Rnal Report of the
Philadelphia integrated Environmental
Management Project
Executive Summary
EPA
Regulatory Integration: Division
Office of Policy Analysis
Office of Policy, Planning, and Evaluation
U.S. Environmental Protection-Agency
December 1986
-------
EXECUTIVE SUMMARY
Philadelphia Integrated
Environmental Management Project
INTRODUCTION
The Philadelphia Integrated Environmental Management Project
(IEMP) was an innovative study designed to analyze and address
the problems posed by toxic chemicals in the Philadelphia area.
This project became a prototype for a series of lEMPs that the
U.S. Environmental Protection Agency (EPA) has conducted over the
last four years. The Project's goals were, therefore, both to
provide insights on particular environmental issues and to
develop new general methods. Our objectives were:
• To develop a methodological approach to evaluating and
comparing the risks to human health caused by exposure
to toxic pollutants in the environment, as measured by
cancer and, to a lesser degree, other chronic health
effects
• To use this evaluation to help local officials in the
process of setting priorities for more detailed analy-
sis and, where appropriate, regulatory controls
• To involve local government agencies in the development
and review of the analysis
The concept of integrated environmental management developed
out of EPA's recognition that there are drawbacks to the tradi-
tional approach that EPA and the states have used to develop
environmental regulations. That approach has focused on indi-
vidual industries, pollutants, and media. While very useful, it
has nevertheless limited our ability to determine where among the
various media our resources are best employed to get the most
health protection. It does not ensure that pollution controls
are not merely shifting risk from one medium to another. In
addition, we have often used national standards that do not ade-
quately address all site-specific situations.
In contrast, the IEMP approach provides a multimedia analy-
sis of issues and accounts for transfers of toxics across media—
in land, air, surface water, and ground water. It is founded
principally on the concepts of risk assessment and risk manage-
ment. Risk assessment requires a systematic evaluation of the
potential for adverse effects to human health from exposure to
-------
pollution. Risk management is a process for evaluating pollution
controls in terms of their projected cost and the level of risk
reduction they afford. Finally, we use the IEMP approach to
focus on the issues of one community at a time, so we can develop
environmental management strategies tailored to each area's
unique problems and characteristics.
Risk Assessment
The IEMP uses risk assessment techniques to evaluate and
compare the potential problems from toxic pollutants in the air,
land, and water. In Philadelohia, we focused specifically on
assessing the risks in air, surface water, and drinking water.
At times this work was qualitative in nature. However, in other
situations, we were able to provide quantitative estimates of the
probability that an individual would contract cancer and of the
expected cancer incidence in the entire population. Our tech-
niques for estimating health effects are in accordance with stan-
dard EPA practices, as described in the proposed EPA risk assess-
ment guidelines. Despite these standards, our analysis has some
significant uncertainties, which we elaborate below. The goals
of our risk assessment in Phase I of the Philadelphia IEMP were
(1) to determine which toxics issues were suitable for a more
detailed examination of risks and control options in Phase II and
(2) to identify study topics for which we would initiate ambient
monitoring programs to better assess exposure levels and, in some
cases, potential risks.
Risk Management
Risk management is the process by which policymakers balance
programs to reduce human health risks against the available
resources to support those programs. In its simplest form, it
requires an examination of how large the risks are, how much the
risks can be reduced by various regulatory controls, and the
costs of these controls. However, it often entails a lot more.
It can involve consideration of the strength of evidence we have
that a health problem could exist or whether effective regulatory
controls can be enforced. Risk management is a process that
requires an assessment of all pertinent information before deci-
sions are made to control pollution. Sometimes the appropriate
decision may be to conduct further analysis of the problem. In
Phase II of this study, we dedicated a significant amount of
effort to providing the data and analysis that serve risk manage-
ment needs. This is especially true of areas where we carried
out further problem definition through ambient monitoring or
performed cost-effectiveness analyses to allow local policymakers
to examine risk reductions and costs of controls for general
policy evaluation.
-------
Choice of Location
Philadelphia was chosen as the site of this pioneering study
because it provided a good setting for the examination of multi-
media issues. The City offered the benefits of several rela-
tively good toxics databases and local officials who we believed
provided a strong technical resource for project implementation.
Philadelphia was not chosen because it had significant environ-
mental problems. Local officials had already demonstrated exem-
plary leadership in some areas. EPA and local officials wanted
to explore ways to identify, assess, and manage human health
risks in an area where significant progress had been made in
understanding and controlling toxics issues.
Conclusions
In the following sections we discuss the methodological and
substantive conclusions from the Philadelphia TEMP. First, we
review our activities in Phase I, largely devoted to screening
pollution issues, and report our findings. Second, we present
findings from our Phase II risk assessment and control-options
analysis. We also discuss the limitations in conducting risks
assessments of toxics. Third, we summarize our methodological
findings from the ambient monitoring programs and the substantive
insights gained from these activities. Finally, we close with
observations on our application of the IEMP methodology in
Philadelphia.
SUMMARY OF PHASE I OF THE PHILADELPHIA
IEMP PROCESS AND FINDINGS
Phase I consisted of three major activities:
• Establishing two intergovernmental committees. The
first was the Steering Committee, which included pri-
marily senior appointed officials from all participat-
ing jurisdictions and EPA. The other group, the Tech-
nical Committee, consisted of technical staff from the
environmental agencies. The Steering Committee
directed the study. The Technical Committee reviewed
and guided the technical and scientific activities.
• Gathering and reviewing data. We collected data to
assess the potential effects of human exposure to toxic
chemicals. In some instances, data directly useful for
measuring exposure were already available, e.g., data
-------
on the quality of finished drinking water. In other
cases, especially in the case of air, we estimated the
expected concentration of a compound in the environment
using data on pollutant sources and the manufacturing
processes, which were used to run fate and transport
models. After gathering available data, we organized
it to facilitate easy review by those on our project
committees who had considerable knowledge about the
quality of the data. We relied heavily on their expert
judgment.
• Developing a screening process for identifying critical
toxics issues for detailed assessment in Phase II.
Once we had collected,organized, and reviewed the
preliminary data, we narrowed the project's geographic
and analytical scope. We set project priorities on the
basis of the primary criteria: quantitative and quali-
tative measures of risk. Then we applied secondary
criteria. These criteria included EPA, state, and
local program objectives, the analytical feasibility of
examining the issues, and our ability to control
environmental impacts.
After applying the screening criteria, which narrowed down
our original list of 170 pollutants associated with over
376 ooint and area sources, we arrived at a subset of 17 sources
of nine toxic chemicals released to the ambient air and surface
water and contained in drinking water. These sources and chemi-
cals constituted our Phase II study topics, which, along with our
analytical activities and objectives, are set out in Table 1.
We emphasize that not all study topics had the same objec-
tives. We also note that we could not analyze each issue with
the same degree of technical rigor. Finally, the Phase I results
warrant two concluding remarks:
• We successfully identified a manageable set of multi-
media topics that could be usefully examined with
available EPA resources and analytical methods and that
would address issues of interest to the community. Our
Phase II efforts would focus on toxic chemicals in air
from point and area sources, toxics in finished drink-
ing water, and intermedia transfers of toxics from the
City's major POTW.
• We had to exclude some important environmental issues
early in Phase I of the Philadelphia IEMP, for several
reasons. First, limited resources required that we set
priorities among possible topics. Second, limited data
-------
Study Topics
Taole 1
PHILADELPHIA IEMP
OVERVIEW OF PHASE II STUDY TOPICS AND ACTIVITIES
Analytical Activities
Objectives
Riak And Control-Option* Analysis
-Identified through screening of
available data
-Addressed 7 of the 11 initial
screening issue papers
- Benzene emissions
- Area sources of solvents
- Refinery, pipeline, and
terminal emissions
- Baxter drinking water
- Queen Lane/Belmont drinking
vtater
- Multimedia discharges of
1,2-OCE and 1,2-OCP
- Haloforms (chloroform)
—Monitor ambient air
—Monitor ambient water
—Validate emission estimates
—Recalculate exposures and risks
using air dispersion models
—Identify feasible controls
—Calculate control costs and
removal efficiencies
—Develop model to evaluate and
rank cost-effective control
option strategies
—Provide results in useful format
for review by decision makers
—Determine baseline exposures
—Identify which sources, pollu-
tants, and exposure pathways
contribute most significantly to
estimated human health risk
—Develop cost-effective strategies
for reducing risks to human
health:
- Aggregate excess cancer
incidence
- Risks to the Most Exposed
Individual
—Examine noncarcinogenic risks
—Identify limitations of analysis
Monitoring
1. Benzene Emissions (specifically
from gasoline marketing)
—Identified by Philadelphia as
a topic of interest
2. formaldehyde Ralessaa to the
Aabient Air
—Identified by Philadelphia as
a topic of interest
3. Coabustion of Used Oil
—Identified by all study
participants aa a topic of
interest
-Short-term benzene ambient air —Determine significance of
monitoring at selected intersec- observed concentrations
tions with and without service
stations - Philadelphia air guidelines
- Cancer risk
—Ambient air monitoring —Determine significance of
observed concentrations
- Philadelphia air guidelines
- Cancer risk
-Sampling at points of distribu- —Compare observed used oil concen-
tion and use trations with EPA fuel specifica-
tions
-Analysis of samples for metals
and organics
4. Air Emissions froa Landfills
(focusing specifically on New
Jersey landfills)
—Identified by New Jersey and —Short-term monitoring at a select —Identify compounds and observed
EPA as a topic of interest number of sanitary and hazardous concentrations
waste landfills in New Jersey
—Satisfied EPA interest in test- (but within geographic boundary —Determine significance of
of project)
ing the applicability of a
mobile monitoring system for
use in future geographic
studies
measured concentrations
-------
restricted the extent of our analysis. Third, certain
research and analytical methods now in practice were
not available at the time of the study. Changes in any
of these factors probably would have altered our selec-
tion of issues. We wish to emphasize that the absence
of an issue from our list of study topics does not mean
that the issue is insignificant.
PHASE II FINDINGS
In the following sections we present the findings of the
analytical efforts to characterize risks to human health and
examine the cost-effectiveness of control options to reduce these
potential hazards. We also discuss our conclusions from the
ambient monitoring programs. Tn each area, we summarize both
methodological and quantitative results. Since the Philadelphia
IEMP was a pilot study, we believe that both types of findings
are important for consideration.
Baseline Risk and Control-Options
Analysis for Policy Development
Methodology
We were successful in designing an analytical approach that
could be used by decision makers to identify which environmental
issues present the most significant risks, roughly quantify the
magnitude of these risks, and evaluate the cost-effectiveness of
various control strategies to reduce the risks that we could
quantify. We were successful for several reasons:
• We were able to enlist the critical participation of
the members of our project committees, most notably the
representatives of the Philadelphia Water Department
and Air Management Services (AMS).
• The AMS emissions inventory allowed us to select a
limited number of pollutants and sources that contrib-
uted most significantly to ambient "air releases of the
toxic pollutants included in our analysis. Without
this inventory, it would have been extremely difficult
to narrow the scope of the project.
• We employed existing EPA analytical methods used in
Agency research and rulemaking activities to generate
much of the data and our analytic results. We tailored
their application to our site-specific needs.
-------
Limitations and Uncertainties
The success of providing an analytical approach to evaluate
the risks to human health and the cost-effectiveness of alter-
native control options to reduce these risks must be appreciated
in the context of the uncertainties associated with this work.
We discuss below the limitations in the project scope, exposure
data, toxicological data, and estimates of the cost of controls.
Limitations in Scope. The Philadelphia IEMP was not an
epidemiological study. We did not collect data on diseases that
occurred in local populations or attempt to trace their causes,
environmental or otherwise. Instead, we combined local data and
engineering estimates of environmental exposure to toxic chem-
icals with toxicological data to estimate the risks to human
health.
The Philadelphia IEMP only attempted to estimate the health
risks from exposure to toxic chemicals in the ambient environ-
ment. For example, we did not estimate risks resulting from
occupational and indoor air exposures. We also did not include
exposures through the food chain. The omission of these routes
of exposure in our study does not imply that they are unimpor-
tant. It is quite possible that risks from any of these exposure
pathways could exceed risks from the exposures we considered. We
decided not to assess these exposure routes because of resource
constraints and because these areas were generally outside of
EPA's traditional regulatory purview and area of expertise.
We chose not to analyze the exposure and risks from conven-
tional pollutants in air and water (such as ozone and sulfur
oxides in air, and oxygen-depleting substances and oil and grease
in water) because we believed that we could make a more signifi-
cant contribution by concentrating on toxic chemicals, which are
neither as well understood nor as well regulated. Future proj-
ects, however, may want to consider these pollutants.
Limitations in Exposure Data. Another important limitation
to our analysis is that we did not examine exposures associated
with all sources and pollutants. While we tried to identify and
assess the cancer risks from the most significant sources and
pollutants, some of those for which we were unable to estimate
exposure, such as combustion of used oil, may also pose health
risks.
Even where exposure data were available, those data varied
significantly in quality. As a result, the exposure estimates
vary in their reliability. Those based on extensive monitoring,
-------
such as for trihalomethanes in drinking water and selected chlor-
inated solvents in air, are the best exposure estimates we have.
The exposure data from short-term ambient air monitoring for
benzene from service stations and formaldehyde are much more
uncertain.
Exposure estimates derived from modeling also vary in their
reliability. Estimates of exposure to toxic organic chemicals in
air derived from dispersion models are dependent primarily on the
quality of the emissions data and a few other factors, such as
meteorological data. In general, the uncertainties introduced by
the exposure data are probably much smaller than those associated
with the dose-response information used to estimate human health
risks.
Limitations in Toxicological Data. Estimates of health
effects are designed to be conservative in several ways. When
evaluating potential health hazards from a chemical, EPA scien-
tists assume that health effects observed in laboratory animals
are a reasonable indicator of potential effects in humans. In
converting the animal data to estimate predicted human responses,
and in extrapolating from high doses to low doses, we use models
that yield a plausible, upper-bound estimate of potency rather
than a "best guess" estimate.
Many substances of potential concern have never been evalu-
ated scientifically, or have not been evaluated in sufficient
detail to allow estimation of effects on humans. For example,
lead (present in air, water, and dust) is thought to pose a
health risk to children at ambient levels; however, at the time
we conducted this project, we had no way of estimating individual
risks or numbers of possible cases.
The addition of new toxicological data and revised scienti-
fic interpretations of previous animal studies often leads to
revised potency values. This makes risk assessment of various
chemicals subject to changes in scientific understanding. Since
the time we completed our Phase II analysis, several unit risk
factors used have been revised by EPA's Carcinogen Assessment
Group (GAG). However, these revisions do not appear to change
our findings dramatically.
We relied on CAG unit risk factors for all chemicals except
1,2-dichloropropane. We developed the inhalation and ingestion
unit risk factors for 1,2-dichloropropane using the (potency)
qi* value found in EPA's Drinking Water Criteria Document on
1,2-dichloropropane (March 2, 1984). The Drinking Water Criteria
Document is nearing completion of external review and, as a
result, the unit risk factor for 1,2-dichloropropane could
change.
-------
Despite these uncertainties, our risk estimates are useful
policy analysis results for comparing issues with one another;
setting priorities among environmental issues and concerns that
we examined; and roughly assessing the potential magnitude of the
overall risks from particular pollutants, sources, and pathways.
Limitations In Estimates of the Cost of Controls. For a
variety of reasons, we were unable to obtain complete site-
specific control cost information for all plants in our control-
options analysis. We were able to gather, with the assistance of
AMS, detailed information about the manufacturing process and
variations in production levels for many plants. Many of the
estimates are based on best engineering judgments, using standard
cost estimation techniques employed in EPA regulatory activities.
Risk Assessment and Control-
Options Analysis Results
Within the limitations described earlier, we are able to
draw the following conclusions about the nine pollutants and 17
sources we investigated.^- The reader should be careful not to
construe any risk estimates presented below as predictions of
actual cancer risk in Philadelphia. Actual risks may be signifi-
cantly lower; in fact, they could be zero. The unit risk factors
used in this analysis are based on conservative assumptions that
generally produce upper-bound estimates. Because of limitations
in data and methods in several areas of the analysis, such as
exposure calculations and pollutant selection, risk estimates
were calculated as aids to policy development. The proper func-
tion of the estimates is to help local officials select and eval-
uate issues, set priorities, and develop control strategies for
the topics examined.
1. Our upper-bound estimate of aggregate excess cancer
incidence for the general population of about 1.7 mil-
lion in Philadelphia was close to three cases per
year.2 Drinking water accounted for over four-fifths
of the estimated excess cancer risks we found in our
analysis. Table 2 shows the upper-bound excess cancer
risks by source category and exposure pathway. The
^•Unless otherwise noted, all conclusions on risk and control
strategies are based on the results from our original (1984)
analysis using the environmental and health data available at
that time.
2it is important to consider the opening caveat in the beginning
of this section as each, risk number is read.
-------
10
Table 2
PHILADELPHIA IEMP
PHASE II RESULTS INTENDED FOR POLICY DEVELOPMENT1
UPPER-BOUND ESTIMATES OF EXCESS ANNUAL CANCER INCIDENCE BY SOURCE
CURRENT CONTROL
(1984 analysis)2
Sources
Air (point)
Pharmaceutical Manufacturer
Chemical Manufacturer
Garment Manufacturer
Plastic Cabinet Manufacturer
Industrial Dry Cleaner
Refinery B
Refinery A
Northeast MPCP3
Subtotal
Air (area)
Degreasing
Dry Cleaning
Other Industrial Usage
Gasoline Marketing (excluding self-service)
Sewer Volatilization (NEMPCP)
Delaware River (receiving NEWPCP effluent)
Subtotal
Drinking Water
Baxter DWTP
Belmont DWTP
Queen Lane DWTP
Subtotal
Total4
Estimated Excess
Annual Cancer
Incidence5
(cases/year)
.006
.007
.001
.000
.000
.066
.007
.090
.177
.049
.064
.003
.053
.021
.024
.214
1.221
.447
.770
Percentage
Breakdown of
the Total 2.8
Caaea
.28
.28
.08
.08
.08
2.38
.28
3.28
6.38
1.78
2.38
.18
1.98
.78
.88
43.18
15.88
27.28
7.68
2.438
86.28
2.8
100.08 (Note:
1008 is equal
to 2.8
cases/year)
WPCP - Water Pollution Control Plant.
DWTP = Drinking Water Treatment Plant.
Note: Numbers have three decimal places not aa an indication of precision, but to
identify source contribution to the risks.
UNIT RISK FACTORS USED IN THIS ANALYSIS ARE BASED ON CONSERVATIVE ASSUMPTIONS THAT GENERALLY
PRODUCE UPPER-BOUND ESTIMATES. BECAUSE OF LIMITATIONS IN DATA AND METHODS IN SEVERAL AREAS OF
THE ANALYSIS, SUCH AS EXPOSURE CALCULATIONS AND POLLUTANT SELECTION, RISK ESTIMATES WERE CALCU-
LATED AS AIDS TO POLICY DEVELOPMINT, NOT AS PREDICTIONS OF ACTUAL CANCER RISKS IN PHILADELPHIA.
ACTUAL RISKS MAY BE SIGNIFICANTLY LOWER; IN FACT, THEY COULD BE ZERO. THE PROPER FUNCTION OF THE
ESTIMATES IS TO HELP LOCAL OFFICIALS SELECT AND EVALUATE ISSUES, SET PRIORITIES, AND DEVELOP
CONTROL STRATEGIES FOR THE TOPICS EXAMINED.
2The risk estimates presented in this table were calculated using unit risk factors from 1984.
'Recent reductions in discharges to the NEWPCP may result in lower risk numbers than presented in
this table.
^Columns may not sum due to rounding.
5IEc, Inc., Coat-Effectiveness Analysis of Strategies to Reduce Human Health Risk in Philadelphia.
U.S. EPA. May 1985.
-------
11
compound responsible for most of the estimated upper -
bound excess cancer incidence in drinking water was
chloroform, a trihalomethane. Chloroform's presence in
finished drinking water is a byproduct of the process
for disinfecting the water. Chloroform concentrations
in the Philadelphia drinking water are less than half
of EPA's primary interim drinking water standard.
For the nine pollutants and 17 sources we investigated,
we estimated an upper-bound incidence of nearly three
cases per year of cancer in a city that had about 4,500
die of cancer in 1984.3,4 This point helps place our
analysis in context. Additional perspective on our
findings could be gained by examining estimated cancer
risks from environmental exposures developed independ-
ently of this study. For example, recently available
data suggest that it is quite possible that the risks
from indoor air exposures may be significantly higher
than those from ambient air exposures and those of the
other risks we examined in this study.
Less than one-fifth of the estimated three cases
(upper-bound) per year were attributable to exposures
in air and are divided about evenly between point and
area air sources.3 Table 2 shows the different air
source categories we analyzed and the estimated cancer
incidence associated with exposures to ambient air
releases from^sach. The traditional point sources
(i.e., smokestack industries) were accountable for a
modest percentage (nearly 3 percent) of the risks we
could quantify.
Roughly one-third of the 0.4 estimated annual cancer
cases (upper-bound) from exposure to air toxics risks
is attributable to intermedia transfers (from water to
air) resulting from industrial wastewater discharges to
the Northeast Water Pollution Control Plant (NEWPCP).3
Roughly two-thirds of the estimated upper-bound excess
cancer risks from these intermedia transfers occur at
the sewage treatment plant itself. The remaining one-
third is split fairly evenly between volatilization
from the major sewer line into the NEWPCP and volatil-
ization from the discharge to the Delaware River.
is important to consider the opening caveat in the beginning
of this section as each risk number is read.
We should note that on average the annual number of cancer cases
could be nearly twice the number of annual deaths from cancer
(based on 1983 American Cancer Society national data).
-------
12
Recent reductions in discharges to the NEWPCP and
operational changes may result in lower risk numbers
than presented in the report.
We found that the cumulative upper-bound cancer risks
from multiple chemical exposures are greater than
single-chemical risks by a factor of 10 and, in some
cases, by a factor of 100. EPA's regulatory actions
often focus on individual pollutants, such as the list-
ing and control decisions for hazardous air pollutants
under Section 112 of the Clean Air Act, and do not
account for the cumulative risks that may occur from
exposure to multiple chemicals.
We found no concentrations of toxic chemicals that
warrant increased concern about noncarcinogenic
effects. However, independent monitoring information
(e.g., AMS's breathing zone study) indicated that ben-
zene concentrations could pose some concern. A better
determination of the significance of these concentra-
tions will depend on where the "no-effect" thresholds
for benzene, which are currently under internal EPA
review, are finally set.
People could experience upper-bound individual lifetime
risks of cancer of around 1 in 10,000 from their drink-
ing water and in some cases of air exposure.5 However,
upper-bound risks to the maximum exposed individuals in
air generally are above 1 in 100,000. Table 3 shows
the upper-bound risks to the maximum exposed individual
(MED from air and drinking water.
Decision makers wanting to achieve different levels of
risk reduction will need to employ different control
strategies. For example, to reduce risks by less than
20 percent, the most cost-effective control strategies
focus solely on lowering ambient air releases of
toxics. However, to achieve significantly greater
reductions in risk, some previously recommended air
controls may be replaced by controls at the drinking
water treatment plants. Controls at the drinking water
treatment plants can achieve large, discrete reductions
in risk, as opposed to the smaller, incremental reduc-
tions, achievable through air emission controls at
different sources.
It is important to consider the opening caveat in the beginning
of this section as each risk number is read.
-------
13
Table 3
PHILADELPHIA IEMP
PHASE II RESULTS INTENDED FOR POLICY DEVELOPMENT1
UPPER-BOUND ESTIMATES OF CUMULATIVE LIFETIME CANCER RISKS
TO THE MOST EXPOSED INDIVIDUAL
(1984 analysis)2
Cumulative Lifetime
Cumulative Lifetime Cumulative Lifetime Total Cancer Risk
Cancer Risk Cancer Risk (Upper-Bound) Inhalation
ME I Location (Upper-Bound) Inhalation (Upper-Bound) Inqestion and Inpestion
Northeast WCP3 5.6 x 10~5 1.0 x 10"* 1.6 x 1CT4
Refinery 8 1.4 x 10"5 1.0 x 10~4 1.1 x 10~4
Chemical Mfr. 2.2 x UP* 1.0 x 10'4 3.2 x 10'4
Plastic Cabinet Mfr. 6.5 x 10~6 1.0 x HT4 1.1 x 10~4
Pharmaceutical Mfr. 4.5 x 10"5 1.0 x HT4 1.4 x 10"4
Garment Manufacturer 1.2 x 10'5 1.0 x 10"4 1.1 x 10'4
Refinery A 3.0 x HT5 1.0 x 10"4 1.3 x 10"4
Industrial Dry Cleaner 2.2 x UP5 1.0 x 10'4 1.2 x ID'4
UNIT RISK FACTORS USED IN THIS ANALYSIS ARE BASED ON CONSERVATIVE ASSUMPTIONS THAT GENERALLY
PRODUCE UPPER-BOUND ESTIMATES. BECAUSE OF LIMITATIONS IN DATA AND METHODS IN SEVERAL AREAS OF
THE ANALYSIS, SUCH AS EXPOSURE CALCULATIONS AND POLLUTANT SELECTION, RISK ESTIMATES WERE
CALCULATED AS AIDS TO POLICY DEVELOPMENT, NOT AS PREDICTIONS OF ACTUAL CANCER RISKS IN PHILA-
DELPHIA. ACTUAL RISKS MAY BE SIGNIFICANTLY LOWER; IN FACT, THEY COULD BE ZERO. THE PROPER
FUNCTION OF THE ESTIMATES IS TO HELP LOCAL OFFICIALS SELECT AND EVALUATE ISSUES, SET PRIORI-
TIES, AND DEVELOP CONTROL STRATEGIES FOR THE TOPICS EXAMINED.
2The risk estimates presented in this table were calculated using unit risk factors from 1984.
'Recent reductions in discharges to the NEWPCP may result in lower risk numbers then presented in
this table.
Source: lEc, Inc., Cost Effectiveness Analysis of Strategies to Reduce Human Health Risk in
Philadelphia. U.S. EPA, May 1985.
9. Philadelphia would have to implement controls at the
drinking water treatment plants to reduce most of the
estimated upper-bound excess cancer incidence for the
sources and pollutants considered in our analysis. It
is important to recognize, however, that the upper-
bound estimated individual risk associated with inges-
tion of Philadelphia drinking water is smaller than the
current nationally accepted risk for drinking water,
based on the concentrations established by EPA's pri-
mary interim drinking water standards.
10. The City would need to make significant expenditures
for controls at its treatment plants to reduce most of
the estimated risks from drinking water. This situa-
tion presents local decision makers with a difficult
-------
o
Figure 1
Philadelphia IEMP
Phase II Results Intended for Policy Development1
Cost vs. Reduction in Cancer Incidence
Air and Drinking Water
Percentage Reduction in Cancer Cases
30 40 50 60 70
Total
Annual Cost
(millions of dollars)
Percentage
of Maximum
Control Cost
-10
1 1.5 2
Number of Cases Reduced per Year
2.5
2.8
1 The unit risk factors used in this analysis are based on conservative assumptions that generally produce upper-bound estimates. Because of limitations in data
and methods in several areas of the analysis, such as exposure calculations and pollutant selection, risk estimates were calculated as aids to policy development,
not as predictions of actual cancer risks in Philadelphia. Actual risks may be significantly lower; In fact, they could be zero. The proper function of the estimates
is to help local officials select and evaluate issues, set priorities, and develop control strategies for the topics examined.
-------
15
risk management decision, involving the examination of
the amount of risk reduced, the control costs, and
other factors. Reductions in the upper-bound estimated
cancer risks from drinking water can be achieved by
reducing chloroform levels through the use of granular
activated carbon (GAC) filters. These controls reduce
the chloroform that forms in drinking water following
disinfection.
In our cost-effectiveness analysis, we incrementally
added GAC controls to each drinking water treatment
plant to achieve increasing levels of risk reduction.
The results of our analysis suggest that implementing
GAC at the Belmont drinking water plant only could
reduce approximately one-fifth of the estimated drink-
ing water risks. Implementing GAC at all three drink-
ing water plants could reduce nearly all the estimated
cancer risk. The annual costs of these controls would
range from roughly $11 million to $58 million. The
average cost per cancer case avoided ranges from around
$24 million to $26 million, depending on the drinking
water treatment plant controlled. The value of the
reduction depends on what other risk reduction oppor-
tunities are possible for these same expenditures and
the value of other social services that could use these
resources.
11. For the range of control options we examined, we did
not see any dramatic increase in the cost per case
avoided as we pursued additional control strategies to
raduce further increments of risk. Figure 1 illus-
trates our results. Typically one would not expect
such a linear relationship; rather, one would expect
that, as costs increase, a larger fraction of the
potential risk could be reduced for a significantly
smaller portion of the cost. We attribute this result
to two factors affecting the dominant sources of risk
(i.e., the drinking water treatment plants). First,
the available inexpensive controls, such as switching
from chlorine to chloramines as the residual disinfec-
tant, have already been implemented. Second, as noted
above, the increase in risk reduction would be achieved
by applying the same control technology (i.e., GAC
filters) to additional drinking water treatment plants.
In practice, the City would most likely implement
controls at all three drinking water treatment plants
to further remove contaminants in the City's drinking
water supply.
-------
16
12. Some controls to reduce air risks would save money.
These controls included product and solvent recovery.
These controls would reduce the estimated cancer risks
from inhalation exposures by close to a tenth of the
air risks we quantified and save over one hundred
thousand dollars annually.
13. From a different vantage point, decision makers who are
concerned about the risks to the MEI could realize a
risk reduction of roughly an order of magnitude by
implementing controls costing about $23 million
annually. Again, the value of that purchase depends on
the community's priorities and the importance of pro-
viding everyone with protection above the current
level.
14. Our evaluation of the risks and costs of control was
initially completed in 1984. Our 1986 revision of the
risk estimates, using more current environmental and
health data, indicates that there would be little, if
any, change in the conclusions discussed above. The
major changes noted in our analysis were in the cancer
unit risk factors and reductions in the discharges from
the chemical manufacturer to the NEWPCP occurring after
completion of our monitoring programs in Philadelphia.
The reduction in the industrial indirect discharges
have undoubtedly led to lower volatilization rates from
sewer lines, the sewage treatment plant, and the
Delaware River.
Readers of these conclusions should recognize that local
officials will be considering other information outside our anal-
ysis in using this study's findings. They will evaluate the
control-options analyses we performed in a broader context of
other health, safety, and social services that they provide their
citizens. We provide our findings to local policymakers to
assist them in their continuing efforts to set priorities among
environmental issues.
Monitoring Activities
Our completion of various ambient monitoring programs leads
us to the following conclusions about our approach and about the
substantive environmental issues we addressed in Philadelphia.
-------
17
Insights for Future IEMP Studies
• Our ability to quantify differences between monitoring
and modeling results highlighted the advantages of
using both techniques together to improve exposures
assessments substantially. Modeling, which relates
source releases to ambient pollutant levels, is limited
by the available information on sources and pollutant
loads. Monitoring, on the other hand, provides infor-
mation about pollutants and sources for which one may
know very little, but is not necessarily well-suited to
pinpoint the sources of toxic releases. The two in
combination complement the strengths of each and reduce
the inherent weaknesses of. both when used indepen-
dently.
• Difficult technical and scientific problems in deter-
mining ambient toxic pollutant concentrations can con-
strain attempts to quickly fill in missing data. In
addition, one may be forced to use new and unproven
monitoring techniques to gather the needed information.
For example, our experience in Phase II showed that the
ROSE system is not well-suited for measuring ambient
air releases from landfills. In another example, the
monitoring technique used by AMS to measure formalde-
hyde levels in the ambient air has not yet been proven
to be reliable for quantifying formaldehyde concentra-
tions at low levels representative of ambient condi-
tions. Finally, in the two years since our use of
Tenax for ambient air monitoring, its use, effective-
ness, and reliability have been called into question by
scientists inside as well as outside EPA. EPA is cur-
rently exploring alternatives to Tenax.
Monitoring Results
We gained important insights on environmental exposures from
our examination of benzene, formaldehyde, used oil, and landfill
air emissions. The most important findings are stated below.
Benzene in the Ambient Air
• Average benzene concentrations at the Philadelphia
intersections that we examined ranged from 5.4 micro-
grams per cubic meter (ug/m^) to 22.3 ug/m^. However,
we were unable to distinguish between the influence of
traffic versus gasoline marketing on ambient
-------
18
concentrations. From benzene concentrations at busy
traffic intersections, we calculated rough
approximations of average individual cancer risks
(upper-bound, lifetime) that were in a range comparable
to drinking water and air risks we studied in more
depth.
• A breathing zone study showed benzene concentrations
during 14 refueling episodes that ranged from a low of
1.5 ug/m^ to a high of more than 88,714 ug/m^. Because
of the small numbers of sampling events, these values
are only a limited indication of average benzene con-
centrations from refueling.
Formaldehyde in the Ambient Air
At the four sites monitored, average ambient formaldehyde
concentrations ranged 2.8 ug/m^ to 3.6 ug/m^. These concentra-
tions are significantly lower than those measured in several
major cities nationwide.
• The average ambient formaldehyde concentrations are
below the AMS ambient air quality guideline for formal-
dehyde. However, when we considered the individual
sampling events at each site, we found a few days on
which the ambient concentrations exceeded the guide-
line .
• Our 1984 rough approximations of average upper-bound
lifetime individual cancer risks at each of the four
monitoring locations showed risks roughly comparable to
most of the air risks examined above in the control-
options work. If we employ recent changes (1986) in
the unit-risk factor for formaldehyde, we find the risk
to be in a range that is roughly twice to 30 times
that high. Therefore, we now see that formaldehyde may
be a more significant health concern than many of the
pollutants we looked at in our control-options work.
Combustion of Used Oil
A short-term sampling program (45 samples collected and
analyzed) was conducted in coordination with Philadelphia, the
State of Pennsylvania, and the State of New Jersey. Because of
the limited data, we have not estimated exposure and risk. How-
ever, the data suggest that contaminated fuel oil contains lead
at levels that warrant further investigation. For all samples,
-------
19
the lead concentrations exceeded the recently promulgated EPA
standards for burning used oil. However, the lead concentrations
in used oil will fall dramatically as the lead in gasoline is
phased out.
Air Emissions from Landfills
As noted above, the EPA field monitoring using the ROSE
system experienced serious equipment problems and resulted in
poor results. Better data were collected by New Jersey using
different monitoring equipment. While the New Jersey data are
too limited to assess exposure or risk: adequately, the results do
indicate that a fairly large number of toxic pollutants can be
emitted from landfills and that concentrations of volatile
organic compounds (VOCs) around landfills can be above urban
background levels.
OBSERVATIONS FOR THE FUTURE
The Philadelphia IEMP has shown us that exercises to set
priorities across numerous environmental issues can be managed to
provide useful information to local decision makers. The project
has also revealed important practical limitations in our ability
to apply the IEMP methodology. Three of the most important les-
sons we learned are discussed below.
• It is essential to define the objectives and scope of
the project in its early stages. In Philadelphia we
learned that, because of methodological and resource
constraints, we cannot analyze every environmental
issue. It is therefore important to identify at the
beginning of an IEMP the areas of greatest concern to
the project participants and to set priorities among
this subset of issues for analysis. This allows the
limited resources available to be used well on a man-
ageable set of topics, rather than superficially on
every potential issue in a community.
• The data and analytical methods available are very
significant influences on the topics we analyze and how
we analyze them. Major advances in methods for assess-
ing noncarcinogenic risks and ecological damage could
greatly increase the ability of lEMPs to aid local
decision makers in priority setting. Some environ-
mental concerns may be given a low priority because we
either know little about them or do not currently have
-------
20
the capability to analyze these issues. Users of this
analysis should not misconstrue our results as state-
ments about priorities based on a comprehensive
assessment.
When we have reasonable knowledge about an issue, we
can conduct monitoring or perform analyses that may
shed new light on a subject. Or, when data is the only
concern, we can collect some data even with our limited
resources if it appears that the increased base of
knowledge will improve local decision making. For the
lEMPs, methodological problems can be much more
intractable, especially on very complex topics.
Two new methods, beyond the scope of any particular
IEMP, would greatly enhance our assessments: (1) an
ability to quantify the noncarcinogenic risks for com-
pounds present in the environment at concentrations
above the threshold, in ways analogous to carcinogenic
risk assessment, and (2) an ability to assess ecosystem
effects. Estimates of noncancer effects would allow
comparisons between cancer and noncancer problems and
provide insightful information for those responsible
for public health protection policies. We then could
treat noncancer health concerns more equally with can-
cer concerns, which often seem more important because
they can be quantified. Methods for quantitatively
measuring risks to ecosystems currently do not exist in
any very sophisticated form. It is, therefore, diffi-
cult to assess very rigorously the cost-effectiveness
of control options that reduce ecological risks or to
clarify for local decision makers the implications of
environmental issues that span cancer, noncancer, and
ecological concerns.
Finally, we found it essential to have representatives
from state and local public health and environmental
agencies participate in the project. Local participa-
tion in this IEMP greatly enhanced the quality of the
analysis performed. It also increased the amount of
available data, facilitated the communication of com-
plex issues (such as risk assessment), and built credi-
bility into the project. Also, since we cannot practi-
cally analyze all environmental issues, local involve-
ment ensures that our limited resources are used for
priority setting and issue evaluation in the areas most
critical to the host community. We believe direct
local involvement in this and future projects is vital
to achieving sound results, essential to gaining local
-------
21
acceptance of the project findings, and critical for
using the IEMP approach to aid better environmental
policymaking in specific geographical locations.
------- |