United States Science Advisory EPA-SAB-EEC45-017
Environmental Board (1400) SEPTEMBER 1M6
Protection Agency Washington, DC
vvEPA AN SAB REPORT:
IMPROVING THE USE
CLUSTER SCORING SYSTEM
RECOMMENDATIONS FOR THE USE
CLUSTER SCORING SYSTEM
PREPARED BY THE ENVIRONMENTAL
ENGINEERING COMMITTEE
-------
September 26, 1995
EPA-SAB-EEC-95-017
Honorable Carol M. Browner
Administrator
U.S. Environmental Protection Agency
401M Street, S.W.
Washington, D.C. 20460
RE: Review of the Use Cluster Scoring System: A Risk Screening Method for Groups
of Commercial Chemicals by Use
Dear Ms. Browner:
At the request of the Office of Pollution Prevention and Toxics (OPPT), a Subcommittee
of the Environmental Engineering Committee (EEC) of the Science Advisory Board (SAB)
reviewed the Use Cluster Scoring System (UCSS). The Subcommittee began its review at the
October 26-28, 1994 EEC meeting, received additional documents during the winter, and
completed its review at the June 22-23, 1995 meeting.
Clustering chemicals by intended functions could provide efficient risk screening, as well
as improved pollution prevention opportunity identification. The UCSS uses algorithms to score
chemicals in each cluster according to health and ecological risks. As a result, priorities can be
established to address the problems of these clusters in greater detail.
The scope of this review included a response to the following questions which constituted
the Subcommittee's charge:
a) How can the clustering process and the methods used to define the chemicals
within a use cluster be improved?
b) Are the data sources proposed for the UCSS appropriate, and are there additional
useful ones?
c) Are the proposed data sources appropriate to bin chemicals into high, medium, and
low risk categories?
-------
d) Are there other general areas/criteria (e.g., waste generation) that should be in-
cluded in the UCSS?
e) Is the manner in which chemicals and clusters are scored appropriate; and how can
uncertainties within the system algorithms and the data sources be treated?"
The findings and recommendations of the Subcommittee are:
a) The proposed UCSS is an important effort to make screening level risk evaluations
without the benefit of a complete risk assessment.
b) The methods used to define chemicals in a use cluster need to be improved. For
example, a cursory search by the Subcommittee members uncovered a large
number of process chemicals in use that were not included in the screening level
risk management case study. Therefore, to ensure that each use cluster is as
accurate and complete as possible, the Subcommittee recommends the use of
several additional data sources including consultation with trade organizations,
individual companies, and other federal agencies. Once all chemicals used in a
process are identified, their respective functionality needs to be consistently
defined.
c) The inclusion of ecological and health concerns, as well as potential for pollution
prevention, is a significant feature of the UCSS. Omission of several important
data sources for the ecological and health concerns severely limits the utility of
UCSS. It should be noted that ecological effects information is often unavailable
for many of the chemicals being evaluated. Information on human health effects
relatively more abundant. The data sources should be expanded to include data
bases from TSCA, as well as QSAR, ECOSAR, and AQUIRE.
d) The basic elements of the scoring system are logically arranged and rely heavily on
surrogate exposure measures such as chemical use volumes, number of workers,
and octanol-water partition coefficient. A large value of any one of these
parameters results in a high exposure estimate. It appears that chemical use
volumes are the primary driver of the scoring system. Such a scoring approach
leads to severe under- or over-estimates of potential risks and an inappropriate
ranking of clusters. The Subcommittee recommends that partitioning behavior,
environmental persistence, release or waste concentrations be given more weight
and appropriate level of effort be allocated to generate the databases for the
UCSS.
-------
e) The UCSS be expanded to include consideration of the cost and efficacy of
substitute chemicals to provide a measure of performance potential. Some cost-
benefit analysis needs to be considered for the overall cluster evaluation. This
analysis could use a relative weighting factor to provide further guidance for prior-
itizing clusters for evaluation in greater detail. No specific method for cost-benefit
analysis has been recommended by the Subcommittee acknowledging that the
Agency will be able to adopt an appropriate method based on its previous
experiences.
The Subcommittee appreciates the opportunity to review the scoring system, and we look
forward to a written response to its recommendations for improving the UCSS.
Sincerely,
Dr. Genevieve M. Matanoski, Chair
Executive Committee
Dr. Ishwar P. Murarka, Chair
Environmental Engineering Committee
Dr. James H. Johnson Jr., Chair
Use Cluster Scoring System Subcommittee
-------
NOTICE
This report has been written as part of the activities of the Science Advisory Board, a
public advisory group providing extramural scientific information and advice to the Administrator
and other officials of the Environmental Protection Agency. The Board is structured to provide
balanced, expert assessment of scientific matters related to problems facing the Agency. This
report has not been reviewed for approval by the Agency and, hence, the contents of this report
do not necessarily represent the views and policies of the Environmental Protection Agency, nor
of other agencies in the Executive Branch of the Federal government, nor does mention of trade
names or commercial products constitute a recommendation for use.
-------
ABSTRACT
At the request of the Office of Pollution Prevention and Toxics (OPPT), a Subcommittee
of the Environmental Engineering Committee (EEC) of the Science Advisory Board (SAB)
reviewed the Use Cluster Scoring System (UCSS) which is being developed by OPPT. The
primary purposes for clustering chemicals by intended functional use are to: a) efficiently screen
large numbers of commercial chemicals; and b) identify opportunities to prevent pollution within
the resulting use clusters. Algorithms are used to score chemicals in each cluster according to
their health and ecological risks in order to set priorities for future evaluation of these clusters.
The review focused on: a) improvement of the clustering process; b) validity of the UCSS
data sources; c) capability of UCSS to sort chemicals by risk categories; d) other general areas
and criteria that could be included in UCSS; and e) the soundness of the system algorithms.
The proposed UCSS is an important initial effort to construct a screening-level risk
statement for clusters of chemicals by use. The Subcommittee report suggests ways to enhance
that effort.
The primary recommendations are that the Agency should: a) identify and use additional
significant data sources for the ecological and health concerns; b) modify the surrogate exposure
measures so that the score is not dominated by wide use in industry, but includes the impacts of
bioaccummulation, persistence and exposure pathways; c) include measures of performance
potential in the chemical-specific analysis; and d) use a cost-benefit analysis for the overall cluster
evaluation.
Keywords: chemical use, scoring systems, pollution prevention, ranking, relative risk
-------
U.S. ENVIRONMENTAL PROTECTION AGENCY
Science Advisory Board
Environmental Engineering Committee (FY95)
Members and Consultants Participating in This Review
CHAIR
Dr. Ishwar P. Murarka, Target Manager, Environmental and Health Science Business Unit,
Electric Power Research Institute, Palo Alto, CA
MEMBERS
Dr. Linda M. Abriola, Associate Professor, Dept. of Civil and Environmental Engineering,
University of Michigan, Ann Arbor, MI
Dr. Calvin C. Chien, Environmental Fellow, Corporate Remediation Group,
E.I. DuPont Company, Wilmington, DE
Dr. Hilary I. Inyang, Associate Professor, Department of Civil Engineering, University of
Massachusetts, Lowell, MA
Dr. James H. Johnson, Jr., Acting Dean, School of Engineering, Howard University, Washington,
DC
Dr. Wayne M. Kachel, Technical Director, Lockheed Martin Corporation,
Oak Ridge, TN
Dr. Jo Ann Lighty, Associate Professor, Department of Chemical and Fuels Engineering,
University of Utah, Salt Lake City, UT
Dr. James W. Mercer, President, GeoTrans, Inc., Sterling, VA
Dr. Frederick G. Pohland, Weidlein Chair of Environmental Engineering,
Department of Civil and Environmental Engineering, University of Pittsburgh, Pittsburgh, PA
Dr. Robert B. Pojasek, Senior Program Director, Cambridge Environmental, Inc., Cambridge,
MA
Dr. Wm. Randall Seeker, Senior Vice President, Energy & Environmental Research Corp., Irvine,
CA
m
-------
CONSULTANTS
Mr. Richard A. Conway, Senior Corporate Fellow, Union Carbide Corporation,
So. Charleston, WVA
Dr. Rolf Hartung, Professor, Environmental Toxicology, Department of Environmental and
Industrial Health, University of Michigan, Ann Arbor, MI
Dr. Richard Kimerle, Senior Science Fellow, Monsanto Corporation, St. Louis, MO
Ms. Lynne Preslo, Senior Vice President, Earth Tech, Berkeley, CA
Dr. Walter M. Shaub, President, CORRE, Inc., Reston, VA
Dr. Michael G. Trulear, Director, Research and Product Development, ChemTreat, Inc.,
Richmond, VA
Dr. C. Herb Ward, Foyt Family Chair of Engineering and Director
Energy & Environmental Systems Institute, Rice University, Houston, TX
SCIENCE ADVISORY BOARD STAFF
Mrs. Kathleen W. Conway, Designated Federal Official, U.S. EPA,
Science Advisory Board, 401 M Street, SW., Washington, DC
Mrs. Dorothy M. Clark, Staff Secretary, U.S. EPA, Science Advisory Board,
401 M Street, SW., Washington, DC
IV
-------
TABLE OF CONTENTS
1. EXECUTIVE SUMMARY 1
2. INTRODUCTION 3
3. FINDINGS AND RECOMMENDATIONS 4
3.1 How Can the Clustering Process and the Methods Used to Define the
Chemicals Within a Use Cluster be Improved? 4
3.2 Are the Data Sources Proposed for the UCSS Appropriate, and Are There
Additional Useful Ones? 5
3.3 Are the Proposed Data Sources Appropriate to Bin Chemicals Into High,
Medium and Low Risk Categories? 8
3.4 Are There Other General Areas/Criteria (e.g., waste generation) That Should
Be Included in the UCSS? 8
3.5 Is the Manner in Which Chemicals and Clusters are Scored Appropriate; and
How Can Uncertainties Within the System Algorithms and the Data
Sources be Treated? 9
3.6 Additional Comments 9
Figure 1. Schematic Diagram of Chemical Use Clusters Scoring Algorithm
(Source: U.S. EPA, 1993) 10
Figure 2. Schematic Diagram of Recommended Changes to the
Chemical Use Clusters Scoring Algorithm (Modified from: U.S. EPA, 1993) 13
REFERENCES CITED R-l
ACRONYMS R-2
APPENDIX A A-l
APPENDIX B B-l
APPENDIX C C-l
APPENDIX D D-l
-------
1. EXECUTIVE SUMMARY
At the request of the Office of Pollution Prevention and Toxics (OPPT), a Subcommittee
of the Environmental Engineering Committee (EEC) of the Science Advisory Board reviewed the
Use Cluster Scoring System (UCSS) being developed by the OPPT. The primary purposes for
clustering chemicals by intended functional use are to: (1) efficiently screen large numbers of
commercially used chemicals; and (2) identify opportunities to prevent pollution within the use
clusters. Algorithms are used to score chemicals in each cluster according to their health and
ecological risks in order to set priorities for the evaluation of clusters in greater detail. To make
this effort even more useful, the Subcommittee made several observations and recommendations
including:
a) The proposed UCSS is an important effort to methodically use existing
information to make a screening-level risk statement about clusters of chemicals by
use.
b) The inclusion of ecological and health concerns as well as potential for pollution
prevention in the UCSS is a significant feature. However, the UCSS, as well as its
use in the screening-level Risk Management (RM-1) dossiers, do not take
advantage of several important data sources for the ecological and health concerns.
This omission severely limits the utility of UCSS. It should be also noted that
ecological effects information is often unavailable for many of the chemicals being
evaluated. Information on human health effects is relatively more abundant.
In addition, the formation of clusters in the RM-1 dossiers were not well
delineated on the basis of functionality and substitutability as required by the
UCSS. Therefore, to ensure that each use cluster is as accurate and as complete as
possible, the Subcommittee recommends the use of several additional sources
including consulation with trade organizations, individual companies and other
federal agencies.
c) The Subcommittee noted with concern that the exposure component of the
algorithm by means of surrogate exposure measures (e.g., chemical use volume
and octanol-water partition coefficient) drives the scores for clusters. As a result,
if a chemical is widely used in the industry, it could receive a high ranking for
human and ecosystem exposure, regardless of its tendency to bioaccumulate, its
tendency to persist in the environment, or its potential exposure pathways and
concentrations of environmental releases. This limitation is judged to be a serious
flaw that must be remedied prior to the application of the UCSS.
-------
d) The Subcommittee recommends that some measure of performance potential be
included in the chemical-specific analysis. The performance potential should
consider factors such as the cost and efficacy of the substitute. A cost-benefit
analysis needs to be considered for the overall cluster evaluation. This analysis
could use a relative weighting factor to compare different clusters and provide
further guidance for prioritizing clusters to be evaluated in more detail. No
specific method for cost-benefit analysis has been recommended by the
Subcommittee acknowledging that the Agency will be able to adopt an appropriate
method based on its previous experiences.
-------
2. INTRODUCTION
At the request of the Office of Pollution Prevention and Toxics (OPPT), a Subcommittee
of the Environmental Engineering Committee (EEC) of the Science Advisory Board (SAB)
reviewed the Chemical Use Cluster Scoring Methodology (September 30, 1994 - draft) under
development by the OPPT. The scientific and technical aspects of this methodology, termed the
Use Cluster Scoring System (UCSS), were reviewed on October 27, 1994, in light of its intended
uses. To aid in the evaluation, two pilot studies on the use of UCSS for First-Level Risk
Management (RM-1) were provided to EEC. The studies were entitled "Corrosion Inhibitors
Used in Industrial Boilers" and "Positive Photoresist Developers for Semiconductors." The
appropriateness of the use of UCSS for these RM-1 assessments is included in the review.
OPPT's primary purposes in clustering chemicals around intended functional use are to: a)
efficiently screen large numbers of commercial chemicals by risk factors; and b) identify pollution
prevention opportunities within the resulting use clusters. By applying algorithms to score
chemicals in each cluster on the basis of health and ecological risks, clusters can then be
prioritized for further evaluation.
OPPT requested particular consideration by the Subcommittee to the following aspects of
the use cluster methodology:
a) How can the clustering process and the methods used to define the
chemicals within a use cluster be improved?
b) Are the data sources proposed for the UCSS appropriate, and are there
additional useful ones?
c) Are the proposed data sources appropriate to bin chemicals into high,
medium, and low risk categories?
d) Are there other general areas/criteria (e.g., waste generation) that should
be included in the UCSS?
e) Is the manner in which chemicals and clusters are scored appropriate; and
how can uncertainties within the system algorithms and the data sources be
treated?
-------
3. FINDINGS AND RECOMMENDATIONS
The Subcommittee recognizes that a complete risk assessment on every chemical is not
practical. Therefore, a rank ordering of groups of chemicals could enable the EPA and industry
to identify the best candidates for reduction and replacement in a pollution prevention program on
the basis of potential for adverse health and ecological effects. Accordingly, the UCSS is a
method of selectively assessing the impacts of a chemical within a designated use area, thereby
somewhat simplifying the exposure assessment. An excellent review of chemical use clustering
methods is provided by Davis et al. (1993). The Subcommittee notes and concurs with the
inclusion of ecological impacts with health concerns and potential for pollution prevention.
The following suggestions to enhance the scientific merit and usefulness of the scoring
system are presented in response to the questions posed in the Subcommittee's charge.
3.1 How Can the Clustering Process and the Methods Used to Define the Chemicals
Within a Use Cluster be Improved?
The Subcommittee recommends improvement in two specific areas, i.e., the methods used
to define chemicals in a cluster, and the functionality and substitutability of chemicals in a cluster.
The methods used to define chemicals within use clusters are deemed inadequate. For example, in
the First-Level Risk Management (RM-1) case study for boiler water treatment processes, a
cursory search of the literature by Subcommittee members uncovered a number of process chem-
icals in use that were not included in the analysis. Moreover, several of the higher scoring
chemicals which were included in the corrosion inhibitor cluster may actually not be used as
corrosion inhibitors in industrial boilers. Therefore, to ensure that each use cluster is as accurate
and as complete as possible, the Subcommittee recommends the use of several other data sources,
such as:
a) Trade associations for each use area who could poll their membership with regard
to chemical use. The EPA Design for Environment program is an example of
where this approach has been used effectively.
b) Industries that produce and supply alternative chemicals. For example, EPA,
Research Triangle Park, North Carolina has held several workshops on low- and
no-VOC coatings with these groups in attendance. The National Roundtable of
Pollution Prevention Programs currently is documenting these workshops.
c) DOD and DOE who currently have alternative chemical initiatives, such as with
personnel at Wright-Patterson AFB and Idaho Falls National Laboratory,
-------
respectively.
After identifying the chemicals used in a particular industrial process, their respective
functionality needs to be defined. In the RM-1 for semiconductors, several forms of functionality
are listed , e.g., photoresists, developers, buffers and surfactants, and other wetting agents.
However, the table of chemicals for the use cluster does not designate the chemicals by
functionality which is important because substitutes can only be identified within a listing of
chemicals by using functionality. Hence, the utility of the UCSS in this RM-1 is not apparent.
In contrast, in the RM-1 for "Corrosion Inhibitors Used in Industrial Boilers," an attempt
was made to assign the component chemicals of a formulation to a use cluster on the basis of their
function in the formulation. In most cases, classifying them by their function in a formulation will
be very difficult and may result in chemicals being misclassified. In addition, components in
formulations frequently function synergistically and their real function is often not apparent.
A second issue for each use cluster is the assumption that chemicals in a use cluster can
substitute for each other without the loss of effectiveness. This assumption is not valid in the RM-
1, "Analysis for the Corrosion Inhibitors Used in Industrial Boilers." In this RM-1, the effec-
tiveness of chemicals in the clusters is not equal and, in most cases, the chemicals also are not
substitutable (See Appendix A).
3.2 Are the Data Sources Proposed for the UCSS Appropriate, and Are There Additional
Useful Ones?
Several of the data sources are considered incomplete and the uses inappropriate. The
Subcommittee recognizes that screening for priority setting needs to address large numbers of
substances with maximum use of available data. However, data quality and relevance need to be
assured, since the inclusion of inappropriate and spurious data will lessen the overall accuracy and
usefulness of the scoring system. Hence, the effort could be strengthened by considering other
databases such as:
a) TSCA information on chemicals solicited from chemical suppliers;
b) Stanford Research Institute databases on chemicals and their industrial use;
c) information solicited directly from industries; and
d) information solicited from professional and trade associations, e.g., the American
Chemical Society.
-------
The Subcommittee also noted a relative dearth of data sources for ecological data
elements in the UCSS. Therefore, it was very difficult to judge this aspect of the results and
utility of the current approach. In that respect, Table 1 developed based upon the experience of
Subcommittee members, provides a qualitative assessment of the ecological effects data elements
currently utilized by the UCSS. A review of how other scoring procedures incorporate ecological
data could lead to the adoption of a better approach that would significantly improve the section
on the potential for ecological effects.
Table 1. Assessment of Ecological Data Elements in UCSS
Element
AWQC, acute and chronic
RQ for aquatics
QSAR
SATeam
Chemical EcoEstimate
Assessment
Very limited use because so few exist
Not a very reliable source
of direct toxicity
Good, if used properly
Acceptable as a last resort
No reason to consider
With respect to the currently used data elements in the UCSS, reliable and high-quality
measured acute toxicity data sources are readily available for every chemical. These include
QSAR (Environmental Estimator and Modeling), ECOSAR, AQUIRE (Aquatic Information
Retrieval) and HSDR (Hazardous Substance Data Bank). Appendix B summarizes the
information available in these sources, and Appendix C provides additional sources on predictive
environmental fate properties and toxicity and environmental fate data sources. Chronic NOEC
(no observable effect concentration) measured data are available through literature searches for
some of the chemicals. NOECs also can be obtained using established acute-to-chronic ratios.
Other important peer-reviewed data bases which could be cited include EPA's Drinking
Water Standards, Threshold Limit Values (TLV) of the ACGIH, and Permissible Exposure Limits
(PEL) of OSHA. Although the TLVs and the PELs are designed for the protection of workers
during a 40-hour work week for a working lifetime, they can be used to estimate values
approximating RfCs (Reference Concentrations) by the application of time-weighted averaging
(Haber's Rule), and the application of a safety factor often to account for the greater variability
of the general population when compared to workers. In application, this results in the following:
Pseudo RfC = TLV x 0.1 x _8_ x J_ = TLV x 0.024
24 7
-------
The components of the exposure portion of the algorithm have only remote relationships
to exposures. It is uncertain that the exposure portion of the algorithm will allow the construction
of even an approximate view of the chemical constituents within clusters. Consequently, it also
will be virtually impossible to estimate relationships among clusters. The exposure potential
elements of persistence, bioconcentration, and some estimate of volume used/released are
appropriate to include in the UCSS. However, there is a need to enhance the extent of the data
base for bioconcentration and persistence. The Subcommittee suggests that other published
procedures be reviewed to select options to derive estimates and highlight reasonable approaches
to estimate bioconcentration and persistence parameters. Sources to consider include the
following:
a) The European Community scoring procedure (Van der Zandt and Van Leeuwen,
1992) provides a useful model for a more sophisticated procedure which carries
exposure estimates to environmental media concentrations. Effects data are then
compared to exposure concentrations, and potential adverse effects evaluated via
the margin of safety.
b) The University of Tennessee Chemical Ranking System (Davis et al., 1993) offers
the advantage of a minimum data base requirement with reliance on QSAR to fill
data gaps.
c) In February 1995, the Society for Environmental Toxicology and Chemistry held a
workshop on Ranking and Scoring Methodologies which resolved some of the
issues identified in the UCSS. Agency personnel should review the outputs of this
important workshop when published.
-------
3.3 Are the Proposed Data Sources Appropriate to Bin Chemicals Into High, Medium and
Low Risk Categories?
The Subcommittee considered the advantages and disadvantages of relying upon a high,
medium, and low categories approach. With the frequent use of high, medium, and low value
judgments, the model has the appearance of being subjective rather than objective. In many
instances, value judgments are made by arbitrarily dividing numerical data into these three groups.
The rationale for these groupings is not apparent, and ignores a great deal of information that is
very useful in understanding why one chemical might be better or worse for the environment.
The Subcommittee suggests that consideration be given to eliminating the high, medium,
and low categories in favor of rank ordering the raw data. Chemicals could then be ranked on the
basis of each characteristic. The rank orders for each characteristic for any one compound could
then be combined to develop a numeric score for an individual compound as is currently done,
e.g., chemicals could be ranked by Use Volume. The chemical with the highest Use Volume
could be assigned a value of 1, while that with the lowest, a much higher number. This approach
is expected to provide a more meaningful relative ranking of the impact of chemical use on the
environment.
3.4 Are There Other General Areas/Criteria (e.g., waste generation) That Should Be
Included in the UCSS?
The Subcommittee recommends several enhancements to improve the scoring of
chemicals and clusters. These include more appropriate data sources for the exposure side of the
scoring algorithm, a measure of performance potential of chemicals, use of high quality, consistent
data sources for ecological and health risks, consideration of pollution prevention alternatives, and
use of a credibility check to ascertain reasonableness of UCSS outputs.
Potential for pollution prevention, e.g., recycling, should also be included in the screening
evaluation. Trade organizations could be consulted at the UCSS stage, informed of the cluster(s)
being examined, and asked if any new emerging technologies are being evaluated which have the
potential for major source reduction.
-------
3.5 Is the Manner in Which Chemicals and Clusters are Scored Appropriate; and How
Can Uncertainties Within the System Algorithms and the Data Sources be Treated?
The basic elements of the Chemical Use Clusters Scoring System (UCSS) are illustrated in
Figure 1. The system currently considers four elements: a) Human Risk Potential, b) Ecological
Risk Potential, c) EPA Interest, and d) Cluster Pollution Prevention Potential. The system scores
both the Human and Ecological Risk Potentials three times higher than the latter two, i.e., EPA
Interest and the Pollution Prevention Potential. The Subcommittee recommends the inclusion of
an additional screening element to provide a measure of chemical performance.
3.6 Additional Comments
While the basic elements of the UCSS algorithm are logically arranged, the Subcommittee
recommends several enhancements to improve the scoring of chemicals and clusters. These
include: incorporation of additional data for the exposure portion of the scoring algorithm, a
measure of performance potential of chemicals, use of high quality, consistent data sources for
ecological and health risks, consideration of pollution prevention alternatives, and use of a
credibility check to ascertain reasonableness of UCSS outputs. The following enhancements in
the scoring algorithm have the potential to improve the ultimate application of the UCSS.
-------
Figure 1. Schematic Diagram of Chemical Use Clusters Scoring Algorithm
(Source: U.S. EPA, 1993)
"INSERT HARDCOPY HERE***
(Later add electronically)
10
-------
a) As designed and implemented, the cluster scoring system heavily weights exposure
data. A qualitative exposure ranking of high, medium, or low weighs both human
hazard and ecological hazard information. Unfortunately, there are no easily
accessible data on exposure. In place of exposure pathway and exposure level
(concentration) information, the cluster scoring system employs surrogate expo-
sure measures, which include chemical use volumes, number of workers, and
octanol-water partition coefficient. A large value of any one of these parameters
results in a high-exposure estimate. For example, if a chemical is used widely in
the industry, it could receive a high ranking for human and ecosystem exposure,
regardless of its tendency to bioaccumulate, its tendency to persist in the environ-
ment, or the potential pathways and concentrations of environmental releases.
It appears that chemical use volumes are the primary drivers of the scoring
system. Such a scoring approach may lead to severe under- or over-estimates of
potential risk and an inappropriate ranking of clusters. It is recommended that
partitioning behavior, environmental persistence, release or waste system
concentrations be given more weight in the exposure estimation scoring process,
and a major effort be employed to improve the data bases in these areas.
Also, care should be used in determining what information should be
entered into the model. Parameters that are functionally related to one another
should not be included in the model. For example, "Reportable Quantities" were
likely developed on the basis of "Reference Doses" or some other toxicological
parameter. The base parameter should be used in favor of the derived parameter.
b) Some measure of performance potential, i.e., the effectiveness of a substitute,
should be included in the chemical-specific analysis. Figure 2 shows a suggested
screening algorithm. The performance potential measure could consider factors
such as cost of substitution and efficacy of substitute, if these data are available. If
not, end-user opinion could be solicited to establish a relative score.
In addition, some type of cost/benefit analysis should be considered for the
overall cluster evaluation (See Figure 2). This analysis could be used as a relative
weighting factor to compare the different clusters and provide further information
for identifying clusters to be evaluated in more detail. No specific approach or
methodology is suggested.
c) Cluster scoring for ecological effects is appropriate when the data are available to
fill in the matrix. Uncertainty of data in a procedure designed for screening
purposes can be handled adequately with some general expression of quality. For
11
-------
aquatic toxicity data, actual tests are of higher quality than QSAR-derived data.
When QSAR data are used, its use should follow the current accepted practices.
d) The evaluation of human health hazards for specific chemicals is probably the
strongest segment of the UCSS. However, the current procedure for health
hazard scoring contains a number of inconsistencies. The Subcommittee concurs
that Reference Dose (RfD) and Reference Concentration (RfC) values are
generally of high quality and have undergone internal review by EPA, but often
have not benefited from formal external peer review. The present methodology
shows an inconsistent and unsupportable set of relationships in scoring thresholds
between RfD and RfC values. As an alternative, equivalent exposures can be
approximated by assuming that an adult consumes 2 liters of water per day, inhales
20 m3 of air, and weighsTO kg. Under such conditions a daily dose of 0.001 mg/kg
of body weight would correspond to a dose of 0.07 mg/20m3 or 0.0035mg/m3 of
air inhaled, or 0.035 mg/L of water consumed.
Reportable Quantities (RQ) and Threshold Planning Quantities (TPQ) are
values that are derived through their own methodologies. Their construct usually
considers a variety of potential effects, including flammability, explosivity, toxicity,
and other conditions. When they are based upon toxicity alone, they also
incorporate severity of responses and their own scaling factors. Consequently, the
reportable quantity has become more of an administrative device and is no longer a
translatable descriptor of toxicity. Although these RQ and TPQ values have
received some peer review, it was for a different purpose.
12
-------
Figure 2. Schematic Diagram of Recommended Changes to the
Chemical Use Clusters Scoring Algorithm (Modified from: U.S. EPA, 1993)
"INSERT FROM HARDCOPY***
(Add electronically later)
13
-------
Table 9 of Reference 1 cites a number of instances of using NOAEL and
LOAEL values. The ratios of the associated scoring thresholds suggest that, in
fact, these were used to calculate values equivalent to RfDs, using the usually
accepted uncertainty factors. Methodologies also exist to estimate NOEC or
NOAEL values from more acute toxicological data, albeit with increasing
uncertainty. Major data bases that can assist in these endeavors are the Registry of
Toxic Effects of Chemical Substances (RTECS) by NIOSH, the Hazardous
Substance Data Base by the National Library of Medicine, and Toxline. The units
of measurements for these data can be made equivalent in terms of dose in mg/kg-
day, or their equivalent water or air concentrations. Thus, it will be possible to
enter the dose rate directly into the scoring algorithm.
The RfC, RfD and derived limiting values relate to dose rates that must be
exceeded by some amount before an effect may be expected. Thus, arranging the
RfDs or their equivalents from lower to higher doses creates an ordinal grouping
from highest to lowest toxicity potential.
The careful distinction between RfC and RfD values also indicates one of
the differences between exposure media. The contaminant releases and/or their
subsequent fate and transport are often media specific, so that it may be desirable
to differentiate between potential exposures through air and water (consider the
use of Henry's constant or McKay's modeling approaches). This can also have
many benefits for the assessment of ecological effects.
The data quality will vary as more and more steps are entered in the deri-
vations of the RfDs or their equivalents. It is desirable to carry an index of
uncertainty along with the individual estimates.
e) The current model does not address the "no" chemical alternative which exists for
many of the use categories. Information on this alternative can be obtained from
industry trade associations, the National Roundtable for Pollution Prevention
members, and the Office of Pollution Prevention of EPA. These alternatives
should be identified and listed on the cluster for each use. The Subcommittee
recognizes that substitutes are not always the best means for pollution prevention,
and often only shift or delay the problems. Generally, there are no drop-in
substitutes, and often the substitutes may cause other problems, for example,
trading a human health exposure for an aquatic toxicity problem.
14
-------
f) A reality check must be conducted on the model output to assure that it is
reasonable and thus credible. The Subcommittee recommends the use of external
panels inclusive of affected industries, to judge merits of rank ordering outcomes
from the cluster scoring.
The Subcommittee was not able to ascertain reliability of the data and
impacts on the UCSS output, and therefore, proposes the application of the
following as a means to determine the order of magnitude of the variability of the
data. If the values are ±30 percent or so, the Pareto Principle (80/20 Rule) will
negate the uncertainty, i.e., 20% of the chemicals will cause 80% of the problems
(see, for example, Ishikawa, 1968). Rank ordering of the major impact chemicals
will not be changed by the uncertainly in the data with the ± 30 percent level. Even
at the +100 percent level, the impact will not be overriding. The work group
should consult the literature on the use of Pareto Analysis for rank ordering.
As a final point, the Subcommittee recommends that coordination of the UCSS with other
scoring activities within the Agency be undertaken.
15
-------
REFERENCES CITED
Davis, G.A., et al., 1993. Chemical Ranking for Potential Health and Environmental Impacts,
University of Tennessee, Center for Clean Products and Clean Technologies, Knoxville, TN.
Ishikawa, K., 1968.Guide to Quality Control, Quality Resources Publishing Co., White Plains,NY
1968
Van de Zandt, D.TJ. and CJ. Van Leeuwen, 1992. A Proposal for Priority Setting of Existing
Chemical Substances. Netherlands Ministry of Housing, Physical Planning and the Environment,
Risk Assessment and Environmental Quality Division.
DOCUMENTS REVIEWED
EPA, 1994. Chemical Use Clusters Scoring Methodology (Draft), September, 1994, Chemical
Engineering Branch, Office of Pollution Prevention and Toxics, U.S. Environmental Protection
Agency, Washington, DC.
EPA, 1994. RM-1 Use Cluster Dossier: Corrosion Inhibitors Used in Industrial Boiler (Draft
Report) September 1994,, Office of Pollution Prevention and Toxics, U.S. Environmental
Protection Agency, Washington, DC.
EPA, 1994. RM-1 Use Cluster Dossier: Positive Photoresist Developers for Semiconductors,
August, 1994. Office of Pollution Prevention and Toxics, U.S. Environmental Protection Agency,
Washington, DC.
R-l
-------
ACRONYMS
AQUIRE: a data base focused on the toxic effects of chemical substances on freshwater and
salt-water organisms (excluding aquatic mammals, birds, and bacteria)
ECOSAR: a computer program for estimating the ecotoxicity of industrial chemicals based on
structure activity relationships (SARs)
EEC: Environmental Engineering Committee, Science Advisory Board, U.S. EPA
HSDB: Hazardous Substance Data Base by the National Library of Medicine, a factual,
non-bibliographic database containing information concerning toxicological and
environmental effects of approximately 4300 chemicals
LOAEL: Lowest observable adverse effect level
NOEC: No Observable Effect Concentration
NOAEL: No Observable Adverse Effect Level
OPPE: Office of Policy, Planning and Evaluation, U. S. EPA
PEL: Permissible Exposure Limits (PEL) of OSHA designed for the protection of
workers during a 40-hour work week for a working lifetime
QSAR: an on-line chemical property system that uses Quantitative Structural Activity
Relationship methods
RfC: Reference Concentrations
RfD: Reference Doses
RM-1: First-Level Ri sk Management assessment
RQ: Reportable Quantities
RTECS: NIOSH's Registry of Toxic Effects of Chemical
Substances, a major data base
SAB: Science Advisory Board, U.S. EPA
SETAC: Society for Environmental Toxicology and Chemistry
TLV: ACGIH Threshold Limit Values designed for the protection of workers during a
40-hour work week for a working lifetime
TPQ: Threshold Planning Quantities
TSCA: Toxic Substances Control Act
UCSS: Use Cluster Scoring System
R-2
-------
APPENDIX A
SUBCOMMITTEE REVIEW OF
RM-1 -- USE CLUSTER DOSSIER: CORROSION INHIBITORS
USED IN INDUSTRIAL BOILERS
DECEMBER 5, 1994 DRAFT REPORT
The following review of the draft RM-1 dossier on Corrosion Inhibitors Used in Industrial
Boilers was conducted to assist in the SAB review of the UCSS and its implementation into RM-
1. The comments provided in the review should be used to improve the RM-1, and to reinforce
the recommendations made in the main body of this report for improving the overall usefulness of
the UCSS.
Review of the RM-1 dossier uncovered several major areas of concern with corresponding
opportunities for improvement. The identified areas of concern that will be discussed in the
review are as follows:
1. Cluster Chemical Selection/Identification
2. Chemical Volume/Exposure Data
3. Process Regulatory Trends
Cluster Chemical Selection/Identification
The most serious concern identified in the review of the corrosion inhibitor is the selection
of chemicals defining the use cluster. The Subcommittee is concerned that several of the
chemicals included in the use cluster are not used as corrosion inhibitors in industrial boilers.
Two chemicals of primary concern with respect to this issue are arsenic and antimony.
Several experts in the field of industrial boiler water treatment were consulted to verify the
use of arsenic and antimony in industrial boilers. An attempt was made to verify the use of these
chemicals by consulting the original reference cited in the RM-1 [Kirk-Othmer. 1991. Corrosion
and Corrosion Control (Vol. 7), Steam (Vol 21). Kirk-Othmer Encyclopedia of Chemical
Technology. 3rd and 4th Editions. New York, NY: John Wiley and Sons.] Although listed as
corrosion inhibitors which have been used in aqueous systems, neither arsenic nor antimony could
be confirmed for cluster use (Kirk-Othmer, 3rd and 4th Editions). A number of other cluster
chemicals including chromate, sulfur dioxide and benzotriazole are also questionable as to their
use in industrial boilers.
A-l
-------
As discussed in the Draft Corrosion Inhibitor Dossier [Submitted by Versar, Inc. 1994 to
the Risk Analysis Branch, Office of Pollution Prevention and Toxics, U.S. Environmental
Protection Agency] arsenic and antimony ranked high for human health concerns, with antimony
also high for ecological concern. If these chemicals were not included in the use cluster, it is pos-
sible that the corrosion inhibitor cluster would not have scored sufficiently high to warrant a RM-
1 screening level assessment. The possibility of including chemicals not representative of a use
cluster presents a potentially serious flaw to the UCSS and points out the need to include trade
organizations and other outside expert advice in the UCSS process.
Volume/Exposure Data
The UCSS methodology is judged to be inappropriately weighted with respect to volume
and exposure data which drive the scoring of clusters. To compound further, the data generally
are of high uncertainty. Table 6 of the dossier summarizes the corrosion inhibitor exposure data.
Limitations in the quality and accuracy of the data are recognized and discussed in the draft RM-1
in Section 7.2. A case in point illustrating the limitations of the data is the use volume estimate of
2,300,000 Ibs in Table 6 for hydrazine (although the use is on a volume basis, the actual quantities
are given in weight). Hydrazine is typically applied at part per billion (ppb) levels as a boiler
water oxygen scavenger and is primarily used only in high pressure boilers found mostly in the
power generation industry. Hydrazine use in industrial boilers is limited (41 sites per Table 6),
and suspected to be much less than 2,300,000 Ibs. Other data entries in Table 6 which are suspect
include the use volume estimate for arsenic (2.25 million Ibs), the use volume estimate for
ethylenediamine (1.93 million Ibs) and the number of sites estimate for cyclohexylamine (107,
considered low).
As presented in the UCSS review, several recommendations are made for improving the
quality of the use volume/exposure data, and for reducing the significance of these data in driving
the UCSS methodology.
Process Regulatory Trends
In a number of sections of the draft RM-1, references and parallels are drawn to
regulatory actions, pollution and process trends in industrial cooling water treatment. The
Subcommittee recommends that caution be used in making these comparisons, and that attention
be given to the fact that the technology used for boilers varies with the use of the boiler (Kirk-
Othmer, 3rd Edition). Comments and suggestions with respect to specific sections of the draft
RM-1 which discuss trends are as follows:
a) Section 4.3, page 7
A-2
-------
The trends discussed with respect to new inhibitors to replace chromium com-
pounds pertains to cooling water corrosion control. Parallels to boiler water are
overstated.
b) Section 5.0, page 9
Corrosion inhibitors (e.g., phosphate and zinc) are added to once-through cooling
water. The relevance of the last three paragraphs in Section 5 to boiler water
treatment is not clear and should be restated.
A-3
-------
APPENDIX B
Ecotoxicity Data Sources
Name: QSAR - Environmental Estimation and Modeling
Source: TDS Numerica™
Producer: Hunter Systems and the IPA at Montana State University
Summary of Content: QSAR is an on-line chemical property evaluation system that uses Quantitative
Structural Activity Relationships methods. This system is a tool for estimating chemical properties and
predicting environmental fate, persistence, and the transport of chemical substances in the
environment. Calculations are based on the chemical structure and property values stored in the data
base or entered by the user. Properties estimated include log P, water solubility, pKa, bioconcentration
factor, soil adsorption and vapor pressure. Environmental partitioning and aquatic toxicity estimates are
also provided. Compounds are identified by CAS number or SMILES notation.
Name: ECOSAR
Source: National Technical Information Service (NTIS), Item Number PB94-500485
Producer: USEPA, Office of Pollution Prevention, and Toxics, February, 1994
Summary of Content: ECOSAR is a computer program for estimating the ecotoxicity of industrial
chemicals based on structure activity relationships (SARs). Originally developed to estimate the aquatic
toxicity of chemicals reviewed by the USEPA in response to P re-Manufacture Notices mandated by
Section 5 of the Toxic Substance Control Act, such estimations have been found to have a wide use in
hazard assessment, ecological risk assessments, and general aquatic toxicology.
Name: AQUIRE (Aquatic Information Retrieval)
Sources: Chemical Information Systems; TDS Numerica™
Producer: USEPA Environmental Research Laboratory, Duluth, MN (ERL-D) - sponsored by
the Office of Toxic Substances of the USEPA
Summary of Content: The AQUIRE database focuses on the toxic effects of chemical substances on
freshwater and saltwater organisms (excluding aquatic mammals, birds, and bacteria). Data types
include acute and chronic toxicity, bioaccumulation, and sub-lethal effects. Fathead minnow toxicity
data generated by the EPA Duluth laboratory is included.
Name: HSDB (Hazardous Substance Data Bank)
Sources: STN International (private); TOXNET (public)
Producer: Toxicology Information Program of the National Library of Medicine
Summary of Content: The Hazardous Substance Data Bank (HSDB) is a factual, non-bibliographic
database containing information concerning toxicological and environmental effects of approximately
4300 chemicals. Data regarding environmental impact, human exposure, detection methods,
manufacturing and use information, and emergency handling procedures may also be available. Some
aquatic toxicity data is available in the "Wildlife Toxicity:" field. This data base contains narrative
descriptions of environmental fate and impact. The data sources are identified. This database is
updated quarterly.
B-l
-------
APPENDIX C
PREDICTIVE PROGRAMS FOR
PHYSICAL AND ENVIRONMENTAL FATE PROPERTIES
ECDIN (Environmental Chemicals Data & Information Network)
Source: IDS Numerica™
Producer: Environmental Research Programme of the Joint Research Centre of the
Commission of the European Communities
This system contains important experimental and regulatory information from the U.S.,
Canada, Eastern and Western Europe and the Pacific Rim. Over 25,000 chemicals are included,
with data including chronic and acute toxicity, concentration and fate in the environmental,
physical and chemical properties. In addition, legislation and rules, occupational safety and
health, and detection methods are available.
ENVIROFATE (Environmental Fate and Physical Property Database)
Source: Chemical Information Systems
Producer: Office of Toxic Substances of the U.S. EPA/Syracuse Research Corporation
(SRC)
ENVIROFATE includes data on environmental transformation rates (biodegradation, oxidation,
hydrolysis, photolysis) and physical-chemical properties (water solubility, log partition coefficient,
vapor pressure). Bioconcentration data is also available. Most of the data in ENVIROFATE is
also present in the Environmental Fate Database.
Environmental Fate Database (Components: CHEMFATE, BIODEG, BIOLOG,
DATALOG)
Source: TDS Numerica™
Producer: Syracuse Research Corporation
C-l
-------
CHEMFATE contains experimental data regarding chemical properties, environmental transport
properties, octanol/water partition coefficients, degradation studies, and field monitoring covering
more than 1200 chemicals.
BIODEG contains experimental biodegradation data for more than 800 substances. Full reference
information is provided.
BIOLOG and DATALOG contain bibliographic references covering biodegradation data (for
more than 12,000 chemicals) and environmental fate data (for more than 15,000 chemicals),
respectively. References for BIODEG and CHEMFATE are included in these systems, along with
additional sources of data.
Log P and Related Parameters Database
Source: IDS Numerica™
Producer: Medicinal Chemistry Project at Pomona College
The Log P and Related Parameters Database provides experimentally determined partition
coefficients (Log P) for 14,000 organic compounds using more than 300 solvent pairs including
octanol/water (Low Kow), the standard for this system. Acid dissolution constants (pKa),
references, comments, and evaluations by the database authors, and a large compilation of related
parameters also can be retrieved. Log P values are used in environmental analysis and in models
that predict biological activity.
OHM/TADS (Oil and Hazardous Materials/Technical Assistance Data System)
Source: Chemical Information Systems, Inc.
Producer: Superfund Program of the U.S. Environmental Protection Agency
OHM/TADS provides access to a data file created to aid emergency response teams in the
retrieval of chemical-specific response information. Types of information include reactivity
information, detection information, chemical and physical properties, environmental fate and
chemical data, toxicological data, and hazard information. The system was last updated in 1985,
and contains records for 1,482 substances.
ISHOW (Information System for Hazardous Organics in Water)
Source: Chemical Information Systems, Inc.
C-2
-------
Producer: Office of Toxic Substances of the U.S. Environmental Protection Agency
C-3
-------
The ISHOW system covers six types of physical property data: melting point, boiling
point, vapor pressure, water solubility, log partition coefficient (Log P) and dissertation constants.
Bibliographic references are included. Approximately 5,400 chemicals (16,600 records) were
included as of December, 1992. ISHOW was last updated in 1985, and no further updates are
anticipated.
C-4
-------
APPENDIX D
DATABASE SOURCES
National Technical Information System
5285 Port Royal Road
Springfield, VA 22161
(703) 487-4650
Chemical Information Systems, Inc.
A Division of PSI International, Inc.
810 Gleneagels Court, Suite 300
Baltimore, MD 21286
(800) 247-8737
(410) 196-0712
TDS Numerica™
Technical Database Services, Inc.
135 West 50th Street, Suite 1170
New York, NY 10020-1170
(212)245-0044
STN International
c/o Chemical Abstracts Service
P.O. Box 3012
Columbus, OH 43210
(614)447-3600
General Sciences Corporation
A subsidiary of Science Applications International Corporation (SAIC)
6100 Chevy Chase Drive
Laurel, MD 20707-2929
(301)953-2700
D-l
-------
DISTRIBUTION LIST
The Administrator
Deputy Administrator
Regional Administrators, Regions 1-10
Assistant Administrators
Director, Office of Air and Radiation
Director, Office of Atmospheric and Indoor Air Programs
Director, Office of Policy Analysis and Review
Director, Office of Environmental Processes and Effects Research
Director, Office of Policy Analysis
Director, Climate Change Division
EPA Headquarters Library
EPA Regional Libraries
EPA Research Laboratory Directors
National Technical Information Service
Congressional Research Service
Library of Congress
Office of Technology Assessment
------- |