United States
              Environmental Protection
              Agency
              Office of Air Quality
              Planning and Standards
              Research Triangle Park NC 27711
EPA-450/1-90-0043
September 1990
&EPA
CANCER RISK
FROM OUTDOOR
EXPOSURE TO
AIR TOXICS
              Volume I
              Final Report

-------

-------
                                       EPA-450/l-90-004a
                                       September  1990
     CANCER RISK FROM OUTDOOR EXPOSURE
               TO AIR TOXICS
                  VOLUME I

                FINAL  REPORT
    U.S. Environmental Protection Agency
Office of Air Quality Planning and Standards
Research Triangle Park,  North Carolina  27711

-------

-------
                            ACKNOWLEDGEMENTS

      This report was prepared under the guidance of Joseph Padgett,
U.S. Environmental Protection Agency, Office of Air Quality Planning and
Standards, Research Triangle Park, North Carolina.  Key individuals
assisting Mr. Padgett in the development of this report were Bob Faoro
(Monitoring and Reports Branch), Fred Hauchman (Pollutant Assessment
Branch), and Tom Lahre (Pollutant Characterization Branch).

      Several offices within EPA provided technical guidance and
support.  Special recognition is due to Penny Carey (Office of Mobile
Sources) for her assistance on the mobile sources portion of the report.
Others who contributed welcomed support include Brenda Riddle, Warren
Peters, Bob Lucas, Scott Voorhees, Beth Hassett-Sipple, K.C. Hustvedt,
Joellen Lewtas, Ila Cote, Shiva Garg, and James Hardin.

      Finally, the analysis for this study was conducted by Pacific
Environmental Services, Inc., Durham, North Carolina.  Special
recognition is due to Ken Meardon, who performed the analysis and
authored the report.  Assistance to Mr. Meardon was provided by Karin
Gschwandtner.

-------

-------
                            TABLE  OF  CONTENTS
ACKNOWLEDGEMENTS

GLOSSARY

ACRONYMS
iii

 xi

xix
CHAPTER ES    EXECUTIVE SUMMARY
                Technical Approach
                  Sources of Information
                  Methodology
                  Pollutant and Sources Not Evaluated
                  Additive Risk
                Results
                  Magnitude of the Problem
                  Nature of the Cancer Risk
                Comparison with 1985 Six-Month Study
ES-3
ES-3
ES-3
ES-6
ES-6
ES-6
ES-7
ES-10
ES-13
CHAPTER 1.0    INTRODUCTION
                Background
                Purpose of Current Study
                Other Studies or Reports on Air Toxics
                   Indoor Air Pollution
                   Noncancer Health Risk Study
                  -SARA Title III
                Outline of the  Report
 1-1
 1-3
 1-7
 1-7
 1-9
 1-9
 1-11
 CHAPTER  2.0    SCOPE  OF  STUDY AND ANALYSES

                 Data Base
                 Annual  Cancer  Incidence Analysis
                   Methodology
                 Limitations and Uncertainties
                   Limitations
                  'Uncertainties
 2-1
 2-10
 2-13
 2-36
 2-36
 2-40
 CHAPTER 3.0    THE MAGNITUDE  AND NATURE  OF  THE  CANCER RISK

                 Magnitude  of the Cancer Risk Problem
                   Annual Cancer Cases
                   Lifetime Individual Risk
  3-2
  3-2
  3-7

-------
                      TABLE OF CONTENTS  (concluded)
 CHAPTER 3 continued
                 Nature of the Cancer Risk Problem
                   Individual  Pollutants
                   Source Categories
                   Geographic  Variation
                 Comparison with  the Results  from the  1985
                 Six-Month Study
                   Magnitude of the  Problem
                   Nature of the  Problem
3-21
3-21
3-24
3-31

3-50
3-50
3-55
CHAPTER  4.0    SUMMARY AND  CONCLUSIONS

                Magnitude  of  the Cancer  Risk
                  Annual Cancer Incidence
                  Lifetime Individual Risks
                Nature of  the Cancer Risk
                  Individual  Pollutants
                  Sources
                  Geographic  Variability
4-1
4-1
4-3
4-4
4-4
4-6
4-9
APPENDIX A    COMMENTS RECEIVED ON THE EXTERNAL REVIEW DRAFT

APPENDIX B    CANCER RISK REDUCTION ANALYSIS FOR SELECTED POLLUTANTS

APPENDIX C    SUMMARIES OF POLLUTANT-SPECIFIC AND SOURCE-SPECIFIC
              STUDIES (Including Noncancer Health Risk Project on Air
              Toxics)

-------
                            LIST OF TABLES
1-1

2-1
2-2
2-3

2-4

2-5

2-6
2-7

2-8

2-9


2-10

2-11
3-1

3-2

3-3



3-4

3-5

3-6

3-7
Summary of Estimated Nationwide Annual
  Cancer Cases by Pollutant

Main Conclusions of the 1985 Six-Month Study

List of Reports Used in Study
EPA Source Category and Pollutant Studies
Number of Pollutants Included in Cancer Incidence
  Estimates, by Study
Number of Studies that Included Specific
  Pollutant in Cancer Risk Estimate, by Pollutant
Distribution of Source Categories by Number
  of Studies
Unit Risk Factors Used to Compare Cancer Risks
Unit Risk Factors Used to Estimate Cancer Risk
  from PIC
Pollutants with Unit Risk Factors Different
  from Those Used in this Report
Effect of Changes in Unit Risk Factors Used
  in this Report on Original Estimates of
  Annual Cancer Cases
Effect of Unit Risk Factors on Estimated Annual
  Cancer Cases:  The South Coast Study
Selected Limitations of Modeled and Ambient-
  Measured Concentrations for Estimating
  Cancer Risk

Summary of Estimated Nationwide Annual
  Cancer Cases by Pollutant
Summary of Maximum  Individual Risks of Cancer
  as Reported  in the Various Studies
Distribution  of Maximum Lifetime  Individual
  Cancer Risks to the  Most  Exposed  Individual
  from Hazardous Waste Combustors-Boilers  and
  Furnaces
Maximum Lifetime  Individual Cancer  Risks from
  Coke Oven  Emissions
Distribution  of Maximum  Individual  Cancer  Risk
  at 22 Drinking Water Aerators
Areawide Lifetime  Individual Cancer Risks  for
  Selected  Cities
Summary of  Lifetime Individual  Cancer Risks
  for  Selected Cities
Page


ES-8

 1-4

 2-3
 2-5

 2-7

 2-8

 2-11
 2-19

 2-23

 2-26


 2-27

 2-28


 2-31


 3-3

 3-10



 3-16

 3-17

 3-18

 3-19

 3-20
                               vn

-------
 3-8

 3-9


 3-10


 3-11

 3-12

 3-13


 3-14


 3-15

 3-16


 3-17


 3-18

 3-19


 3-20


 3-21


 3-22


 3-23

 3-24


 3-25

3-26

3-27
3-28
          LIST OF TABLES .(continued)

 Relative Contribution of Individual Pollutants
   to Total  Estimated Cancer Cases                    3-22
 Summary of Estimated Cancer Cases Based on
   Modeled Ambient Concentrations, by Source
   Category                                           3-25
 Distribution of Estimated Cancer Cases from
   Secondary Formaldehyde Formation Among Source
   Categories                                         3-30
 Contribution of Area vs. Point Sources to
   Nationwide Annual  Cancer Cases                     3-32
 Comparison  of Measured Ambient Concentrations
   of Selected Pollutants in Selected Cities          3-33
 Intracity Comparison of Ambient Concentrations
  Ug/nr)  for  Selected  Pollutants  in  Three  Cities,
  by Location                                         3.34
 Variation in Annual  Cancer Cases and Cancer
   Rates Due to Exposure to Outdoor Air Toxics
   by Geographic Locales                              3-35
 Variation in  Maximum Lifetime  Individual  Risk,
   by Location                                        3.37
 Estimates of Maximum Lifetime  Individual
   Cancer Risks in  Neighborhoods Surrounding
   Facilities  in the  Kanawha Valley                   3-40
 City-to-City  Variation in  Relative Contribution
   of Selected Pollutants to Total  Annual
   Cancer Incidence                                    3.41
 Maximum Lifetime Individual  Cancer Risks  in
   Baltimore by Individual  Pollutant                   3-43
 Relative Contribution  of Individual  Pollutants to
   Maximum Lifetime Individual  Risk of  Cancer  in
   the Southeast  Chicago  Area                          3-44
 Areawide  Lifetime  Individual Risks  of  Cancer:
   Monitored vs.  Modeled  Ambient Air  Concentra-
   tions  in  Philadelphia                               3-45
 Estimates of  Multi-Pollutant Lifetime  Cancer
   Risks  to  the Most  Exposed  Individual
   to Various  Sources in  Philadelphia                  3-46
 Estimated Cancer Risk  to Maximum  Exposed
   Individuals  to Organic Gases  in  Santa Clara
   for Selected Sources                                3-47
 Areawide  Individual  Risk of Cancer from Lifetime
   Exposure  to Organic  Gases in Santa Clara            3-48
 Estimates of Areawide  Lifetime  Individual Risks
 '  of Cancer Across Area and Point Sources in the
   Kanawha Valley                                     3.49
 Comparison  of Annual  Cancer Cases Per Million
   Population with 1985 Six-Month Study               3-51
Comparison of Annual  Cancer Cases with 1985
  Six-Month Study                                    3-52
Comparison of Unit Risk Factors                      3-54
Contribution of Sources to Estimated Annual
  Cancer Cases and Areawide Lifetime Individual
  Risks in Southeast  Chicago                         3-59

-------
                                             LIST  OF  FIGURES
                Figure No.                    Title                               Page

                  ES-1        Relative Contribution to Total Estimated
                                Nationwide Cancer Cases Per Year, by
                                Pollutant                                         ES-11
                  ES-2        Relative Contribution by Source Categories
                                to Total Estimated Nationwide Cancer
                                Cases Per Year                                    ES-12

                  1-1         Relationship of this Study to Other Air
                                Toxic Risk Studies                                 1-8

                  2-1         Illustrative Relationship of Pollutants
                                and Source Categories Covered by Five
                                Hypothetical Studies                               2-15

                  3-1         Relative Contribution to Total Estimated
                               Nationwide Cancer Cases Per Year, by Pollutant      3-23

                  3-2         Relative Contribution by Source Categories to Total
                              Estimated Nationwide Cancer Cases Per Year           3-27
_

-------

-------
                                GLOSSARY


Ambient fair) monitoring.  The collection of ambient air samples and the
analysis thereof for air pollutant concentrations.

Acute exposure.  One or a series of short-term exposures generally lasting
less than 24 hours.

Additivltv.   A  pharmacologic  or  toxicologic  interaction in  which the
combined effect  of two  or more chemicals is approximately  equal  to the
sum of  the  effect of each chemical alone.   (Compare  with:   antagonism,
synergism.)

Adverse  effect.    A  biochemical  change,  functional  impairment,  or
pathological lesion that either singly or in combination adversely affects
the performance  of the  whole organism,  or reduces an organism's ability
to respond to  an  additional  environmental challenge.

Aggregate  risk.    The  sum of  individual  increased risks of  an adverse
health  effect  in  an exposed  population.

Annual  incidence.  The  number of new  cases  of  a  disease  occurring or
predicted to occur in a  population over  a year.

Antagonism.    A  pharmacologic  or  toxicologic  interaction  in  which the
combined effect  of two chemicals  is less than  the sum of the effect of
each  chemical  alone; the  chemicals  either interfere  with  each other  s
actions, or  one  interferes with the action  of the other.  (Compare  with:
additivity,  synergism.)

Areawide average  Individual  risk.  Average individual  risk to  everyone in
an area (but not  necessarily the actual risk to anyone).   May  be computed
by  dividing lifetime aggregate incidence  by the population  within the
area.

Areawide  incidence.   Incidence over  a broad  area,  such as  a city or
county, rather than at  a particular location,  such as  an  individual  grid
cell.

Background.    A  term   used   in  dispersion  modeling  representing the
contribution  to  ambient concentrations from  sources  not specifically
modeled.in the analysis, including natural  and  manmade sources.

Bioassav.   A test conducted in living organisms  to determine  the  hazard
of potency of a  chemical by its effect  on  animals,  isolated  tissues,  or
microorganisms.

Box model.  A  simplified modeling technique  that assumes uniform emissions
within an  urban area and uniformly mixed concentrations within a specified
mixing depth.

 Cancer.   A malignant new growth.  Cancers are  divided into  two  broad
 categories:  carcinoma and sarcoma.
                                    XI

-------
 Carcinogenic.  Able to produce malignant tumor growth.  Operationally most
 benign tumors are usually included also.

 Carcinogenic process.   A series of  stages  at  the cellular  level  after
 which cancer will  develop  in an organism.  Some  believe there are at least
 3 stages,  initiation, promotion, and progression.  While hypothesized as
 staged process, little is known about specific  mechanisms of action.

 Chronic exposure.   Long-term exposure  usually  lasting  six months to  a
 lifetime.

 Comparative potency  factor.   A  cancer unit risk  factor for a  complex
 substance or mixture that is  extrapolated  from human  risk data for  a
 reference substance and  the ratio of short term bioassay responses of  the
 complex substance  to the reference  substance.   The  EPA  is  developing
 comparative potency factors  for various classes of  POM.

 Confidence limit.   The confidence interval  is a range  of values that  has
 a specified probability  (e.g., 95 percent) of  containing a given parameter
 or characteristic.  The  confidence limit referees to the  upper value of
 the range (e.g.,  upper confidence limit).

 Criteria  pollutants.  Pollutants defined pursuant to Section  108  of  the
 Clean  Air Act and  for which  national  ambient air quality  standards  are
 prescribed.  Current  criteria pollutants include particulate matter,  SO  ,
 NOX, ozone, CO and  lead.                                              x

 Dispersion modeling.  A means of  estimating ambient  concentrations at
 locations  (receptors) downwind of a source, or an array of sources, based
 on  emission rates, release specifications and meteorological factors such
 as  wind speed,  wind direction, atmospheric stability,  mixing  height  and
 ambient temperature.

 Dose-response relationship.  A relationship between:   (1)  the dose, often
 actually   based  on  "administered  dose"  (i.e.,  exposure)  rather than
 absorbed  dose,  and  (2)   the  extent  of toxic  injury  produced by that
 chemical.  Response  can  be expressed either as the severity of injury or
 proportion of exposed subjects affected.  A dose-response  assessment is
 one of  the four steps in a risk assessment.

 Excess  risk.  An  increased risk  of disease  above the  normal  background
 rate.

 Exposure.   Contact of an  organism with a chemical, physical,  or  biological
 agent.  Exposure is quantified  as the amount of the agent  available  at the
 exchange boundaries of the organism (e.g., skin, lungs, digestive tract)
 and available for absorption.

 Exposure  assessment.    Measurement  or estimation  of  the  magnitude,
 frequency,  duration  and route  of  exposure  of  animals  or ecological
 components to substances  in the environment.  The exposure assessment also
describes   the nature  of  exposure  and  the size and nature of the exposed
populations, and is one of four steps in risk assessment.
                                   xn

-------
Human  Exposure Model  (HEM).    A mathematical  model  used in  exposure
assessments for  toxic  air pollutants to  quantify the number  of people
exposed to  pollutants  emitted by  stationary sources and  the  pollutant
concentrations  they   are   exposed   to.      Input  data   include  plant
characteristics such as location, emission,  parameters,  etc.  as well  as
Bureau  of Census data used  in  the estimation  of persons exposed  and
appropriate meteorological data.

Incidence.  The number of new cases of a disease within a specified time
period.  It is frequently presented  as the number  of new cases per 1,000,
10,000, or  100,000.   The incidence ra.te is a  direct estimate  of  the
probability  or risk of developing  a  disease  during a  specified  time
period.

Individual risk.  The  increased  risk for  a  person exposed to a specific
concentration  of  a  toxicant.   May  be expressed  as a lifetime individual
risk or as an annual individual risk, the  latter usually computed as 1/70
of the  lifetime risk.

Lifetime.  Covering the lifespan of an organism (generally considered 70
years  for humans).

Limited evidence.   According to the  USEPA  carcinogen  risk assessment
guidelines,  limited evidence is  a collections  of  facts  and  accepted
scientific  inferences  that suggests the  agent may be causing an effect
but the suggestion  is  not  strong enough to be an  established fact.

Lowest-observed-adverse-effect level (LOAED. The 11owest dose or exposure
level  of a  chemical   in  a study at which  there  is  a statistically or
biologically  significant  increase  in  the  frequency or  severity  of an
adverse effect in the  exposed population as  compared  with  an appropriate^
unexpected  control  group.

Lowest-observed  effect level  fLOEU.   In  a study,  the  lowest dose or
exposure  level at which a statistically or biologically significant effect
is  observed  in  the  exposed population  compared with  an  appropriate
unexposed control group.

Malignant.   A condition  of a neoplasm (tumor)  in which it has  escaped
normal growth regulation  and  has demonstrated the ability to invade local
or  distance  structures,  thereby disrupting  the  normal  architecture or
 functional  relationship of the tissue  system.

Maximum individual  risk (MIR).  The increased risk  for  a  person exposed
 to  the highest measured or predicted concentration of a  toxicant.

 Maximum likelihood estimate  fMLE^.  A statistical best estimate of  the
 value of a parameter from a given  data set.

 Mobile source.  Any motorized vehicle, such as  cars,  trucks,  airplanes,
 trains.  Sometimes refers specifically to highway vehicle sources.
                                   xm

-------
 Monitoring.    The  collection  and  analysis  of  ambient  air  samples.
 Sometimes refers specifically to  just  sampling and not to analysis.   Can
 also refer to source (stack)  sampling.

 Motor vehicle.   On-road or off-road  cars,  trucks or motorcycles.

 Multistage model.    A mathematical  function  used to  extrapolate  the
 probability of  incidence  of disease from a bioassay in animals using high
 doses,  to that  expected to be observed at the low  doses  that  are  likely
 to be found in  chronic human  exposure.   This model is commonly used  in
 quantitative carcinogenic  risk assessments  where the chemical agent  is
 assumed to  be   a  complete carcinogen  and  the   risk  is   assumed  to  be
 proportional  to the dose  in the low  region.

 Hutaoenlc.   Ability to cause  a permanent change  in  the structure of DNA.
 More specific than,  but often  used interchangeably  with,  genotoxic.

 Noncancer  risk.   Risk of  a health effect other than cancer.

 Nonthreshold toxicant.  An agent considered to produce a toxic effect from
 any  dose;  any level  of exposure is deemed to involve some risk.  Usually*
 used only  in regard  to carcinogenesis.

 Nontradltional  sources.   Sources  not  usually included  in  an emission
 inventory,  such  as  wastewater treatment  plants,  groundwater aeration
 facilities, hazardous waste combustors,  landfills, which are  air emitters
 due  to  intermedia  transfer from water or solid waste.

 No-observed-adverse-effect level (NOAEU.  The highest experimental dose
 at which there  is  no  statistically or biologically  significant increases
 in frequency or severity of adverse health effects, as seen in the exposed
 population  compared  with  an appropriate,  unexposed population.  Effects
 may  be produced at this level,  but they  are not considered to be adverse.

 No-observed-effect  level  fNOEn.   The highest experiment dose at.which
 there  is  no statistically or  biologically  significant increases  in
 frequency or severity  of toxic effects seen  in the  exposed compared with
 an appropriate,   unexposed  population.

 Normalized  modeling.   Modeling  of unit  weights  (e.g.,  1  Mg/yr)  of
 emissions from each source, rather than modeling  of  actual   emissions, and
 displaying incremental receptor concentrations or receptor coefficients.
 Thereafter, the  resulting  normalized receptor coefficients are adjusted
 by actual emission rates to simulate different emission scenarios rather
 than re-running the model   over and over with different emissions totals.
This process assumes linearity between  emissions and modeled ambient air
concentrations,  which does not always hold if stack and exhaust parameters
change.

Photochemlcallv formed pollutant.   A secondarily formed pollutant due to
atmospheric photochemistry.  Some  examples  are formaldehyde and PAN.

Potency.  A comparative expression of chemical or drug activity measured
in terms  of the relationship  between  the incidence  or  intensity  of a
                                  xiv

-------
particular effect and  the  associated  dose of a chemical, to  a  given or
implied standard or reference.

Receptor.  A particular point  in space where a monitor is located or where
an exposure or risk is modeled.

Receptor grid. An array of receptors.   Generally synonymous with network.

Receptor modeling.   A technique  for  inferring source culpability  at a
receptor(s)  by  analysis  of the  ambient  sample composition.  There are
various receptor models  employing microscopic and  chemical  methods for
analysis.

Reference dose (RfD).  An estimate (with uncertainty spanning perhaps an
order  of  magnitude)   of  the  daily  exposure  to  the  human  population
(including  sensitive  subpopulations)   that is  likely  to  be  without
deleterious effects during a  lifetime.   The RfD  is reported in  units of
mg  of  substance/kg  body weight/day   for  oral   exposures,  or  mg  of
substance/m3 of air breathed  for inhalation exposures.

Risk.   The  probability  of   injury,   disease,  or  death  under  specific
circumstances.  In  quantitative terms, risk is expressed in values ranging
from zero  (representing  the  certainty that harm will  not  occur)  to one
(representing the certainty that harm will occur).

Risk  assessment.    The  scientific  activity  of  evaluating  the  toxic
properties of a  chemical  and the conditions of human  exposure  to it in
order  both  to  ascertain  the likelihood  that exposed  humans  will  be
adversely  affected,  and  to characterize the nature of the  effects that
they may experience.  May contain some or all of the following four steps:


      Hazard  identification.   The  determination  of whether a particular
      chemical  is  or  is not  causally  linked   to   particular  health
      effect(s).

      Dose-response assessment.  The determination of the relation between
      the magnitude of exposure and the probability of occurrence of the
      health  effects in question.
      Exposure  assessment.
      exposure.
The determination  of the . extent of  human
      Risk characterization.  The description of the  nature and often the
      magnitude of human risk, including attendant uncertainty.

Risk characterization.  The  final  step  of  a risk assessment, which is a
description of the nature and often the magnitude of human risk, including
attendant uncertainty.

Risk management.   The.decision-making process  that  uses  the results of
risk assessment to  produce  a decision about environmental  action.  Risk
management  includes  consideration  of   technical,   scientific,   social,
economic,, and political information.
                                   xv

-------
Route  of exposure.  The  means  by which toxic agents  gain  access to an
organism (e.g.,  ingestion,   inhalation,  dermal   exposure,  intravenous,
subcutaneous,  intramuscular,  intraperitoneal administration).

Scoping  study.  Also known as screening study.  An assessment of analysis
using  tentative or preliminary  data  whose results are  not accepted as
absolute indicators of risk  or  exposures,  but rather,  are taken as an
indication of  the  relative importance  of various  sources, pollutants and
control  measures.   Most urban air  toxics  assessments  conducted to date
have been considered to be scoping studies,  useful for  pointing out where
more detailed  work is needed  prior  to  regulation.

Species  profile.   A  set of apportioning  factors that  allow  one  to
subdivided VOC or PM emission  totals into individual  chemicals or chemical
classes.  Generally, species  profiles  are multiplicative in nature.

Subchronic exposure.   Exposure  to a substance spanning approximately 10
percent  of the lifetime of organism.

Synergism.   A pharmacologic or toxicologic  interaction   in which  the
combined effect of two  or more  chemicals  is greater than the sum of the
effect of each chemical alone.   (Compare with:  additivity, antagonism.)

Threshold Limit Value fTLV).   The concentration of a  substance below which
no  adverse  health  effects  are  expected to occur  for workers  assuming
exposure for 8 hours per  day, 40 hours per week.   TLVs are published by
the American  Conference of Governmental  Industrial  Hygienists  (ACGIH).
This listing may be useful in  identifying substances  used in  the workplace
and having the potential to be emitted into the ambient air.

Threshold toxicant.  A substance showing an  apparent level of effect that
is a minimally effective dose, above which  a response  occurs; below that
dose no  response is expected.

Transformation.  The conversion, through chemical or physical processes,
of one compound or several compounds into other compounds as a result of
aging and irradiation in the  atmosphere.

Transport.   The  movement of pollutants by  wind  flow.    Transport  is
characterized  for  modeling purposes by wind speed and wind direction.

Unit  cancer  risk.   A  measure   of  the probability of  an  individual's
developing cancer  as  a result of exposure to a  specified  unit ambient
concentration.  For example,  an inhalation  unit cancer risk of 3.0 x
10   near a point  source  implies that if 10,000  people  breathe  a given
concentration  of a carcinogenic  agent (e.g., 1  /ig/m3) for 70 years, three
of the 10,000 will  develop cancer as a  result of this exposure.  In water
the exposure unit  is usually  1 /tg/1, while  in air it is 1
Wei ght-of -evi dence .   The  extent to which the  available  biomedical  data
support the hypothesis that a substance causes an effect in humans.  For
example, the  following factors  increase  the weight-of-evidence  that  a
chemical poses  a  hazard  to humans; an increase  in  the  number of tissue
                                   xvi

-------
sites affected by the agent;  an  increase in the number of animal species,
strains, sexes, and  number of experiments  and  doses  showing a response;
the occurrence of a clear-cut dose-response relationship as well as a high
level of statistical  significance in the occurrence of the adverse effect
in  treated  subjects  compared with untreated  controls;  a  dose  related
shortening of the time of occurrence of the adverse effect;  etc.
                                   xvi i

-------

-------
                                ACRONYMS
ATERI'S
BaP
BID
CAG
CDD/CDF

DOE
EDB
EDC
EPA
HEM
IACP
IEMP
LOAEL
LOD
ME I
MIR
MLE
MWC
 NESHAP

 NRC
 OAQPS
 PCB
 PIC
 POHC
 POM
Air Toxic Exposure and Risk Information System
Benzo(a)pyrene
Background information document
Carcinogen Assessment Group
Chlorinated dibenzo-p-dioxins and chlorinated
dibenzofurans
Department of Energy
Ethylene dibromide
Ethylene dichloride
Environmental Protection Agency
Human  exposure model
Integrated Air Cancer  Program
Integrated Environmental Management  Project
Lowest-observed-adverse-effect -1 eve!
Limit  of detection
Maximum exposed  individual
Maximum individual  risk
Maximum likelihood  estimate
Municipal  waste  combustor
 National emission standard for hazardous air
 pollutants
 Nuclear Regulatory Commission
 Office of Air Quality Planning and Standards
 Polychlorinated biphenyl
 Products of incomplete combustion
 Principal organic hazardous constituents
 Polycyclic organic matter
                                    xnx

-------
 POTW
 PUL
 RCRA
 SARA
 SHED
 STAPPA/ALAPCO

 STB
 TCCD
 TSDF

TSP
VOC
g/mi 1 e
 Publicly owned  treatment  works
 Plausible upper limit
 Resource Conservation  and Recovery Act
 Superfund Amendments and  Reauthorization Act
 SAI Human Exposure Dosage Model
 State and Territorial  Air Pollution  Program
 Administrators  and the Association of Local Air
 Pollution Control Officials
 Science  and Technology Branch
 Tetrachlorinated dibenzodioxin
 Treatment, storage, and disposal facilities (for
 hazardous waste)
 Total  suspended particulates
 Volatile  organic compound
microgram per cubic meter
grams  per mile
                                   xx

-------
                           EXECUTIVE SUMMARY
      This report presents an analysis of cancer risks in the United
States from outdoor exposures to airborne toxic pollutants.   It is
intended to provide updated information to suggest priorities for air
toxics control.  This study is an update of an EPA report issued in 1985
entitled The Air Toxics Problem in the United States:  An Analysis of
Cancer Risks for Selected Pollutants (EPA-450/1-85-001,  May 1985), known
as the "Six-Month Study."

      This analysis is based primarily on information  derived from
recent studies and reports.  Results are expressed as  cancer risk from  .
individual pollutants and source categories in terms of excess lifetime
individual cancer risks1  and nationwide annual  cancer  cases.

      Health risks due to indoor exposure and noncancer health effects
resulting from outdoor exposure are not included in this analysis, but
are addressed in separate studies.2  Risks from indoor exposures to
certain pollutants can be significant because of higher indoor
concentrations and the fact that most people spend much of their time
indoors.  Noncancer risks from outdoor exposure also may be significant,
but more information is needed to adequately quantify these risks.

      About 90 toxic air pollutants and 60 source categories were
addressed in one or more of the studies examined.  Additional risks
associated with other pollutants and sources are not characterized.  Of
particular concern is the absence of information on pollutants
secondarily formed in the atmosphere.  Only one (formaldehyde) is
considered in this analysis.

      Significant uncertainties are associated with estimating risk.
These are due to both data limitations and assumptions inherent in our
current risk assessment methodology and the methodology required to
combine and extrapolate information from  individual studies to develop .
national estimates.

      Assumptions about cancer potencies  of various chemicals or
chemical mixtures are generally considered to overestimate the risk, as
do some assumptions about exposures.  Uncertainties such as those due to
missing pollutants, uncharacterized sources, long-range transport of
     1 "Lifetime individual risk" is a measure of the probability that
an individual will develop cancer as a result of exposure to an air
pollutant over a lifetime  (i.e., a 70-year period).

     2 See Report to Congress on Indoor Air Quality (EPA-400/1-89-001,
August,  1989) for current estimates of cancer public health risks from
exposure to  indoor air toxics.   EPA also is evaluating the noncancer
public health risks resulting from short-term and long-term outdoor
exposures to toxic air pollutants.  This latter study is discussed in
Appendix C of this report.

                                  ES-1

-------
 pollutants, and pollutant transformation in the atmosphere will
 underestimate the risk.

       Major findings on national cancer incidence and lifetime
 individual risk, which are subject to uncertainties and data limitations
 as noted above, are highlighted below.

 Cancer Incidence

       •   Based on the pollutants and source categories examined,  total
          excess cancer cases were estimated to be between 1,700 and
          2,700 per year nationwide.   This is equivalent to between 7 and
          11 cancer cases per year per million population.

       •   Of the approximately 90 pollutants evaluated,  12 accounted for
          over 90% of total  annual  cancer incidence.   Of these,  PIC
          (products of incomplete combustion) were responsible  for about
          35% of the total.   Other major contributors  include 1,3-
          butadiene, hexavalent chromium,  benzene,  formaldehyde,  and
          chloroform.

       •   Motor vehicles accounted for almost 60% of total  cancer
          incidence.  Other  area sources accounted for approximately 15%
          of the total.   Point sources accounted  for the remaining 25% of
          the total  annual cancer incidence.

 Lifetime  Individual  Risk

       •   Maximum lifetime individual  risks  exceeding  10~4 (exceeding 1
          chance in  10,000 of contracting  cancer)  from multi-pollutant
          exposures  were reported in  almost  all studies.   Risks  of 10"3
          or greater from individual  pollutants were reported adjacent to
          various types  of sources.

       •   The  relative contribution of pollutants  and  sources to risk  in
          a  specific urban area  can vary significantly.   However,  the
          areawide lifetime  individual  risks  in urban  areas  from the
          combined exposure  to many pollutants generally  are  in the  10"4
          range,  but varied  from 10~s  to 10"3.  These levels result from
          exposure to emissions  from mobile and stationary sources
          combined.

      The numerical estimates presented in this report  should be  viewed
only as rough  indications of the potential for cancer risk caused by a
limited group  of pollutants found in  the ambient air.   Many of the risks
cited in  this  report are almost  certainly inaccurate  in  an absolute
sense.  The best use of the risk estimates is in describing the broad
nature of cancer risk posed by  these  toxic air pollutants and by making
relative  comparisons of risks between pollutants and sources.

      The technical approach for this study, including a description of
the methodology  and a discussion of uncertainties and  assumptions, is
presented in the next section.  Additional information on major findings
Is provided under Results, and  a comparison with the findings of the
                                  ES-2

-------
1985 Six-Month Study is presented under Comparison With 1985 Six-Month
Study.
TECHNICAL APPROACH

Sources of Information

      This study is based on information contained in 10 area-specific
or national air quality based risk-related reports on air toxics, 14 EPA
source category and pollutant-specific studies, risk assessments per-
formed for the development of National Emission Standards for Hazardous
Air Pollutants (NESHAP), and source specific risk data contained in the
EPA Air Toxic Exposure and Risk Information System (ATERIS) data base.
These reports and studies are described in Chapter 2 of this report.
They represent a much larger data base and more comprehensive coverage
than used for the 1985 Six-Month Study.

      Additional information on air toxics emissions data is being
collected under Title III of the Superfund Amendments and
Reauthorization Act (SARA).  However, in their present form, these data
can not be used to estimate risks.  Therefore, this study does not
present risk estimates based on the SARA Title III emissions data.

Methodology

      Estimates of annual cancer incidence were derived by first
developing estimates of the annual cancer cases per million population
for each pollutant/source category combination (e.g., 1,3-butadiene
emissions from mobile sources) reported in the data sources.  These were
modified as necessary to reflect updated unit risk and emission factors.
Estimates of total nationwide annual  incidence then were calculated, in
most instances, by multiplying the annual cancer cases per million
population by the total U.S. population and then summing across all
pollutant/source categories.  Lifetime individual cancer risk estimates
either were obtained directly from each study or modified based on
updated information.

      Because studies were of varying quality and most were concerned
with specific geographic areas, source categories, and/or pollutants, a
number of factors had to be examined  to evaluate study results before
they could be combined and extrapolated to obtain national cancer
incidence estimates.  These include the geographic scope of the study,
source category definitions, unit risk factors, method of estimating
ambient concentrations  (modeled vs. monitored), and emission estimates.
These factors are discussed below.

      Geographic Scope of Studies.  Cancer rates for a pollutant and
source category were extrapolated to  nationwide estimates based on the
geographic scope of each study examined.  Most pollutant/source
categories were included in at least  one study that was nationwide in
scope and this permitted a direct extrapolation to total nationwide
estimates.
                                   ES-3 .

-------
       A few pollutants and source categories were included only in a
 study of limited geographic scope.  In such instances,  it was determined
 whether the pollutant/source category might be unusually concentrated in
 the area studied or was fairly common across the United States.  This
 information was used to determine how study results could be
 extrapolated to obtain total nationwide estimates.

       Source Category Definitions.  Source category definitions in each
 study were examined to minimize the possibility of  double-counting.
 This was especially difficult for the heating/combustion source category
 because the various studies used different terminology  and not all
 reports clearly indicated what was or was not covered.

       Unit Risk Factors.   The unit risk factor is defined as an estimate
 of the probability that an individual  will  develop  cancer when exposed
 to a pollutant at an ambient concentration of one microgram per cubic
 meter (/tg/nr)  for 70 years.   These are  either  upper-bound  values or
 maximum likelihood values.    The  estimate  of cancer risk  for each
 pollutant,  considering the unit risk factor alone,  is conservative;  that
 is,  while the  actual  risk may be higher,  it is more  likely to  be lower
 and  may even be as low as zero.   The weight-of-evidence that a pollutant
 causes cancer  varies from proven human  carcinogen (e.g.,  benzene)  to
 probable human carcinogen (e.g.,  1,3-butadiene)  to  possible human
 carcinogen  (e.g., vinylidene chloride).   All  were included in  this
 analysis as carcinogens.

       The cancer rates presented in the studies  were updated,  as
 necessary,  based on  common  unit  risk factors  used by EPA.   With one
 exception,  this adjustment  generally had  little  effect  on  the  magnitude
 of the total risk estimated by the various  studies.  The  exception was
 the  South Coast study  where the  estimated  cancer risk was  10 times
 higher than the adjusted  estimate based  on  EPA factors.

       Although the'unit risk factors  used  in this report  come  from EPA
 studies,  not all  of  them  have been officially  approved  by  EPA.   In
 addition, many of the  unit  risk  factors  remain  uncertain  and are subject
 to change as further evidence of carcinogenicity is obtained.   For many
 substances,  this  factor probably  has the greatest potential  for error in
 estimating  cancer risk.   This  is  a significant  issue and  affects
 pollutants  such  as formaldehyde,  vinyl  chloride,  products  of incomplete
 combustion  (PIC4), and diesel particulate (which  is  included with PIC).

       Of particular concern  are  the  unit risk  factors for  PIC mixtures
 since  these mixtures are  responsible for about one-third of the cancer
 cases  estimated  in this study.  While many  unit  risk factors used in
 this study  have  been approved  by  EPA, PIC  is an  important  exception.
     3 "Maximum likelihood estimate" refers to a statistical  best
estimate of the risk.

     4 "PIC" is primarily composed of "polycyclic organic matter" (POM).
Benzo(a)pyrene (BaP), which is used as a surrogate for PIC exposure, is
a component of POM.

                                  ES-4

-------
There are no current EPA-approved unit risk factors for these mixtures
or for individual PIC components, although unapproved unit risk factors
are available for some of the compounds (e.g.,  benzo(a)pyrene).  A
number of methods have been used to estimate the aggregate carcinogenic
potency of various PIC mixtures.  The values and method selected for
this study were based on a review of ten EPA studies that included risk
estimates for PIC.  In view of the potentially high risks associated
with PIC, there is a need for more thorough review to establish an EPA-
approved risk methodology and unit risk factor(s) for PIC mixtures that
can be used in future studies of this type.

      Modeled vs. Measured Ambient Concentrations.  Cancer risk
estimates can be derived from either modeled or measured ambient
concentrations.  Each method has advantages and limitations.  Both
ambient and modeled estimates were given equal weight in estimating
cancer risk unless there were clear reasons to prefer one estimate to
another.

      A limitation of dispersion models is the need for accurate
estimates of pollutant emissions.  A limitation of monitoring for many
analyses is the difficulty of monitoring at enough locations to
characterize the variability of  ambient concentrations.  This is true
primarily for point source analyses.   In urban areas where the object is
to estimate average individual risk over a wide area, the specific
location of the monitor may not  be as  critical.   In this case, the use
of measured ambient concentrations for risk estimation should provide
more credible and reliable results than reliance  on modeled estimates.

      Modeling and monitoring produced similar risk estimates for some
pollutants, such  as for cadmium, methylene chloride, and trichloro-
ethylene.  For others, such as  chloroform, ethylene dibromide, and
formaldehyde, risk estimates based on  measured ambient concentrations
were greater than model-based estimates.

      For formaldehyde, the difference in  results probably  is due to  the
fact that this pollutant  is formed primarily  in the atmosphere from
other volatile organic compounds.  Models  are not yet  available which
can properly account  for  this,  whereas ambient measurements do.  The
reasons  for the  different  results  for  chloroform  and ethylene  dibromide
are not  clear, but  a  likely possibility  is that the modeled results do
not  include all  sources of emissions  of  these pollutants.

      Emission Estimates.  Modeled ambient concentrations,  and therefore
cancer  risk estimates, are directly  proportional  to  source  emissions.
The  quality of emissions  data can  vary significantly.  Three  pollutants
for which large  uncertainties are  associated  with emissions  estimates
are  dioxin, gasoline  vapors,  and hexavalent  chromium.  These  uncer-
tainties are recognized by reporting  the risk from these pollutants  as  a
range.

      The uncertainty in  dioxin emissions estimates  is associated with
emissions from hazardous  waste  treatment,  storage, and disposal  facili-
ties  (TSDFs).  The  range  of  emission estimates  for gasoline vapors  is
                                   ES-5

-------
 due to the uncertainty as to the fraction of vapors that is carcino-
 genic.

       With respect to chromium,  only hexavalent chromium is known to be
 a carcinogen,  but only total chromium is measured.   Information has  been
 developed on the percent of total  chromium emissions that is hexavalent
 for specific emission sources,  such as cooling towers,  and this has  been
 used in modeling studies.  Studies based on measured ambient concentra-
 tions assume that a fraction of the measured chromium is hexavalent,  but
 this is not well-defined.

 Pollutants and Sources Not Evaluated

       Although approximately 90  different toxic air pollutants  and over
 60 source categories were addressed in one or more  of the studies used
 in this report,  there are thousands of airborne chemicals that  are
 potentially toxic,  but have neither adequate exposure nor health  effects
 data.   Also,  reliable quantitative emission estimates remain unavailable
 for many potentially important  source categories.   The  lack of  data  for
 these  pollutant  and source categories could result  in a significant
 underestimate  of risk.

       There also is a lack of information on risks  associated with
 pollutants formed photochemically  in the atmosphere (i.e.,  secondary
 formation).  There  is evidence that the  mutagenicity of mixtures  of  some
 pollutants increases greatly as  they undergo transformation  in  the
 atmosphere,  but  insufficient data  are available to  derive cancer  risk
 estimates.   Data on only one secondarily formed pollutant (formaldehyde)
 are included in  this study.

 Additive  Risk

       Total  nationwide  annual incidence  was  calculated  by summing  the
 risks  for all  pollutants  and source categories.  In  addition, additive
 lifetime  individual  risks  were obtained  by  summing  risks  for different
 pollutants  at  the same  geographic  location.   This is  the  accepted
 approach  and was  used  in  all  of  the studies  reviewed.

       It  should  be  noted that the  assumption  of additivity can  lead to
 substantial errors  in risk estimates  if  synergistic  or  antagonistic
 interactions occur.  Although dose  additivity has been  shown to predict
 the acute  toxicity  of many mixtures  of similar  and dissimilar compounds,
 some marked exceptions  have  been identified.  In some cases, risks would
 be greatly overestimated and, in other cases, greatly underestimated.
The available  data  are  insufficient  for  estimating the magnitude of
these  errors.
RESULTS

      From the foregoing discussion, it is clear that there are numerous
assumptions and significant uncertainties associated with the risk
estimates in this study.  In addition, potential sources of error are
important to recognize and are discussed in detail  in this report.  In
                                  ES-6

-------
spite of these potential sources of error, it was concluded that point
estimates would be a more useful way to compare risks among various
pollutants and sources than by expressing broad ranges of risks.
Nevertheless, .for reasons discussed below, ranges were expressed for
several pollutants.

Magnitude of the Problem

      For the pollutants and source categories examined, the total
nationwide cancer incidence due to outdoor concentrations of air toxics
in the U.S. was estimated to range from approximately 1,700 to 2,700
excess cancer cases per year (see Table ES-1).  This estimate is based
on the most recent available unit risk factors and, in general, 1986
emissions data.  It is roughly equivalent to between 7 and 11 annual
cancer cases per million population (1986 population of 240 million).

      The number of deaths resulting from these projected cancer cases
is unknown.  By way of comparison, the American Cancer Society has
estimated total cancer deaths in the U.S. in 1989 to be 500,000.

      The range of estimated excess cancer cases per year in this study
is due primarily to the following uncertainties:  (1) the unit risk
factor for diesel particulate (which is included in the estimated cancer
risk from PIC); (2) dioxin emissions from TSDFs; (3) the cancer-causing
portion of gasoline vapors; and (4) the fraction of total chromium that
is hexavalent.

      Maximum lifetime individual risks of 1 x 10"4  (1  chance in 10,000
of contracting cancer) or greater were reported in almost all of the
studies examined.  Maximum lifetime individual risk levels exceeding 1 x
10"4  were reported  for multi-pollutant  exposures  from such  sources as
major chemical manufacturers, waste oil incinerators, hazardous waste
incinerators, municipal landfills, publicly owned treatment works
(POTWs), and TSDFs.

      Maximum individual risks of 1 x.10"4 or greater were  reported
adjacent to at least  one source for each of 16 pollutants  included in
the NESHAP/ATERIS data base.  Twelve of these pollutants were estimated
to be responsible for maximum individual risks of 1 x 10   or greater.

      For the urban areas studied, areawide lifetime individual risks
from all pollutants for point and area sources combined generally were
in the 10"4 range,  and ranged from 10"5 to 10"3.   Lifetime  individual
risks in four urban areas6 due to multi-pollutant exposure (9 to 16
pollutants) ranged from 10"4 to  10"3 based on measured ambient
concentrations.  The  contribution of specific area and point sources to
     5 Acetaldehyde, acrylonitrile, arsenic, benzene, 1,3-butadiene,
carbon tetrachloride, chloroform,  hexavalent chromium, coke oven
emissions, ethylene dichloride, epichlorohydrin, ethylene oxide,
methylene chloride,.p-dichlorobenzene, styrene, and  vinylidene  chloride.
     6 Los Angeles, Baton Rouge, Boston, and Chicago.

                                   ES-7

-------
                              TABLE ES-1

  SUMMARY OF ESTIMATED NATIONWIDE ANNUAL CANCER CASES BY POLLUTANT

POLLUTANT
1. Acrylonitrile
2. Arsenic
3. Asbestos
4. Benzene
5. 1,3-Butadiene
6. Cadmium
7. Carbon tetrachloride
8. Chloroform
9. Chromium (hexavalent)
10. Coke Oven Emissions
11. Dioxin
12. Ethyl ene di bromide
13. Ethyl ene di chloride
14. Ethyl ene oxide
15. Formaldehyde
16. Gasoline vapors
17. Hexachlorobutadiene
18. Hydrazine
19. Methyl ene chloride
20. Perchloroethylene
21. PIC
22. Radionuclides0
23. Radon0
24. Trichloroethylene
25. Vinyl chloride
26. Vinyl idene chloride
27. Miscellaneous11
Total s
EPA
CLASSIFICATION8
Bl
A
A
A
B2
Bl
B2
B2
A
A
B2
B2
B2
B1-B2
Bl
B2
C
B2
B2
B2
_b-
A
A
B2
A
C


ESTIMATED ANNUAL
CANCER CASES
13
68
88
181
266
10
41
115
147-265
7
2-125
68
45
6
124'
19-76
9
6
5
6
438-1120
3
2
7
25
10
15
1,726 - 2,706
NOTE:  Values in this figure are not absolute predictions of cancer-
       occurrence and are intended to be used in a relative sense only.
       The dose-response relationships and exposure assumptions have a
       conservative bias, but omissions due to uncharacterized pollutants
       (either directly emitted or secondarily formed) and emission
       sources, the long-range transport of pollutants, and the lack of
       knowledge of total risk from multi-pollutant exposures will offset
       this bias to an unknown extent.
                                   ES-8

-------
FOOTNOTES TO TABLE ES-1

a For a discussion of how EPA evaluates suspect carcinogens and more
  information on these classifications, refer to "Guidelines for
  Carcinogen Risk Assessment" (51 Federal Register 33992).

  The EPA classifications used in this report are:

      A =  proven human carcinogen;  B =  probable human carcinogen (Bl
      indicates limited evidence from human studies and sufficient
      evidence from animal studies; B2 indicates sufficient evidence
      from animal studies, but inadequate evidence from human studies);
      C =  possible human carcinogen

b EPA has not developed a classification for the group of pollutants
  that compose products of incomplete combustion  (PIC), although EPA
  has developed a classification for some components, such as
  benzo(a)pyrene  (BaP), which is a B2 pollutant.

c From sources emitting significant amounts of radionuclides (and
  radon) to outdoor air.  Does not include exposure to indoor
  concentrations  of radon due to radon in soil gases entering homes
  through foundations  and cellars.

d Includes approximately 68  other  individual pollutants,  primarily from
  the TSDF study  and the Sewage Sludge Incinerator study.
                                   ES-9

-------
 these levels of risk generally could not be identified, but the
 relatively narrow range of the areawide lifetime individual risks in the
 urban areas studied suggests that other large urban populations may be
 subject to similar risk levels.

 Nature of the Cancer Risk

       Individual  Pollutants.  Available information suggests that 177  of
 the approximately 90 pollutants included in the data sources may each
 account for risks of at least 10 cancer cases per year.  Of these,  13
 pollutants  may each account for 40  or  more cases  per year.   The
 relative contributions of these pollutants to the total annual  cancer
 cases are shown  in Figure ES-1.
                                                   '  - •.
       The pollutants found in this study to be the primary contributors
 to annual cancer  incidence also were frequently associated with high
 maximum individual  risks.   Other individual compounds,  such as
 epichlorohydrin and styrene, which account for smaller  aggregate  cancer
 incidence,  also are associated with  high individual  risks (greater  than
 1  x lO'4).

       Source Categories.   Many types of sources contribute to aggregate
 incidence and lifetime individual  risk.   Figure ES-2 illustrates  the
 relative contribution to  total  annual cancer incidence  for each of  the
 source categories  evaluated.

       On an  individual  source category  basis,  motor  vehicles were the
 largest  contributor to nationwide  annual  incidence,  contributing
 approximately 58%  of the  total  [including  approximately 35% of  the  total
 contribution  for which secondarily formed  formaldehyde  (shown as  a
 separate category)  is responsible].   Electroplating  (6%)  was another
 large  contributor  as a result of chromium  emissions.  Other major
 contributors  are TSDFs (5%); woodsmoke  (5%);  asbestos,  demolition (4%);
 gasoline marketing  (3%);  cooling  towers  (3%);  and  solvent  use/degreasing
 (3%).

       A  significant  portion  of  the cancer  risk  from  most  sources  usually
 was due  to a  few pollutants,  even  where  a  source emitted many different
 pollutants.   For example,  over  70  pollutants were  included  in the
 analysis  of  hazardous  waste  combustors,  but  only two pollutants (cadmium
 and hexavalent chromium) were estimated  to  be responsible  for almost 90
 percent  of the estimated cancer cases in this source category.
 Similarly, three pollutants  (cadmium, hexavalent chromium,  and  arsenic)
 were responsible for almost  90  percent of  the estimated cancer  cases
 from hazardous waste boilers  and furnaces.
       Acrylonitrile, arsenic, asbestos,  benzene,  1,3-butadiene,
cadmium, carbon tetrachloride, chloroform, hexavalent chromium, dioxin,
ethylene dibromide, ethylene dichloride,  formaldehyde, gasoline vapor,
PIC, trichloroethylene, and vinyl chloride.

     8 Arsenic,  asbestos,  benzene,  1,3-butadiene,  carbon tetrachloride,
chloroform, hexavalent chromium, dioxin,  ethylene dibromide, ethylene
dichloride, formaldehyde,  gasoline vapor, and PIC.

                                 ES-10

-------
o
a.
                  NOTE: Values in this figure are not absolute predictions of cancer occurrence
                      •  and are intended to be used in a relative sense only. The dose-response
                        relationships and exposure assumptions have a conservative bias, but
                        omissions due to uncharacterized pollutants (either directly emitted or
                        secondarily formed) and emission sources, the long-range transport
                        of pollutants, and the lack of knowledge of total risk from multi-
                        pollutant exposures will offset this bias to an unknown extent
o
CO
                                          0)     W

                                         .2     .2

                                                    S •


                                                   '
                                                     -S
                                                  9

42
i
                            3
o
CM
3-Butadiene
um, hexavalent
40
30
2
0
PERCENT CONTRIBUTION
 O
 CD
5
 c
 O
to

T3
 CD
 es
.is
 w   «
LU   j|
lo   "o
2   °-
•"   ^
 2   ^.
 C   CO
 O   CD
 s  t
^2  CD
5  Q.
 5  w
 O  CD
0  %
 |  O

 'I  8
 CD  e
 CC  co
  .  O
ES
                                                                                                     o>
                                                                                                     LL
                           iNvirmod ivnaiAiaNi
                                                     ES-ll

-------
                                          CO
                                          CO

                                         •g  CO
                                          o P
                                          o>>-
                                          CD  *-
                                          CD  CD
                                          O  O>
                                         c  §
                                         §0


                                         II
                                         •fe  C
                                         o .2
                                         O to
                                         CD  °^
                                         &  E
                                         CO
                                         LU


                                         I
                                         D)
                                         U_
ES-12

-------
      Comparing aggregate source categories, mobile sources were
estimated to contribute approximately 58 percent and stationary sources
approximately 42 percent of the total annual incidence.  Area sources
were responsible for approximately 75 percent and point sources 25
percent of the total annual incidence associated with outdoor exposure
to air toxics.

      Geographic Variability.  Ambient concentrations of individual air
toxics vary on a city-to-city basis as well as on an intra-city basis.
For the cities included in this study, the variation among cities ranged
from a factor of 2 for benzene to almost 20 for chloroform.  Similar
variations were found within cities.  Many factors could account for
this.  These include meteorological conditions, the location of sources
relative.to the population, and, where cancer risks were estimated from
measured ambient concentrations, the location of the monitors.
COMPARISON WITH 1985 SIX-MONTH STUDY

      The present study shows approximately 500 to 900 more cancer cases
per year than reported in the 1985 Six-Month Study published in May,
1985.  This apparent increase is due primarily to the inclusion of more
pollutants, a better accounting of emission sources, and, in some cases,
substantial increases in unit risk estimates.

      The present study shows additive lifetime individual risks to be
similar to those estimated in the 1985 Six-Month Study.  However, the
broader scope of the present study has identified additional source
types (e.g., TSDFs, POTWs) that can cause high lifetime individual
risks.

      The individual compounds found in the present study to be the most
important contributors to cancer risk are, for the most part, the same
as those identified in the 1985 Six-Month Study.  The most important
addition is 1,3-butadiene.  Dioxin also may be an important contributor,
but the uncertainty associated primarily with estimates of dioxin
exposure from TSDFs makes it difficult to conclude this at the present
time.  Several pollutants (asbestos, ethylene oxide, and
trichloroethylene) appear to be somewhat less of a factor in terms of
aggregate cancer risk, but not necessarily in terms of maximum
individual risk.

      The 1985 Six-Month Study found that area sources accounted for
over 75 percent and point sources accounted for less than 25 percent of
the total annual cancer incidence.  This finding was essentially
confirmed by the results of the present study.  Findings  in the present
study on the geographic variability of risk also are reasonably
consistent with those in the 1985 Six-Month Study.
                                  ES-13

-------

-------
                            1.0   INTRODUCTION
Background
      The U.S. Environmental Protection Agency (EPA) initiated a broad
"scoping" study in November, 1983, with a goal of gaining a better
understanding of the size and causes of the health problems caused by
outdoor exposure to air toxics.  This broad scoping study, often
referred to as the Six-Month Study1,  was published in May, 1985,  and is
hereafter referred to in this report as the 1985 Six-Month Study.
      The objective of the 1985 Six-Month Study was to assess the
magnitude and nature of the air toxics problem by developing
quantitative estimates of the cancer risks posed by selected air
pollutants and their sources from a national and regional perspective.
It was designed to answer four basic questions:
      1.    What is the approximate magnitude of the air toxics problem,
            as measured by the estimated cancer risks associated with
            air pollution?
      2.    What is the nature of the air toxics problem; that is, what
            pollutants and sources appear to increase the incidence of
            cancer and what are their relative importance?
      3.    Does the cancer risk problem vary geographically and, if so,
            in what ways?
      4.    Are current air toxics data bases adequate for assessing the
            cancer risk from air toxics?  If not, what are the
            significant data gaps?
     1 Haemisegger,  E.  et.  al_.   The  Air  Toxics Problem  in  the United
States:  An Analysis of Cancer Risks for Selected Pollutants.  EPA-450/1-
85-001.  May  1985.  ,
                                   1-1

-------
       These questions  were  answered  primarily  by  conducting  three
 analyses  to estimate cancer incidence  (i.e., cancer  cases  per year)  and
 lifetime  individual risks.2  One analysis estimated national exposure
 and  risks from about 40  pollutants being  considered  for  listing  under
 Section 112 of the  Clean Air Act.3  The risks estimated  in this
 analysis,  which is  referred to  as the  NESHAP (National Emission
 Standards for  Hazardous  Air Pollutants) Study, were  national in  scope
 and  considered emissions obtained from traditional air pollution
 inventories.   The emphasis  of the NESHAP  Study was on large  point
 sources,  but both mobile and area sources were also  covered.  The second
 analysis  provided a more detailed estimate and analysis  of exposure  and
 risk in 35 counties for  approximately  20  pollutants.4  This second
 analysis,  which is referred  to  as the  35-County Study, was designed  to
 examine risk from air  toxics on a more local perspective than the NESHAP
 Study.  The analysis in  the  35-County  Study included sources not usually
 considered in  previous studies, such as publicly owned treatment plants
 (POTWs) and waste oil  combustors.  The third analysis, which is  referred
 to as the  1985  Ambient Air Quality Study, estimated cancer risks based
 on ambient air  quality data  for fourteen pollutants.5  Quantitative risk
       "Lifetime  individual  risk" is  a  measure of  probability  that an
individual will  develop cancer as a  result of exposure  to the ambient
concentration  of an  air  pollutant  over  a lifetime  (i.e.,  a  70-year
period).
     3 Schell,  R.M.  Estimation of  the Public Health Risks Associated with
Exposure to  Ambient Concentrations of 87 Substances.   OAQPS,  U.S. EPA,
July 1984.  Revised February  1985.
     4 Versar;  American  Management System, Inc. Hazardous Air Pollutants:
An Exposure and  Risk Assessment for 35 Counties.   U.S.  EPA Contract No.
68-01-6115, September 1984.
     5 Hunt,  Bill, et..al.  Estimated Cancer  Incidence Rates  from Selected
Toxic Air Pollutants Using Ambient Data.  U.S.  EPA, revised March 1985.

-------
assessments available from other EPA activities for asbestos,  radio-
nuclides, and gasoline marketing supplemented these three analyses in
the 1985 Six-Month Study.  Information available on several  source
categories for which data at that time were insufficient to perform a
quantitative risk assessment were also analyzed and summarized in the
study.  The main conclusions reached in the 1985 Six-Month Study are
summarized in Table 1-1.
Purpose of Current Study
      The primary objective of the current study is to evaluate the
magnitude and nature of the cancer problem associated with outdoor
concentrations of air toxics in the United States.  The magnitude of the
cancer problem is addressed in terms of annual cancer incidence (i.e.,
the number of cancer cases per year nationwide) and lifetime individual
risk  (i.e.,  areawide and maximum individual risk6).  The nature of the
cancer problem is addressed primarily by examining the relative
contributions of pollutants and sources to annual cancer  incidence  and
the geographic variability of cancer risk and  important contributors to
that  risk.   In addition, the results of this study are compared with
those of the 1985 Six-Month Study.  Finally, while the current study
does  not include  a  reevaluation of  EPA's air toxics control strategy,
the study seeks to  present  information on the  magnitude  and nature  of
the air  toxics problem  due  to outdoor  exposure that may  be used to  help
set priorities for  the  control  of air  toxics  and  to better define
research and data needed to  support a  more effective  control  program.
      6 "Areawide" individual risk refers to the average lifetime individual
 risk to everyone  in  an area.   "Maximum"  individual  risk refers  to  the
 maximum level of risk to which a person could be exposed,  and is located
 at a specific point within an area.
                                   1-3

-------
                       TABLE 1-1

     MAIN CONCLUSIONS OF THE 1985 SIX-MONTH STUDY
Nationwide annual cancer cases were estimated as 1,300 to 1,700 (5
to 7.4 cases of cancer per year per million population) for the 15
to 45 pollutants examined in each analysis.

Maximum lifetime individual risks of 1 x 10"4  (1 in ,10,000) or
greater in the vicinity of major point sources were estimated for
21 pollutants, about half of those that were studied.  Maximum
lifetime individual risks of 1 x 10"3  (1  in  1,000) or greater were
estimated for 13 pollutants.
                                                               -3
Additive lifetime individual risks in urban areas due to
simultaneous exposure to 10 to 15 pollutants ranged from 1 x 10
to 1 x 10  .   These  risks,  which  were calculated  from  monitoring
data, did not appear to be directly related to specific point
sources.  Instead, they represent a portion of total risks
associated with the complex pollutant mixtures typical of urban
ambient air.

Thirteen specific pollutants8 were identified as  possibly
important contributors to aggregate cancer cases from air toxics.
Although little aggregate cancer incidence (less than 1 cancer
case per year total) was found for 21 low production organic
chemicals, some of these compounds appear to be associated with
high individual risks.  The low aggregate incidence for these
compounds may be due in part to the' lack of data concerning their
emissions and toxicity.

A wide variety of sources was found to contribute to cancer risk
from air toxics, with combustion/incineration probably the largest
single source of risk.  Among this wide variety of sources were
sources,that have not historically been part of emission
inventories, such as publicly owned treatment works (POTWs) and
hazardous waste treatment,  storage, and disposal  facilities
(TSDFs), which were found to possibly pose important risks in some
locations.

Both point sources (major industrial  sources) and area sources
(smaller sources that may be widespread across a given area, such
as solvent usage and motor vehicles)  appear to contribute
significantly to cancer risk from air toxics.  Large point sources
tended to be associated with many high individual risks, while
area sources appeared to be responsible for the majority of
aggregate cancer cases.
     *
Where it could be analyzed, large city-to-city and neighborhood-
to-neighborhood variation in pollutant levels and sources was
found.
                            1-4

-------
                                   TABLE 1-1  (concluded)

                        MAIN CONCLUSIONS OF THE 1985 SIX-MONTH STUDY
                   Major weaknesses and data gaps in the air toxics data bases at the
                   Federal, State, and local levels made it impossible to accurately
                   characterize most local air toxics problems.  Problems identified
                   with the few available air toxics data bases were inconsistencies
                   and anomalies  in the emission inventories, lack of sufficient data
                   to develop population exposure estimates, and lack of compounds
                   for which adequate health effects tests have been performed.

                   EPA's criteria pollutant15 program appears to have reduced air
                   toxics  levels.  One analysis estimated the cancer rate from 16 air
                   toxics  in 1980 was less than half that for 1970  (6.8 vs.  17.5
                   cancer  cases per year per million population).


        SOURCE:   Haemisegger, E. et. al_.  (1985) pp. 94-96.

        a The thirteen pollutants were:  chromium, arsenic, asbestos, products of
          incomplete  combustion  (PIC), formaldehyde, benzene, ethylene oxide, gasoline
          vapors,  chloroform, carbon tetrachloride, perch!oroethylene,
          trichloroethylene, and vinylidene chloride.

        b  EPA's criteria pollutants are:  carbon monoxide, ozone, lead, total
          suspended particulate, oxides of nitrogen, and  sulfur dioxide.
                                                 1-5
_

-------
      One of the key analyses in this study is based on combining the
risk assessments from a number of information sources and developing an
estimate of total cancer incidence from all of the pollutants and
sources included in the various reports and studies.  Ideally, this
analysis would lead to a point estimate-of total cancer incidence.
While a point estimate seemed reasonable for a large number of
individual pollutants, certain aspects of the risk assessment
methodology for other pollutants did not allow for identifying a point
estimate.  For these other pollutants, only ranges could be identified.
The analysis then tried to narrow the range as much as possible.
      Existing information from various reports and studies available
since the 1985 Six-Month Study was released has been gathered,
organized, and evaluated in order to accomplish these objectives.  Some
of the quantitative estimates of risk used in this study to help
identify high risk air toxics and sources come from studies that are
part of the regulatory decision making process (e.g., background
documents in support of NESHAPs under Section 112 of the Clean Air Act).
Other quantitative risk estimates come from reports or studies that are
of a general "scoping" nature and are not,of the level of detail
necessary for regulatory decisions.  In addition, the risk estimates
contained in these studies and reports are based on an uneven level of
quality, which affects the certainty that one can attach to the risk
estimates.  For these reasons, the quantitative risk estimates can not
be used to support regulatory decisions on source regulation.  These
results should, nevertheless, be useful in developing and evaluating air
toxics control strategies and in establishing priorities within these
strategies.  Since there are limitations in most risk analyses, it is
important for the reader using this report to consider the caveats and
                                   1-6

-------
assumptions associated with the analyses in order to interpret and use
the results properly.
Other Studies or Reports on Air Toxics
      Health risk from air toxics encompasses both cancer and rioncancer
effects, and results from both indoor exposure as well as outdoor
exposure to air toxics.  This report examines cancer risk from outdoor
concentrations of air toxics.  The risk estimates presented in this
study are associated with just one part of the total risk from air
toxics  (see Figure 1-1).  Health risks from indoor exposure to air
toxics  and the noncancer risks from outdoor exposure to air toxics are
the subjects of separate studies, which are discussed below.  Also
discussed are air toxics emissions data recently released under Title
III of  the Superfund Amendments  and Reauthorization Act  (SARA) of 1986.
      Indoor Air Pollution
      Under the Radon  Gas  and  Indoor Air Quality Research Act of  1986
 (Title  IV of SARA),  EPA is  establishing a  research  program  on all
 aspects of  indoor  air  quality. , As part of this  program,  EPA  is  seeking
 to identify  high risk  pollutant  sources and characterize  the  exposures
 and health  risks of various populations to those sources.   Source
 categories  that  have been  identified  are:   environmental  tobacco smoke,
 combustion  appliances, materials and  furnishings,  biological
 contaminants,  consumer products (e.g.,  hair spray,  paint solvents,
 cleaning fluids),  outdoor sources (e.g.,  infiltration of radon,  vehicle
 exhaust, pesticides),  and nonionizing radiation.  The indoor air program
 also addresses generic research activities.  Generic research needs
 emphasize the concept of limiting total exposures and include develop-
 ment of standard measurement protocols, establishment of emission
 reduction baselines, identification of mitigation techniques, and
                                    1-7

-------
               EXPOSURE


            Indoor    Outdoor
    x
    EC
       <
       o
       O
                                         This study
/y///  Indoor study program
/ScS<  Noncancer study
                   "CANCER RISK FROM \
                                        •,


                   OUTDOOR EXPOSURE V


                      TO AIR TOXICS"
Figure 1 -1.  Relationship of this Study to Other Air Toxic Risk Studies
                               1-8

-------
dissemination of information to the public.  The EPA's ultimate goals in
addressing indoor air quality problems are to characterize and
understand the risks to human health that pollutants pose and to reduce
those risks by reducing exposures.  A report to Congress has been
prepared that estimates cancer risk from indoor air toxics.7
      Noncancer Health Risk Study8
      In a separate study, EPA is evaluating the noncancer public health
risks resulting from short-term and long-term outdoor exposure to toxic
air pollutants.  Noncancer effects range from subtle biochemical,
physiological, or pathological effects to gross effects, including
death.  The main focus of the noncancer study is on the evaluation of
risk from exposure to air toxics that are routinely emitted from
industrial or commercial sources.  Excluded from the noncancer analysis
is the consideration of occupational exposures, indoor air pollutants,
criteria air pollutants, secondary atmospheric reaction products, and
accidental releases.  The Executive Summary from the Noncancer Health
Risk study is presented in Appendix C of this report.
      SARA Title III
      Under Title  III of SARA, EPA is collecting air toxics emissions
data from industrial and manufacturing sources that are covered by
certain Standard Industrial  Classification  (SIC) codes, have  10 or more
employees, and  handle listed chemicals above threshold  amounts.  These
data are  collected  as part  of the Toxic  Release  Inventory mandated under
      7  See Report to Congress  on  Indoor Air Quality' (EPA-400/1-89-001,
 August  1989) for  current  estimates of  cancer  public health risks  from
 exposure to indoor air  toxics.
      8 U.S.  Environmental Protection Agency, OAQPS.  Toxic Air Pollutants
 and Noncancer Risks.  Summary of Screening  Study.   External  Review Draft,
 September, 1990.
                                    1-9

-------
Section 313 of SARA.  Based on data contained in EPA's Toxic Release

Inventory Data (1989), approximately 320 toxic chemicals were identified

as being released to the environment and over 2.3 billion pounds of

toxic chemicals were identified as being released to the air from the

reporting facilities in 1987.  Limiting the number of employees and

specifying threshold amounts resulted in excluding smaller producers and

facilities.

      The following items highlight the relationship between the air

toxic emissions data collected under SARA Title III and the risk

estimates presented in this report.

      -  SARA Title III data concern only estimates of air toxics
         emissions and not estimates of cancer risk.  This report
         focuses on estimates of cancer risk from exposure to air
         toxics.  This report does not estimate emissions of air toxics,
         although the studies upon which the risk estimates are drawn
         had to estimate such emissions.

      •  SARA Title III emissions data are limited to industrial and
         manufacturing sources covered by SIC codes 20 through 39.  This
         report is not limited to these sources, but includes
         additional emission sources such as mobile vehicles, treatment,
         storage, and disposal facilities for hazardous wastes (TSDFs),
         and dry cleaners.

      •  Generally, SARA Title III covers many more air toxics than
         this report, which focuses on the subset of pollutants for
         which cancer is the health concern and for which unit risk
         factors are available.

      •  The only information source used in this report that is similar
         to the SARA Title III effort is the Air Toxic Exposure and Risk
         Information System (ATERIS) data base, which includes nation-
         wide emission estimates of many pollutants covered by SARA
         Title III.

      The emission data submitted under SARA Title III were not used in

this study to estimate cancer risk.  The SARA Title III emission data

are not reported in a form that allows estimation of risk.  Thus, these

data could not be used to estimate cancer risk for this study.  However,
                                  1-10

-------
the information on source emissions gathered under SARA Title III may be
useful in identifying sources of concern for future risk assessments.
Outline of the Report        •     •_
      This report is divided into two volumes.   Volume I contains a
glossary of key terms; a list of acronyms; the Executive Summary;
Chapter 1, Introduction; Chapter 2, Scope of Study and Analyses; Chapter
3, The Magnitude and Nature of the Cancer Risk; and Chapter 4, Summary
and Conclusions.  Volume II contains several appendices.  The following
paragraphs describe the remaining chapters of Volume I.  This is then
followed by a brief description of the material contained in Volume II.
      In Chapter 2, "Scope of Study and Analyses," the various reports
and information used, the analytical methodology used to develop
estimates of annual cancer incidence, and major limitations and
uncertainties associated with the risk estimates presented in the study
are discussed.
      The results of the study are presented in Chapter 3, "The
Magnitude and Nature of the Cancer Risk."  The magnitude of risk
estimated, in terms of both estimated annual incidence and lifetime
individual risk, is presented first.  The nature of the cancer risk, in
terms of individual pollutants, source categories, and geographic
variability, is then presented.  The results are then compared with
those reported  in the 1985 Six-Month Study.
      Chapter 4, "Summary and Conclusions," summarizes the results of
the study and presents the conclusions drawn from  it with regard to
the magnitude and nature of the cancer risk from outdoor air toxics.
      As noted  above, Volume  II of this report contains the appendices.
Appendix A lists the  individuals who commented on  the external review
draft of this report  and a summary of their comments.  Appendix  B
                                   1-11

-------
provides detailed summaries of the analyses conducted for determining
the estimates of cancer cases per year per million population that would
be used in estimating total nationwide annual cancer incidence and the
resulting estimates of total nationwide cancer incidence for each of
those pollutants initially identified as possibly resulting in at least
ten cancer cases per year nationwide.  Appendix C provides summaries of
the 14 EPA studies that focused on individual pollutants and source
categories which formed part of the data base.  As noted earlier, the
Executive Summary to the Noncancer Health Risk study is also provided in
Appendix C.
                                  1-12

-------
                    2.0  SCOPE OF STUDY AND ANALYSES

      The purpose of this chapter is to provide the reader with an
overview of the scope of the study,  the analyses performed in estimating
the cancer risk from air toxics, and an understanding of the limits and
uncertainties associated with it.  The scope is described by a
discussion of the data base used.  This discussion identifies for the
reader the various reports and studies included and the pollutants and
source categories examined.  Next, the methodology used to derive the
nationwide estimates of annual cancer incidence is described.  This
description gives the reader an understanding of the major components of
the annual cancer incidence analysis, as well as some of its boundaries.
Following the description of this analysis, limits and uncertainties
associated with the risk estimates presented in this report are
identified.  By keeping in mind the scope of the study and the limits
and uncertainties associated with these risk estimates, the reader will
be able to more properly interpret and use the results of the study.
Data Base
      A number of reports dealing with air toxics have been completed by
EPA or other agencies since the 1985 Six-Month Study was published.  A
list of these reports was compiled and circulated to EPA Regional
Offices, the State and Territorial Air Pollution Program Administrators
and the Association of Local Air Pollution Control Officials (STAPPA/
                                   2-1

-------
ALAPCO), and others to identify any additional reports that might be
included in this study.  In addition to the reports, information from 14
individual source category- and pollutant-specific studies being
conducted by EPA was included in this study.  Two of these studies (the
Superfund study and the Woodstove study) did not provide estimates of
cancer risk that could be used to estimate nationwide cancer risk.  Risk
estimates based on the NESHAP (National Emission Standards for Hazardous
Air Pollutants) analysis used for the 1985 Six-Month Study and
supplemented by data contained in the Air Toxic Exposure and Risk
Information System (ATERIS) data base developed by the EPA's Office of
Air Quality Planning and Standards (OAQPS)1 were also used in the
analysis.  As a result, the magnitude and nature of the cancer risk
posed by air toxics were evaluated based upon information contained in
ten reports, twelve source category- or pollutant-specific studies, and
the NESHAP/ATERIS data base.  The ten reports are listed in Table 2-1
and the fourteen source categories and pollutants for which information
was obtained from other EPA studies are listed in Table 2-2.
      The purposes of these reports and studies vary.  Some were under-
taken as general scoping studies to estimate cancer risk from air toxics
in a specific locale (e.g., the"Integrated Environmental Management
Project (IEMP) studies, the South Coast Air Basin study) or on a
national basis (e.g., the Mobile Source study, the Ambient Air Quality
study).  Some studies were undertaken to estimate cancer risk from a
specific source category (e.g.,  publicly owned treatment works (POTWs),
sewage sludge incinerators, mobile sources) or a specific pollutant
     1  This is referred to in this study as the NESHAP/ATERIS data base.
The NESHAP risk estimates  from  the  1985 Six-Month Study were updated by
applying  new unit  risk factors  for those  pollutants  whose  unit  risk
factors had changed since the original analysis.
                                  2-2

-------
                               TABLE 2-1

                     LIST OF REPORTS USED IN STUDY
 1.    U.S.  EPA,  Region III.   Kanawha Valley Toxics Screening Study,
      Final  Report.   July 1987.   (lEMP-Kanawha Valley)

 2.    U.S.  EPA,  OPPE.  Santa Clara Valley Integrated Environmental
      Management Project:  Revised Stage I Report.  May 30,  1986.
      (lEMP-Santa Clara)

 3.    U.S.  EPA,  OPPE.  Baltimore Integrated Environmental  Management
      Pro.iect:   Phase I Report.   May 1987.  (lEMP-Baltimore)

 4.    U.S.  EPA,  Region V.  Estimation and Evaluation of Cancer Risk
      Attributable to Air Pollution in Southeast Chicago (Draft).
      January 1989.   (Southeast  Chicago)

 5.    U.S.  EPA,  OAQPS.  Analysis of Air Toxics Emissions,  Exposures,
      Cancer Risks and Controllability in Five Urban Areas.   Volume I,
      Base  Year  Analysis and Results.  EPA-450/2-89-012a.   July 1989.
      (5 City)

 6.    U.S.  EPA,  OAQPS.  Updated  Estimated Cancer Incidence for Selected
      Toxic Air  Pollutants Based on Ambient Air Pollution  Data.  August
      1989.    (Ambient Air Quality)3

 7.    South Coast Air Quality Management District.  The Magnitude of
     'Ambient Air Toxics Impacts from Existing Sources  in  the South
      Coast Air  Basin.  1987 Air Quality Management Plan Revision
      Working Paper No. 3.  June 1987.  (South Coast)b

 8.    U.S.  EPA,  OPPE.  Final Report of the Philadelphia Integrated
      Environmental  Management Pro.iect.  December 1986.  (IEMP-
      Philadelphia)

 9.    American Management Systems.  Updated 35-Countv Study.  March
      1988.   (35-County)  This report was prepared under contract to the
      U.S.  EPA.

10.    U.S.  EPA,  Office of Mobile Sources.  Air Toxics Emissions from
      Motor Vehicles.  EPA-AA-TSS-PA-86-5.  (Mobile Sources)0
                                   2-3

-------
FOOTNOTES TO TABLE 2-1

a The VOC data used in this study was obtained from either: (1) J.J.
  Shah and E.  K. Heyerdahl, National Ambient Volatile Organic Compounds
  (VOC's) Data Base Update, U.S. EPA, Atmospheric Sciences Research
  Laboratory,  Research Triangle Park, NC, February 1988, or (2) A.
  Pollack, Systems Applications, Inc., Updated Report on the Interim
  Data Base for State and Local Air Toxic Volatile Organic Chemical
  Measurements, prepared for Bob Faoro, U.S. EPA, OAQPS, Research
  Triangle Park, NC,  August 1988.  The trace metal data were obtained
  from the Aerometric Information Retrieval  System, U.S. EPA, OAQPS,
  Research Triangle Park, NC,  March 1988, and the benzo(a)pyrene (BaP)
  data from J.  Bumgarner, Environmental Monitoring and Systems
  Laboratory,  U.S. EPA,  Research Triangle Park, NC, September 1988.

b Reprinted by the U.S.  Environmental Protection Agency as Multiple Air
  Toxics Exposure Study, Working Paper No. 3, South Coast Air Basin,
  EPA-450/4-88-013, November 1988.

c Information  in this study has been updated in this report using "Air
  Toxics Emissions from Motor Vehicles," prepared by Penny Carey and
  Joseph Somers.  This paper was presented at the 81st Meeting of APCA,
  Dallas, Texas, June 19-24, 1988.
                                   2-4

-------
                                TABLE  2-2
                EPA SOURCE CATEGORY  AND  POLLUTANT  STUDIES
Source Categories
 1.   Coal and oil combustion
 2.   Drinking water aerators
 3.   Gasoline marketing
 4.   Hazardous waste combustors
 5.   Municipal incinerators
 6.   Municipal solid waste landfills
 7.   Publicly owned treatment works (POTWs)
 8.   Sewage sludge incinerators
 9.   Superfund sites
10.   Treatment, storage, and disposal  facilities for hazardous waste
      (TSDFs)
11.   Waste oil combustors
12.   Woodstoves
Pollutants
13.   Asbestos
14.   Radionuclides
NOTE:  The references used to obtain risk estimates from these source
       category and pollutant studies are identified in Appendix C.
                                   2-5

-------
(e.g., radionuclides, asbestos).  The ATERIS data base contains
information generated from assessments of potentially toxic air
pollutants performed by OAQPS.  The ATERIS contains data from all  stages
of air toxics analyses, from the very preliminary to the more detailed
analyses.  The data contained in the ATERIS are intended for the
relative comparison and ranking of source categories and pollutants on a
nationwide basis.  The information in ATERIS is not considered an
authoritative source for verified estimates of risk attributable to
individual point sources.
      The number of pollutants and source categories included in the
individual studies varied.  As shown in Table 2-3, the number of
pollutants contributing to the estimated cancer risk in a study varied
from one (the Asbestos study) to 74 (the Hazardous Waste Combustor
study).  The study for the treatment, storage, and disposal facilities
for hazardous waste (TSDFs) used an initial list of 84 potential air
pollutants, 74 of which were identified as being emitted.  Of these 74
pollutants, risk estimates for 42 were made on the basis of available
EPA unit risk factors.  Most studies included 9 to 20 individual
pollutants in their risk estimates.
      A total of 90 different pollutants2 were included in the 22
studies and reports (see Table 2-4).  Forty-eight of the pollutants were
included in one or two studies.  Most of these 48 pollutants were found
in the NESHAP/ATERIS data base, the Hazardous Waste Combustor study, the
Sewage Sludge Incinerator study, or the TSDF study.  Another 22
pollutants were found  in three to six studies.  Twenty pollutants were
included in more than  six studies.
     2 Not all  of  these  pollutants,  however,  have EPA-derived unit risk
factors, as shown  in Tables 2-6 and 2-7 and as indicated in Table 2-4.
                                   2-6

-------
                              TABLE 2-3
NUMBER OF POLLUTANTS INCLUDED IN CANCER INCIDENCE ESTIMATES, BY STUDY
       STUDY
 NUMBER OF POLLUTANTS
INCLUDED IN RISK ESTIMATE
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
NESHAP/ATERIS Data Base
Ambient Air Quality
35-Coimty
5-City
I EMP- Baltimore
lEMP-Kanawha Valley
lEMP-Philadelphia
lEMP-Santa Clara
South Coast
Southeast Chicago
Mobile Sources
Asbestos
Coal and Oil Combustion
Drinking Water Aerators
Gasoline Marketing
Hazardous Waste Combustors
Municipal Waste Combustors
POTWs
Radionuclides
Sewage Sludge Incinerators
TSDFs
Waste Oil Combustors
45
20
19
19
9
18
9
14
15
30
9
1
9
10
4
74
10
7
2
33
42
8
     NOTE:   The Municipal  Solid Waste Landfills,  Superfund Sites,  and
            Woodstove studies do not include estimates  of cancer
            incidence.
                                 2-7

-------
                                   TABLE 2-4

             NUMBER OF STUDIES THAT INCLUDED SPECIFIC POLLUTANT IN
                      CANCER  RISK  ESTIMATE, BY POLLUTANT
    POLLUTANT
NUMBER OF
 STUDIES
POLLUTANT
NUMBER OF
 STUDIES
 1. Acetaldehyde             3
 2. Acrolein3                1
 3. Acrylamide               4
 4. Acrylonitrile            8
 5. Aldrin                   3
 6. Ally! chloride           3
 7. Aniline                  2
 8. Arsenic                 13
 9. Asbestos                 4
10. Benz(a)anthracene        2
11. Benzene                 17
12. Benzidine                2
13, Benzo(a)pyrene  (BaP)     6
14. Benzyl chloride3         2
15. Beryllium               12
16. Bis(2-chloroethyl)
      ether                  1
17. Bis(chloromethyl)
      ether                  2
18, Bis(2-ethyhexyl)
      phthalate              1
19. 1,3-Butadiene            9
20. Cadmium                 15
21. Carbon tetrachloride    15
22. Chlordane                3
23. Chloroform              15
24. Chloromethane            2
25. Chlorophenolsb           4
26. Chromium (VI)           13
27, Coke Oven Emissions      3
28. DDT                      2
29. Dibenz(a,h)anthracene    3
30. l,2-Dibromo-3-chloro-
      propane                4
31. p-dichlorobenzene        1
32. 1,2-Dichloropropane      3
33. Dieldrin                 2
34. Diethylstilbestrol       2
35. Diethanolamine8          1
36. Dimethylnitrosamine      1
37. 2,4-Dinitrotoluene       2
38. Dioctyl phthalate3       1
39. 1,4-Dioxane              2
40. Dioxin                   6
41. 1,2-Diphenyl
      hydrazine              2
                  42.
                  43.
                  44.
                  45.
                  46.
                  47.
                  48.
                  49.
                  50.
                  51.
                  52.
                  53.

                  54.
                  55.

                  56.

                  57.
                  58.
                  59.
                  60.
                  61.

                  62.
                  63.
                  64.
                  65.

                  66.
                  67.
                  68.
                  69.
                  70.

                  71.
                  72.
                  73.
                  74.
                  75.
                  76.
                  77.
                  78.
                  79.
                  80.
                  81.
Epichlorohydrin
Ethyl acrylate3
Ethylene dibromide0
Ethylene dichlorided
Ethylene oxide
Formaldehyde
Gasoline vapors
Heptachlor
Heptachlor epoxide
Hexachlorobenzene
Hexachlorobutadi ene
gamma-Hexachlorocy-
clohexane (lindane)
Hexachloroethane
Hydrazi ne/Hydrazi ne
sulfate
4,4 Isopropylidene
  diphenol3
Methyl chloride
3-Methylchloanthrene
Methyl hydrazine
Methylene chloride
4,4-Methylene
  dianiline3
Nickel (subsulfide)
Nitrobenzene3
2-Nitropropane
n-Nitroso-n-
  butylamine
n-Ni tro-n-methylurea
n-Ni trosodi ethyl ami ne
Nitrosomorpholine3
N-Ni trosopyrroli di ne
Pentachloronitro-
  benzene
Perchloroethylene
PICe
PCBf
Pronamide
Propylene dichloride3
Propylene oxide
Radionuclides
Radon
Reserpine
Styrene
Terephthalic acida
     4
     2
    12
    14
     7
    11
     8
     2
     2
     4
     2

     2
     2
     1
     3
     3
     1
    13

     1
     6
     2
     3

     1
     1
     2
     1
     1

     1
    16
     8
     7
     1
     2
     2
     2
     1
     2
     4
     1
                                         2-8

-------
                            TABLE 2-4 (concluded)

             NUMBER OF  STUDIES  THAT  INCLUDED  SPECIFIC  POLLUTANT  IN
                      CANCER RISK ESTIMATE, BY POLLUTANT
NUMBER OF
POLLUTANT STUDIES
82,

83.
84.
85.
1,1,2,2-Tetrachloro-
ethane
Thiourea
Titanium dioxide3
Toxaphene

3
2
1
2
86.
87.
88.
89.
90.
, NUMBER OF
POLLUTANT STUDIES
1,1,1 -Tri chl oroethane3
Trichloroethane
Trichloroethylene
Vinyl chloride
Vinyl idene chloride
1
2
16
11
6
a EPA-derived unit risk factors not available.

b Includes pentachlorophenol  and trichlorophenol.
  factor for trichlorophenol  is available.

c 1,2-dibromoethane.

d 1,2-dichloroethane.

e PIC = products of incomplete combustion.
                                        /
f PCB = polychlorinated biphenyls.
Only an EPA unit risk
                                         2-9

-------
      As shown in Table 2-5, 65 source categories were identified from
among the studies and reports.  Two of the source categories are general
in their coverage.  These are:  (1) chemical manufacturing (unspecified)
and (2) unspecified sources.  Forty-five of the source categories were
identified as being in only one study.  Most of these source categories
were identified in the NESHAP/ATERIS data base.  Nine of the source
categories were included in four or more studies, with gasoline
marketing included in the most (nine) studies.  It is likely that some
of the specified so'urce categories are included in the "chemical
manufacturing (unspecified)" and the "unspecified" categories.
Annual Cancer Incidence Analysis
      The total nationwide estimate of cancer incidence was based on the
estimated cancer incidence from all pollutants for which unit risk
factors have been developed by EPA and from all source categories
covered by the studies and reports in the data base.  It is important to
understand that not all of the unit risk factors developed by EPA have
undergone the same level of scrutiny.  In general, many of the unit risk
factors (e.g., those for benzene and carbon tetrachloride) have been
"verified" by the Agency, having undergone review by an Agency work
group, the Carcinogen Risk Assessment Verification Endeavor.  Such unit
risk factors are identified in Table 2-6 by reference to the Integrated
Risk Information System.  Most of the unit risk factors, however, have
not been Agency-verified.  The non-verified unit risk factors have
undergone various levels of review.  Some have received review by the
Office of Health and Environmental Assessment.  Others have received
little review.  Among the least reviewed unit risk factors are those
estimated for the group of compounds referred to in this study as
                                  2-10

-------
                                    TABLE 2-5                ,u,    ',-,

            DISTRIBUTION OF SOURCE CATEGORIES  BY  NUMBER OF STUDIES
    SOURCE CATEGORY
NUMBER OF
 STUDIES
SOURCE CATEGORY
                       -NUMBER OF i;r
                         .STUB-IE$-'\.
1.  ABS/SAN production8             1
2.  Acrylic fiber  production       1       34.
3.  Acrylonitrile  monomer          1       35.
4.  Asbestos, demolition           1       36.
5.  Asbestos, fabrication          1       37.
6.  Asbestos, manufacturing        1       38.
7.  Asbestos, milling              1       39.
8.  Asbestos, renovation           1
9.  Benzene fugitives              1       40.
10. Benzene storage                I,   -   41.
11. Benzene usage                   1       42.
12. 1,3-Butadiene  production       1
13. 1,4-Butanediol                  1       43.
14. Cadmium pigment mfg.           1
15. Cadmium stabilizer mfg.        1       44.
16. Carbon tetrachloride
      production                    1       45.
17. Chemical manufacturing                46.
      (unspecified)                6       47.
18. Chlorinated drinking
      water                         1       48.
19. Chlorine production            1       49.
20. Chlorinated hydrocarbon               50.
      production                    1
21. Chloroflourocarbon                    51.
      production                   .1       52.
22. Chlorinated hydrocarbon        1       53.
23. Chlorinated hydrocarbon               54.
      users                         1
24. Coal  and Oil  Combustion/              55.
      Heating                       5       56.
25. Commercial  Sterilization/             57.
      Hospitals                     2       58.
26. Cooling Towers                 3       59.
27. Drinking Water Aerators        2
28. Drycleaning                     4       60,
29. EBS  production13                1       61.
30. EDB  manufacturing0        "     1       62,
31. Electroplating                 4       63,
32. ETO  production01                1       64,
33. Formaldehyde  production        1       65.
                                          r:v
                                            >\»

                                           '*'..' '*'
                                          v.V/.tfff
                                              4^-T.i.^ •
                                              ;V-. ••> '•*'•;
                              •- v*?'*
                              v^i' .•"
Gasoline Marketing    ••j.-jj_
Glass mfg.         ' T  .    ';
Hexamethyl enetetram: mfg.',
Ind. solvent  coatings .
Iron and Steel  mfgi'v   "
Mel ami ne fjormal dehyde '.-..;
  resin          •'•••••• , V
4,4-Methylenedianiline      L
Motor vehicles     .   .   ";.-;. 8:
Municipal solid waste  -   ;•,
  landfills    ' -.  .  .:": :• -•;-v',2;
Municipal waste    ...   ',...,.-•-  •
  combustors         :./  ..  .•:*, 4
Nitrile elastomer    -  ,;*,'>7-y
  evaporation.     '    f. ,  :'.'..,,1,
Other organic evaporation, ;.; 1
Pentaerythitol  production .-I,
Pesticide Production/   ., ';  ;
  Usage           .     ,•-.,,.'•.'  ,2
Petroleum Refineries,   ,.':  - 2
Pharmaceutical  mfg.   ''.:•!
Phenol formaldehyde.  :       4- '
  resins           '    ;  ' /;;,"!.-.-w
Phthalic anhydride      "  ; l:-":-i
Polyacetal resins         ;  1;U::
Polybutadiene production  ''!,.;
Publicly owned  treatment   .>.;•/•;•';*'.
  works            :    i;/ ;  :6  .,"
Pulp and paper  mfg;   "  ':T-..-"-i'f;-:':
Sewage sludge incinerators  2  >;
Solvent Use/Decreasing    ,  5' .!
SBR production            ;  1  .v
Stripping  (paint, photo--;   ";..';""
  resist)              h-,'-••r-1^,
TSDFsf          '.    --..^;-..-.- ^2-r-r,
Trimethylopropane   '   "   :Vvi^:;;;
Unspecified                • 7"T^;'i
Urea Formaldehyde         .,„" T'"^
Waste Oil  Burning  '      ,:  3..'•.' •
Woodsmoke                  '„ 1' :'-7'
                                          2-11
                                                                      ••:.-,• ,  ., ;&. -•••'.•
                                                                       • • • .  ^ -" ./v1-;^'--' *,

-------
FOOTNOTES TO TABLE 2-5
a ABS/SAN - acrylonitrile butadiene styrene/styrene acrylonitrile
b EBS = ethyl benzene styrene
0 EDB = ethylene dibromide
d ETO - ethylene oxide
e SBR - styrene butadiene rubber
f TSDF - treatment,  storage, and disposal  facilities for hazardous waste
                                        2-12

-------
products of incomplete combustion (PIC).3   Finally,  many of the  unit

risk factors remain uncertain and are subject to change as further

evidence of carcinogenicity is obtained.

      Primarily due to the limited time and resources available for

the report, the annual cancer incidence analysis was limited in two

aspects:

      .  The analysis did not try to verify the results of the various
         studies.  Any errors that might be contained in the studies
         would, therefore, be carried over into this study.  In a few
         instances, some information was double-checked .as calculations
         suggested a possible error or two.  Double checking of
         information was the exception, however, and not the rule.

      •  The initial analysis was carried out on the basis of readily
         available background documents and reports.  In some instances,
         the documents and reports did not provide all of the necessary
         level of detail that would have been preferred.  This left a
         level of uncertainty in trying to compare data and resolve
         differences.  In general, these instances have been identified
         in the pollutant-by-pollutant analysis summaries, which are
         found in Appendix B.


      Methodology

      The annual cancer incidence analysis began by assembling the

annual cancer incidence estimates for each pollutant by source category

from each of the 22 studies.  Because the 22 studies varied in

geographic scope and thus population exposed, the annual cancer

incidence estimates were of limited value by themselves, especially

where the study was of limited geographic scope.  Therefore, an attempt

was made to correct for geographic scope by calculating the cancer

incidence per year per million population for each pollutant in each
     3  In this  study,  PIC  refers  to  the  large  number  of  primarily
particulate  compounds  that result  from  incomplete combustion.   PIC is
composed  primarily  of  "polycyclic  organic  matter"  (POM).    Some studies
use the term POM  when  estimating  the risk  from this class of compounds.
In addition,  some studies use benzo(a)pyrene (BaP), which is a component
of POM, as a surrogate to estimate risk  from PIC.

                                  2-13

-------
 source category for each study.   For the smaller,  localized studies,  the
 population reported in the studies was used to calculate the annual
 cancer incidence per million population.  For each nationwide study,  a
 1986 population of 240 million was used rather than trying  to determine
 the base year for each nationwide study.4
       The various pollutants and source categories frequently
 "overlapped"  between reports;  that is,  the same pollutant/source
 category combination (e.g.,  1,3-butadiene  emissions from motor vehicles)
 was included  in'more than  one  study or report.   Figure  2-1  illustrates
 this overlap  in a simplified diagram for five hypothetical  studies.
 Studies  No. 1  and 2 represent  some of the  larger studies, such as the
 35-County study or the 5-City  study.   Study No.  3  represents  a source
 category specific study.   Studies No.  4 and 5 represent  pollutant
 specific studies.   For example,  Study No.  2 is  seen in  Figure  2-1 to
 cover three of the same pollutants for  two source  categories  as Study
 No.  1, and the same three  pollutants  for one  source category  as Study
 No.  3.   Study  No.  5 overlaps one  pollutant/source  category  combination
 with  Study No.  3.   Study No. 4 covers  some of the  same  source  categories
 found  in  Studies  No.  1  and 2,  but for  a different  pollutant.
       Where overlaps  of pollutants  and  source  categories  occurred, the
 estimates of annual  cancer incidences per  million  population from each
 study  were compared.   If the estimates  were the  same (or  essentially the
 same)  for a pollutant/source category across  all studies, additional
 analysis  to identify  potential causes for  differences was obviously
       The risk estimates in all of the studies and reports used in this
study are based on 1980 to 1987  data  (i.e., emission inventories, ambient
measurements, populations, etc.).  For purposes  of this study, these data
were treated as applying to the same time frame.  The risk estimates can
be considered as mid-1980 numbers, or 1986 estimates.
                                  2-14

-------

                                        0  CO
                                        CO  CD
                                        •a  T3
                                        C  3
                                        I  8
                                        I  "
                                        O  ^
                                        i^_  O-
                                       •f  5
                                        g  iZ
                                        "CD  "S
                                        ->  O
                                        i=  co
                                        CO  CD
                                        3.  "v—
                                        =  O
                                            D)
                                        CD
                                        3
                                        D)
                                            O
2-15

-------
unnecessary.  If differences in the estimates of annual cancer incidence
per million population were found within a pollutant/source category
combination, a reduction analysis5 (as discussed below) was conducted to
resolve the differences and develop a point estimate6 of the annual
cancer incidence per million population for that pollutant/source
category combination.  If a pollutant/source categpry combination was
unique to an individual study, the estimate of annual cancer incidence
per million population for that pollutant/source category was considered
the best available estimate.  Once the point estimates of annual cancer
incidence per million population were identified, they were adjusted, as
necessary, to common unit risk factors for each pollutant.  (This
adjustment is discussed later in this chapter under the Reduction
Analysis section.)
      In extrapolating the estimates of annual cancer incidence per
million population to total nationwide annual incidence, the geographic
scope of the study was considered.  Most pollutants and source
categories were in at least one study that was nationwide in scope.
This enabled, in most instances, a direct extrapolation to total
nationwide estimates (i.e., multiplying the cancer rate by the total
U.S. population of 240 million).  A few pollutants and source categories
were included only in a study of limited geographic scope.  In such
instances, an attempt was made to determine whether the pollutant/source
     5 This type of reduction analysis was not undertaken for individual
risks because  individual  risks  are  site-specific numbers that cannot be
extrapolated to a  nationwide  estimate  of individual  risk.   Instead, the
study presents the estimates  of individual  risk  as  found in each of the
studies used in the data base for this study.
     6 In  some  instances,  it was not possible to develop a point estimate.
In such cases,  the range of estimates  for the cancer rate was narrowed as
much as possible.
                                  2-16

-------
category was unique to the geographic area, unusually concentrated in
the area, or fairly common across the United States.  If it was unique
to the area or appeared to be unusually concentrated in the area, then
generally only the cancer incidence estimated in the study for that   -.
category was included in the total nationwide estimate.  If the
pollutant/source category appeared to be fairly widespread, the estimate
.of annual cancer incidence per million population was extrapolated to a
total nationwide estimate (i.e., multiplied by 240 million population).
      Once this was done, the estimates of risk for each pollutant/
source category combination were summed to calculate the nationwide
estimate of annual cancer incidence.
      Reduction Analysis.  As noted previously, a large number of
pollutant/source category combinations with discrepant estimates of
annual cancer incidence per million population were identified.  An
analysis was undertaken in an attempt to derive a single estimate of the
annual cancer incidence per million population.
      A decision was made to limit the number of pollutant/source
category combinations for which the reduction analysis would be
conducted.  It was decided to analyze the estimates of annual cancer
incidence per million population of pollutant/source category
combinations for those pollutants that could potentially result in 10 or
more cancer cases per year nationwide based on information in any one
study.  These pollutants were identified in one of two ways:
      (1)  by the total number of annual cancer cases estimated for them
           in studies that were nationwide in scope (e.g., the Ambient
           Air Quality study, the Mobile Source study); or
      (2)  by the calculated number of cancer cases per year per million
           population which when extrapolated nationwide might result in
           10 or more cancer cases per year for the smaller geographic
           studies (e.g., the four IEMP studies).

                                  2-17

-------
A total of 23 pollutants were identified.  It is these 23 pollutants
that are presented in Appendix B.
      The reduction analysis looked to identify and reduce the
discrepancies by analyzing the following set of factors:
      •  unit risk factors
      •  emission factors
      •  modeled vs. ambient-measured concentrations
      •  source category definition and coverage
      •  geographic scope of the study
      •  study specific considerations
Each of these factors are discussed below as to how they were used and
considered in the reduction analysis.
Unit Risk Factors.7  Perhaps the most obvious reason that two estimates
of annual cancer incidence per million population would differ is that a
different unit risk factor had been used.  Unit risk factors have
changed in the past and may change in the future.  Thus, the first step
in the analysis was to put these estimates on the same "footing"; that
is, making sure the risk estimates are compared using the same unit risk
factors.  The unit risk factors used in each study8 were compared to
those identified in Table 2-6 and Table 2-7.  Table 2-7 shows the unit
risk factors used to estimate the cancer risk from PIC.  The pollutants
     7  The  unit  risk  factor   is   a   quantitative  estimate  of  the
carcinogenicity potency  of a pollutant.   It is often  expressed  as the
chance of contracting cancer from a 70-year lifetime continuous exposure
to a concentration of one microgram per cubic meter  (1  /*g/m3) of a given
pollutant.   For  example,  benzene  has  a  unit  risk  factor of  8.3x10
Ug/m3)"1.   In  a  population of  100,000 people  exposed to  10  ng/m  of
benzene  for 70 years,  the upper-bound  estimate  of  cancer cases  is
calculated to be 8.3  cancer cases  over 70 years (10 ^g/m3 x 100,000 people
x 8.3xlO"6 (/ig/m3)"1 = 8.3 cancer  cases over 70 years).

     8 Unit risk factors  used in the Municipal Waste Combustor study were
not in the available  reports, and  were assumed to  be the same as those in
Table 2-6.  Unit factors  for radionuclides and radon were accepted "as is"
in the reports.
                                  2-18

-------
                  TABLE 2-6



UNIT RISK FACTORS USED TO COMPARE CANCER RISK
EPA
POLLUTANT CLASSIFICATION
(CAS NO.)
•1.
2.
3.
4.
5.
6.
7.
8.
9.

10.
11.
12.
13.

14.

15.

16.
17.
18.

19.
20.
21.
22.
23.
24.
25.

26.

27.

28.
29.

30.

31.

32.
33.
Acetaldehyde (75-07-0)
Acryl amide (79-06-1)
Acrylonitrile (107-13-1)
Aldrin (309-00-2)
ATlyl chloride (107-05-1)
Aniline (62-53-3)
Arsenic (7440-38-2)
Asbestos (1332-21-4)
Benz (a) anthracene
(56-55-3)
Benzene (71-43-2)
Benzidine (92-87-5)
Beryllium (7440-41-7)
Bi s (2-chl oroethyl ) ether
(111-44-4)
Bi s (chl oromethyl ) ether
(542-88-1)
Bi s (2-ethyl hexyl )phthal ate
(117-81-7)
1,3-Butadiene (106-99-0)
Cadmium (7440-43-9)
Carbon tetrachloride
(56-23-5)
Chlordane (12789-03-6)
Chloroform (67-66-3)
Chloromethane (74-87-3)
Chromium (VI) (7440-47-3)
Coke Oven Emissions
DDT (50-29-3)
Dibenz (a, h) anthracene
(53-70-3)
l,2-Dibromo-3-chloro-
propane (96-12-8)
1 , 2-Di chl oropropane
(78-87-5)
Dieldrin (60-57-1)
Di ethyl sti 1 besterol
(56-53-1)
Dimethyl ni trosami ne
(62-75-9)
2,4-Dinitrotoluene
(121-14-2)
1,4-Dioxane (123-91-1)
Dioxin (1746-01-6)
B2
B2
BI
B2
B2
B2
A
A
B2

A
A
B2
B2

A

B2

B2
BI
B2

B2
B2
--
A
A
B2
B2


B2
C

B2
--



B2

B2
B2
UNIT RISK
a FACTORS
(tta/m3r1
2.2xlO'6
l.lxlO'3
6.8xlO'5
4.9xlO'3
5.5xlO'8
7.4xlO'6
4.3xlO'3
7.6xlO'3
8.9xlO'4
f
8.3xlO"6
6.7xlO"2
2.4xlO'3
3.3xlO'4

2.7xlO"3
*7
2.4xlO'7
/
2.8xlO"4
1.8xlO"3
1.5xlO"5

3.7xlO"4
2.3xlO'5
3.6xlO'6
1.2xlO'2
6.2xlO'4
S.OxlO'4
1.4xlO'2


6.3xlO'3
1.8xlO'5
"2
4.6xlO'3
l.4x!0'1
o "
1.4xlO'2
c
8.8xlO'5
£
1.4xlO'6
3.3xl01
REFERENCE
1
2
1
1
3
2
1
lb
2

1
1
1
1

2°

4

1
1
1

1
1
4
1
1
2
2


2
5

2
4

1

2

2
2
                     2-19

-------
           TABLE  2-6  (continued)



UNIT RISK FACTORS USED TO COMPARE CANCER RISK
EPA UNIT RISK
POLLUTANT CLASSIFICATION3 FACTORS
(CAS NO.) (Ua/m3r1
34. 1,2-Diphenyl hydrazi ne
(122-66-7)
35. Epichlorohydrin (106-89-8)
36. Ethyl ene di bromide
(106-93-4)
37. Ethyl ene di chloride
(107-06-2)
B2

B2
B2

B2

38. Ethylene oxide (75-21-8) B1-B2
39. Formaldehyde (50-00-0)
40. Gasoline vapors
(8006-61-9)
41. Heptachlor (76-44-8)
42. Heptachlor epoxide
(1024-57-3)
43. Hexachl orobenzene
(118-74-1)
44. Hexachl orobutadi ene
(87-68-3)
45. gamma-Hexachloro-
cyclohexane
(lindane) (58-89-9)
46. Hexachl oroethane
(67-72-1)
47. Hydrazine (302-01-2)
48. Methyl chloride (74-87-3)
49. 3-Methylchloanthrene
(56-49-5)
50. Methyl hydrazi ne (60-34-4)
51. Methyl ene chloride
(75-09-2)
52. Nickel (subsulfide)
(12035-72-2)
53. 2-Nitropropane (79-46-9)
54. n-Nitrosodi-n-
butylamine (924-16-3)
55. n-Nitrosodiethylamine
(55-18-5)
56. n-Nitroso-n-methylurea
(684-93-5)
57. n-Nitrosopyrrolidine
(930-55-2)
58. Pentachloronitro-
benzene (82-68-8)
59. Perch! oroethyl ene
(127-18-4)
Bl
B2

B2
B2

B2

C



C
C

B2
C
B2

B2
B2

A

B2

B2
B2

B2

B2


C
B2

2.2xlO'4

1.2xlO"6
2.2xlO'4

2.6xlO'5

l.OxlO'4
l.SxlO'5
6.6xlO'7

l.SxlO'3
2.6xlO'3

4.9xlO"4

2.2xlO'5



3.8xlO'4
4.0xlO'6

2.9xlO"3
3.6xlO'6
2.7X10'3

S.lxlO'4
4.7xlO"7

4.8X10'4

2.7xlO'3

1.6xlO'3
4.3xlO"2

8.6xlO"2

6.1xlO'4


7.3xlO'5
5.8xlO'7

REFERENCE
1

1
1

1

2
1
2

1
1

4
~
1



2
1

4d
2
2

2
2

1

2

1
2

2

1


2
2

                    2-20

-------
                                  TABLE 2-6  (concluded)

                   UNIT RISK FACTORS  USED TO  COMPARE CANCER  RISK
EPA
POLLUTANT CLASSIFICATION
(CAS NO.)
60.
61.
62.
63.
64.
65.

66.
67.
68.

69.

70.

71.
72.

PCB's (1336-36-3)
Pronamide (23950-58-5)
Propylene oxide (75-56-9)
Reserpine (50-55-5)
Styrene (100-42-5)
1,1,2,2-Tetrachloro-
ethane (79-34-5)
Thiourea (62-56-6)
Toxaphene (8001-35-2)
1 , 1 , 2-Tri chl oroethane
(79-00-5)
Tri chl oroethyl ene
(79-01-6)
2,4,6-Trichlorophenol
(88-06-2)
Vinyl chloride (75-01-4)
Vinyl idene chloride
(75-35-4)
B2
C
B2
B2
B2

C
B2
B2
C

B2

B2

A
C

UNIT RISK
a FACTORS
(tta/ni3r1
1.2xlO'3
4.6xlO'6
3.7xlO"6
S.OxlO'3
5.7xlO'7

5.8xlO'5
5.5xlO'4
3.2xlO"3
1.6xlO'5
£.
1.7xlO"6
£
5.7xlO"6

4.1xlO'6
5.0xlO'5

REFERENCE
2
2
6
2
6

1
2
1
1

2

2e
£
f
1

a  For a discussion of  how EPA evaluates suspect carcinogens and more information on these
   classifications, refer to "Guidelines for Carcinogen Risk Assessment" (51  Federal Register
   33992).  The EPA classifications used in this report are:

   A = proven human carcinogen

   B =   probable human carcinogen (B1  indicates  limited evidence from human studies and sufficient
         evidence from animal  studies;  B2  indicates sufficient  evidence  from animal studies but
         inadequate evidence from human studies)

   C =   possible human carcinogen
   Derived from 2.3 x 10   per fibers per ml (millimeter), which is the unit risk factor  reported in
_  IRIS (Integrated Risk  Information System)

c  IRIS currently reports a unit risk factor of 6.2x10"2  (jig/m3)  .
j                                                    "?      3 -1
   IRIS currently reports a unit risk factor of 4.9x10    (jig/m)  .

e  IRIS currently reports a unit risk factor of 3.1x10    (jig/m >  •

   An alternative unit risk factor of 4.2x10   (;tg/m )    has been developed by ORD.   U.S.
   Environmental Protection Agency, Office of Research and Development, Office of Health  and
   Environmental Assessment.  Health Effects Assessment Summary Tables First Quarter FY89.  January
   1989.
                                              2-21

-------
REFERENCES TO TABLE  2-6

1.  U.S.  EPA, Office of  Health  and  Environmental Assessment. Cincinnati,
    Ohio.  Integrated Risk  Information System  (on-line data base).

2.  U.S.  Environmental Protection Agency.  Hazardous Waste TSDF  -
    Background  Information  for  Proposed RCRA Air Emission Standards.
    Volume II - Appendices.  Preliminary Draft.  March 1988.  pp. E-8
    through E-13.

3.  Schell, R.M. Estimation of  the  Public Health Risks Associated with
    Exposure to Ambient  Concentrations of 87 Substances.  OAQPS, U.S.
    EPA, July,  1984.   Revised February 1985.

4.  U.S. EPA, Office  of  Solid Waste.  Draft Supplemental Rule for
    Hazardous Waste  Incinerators.  Appendix B, Unit Risks for
    Carcinogenic Constituents.  January 16, 1989.

5.  lEMP-Philadelphia.   Developed from EPA's Drinking Water Criteria
    Document.  March  2,  1984.

6.  U.S. Environmental Protection Agency, Office of Research and
    Development, Office  of Health and Environmental Assessment.  Health
    Effects Assessment Summary Tables Third Quarter FY90.  July 1990.

7.  U.S. Environmental Protection Agency, Office of Research and
    Development, Office  of Health and Environmental Assessment.  Health
    Effects Assessment Summary Tables First Quarter FY89.  January 1989.
                                  2-22

-------
                                TABLE 2-7

                   UNIT RISK FACTORS  USED  TO  ESTIMATE
                          CANCER RISK FROM PIC
SOURCE
CATEGORY
Unspecified
Unspecified6
Coke Ovens0
Municipal Incinerators0
Industrial power plants,
oil0
Utility power plants, oil0
Industrial power plants,
coal0
Utility power plants, coal0
Residential Heating0
Oil
Coal
Wood
Gasoline vehicles
Diesel vehicles
Sewage Sludge Incinerators
Hazardous Waste Combustors
COMPONENT
BaP
PIC
POM
POM

POM
POM

POM
POM

POM
POM
POM
POM
POM
BaP
PIC.
UNIT RISK
FACTOR
Ug/m3)-1
1.7xlO"3a
4.2xlO'1
6.5xlO'5
8xlO'8

S.OxlO"7
S.OxlO"7

8xlO"8
8xlO"8

9xlO"6
l.OxlO"5
l.OxlO"5
2.5xlO"4
2.0x10"* to 10x10"*
3.7xlO"6
l.OxlO'5
REFERENCE
1
2
3
3

3
3

3
3
3



4
4
5
5
  Based on inhalation study,
  3.3xlO"3  (jig/m3)"1.
Oral study suggests a unit risk factor of
b This unit risk factor for products of incomplete combustion (PIC) was
  based on relating lung cancer deaths to benzo(a)pyrene (BaP)
  concentrations where BaP serves as a surrogate for the large category
  of BaP-related pollutants referred to in the Six-Month Study as PIC.
  For a more detailed explanation of its derivation, refer to pages 20
  to 24a of the Six-Month Study.

0 These factors have been adjusted such that they are applied to the
  total particulate concentration to estimate risk from the POM fraction
  of the particulate matter.
                                  2-23

-------
REFERENCES TO TABLE 2-7
1.




2.



3.



4.


5.
U.S. Environmental Protection Agency.  Hazardous Waste TSDF -
Background Information for Proposed RCRA Air Emission Standards,
Volume II - Appendices.  Preliminary Draft.  March 1988.  pp. E-8
through E-13.

U.S. EPA, Office of Policy, Planning, and Evaluation.  The Air
Toxics Problem in the United States:  An Analysis of Cancer Risks
for Selected Pollutants.  EPA-450/1-85-001.  May 1985.

U.S. EPA.  Analysis of Air Toxics Emissions, Exposures, Cancer Risks
and Control!ability in Five Urban Areas.  EPA-450/2-89-012a.  July
1989.
U.S. EPA, Office of Mobile Sources.
Vehicles, September 1987.
Air Toxics Emissions From Motor
Memorandum.  Shiva Garg, US EPA, Office of Solid Waste and Emergency
Response, to Joseph Padgett, US EPA, Office of Air Quality Planning
and Standards.  Review of OAQPS Report on Six-Month Study of Impacts
of Air Toxics on Cancer Incidence.  March 3, 1989.
                                  2-24

-------
in each study that had different unit risk values than those shown in
Table 2-6 are identified in Table 2-8.
      The estimates of cancer incidence for PIC reported in Chapter 3
and the Executive Summary are based primarily on the unit risk factors
specific to individual source categories that are shown in Table 2-7.
The unit risk factor for PIC of 4.2 x 10"1  Ug/m3)'1  for unspecified
sources  was used only if a source-specific PIC unit risk factor was not
available.  The method used to calculate this PIC unit risk factor was
unusual, and any risk estimate based on its use should be treated as a
very preliminary estimate.  Some of the studies, such as the Ambient Air
Quality study, used this unit risk factor to estimate risk from PIC
using benzo(a)pyrene (BaP) ambient-measured concentrations as a
surrogate for PIC exposure.  Some studies also used this unit risk
factor for purposes of comparing cancer incidence estimates using
various methodologies.  For a discussion of these methodologies, please
refer to the section on PIC found in Appendix B.
      If a pollutant's unit risk factor differed from that in Table 2-6,
the estimated annual cancer incidence was adjusted to reflect the unit
risk factor in Table 2-6.  In general, there was little net effect on an
individual study's overall estimate of cancer cases as a result of this
modification (see Table 2-9).  The one exception to this was the South
Coast study.  The decrease in estimated annual cancer cases for the
South Coast study was due to large differences between the California
Department of Health Services (DOHS) unit risk factors used for several
pollutants in that study and EPA's unit risk factors for those
pollutants.  As seen in Table 2-10, adjusting the South Coast study's
estimates of cancer cases by using the unit risk factors in Table 2-6
                                  2-25

-------
                                           TABLE  2-8

                POLLUTANTS WITHIN IT  RISK  FACTORS DIFFERENT  FROM
                                THOSE USED  IN  THIS  REPORT
        STUDY
POLLUTANTS WITH  DIFFERENT
    UNIT RISK FACTORS3
 1.     Ambient Air Quality
 2.     NESHAP/ATERIS
 3.     Asbestos
 4.     Coal and Oil Combustion
 5.     Drinking Water Aerators
 6.     Gasoline Marketing
 7.     Hazardous Waste Combustors

 8.     Mobile Sources
 9.     Municipal Waste Combustors
10.     POTWs
11.     Radionuclides
12.     Sewage Sludge Incinerators
13.     TSDFs
14.     Waste Oil Combustors
15.     35-County
16.     5-City
17.     lEMP-Baltimore
18.     IEMP-Kanawha Valley

19.     lEMP-Philadelphia
20.     lEMP-Santa Clara

21.     Southeast Chicago

22.     South Coast
None
(see footnote b)
None
Beryllium,  Formaldehyde
EDC, Perchloroethylene, TCE, Vinyl chloride
None
BaP, Methylene chloride.
 Perchloroethylene,  TCE,  Vinyl chloride
Asbestos, Benzene, gasoline vapors, EDB,  BaP
None
Methylene Chloride,  TCE

BaP, Cadmium, PCBs,  TCE
None
TCE, PCBs
Benzene, BaP, Methylene chloride, TCE
Benzene, Methylene chloride, TCE
Benzene, Perchloroethylene, TCE
Benzene, Perchloroethylene, Vinyl chloride,
   Methylene chloride, TCE, BaP, Allyl chloride
EDC, TCE, Perchloroethylene
Benzene, Gasoline vapors, Methylene chloride,
Perchloroethylene, TCE, BaP
Acrylamide, 1,3-butadiene, PCB's,
   Propylene oxide
Benzene, BaP, Chromium, EDB,
   Methylene chloride, nickel, TCE
   EDC  * ethylene dichloride
   PCBs = polychlorinated biphenyls
BaP = benzo(a)pyrene
TCE = trichloroethylene
   Except for  some methylene chloride source categories  in the ATERIS data base,  all  of  the  unit
   risk factors  in the ATERIS data base are the same as  those in Table 2-6.  For  the  NESHAP  study as
   reported in the 1985 Six-Month Study, 21 unit risk factors have changed.  The  more important one
   in terms of either annual cancer cases or percent change are:  acrylamide, 1,3-butadiene,
   ethylene dibromide, nickel subsulfide, trichloroethylene, and vinyl chloride.   For a  complete
   listing, see  Table 3-27.
                                              2-26

-------
                               TABLE 2-9

          EFFECT OF CHANGES IN UNIT RISK FACTORS USED IN THIS
          REPORT ON ORIGINAL ESTIMATES OF ANNUAL CANCER CASES
                                  ESTIMATED ANNUAL CANCER CASES
                         Using Risk Factors As         Using  Table 2-6
   STUDY                   Reported in Study          Unit Risk  Factors
1.
2.
3.
4.
5.
6.
7.

8.
9.
10.
11.
12.

13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
Ambient Air Quality
NESHAP/ATERIS
Asbestos
Coal and Oil Combustion
Drinking Water Aerators
Gasoline Marketing
Hazardous Waste
Combustors
Mobile Sources
Municipal Waste Combustors
POTWs
Radionuclides
Sewage Sludge
Incinerators
TSDFs
Waste Oil Combustors
35-County
5-City
lEMP-Baltimore
lEMP-Kanawha Valley
lEMP-Philadelphia
lEMP-Santa Clara
Southeast Chicago
South Coast
2,022
504a
82
11.1
0.021
24-75
0.3-9

628-1,874
1.7-2.3
1.5
16
13

140
0.10-0.56
469-553
92.6
2.8-7.0
1.8
0.37
2.2
1.21
162-221
2,022
496b
82
12.1
0.021°
24-75
0.3-9C

601-1,852
1.7-2.3
1,3
16
13

140
0.10-0.56
463-546
90.4
2.95-7.15
1.77
0.42
1.85
1.26
19-33
NOTE:  The reports on Municipal Solid Waste landfills, Superfund sites,
       and Woodstoves did not include estimates of annual cancer cases.

a Based on original  NESHAP study as reported in the Six-Month  Study.

b Incorporates revised NESHAP study estimates and ATERIS data  base risk
  estimates.

c The net effect of adjusting unit risk factors cannot be determined
  as cancer risk attributable to individual organic compounds was not
  available.  The effect is expected to be small.
                                  2-27

-------
                                  TABLE  2-10

         EFFECT OF UNIT RISK FACTORS ON ESTIMATED ANNUAL CANCER CASES:
                             THE SOUTH COAST STUDY




POLLUTANT

Benzene
Carbon tetrachloride
Chi orof orm
Ethyl ene di bromide
Ethyl ene di chloride
Methyl ene chloride
Perch! oroethyl ene
Tri chl oroethyl ene
Arsenic
Beryl 1 i urn
Cadmi urn
Chromium
Nickel
Total Annual Cancer Cases


South
Unit
Ambient
Measured0
99
1.4
1.3
0.37
-
8.0
0.59
0.33
1.5
0.09
0.49
108
0.37
221
ESTIMATES OF
ANNUAL CANCER CASES USING.
Coast Study
Risk Factors3
Model
Predicted0
55
0.001
0
0.007
0.007
3.4
0.43
-
0.0001
0.0003
0.96
102
0.09
162
EPA Unit

Ambient
Measured6
16
1.4
1.3
1.1
-
0.92
0.59
0.43
1.5
0.09
0.49
8.6
0.56
33


Risk Factors'3

Model
Predicted6
8.6
0.001
0
0.02
0.007
0.39
0.43
-
0.0001
0.0003
0.96
8.2
0.14
19
a The unit risk factors used in the South Coast study, which are California
  Department of Health Services' unit risk factors, are found on page V-10 of
  the South Coast study.

b The EPA unit risk factors used to adjust the estimates of annual cancer cases
  are found in Table 2-6 of this report.

c Based on dividing estimated lifetime  (70-year) cancer cases in Table VI-3, p.
  VI-11, of the South Coast study by 70.

d For each pollutant, the annual cancer cases in this column were calculated as
  follows:  the estimate of annual cancer cases using the South Coast study's
  unit risk factors was multiplied by the ratio of the EPA unit risk factor to
  the California Department of Health Services unit risk factor for that
  pollutant.
                                        2-28

-------
reduced total estimated annual cancer cases by approximately 85 percent.
The issue of which unit risk factor, DOHS or EPA, more likely represents
actual risk is beyond the scope of this study.  For purposes of this
study, all cancer risks are evaluated and reported"(unless otherwise
noted) on the basis of the unit risk factors presented in Table 2-6 and
Table 2-7; the unit risk factors from the South Coast study were not
used to estimate nationwide cancer risk.
Emission Factors.  A second basic reason for different cancer risk
                                                u
estimates is that different pollutant emission factors have been used."
Where emission factors could be compared, the most recent emission
factor was selected in the calculation of cancer risk.  (This selection
assumes that the more recently developed emission factor is a better
(more accurate) factor than the previous emission factor.)  In these
instances, appropriate adjustments were made to the cancer risks based
on "older" emission factors, and the "new" set of estimated annual
cancer incidences per million population were compared.  Unfortunately,
except for motor vehicles, pollutant emission factors for most source
categories were either not readily available in that they were not
included  in the final report or were reported in only one of the
studies,  and a comparison could not be made.  Thus, it was generally
very difficult to say anything about the effect, if any, pollutant
emission  factors had on discrepant estimated annual cancer incidences
per million population.
       In  several instances, the studies referred to more recently
developed emission  factors that were used  (i.e., the 5-City study) or
not used  because it was beyond the scope of the  study  (i.e., the 35-
County study).  Such qualitative statements were used to some extent in
                                   2-29

-------
selecting some cancer risk estimates as better than others.  In summary,
except for motor vehicles, trying to identify differences in emission
factors as a source of discrepancy was not very successful.
Modeled vs. Ambient-Measured Concentrations.  Cancer risk estimates can
also vary depending on whether they are derived from modeled concentra-
tions or from ambient-measured concentrations.  Both methods of
obtaining ambient concentrations from which cancer risks can be
estimated have their own inherent set of limitations (see Table 2-11).
It was beyond the scope of this project to analyze the various limita-
tions of the two techniques for estimating ambient concentrations.  For
example, this study did not try to determine whether the most appropriate
models were used in the studies or to try to "correct" the cancer  •
estimates to a single model.  Similarly, it was beyond the scope of this
project to try to determine whether the proper sampling technique was
used to obtain the ambient samples or whether the sampling point
locations were likely to obtain representative samples.
      The study did, however, attempt to use several guidelines or
"thought processes" in evaluating and comparing cancer risks obtained
from modeled concentrations and from ambient-measured concentrations.
These were:
      •  Unless otherwise noted in a study, all models were assumed to
         be appropriate and their results were given equal weight.
      •  Where modeled and ambient-measured concentrations were used
         and risk estimates made, an attempt was made to identify
         potential causes for discrepancies based upon known emission
         sources.  For modeled estimates, this meant trying to identify
         emission sources included in the inventory and emission sources
         that were excluded.  For ambient-measured concentrations, this
         meant trying to determine if the locations from which the data
         were obtained contained known point sources that might
         influence or bias the data.
                                  2-30

-------
                                          TABLE  2-11                        ,

    SELECTED  LIMITATIONS OF MODELED AND AMBIENT-MEASURED  CONCENTRATIONS
                                FOR  ESTIMATING  CANCER RISK
Modeled Concentration Limitations
1.    Many of the dispersion models assume flat terrain and average meteorological conditions.
      Rough terrain in the area surrounding a source,  such  as a valley, can result in concentrations
      that are up to one to two orders of magnitude higher  or lower than concentrations predicted in
      gently rolling terrain.

2.    Dispersion modeling often extends to only 20 kilometers from the source.  This technique  can
      lead to understating risk if extending dispersion increases significantly the number of people
     •exposed.

3.    Dispersion modeling estimates are rarely based on site-specific meteorology.  Often, data from
      hundreds of kilometers away must be used.

4.    Dispersion models do not consider increases in concentrations that could result from re-
      entrainment of toxic particles from streets, rooftops, etc.  In addition, models do not
      account for background concentrations, secondary formation of pollutants, and emissions from
      other sources not explicitly included in the analyses.

5.    Emission estimates are generated from data and assumptions that could be in error.  For
      example, although some of the studies (e.g., the 35-County study) incorporate plant-specific
      emission estimates whenever possible, the pollutant releases fo.r other sources are frequently
      estimated by applying speciation factors against the  volatile organic compound (VOC) and  total
      suspsended .particulate (TSP) data in the National  Emission Data System (NEDS).  Unfortunately,
      some of the information in NEDS is of questionable consistency and quality for the purposes of
      quantitative risk assessment.

Ambient-Measured Concentration Limitations

1.    A basic limitation is the extrapolation of measurements from a limited number of sites to a
      much larger geographic area in order to estimate population exposure.  This affects both
      estimating exposure within a city from a limited number of sites to estimate average exposure
      within the city and estimating nationwide cancer risk from a limited number of geographic
      areas.

2.    Ambient-measured data collected over long periods  of  time (e.g., at least one year) are
      frequently unavailable, which limits the ability to make statements as to long-term exposures
      upon which cancer risk estimates based.

3.    All  ambient-measured data are subject to errors  in sampling and analytical methods.

4.    Ambient data may underestimate "true" maximum individual risk (MIR) concentrations because
      sampling is limited to a small number of fixed monitoring sites.
                                              2-31

-------
       •   In  the  absence  of  evidence  to the contrary, cancer  incidence
          estimates  based on modeled  concentrations and those based on
          ambient-measured concentrations cannot be summed to obtain a
          total risk estimate  (i.e.,  they are not mutually exclusive).
          For only one  pollutant,  1,3-butadiene, were these two estimates
          summed.  This was  done because the locations for the ambient
          data were  not identified as having known point sources of 1,3-
          butadiene.  Therefore, it was felt that a better estimate would
          be  obtained by  assuming  the ambient data reflected  background
          and area-type emissions  of  1,3-butadiene (which would include
          motor vehicle emissions)  to which the cancer risk from the
          modeled point sources could be added.

       •   The South  Coast study noted several pollutants for  which large
          discrepancies between modeled and ambient-measured  concentra-
          tions occurred  and offered  potential reasons for such.  Other
          studies also  noted where they believed one methodology may be
          underestimating risk.  In each case, the studies identified the
          modeled estimate as  possibly underestimating risk.  The reason
          most frequently cited for this underestimation was  an incom-
          plete inventory of emission sources.  Modeling biases can also
          lead to the underestimation of risk.  These discrepancies and
          their reasons are  noted  in  the pollutant-by-pollutant analysis
          section found in Appendix B.  These reasons were considered in
          evaluating  which risk estimates were "better" than  others.

          For formaldehyde and carbon tetrachloride, the ambient-measured
          concentrations  and.derived  cancer estimates were selected and
          evaluated.  For formaldehyde, this was done because it is well
          established that formaldehyde is formed in the atmosphere
          (secondary  formation).   Ambient-measured data can account for
          this atmospheric-formed  formaldehyde, whereas models do not.
          In  the case of  carbon tetrachloride, it is also well known that
          carbon tetrachloride remains in the atmosphere long after it
          has  been emitted.  Thus  again, ambient-measured data can
          account for this "retention" of carbon tetrachloride more
          readily than  models.

       •   For comparing between ambient-measured data, the geographic
          coverage of the  study was considered.  It was assumed that risk
          estimates based  on ambient-measured data from more geographic
          locations were  better estimates from which to estimate nation-
         wide risk than  were estimates from single geographic locations.
         This led to selecting the Ambient Air Quality study results as
         the  best estimates of nationwide risk from those estimates
         based on ambient-measured data.  In fact, most if not all of
         the  smaller geographic ambient data fell within the range of
         data used in  the Ambient Air Quality study.

Source Category Definition  and Coverage.  One of the basic steps in

reducing the data was  to  determine the various source categories (e.g.,

motor vehicles, electroplating, municipal  landfills) covered by the
                                  2-32

-------
studies, and then to assign the risk estimates for each pollutant to
that source category.  This is necessary to avoid double-counting.  In
this study, specific source categories were used to aggregate and
compare the risk data.  For most of the source categories and studies,
assigning risk estimates to the appropriate source category was
relatively easy, as most of the specific source categories were
developed on the basis of the source categories reported in the studies.
In certain instances, however, it was difficult to determine whether or
not a source category in one study was the same as in another.  For
example, the source categories "heating," "combustion," "residential,
heating," "coal and oil combustion," and "oil combustion," all appeared
in one or more studies.  In this instance, it was very difficult to
determine whether or not the same types of emission sources were being
covered.
      Another aspect to source category definition was whether or not
the studies included all of or just some (and which ones) of the types
of emission sources in a particular source category.  For example, some
motor vehicle pollutants are exhaust and evaporative emissions as well
as tire wear emissions.  Some studies reported only the risk from
                                                                      /
exhaust and evaporative emissions, while one study included those from
tire wear.  The ability of determining the specific types of sources
covered by each study for each source category met with varying success,
because the information needed to ensure an accurate accounting was not
always  reported in the available material.  In certain cases, we were
able to obtain  information beyond that which was published.  Thus,
assumptions as to which source categories are mutually exclusive or not
and whether the same  set of emission sources are covered in a particular
source  category remain, in certain  instances, highly uncertain.
                                  2-33

-------
       Finally, plant location information from the  NESHAP/ATERIS  data
 base and various EPA documents was used to determine  whether specific
 plants were located in counties covered by the 35-County  study, in  the
 five cities covered in the 5-City study,  in the four  IEMP study cities,
 in the South Coast geographic area,  and in the Southeast  Chicago  study
 area.   The relationship of plant locations to  the geographic study  area
 of these other studies was used to assess the  potential relationship of
 the risk data (whether they were mutually exclusive and could be  added,
 or whether they were duplicative).   Evidence of a match was  assumed to
 infer  a likelihood of double-counting  if  the two risk estimates were
 added.   If no plant location match was  found,  it was  assumed to infer a
 likelihood of mutual  exclusi'veness.
 Geographic Scope of the Study.   As the  primary purpose of this study is
 to evaluate nationwide risk,  modeled risk estimates from  studies  that
 already have a nationwide  scope  were generally preferred  as  better
 estimates  of nationwide risk than those nationwide risks  that could be
 extrapolated from the  studies with smaller  geographic scopes.  This is a
 somewhat difficult  "preference"  assumption  to  make.  The  smaller
 localized  studies  frequently are  based on much more detailed and  site-
 specific data  than  are the  nationwide studies.   Thus,  those studies may
 do  a somewhat  better job at  estimating likely  levels of risk.  At the
 same time,  because  they take  into account site-specific data, they are
 likely  to  be  less representative of conditions nationwide and thus can
 not be  simply  extrapolated  nationwide.   As this  study is  in itself a
 broad scoping  type  of  study, the broader scoping nature of the
 nationwide  studies  are more  consistent with the goals  of this study.
Therefore,  based upon  these  considerations, the results of the
                                  2-34

-------
nationwide studies received preference in developing point estimates of

cancer risk.

      As noted earlier in this chapter, risk estimates for a few source

categories were available only from the smaller, localized studies.

Nationwide risks were extrapolated from these studies in some instances

(e.g., petroleum refineries).  In other instances, so little was known

about the emission source that the cancer risk from only that study was

used in estimating the total nationwide risk.  Such sources are part of

one of the general source categories (e.g., unspecified sources,

miscellaneous).

      As noted earlier, for ambient-measured risk estimates, those from

the Ambient Air Quality study were generally assumed preferable to those

extrapolated from the smaller, localized studies because of its broader.

geographic scope.

Miscellaneous Specific Considerations.  As the studies and various risk

estimation methodologies were reviewed, several additional factors were

considered in evaluating the data.

      .  The 35-County study noted that the counties studied were
         selected, in part, because of the presence of known emission
         sources of the pollutants being considered.  Thus, the
         estimates of annual cancer incidence per million population
         calculated for the 35-County study may be higher than the
         nationwide population-weighted average.  Applying, the 35-
         County study's rates directly to the total U.S. population
         could result in an overestimation of cancer risk.

      •  Several methodologies exist for estimating risk from PIC.  Each
         methodology has its own inherent limitations, and no methodo-
         logy has been shown to be better than another.  The current
         trend in estimating risk has been toward using individual
         source category emission factors and developing unit risk
         factors that are based upon the mixture of components emitted
        .from the source category.  For purposes of this study, the
         modeled estimates of risk from PIC were selected from those
         estimates using this type of risk estimation methodology;
                                  2-35

-------
 Limitations and Uncertainties
       Limitations
       Consistent with the purposes of this study,  the analyses  in  this
 report consider cancer risk only from air toxics.   Noncancer effects  of
 air toxics are not included.  As noted in Chapter  1,  other studies are
 being undertaken to examine other health effects,  such as  subtle
 biochemical,  physiological, or pathological  effects to gross effects,
 including death.
       The only pathway considered in  this report is inhalation.
 Potential  health risks from ingestion of air pollutants  that ultimately
 reach humans  through  the  diet or that are directly ingested  are not
 examined.   Neither are the  potential  environmental  effects of direct
 deposition and urban  runoff of air pollutants to surface water
 addressed.
       Estimates  of cancer risk are based  on  concentrations of air  toxics
 found in the  ambient  air.   It was  not  the  purpose  of  this study to
 estimate cancer  risk  based  on exposure to  indoor concentrations of air
 toxics.  As noted  in  Chapter  1,  a  separate program  has been  initiated to
 quantify the  risk  from indoor exposure to  air toxics.
      Although quantitative risk estimates are reported  in this study,
 it is important  to  remember that the reports and studies used do not
 cover either all known or potential air toxics or all sources of air
toxics which contribute to outdoor exposure.  As noted earlier in this
chapter, the cancer risk estimates in  the reports and studies reviewed
cover 90 compounds  in approximately 65 source categories.  These
compounds represent only a fraction of the total  number of compounds
present in the ambient air.   Based on a review of studies directed  at
                                  2-36

-------
identifying compounds in the ambient air,  more than 2,800 compounds have
been identified as existing in the atmosphere,9 some  of which  may  be
toxic at ambient levels.  One major factor preventing analysis of more
pollutants is the lack of measurement techniques to obtain ambient
measurements for a number of pollutants.  A second major factor is the
lack of data on cancer risk associated with ambient concentrations of
other pollutants.  Only about 10 percent (approximately 300) of the
2,800 plus atmospheric pollutants have been tested for mutagenicity or
carcinogenicity.  Of these, 97 have tested positive in whole animal
bioassays.  The mutagenicity or carcinogenicity of the other 2,500
atmospheric compounds is unknown.  The impact on cancer incidence from
these other atmospheric compounds is currently impossible to estimate.
      Despite the fact that more than 2,800 chemicals have been
identified in ambient air, a large number of unknown compounds are still
likely to exist.  Indeed, atmospheric chemists studying the reactions of
most common urban pollutants are often able to account for only about
one-half of the carbon in their studies.  The impact of the unidentified
organic products on cancer incidence is unknown.  However, the compounds
for which risk information is available were selected based on evidence
that led to their being suspected carcinogens.  Thus, it is possible
that the cancer risk associated with the 90 or so compounds for which
cancer risk data have been obtained represents a much larger proportion
of the total risk than might be suggested by a simple comparison of the
90 compounds to the total number of atmospheric compounds.
     9 Graedel, T.E.,  D.T. Hawkins, L.D. Claxton.  Handbook of Atmospheric
Compounds:   Sources,  Occurrence,  and  Bioassav.    HERL-051a.    (1985:
Academic  Press, New York).
                                  2-37

-------
       The quantitative risks reported in this study are based,  in  part,
 on unit risk factors that are either upper-bound estimates or maximum
 likelihood estimates of the carcinogenicity potential  of the air toxic
 pollutants.  In either case, the quantitative estimates based on these
 unit risk factors are conservative in that  actual  cancer cases  for these
 pollutants may be higher,  but are more likely to be lower than  the
 estimates presented in this study.   Thus, the aggregate cancer  risk,
 which is  based on the summation  of individual  pollutant's  cancer risks,
 represents a likely overestimate for those  pollutants  considered.
       The amount and quality of  information  concerning pollutants  and
 their risk from specific  source  categories  vary  considerably.   For
 example,  information on the types  of pollutants  emitted  from  motor
 vehicles  is fairly well established.   In addition,  emission  factors for
 most  motor vehicle pollutants  have  been estimated much more closely than
 for other source categories because,  in part,  of the relative ease with
 which motor vehicles  can be tested.   On the  other hand,  the types  of
 pollutants  from source categories,  such as TSDFs and Superfund  sites,
 are much  more  likely  to vary because  the materials  that  give  rise  to the
 pollutants  vary from  one site  to another.  Also, the emission levels of
 pollutants  from such  source categories are much more difficult to
 establish because  the test methodologies are not as easy to apply  as
those for motor  vehicles.   It  should  be noted that  there is considerable
uncertainty associated with the estimates of risk attributed to
 individual  pollutants emitted  from TSDFs.  It is possible that dioxin,
the TSDF pollutant for which the largest risk is estimated, may be
emitted in much smaller quantities, if at all, from TSDFs.  Finally,
estimates of risk for some source categories may suffer simply from a
lack of a complete accounting of pollutants.
                                  2-38

-------
      Reliable quantitative estimates remain unavailable for many
potentially important source categories (e.g., Superfund sites) and, in
some instances, sources may have been missed in the source assessment.
In addition, quantitative estimates of risk from pollutants formed or
transformed in the atmosphere (secondary formation) remain unquantified
for almost all pollutants.  The most important secondary pollutant for
which cancer risks have been quantified to date is formaldehyde.
Formaldehyde is both emitted directly into the atmosphere and formed in
the atmosphere, and atmospheric formation of formaldehyde has the
potential to produce many times the amount directly emitted from most
sources.. The gas-phase transformation products of a variety of common
urban pollutants and air toxics have been shown to be potentially
hazardous.  The normal atmospheric reactions of these pollutants produce
a variety of oxygenated and nitrogenated products, such as glyoxal and
peroxyacetylnitrate (PAN), and a variety of unidentified species, which
have been shown to be mutagenic.  The total mutagenicity of the
transformation products is often many times greater than the
mutagenicity of the original pollutants.  The fact that a gas-phase
product  is mutagenic in a bacterial test system suggests, but does not
establish, that a human health risk may arise from exposure to such
products.  It is not currently possible to quantify the risk from
exposure to the unidentified, potent gas-phase mutagens produced in
these photochemical reactions.  Nevertheless, the evidence to date
clearly  suggests that the transformation of ubiquitous, often innocuous,
urban pollutants may add a significant additional risk component to any
assessment of urban exposure and risk.
                                  2-39

-------
      Uncertainties

      There are many uncertainties associated with the methodologies

that are available for making the risk estimate used in the reports and

studies that formed the data base for this report.  The following

uncertainties are among the more important ones to keep in mind.  The

list is not all inclusive, and additional uncertainties are identified

throughout the report.

      •  Cancer incidences presented in this report are based on the
         assumption that emission levels and ambient levels for each
         pollutant either "averages out" over a 70-year period to equal
         the concentrations used in the calculations of annual incidence
         or remain constant for that period of time.  In reality,
         emissions and air quality will vary from year to year.  Because
         the amount and direction of variation is unknown, it is unclear
         how much this assumption affects the results.

      •  All of the analyses assume exposure to air toxics occurs where
         people reside.  This assumption does not consider the
         possibility that people may move throughout the urban area and
         change their homes several times during their lives.  In
         addition, few plants may operate or be expected to emit at the
         same level for 70 years, though the areas in which they are
         located may remain industrial.  Thus, future exposures may be
         either worse or better than the old environment.  Because
         exposures are simulated over a 70-year period, it is unclear
         how much this assumption affects the results.

      •  All of the risks assume continuous outdoor exposure.  This
         assumption ignores the fact that people spend the majority of
         their time indoors, and thus are exposed to indoor atmospheres,
         which can be significantly different from the outdoor
         atmosphere.  Indoor concentrations of certain pollutants (e.g.,
         radon, tobacco smoke, formaldehyde, and other VOCs) are
         commonly several times higher than outdoor concentrations.
         Estimated cancer risk to such indoor pollutant concentrations
         suggest that cancer risks based solely on outdoor exposure may
         be understated for such pollutants.  On the other hand, the
         extent to which certain pollutants (e.g., trace metals)
         penetrate indoors is large unknown.  If emissions of a
         pollutant do not penetrate completely indoors and if there are
         no indoor sources of that pollutant, then cancer risks based
         solely on outdoor exposure will have been overstated.

      •  All risks are assumed to be additive.  This assumption can lead
         to substantial errors in risk estimates if synergistic or
         antagonistic interactions occur.  Although dose additivity has
         been shown to predict the acute toxicities of many mixtures of
                                  2-40

-------
.similar  and  dissimilar  compounds,  some marked exceptions have
 been  noted.   Consequently, additivity assumptions may
 substantially overestimate risk  in  some cases and underestimate
 it  in others.   The  available data  on mixtures are insufficient
 for estimating the  magnitude of  these errors.  Based on current
 information,  additivity assumptions are plausible for component
 compounds  that induce similar  types of effects at the same
 sites of action.

 Unit  risk  factors used  in this study have been generated, in
 most  instances,  using EPA approaches or models.  Most of the
 resulting  unit risk factors are  generally regarded  either as
 plausible, upper-bound  estimates or as maximum likelihood
 estimates.   The linearized multistage procedure  used to derive
 these factors leads to  a plausible  upper limit to the risk  that
 is  consistent with  some proposed mechanisms  of carcinogenesis.
 Such  estimates,  however, do not  necessarily  give a  realistic
 prediction of the risk.  The true  value of the risk is unknown,
 and may  be as low as zero.

 Cancer unit  risk values are subject to much  uncertainty and in
 many  cases are preliminary estimates.  The risk  estimates in
 the reports  are based on layers  of assumptions concerning the
 health effects of chemicals, the degree of human exposure,  and
 the way  these substances interact  inside the human  body.  For
 example, the weight of  evidence  of carcinogenicity  for the
 compounds  identified in this report varies greatly, from very
 limited  to very substantial.   Further, the extent of evaluation
 and health review performed varies  considerably  among
 compounds.   As additional scientific information is acquired,
 these values could  change significantly, as  they have in the
 past, and  thus the  magnitudes  and  relative importance of
 particular pollutants can change.

 In  developing its unit  risk factors, EPA uses a  nonthreshold,
 multistage model, which is linear  at low doses,  to  extrapolate
 from  high-dose response data to  the low doses typically caused
 by  exposure  to ambient  air.  In  other words, carcinogenic sub-
 stances  are  assumed to  cause some  risk at any exposure level.
 If  the true  dose-response relationship at low doses is less
 than  linear, then the unit risk  estimates err on the high side.

 Many  of  the  individual  pollutants  have specific  uncertainties
 that  affect  their potential contribution to  cancer  risk.
 Chapter  3  and Appendix  B identify  these uncertainties.
 Chromium,  formaldehyde, and PIC  are three of the major
 contributors, based on  this study,  to cancer risk.  Each have
 specific uncertainties  that may  significantly affect the
 estimate of  cancer  risk attributed to them.  These
 uncertainties are  highlighted  below.

 In  the case  of chromium, only  the  hexavalent form  has been
 proven to  be carcinogenic.  The  percentage of total chromium
 that  is  hexavalent  is  known to vary considerably depending  on
                          2-41

-------
 the source.   For example,  hexavalent  chromium  is  less than  1
 percent of total  chromium  emissions from  coal  and oil burning
 combustion,  while it  is  nearly  100 percent  of  total chromium
 emissions  from cooling towers and electroplating.
 Nevertheless,  considerable uncertainty  remains as to the
 exposure to  hexavalent chromium versus  total chromium
 emissions.

 In  the  case  of formaldehyde, a  number of  unresolved issues
 concerning  its carcinogenesis have been the subject of a
 considerable amount of scientific debate.   At  the center of
 this  debate  are questions  concerning  such issues  as the
 mechanism  of action of formaldehyde at  the  molecular level, the
 shape of the dose-response curve, the importance  of irritation
 and the role of the mucus  blanket, and  the  significance of
 endogenous formaldehyde.   The EPA has determined  that the
 95-percent upper  confidence limit on  risk for  formaldehyde,
 based on data  from a  24-month animal  study  conducted by the
 Chemical Industry Institute of  Toxicology (CUT),  is the
 appropriate  statistical  estimate to use in  assessing human
 risk.   This  is consistent  with  the EPA  Guidelines  for
 Carcinogen Risk Assessment, which state that in the absence of
 compelling biological information on  the mechanism of action,
 the linearized multistage  procedure should  be  used to derive an
 upper bound  estimate  of  risk.   The EPA does not recommend the
 use of  maximum likelihood  estimates of cancer  risk based on
 animal  data;  such estimates are highly unstable (i.e., small
 changes  in the data may  cause orders-of-magnitude  fluctuations
 in  the  risk  estimates).  The EPA is currently  evaluating new
 scientific data on formaldehyde and will publish  an update to
 the 1987 assessment at some time in the future.

 There has also been disagreement over whether  to  consider the
 incidence of both malignant and benign tumors  in  rats or
 whether  only the malignant  tumors are significant.  The unit
 risk  factor  based on  total   tumors is  approximately 14 times
 higher than  the unit  risk  factor based on malignant tumors
 only.  The current consensus is that  only the malignant tumors
 should be used  to assess the human cancer risk from
 formaldehyde.   There  appears to be little evidence that benign
 tumors progress to any of the malignant tumors seen in the CUT
 study.  The  unit risk factor based on malignant tumors only is
 used  in this report to estimate cancer incidence  from exposure
 to  formaldehyde.

 In  the case  of PIC, there are several  sources of uncertainty.
There are a  number of methodologies available to estimate risk
 from PIC.  Some of these methodologies use BaP as a surrogate
for both PIC emissions and  unit risk value.   Others use PIC-
 specific emission factors and unit risk factors or comparative
potency factors.  The estimate of cancer incidence is'seen to
vary by a factor of 200 depending on which methodology is used.
While no one methodology has been shown to be a better
methodology  for estimating  risk from PIC,  this study uses the
                         2-42

-------
        methodology that relies on PIC-specific emission factors and
        unit risk factors or comparative potency factors.

        The unit risk factors for estimating risk from PIC are of
        particular concern.  As noted previously in this chapter, many
        of the unit risk factors used in this study have been approved
        by EPA.  The most important exception to the use of EPA
        approved unit risk factors are the group of compounds known as
        PIC.  There is no current EPA unit risk factor for this group,
        although unit risk numbers are available for some of the
        compounds (e.g., BaP) that compose PIC.  The 1985 Six-Month
        Study used a unit risk factor of 4.2X10"1  Ug/nr)~  for PIC.
        This unit risk factor was derived in a highly unusual manner,
        and represents an initial attempt at quantifying the potential
        risk from PIC.  Any estimate based upon this unit risk factor-
        is highly tentative.

        Other unit risk factors for estimating the cancer risk from PIC
        have become available more recently.  These unit risk factors
        represent estimates of risk from PIC mixtures emitted from
        specific source categories (e.g., motor vehicles, hazardous
        waste incinerators).  Some of the more recent unit risk  factors
        were estimated using what is known as the comparative potency
        approach.10  Even though the  more recent factors are  also
        uncertain and have not received the same level  of scrutiny by
        EPA as for other unit risk factors, it was felt  that they were
        an  improvement over the PIC unit risk factor used in the 1985
        Six-Month Study.  Thus, the risk estimates from PIC presented
        in  this report reflect the use, where possible,  of the more
        recently developed unit risk factors for specific PIC mixtures.

        Another source of  uncertainty associated with PIC is the
        selection of  the appropriate unit risk  factor for diesel
        particulates, which  are  included with this group of compounds.
        Unit risk  factors  identified for diesel particulates  range  from
        2xlO~5 to IxlO"4  Ug/m3)"1.  The  EPA  has  not yet  determined  a
        single  best estimate  of  the  unit risk  factor  for these
        particulates.  Thus,  the  estimate of risk  from  all sources  of
        PIC includes  the range  of risk  created  by  the range  in  the  unit
        risk factor for  diesel  particulates.

        Major  uncertainties  exist for many  other  chemicals addressed in
        this report.   For  example, there is considerable debate in  the
        scientific  community concerning the mechanism of carcinogenic
        action  and the  estimation of cancer potency  for dioxin.
        Another unresolved issue concerns  the  relevance to man  of  the
        kidney pathology observed in rats  following  exposure  to
        gasoline  vapors.   A detailed discussion of the  uncertainties
         associated with  risk estimates  for these  and other chemicals is
        outside of the scope of this report.
     10  For a brief discussion on the  comparative  potency approach,  see
page B-110 of Appendix B.
                                  2-43

-------
      In summary, the portion of the entire cancer risk represented by
pollutants and source categories not covered in this study is unknown.
It is expected that the pollutants and source categories covered are
among the most likely major contributors to cancer risk based on our
current state of knowledge regarding carcinogenicity of pollutants and
sources that emit those pollutants.  As new information is obtained,
other pollutants and sources may be found to be as important, or even
more important, contributors to cancer risk.
      As a result of the limitations and uncertainties identified above,
the numerical estimates presented in this report should be viewed only
as a rough indication of the potential for cancer risk caused by a
limited group of pollutants found in the ambient air.  Many of the risks
cited in this report are almost certainly inaccurate in an absolute
sense.  The best use of the risk estimates is in describing the broad
nature of cancer risk posed by these air toxics and by making relative
comparisons of risks across pollutants and sources.
                                  2-44

-------
                               CHAPTER 3.0
               THE MAGNITUDE AND NATURE OF THE CANCER RISK

       In  thfs  study,  the  magnitude  and  nature of the air  toxics problem
 were  evaluated based  upon the  results of  a diverse collection of reports
 and studies.   These reports  and  studies cover many pollutants and  source
 categories,  They also  cover varying geographic areas,  ranging from
 city-specific  studies to  nationwide studies.  The methodology used to
 estimate  the magnitude  and nature of the  cancer risk nationwide from
 this  diverse collection of reports  and  studies was described in the
 previous  chapter.
       In  this  chapter,  the overall  magnitude  of the cancer  risk is
.presented first.   The magnitude  of  the  cancer risk is  presented in terms
 of annual cancer  cases  and lifetime individual risk.   The nature of the
 cancer risk problem  is  then  described  in  terms of individual pollutants,
 source categories, and  geographic variability.  Finally,  the results of
 this study are compared with those  of  the 1985 Six-Month  Study.
       It  is important to  understand that  these estimates  reflect the use
 of either an upper bound  or a  maximum  likelihood estimate of unit  risk;
 that is,  for the  pollutants examined,  the actual cancer risk may be
                                           /
 higher but is  more likely to be  lower.  As discussed  in Chapter 2, this
 occurs because of the manner in  which  EPA calculates  the  unit risk
 factors for toxic pollutants.
                                   3-1

-------
Magnitude of the Cancer Risk Problem
      The magnitude of the cancer risk is presented first in terms  of
total nationwide cancer cases per year and then in terms of lifetime
individual risk.  Both measures of the magnitude of cancer risk play an
important role in the understanding of the problem and in the
development of air toxic control strategies and regulations.  Detailed
analyses for those pollutants that were initially identified as
potentially resulting in ten or more cancer cases per year nationwide
are found in Appendix B.
      Annual Cancer Cases
      The estimates of nationwide annual cancer cases for 26 specific
pollutants are presented in Table 3-1.  The remaining pollutants are
grouped together under "Miscellaneous."  Annual cancer incidence was
calculated by dividing the estimated lifetime incidence levels by 70
years.1
      Both range and point estimates of nationwide annual cancer cases
are presented in Table 3-1.  These estimates were derived, in most
instances, from annual cancer incidence estimates based on both modeled
and ambient-measured concentrations.  The estimates under the column
"Range" reflect a narrowing of  the total range of nationwide annual
cancer incidence that can be calculated from the various studies.  As
seen in Table 3-1, the range of estimates is about two-fold in size,
being approximately 1,400 to 2,900 cancer cases per year.
      1 The  unit  risk  factors  used in this study represent the chance of
 contracting cancer  from  a  lifetime  (70 years)  exposure  to  a given
 concentration  of  that  pollutant.   It was  assumed that  the resulting
 lifetime  incidence levels  could be divided  by 70  to  represent annual
 incidence levels.
                                   3-2

-------
                                  TABLE 3-1

      SUMMARY OF ESTIMATED NATIONWIDE ANNUAL CANCER CASES BY POLLUTANT
                                          ESTIMATED ANNUAL CANCER CASES6
  POLLUTANT
      EPA
CLASSIFICATION6
RANGE
  Totals
                     1,366-2,909
POINT0
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
Acrylonitrile
Arsenic
Asbestos
Benzene
1,3-Butadiene
Cadmium
Carbon tetrachloride
Chloroform
Chromium (hexavalent)
Coke Oven Emissions
Dioxin
Ethyl ene di bromide
Ethylene dichloride
Ethyl ene oxide
Formaldehyde
Gasoline vapors
Hexachl orobutadi ene
Hydrazine
Methyl ene chloride
Perchl oroethyl ene
PIC9
Radionuclides1
Radon1
Tr i chl oroethyl ene
Vinyl chloride
Vinylidene chloride
MiscellaneousJ
Bl
A
A
A
B2
Bl
B2
B2
A
A
B2
B2
B2
B1-B2
Bl
B2
C
B2
B2
B2
A
A
B2
A
C
~
13-14
8-68
82-126
143-181
244-266
6-16
31-47
29-115
113-283
7-11
2-125
25-68
16-45
5-6
124-240
19-76
9
6
3-6
6-13
438-1120
1-3
2
5-13
13-25
0.5-10
15
13
68
88
181
266
10
41
115 „
147-265d
7
2-1256
68
45
6
124 .
19-76f
9
6
5
438-1120h
3
2
7
25
10
15
                1,726-2,706
NOTE:  Values in this figure are not absolute predictions of cancer occurrence
       and are intended to be used in a relative sense only.  The dose-
       response relationships and exposure assumptions have a conservative
       bias, but omissions due to uncharacterized pollutants (either directly
       emitted or secondarily formed) and emission sources, the long-range
       transport of pollutants, and the lack of knowledge of total risk from
       multi-pollutant exposures will offset this bias to an unknown extent.
a These estimates  are  based  on  unit  risk  factors that may overstate  the  actual
 risk.  The  unit  risk factors  for arsenic,  benzene, cadmium,  and  hexavalent
 chromium  are maximum^likelihood estimates.  The  unit  risk  factor for

                                         3-3

-------
 FOOTNOTES TO TABLE 3-1  (continued)

  asbestos is a "best" estimate, very similar to the value that would be
  obtained if the maximum likelihood estimate was calculated.   Unit risk
  factors for PIC have no classification.  All other unit risk factors are
  upper-bound estimates.

 b For a discussion of how EPA evaluates suspect carcinogens and more
  information on these classifications, refer to "Guidelines for Carcinogen
  Risk Assessment" (51 Federal Register 33992).  The EPA classifications used
  in this report are:

  A «  proven human carcinogen

  B -  probable  human carcinogen  (Bl  indicates  limited evidence  from  human
       studies and  sufficient  evidence  from  animal studies;  82  indicates
       sufficient evidence from  animal  studies  but inadequate evidence from
       human  studies)

  C =  possible  human carcinogen

 0 If a range is  shown, it was considered unreasonable to select a point
  estimate.

 d Range primarily reflects uncertainty with  the exposure to hexavalent
  chromium from  cooling  towers.   Some uncertainty to actual  exposure to
  hexavalent chromium from all sources exists because the percent of total
  chromium that  is  hexavalent is still  being evaluated for most sources.

 e Range reflects great uncertainty associated with exposure to  dioxin from
  treatment,  storage, and disposal  facilities,  and from  municipal  waste
  combustors. Other uncertainties associated with dioxin estimates include
  sampling method and extrapolation  from tetrachlorinated dibenzodioxin  (TCDD)
  to the other dioxin subspecies.

 f Range reflects different assumptions  as to which portion of gasoline vapors
  is carcinogenic.   The  upper end  of the range assumes  all  gasoline vapor  is
  carcinogenic;  the lower end assumes only the C6 and higher fraction of the
  gasoline vapor is carcinogenic.

9 PIC (products  of  incomplete combustion)  is a group of  pollutants that  have
  not been very  well  defined and for which EPA has not developed a
  classification.   It is composed  of some pollutants,  such as BaP,  for which
  EPA has developed a classification.   BaP is a B2 pollutant (probable human
  carcinogen).

h Range reflects the  use of  two  unit risk factors for diesel particulates.

1 From sources emitting  significant  amounts  of radionuclides (and  radon) to
  outdoor air.   Does  not include exposure  to indoor  concentrations of radon
  due to radon in soil gases entering homes  through  foundations  and cellars.

* Includes individual  pollutants primarily from the  TSDF study  and the
  Sewage Sludge  Incinerator  study.
                                         3-4

-------
      The estimates under the column entitled "Point" reflect an attempt
to derive a single estimate of nationwide annual cancer incidence.  For
most pollutants, a reasonable point estimate could be selected.  Point
estimates are reported in this column when either only one study
reported that pollutant  (and as a single point estimate) or the analysis
for that pollutant (see  Appendix B for discussion) suggested that a
single point estimate was a better indicator of risk than a range.  For
four pollutants, as discussed below, this could not be done.  For these
four pollutants, a narrower range was estimated.  As seen in Table 3-1,
the "point" estimates narrow the overall range slightly, to between
approximately 1,700 and  2,700 cancer cases per year nationwide.  Between
25 and 40 percent of this range is attributable to the cancer risk
estimated for products of incomplete combustion (PIC).  As noted in
Chapter 2, there are many uncertainties associated with estimates of
cancer risk from PIC.  .
        As noted above,  a point estimate was not reasonable for four
pollutants.  These four  pollutants are PIC, dioxin, gasoline vapors, and
hexavalent chromium.   In the case of PIC, the large range is created
primarily by the uncertainty of the unit risk factor associated with
diesel  particulates,  which are included in the estimates of risk for
PIC.  A single unit risk factor has not been identified by EPA's Office
of Research and Development for diesel  particulates.   Instead, a range
of unit risk factors, from 1.0 x 10"9 to  2.0 x 10"5 (^g/m3)"1, has been
identified by EPA's Office of Mobile Sources.   This' range was used to
estimate the cancer risk from diesel  particulates and is reflected in
the total estimated cancer risk from PIC.
      In addition,  uncertainty lies in  the methodologies available for
estimating risk (as discussed in Chapter 2 and Appendix B)  for PIC.   One
                                  3-5

-------
methodology uses benzo(a)pyrene (BaP) concentrations and BaP unit risk
factors to estimate risk from PIC.  This methodology assumes BaP is the
only carcinogenic component of PIC.  More recently developed
methodologies use source-specific PIC emission factors and unit risk
values to estimate risk.  None of the methodologies have undergone a
high degree of scrutiny at this time.  Thus, although this report uses
the more recently developed methodologies to estimate cancer risk from
PIC, sufficient uncertainty remains concerning all methodologies
associated with PIC risk estimation such that a range might still have
been selected as the reasonable best estimate for PIC, even if a single
unit risk factor could be identified for diesel particulates.
      The range for dioxin is the result of difficulties with the
sampling methodologies used to estimate emissions of dioxin and with the
methodology used to extrapolate risk from tetrachlorinated dibenzodioxin
(TCDD) to the other dioxin subspecies.  In addition, much of the range
is the result of the uncertainty associated with the risk of dioxin from
treatment, storage, and disposal facilities (TSDFs) for hazardous waste.
Although the TSDF study allows the calculation of a single point
estimate (of 91 cancer cases per year), the underlying emissions data
are very uncertain.  Actual cancer cases attributable to dioxin
emissions from TSDFs could be considerably less.  Finally, early data on
municipal waste combustors (MWCs) showed a wide range of estimated
annual cancer cases (approximately 2 to 20).  Recent revisions to the
MWC study suggest that the estimated risk attributable to dioxin may be
one-half this estimate.  For these reasons, no attempt was made to
develop a point estimate for dioxin.
      For gasoline vapors, the range in estimated risk reflects the
uncertainty over quantifying the emissions that are associated with the
                                   3-6

-------
cancer-causing portion of gasoline vapors.   It has been suggested that
only a portion (i.e., only those C6 and  higher components),  rather than
all, of total gas vapors are carcinogenic.   At this time, it is
uncertain as to which provides a better estimate of the emissions of
concern.
      For chromium, the range reflects uncertainty over the ratio of
hexavalent chromium to total chromium emissions for various chromium
emission sources.  Several studies (e.g., the 5-City study) attempt to
consider available information on the estimated ratios of hexavalent
chromium to total chromium for specific sources.  For cancer risk
estimates based on ambient-measured concentrations of chromium,
estimating cancer risk is complicated by the fact that the sources that
contribute to the ambient measured chromium concentrations are not
identified.  Thus, estimating what fraction of total measured chromium
may be hexavalent is even more difficult and uncertain.  This degree of
uncertainty makes any single estimate untenable, and therefore a range
has been retained at this time.
      Lifetime Individual Risk
      In addition to annual incidence, the magnitude of cancer risk from
air toxics can be described in terms of an individual's lifetime risk.
The lifetime individual risk is a measure of the probability that an
individual will develop cancer as a result of exposure to the ambient
concentration of an  air pollutant over a lifetime  (i.e., a 70-year
period).2  The ambient concentration used to calculate lifetime
     2 Lifetime individual risk is calculated as follows:
 Lifetime  individual  risk  =  (exposure concentration) x  (unit risk factor)
                                   3-7

-------
individual risk may be measured or model-predicted.   Lifetime individual
risk at a particular location is calculated by multiplying the unit risk
factor by the estimated long-term average exposure at that location.
Where the average ambient concentration is representative of an entire
geographic locale (e.g., a city), the term "areawide" or "urban-wide"
lifetime individual  risk can be used.
      Frequently, the lifetime individual risk is reported as "maximum
individual risk" (MIR).  Maximum individual risk refers to an estimate
of the maximum level of lifetime individual risk to which a person could
be exposed.  The MIR is calculated at the specific location near an
emission source where the highest long-term average concentration is
predicted.  It is best characterized, especially when developed as part
of preliminary risk assessments, as a rough measure of the potential
maximum individual lifetime cancer risk associated with exposure to the
maximum modeled long-term concentration.  The MIR is not an appropriate
measure of the risk level affecting the entire population residing near
a particular facility, but rather only to the individuals residing at
the specific point of estimated maximum exposure.
      The highest predicted modeled concentration may or may not always
occur at a point where an individual actually lives.  When the highest
predicted modeled concentration is found to occur in an inhabited area,
the term "maximum exposed individual" (MEI) may be used to refer to the
maximum individual risk to which an individual is exposed.
      Highly spatially-resolved models are recommended for calculating
MIRs.  These models, such as EPA's HEM-SHED, calculate individual risks
close in (<1 kilometer) for all potential receptor locations around
specific point sources.  Some model-based studies, however, use a larger
                                   3-8

-------
spatial setting of grid cells, and thus may not identify the highest
modeled concentrations.  Similarly, measured concentrations are unlikely
to identify the highest concentrations because of the too few monitoring
sites generally used in most studies.  Thus, in monitoring-based studies
and model-based studies, it is often more appropriate to refer to the
"maximum individual risks" reported as either "highest observed
individual risks" or "highest grid-cell individual risks," respectively.
      For purposes of this study, the term "maximum individual risk"
(MIR) is used to refer to the highest lifetime individual risk reported
in the various studies and reports.  MIRs were estimated for individual
sources (e.g., waste oil combustors, POTWs), individual pollutants
(e.g., arsenic, benzene), and locations (e.g., traffic intersection,
                                                  \
geographic locale).  MIRs for individual sources reflect the aggregate
risk associated with multiple pollutants emitted from that source.  MIRs
for individual pollutants reflect the risk for that pollutant either
from an individual plant within a particular source category (e.g.,
waste oil combustors) or from sources across multiple source categories
(e.g., the Ambient Air Quality study).  MIRs for locations reflect the
aggregate risk associated with multiple pollutants and sources.
      Table 3-2 summarizes the maximum individual risks reported in the
various studies for individual pollutants and facilities.  Almost all of
the studies reported maximum  individual risks of at least 1.0 x 10~4.
Many studies  showed maximum individual risks of 1.0 x 10"3 or higher.
Where  appropriate and where possible, footnotes are used to further
clarify the types of lifetime individual risks and the procedures used
to calculate  them.
                                   3-9

-------
                                      TABLE  3-2

              SUMMARY  OF MAXIMUM INDIVIDUAL RISKS OF CANCER AS
                         REPORTED  IN THE VARIOUS STUDIES
STUDY
Waste Oil
Combustion
Hazardous
Waste
Combustors
TSOF
Sewage Sludge
Incinerators
Municipal Waste
Combustors
POTW
Coal and Oil
Combustion
Drinking Water
Aerators
Asbestos*1
South Coast
Southeast
Chicago
INDIVIDUAL SOURCE/
POLLUTANT/LOCAT I OH
Individual source3
Arsenic
Cadmium
Individual source
Individual source0
Individual source
Individual source6
Individual source
Individual source9
Arsenic
Beryllium
Cadmium
Hexavalent chromium
POM
Formaldehyde
Individual source
Fabricating
Hilling
Renovation
removal
dispose
Demolition
removal
disposal
Benzene
Hexavalent chromium
Grid cell (populated)1
MAXIMUM
LIFETIME
INDIVIDUAL
RISK
1.8X10"4
1.6x10"*
2.1X1Q"5
<1x10"7 to 1x10"4
2x1 O"2
5x10"2
10~5 to 10~3
4.5x10"2
7x10";!
4x10";?
2x10";!
1x10";
8x10"'
5x10"°
1x10"'
2x10"8 to 2x10"5
2x1 0'3.
3x10"S
6x10"£
3x10"s
4x10"!j
7x10"3
10"* to 10"!?
10"4 to 10"-5
9x10"4
REFERENCES
1
2
3
4.
S
6
7
8
9
10
11
IEMP-BaltimoreJ
lEHP-Santa  Clara
           Benzene
         Chloroform
 Hexavalent chromium
        Five others

Traffic intersection
           Benzene
     Ethylene oxide
 1.0x10
 1.1x10
<3.6x10
<6.8x10

   3x10
   2x10
   2x10
-4
-4
-4
-5

-4
-4
-4
                                                12
                                                                    13
                                         3-10

-------
             TABLE 3-2  (concluded)

SUMMARY OF MAXIMUM INDIVIDUAL RISKS OF CANCER AS
         REPORTED  IN'THE VARIOUS  STUDIES
STUDY
lEMP-Kanawha Valley1
lEHP-Philadelphia
Ambient
Air
Quality"1














NESHAP/ATERIS
Data Basen





















POLLUTANT/SOURCE
CATEGORY
Institute
Chemical mfg.
PIC
Arsenic
Cadmium
Hexavalent chromium
Chloroform
Benzene -
1,3 butadiene
Carbon tetrachloride
Ethylene dibromide
Ethylene dichloride
Formaldehyde
Hethylene chloride
Styrene
Perch loroethylene •
Tried loroethylene
Vinyl chloride
Vinylidene chloride
Acetaldehyde
Acrylonitrile
Arsenic
Benzene
Beryllium
Butadiene
Cadmium
Carbon tetrachloride
Chloroform
Hexavalent chromium
Coke oven emissions
Ethylene dichloride
Epichlorohydrin
Ethylene oxide
Hexach lorobenzene
Formaldehyde
Methylene chloride
Perch loroethylene
p-dich lorobenzene
Styrene
Tri ch I oroethy I ene
Vinyl chloride
Vinylidene chloride
MAXIMUM
LIFETIME
INDIVIDUAL
RISK
8x10"3
2.2x10"4
8.4x10"-*
3.9x10"*
3.3x10"*
3.7x10",
6.4x10"*
1.7x10"*
1.3x10"*
5.2x10";!
7.9x10"^
1.1x10,
3.0x10"*
9.6x10"*
.8x10"°
.2x10";!
.2x10";!
.Ox10"j!
.7x10"5
5.0x10^*
3.8x10",
1,2x10"*
6.0x10";:
1.9x10";
3.2x10-,!
1.2x10-*
5.7x10-^
2.0x10-^
1.8x10-!
3.4x10-^
1.1x1
-------
FOOTNOTES  TO  TABLE 3-2

a  The MIRs for the individual source and the  pollutants are assumed representative of the entire
   population of waste oil combustors.

k  Range covers individual HIRs for each modeled facility  in the source category.

c  MEI to "highest annual average ambient concentration around a TSDF."

^  Based on 10th percentile of all sewage sludge incinerator test data for a non-specified facility.

e  For existing facilities.  Range associated  with  MIRs at different types of municipal waste
   combustors.

f  MIR is associated with one specific POTW.   Other POTWs  have  lower MIRs.

9  HIRs were calculated for three types of boilers  (industrial, commercial, utility) and two types
   of firing (oil-fired and coal-fired) for each type of boiler.  This MIR is_associated with an
   oil-fired, commercial boiler.  The range of MIRs estimated was from 2 x 10"   (oil-fired, utility
   boiler) to 7 x 10~5 (oil-fired, commercial  boiler). The MIRs for the individual pollutants are
   associated with oil-fired commercial boilers except for POM  (coal-fired commercial boiler) and
   radionuclides (coal-fired industrial and utility boilers).   For  additional  information, see
   Appendix C, page C-10.

^  MIR not absolute maximum, but reasonable estimate of highest risk expected.

*  An MIR of 5 x 10"^ was estimated for a grid cell, but census data indicated that no one was
   living in that grid cell.

J"  Based on highest average value reported for the  pollutant at any of the monitoring sites.

k  MIR for the traffic intersection is associated with risks from four pollutants.  The MIR for
   benzene is based on maximum concentration at a traffic  intersection.  The  MIR for ethylene oxide
   is based on maximum concentration  at a hospital.

1  Site of MIR is near a  specific facility in Institute, WV, and  is based on  exposure to six
   pollutants.

m  Based on highest arithmetic mean concentration observed.

n  The  lifetime individual  risks  from the ATERIS database are  highly uncertain.  The ATERIS contains
   data from all stages  of  air toxics analyses, from the very  preliminary  to  the more detailed.
                                                3-12

-------
REFERENCES TO TABLE 3-2

 1.  Peters, W.,  Duggan,  G.,  and  R.  Fegley.  Waste Oil Combustion Cancer Risk Assessment.  Technical
     Staff Paper.  October 1987.   page 3.

 2.  U.S. EPA, Office of  Solid Waste.  Regulatory Impact Analysis for Hazardous Waste Boilers and
     Industrial Furnaces.  Draft.  Exhibits 7-6, 7-9, 7-12, and 7-14.

 3.  U.S. EPA, OAQPS.  Hazardous  Waste TSDF - Background Information for Proposed RCRA Air Emission
     Standards.  Volume 1 - Chapters.  Preliminary Draft EIS.  March 1988.  p. 6-10.

 4.  U.S. Environmental Protection Agency.  Standards for the Disposal of Sewage Sludge.  Proposed
     Rule.  February 6, 1989.  54 FR 5783.

 5.  U.S. EPA, Office of  Solid Waste and Emergency Response.  Municipal Waste Combustion Study:
     Report to Congress.   EPA-530-SW-87-021a.   June  1987.  p. 86.

 6.  Memorandum.   R.B. Lucas, U.S. EPA, Chemicals and Petroleum Branch, to J. Padgett, U.S. EPA,
     OAQPS.  New Study on the Air Toxics Problem in  the United States - POTW Emissions.  July 29,
     1988.  3 pages.

 7.  Peters, W.D., U.S. EPA,  Pollutant Assessment Branch.  Coal and Oil Combustion.  July 25, 1988.
     6 pages.

 8.  Memorandum.   W.D. Peters, U.S.  EPA, PAB, and S.W. Clark, U.S. EPA, STB to R.G. Kellam, Program
     Analysis and Technology Section,  and A.H.  Perler, Science and Technology Branch.  Risks
     Associated With Air  Emissions from Aeration of  Drinking Water.  November 10, 1985.

 9.  U.S. EPA, ESED.   National Emission Standards for Asbestos - Background Information for Proposed
     Standards.  Draft.  March 5,  1987.

10.  South Coast study, p. vi-6.

11.  Southeast Chicago study, p.  43.

12.  lEMP-Baltimore study. Tables V-8  and V-14.

13.  lEMP-Santa Clara study,  pp.  3-82  and 3-112.

14.  lEMP-Kanawha Valley study, p. 4-94.

15.  lEMP-Philadelphia study, p.  VI-22.

16.  Ambient Air Quality study. Tables 9 and  10.

17.  ATERIS Database printouts.  1989.

18.  Table E-1. Major Lifetime Risk  and Cancer  Incidence for the Four Major AN Source Categories.
     (Personal communication from I la  Cote, USEPA,  to Ken Meardon, PES).

19.  U.S. EPA.  National  Emission Standards for Hazardous Air Pollutants; Benzene Emissions from
     Haleic Anhydride Plants. Ethvlbenzene/Styrene  Plants. Benzene Storage Vessels. Benzene
     Equipment Leaks, and Coke By-Product Recovery  Plants.  Proposed rule and notice of public
     hearing.  July 28, 1938.  53 FR 28496.

20.  U.S. EPA.  Coke Oven Emissions  from Wet-Coal Charged By-Product Coke Oven Batteries --
     Background  Information for Proposed Standards.  EPA-450/3-88-028a.  April 1987.  p. E-30.

21.  Memorandum.  L.J. Zaragoza,  Pollutant Assessment Branch.  Hexachlorobenzene Exposure and Risk
     Assessment.  December 1.1, 1984.  Docket  No. A-84-39,  Item II-B-1.
                                               3-13

-------
      The NESHAP/ATERIS data base showed maximum individual  risks of 1.0
x 10"3 or  higher  for  12  pollutants in at least one location.3  These
risk estimates are associated with individual sources within specified
source categories.  The EPA re-estimated the maximum individual  risk for
182 of the 205 facilities that were identified as having maximum
individual risks of 1 x 10"3  or  higher  in the ATERIS  data base.   This
was done by collecting current information on emissions from these
facilities under Section 114 of the Clean Air Act.   Overall, the new
analysis showed that estimates of maximum individual risk increased for
12 facilities and decreased for 170.  One reason for the generally lower
risk estimates is that some of the companies have taken steps since the
previous emissions data were collected to reduce emissions through
process changes and control devices.  The new risk estimates generally
suggest that the maximum individual risk estimates in the ATERIS data
base are too high.  Nevertheless, the estimates for  some of the  182
facilities analyzed continue to be of serious concern.
      On a source category basis, some of the source-specific studies
identified a single maximum  individual risk value and others reported a
range of MIR values.  In some instances, only the highest maximum
individual risk associated with a specific facility  was reported (e.g.,
POTWs and TSDFs).  Other facilities in such source categories would have
lower MIRs than those shown  in Table 3-2.  In other  instances, a single
MIR was reported that could  be expected at a typical, but unspecified
facility  in a source category (e.g., waste oil combustors).  Where a
     3 The ATERIS contains  data  from all  stages of air toxics analyses,
from the very preliminary to the more detailed.   It is not considered an
authoritative  source  for   verified  estimates  of risk  attributable to
individual  point sources,  and should  not be  relied upon  as credible
estimates of individual source cancer risks.   Therefore,  the estimates of
MIR in the ATERIS data  base are subject to significant uncertainty.
                                  3-14

-------
range of MIRs are shown, the values cover either all of the MIRs for
each of the facilities modeled (e.g., hazardous waste combustors,
drinking water aerators) or the highest MIR expected at typical
facilities within the source category (e.g., municipal  waste
combustors).
      Maximum individual risks associated with individual sources within
a source category can vary.  Table 3-3 illustrates this variation using
the distribution of MIR values associated with hazardous waste
combustors.  Under current conditions (referred to as "baseline" in the
table), the majority of these hazardous waste combustors have MIR values
of less than 10"7; however,  several may have MIR values  of  10"4,
depending on the type of waste being burned.  After compliance with
proposed regulations, the highest MIR value decreases to 10"5,  and  there
is a reduction in the number of hazardous waste combustors associated
with each level of MIR risk.  A second illustration is  provided in Table
3-4, which shows the distribution of MIRs associated with coke ovens.
      Another way to examine MIR is to look at the number of people
exposed to various MIRs.  This is illustrated in Table  3-5, which shows
the distribution of people exposed to MIRs for drinking water aerators.
      Areawide lifetime individual risks are shown in Table 3-6.  These
risks are in the 10"5  to  10~4 range.   Compared to the maximum individual
risks for the corresponding cities shown in Table 3-2,  the areawide
risks are approximately one order of magnitude lower.
      Lifetime individual risks were calculated in the  Ambient Air
Quality study at monitoring sites in four cities (see Table 3-7).
Lifetime individual  risks in the Ambient Air Quality study were defined
as "the sum of lifetime individual risks for metals, BaP, VOC, and PIC
at a monitoring site within a city where a battery of air toxic
                                  3-15

-------
                                                TABLE 3-3

     DISTRIBUTION  OF  MAXIMUM  LIFETIME  INDIVIDUAL CANCER RISKS TO THE MOST  EXPOSED
             INDIVIDUAL  FROM  HAZARDOUS WASTE COMBUSTORS  - BOILERS AND  FURNACES
MAXIMUM
INDIVIDUAL
RISK
>1 X 10"4
1 X 10~4
1 X 10"5
1 X 10"6
1 X 10"7
<1 X 10~7
Total
>1 X 10~4
1 X 10'4
1 X 10~5
1 X 10"6
1 X 10~7
<1 X 10~7
Total
TYPE OF
WASTE
Base Case






High Risk






CONTROL DEVICE PERFORMANCE
Base Case"
Baseline
0
0
10
61
103
778
952
0
19
100
167
198
468
952
After Compliance
0
0
6
48
56
650
795C
0
0
73
52
35
595
755C
Pessimistic"
Baseline
0
0
10
65
101
777
952
0
21
102
167
207
456
953
After Compliance
0
0
6
48
72
634
759C
0
0
73
58
36
585
752C
NOTE:  Numbers in table indicate the number of hazardous waste combustors associated with each maximum individual
      risk level.

SOURCE:  U.S. EPA, Office of Solid Waste.  Regulatory Impact Analysis for Hazardous Waste Boilers and Industrial
        Furnaces. Exhibits 7-6, 7-9, 7-12, and 7-14.

8 "Base case" assumes "typical" removal efficiencies for control  devices.

  "Pessimistic" assumes removal efficiencies of control devices for toxic metals and hydrogen chloride are several
  percentages points lower than in the base case in most cases.  For organic  compounds the difference is several
  fractions of a percent in most instances.

c Difference in total device due to some devices that discontinue burning due to the regulations.
                                                   3-16

-------
                                TABLE  3-4

   MAXIMUM  LIFETIME  INDIVIDUAL  CANCER RISKS  FROM  COKE OVEN  EMISSIONS
MAXIMUM LIFETIME
INDIVIDUAL RISK
> 10"2
10'3 to 10'2
10'4 to 1(T3
NUMBER OF
COKE OVENS
13
25
5
SOURCE:  Appendix E. Coke Oven Emissions Risk Assessment for Wet-Coal
         Charged Coke Oven Batteries.
                                  3-17

-------
                                TABLE 3-5

             DISTRIBUTION OF MAXIMUM  INDIVIDUAL CANCER RISK
                      AT 22  DRINKING WATER AERATORS
MAXIMUM
INDIVIDUAL RISK
1.9xlO'5
1.3X10'5
9.5X10'6
4.6xlO'6
2.9X10'6
l.SxlO'6
1.4X10"6
l.lxlO'6
l.OxlO"6
1(T7
lO'8
NUMBER OF
FACILITIES
1
1
1
1
1
1
1
1
1
11
2
NUMBER OF PEOPLE
EXPOSED
439
7
4
28
33
30
2
1
11
208
13
SOURCE:  Memorandum.  W.D. Peters, US EPA, Pollutant Assessment Branch,
         and S.W. Clark, US EPA, Science and Technology Branch, to R.G.
         Kellam, US EPA, Pollutant Assessment Branch, and A.M. Perler,
         US EPA, Science and Technology Branch.  Risks Associated with
         Air Emissions from Aeration of Drinking Water.  November 13,
         1985.  Table 5.
                                  3-18

-------
                                                           TABLE  3-6

                        AREAWIDE LIFETIME  INDIVIDUAL  CANCER  RISKS  FOR SELECTED CITIES
CITY/LOCALE
Philadelphia6
Santa Clarac
Southeast Chicago
City Ae
City Be
City Ce
City De
City Ee
Kanawha Valley*
Baltimore9
South Coast*1
AREAWIDE
LIFETIME
INDIVIDUAL RISK
4.0x10~5, 1.2x10"4
4x10"5
2.2x10"4
1.4x10"4
4.3x10~4
2.0x10"4
7.0x10"4
2.7x10"4
5.0x10~4, 1.2x10~3
1.3x10"4, 3.3x10"4
1.2x10~4, 2.1x10"4
NUMBER OF
POLLUTANTS
7
--
30
20
20
20
20
20
18
9
12
MAJOR POLLUTANT
CONTRIBUTORS TO
INDIVIDUAL RISK8
Benzene, carbon tet.
1,2 dichloropropane
Carbon tetrachloride
Coke oven emissions,
Cr+6
Formaldehyde,
PIC, 1,3-butadiene
1,3-butadiene, PIC,
Cr+6, formaldehyde
1,3-butadiene,
formaldehyde
PIC, 1-3-butadiene,
Cr+6
PIC, formaldehyde,
Cr+6
Ethylene oxide,
1,3-butadiene
Benzene, Cr+6
Benzene, Cr+6
                 Note:   In some  instances, the areawide  lifetime individual  risk was calculated by dividing total
                        lifetime cancer cases by exposed population.   Where  possible and as appropriate, these
                        estimates were adjusted based on unit risk factors used in this study.

                 a Cr+6 = hexavalent chromium
                   Carbon tet. = carbon tetrachloride

                 k IEMP Philadelphia study, p. V-27,  Lower estimated based  on modeled data; higher  estimate, on
                   monitored data.

                 c IEMP Santa Clara study, p. 3-80.

                   Southeast Chicago study, p. 38.

                 e Five City study, p. 53.

                 * IEMP Kanawha  Valley study, pp. 4-116  and 4-117.  Higher estimate based on box model
                   concentrations; the lower, on Gaussian model analysis.  These estimates are for the
                   entire Kanawha Valley study area.

                 9 IEMP Baltimore study. Tables V-7 and  V-13.  Range created by range of estimated risk
                   for hexavalent chromium and cadmium.

                 h South Coast study, p. VI-11.  Lower estimate based on modeled data; higher estimate on
                   ambient measured data.
                                                               3-19
_

-------
                               TABLE 3-7
              SUMMARY OF LIFETIME INDIVIDUAL CANCER RISKS
                          FOR SELECTED CITIES
CITY
Los Angeles
Baton Rouge
Boston
Chicago
LIFETIME
INDIVIDUAL
RISK3
6.6 x 10'4
3.8 x 10'4
3.0 x 10"4
3.2 x 10'3
NUMBER OF
POLLUTANTS"
17
16
11
14
MAJOR POLLUTANTS
CONTRIBUTING TO
LIFETIME INDIVIDUAL
RISK
Formaldehyde, PIC
Ethylene Dichloride,
PIC
PIC, Chromium
(hexavalent)
PIC, Formaldehyde
SOURCE:  Ambient Air Quality Study, Table 8.
a These risks are the sum of the lifetime individual  risks  for a number
  of pollutants using the estimated annual average concentration at a
  monitoring site within each of the four cities.
b Includes nickel, but no cancer incidence was attributed to nickel.
                                  3-20

-------
pollutants is being monitored."  As seen in Table 3-7, lifetime
individual risks on the order of 10"4 and higher were  found.  The
magnitude of lifetime individual risks is affected by the number of
pollutants as well as the particular pollutants included.  The number of
pollutants monitored ranged from 11 in Boston up to 17 in Los Angeles.
None of the cities had data on 1,3-butadiene, a pollutant found in this
study to be one of the major contributors to risk.  In addition,
formaldehyde data, another major contributor to risk, were unavailable
for Baton Rouge and Boston.
Nature of the Cancer Risk Problem
      The nature of the cancer risk problem is examined by looking at
the relative contributions of individual pollutants and source
categories to total estimated nationwide annual cancer incidence.   In
addition, the geographic variability of the cancer risk is examined by
comparing reported ambient concentrations of selected pollutants,
estimated annual cancer incidences, and estimated lifetime individual
risks.
      Individual Pollutants
      Table 3-8 presents the percent contribution of individual
pollutants to the total estimated cancer cases.  The percent
contributions were calculated using the point estimates presented in
Table 3-1.  Where a range is indicated in Table 3-1, the midpoint was
used to estimate the pollutant's potential  relative contribution.
Figure 3-1 illustrates the results presented in Table 3-8.
      Based on the estimates in Table 3-1,  five pollutants -- PIC, 1,3-
butadiene, chromium, benzene, and formaldehyde -- account for
approximately 70 percent of the total estimated annual cancer cases.
The reader is reminded that there is considerable uncertainty associated
                                  3-21

-------
                                 TABLE 3-8

          RELATIVE CONTRIBUTION  OF  INDIVIDUAL  POLLUTANTS TO
                         TOTAL  ESTIMATED  CANCER CASES


1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.


POLLUTANT
PIC
1,3-Butadiene
Chromium (hexavalent)
Benzene
Formaldehyde
Chloroform
Asbestos
Arsenic
Ethyl ene di bromide
Dioxin
Gasoline vapors
Ethyl ene dichloride
Carbon tetrachloride
Vinyl chloride
Acrylonitrile
Cadmium
Vinyl idene chloride
Hexachl orobutadi ene
Trichloroethylene
Coke Oven Emissions
Perch! oroethyl ene
Hydrazine
Ethyl ene oxide
Methyl ene chloride
Radionuclidesa
Radon3
Miscellaneous
Totals
PERCENT
CONTRIBUTION
35.2
12.0
9.3
8.2
5.6
5.2
4.0
3.1
3.1
2.9
2.1
2.0
1.9
1.1
0.6
0.5
0.5
0.4
0.3
0.3
0.3
0.3
0.3
0.2
0.1
0.1
0.7
100.0
NOTE 1:  Values in this figure are not absolute predictions of cancer occurrence and
         are  intended to  be used in a relative sense only.  The dose-response
         relationships and exposure assumptions have a  conservative bias, but
         omissions due to uncharacterized  pollutants 
-------
o
co
o
CM
o
T—
o





0
CL
3*.



"p
NOTE: Values I" this figure are not absolute predictions of cancer occurrence
and are Intended to be used in a relative sense only. The dose-response
relationships and exposure assumptions have a conservative bias, but
omissions due to uncharacterized pollutants (either directly emitted or
secondarily formed) and emission sources, the long-range transport
of pollutants, and the lack of knowledge of total risk from multi-
pollutant exposures will offset this bias to an unknown extent

| 1 ,3-Butadiene
mium, hexavalent
.
u
; §
i
= pC".?V
< • {•'.'•

'/ ' ' • J • • '
Hi


Others
Acrylonitrile
Cadmium

\\\\\§ . Benzene
Formaldejiyde
i

'.'.' i .'.'.
i'.'.' s.'.'.
s •' !"•
* " JiL
£ 	
S
~>, '.'.'
b
-

Vinylidene chloride
Hexachlorobutadiene
Trichloroethylene
•i
CD
1
= '

I
R"5!
yi


Coke oven emissions
Perchloroethylene
Hydrazine

CD
C
IT
? ••
« . .


c
1
111



Ethylene oxide
Methlyene chloride
Radon
Radionuclides
56 other pollutants





Gasoline vapors
Ethylene dichloride
Carbon tetrachloride
Vinyl chloride
S . . i
: . . . t
ititiip

CO
03
O
k.' '.
0 10 20 30 4(
PERCENT CONTRIBUTION (%)
                                                CD
                                               •a
                                               T3
                                               CD

                                               •S •£

                                               .i §
                                               •4—' "tli
                                               CO _5

                                               m o
                                               "CO D-

                                               "o >•
                                               i   O


                                               Q S3
                                               C CD
                                               g >-

                                               "3 CD
                                               .g Q_

                                               4±; CO
                                               ^ CD


                                               8 %

                                               "^
                                               .> CD

                                               I i
                                               CD CO
                                               DC O
                                               CO

                                                CD
                                                D)

                                               LJL
iNvinnod nvnaiAiQNi
                3-23

-------
with  the  absolute  risk  estimates for many of the pollutants examined in
this  study.  The relative  contribution of any one pollutant is subject
to  change due  to the  uncertainties associated with the risk estimates.
In  addition, the relative  contribution may change as new information on
as  yet  unquantified air toxics  is obtained.
      Source Categories
      As  noted earlier, over 60 source categories were identified in the
studies and reports used.  Table 3-9 summarizes the estimated annual
cancer  cases associated with 40 specified-types of source categories,
one aggregate  category, and two unspecified categories.  The relative
contributions  of the  individual source categories are illustrated in
Figure  3-2.  It is important to remember that not all source categories
that  emit air  toxics  were  covered in the studies and reports used for
this  study.  Thus, the  relative contributions presented in this study
can only  reflect the  source categories that were covered.  As new
information is developed,  these relative contributions could change,
perhaps significantly.
      The estimates presented in Table 3-9 reflect the range estimates
for modeled estimates only.  (By their nature, ambient-measured data do
not distinguish between sources.)  Although secondary formaldehyde is
not a modeled  source  category per se, two studies (the 5-City study and
the Southeast  Chicago study) attributed,the difference between ambient-
measured  concentrations and the modeled concentrations to the secondary
formation of formaldehyde.  It is these results that are included in
Table 3-9.  Based on  the range estimate for the modeled estimates,
between 1,430  and 2,538 cancer cases per year are estimated.
      Individual source categories have frequently been grouped in two
ways:    (1) mobile vs. stationary and (2) point vs. area sources.  Mobile
                                  3-24

-------
                                    TABLE 3-9

SUMMARY OF ESTIMATED CANCER CASES BASED ON MODELED AMBIENT CONCENTRATIONS,  BY
                               SOURCE  CATEGORY
ANNUAL
SOURCE CATEGORY CANCER CASES
1.
2.

3.
4.
5.
6.
7.
8f
9.
10.

11.
12.
13.

14.
15.
16.
17.
18.

19.
20.
21.
22.

23.
24.
25.

26.
27.

28.
29.
30.

31.
32.
33.
34.
35.
36.
37.

Motor Vehicles
Secondary
Formaldehyde
Electroplating
TSDFs
Woodsmoke
Asbestos, Demolition
Unspecified (point)
Cooling Towers
Gasoline Marketing
Solvent Use/Oegreasing

Unspecified (area)
PVC/EDC/Vinyl Chloride
Iron and Steel

Sewage Sludge Incinerators
Municipal Waste Combustors
Petroleum Refineries
1,3-Butadiene Production
Styrene'butadiene Rubber
Production
Coal and Oil Combustion
POTWs
Smelters
Commercial Sterilization/
Hospitals
Pesticide -Production/Usage
D rye lean ing
Pulp and Paper Manufac-
turing
Chlorinated Drinking Water
Ethylene Dibromide
Production
Polybutadiene Production
Ethylene Oxide Production
Ethylene Dichloride
Production
Waste Oil Burning
Asbestos Manufacturing
Asbestos Renovation
Glass Manufacturing
Hazardous Waste Combustors
Paint Stripping
Pharmaceutical Manufac-
turing
769-1,461

106-154
120
49-140
89
81
27-92
0.01-111
24-75
22-36

21
19
17-18

13 •
2-22
8-14
10

10
8-10
6
3-4

3-4
3.4
3

2.1
1.7

1.5
1,2
1.2

0.8
0.6
0.5
0.4
0.4
0.3
0.22

0.2-0.4
(percent of
total)
(54-58)
_„,.
(7.4-6.1)
(8.4-4.7)
(3.4-5.5)
(6.2-3.5)
(5.6-3.2)
(1.9-3.6)
(0.0-4.4)
(1.7-3.0)
(1.5-1.4)

(1,5-0.8)
(1.3-0.7)
(1.2-0.7)

(0.9-0.5)
(0.1-0.9)
(0.6-0.6)
(0.7-0.4).

(0.7-0.4)
(0.6-0.4)
(0.4-0.2)
(0.2-0.1)

(0.2-0.2)
(0.2-0.1)
(0.2-0.1)
,
(0.1-0.08)
(0.1-0.08)

(0.1-0.06)
(0.08-0.05)
(0.08-0.05)

(0.08-0.05)
(0.04-0.02)
(0.04-0.02)
(0.03-0.02)
(0.03-0.02)
(0.02-0.01)
(0.02-0.01)

(0.01-0.02)
PRINCIPAL POLLUTANTS
PIC, 1,3-butadiene

Formaldehyde
Hexavalent Chromium
Dioxin
PIC
Asbestos
Arsenic, formaldehyde
Hexavalent Chromium
Gasoline Vapors, Benzene
Perch loroethylene, Methylene
Chloride
Carbon tetrachloride
Vinyl chloride
Coke Oven Emissions, Benzene,
PIC
Cadmium, Vinyl Chloride
Dioxin
Gasoline Vapors, Formaldehyde
1,3-butadiene

1,3-butadiene
Arsenic
Vinyl chloride
Formaldehyde

Ethylene Oxide
Benzene
Perch I oroethy I ene

Chloroform
Chloroform

Ethylene Dibromide
1,3-butadiene .
Ethylene Oxide

Ethylene Dichloride,
Arsenic
Asbestos
Asbestos
Arsenic
Hexavalent Chromium
Methylene chloride
_
Chloroform
                                        3-25

-------
                                    TABLE 3-9

SUMMARY OF ESTIMATED CANCER CASES BASED ON MODELED AMBIENT CONCENTRATIONS,  BY
                          SOURCE  CATEGORY  (concluded)
ANNUAL (percent of
SOURCE CATEGORY CANCER CASES total)
38. Benzene Fugitives
39. Nitrite Elastomer Produc-
tion
40. ABS/SAN Production
41. Asbestos Fabrication
42. Benzene Storage
43. Other
Total
NOTE: Values in this
0.2 (0.01-0.01)
0.16 (<0.01)
0.13 (<0.01)
0.13 (<0.01)
0.1 (<0.01)
6-13 (0.4-0.5)
1,430-2,538
figure are not absolute pr
PRINCIPAL POLLUTANTS
Benzene
Acrylonitrile
Acrylom'trile
Asbestos
Benzene
Hexavalent Chromium, radon

'edictions of cancer
       occurrence and are intended to be used in a relative sense only.  The
       dose-response relationships and exposure assumptions have a
       conservative bias, but omissions due to uncharacterized pollutants
       (either directly emitted or secondarily formed) and emission sources,
       the long-range transport of pollutants, and the lack of knowledge of
       total risk from multi-pollutant exposures will  offset this bias to an
       unknown extent.

  Estimated  incidences  is  approximately  equally  divided between  point  and  area
  sources.
                                        3-26

-------
                                             CO
                                             o
                                             a  «
                                             .9  CD
                                             O >
                                             s? ®
                                             .2 Q_
                                             r^  w
                                             O  Q)
                                             O  CD
                                             C/D  O
                                             11
                                             £=  o
                                             c; '.£5
                                             O  Cd
                                             O Z
                                             CD -O
                                             >  CD
                                             CD
                                             CO
                                             CD
                                             •5
                                             D)
                                             U_
3-27

-------
sources are represented by the "motor vehicle" source category in
Table 3-9.  All other source categories, except secondary formaldehyde,
make up stationary sources.  It is sometimes difficult to distinguish
between point and area sources.  The primary distinguishing feature is
the number of sources in a source category.  "Area" sources are
generally considered to have too many individual sources to develop
source-specific data from which to estimate risk for each source.  Point
sources, in contrast, are "few" enough in number to be located
individually to allow source-specific data to be developed, from which
risks can be estimated for each individual source.  To estimate risks
from sources, EPA has developed model algorithms for area sources and
for point sources.
      For purposes of this study, source categories have been designated
as either an "area" source or a "point" source depending on whether the
cancer risks were estimated using an area source model algorithm or a
point source model algorithm.  Using this basis, the following
individual source categories are considered area sources:
      •  motor vehicles
      •  woodsmoke
      •  asbestos, demolition and renovation
      •  gasoline marketing (service stations only)
      •  coal and oil combustion (residential only)
      •  solvent use/degreasing
      •  drycleaning
      •  pesticide usage
      •  chlorinated drinking water
      •  paint stripping
                                  3-28

-------
All other sources (except secondary formaldehyde) are considered to be
point sources in this study.
      The estimated cancer incidence-from secondary formation of
formaldehyde for selected source categories is shown in Table 3-10.
This apportionment was based on the relative percent contributions
calculated for specific source categories in the 5-City study.  The
cancer incidence from secondary formaldehyde was distributed among the
major classifications (i.e., mobile vs. stationary and area vs. point),
as discussed below.  Approximately one-half of the cancer incidence from
the "other" source category was from point sources and one-half from
area sources.  The specific source categories in Table 3-9, however, do
not include the estimated cancer incidence from secondary formaldehyde
shown in Table 3-10.  Instead, a separate "source category" for
secondary formaldehyde is shown.
      Examining mobile versus stationary sources, approximately 58
percent of the estimated total annual incidence is estimated to occur
from motor vehicles (including cancer risk from secondary formaldehyde).
Stationary sources account for approximately 42 percent of the total.
Of the major stationary sources, two of the top six -- electroplating
and cooling towers — are related to hexavalent chromium, which accounts
for the entire estimated risk from these two source categories.  The
second largest stationary source category is TSDF, in which dioxin is
estimated in the study on TSDFs to contribute 65 percent of the total
estimated 140 annual cancer cases.4  In general,  while many stationary
source categories emit a number of different pollutants, the majority of
risk is attributable to a select few in each source category.
     4 Because of the great  uncertainty  associated  with the estimate of
dioxin emissions from TSDFs, this estimate could be substantially lower.
                                  3-29

-------
                                  TABLE 3-10

             DISTRIBUTION OF ESTIMATED CANCER CASES  FROM SECONDARY
                 FORMALDEHYDE FORMATION AMONG SOURCE CATEGORIES8
      SOURCE
      CATEGORY
ANNUAL CANCER
CASES IN U.S.
  PERCENT
CONTRIBUTION
Area Sources
      Motor Vehicles
      Solvent Use
      Gasoline Marketing

      Area Source Subtotals
     45
     38
     10

     93
    34.8%
    28.9%
     8.3%

    71.9%
Point Sources
      Petroleum refining
      Chemical Manufacturing

      Point Sources Subtotal
      8
      5

     13
     5.9%
     3.7%

     9.6%
                                     24
                             18.4%
Total Secondary Formaldehyde
    130
   100%
8 Distribution of secondary formaldehyde based on data from the 5-City
  study.
                                  3-30

-------
       Table 3-11 presents the results in Table 3-9 on an area versus
 point source basis.  Area sources are found to contribute approximately
 75 percent off the total  number of annual  cancer cases (including those
 from secondary formaldehyde)  with point sources contributing
 approximately 25 percent  of the total.   Of the area sources, the major
 source is mobile sources, contributing  78  percent of the total  annual
 incidence attributed to area  sources  (including the estimated 45 annual
 cancer cases attributed to mobile sources  from secondary formaldehyde,
 as shown  in Table 3-10).   For point sources,  the largest category is
 electroplating,  which accounts for almost  25  percent of the  total  point
 source-related annual  incidence.   Although the estimates in  Table 3-11
 add up to "100 percent of the risk,"  the reader is  reminded  that this
 study does  not include risk estimates from all  known sources.   The
 relative  contributions of the types of  sources,  therefore, could change
 as additional  data  on  other sources are obtained.
       Geographic  Variation
       As  has  been stated,  the primary purpose  of this  study  is  the
 estimation  of nationwide  cancer risks.  The various  studies  used to  meet
 this  goal illustrate the  variation in exposure  to different  pollutants
 and  in  the  resulting cancer risk that exists between geographic  areas on
 a  county-to-county  and  city-to-city basis  as well as on  an intra-city or
 intra-region  basis.  Table 3-12 presents ambient-measured concentration
 data  for  several selected pollutants and cities.  As can be  seen  in  this
 table, the  variation in ambient concentrations depends on the pollutant
 considered.   For example,  for the selected cities, ambient benzene
concentrations differ by less than a factor of 2.  Two of the pollutants
vary by factors of approximately 4 to  5.  For the other two selected
pollutants, ambient concentrations vary by factors of approximately  12
                                  3-31

-------
                                         TABLE 3-11

                      CONTRIBUTION OF AREA  VS.  POINT  SOURCES
                          TO  NATIONWIDE  ANNUAL  CANCER CASES
   SOURCE TYPE/
INDIVIDUAL SOURCE
    CATEGORY
ANNUAL
CANCER
CASES3
PERCENT  CONTRIBUTION TO..
 Nationwide  Source Type
    Total       Total
Area Sources

        Mobile Vehicles
        Woodsmoke
        Asbestos, demolition
        Gasoline Marketing
        Solvent Use/Degreasing
        Unspecified/Other
    Commercial  Sterilization/
         Hospital
        Pesticide Usage
        Drycleaning
        Chlorinated Drinking Water
    Coal  and Oil Combustion
         (residential onl>)
        Asbestos, renovation
        Paint  Stripping
        Secondary Formaldehyde
 1,115
    89
    81
    46
    29
    21
     3.5
     3
     3
     2

     2
     0.4
     0.3
    93
    56.2
     4.5
     4.1
     2.3
     1.5
     1.1

     0.2
     0.2
     0.2
     0.1

     0.1
     0.02
     0.01
     4.7
75.0
 6.0
 5.4
 3.1
 1.9
 1.4

 0.2
 0.2
 0.2
 0.1

 0.1
 0.03
 0.02
 6.2
Subtotal  Area Sources
Point Sources
 1,487
                                                                         75.0
                                                                                     100
Electroplating
TSDFs
Unspecified
Cooling Towers
Chemical Users and Producers
Iron and Steel
Coal and Oil Combustion
(non- residential)
Sewage Sludge Incinerators
Municipal Waste Combustors
Petroleum Refineries
Miscellaneous
POTWs
Manufacturing
Gasoline Marketing
Secondary Formaldehyde
ubtotal Point Sources
120
94
59
56
43
17

8
13
12
11
11
6
6
3
38
497
6.0
4.8
3.0
2.8
2.2
0.9

0.4
0.7
0.6
0.6
0.5
0.3
0.3
0.2
1.9
25.0
24.1
19.0
11.9
11.2
8.7
3.5

1.5
2.6
2.4
2.2
2.2
1.1
1.3
0.6
7.5
100
TOTAL - All  sources
 1,984
                                                                        100
  Based on raid-point of estimate from Table 3-9.
                                             3-32

-------
                                TABLE 3-12
             COMPARISON OF MEASURED AMBIENT  CONCENTRATIONS OF
                 SELECTED  POLLUTANTS IN SELECTED CITIES
POLLUTANT

Benzene
Chloroform
Ethylene dibromide
Methylene chloride
Perch loro«ithylene

A
8.2
1.0
0.06
6.1
1.5

B
7.9
6.2
0.05
3.7
3.6

c-
8.4
4.6
0.2
2.1
5.9
CITY
D
10.8
4.6
0.04
5.2
4.5

E
9.8
17.4
0.07
7.6
2.4

F
13.10
16.6
-
-
5.8

G
8.8
1.2
-
-
0.5
         NOTE: All numbers are in terms of
         SOURCE:  Ambient Air Quality Study, data worksheets.
Key; A = Bakersfield, CA
    B = Newark, NJ
    C = Philadelphia, PA
    D = Elizabeth, NJ
                               -E = Camden, NJ
                                F = Baltimore, HO
                                G = Baton Rouge, LA
for perch!oroethylene and 17 for chloroform.  The  degree of variation
presented  in  Table 3-12 depends upon the pollutants  and  cities selected.
Nevertheless,  the point is still the same regardless of  which pollutants
or cities  are selected -- ambient concentrations vary between cities.
      Ambient concentrations can also vary within  a  city or within a
specified  geographic locale (e.g., the South Coast Air Basin, the
Kanawha Valley).   Table 3-13 presents ambient-measured concentrations
for selected  pollutants at different locations in  Baltimore,  the South
Coast Air  Basin,  and the Kanawha Valley.  In general,  the variation in
ambient concentrations within each area is approximately the  same as the
variation  in  Table 3-12 for comparable pollutants.
      As might be expected,  the variations in ambient  concentrations for
pollutants can lead to variations in the number of cancer cases between
geographic areas  and in the estimates of cancer cases  per year per
million population.  Results from eight studies are  presented in Table
3-14.  As  seen in this table,  annual cancer cases  varied from a low of
0.03 per year  to  a high of 128 per year.  This reflects  a combination of
                                   3-33

-------
5?    g
                                                                  5    M

                                                                 I    £
                                                                             a
                                                                  -
                                                                       8.,-
                                                                       +-»    -M
                                                                                      .8
                                                                                      •8
                                                                                         -§
                                                                                 «-    -i-  c
                                                                                  3    4-»
                                                                                  O    0)
                                                                                 S    JB
                                                                                 _,    O •—
                                                                                       O -t->  E
                                                                     3-34

-------
                                                 TABLE 3-14

                                   VARIATION IN ANNUAL CANCER CASES  AND
                                CANCER RATES DUE TO EXPOSURE TO OUTDOOR
                                     AIR TOXICS BY  GEOGRAPHIC LOCALES
GEOGRAPHIC
LOCALE
Baltimore
Kanawha Valley
Philadelphia
Santa Clara
City A
City B
City C
City D
City E
Southeast
Chicago
South Coast
County A
B
C
D
E
F
G
H
I
J
K
L
M
N
0
P
Q
R
ESTIMATED ANNUAL
CANpER CASES
2.95-7.15
1.77
0.42
1.85
3.1
67.7
1.1
14.1
6.7
1.26

19-33
128.3
38.4
24.1
22.8
14.6
12.4
7.8
7.6
6.6
5.5
4.8
3-1
1.6
1.1
0.6
0.2
0.1
0,03
CANCER CASES PER YEAR
PER MILLION POPULATION
1.9-4.7
17.8
0.26
1.4
2.1
6.2
2.8
10.0
3,9
3.2

1.7-3.0
17.2
20.4
12.5
38.3
11.3
14.6
9.2
9.7
8.7
3.5
19.2
2.8
1.9
3.0
4.2
1.2
0.6
1.6
b STUDY
lEMP-Baltimore
lEMP-Kanawha Valley
lEMP-Philadelphia
lEHP-Santa Clara
5 City
5 City
5 City
5 City
5 City
Southeast Chicago

South Coast
35 -County3

















                       Selected counties.

                     " Derived by dividing estimated annual cancer cases by the population in
                       the geographic locale.
                                                     3-35
_

-------
different exposure levels and the size population exposed to those
levels.  Cancer rates varied from a low of 0.26 to a high of 38.3 cancer
cases per year per million population.  These are equivalent to areawide
lifetime individual risks of approximately 2 x 10"5 to 3 x 10"3 for the
exposed populations.  In general, a lower absolute number of cancer
cases corresponded to a lower cancer rate.  A notable exception is
County K from the 35-County study, where a "modest" number of estimated
annual cancer cases (4.8) had one of the highest cancer rates (19.2
cancer cases per year per million population).
      Variation in lifetime individual risk between and within cities
can also be examined.  Areawide lifetime individual risks for selected
cities were presented earlier in Tables 3-6 and 3-7.  The lifetime
individual risks among the cities/locales shown in these two tables,
however, were essentially on the same order of magnitude with one
another (approximately 10"4).  The areas  shown  in  Tables 3-6  and 3-7  are
urban areas, with the exception of the Kanawha Valley, which is
classified primarily as rural.  The Kanawha Valley, on the other hand,
is a fairly highly industrialized area.  Because of the large number of
factors that differed in deriving these risk estimates,  it is difficult,
if not impossible, to say why such a relatively narrow range is
observed.  The small range may point to a relatively consistent areawide
lifetime individual risk regardless of the urban area or industrialized
area in which one lives.
      Table 3-15 presents maximum individual  risks associated with
various cities or specific geographic locales.  The maximum individual
risks in Table 3-15, however, are not necessarily directly comparable to
each other, because they vary in manner in which they were estimated and
                                  3-36

-------
                                         TABLE 3-15

          VARIATION  IN  MAXIMUM LIFETIME INDIVIDUAL  RISK,  BY  LOCATION
MAXIMUM LIFETIME
LOCATION INDIVIDUAL
RISKS
South Coastb
Southeast Chicago0
Baltimore
Philadelphiad
Santa Clara
Kanawha Valley6
Belle
Charleston
Institute
Nitro
lo-3
1 x 10"3
r x io"4
7 x IO'6 to
3 x IO"4
3 x IO"4

3 x 10'3
6 x 10'3
8 x 10'3
8 x 10'6
COMMENT REFERENCE
MEIa to hexavalent
chromium
MEI, additive,
30 pollutants
MEI to benzene and
to chloroform
MEI, additive,
9 pollutants
MEI to emissions at
a traffic intersection

additive,
3 pollutants
additive,
4 pollutants
additive,
6 pollutants
additive,
9 pollutants
1
Z
3
4
5
6




NOTE:  Values have been adjusted to reflect unit  risk factors used in this study wherever possible.

a  MEI  = maximum exposed individual

   Maximum exposed individual  risk based on model-predicted exposures in the South Coast study
   rather than the monitored exposures in the.study.

c  8 x  10'  of additive lifetime risk was attributed to five pollutants from steel mills.  Southeast
   Chicago study, p. 43.

   MEI  risks were calculated for eight locations  in the city.  The range reflects the high and low
   MEI  risks estimated.

e  The  individual risks calculated in the IEMP-Kanawha Valley Study were for neighborhood sites with
   suspected highest exposures from point source  pollutants.
                                             3-37

-------
REFERENCES TO TABLE 3-15
1.  South Coast study, p. VI-6.
2.  Southeast Chicago study, p. x.
3.  IEMP - Baltimore study, Table V-8.  Based on exposure at the site of
    maximum average concentration.
4.  IEMP - Philadelphia study, p. VI-49.
5.  IEMP - Santa Clara study, p. 3-82,
6.  IEMP - Kanawha Valley study, pp. 4-73, 4-84, 4^95,  and 4-99.
                                  3-38

-------
 the number of pollutants included in the risk estimate.  For the
 locations shown, maximum individual  risks are in the neighborhood of
 10"3 and  10~4.   The results from the  lEMP-Philadelphia study illustrate
 that the MEI level can vary within subareas of a city by a factor of 40
 (7 x 10'6 to 3 x  10'4),  while the lEMP-Kanawha Valley study's results
 show a 1,000-fold difference in maximum individual  risk for one of the
 locales in  the Kanawha Valley versus the other three locales (10~6 vs.
 10"3).  Table 3-16 shows additional details on the variation in maximum
 individual  risk within the Kanawha Valley.
       The pollutants  that  are the most  important contributors  to annual
 incidence and  the source categories  that emit those  pollutants and to
 the  individual  risk  in  a geographic  area can  also vary  from one area to
 another.  Table  3-17  illustrates some of the  annual  incidence  variation
 for  five  pollutants and the  source categories that emit two of these
 pollutants  across six selected  cities.   For example,  among  the
 individual  pollutants,  benzene  is  estimated to contribute a relatively
 consistent  percentage,  between  approximately  5 and'10 percent  of  total
 cancer  cases for  the  six cities.   In contrast, 1,3-butadiene is seen  in
 Table 3-17  to contribute a much  wider range,  between 6  and  48,  percent
 of total cancer cases across  the  five cities.  Among source categories,
 road vehicles are  consistently  a major contributor,to annual incidence
 attributed  to benzene in each city, contributing  between 45 and 81
 percent of  the total cancer cases  attributed  to benzene, and are the
most important source category of  benzene-related incidence in five of
the six cities.   In contrast, the  relative contribution of  "iron and
steel" to benzene  incidence varies dramatically between cities, ranging
from 0 percent in four of the six cities to 25 percent in City D and
over 50 percent in Southeast Chicago.  Along the same lines, 100 percent
                                  3-39

-------
                                    TABLE 3-16

     ESTIMATES  OF MAXIMUM LIFETIME  INDIVIDUAL CANCER RISKS  IN
    NEIGHBORHOODS SURROUNDING  FACILITIES IN  THE KANAWHA VALLEY
TYPE OF
INDIVIDUAL RISK
Highest
(number of people
exposed)
Range of Maximum
Risks in
Remaining
Neighborhoods
Average Risk
Population in
locale
(number of people)


Belle
2.8x10"3
(600)
7x1 0"5. to
4x1 O'4


2.2x10"4
15,530
LOCALE
Charleston/
South Charleston Institute Nitro
6x10"3 8x10"3 8x10"6
(2,700) (1,300) (1,453)
2x10~4 to 3x10"3 2x10"4 to 2x10"3 9x10"7 to 6x10"6


2.8x10"4 1.U10"3 3.2x10"6
51,750 22,390 9,990
NOTE 1:  Values have been adjusted to reflect the unit risk factors used in this study.

NOTE 2:  These risk estimates are based on pollutants from point sources; risk from area
        source pollutants are not included.

SOURCE:  lEMP-Kanawha Valley Study, pp. 4-73, 4-77, 4-84, 4-88, 4-95, 4-97, 4-102, and 4-106.
                                         3-40

-------
                                   TABLE 3-17

       CITY-TO-CITY  VARIATION IN  RELATIVE CONTRIBUTION OF  SELECTED
               POLLUTANTS TO  TOTAL ANNUAL CANCER  INCIDENCE
ITEM
Pollutant
(Source Category)
Benzene
(Road vehicles
(Gas marketing
(Iron & Steel/Steel
Mills
1,3-Butadiene
(Road vehicles
(Chemical Mfg.
Formaldehyde
Chromium
Methylene chloride
CITY3

A
8.8%
73%
7%
0

19.3%
100%
0
34.5%
5.2%
2.3%

B C
10.1% 4.8%
81% 67%
3.6% 0
0 0

23.9% 48.4%
100% 17%
0 83%
23.0% 24%
21.3% 16%
3.9% 1.6%

D
9.6%
56%
1%
24%

16%
100%
0
6.9%
14%
2.2%

E
7.0%
63%
3%
0

13%
100%
0
18.9%
16%
3%

SOUTHEAST
CHICAGO
4.8%
45%)
2.3%)
52%)

6.4%
100%)
0)
17%
16%
0.2%
5-City study.  Derived from data worksheets.

Southeast Chicago study, p. 33.  Relative contributions have been adjusted
based on unit risk factors listed in Table 2-6  in this report.
                                     3-41

-------
of the estimated 1,3-butadiene related cancer cases are due to emissions
from road vehicles in five of the six cities.  For City C, however, over
80 percent of the 1,3-butadiene related cancer cases are attributed to
chemical manufacturing and less than 20 percent to road vehicles.
      Table 3-18 illustrates the variation in maximum individual risk
for individual pollutants.  The data are taken from the lEMP-Baltimore
study, and show the exposed population to each pollutant as well.  The
range in "maximum" individual risks is from 10"6 to  10"4.  In Table 3-19,
the individual pollutant contributors to the highest estimated
individual risk grid cell in the Southeast Chicago area are presented.
As seen in Table 3-19, coke oven emissions contribute over 77 percent of
the total individual risk.
      Table 3-20 illustrates the areawide lifetime individual risk
associated with individual pollutants based upon data from the  IEMP-
Philadelphia  study.  Both monitored and modeled results are presented.
The range of  areawide lifetime individual risks for individual
pollutants is from 10"6 to 10"5.
      Variation in lifetime individual risk across source categories  is
illustrated in Tables 3-21 through 24.  Tables 3-21 and 22 report
maximum individual risks  for two cities.  Specific sources show maximum
individual risks in the range of 10"7 to  10"4.
      Tables  3-23 and 24  show areawide lifetime individual risks for
specific sources.  Table  3-23 shows area and point sources in Santa
Clara, while  Table 3-24 shows area and point sources for  the Kanawha
Valley.  In the Santa Clara study, area and point  sources are found to
be the major  contributor  to total areawide lifetime individual  risk.  On
the other hand, point sources are found in the Kanawha  Valley to be the
major contributor to total areawide lifetime individual risks.   Both
                                  3-42

-------
                                  TABLE 3-18

             MAXIMUM LIFETIME INDIVIDUAL CANCER RISKS IN BALTIMORE
                            BY  INDIVIDUAL POLLUTANT8
MAXIMUM LIFETIME
POLLUTANT INDIVIDUAL RISK
._.
Benzene
Tri chl oroethyl ene
Perch! oroethylene
Ethyl ene di chloride
Chl orof orm
Carbon tetrachloride
1,2-dichloropropane
Chromium (hexavalent)c
Cadmium6

l.OxlO'4
6.7xlCT6
5.4xlO'6
6.8xlO"5
l.lxlO'4
Z.lxlO'5
3.6xlO'5
0 to 3.6xlO'4
0 to 3.6xlO'6
EXPOSED
POPULATION15
i
48,771
48,771
14,270
12,880
23,997
23,997
16,848
490,690
118,411
NOTE:  Values have been adjusted to reflect the unit risk factors used
       in this study.

SOURCE: ' lEMP-Baltimore Study, Tables V-8 and V-14..

a Except for  cadmium,  individual  risks were  calculated  using  the maximum
  observed  ambient  concentration  measured  across all monitoring sites.
  Measured  cadmium  concentrations were below detection  limits.  For  screening
  purposes, the Baltimore  study calculated risks assuming  a range  in  ambient
  concentrations  from  zero to  the upper  end  of the detection  limit (about
  0.002
b The  exposed  population  is  in  the  grid cell  at the monitoring  site  of maximum
  concentration.

c Range  indicates  possible ambient  levels of  hexavalent chromium,  from 0
  percent  to 100 percent.

d Range  indicates  possible ambient  levels from 0.00 ng/mz to the upper end of
  the  detection limit  (about 0.002  /t
                                        3-43

-------
                            TABLE 3-19

         RELATIVE  CONTRIBUTION OF  INDIVIDUAL  POLLUTANTS TO
            MAXIMUM LIFETIME INDIVIDUAL RISK OF CANCER
                   IN THE  SOUTHEAST CHICAGO AREA
POLLUTANT
   MAXIMUM
INDIVIDUAL RISK
% CONTRIBUTION
Coke Oven Emissions3
Benzene
Chromium
Formaldehyde
POM
Arsenic
Cadmium
Carbon tetrachloride
7xlO'4
6xlO'5
5xlO'5
4xlO'5
3xlO"5
2xlO'5
IxlO'5
IxlO'5
77%
7
6
4
3
2
1
1
Others
    
-------
                             TABLE 3-20

            AREAWIDE LIFETIME INDIVIDUAL RISKS OF CANCER:
          MONITORED VS. MODELED AMBIENT AIR CONCENTRATIONS
                           IN PHILADELPHIA

POLLUTANT

Chloroform
Ethylene dichloride
Carbon tetrachloride
Benzene
Trichloroethylene
1 , 2-di chl oropropane
Perch! oroethyl ene
CUMULATIVE
AREAWIDE LIFETIME

MONITORED
6.9xlO'6
L.OxlO'5
2.7xlO"5
S.OxlO"5
2.7xlO"6
2.2xlO'5
2.8xlO'6
1.2xlO"4
INDIVIDUAL RISK

MODELED
4.6X10"6
l.OxlO'5
r.5xl(T6
1.9xlO'5
1.7xlO'6
9.0xlO'6
2.1xlO'6
4.8xlO"5
NOTE:  Values have been adjusted to reflect the unit risk factors
       used in this study.

SOURCE:  lEMP-Philadelphia, p. V-27.
                                 3-45

-------
                               TABLE 3-21

                  ESTIMATES  OF  MULTI-POLLUTANT  LIFETIME
               CANCER RISKS  TO  THE  MOST  EXPOSED INDIVIDUAL
                   TO VARIOUS SOURCES IN PHILADELPHIA
MEI
LOCATION
Northeast Water Control Plant
Refinery B
Chemical Manufacturer
Plastic Cabinet Mfr.
Pharmaceutical Mfr.
Garment Mfr.
Refinery A
Industrial Dry Cleaner
MAXIMUM
INDIVIDUAL
RISK
6.2 x 10'5
1.4 x 10'5
2.3 x 10"4
8.2 x 10"7
3.2 x 10'4
1.7 x 10'5
3.1 x 10'5
2.8 x 10"5
COMMENT
8 pollutants
3 pollutants
3 pollutants
1 pollutant
3 pollutants
1 pollutant
3 pollutants
1 pollutant
NOTE:  Where possible and as needed, the values have been adjusted to
       reflect the unit risk factors used in this study.

SOURCE:  lEMP-Philadelphia Study, p. VI-49.
                                  3-46

-------
                     TABLE 3-22

       ESTIMATED  CANCER RISK TO MAXIMUM  EXPOSED
      INDIVIDUALS TO ORGANIC GASES  IN SANTA CLARA
                 FOR SELECTED SOURCES
SOURCE
TYPE
Traffic Intersections
Hospitals
Pharmaceutical Manufacturer
Computer Equipment Mfr.
Industrial Facility
Fuel Pipeline
Drycleaners
Sewage Treatment Plants
Gasoline Station Pump
Groundwater Aeration
MAXIMUM
INDIVIDUAL RISK
3 x 10'4
2 x 10"4
1 x 10'4
4 x 10'5
3 x 10'5
2 x 10'5
1 x 1CT5
5 x 10'6
4 x 1(T6
2 x 1(T7
SOURCE:  lEMP-Santa Clara study, p. 3-82.
                           3-47

-------
                   TABLE 3-23

  AREAWIDE LIFETIME INDIVIDUAL RISK OF CANCER
    FROM LIFETIME EXPOSURE TO ORGANIC GASES
                IN SANTA CLARA
SOURCE
CATEGORY
Burning of Waste Material
Combustion of Fuels
Degreasers
Drycleaners
Fuels Distribution .
Industrial Solvents Coating
Mobile Sources
Off-Highway Mobile Sources
Other Chem./Indust.
Other Organics Evaporation
Pesticides Usage
Area Source Total
25 Point Sources Total
Carbon Tetrachloride
AREAWIDE
INDIVIDUAL
RISK
4 x 10'8
1 x 10'6
8 x 10'7
8 x 10'7
1 x 10"6
3 x 1(T6
1 x 1(T5
3 x 10'7
2 x 10'7
4 x 10'7
8 x 1CT7
2 x 10"5
6 x 10"6
1 x 10'5
TOTAL
4 x 10
                                       -5
SOURCE:  lEMP-Santa Clara study, p. 3-80.
                        3-48

-------
                                  TABLE 3-24

           ESTIMATES OF AREAWIDE LIFETIME INDIVIDUAL RISKS OF CANCER
              ACROSS AREA AND POINT SOURCES IN THE KANAWHA VALLEY
SOURCE TYPE
Area
Gasoline Marketing
Heating
Road Vehicles
Solvent Use
Waste Oil Burning
Area Subtotal
Point
TOTAL
Population in
Locale
LOCALE
Belle

l.lxlO"6
l.SxlO'5
S.lxlO'5
7.7xlO"6
6.8xlO'7
5.3xlO'5
2.2xlO~4
S.OxlO'4
15,530
Charleston/
South Charleston

1.4xlO"6
1.3xlO"5
5.1xlO'5
l.SxlO'5
1.2xlO'6
S.OxlO'5
2.8xlO"4
3.6xlO'4
51,750
Institute

7.9xlO"7
l.SxlO'5
2.6xlO'5
6.3xlO"6
6.5xlO'7
4.7xlO"5
l.lxlO'3
l.lxlO'3
22,390
Nitro

3.4xlO'7
4.2xlO'6
9.8xlO'6
l.Sxl'O'6
2.4xlO'7
1.6xlO"5
S.OxlO'6
1.9xlO"5
9,990
SOURCE:  lEMP-Kanawha Valley Study.   Tables 32,  40,  45,  52,  and 54.   Values
         for area sources could not  be adjusted  using the unit risk factors
         in this study.  However,  the net effect is  expected to be small.
         Values for point sources  have been adjusted using the unit risk
         factors in this study.
                                        3-49

-------
studies show similar risks from area sources (10~5 range).  The
difference in relative contributions is due to the presence or absence
of point sources.  The Kanawha Valley is a relatively heavily
industrialized area, with significant point sources,  whereas the Santa
Clara area is much less industrialized.  Thus, the relative
contributions of point and area sources to total  areawide lifetime
individual risks is consistent with the character of the two study
areas.
      The areawide lifetime individual  risk data  from area sources in
the Kanawha Valley Study show relatively consistent percentage
contribution among the same source category between locales.  Among
area-type sources, "mobile sources" as a source category is found to be
the largest contributor to areawide lifetime risks in both the Santa
Clara and Kanawha Valley studies.
Comparison with the Results from the 1985 Six-Month Study
      The results of the present study are compared with the results of
the 1985 Six-Month Study.  This is done in two ways.   First, a
comparison of estimated nationwide cancer cases is made to examine the
magnitude of the problem.  Second, a comparison of the nature of the
problem is presented by examining the pollutants  and the source
categories that appear to be the greatest contributors to risk.
      Magnitude of the Problem
      Tables 3-25 and 3-26 compare the cancer rates (i.e., annual cancer
cases per million population) arid annual cancer cases, respectively,
estimated for three studies presented in the 1985 Six-Month Study with
the point (or range) estimates of this study.  As seen in these two
tables, the present study's low end estimated total cancer cases per
year per million population and the nationwide number of annual cancer
                                  3-50

-------
                             TABLE  3-25

COMPARISON OF ANNUAL  CANCER  CASES  PER MILLION  POPULATION
                        WITH 1985  SIX-MONTH STUDY
POLLUTANT
Arsenic
Benzene
1,3-butadiene
Cadmium
Carbon tetrachloride
Chloroform
Chromium (hexavalent)
Dioxin
Ethylene Oxide
Ethylene dibromide
Ethylene dichloride
Formaldehyde
Gasoline vapors
Perch loroethylene
Trichloroethylene
Vinyl chloride
Vinyl idene chloride
Other
NESHAP
0,02
0.14
<0.001
0.04
0.06
<0.01
1.43
--
0.21
0.12
<0.01
0.01
N/A
0.01
0.04
0.05
<0.01
0.11
1985 SIX-MONTH
35 County
0.02
0.39
<0.001
0.02
0.004
0.002
0.29
--
N/A
0.02
0.03
0.21
0.15
0.14
0.15
0.17
N/A
0.35
STUDY
Ambient Air
Quality
0.26
1.02
--
0.06
0.19
0.07
1.05
--
N/A
N/A
0.05
0.83
N/A
0.10
0.08
--
0.27
0.01
THIS STUDY
0.28
0.75
1.11
0.04
0.17
0.48
0.61-1.1
0.008-0.52
0.03
0.28
0.19
0.52
0.08-0.32
0.03
0.03
0.10
0.04
0.13
 Risk Estimates from Other EPA Efforts
                                    a
Asbestos
Gasoline Marketing
PIC
Radionuclides
0.50
0.20
2.65
0.07
0.50
--
2.60
0.07
0.50
0.20
2.68
0.07
0.37
•• --
1.83-4.67
0.02
        TOTAL
                         5.6
                                      4.9
                                                   7.4
                                                              7.2 -  11.3
 NOTE:  Values  in this figure are  not absolute predictions of cancer occurrence
        and are intended to be used  in a relative sense only.  The dose-response
        relationships and exposure assumptions have a conservative bias,  but
        omissions due to uncharacterized pollutants (either directly emitted
        or secondarily formed) and emission sources, the  long-range transport
        of pollutants, and the lack  of knowledge of total risk from multi-
        pollutant exposures will offset this bias to an unknown extent.

 a Except for PIC in the 35-County study, these estimates of cancer incidences
   were not part of the individual results of the NESHAP, 35-County, and
   Ambient Air  Quality studies. The 1985 Six-Month Study included these
   estimates for these pollutants  to provide for a more complete accounting of
   information  available to the 1985 Six-Month Study.

   Includes radon.
                                   3-51

-------
                            TABLE  3-26

          COMPARISON OF ANNUAL CANCER  CASES  WITH
                     1985  SIX-MONTH STUDY
                                1985 SIX-MONTH STUDY
                                                             THIS STUDY
POLLUTANT
Acrylonitrile
Arsenic
Benzene
1,3 butadiene
Cadmium
Carbon tetrachloride
Chloroform
Chromium (hexavalent)
Coke Oven Emissions
Dioxin
Ethylene Oxide
Ethylene dibromide
Ethylene dichloride
Formaldehyde
Gasoline vapors
Hethylene chloride
Perch loroethylene
Trichloroethylene
Vinyl Chloride
Vinyl idene chloride
Other
NESHAP
0.42
4.7
32.3"
0.01
8.5
14
0.27
330
8.6
--
47.8
26.7
0.9
' 1.6
--
1
2.9
9.7
11.7
0.04
2.9
35 County
4.2
1.1
18.5
0.01
1.1
0.2
0.1
13.4
2.4
--
--
1.0
1.5
10
6.8
--
6.7
6.8
8.2
--
0
Ambient Air
Qua I i ty
..
60
234
--
14.6
43
17
242
--
--
--
--
11
191.3
--
7.4
22
18
--
62
1

13
68
181
266
10
41
115
147-265
7
2-125
6
68
45
124
19-76
5
6
7
25
10
30
 Subtotals
                         504
                                      207
 Risk Estimates from Other EPA Efforts
        TOTAL
                        1291
                                      234
                                                   1539
1,195-1,493
Asbestos
Gasoline Marketing
PIC
Radionuc I ides/Radon
115
46
610
16
23.7
--
125.1
3.3
115
46
615.4
16
88
(see gas vapor)
438-1120
5
                                                   1716
1,726-2,706
NOTE:  Values  in this figure are not  absolute predictions of  cancer occurrence
       and are intended to be used in a  relative sense only.   The dose-response
       relationships and exposure assumptions have a conservative bias, but
       omissions due to uncharacterized  pollutants (either directly emitted or
       secondarily formed) and emission  sources, the long-range  transport of
       pollutants, and the lack of knowledge of total risk from  multi-pollutant
       exposures will offset this bias to an unknown extent.

 a Except for  PIC in 35-County study, these estimates of concern incidences were
   not part of the individual results of the NESHAP, 35-County,  and Ambient Air
   Quality studies.  The 1985 Six-Month  study included these  estimates for these
   pollutants  to provide for a more complete accounting of information available
   to the 1985 Six-Month study.
                                      3-52

-------
cases are essentially the same as the original  Ambient Air Quality study
in 1985 Six-Month Study, which was larger than  the other two studies in
the 1985 Six-Month Study.  The upper end of the estimated cancer rate
and annual cancer cases of the present study are approximately 1.5 to 2
times larger than the NESHAP or Ambient Air Quality studies in the 1985
Six-Month Study.  Using a 1986 U.S. population  of 240 million, the
present study estimates up to approximately 500 to 900 more cancer cases
per year nationwide than either the NESHAP or Ambient Air Quality study
results in the 1985 Six-Month Study.
      There are several factors that account or may account for this
apparent increase in estimated risk.  One factor is that this study
includes more pollutants for which risks have been estimated than were
included in the 1985 Six-Month Study/  Most of these pollutants are the
result of the Sewage Sludge Incinerator, Hazardous Waste Combustion, and
TSDF studies being available for inclusion..  On an individual pollutant
basis, the potentially most important addition  from TSDFs is dioxin, for
which up to 92 annual cancer cases were estimated based on data in the
TSDF study.  As shown in Table 3-26 by the large range in risk (2 to 125
annual cancer cases), there is substantial uncertainty associated with
the risk estimate for dioxin in this study.
      A second factor that accounts for an increase in the estimated
cancer risk is the changes, some of .which are significant, that have
occurred to unit risk factors.  Table 3-27 compares the unit risk
factors used in the 1985 Six-Month Study with those used in this study
for those pollutants for which the unit risk factor has changed.  As
seen in this table, the unit risk factors have changed in a few
instances by relatively small amounts (±25 percent).  In some instances,
                                  3-53

-------
           TABLE 3-27



COMPARISON OF UNIT RISK FACTORS
POLLUTANT
Acryl amide
Benzene
BaP
Beryl 1 i urn
1,3-Butadiene
Cadmium
Chloroform
Epichlorohydrin
Ethyl ene di bromide
Ethyl ene oxide
Formaldehyde
Gasoline vapors
Methyl chloride
Methyl ene chloride
Nickel (subsulfide)
Perchl oroethyl ene
Propylene oxide
Styrene
Tri chl oroethyl ene
Vinyl chloride
Vinyl idene chloride
MAY 1985
1.7xlO'5
6.9xlO'6
3.3xlO'3
4.0xlO"4
4.6X10'7
2.3xlO"3
l.OxlO'5
2.2xlO'7
S.lxlO'4
3.6xlO"4
6.1xlO"6
7.5xlO'7
1.4xlO'7
l.SxlO'7
3.3xlO'4
1.7xlO"6
1.2xlO"4
2.9xlO'7
4.1xlO"6
2.6xlO"6
4.2xlO'5
JUNE 1988
l.lxlO'3
8.3xlO"6
1.7xlO"3
2.4xlO'3
2.8xlO"4
l.SxlO'3
2.3xlO'5
1.2xlO'6
2.2xlO"4
-A
1.0x10 4
l.SxlO"5
6.6xlO"7
3.6xlO'6
4.7xlO'7
4.8xlO'4
5.8xlO'7 -
3.7xlO'6
5.7xlO"7
1.7xlO'6
4.1xlO'6
5.0x10'^
% CHANGE
+ 6400
+ 20
48
+ 500
+ 60770
22
+ 130
+ 445
57
72
+ 113
12
+ 2,470
+ 161
+ 45
66
97
+ 97
59
+ 58
+ 19
                 3-54

-------
the change has been large (over 100 percent) and, in the case of 1,3-
butadiene, the change has been over a 60,000 percent increase.
      A third factor that accounts for the apparent increase in
estimated cancer risk is a more complete accounting of sources that
contribute to cancer risk.  As noted above, a potentially significant
                                                                       »
new source category is TSDFs.  Another important source category is
electroplating.
      The more extensive body of information available to this study has
helped provide for a more complete accounting of source categories and
pollutants.  The apparent increase in estimated cancer risk, therefore,
should not necessarily be viewed as a problem that has become worse.
Rather, the estimates in this study, which are based on new and more
complete information, simply suggest that the problem may be larger than
previously thought.
      Nature of the Problem
      The nature of the air toxics problem can be described in several
ways:  which pollutants contribute the most to the cancer risk; which
sources contribute the most to cancer risk; and how does cancer risk
vary from one geographic region to another.  Since this study found
geographic variations to be of a very similar nature as those reported
in the 1985 Six-Month Study, only the first two aspects of the problem
will be compared.
      Individual Pollutants.  For the most part, the same pollutants
found to contribute the largest percentages to total annual cancer risk
in the 1985 Six-Month Study are also found to be among the larger
contributors in the present study.  These compounds include hexavalent
chromium, PIC, asbestos, benzene, carbon tetrachloride, ethylene .
dibromide, arsenic, and vinyl chloride.
                                  3-55

-------
      As seen earlier in Table 3-25,  eight of the  individual  cancer
rates (i.e., annual cancer cases per year per million population)
calculated for pollutants under the present study  fall  within the  range
of cancer rates created by the three studies in the 1985 Six-Month
Study.  The cancer rates for two pollutants (gasoline vapor and PIC)
bound their respective estimates in the 1985 Six-Month Study.  This
indicates that the magnitudes of incidence for these ten pollutants have
been estimated to be approximately the same.  For formaldehyde, the
cancer rate calculated in this study is lower than that in the Ambient
Air Quality study found in the 1985 Six-Month study, but higher than the
other two analyses in the 1985 Six-Month Study.  The decrease most
likely reflects better data and measurement techniques available than
were used in the original Ambient Air Quality study.
      For the other  10 pollutants identified in Table 3-25, the cancer
rates calculated in  this study fall outside the range created  in the
1985 Six-Month Study.  Of these pollutants, four --  ethylene oxide,
trichloroethylene, asbestos,  and radionuclides  --  show a decrease  in the
estimated cancer rate.   For ethylene oxide  and  trichloroethylene,  most
of this  decrease can probably be attributed to  the  change  in the unit
risk factor.  For  asbestos, the change reflects better emission factors.
For radionuclides, a new risk analysis was  conducted using updated
information on the number of  facilities,  radionuclide  emissions to the
air,  and control technologies.  The  net  effect  of  the  updated
information was  a  decrease  in the  estimated risk for radionuclide
exposure.
      Six pollutants (including dioxin,  which  was  not  included in  the
1985  Six-Month Study)  show  an increase  in estimated risk.   For arsenic,
a modest increase  in annual  cancer cases is estimated  (from  60 to  68 per
                                   3-56

-------
year), which is apparently due to higher measured ambient
concentrations.
      For ethylene dibromide (EDB) the estimated cancer rate has
increased in spite of a decrease in the unit risk factor.  This has
likely occurred due to the absence of a risk estimate for EDB in the
1985 Six-Month Study from the Ambient Air Quality study.  As noted in
the present study, modeled estimates appear to underestimate actual
ambient concentrations.  The present study based the risk estimate on
ambient-measured concentrations.  Thus, the net effect is an increase in
the estimated cancer rate for EDB, with an increase in estimated cancer
cases from 27 to 68 per year nationwide.
      For both ethylene dichloride and chloroform, the updated Ambient
Air Quality study's estimates of cancer risk were selected for the risk
estimate.  The increase in the estimated cancer cases from chloroform
can be attributed in part to an increase in its unit risk factor.  For
both pollutants, the increase may be simply attributable to a more
recent and larger data set that shows higher ambient concentrations than
before.
      The most dramatic increase is associated with 1,3-butadiene.  This
has occurred for two reasons.  One reason is the increase in the unit
risk factor, from 4.6 x 10"7  to  2.8 x  10"4,  an increase of over 600
times.  The second reason is that ambient-measured concentrations of
1,3-butadiene were not a part of the 1985 Six-Month Study and the major
source of ambient 1,3-butadiene -- motor vehicles -- were not included
in the other two studies in the 1985 Six-Month Study.  These two factors
combined to increase the estimated nationwide cancer risk due to 1,3-
butadiene from 0.01 cancer cases per year to almost 270 per year.
                                  3-57

-------
      Source Categories.  In the 1985 Six-Month Study,  area and point
sources were found each to account for approximately one-half of the
aggregate incidence in both the NESHAP and the 35-County studies.  When
PIC was included (by using BaP as a surrogate), areas sources were found
to be dominant, -accounting for over 75 percent of the incidence in both
the NESHAP and the 35-County studies.  This result was noted as being
consistent with the fact that PIC was estimated to account for a large
portion of aggregate incidence, and that nearly all BaP emissions
appear to come from area sources (principally motor vehicles and fuel
combustors in small heating units).
      Earlier in this chapter, Table 3-11 summarized the estimated
contributions of individual source categories to total cancer risk by
area vs. point source.  Area sources were estimated to contribute
approximately 75 percent of the total nationwide annual incidence and
point sources, approximately 25 percent.  The two studies, thus, show
essentially identical estimates of the relative contribution of  area vs.
point sources in spite-of some significantly important pollutants and
source categories  included in the current study that were not included
in the 1985 Six-Month Study.
      Table 3-28 presents the results of the Southeast Chicago study in
terms of area vs.  point and mobile vs. stationary  sources.   In that
study, point sources are estimated to contribute approximately 48
percent of the  total estimated annual incidence in  the Southeast Chicago
area, and area  sources  approximately 30 percent.   (Approximately 20
percent was attributed  to  background pollutants, the  sources of  which
were not identified.)   These  relative contributions of area  vs.  point
sources are very different  from  the  nationwide  split  estimated.  The
                                   3-58

-------
                                  TABLE 3-28

                  CONTRIBUTION OF  SOURCES  TO  ESTIMATED  ANNUAL
              CANCER CASES  AND AREAWIDE  LIFETIME  INDIVIDUAL  RISKS
                              IN SOUTHEAST CHICAGO
SOURCE TYPE/
CATEGORY
Point
Steel Mills
Chrome Platers
Other Industrial Sources
Sewage Treatment Plants
Total Point
Area
Home Heating
Consumer Sources
Mobile Sources
Waste Handling
Total Area
Background Pollutants
Mobile
Stationary
Background
ANNUAL
CANCER CASES3

0.41
0.185
0.016
0.001
0.612

0.127
0.05
0.22
0.001
0.398
0.26
0.22
0.79
0.26
AREAWIDE
LIFETIME
INDIVIDUAL RISK

7.3 x 10'5
3.3 x 10"5
2,9 x 10'6
1.8 x 10'7
1.1 x 10"4

2.3 x 10'5
8.9 x 10'6
3,9 x 10'5
1.8 x 10'7
7.1 x 10'5
4.6 x 10'5
3,9 x 10'5
1.4 x 10'4
4.6 x 10'5
PERCENT OF
TOTAL

32
15
1
0.1
48

10
4
17
0.1
31
20
18
62
20
a Southeast Chicago study,  p.  33.   Values  were adjusted to  the unit risk
  factors used in this study.

b Calculated by multiplying annual  cancer  cases by 70 and dividing by
  population of study area  (i.e,,  393,000).
                                        3-59

-------
larger share attributed to point sources in the Southeast Chicago area
is likely due to locally high risk from steel  mills.
      Mobile sources are also seen to be a relatively lower contributor
to total annual incidence versus stationary sources in. the Southeast
Chicago area.  This again is likely due to the locally-high risk from
steel mills.  In addition, risk from heating appears to be higher than
the nationwide estimate, thus further reducing the percent of total
annual incidence attributed to mobile sources.
      The total areawide lifetime individual risks for the Southeast
                                                   i
Chicago study are similar to those reported earlier in this chapter for
the Santa Clara study (Table 3-23) and the Kanawha Valley study (Table
3-24).  Excluding consideration of "background pollutants," areawide
lifetime individual risks from area sources are again in the 10"5  range,
with mobile sources being the major contributor to areawide lifetime
individual risk followed closely by home heating.  Point source
contribution to areawide lifetime individual risk in Southeast Chicago
is higher than the area source contribution, as was seen in the Kanawha
Valley study.  This seems consistent with the relative nature of the
study areas (Southeast Chicago has significant point source contribution
from coke ovens.)
                                  3-60

-------
                      4.0  SUMMARY AND CONCLUSIONS
      In this chapter, the results of this study are summarized and
conclusions are drawn with regard to the magnitude and nature of the
cancer problem associated with outdoor exposures to air toxics in the
United States.  These results are also compared to those contained in
the 1985 Six-Month Study.
      As has been discussed throughout the report, the results of this
study are subject to various limitations and uncertainties.  The
numerical estimates presented in this report, therefore, should be
viewed as rough indications of the magnitude of potential cancer risk
caused by a limited group of pollutants found in the ambient air.  Many
of the absolute values for individual pollutants are almost certainly
inaccurate.  The best use of these estimates is in describing the broad
nature of ;the cancer risk posed by these toxic air pollutants and by
making relative comparisons of risks between pollutants and source
categories.
Magnitude of the Cancer Risk
      Annual Cancer Incidence
      Based on the pollutants and source categories examined, nationwide
annual cancer incidence is estimated to be between 1,700 and 2,700
cancer cases per year (see Table 3-1).  This is equivalent to between
7,2 and 11.3 cancer cases per year per million population (1986
population of 240 million).  Approximately one-third of this cancer risk
                                   4-1

-------
was attributed to exposure to products of incomplete combustion (PIC).
Of all the cancer risks estimated in this study,  the greatest degree of
uncertainty is mostly likely associated with the cancer risk estimate
for PIC.
      A range of annual cancer incidence is reported as the result of
uncertainties associated with primarily four individual pollutants that
also are estimated to be among the largest individual contributors to
cancer risk.  These four pollutants are PIC, dioxin, gasoline vapors,
and hexavalent chromium.  The uncertainties identified are associated
primarily with:  (1) the inability at this time to select a single unit
risk factor from a range of unit risk factors for diesel particulates,
which are included with PIC; (2) the sampling and extrapolation
methodologies for dioxin; (3) the identification of the cancer-causing
portion of gasoline vapors; and  (4) the portion of total ambient
chromium that is hexavalent.  Although point estimates were made for
most pollutants, the lack of a range does not mean there is no
uncertainty associated with the  absolute magnitude of the cancer risk
estimate.
      The 1985 Six-Month Study presented three separate analyses that
showed  a range of cancer rates from approximately 5 to 7.4 cancer cases
per year per million population.  The results of the current  study
estimated a cancer rate of  between 7.2 and  11.3 cancer cases  per year
per million population  (see Table 3-25).  Using a total 1986  U.S.
population  of 240 million,  the results of this study show approximately
500 to  900  more  cancer  cases per year  (comparing lower and upper
ranges).  This  "increase" does not necessarily indicate a growing
problem, but  is  more likely the  result of analysis  of more air toxic
pollutants  than  were considered  in the 1985 Six-Month  Study  and,  in  some
                                   4-2

-------
instances, a better accounting of sources (e.g., sources that emit 1,3-
butadiene).  Even though this study has a broader data base to draw upon
than was available to the 1985 Six-Month Study, it is recognized that
cancer risk estimates are being made for a only a portion of total
ambient air pollutants and for a portion of all sources.  In addition,
quantitative risk estimates from pollutants formed ,or transformed in the
atmosphere (secondary formation) remain unquantified for almost all '
pollutants.  Evidence to date suggests secondarily-formed pollutants may
pose a significant component of total cancer risk.  Based on these
considerations, the actual magnitude of the problem, therefore, can
easily be larger than estimated in this study.  On the other hand, the
estimates presented in this study are based on the use of unit risk
factors that are either upper-bound estimates or maximum likelihood
estimates of the carcinogenicity of a pollutant.  Quantitative estimates
derived from the use of these unit risk factors, therefore, could
overstate the true risk from a pollutant.  The net effect of these and
other uncertainties (e.g., assessing exposures) on total risk is
unknown.  It is expected, nevertheless, that the pollutants and source
categories considered herein are among the major contributors to cancer
risk from air toxics based on our current state of knowledge.
      Lifetime Individual Risks
      Maximum lifetime individual risks of 1 x 10"4  (1  in  10,000)  or
greater were reported in almost all of the studies examined for this
report (see Table 3-2).  Risk levels this high were reported for such
specific sources as major chemical manufacturers,  waste oil
incinerators, hazardous waste incinerators,  publicly owned treatment
works (POTWs), steel mills, hospitals, traffic intersections, and
hazardous waste treatment, storage, and disposal facilities (TSDFs).
                                   4-3

-------
Risk levels reported for areawide lifetime individual  risks,  which
includes risk from point and area sources, were generally around 10
(see Table 3-6).
      On an individual pollutant basis, maximum individual  risks of 1 x
10"4  or  greater  were  reported  for  16 of the pollutants included  in  the
NESHAP/ATERIS data base (see Table 3-2).   Twelve of these pollutants
were estimated to have maximum individual risks of 1.0 x 1()"3 or
greater.  These'estimates of risks are related to specific sources.
However, because of the nature of the  assessments contained in, the
ATERIS data base, there is a very large degree of uncertainty associated
with some of these estimates for specific sources.
      Multi-pollutant lifetime individual risks in four urban areas due
to exposure to 9 to  16 pollutants (at  one monitoring site  in each  urban
area) ranged from 3 x 10"4 to  3  x 10"3  (see Table 3-7).  These estimates
were based on ambient-measured data and generally cannot be related to
specific point  sources.
      While the present study shows the estimate of nationwide  cancer
cases to be somewhat  larger than was estimated  in the 1985 Six-Month
Study,  the maximum and areawide lifetime  individual risks  estimated  in
        *.
the  present study are nearly  identical to those estimated  in the  1985
Six-Month Study.  The broader scope of the present study has resulted  in
identifying additional types  of  sources  (e.g.,  TSDFs, POTWs) that  can
contribute to significant maximum individual  risks.
Nature  of the Cancer Risk
       Individual  Pollutants
      As discussed  in Chapter 2,  there is considerable  uncertainty with
the  absolute risk estimates for  some  of  the  pollutants  examined in this
                                   4-4

-------
 study.   Nevertheless,  the  available  information  indicates  seventeen1 of
 the  approximately  90 pollutants  examined may  each  account  for  10  or more
 cancer  cases  per year  nationwide.  Of  these,  thirteen may  each  account
 for  40  or more  cancer  cases  per  year.  These  thirteen are:   PIC;
 1,3-butadiene;  hexavalent  chromium;  formaldehyde;  benzene; chloroform;
 asbestos; dioxin;  arsenic; ethylene  dibromide; gasoline vapors; ethylene
 dichloride; and carbon tetrachloride.
      The seventeen compounds that are estimated to contribute  at least
 10 excess cancer cases per year  nationwide appear  to be most frequently
 associated with high maximum individual risks.  However, other  compounds
 may  be.the most significant  contributor to the maximum individual risk
 for  a particular city.  For  example, coke oven emissions in the
 Southeast Chicago  study contributed  over 75 percent of the highest
 estimated lifetime individual risk.  Individual compounds, such as
 epichlorohydrin and styrene, that have small  aggregate cancer incidences
 may  also be associated with  high maximum individual risks  (greater than
 1 x  10'4).
      For the most part, the individual compounds  found to be the more
 important contributors to cancer risk in the present study are the same
 as those found in the 1985 Six-Month Study.   The most significant
 difference is the addition of 1,3-butadiene to the list of potentially
 important contributors.  Dioxin may also be a significant contributor,
 but the uncertainties associated with its risk estimates make it
 difficult to conclude this at this time.   Several pollutants, on the
 other hand,  appear to be somewhat less of a factor in terms of aggregate
       Acrylonitrile,  arsenic, asbestos, benzene, 1,3-butadiene, cadmium,
carbon tetrachloride,  chloroform, chromium (hexavalent), dioxin, ethylene
dibromide, ethylene dichloride,  formaldehyde, gasoline vapors, PIC, vinyl
chloride, and vinylidene chloride.
                                  4-5

-------
cancer risk, but not necessarily in terms of maximum individual  risk.
Changes in the pollutants identified in the present study and in the
1985 Six-Month Study as the more important contributors are primarily
due to the broader scope of the present study and to newer estimates of
the unit risk factors for the individual pollutants.
      Sources
      As in the 1985 Six-Month Study, a wide variety of sources
contribute to aggregate incidence and individual risk (see Table 3-11).
Motor vehicles were found to be the largest contributor to nationwide
annual incidence, contributing almost 58 percent of total  incidence
(including  estimated risk attributable  to  the secondary formation of
formaldehyde).  The risk associated with the secondary formation of
formaldehyde was  estimated  to account for  6.5 percent of  the total
estimated  incidence  (130 annual cancer  cases).  Of these  130 annual
cancer cases, 93  are estimated to  be attributable  to volatile organic
compound  (VOC)  emissions from area  sources (including  45  from mobile
sources)  and  37 from point  sources  (see Table 3-10).   Electroplating
 (6%)  was  the  third  largest  contributor  to  aggregate incidence  as  a
result  of chromium  emissions.   The next five major contributors were
TSDFs (5%); woodsmoke  (5%); asbestos,  demolition  (4%);  gasoline
marketing (3%); and solvent use/degreasing (3%).   Unspecified  point
 sources (3%)  and cooling towers (3%)  were the  ninth and tenth  largest
 contributors to total  annual incidence.
       In general, a significant portion of the cancer risk from specific
 sources was usually due to a few pollutants, even where a source emitted
 many different pollutants.  For example, over 70 pollutants were
 included in the  analysis on hazardous waste combustors, but two
 pollutants (cadmium and hexavalent chromium) were estimated to be
                                    4-6

-------
responsible for almost 90 percent of the estimated cancer cases from
hazardous waste incinerators and three pollutants (cadmium, hexavalent,
chromium, and arsenic) for almost 90 percent of the estimated cancer
cases from hazardous waste boilers and furnaces.
      Both mobile and stationary sources were found to contribute
significantly to total nationwide annual incidences.  Considering both
direct emissions to the atmosphere and secondary formation of
formaldehyde, mobile sources were estimated to contribute approximately
58 percent and stationary sources approximately 42 percent of total
annual incidence.  Area sources were found to contribute approximately
75 percent and point sources approximately 25 percent of the total
cancer incidence (see Table 3-11).
      The relative contribution of the aggregate types of sources (i.e.,
point vs. area, mobile vs. stationary) to total annual incidence can
vary significantly for specific geographic areas.  For example, the
Southeast Chicago study showed point sources contributing almost 50
percent (vs. the 20 percent noted above) and stationary sources
approximately 60 percent (vs. 42 percent from above) of the total annual
incidence estimated for Southeast Chicago (see Table 3-28).  These
differences are most likely due to* the significant contribution to risk
from steel mills in the Southeast Chicago area.
      With regard to lifetime individual risk, reported maximum
individual risks usually were associated with specific point sources,
such as industrial  facilities or chemical manufacturers.  Based on the
information in the lEMP-Santa Clara study,  the levels of maximum
individual risk associated with individual  area-type sources (e.g.,
gasoline marketing, degreasers, waste oil burning) appear to be lower
than those found for sources typically included in a point source
                                   4-7

-------
category (see Table 3-22).  However, the lEMP-Santa Clara study found a
maximum lifetime individual risk of 10"4 for at least one traffic
intersection.  On the other hand, not all  point sources have high
maximum lifetime individual risks associated with them.  In fact, the
majority of point sources in some source categories have maximum
individual risks of 10"6  or  less.
      As noted earlier, areawide lifetime individual risks were
generally lower than the maximum individual risk values within
comparable geographic locales.  The relative contribution of area and
point sources to areawide lifetime individual  risks can vary from one
locale to another.  For example, the lEMP-Santa Clara showed area
sources contributing approximately 50 percent of the areawide lifetime
individual risk and point sources approximately 15 percent (see Table
3-23).  (The remaining 25 percent was from carbon tetrachloride, which
was not allocated in that study to either area or point source.)  In
contrast, the Southeast Chicago study shows point sources contributing
approximately 48 percent and area sources approximately 31 percent of
the areawide lifetime individual risk (the remaining 20 percent was from
formaldehyde and carbon tetrachloride, which were not allocated to
either area or point source) (see Table 3-28).
      Among area sources, mobile sources were found to be responsible
for between 50 and 60 percent of the areawide lifetime individual risk
(see Tables 3-23, 3-24,  and 3-28).  Solvent use and heating in the IEMP-
Kanawha Valley study (see Table 3-24) and home heating in the Southeast
Chicago study (see Table 3-28) were identified as having areawide
lifetime  individual risks approximately one-half to one-quarter as large
as those  associated with mobile sources.
                                   4-8

-------
      Geographic Variability
      Exposure to individual air toxics varies on a city-to-city basis
as well  as on an intra-city basis.  For some pollutants,  such as
benzene, the variation appears to be relatively small,  less than a
factor of two (.see Table 3-13).  For other pollutants,  the variation is
higher,  ranging to a factor of almost 20.   For the pollutants compared,
the degree of variation in ambient concentrations for a particular
pollutant apparently can vary by the same  degree within a city as
between cities.
      The variations in ambient concentrations for individual pollutants
can lead to variations in the number of cancer cases and the cancer rate
(i.e., cancer incidence per year per million population)  between
geographic areas.  In spite of the differences in risk attributable to
individual pollutants, areawide lifetime individual risks were found to
be generally the same between the geographic locales examined in this
study (see Table 3-6).  Particular geographic locales may have
substantially higher areawide lifetime individual risk.  If this occurs
in a relatively sparsely populated locale, a low absolute number of
cancer cases would mask a high cancer rate and this higher-than-average
areawide lifetime individual risk.  In a similar manner,  a relatively
low areawide lifetime individual risk may  mask a significant maximum
individual risk that affects a small portion of the local population.
      Most of the geographic locales reviewed in this study showed
comparable maximum or highest estimated lifetime individual risk levels
(see Table 3-15).  However, this does not  mean that the same number of
people are exposed to that level, of risk in each city.
      The pollutants and source categories that are the most important
contributors to risk (annual incidence and maximum individual risk) in a
                                   4-9

-------
geographic area will vary from one area to another.  For some
pollutants, the variation may be relatively small and the primary source
will be the same between areas.  For example, benzene was found to
contribute between approximately 5 and 10 percent of the total annual
incidence in six cities, with between 45 and 80 percent of the benzene-
related cancer incidence attributed to motor vehicles (see Table 3-17).
Other pollutants show a wider range of variation, and the relative
contribution for some pollutants can be dramatically affected by the
presence of major point sources.  For example, in five of the six
selected cities, 1,3-butadiene was estimated to contribute between 6 and
24 percent of the total cancer incidence, all attributable to motor
vehicles.  In the sixth city, over 48 percent of the total cancer
incidence was attributed to 1,3-butadiene.  Of the 1,3-butadiene-
related cancer incidence in this city, over 80 percent was attributed to
chemical manufacturing plants and less than 20 percent to motor
vehicles.
      In general, the results and conclusions of the present  study are
consistent with those drawn in the 1985 Six-Month Study regarding
geographic variability.
                                   4-10

-------