion of soil,  consumption of drinking water and direct
 inhalation of  particles and vapors.   These exposure scenarios
 differed primarily in their consumption of certain foods.
 Specifically,  only the subsistence farmer was assumed to 'consume
 contaminated beef and milk,  while only the subsistence fisher was
 assumed to consume contaminated fish.    Because the drinking
 water supplied to the  area surrounding the facilities  comes from
.Joe Pool Lake,  exposure via contaminated drinking water was
 considered under all of the scenarios.           .

      Although  differences in consumption are the primary
 difference between the scenarios,  other differences exist.   The
 ingestion rate of soil and'above-ground vegetables and the
 inhalation rate of air differ ,'for the  child and the adult'
 scenarios.   Exposure duration is another difference.   The adult
 resident and fisher are assumed to be  exposed for 30 years,  the
 subsistence farmer for 40 years,  and the  child exposed, for  6
 years.   ,    '

      The watersheds and water bodies .considered in the analysis
 were selected  from U.S.  Geologic Survey (USGS)  topographical maps
 and on information, collected during  a visit to Midlothian.   The
 selected water.bodies  and watersheds that were included in  the-
 analysis are those .that would be large  enough.to support fish and
 reflect  the highest  impact from the facilities,  .In addition,  one
 of .the water bodies  selected (i.e.,  Joe Pool Lake)  was identified
 as  the City of  Midlothian's•primary  drinking water source based
 on  information  from the Texas Department  of Health. As a result,
 Joe  Pool Lake was modeled as the drinking water'source.  The

-------
 topographic maps' were used in identifying, the watersheds
 associated with each water body and in estimating water body and,
 watershed surface areas.

      The SCS Lakes 9 & 10 watershed includes Cottonwood Creek and
 portions of the Newton Branch of Soap Creek. The SCS Lake
 watershed  is also a subsection of  the Joe Pool Lake watershed.
 Assuming that the SCS Lake watershed is sufficient to support
 subsistence fishing is conservative because the true viability of
 the  SCS  Lake watershed tp support subsistence fishing is
 unknown.1  Nevertheless,  Region  6  assumed that  these  water bodies
 could potentially support subsistence activity based on their
 size and their nearby proximity to,residential development.
 Furthermore,  these water  bodies are in an area that could be
 significantly impacted by the facilities' emissions due to their
 near central location to  the  facilities being evaluated in the
 study.

     .Contaminants were assumed to be emitted from the four
 facilities 24 hours/day,  7 days/week,  365 days/year.   EPA's air
 dispersion model ISCSTDFT (Industrial Source Complex Sort-Term)
 was  employed to estimate  the transport of the contaminants to the
 surrounding area.   Soil was assumed to become contaminated by wet
 and  dry  deposition of particles and vapors.   Above-ground
 vegetation,  for human and animal consumption,  were assumed to
 become contaminated via deposition of particles on plants,
 transfer of vapor phase contaminates,  and uptake through the
 roots.   Beef and milk were assumed to be contaminated via
 ingestion of contaminated forage (including hay),  silage,  grain,
 and  soil.   Fish and the drinking water source were assumed to be
 contaminated by deposition directly onto the water body and
 through  contaminants transported to the water body via runoff.
                                                     '
     Map 1 identifies the points of maximum air concentration and
 combined deposition based on estimated constituent-specific
 emission rates.   As seen  from this map,  there were three points
 of maximum air concentration and three points of maximum
 deposition identified.  For each compound grouping,  these points
were typically located in close  proximity to the facility
     1 Both of the SCS lakes are privately owned. The property upon which SCS 10 (the northern most
lake) is located is posted "No Trespassing."

-------
 emitting  the compound at  the highest rate.  Map 1 also presents
 the general locations  (Alf Blf .Cj.) of selected site specific
 receptors evaluated in the study. Subsistence farming was not a
 reasonable land"use assumption for location Ca,  and therefore,
 risk to. subsistence farmers  was actually  calculated at a point
 7,640 feet north of this  location depicted  as point C3 on the
'map.                                                     •  -

      Rather than estimate theoretical worst 'case risk, Region 6
 obtained  information  regarding the location of several potential
 resident  and farm locations  likely to be  most impacted by the  •
 facilities.2  This  information was  obtained during  several site
 visits that were conducted during the summer of 1995.  Three
 site-specific residents,  subsistence fishers and subsistence
 farmers were identified and  modeled in the  analysis  (see Table
 1).   Multiple receptors were considered in  order to ensure that
 the maximum media concentrations of each  pollutant were
 considered .because the overall  risks.for  each pathway could vary
 according to which contaminant  was deposited at the highest rate
 or was present  at the highest  ambient .air location.   Resident Al
 and subsistence farmer Al, resident Bl, and resident Cl and      . '
 subsistence farmer C3 are the  receptors located closest to the /
 points of  maximum combined deposition A,  B,  and C,  respectively.
 The exposed individuals assumed to live at  residence Al, Bl,  and
 Cl included, the'adult and 'child resident  and the subsistence
 fisher.    The difference between the adult resident and
 subsistence fisher was that  the fisher was  additionally exposed
-through the consumption of contaminated fish.    .            . .    '
      2  It should be noted that these locations do not necessarily reflect actual residences and farms based on
interviews etc., but rather reflect a reasonably conservative analysis of activities as seen from driving in and about
the study area. For example, residential locations typically correspond to locations of houses or similar
residential-type structures. Farms were estimated based on the presence of livestock or barn type structures in
the area of interest

-------
                              Map  1:
 Points of  Maximum  Combined Deposition and
                      Air  Concentration
                                            Points of Maximum
                                            Combined Deposition
                                                     A  TCDD, 2, 4 -and 2, 6-dinitrotoluene
                                                        hexachlorobenzene, PCNB,
                                                        pentachlorophenol,
                                                        thallium, and beryllium
                                                     B  For all other organic compounds
                                                     C  For all other metals

                                                        Points of Maximum Air
                                                             Concentration


                                                     D  TCDD, 2, 4 -and 2, 6-dinitrotoluene
                                                        heTachlorobenzene, PCNB,
                                                        pentachlorophenol,
                                                        thallium, and beryllium
                                                     E  For all other organic compounds
                                                     F  For all cither Metals
                                                         Approximate Property
                                                         Boundaries
                                                         Hydrography

                                                         Local Roads
                                                         Primary Roads and
                                                         Highways
                                                         Hydrography
                                                         Railroads


                                                         Facility Emission Points
                                                                                    r
                                                           Approximate Scale:
                                                           1 inch = 1.6 miles
                                                                             jm
                                                                ns
\
     fip5>-^-
     tfoi: \ _£fr~
AI  rflLJ	!

-------
3. SUMMARY OF RESULTS                       ,

     .In'.the Superfund program, EPA established a theoretical
value of an exces.s acceptable lifetime cancer risk range  from  one
in ten thousand to one in one million.  This range may be>
expressed as 1 x 10'4  to 1 x  1CT6 (expressed throughout  this  -
report as 1E-4 to 1E-6).  For example, a risk of 1 x  ICr6  means
that 1 person out of one million could develop cancer as  a  result
of a lifetime exposure to a emissions from  the four facilities
studied in this assessment.1  In the Superfund program, EPA  must
consider the need to conduct remedial (clean-up) _action at  a site
if the theoretical risk exceeds 1 x 10~6 and EPA usually requires
remedial action at locations where calculated excess  cancer risks
are greater than 1 x ICr4 (1,excess cancer case in ten thousand
people could potentially occur).      '                      •

     The level of concern for non-carcinogenic contaminants is
determined by calculating a Hazard Quotient  (HQ) or Hazard  Index
(HI).  An HI is the sum of the HQs for several chemicals  that
affect the same target organ.  If the HQ or HI equals or  exceeds
one, there may be concern for potential exposure to site
contaminants.  EPA .typically considers the need for taking  a '
remedial action at locations where the HQ or HI values equal or
are slightly greater than 1.0 for human populations who may
reasonably be expected to be exposed.  EPA usually requires
remedial action at locations where HQ or HI values significantly
exceed one.                                              .

     This risk assessment estimates theoretical ^cancer risk and
the 'potential for theoretical non-cancer health effects from 30
years (beginning today)  of emissions, - associated with CSC, NTCC,
TXI, and Holnam.   No cancer.risk above regulatory levels  of
concern were identified.   Theoretical and conservative modeling
estimates indicate that  there is the pptential for non-cancer
health effects.   However,  as  explained in more detail in  the
Section 5,  actual site data (for soil and water samples near
residents and fishing areas)  indicates that the models used over-
predict media concentrations  of  the principle contaminants,
antimony and cadmium,  which drive the potential for theoretical
non-cancer health effects.
     The most significant theoretical cancer risk is attributed
to the ingestion of fish caught from SCS Lakes 9&10.  Arsenic

-------
 contributes  up  to 80% of the risk from this pathway.   Other
 pathways  that result  in the significant theoretical  risk  are
 subsistence  farming,  and subsistence fishing in Joe  Pool  Lake.  A
 combination  of  organic contaminants"such as dioxin,  BAP and DEHP
 drive  the subsistence farming risk while arsenic again dominates
 the  subsistence fishing risk.

   ,  Risk assessment  results based on theoretical modeling  (Table
 1) show a potential for non-cancer effects  from exposure  to
 antimony  in  drinking  water,  and  cadmium and mercury  through the
 ingestion of fish from SCS  Lakes 9 & 10.  The HQ for antimony is
 estimated to be three for adults and six for children  at  every
 receptor  location.    The HQ for  cadmium-equals  one for the
 subsistence'  fisherman that  fishes SCS Lakes 9 & 10 and the
 mercury HQ equals one for the  subsistence fisherman  that  fishes
 both SCS  Lakes  9&10 and.Joe "Pool Lake.

     The  chronic oral reference  dose for antimony (0.0004
 tng/kg/day) contains an uncertainty of factor of 1,000.  An
 uncertainty  factor of 1000  means that the critical amount of
 antimony  found  in laboratory studies to cause potential non-
 cancer health effects was multiplied by 1000  to account for
 uncertainties in the  studies before that  value  was used in -this
 study to  estimate the potential  for non-cancer  health  effects.
 Critical  health effects from animal studies upon which the
 reference dose  is based include  a decrease  in median life span, a
 decrease  in  non-fasting blood  glucose levels, altered  cholesterol
 levels, and  a decrease in the  mean heart  weight  of males.   Table
 1 presents the  overall results of'the risk  assessment  process.
                  *i              " •      '                     ,
     The  chronic  reference  dose  for cadmium (0.001 mg/kg/day for
 food and  0.0005 mg/kg/day for water)  contains an uncertainty
 factor of  10.   Critical human health-effects  attributed to
 cadmium include anemia and pulmonary disease, edema, pneumonitis,
possible  effects  on the endocrine system, defects in sensory
 function,  and bone damage".

     Citizens in  the  local area also requested  that Region  6
consider risk to  infants  from dioxin via  the breast milk pathway
and risk from a tire  fire that occurred in  December,  1995,  at a
tire shredding  facility located in  the study area.  To address
the risk via the breastmilk pathway,  Region 6 used the Screening
Guidance methodology  to estimate  an infant's daily intake of

-------
 dioxin  if  the mother were  a  resident,  subsistence farmer,  or
 subsistence  fisher.   These estimated intakes  were then compared
 to  an infants background exposure  to dioxin'through ingestion of
 breast  milk. Based on.the  modeled  values,  an  infant's  estimated
 daily intake of  dioxin'is  0.01 pg/kg/day if- the mother is  a"
 resident,  0.45 pg/kg/day if  the mother is a subsistence farmer,
 and 0.38 pg/kg/day if the  mother is  a subsistence fisher;   All of
 these intakes are less than  1% of  the comparison value of  50
 pg/kg/day  which  is the average daily dose an  infant would  obtain
 from background  levels of  dioxin in  breast milk.            .
      Region  6 considered  including the effects  of  the December
 tire  fire  in  this assessment, but was unable  to  complete  the
 evaluation because of  a  lack of data concerning  the actual .
 emission rates of"contaminants during.the  tire fire and the
 uncertainties associated with using a methodology based on long-
 term  chronic  exposures to  estimate the effects from a short-term
 event.    '                      .           '-•--•'-..

      Finally, Region 6 conducted a qualitative analysis of the
 combined effects of windblown cement kiln  dust  (CKD) emissions
 and the contaminant emissions specified in this  study.    This
 qualitative analysis was conducted by comparing  wbest estimates"
 of high end baseline risks outlined in EPA's  Report to Congress
 on Cement  Kiln Dust with the maximum theoretical risk estimates
 presented  in  this report.  A quantitative  analysis  cannot be
 performed  because the  exposure assumptions and fate and transport
•methodologies used in  the  two studies contain some  differences..
 However, the  comparison  does provide a general feel for the
 overall contribution of  CKD emissions to the  theoretical  risk
 estimated  for the area.                        :
     As discussed above, the  most.significant cancer risk
identified in the study was to a subsistence  fisherman at a  level
of 1E-4.  Pathways'contributing to this risk  include ingestion of
fish, ingestion of drinking water, incidental ingestion of soil,. :
ingestion of vegetables,, and inhalation..  The CKD .Report  to
Congress provides a  "best estimate" of high end  baseline risk
from the ingestion of fish contaminated by CKD at 4E-6.  Risk
from ingestion of surface water1 contaminated by CKD emissions are
estimated at 1E-8.  Risk from the ingestion of soil contaminated

-------
by CKD are estimated at 1E-7.  Risk from ingestion of vegetables
is estimated at 2E-6 and risk from inhalation is estimated at 2E-
12.  All of these risks added together do not materially affect
the most significant estimate, contained in this report of 1E-4.
Thus, the uncertainty associated with the failure to
quantitatively assess risk from the emissions of CKD does not
appear to be significant.
                               10

-------
Table 1  Overall Direct and Indirect Cancer Risk
        Across All Carcinogenic Chemicals
Scenario
i
Adult Resident
Child Resident
Subsistence Fisher
Subsistence Farmer
Adult Resident
Child Resident
Subsistence Fisherman,
Subsistence Farmer

Adult Resident
Child Resident
Subsistence Fisherman
Subsistence Farmer
Theoretical
Risks
Point Al
7E-6
3E-6-
SCS
Lakes
9 & 10
9E-5
Joe
' Pool
Lake'-
3E-5
5E-5
Point Bl
3E-5
1E-5
SCS
Lakes
9 & 10
1E-4
Joe
Pool
Lake"
5E-5
4E-5
Point Cl
, 4E-5
2E-5
SCS
Lakes
9 & 10
1E-4
Joe
Pool
Lake
6E-5
6E-5
                       11

-------
 4.  RISK CHARACTERIZATION

 4.1  Limitations

      This section discusses  the  limitations  and uncertainty
 associated with this  screening level  cumulative risk  assessment.
 The degree to which the  uncertainty needs  to be quantified and
 the amount of uncertainty that is  acceptable varies with the
 intent  of the analysis.   For a screening level analysis  such  as
 this, a high degree of uncertainty is often  acceptable,  provided
 that conservative assumptions are .used to  bias potential error
 toward  protecting human  health.

      Uncertainty can  be  introduced into a  health risk assessment
 at  every step in the  process.  Uncertainty occurs because risk
 assessment is a complex  process, requiring the integration of

      •   '  The release of pollutants into the environment;
                      '
      •     The fate and transport of pollutants in a variety of
           different and  variable environments by processes' that
           are often poorly understood or too complex  to  quantify
           accurately;

      •     The potential  for  adverse health effects in humans as
           extrapolated from  animal bioassays; and

      •     The probability of adverse  effects in a human
           population  that  is highly variable genetically,  in age,
           in  activity level,  and in life style.        -   '

     Even  using  the most  accurate data with  the most
 sophisticated models, uncertainty is  inherent in the  process.

 4.2 Uncertainty

     Uncertainty of data used  to estimate  risk or potential
health effects for emission  rate data, exposure.parameters, air
modeling,  and exposure scenarios are described below.
                    •                     '             ,

4.2.1 Emission Rates

     The availability and quality of chemical-specific emission

                               12          ''

-------
 rates presented one of the largest  sources  of  uncertainty
 associated with this screening level  assessment.   For the cement
 manufacturing companies,  the  majority of  the emission rates  were.
 based on trial burn data.   Because  there  was only limited data •
 and information on the quality of the data  obtained during the  .
 trial burns (e.g.,  percent recovered)  and the  representativeness
 of  the operating conditions during  the trial burns,-the
 representativeness of these emission  rates  could  not be  fully
 evaluated.   To address this source  of uncertainty,  the emission
 rates used in the analysis were compared  across available data
'sources - (i.e.  trial burn  data,  TNRCC  data,  Company reported  data)
 to  ensure that the selected emission  rates  were reasonable while
 still being conservative  enough to  allow  for operational upsets  •
 and the uncertainty associated with the quality of the data.
 Region 6  is confident that the rates  presented are as reasonable '
 as  can be provided given  the  availability of accurate data.   ,In
 fact,  one of the outside  reviewers  noted  that .emission rates  for
 dioxin were consistent with EPA's   experience  in  preparing the
 Estimating Exposure to Dioxin-Like  Compounds  (draft)  report.

      Another significant  source of  uncertainty in the overall
 process is  the use  of emission rates  for  CSC that were based  on
 the assumption that .baghouse  and fugitive emissions  contained
 concentrations of contaminants  similar to those found in steel.
 mill baghouse  dust.   Although contaminant concentrations emitted
 to.the atmosphere from the baghouses  are  unlikely to contain
 concentrations greater than those found in  the dust,  the fugitive
 .emissions could contain higher  concentrations  than those found in
 the baghouse dust since they  are emissions  that have not yet  been
 treated.   In Addition,  the volume of  fugitive  emissions  could be
 more or less than assumed  in  this study because CSC's  actual
 fugitive  emissions  have not been measured^  Hence,  the
 uncertainty in the.emission estimates  for CSC are  significant.

      One  area  of  uncertainty  that has  been,  addressed since the -
 review of the  draft report  by outside  experts is  the uncertainty
 associated  with assumed baghouse dus.t  emissions profile.  The
 emissions profile sets  forth  concentrations of contaminants that
 are  very  similar  to CSC actual baghouse dust data with the
 exception of antimony and  hexavalent chromium.

      The  lack  of  any method to check the viability of  actual
 antimony  and hexavalent chromium emissions  is significant because

               :                 13                        •

-------
 both of these contaminants contribute to the overall .cancer risks
 and non-cancer effects estimates.   Antimony emissions  were based
 solely on the baghouse dust profile contained in the Detailed
 Summary of Information Collection Request Responses For Electric
 Arc Furnaces (ICR).   The ICR is based upon data.from both
 stainless and non-stainless_steel mill facilities.   CSC
 reportedly operates  a non-stainless steel mill.  Hexavalent
 chromium emissions were estimated by assuming that  the hexavalent
 chromium emissions constituted only two percent of  total chromium
 emissions.   This assumption of two percent is based on a table
 included in the Agency for Toxic Substances and Disease
 Registry's Toxicological Profile for Chromium.   The actual
 amounts of antimony  and hexavalent chromium emitted by CSC are
 unknown.
 4.2.2 Exposure Parameter Uncertainty

      Another area of  uncertainty includes, the use of  standard  EPA
 default  values in the analysis.   These include inhalation and
 consumption rates,  body weight,  and exposure duration and
 frequency,  which are  standard default  values used in  most EPA
 risk assessments.   These parameters often assume that the exposed
 population  is homogenous,  when in fact variations exist  among  the
 population.   Using a  single  point estimate for these  variables
 instead  of  exposure parameter probability distributions  ignores  a
 variability that may  influence the results by up to a fact'or of
 two  or three.

      Other  parameters that are subject to uncertainty are the
 data used tp estimate the  chemical,concentration in the  media  and
 locations of interests.   The meteorological data from the
 Dallas/Fort Worth National Weather Station provided an
 approximation of the  meteorological conditions at the site as  no
 site-specific data of sufficient quality were available.
 Different meteorologic conditions can  influence the risk results
 by up to an order of  magnitude given the same facility
 characteristics  and surrounding land uses.

      Another area of  uncertainty is the use of EPA verified
.cancer slope factors,  Reference  Doses  and Reference
 Concentration.   These health benchmarks are used as single point
 estimates throughout  the analysis.   These benchmarks  have both
                                14

-------
 uncertainty and variability associated with them.   However,  the '
 EPA has  developed  a process for setting verified health benchmark
 values .to be used  in  all  EPA risk assessments,   With the
 exception of the dioxin and BaP toxicity equivalency methodology.;
 all health-benchmarks used  in this analysis-are verified through
 the EPA's work groups and available, on the  EPA's Integrated  Risk
 Information System.

 4.2.3 Limitations  of  ISCSTDFT Air Modeling

     The indirect  exposure  model used  in this analysis  is EPA's
 current methodology for addressing•a variety of exposure pathways
 important for chemicals.that  bioaccumulate  and  persist  in the
 environment.  Implementation  o'f this methodology requires air
 dispersion modeling results for wet and dry depositions and  air
 concentrations of  particles and vapors  in a variety of  settings.
 ISCSTDFT is  the only  air  dispersion and deposition  model
 currently available to provide such estimates from  combustion
 sources located in both complex and non-complex terrains.
 ISCSTDFT was released as  a  draft and has  not been widely applied
 in  the present form.

 4.2.4 Uncertainty  Associated with Scenarios

     The exposure  scenarios included in this screening  level
 assessment include an adult and child resident,  a subsistence
 fisher and a subsistence  farmer.  Although  a distribution of the
 characteristics (e.g., consumption  rates) of each type  of
 receptor are reasonably well characterized, population
 distributions for  the modeled  behaviors and activities  have not
 been adequately studied.  For  example,  little is known  about the
 fraction of the general population  that consists of  subsistence
 farmers and fishers.  Without population distributions  for these
 receptors,' the number of people  likely  to be exposed to
 contaminated media cannot be determined and, therefore,  the
 appropriateness of'the receptors cannot :be  evaluated from the
 standpoint of population risk.
                    -•         . .     -       -        ^

 5. CONCLUSIONS

    , The results of this conservative screening level risk
assessment are:
                                15

-------
           1.    available site data show that  there  are no  cancer
                risks  or the  potential  for non-cancer  health
                effects  above regulatory levels  of concern  even
                though conservative,  theoretical models estimate
                exposures equal to or slightly above 'threshold
                levels for potential  non-cancer  effects; and

           2.    the  predominant source  of the  theoretical      '
                exposures above threshold levels is  CSC, not the
                cement companies.                      .

     Region  6  arrives at the first conclusion for two reasons.
First,  the models and exposure scenarios upon which the estimates
of risks  and potential  non-cancer health effects are  theorized to
occur are, in  our judgement,  conservative.  The experts who
reviewed  this  report  also commented  at length on the  conservatism
associated with the risk assessment.   Because the risk assessment
is conservative, actual risks and exposures are likely to  be less
than-the  estimated  risk and  exposures.   Given this  conservatism
and  the fact that the theoretical exposures of  concern for
antimony,  cadmium,  and  mercury are in  the "grey" or "borderline"
range  (equal to or  barely over the threshold),  Region 6 cannot
presently justify the necessity  for immediate  regulatory  action.

     Secondly,  actual measured concentrations of those
cpntaminants that result in  exposures  above threshold values
appear to be present  in media at  concentrations less  than  modeled
concentrations.  Assessment  of measured concentrations antimony.
(the contaminant with the greatest exposure)  in the Midlothian
drinking  water supply system results in a hazard quotient  of 0.05
rather than 3  as presented in Section  3.  Secondly, actual
measured  concentrations  in soil of two of the contaminants for
which exposures  are above threshold  levels  (antimony  and cadmium)
are less  than  modeled concentrations in the area north' of  CSC
close to  receptor locations  Cl and C3.   The measured  and modeled
concentrations are  compared  in Table 2  below  along  with
background data.  The fact that the measured  concentrations are
             **            .                            ,
less than the  modeled concentrations is  particularly  interesting
given that CSC has been  operating since  1975  (20 years to  date)
and TXI has been burning waste derived fuel since 1987 (9  years
to date)  and the risk assessment  considers emissions  for 30
years.
                                16

-------
  Table 2   Comparison of Modeled and Measured Concentrations
CONTMNT
Antimony
Cadmium
Mercury
MODELED SOIL
CONG.
(MG/KG)
6.. 3
11 - 50
0.38
MEASURED3
(MG/KG)
' • <3
< 0.095 - 3.6.
• 
-------
Emission rates are compared in Table 4
                               18

-------
     Table 3   Comparison of  Unit Deposition Rates and Air
                         Concentrations
FACILITY
CSC Fugitives
CSC Baghouse A
CSC Baghouse B
CSC Baghouse C
NTCC '
TXI
Holnam
UNIT COMBINED
DEPO.
(g/m2-yr) per 1
g/sec
30'. 8
0.320
0.080
0.078
0.005
0.012
0.001
UNIT AIR CONC.
(ug/m3) per 1
g/sec
18
0.37
0.06
0.063
0.006
0.013
0.001
     As noted in Table 3 above, the deposition rate of
contaminants from CSC are at least an order of magnitude greater
than the contaminant deposition rate associated with the cement
kilns.  CSC's fugitive emissions overwhelm all other deposition
rates by two to three orders of magnitude while Holnam's and
NTCC's deposition rates at this location are almost negligible.
TXI's deposition rate at this location is greater than Holnam'S'
and NTCC's, yet still significantly less than CSC's deposition
rates.            .                '        '    -  •         -   ,    -

     Likewise,  the unit air concentrations associated with
emissions from CSC are at least 100 times greater than those
associated with NTCC and Holnam.  The level of CSC's baghouse
emissions on contaminant air concentrations at this location is
six times the level of TXI,  the next most significant source.
The level of CSC's fugitive emissions is 1000 times the level
TXI's emissions.                .                 .

     A comparison of the emission rates between the four
facilities in Table 4 again shows that CSC's emissions of
antimony and cadmium dominate that of the other facilities.
CSC's estimated emissions of antimony are 186 times that of  TXI
and CSC's emissions of cadmium are almost five times that of TXI.
                               19

-------
     Thus, it is clear that the majority of the potential for
theoretical noncancer health effects associated with antimony and
cadmium result from CSC, not the cement manufacturing facilities.
                               20

-------
Table 4 Comparison of Emission Rates
Constituent
s
Antimony
Arsenic
Barium
Beryllium
Cadmium
Chromium •- VI
Lead
Mercury-
Nickel
Silver
Thallium
Zinc
Chaparral
Estimated
Representati
ve (g/sec)
2.97E-02
1.89E-04
NA . ,
NA
3.02E-03
3.78E-04
5.85E-02
1.06E-05
7.68E-03
NA
NA
5.96E-01
NTCC
Estimated
Representat
ive (g/sec)
9.09E-05
1-07E-05
2.65E-05
1.77E-OS
5".18E-04
:2.65E-05
.. ...4.17E-03
4.67E-04.
2.78E-05
8.96E-05
'1.26E-05
5'.43E-06 -
TXI
Estimated
Representat
ive (g/sec)
1.60E-04
2.13E-04
4.03E-03
2.08E-04
6.50E-04
9.80E-09
1.43E-02
3 .01E-04,
• 3.01E-04
5.33E-05
5.04E-04
2.69E-03
' Holnam
Estimated
Representat
ive (g/sec)
NA
9.00E-05
8.82E-04
2.00E-05
NA
1.11E-05
(total)
8.0.0E-05
2.52E-04
3.78E-04
NA
2.00E-5
8.82E-004
                 21

-------
      Table  5  Example Direct  and Indirect Pathway Chemicals
Direct Exposures
(i.e. /drinking water, inhalation)
. , '-:.•.''••.•••
' 1,3-Dinitrobenzene '.
2, 6-Dinitrobenzene ';•"
2,4-Dinitrotoluene : . '•• •.. •'
Nitrobenzene ".'.'•'
Pentachloronitrobenzene
Bis (2-ethylhexyl) phthalate
Di(n)octyl phthalate :
Hexachlorobenzene
Pentachlorophenol
Antimony
Arsenic
Barium
Beryl ium
'Cadmium
Chromium VI
Lead
Mercury
Silver
Thallium
Indirect Exposures
(i.e., fish, consumption) •
Congeners for (p) dioxins
congeners for dibenzofurans
Benzo (a) pyrene
Benzo (b) f luoranthene
Chrysene
Indeno ( 1 , 2 , 3 - cd ) pyrene
Benzo (a) anthracene
Benzo (k) f luoranthene
Dibenzo (a, h) anthracene
Polychlorinated biphenyls




'
'



Direct Exposure Chemicals; In addition to the chemicals listed in
Table 5 other volatile and semi-volatile constituents were
identified on a facility/source-specific basis.  The additional
chemicals can be extensive and therefore are not presented in
Table 5.  The additional chemicals are in the Region 6 original
Midlothian Cumulative Risk Assessment document  (Section 2.2).
Indirect Exposure Chemicals;  The constituents considered in the
indirect exposure analysis were those recommended in the
Screening- Guidance.  The constituents represent those that are
expected to present the highest risks to human health via
indirect exposures.  Dibenzo(p)dioxins and dibenzofurans were
                                22

-------
converted to 2,3,7,8-tetrachlorodibenzo(p)dioxin toxicity
equivalents (TEQs).  Emissions of polycyclic aromatic
hydrocarbons (PAHs) were converted to benzo(a)pyrene TEQs.
                            References

[NOTE: For purposes of this peer review, the references are not
included]     '          '         •        . "   .
                               23

-------

-------