EPA-450/l-90-004b
September 1990
Cancer Risk from Outdoor Exposure
to Air Toxics
VOLUME II: APPENDICES
FINAL REPORT
Appendix A. Comments Received on the External
Review Draft
Appendix B.
Appendix C.
Cancer Risk Reduction Analysis
for Selected Pollutants
Summaries of Pollutant-Specific
and Source-Specific Studies
(Including Noncancer Health Risk
Project on Air Toxics)
-------
-------
APPENDIX A
COMMENTS RECEIVED ON THE
EXTERNAL REVIEW DRAFT
A-l
-------
Approximately 23 comment letters were received on the September,
1989, external review draft of this report. Commenters included EPA
personnel, industry representatives, State and local agencies, and
university professors. Table A-l lists the commenters and their
affiliations.
The comments were tabulated and sorted by subject areas. The EPA
then reviewed the comments to determine which ones would be incorporated
into the final report. The following paragraphs summarize how EPA
responded to some of the major comments.
Reference Section
Several commenters suggested that a reference section be added to
both Volume I and Volume II. While separate reference sections were not
created, we agree that more complete referencing was needed. In Volume
I and Appendix B (Volume II), references are provided as they occur. In
Appendix C, references are provided at the end of each summary.
Along similar lines, we have improved the citations within the
report to facilitate access to source material for the reader. We have
also increased cross-referencing within the report, particularly in
Chapter 4, to facilitate the location of related information.
Glossary
Several commenters suggested that, to improve the readability of
the report, a more complete glossary be provided, that a list of
acronyms be provided, and that these be placed in Volume I, rather than
as appendices in Volume II. We agree that these suggestions improve the
readability of the report, and have incorporated them in Volume I of the
final report.
A-2
-------
TABLE A-l
LIST OF COMMENTERS
Commenter
Affiliation
Donald J. Ames
Walter J. Bishop
Geraldine V. Cox
Robert Fegley
John L. Festa
Robert C. Kaufman
Maryann Froehlich
John Graham
William Groan
Richard Guimond
Charles E. Holmes
Stacey Katz
Steven D. Lutkenhoff
Bruce K. Maillet
William H. McCredie
John F. Murray
John E. Pinkerton
John Roberts
Robert R. Romano
Stationary Source Division, California Air
Resources Board
East Bay Municipal Utility District
Chemical Manufacturers Association
U.S. EPA, Air Economics Branch
American Paper Institute
U.S. EPA, Regulatory Integration Division
School of Public Health, Harvard University
Hardwood Plywood Manufacturers Association
U.S. EPA, Office of Radiation Programs
Department of Air Pollution Control, Common-
wealth of Virginia
U.S EPA, Office of Technology Transfer and
Regulatory Support
U.S. EPA, Environmental Criteria and Assessment
Office.
Division of Air Quality Control, Commonwealth
of Massachusetts
National Particleboard Association
The Formaldehyde Institute, Inc
National Council of the Paper Industry for Air
and Stream Improvement, Inc.
Engineering Plus, Inc.
Chemical Manufacturers Association
A-3
-------
TABLE A-l
LIST OF COMMENTERS (concluded)
Commenter
Affiliation
Sara D. Schotland
James H. Souther!and
Donald F. Theiler
Dr. Paul Urone
William Waugh
Cleary, Gotlieb, Steen, & Hamilton
U.S. EPA, Pollutant Characterization Section
Department of Natural Resources, State of
Wisconsin
Department of Environmental Engineering
Sciences, University of Florida
U.S. EPA, Health and Environmental Review.
Division
A-4
-------
Terminology
We received numerous comments concerning some of the terminology
used in the report. We have reviewed carefully all of the suggestions,
and have incorporated most of them in the final report. For example,
several commenters did not like the term "best estimate" when referring
to the estimates of nationwide cancer incidence obtained as the result
of the reduction analyses. In the final report, we now use the term
"point estimate," even though for four pollutants our "point" estimate
of nationwide annual cancer incidence is still a range (rather than a
single number).
Several commenters requested that we use the term "upper-bound" to
qualify our nationwide estimates. We have not done this in the final
report. We believe that to describe the estimates as upper-bound would
not be an appropriate descriptor of national estimates aggregated across
a limited set of pollutants and source categories studied. It is
possible that the risk methodologies and as yet unquantified risks from
other pollutants and sources may make the use of "upper-bound"
inappropriate. We agree that the unit risk factors in and of themselves
are upper-bound estimates. However, other factors that enter into
estimating nationwide cancer incidence may make the use of the term
"upper-bound" misleading, especially since it is so closely associated
with unit risk factors.
Several commenters requested that the terminology associated with
lifetime individual risk be reviewed for clarity and consistency. This
has been done, although some of the original detail has been retained.
In describing the risk estimates, we have revised the language to
reflect past EPA descriptions that note the derivation of the unit risk
A-5
-------
factors, and that the actual risk is unknown and may even be as low as
zero.
Specific Pollutants
A number of comments were received on several pollutants, mostly
concerning the uncertainties associated with each one's risk. In
general, the report already identified a number of uncertainties
associated with individual pollutants. In addition, it is not the
purpose of this report to review all of the uncertainties associated
with each individual pollutant. This report rather tries to highlight
\
some of the more important uncertainties in order to give the reader a
feel for the uncertainty associated with the estimates. Other reports
and studies should be reviewed for details on any individual pollutant.
Nevertheless, we have considered each point raised by the
commenters. Those associated with formaldehyde have generally been
incorporated, with the exception that the reported risk estimates
continue to be based on the upper-bound unit risk factor and not the
maximum likelihood unit risk factor. This decision is consistent with
current EPA policy. The other comments generally have not been
incorporated because, in our opinion, they did not add to the sense of
uncertainty already presented in the report.
Source Categories
One commenter questioned the segregation of the individual source
categories between point and area sources. This was reviewed, and we
agree that several individual sources that were identified as area
sources should have been identified as point sources. The,final report
makes these changes. Because of these changes, the final report shows
area sources contributing approximately 75 percent of the total
A-6
-------
estimated national cancer incidence and point sources 25 percent. The
draft external review draft showed a 80 percent contribution by area
sources and 20 percent by point sources.
A brief discussion explaining how the source categories were
divided between area and point has been added. While this should help
the reader understand how we assigned the source categories, a clear
distinction between area and point sources is not always possible.
Several commenters made suggestions concerning individual source
categories. In most instances, these comments were not incorporated
because it was felt that the text already adequately covered the comment
or that the additional detail was not appropriate for this report. One
commenter noted a discrepancy in the estimated cancer incidence for
POTWs. This discrepancy has been corrected in the final report.
One commenter requested that the cancer risk estimates for TSDFs
and sewage sludge incinerators be eliminated from the report because of
the methodologies are flawed and the estimates from them are not
meaningful. We have retained the estimates from these two source
categories. We agree that these two source categories have uncertain
risk estimates, and this has been noted in the appropriate spots in the
report.
ATERIS/SARA Title III
Several comments were received concerning the'ATERIS data base and
the use of toxic emission information received under SARA Title III.
The SARA Title III data are not reported in a form that allows for the
development of risk estimates. It is outside the scope of this study to
develop original analyses based on those data. Thus, the SARA Title III
data are not used in this report. Future updates of this report will
A-7
-------
include the results of any risk analyses based on the SARA Title III
data. In the meantime, the risk estimates from the ATERIS data base
have been retained in the final report. These estimates are available,
and we believe do provide useful information.. In addition, they have
been adequately caveated in an attempt to limit their misuse.
Uncertainty in Cancer Risk Estimates
Several commenters requested that we segregate the cancer
incidence estimates for the individual pollutants on the basis of the
relative uncertainty with each estimate. We agree that this can be a
desirable segregation. However, such an effort is outside the scope of
resources allocated to this study. Further, we believe that there is
sufficient information in the report that allows the reader to gain a
sense of the relative uncertainty of each of the estimates. Thus, the
final report does not incorporate this suggestion.
Perspective of Cancer Estimates to Total Cancer
Several commenters suggested that a brief paragraph relating the
estimates of cancer risk from outdoor exposure to air toxics to
estimates of total cancer incidence. We agree that such a comparison is
useful for the reader, and have addressed this comment in the Executive
Summary.
Maximum Exposed Individual
One commenter suggested that the method for calculating the risk
to the maximum exposed individual should be redone by collecting some
actual data of residential living patterns and human activity patterns.
According to the commenter, it is not defensible at this stage to
continue to use totally unrealistic assumptions, especially since these
MEI/MIR numbers may take on increasing regulatory importance in the
A-8
-------
future. This .comment could not be responded to within the context of
this report. Therefore, the methodology used to estimate the MEI/MIR
estimates has not been changed.
A-9
-------
-------
APPENDIX B
CANCER RISK REDUCTION ANALYSIS
FOR SELECTED POLLUTANTS
B-l
-------
The purpose of this appendix is to present the reduction analyses
for the 23 individual pollutants that were initially identified as
possibly resulting in at least 10 cancer cases per year nationwide. The
analyses derive a point estimate, or as narrow a range as possible, of
the annual cancer cases per million population for each pollutant/source
category combination from the range of estimates found in the various
reports and studies. The specific data on the estimated number of
annual cancer cases and the estimated annual cancer cases per million
population for a pollutant by each source category for each study are
presented in this appendix. The annual cancer cases per million popula-
tion are shown in parentheses in the tables. NOTE: Unless otherwise
noted, all risk estimates have been adjusted based on a consistent set
of unit risk factors.
Please note that the last two columns in each table are "Range"
and "Point Estimate". The numbers in these two columns are estimates of
nationwide annual cancer cases. The estimates are conservative in that
actual risk may be higher, but is more likely to be lower.1 For the
"Range" column, the estimates of nationwide annual cancer cases were
calculated, in most instances, by taking the lowest and highest annual
cancer cases per million population for a source category and multi-
plying it by 240 (1986 U.S. population in millions). The "Total" for
this column simply represents the summation of the low end of the range
and the summation of the high end of the range. The column labeled
"Point Estimate" presents the estimates of nationwide annual cancer
1 The unit risk factors used to estimate cancer risk are based on a
linearized multistage procedure that leads to a plausible upper limit to
the risk that is consistent with some proposed mechanisms of
carcinogenesis. Such an estimate, however, does not necessarily give a
realistic prediction of the risk. The true value of the risk is unknown,
and may be as low as zero.
B-2
-------
incidence based on the results of the reduction analyses. The text
discusses how the ranges and point estimates were selected.
Where the cancer incidence for a pollutant was estimated with both
modeled concentrations and ambient-measured concentrations, this is
shown in the "Totals" row. Separate headings are given for the modeled
concentration-based estimates (i.e., "Modeled") and for the ambient-
measured concentration-based estimates (i.e., "Ambient"). Some studies
estimated cancer incidence using both types of concentrations. For
these studies, entries are made for both "Modeled" and "Ambient" totals.
An index to the pollutants covered in this appendix is presented
below.
Pollutant
Acrylonitrile
Arsenic
Asbestos
Benzene
1,3-Butadiene
Cadmium
Carbon tetrachloride
Chloroform
Chromium
Coke Oven Emissions
1,2 Dichloropropane
Dioxin
Ethylene dibromide
Ethylene dichloride
Ethylene oxide
Formaldehyde
Gasoline vapor
Methylene chloride
Perchloroethylene
PIC
Tri chloroethylene
Vinyl chloride
.Vinylidene chloride
Page Number
B-4
B-9
B-17
B-21
B-31
B-37
B-46
B-52
B-58
B-65
B-67
B-68
B-71
B-76
B-83
B-87
B-94
B-98
B-102
B-107
B-124
B-130
B-135
B-3
-------
Acrvlonitrlle. Point-sources of acrylonitrile include acrylonitrile
monomer production, acrylic and modacrylic fiber production, ABS/SAN
resin production, nitrile rubber production, and acrylamide and
adiponitrile production. Other production processes that consume a
small percentage of acrylonitrile are nitrile barrier resin production,
fatty amine production, and as an absorbent.2 Acrylonitrile emissions
have also been identified as occurring from publicly owned treatment
works (POTWs) and treatment, storage, and disposal facilities (TSDFs).
Seven studies included acrylonitrile as a pollutant of concern
(see Table B-l). Seven specific source categories were examined. Three
of the studies (35-County, the lEMP-Kanawha Valley, and the Southeast
Chicago studies) did not identify the specific types of plants included.
A comparison of plant locations in the NESHAP/ATERIS data base with the
counties included in the 35-County study revealed that some of the
counties examined in the 35-County study had acrylonitrile sources
covered in the NESHAP/ATERIS data base. On the other hand, none of the
plant locations examined in the NESHAP/ATERIS data base were identified
as being in the areas covered by the lEMP-Kanawha Valley study.
The annual incidences were based on modeled estimates of ambient
concentrations. The lEMP-Kanawha Valley used a box model and an,ISCLT
model. The box model was known to likely overestimate actual exposure
levels, but was used in the study to bound the problem. For the TSDF
study, the annual incidence attributable to acrylonitrile was estimated
by assuming the annual incidence from acrylonitrile was proportional to
2 U.S. EPA. Locating and Estimating Air Emissions from Sources of
Acrvlonitrile. EPA 450/4-84-007a. March 1984.
B-4
-------
a:
o
C3
UJ
UJ
o
z
o
<5
Q
13
CO
1
1
0
UJ
1
1 <
ZZ 3-*
a. >
LU
CD
UJ
z
2
CO O
CO O>
(U CO
J= O
4-1
CO
O. 2 0>
s:
4-» 0) 4-»
t- > C
.E S- O
1 CJ
(O
O
0) 0) 4->
CO CO 't-
c
LU
"
^
h-
2
Q- CO
1C 5
CO UJ
UJ I
z <
UI O£
CJ 0
/v {j
O 'I1
O
u
co in iy» N- *o co
-» o o «- o T- o
oi-eacaocsea cao
«-
0
in ro o- *o
CO O «- O T- O »-
o^-o0000 o°
o
X
o *-
CO CO
o
o
O V^
o o in in
O O (MO
o o o
xx ' CD
^^
«^
CO 0
O 0
o cp
rf-\
sr
« o
"~ o
t\j
o
o
o
-i" rvj in po c\J 10
uio rocD l>-co >oo
OO t O OO « O
O Q O O
0 0 0 0
o o o o
CO
fl) t CO
D) CO 0) O) O
at. ja *- o
3 o - co z «-
4-' U- t, ^
£_1_'=E U-0)t? "C§ C» T-4-- .- «
3 u- (o 5 » i_co E c
SCO CICO CO U ' f O
i CO*
0 0
CD
CD
o
J-I
o
^
o
**
^^
CD
o
Q
o
UI
UI
CO Q
_J 3E
V
o
1
B-5
-------
1
.
0 DC
H- O
o o
O
JC
S
to
g
u
CO
a
4>
a
8
§
U)
8 a -
u jc
»* (W W
M y o
C C C T3
- ra ro o>
O JC O
c. u en u
Q. CD GJ -
0> O TJ
g ^ 5 5 .5 "
8
8
*J
8
UJ
1
B-6
-------
its relative contribution to the weighted unit risk factor used to
estimate the total annual incidence from TSDFs. This is a very crude
estimate.
Point Estimate. Excluding for the moment the estimated annual
cancer cases from the chemical manufacturing source category of the
lEMP-Kanawha Valley study and from TSDFs, the estimated annual cancer
cases from the other five studies total between 2 and 3 cancer cases per
year. It is quite likely that there is some double counting between the
NESHAP/ATERIS estimates and the 35-County estimate for point sources
(because, as noted above, some of the acrylonitrile sources identified
in the NESHAP/ATERIS data base are located in counties evaluated in the
35-County study). Double counting is also likely with regard to the
POTW estimates in the 35-County study and the POTW study. With regard
to the chemical manufacturing sources in the lEMP-Kanawha Valley study,
it does not appear that these sources are included in the NESHAP/ATERIS
data base. Since the sources of acrylonitrile emissions in the lEMP-
Kanawha Valley study are point source emissions related to a specific-
type, but unknown, chemical manufacturing facility, it is not possible
to estimate annual cancer cases beyond this study's limited geographic
range, and it would.be unreasonable to apply its annual cancer incidence
per million population to the entire U.S. population to obtain a
national estimate. Taking these things into consideration, the point
estimate of total annual cancer cases from these five studies is
estimated to be approximately two cancer cases per year.
As noted earlier, the estimated cancer cases from TSDFs is a very
crude estimate, but is the only estimate available at this time. Since
the TSDF study is a national estimate, it most likely includes the
Southeast Chicago study area. Even if it does not, the negligible
B-7
-------
estimated cancer cases from TSDFs in the Southeast Chicago study would
not affect the estimate of annual cancer cases from TSDFs (i.e., 11
cancer cases per year). Combining the six studies, a total of 13 cancer
cases per year nationwide from exposure to acrylonitrile emissions is
estimated.
B-8
-------
Arsenic. Arsenic is emitted from a number of point and area sources.
Point sources include smelters, glass manufacturing, and steel mills.
Area sources are primarily combustion related activities.
Thirteen studies included arsenic as a pollutant of concern (see
Table B-2). Of these studies, three estimated annual cancer cases on
the basis of ambient measurements (the lEMP-Santa Clara study, the South
Coast study, and the Ambient Air Quality study) and the other ten used
modeling to estimate ambient concentrations and cancer cases. The South.
Coast study also included an estimate based on modeling.
Ambient Estimates. The South Coast study estimated 1.5 cancer
cases per year, or approximately 0.14 cases per year per million
population, based on an average ambient concentration of approximately
2.4 x 10"3 micrograms per cubic meter (/ig/m3). This estimate was based
on over 300 individual samples at a total of 7 sampling sites; 24 of the
samples were below the minimum detectable limits. The lEMP-Santa Clara
study estimated 0.2 to 0.4 annual cancer cases, or approximately 0.14 to
0.29 cancer cases per year per million population. The lower estimate
reflects half of the minimum detectable limit of the analytical
equipment used (i.e., one-half of 0.0055 jig/m3). The upper estimate
reflects the average of the lower estimate with the samples above the
minimum detectable limits. The Ambient Air Quality study estimate of 68
annual cancer cases, or approximately 0.28 cancer cases per year per
million population, was based on 163 areas with ambient data. Because
of the larger geographic scope of the Ambient Air Quality study, 68
annual incidences was selected as the best estimate of cancer cases from
arsenic based on ambient air quality data.
B-9
-------
TABLE B-2
ESTIMATED ANNUAL CANCER CASES FROM ARSENIC BY SOURCE CATEGORY
SOURCE
CATEGORY
INDIVIDUAL STUDY
Ambient
Air
Quality
NESHAP/ Coal and Oil Hazardous
ATERIS Combustion Waste
Combustors
Municipal
Waste
Combustors
Sewage Waste Oil
Sludge Combustion
Incinerators
Thirty-
five
County
Chcralcal
Hanufacturing
Class
Hanufacturing
Hon- ferrous
Smelters
Coal and Oil
Combustion/
Heating
Hazardous Uastc
Combustors
Municipal Waste
Coccus tors
icircrators
'*i.v= Oil
Combustion
Other
Solvent Use
Voodsmokc
Steel Hills/
Iron and Steal
Zinc Oxide
0.0043
(0.00002)
0.4
(0,0017)
1.1
(0.0046)
5.3
(0.022)
0.005
(0.00002)
0.16
(0.0007)
0.17
(0.0007)
0.087-0. >-.Z
(O.OOU36-0.002)
0.08
(D.00033)
7.46
(0.16)
0.64
(0.014)
33.9
(0.72)
TOTALS
MODELED
AHBIEHT
68
(0.28)
1.6
(0.0067)
5.3
(0.022)
0.005
(0,00002)
0.16
(0.0007)
0,17
(0.0007)
0.087-0.48
(0.00036-0.002)
42
(0.89)
B-10
-------
TABLE B-2 -- concluded
ESTIMATED ANNUAL CANCER CASES FROM ARSENIC BY SOURCE CATEGORY
SOURCE
CATEGORY
Chemical
Manufacturing
Glass
Manufacturing
Non-ferrous
Smelters
Coal and Oil
Combustion/
Heating
Hazardous Waste
Combustors
Municipal Waste
Combustors
Sewage Sludge
Incinerators
Waste Oil
Combustion
Other
Solvent Use
Woodsmoke
Steel Mills/
Iron and Steel
Zinc Oxide
TOTALS
MODELED
AMBIENT
INDIVIDUAL STUDY
IEMP- IEMP- Southeast
Five City Kanawha Santa Clara Chicago South Coast
Valley
0.01
(0.0006)
0.37 0.018
(0.023) (6.18)
0.0004
(0.004)
0.76 0.0014
(0.048) (0.004)
0.0009
(0.000057)
0.013
(0.0008)
0 ' 0.02
(0.05)
1.14 0.018 0.021
(0.072) (0.18) (0-055)
0.2-0.4 " 1.5
(0.14-0.29) (0.143)
NATIONWIDE
POINT
RANGE3 ESTIMATE13
0.0043 0.0043
0.4 0.4
0.1-1.1 0.1-1.1
5.3-43 5.3
0.005 0.005
0.16 0.16
0.17 0.17
0.09-3.4 0.5
1-173 1-34
0.01 0.01
0.2 0.2
c c
0.08 0.08
7.5-222 8-42
34-70 68
B-ll
-------
Footnotes to Table B-2.
NOTE:
Mumbers in parentheses are annual cancer cases per million population. For nationwide
studies, annual cancer cases were divided by 240 million, unless otherwise noted. For
studies with smaller geographic scopes, the annual cancer cases were divided by the study's
population.
MOTE: An "x" in a column indicates that the source category was considered in the study, but a
specific cancer risk for the source category was not indicated.
8 The numbers in this column were calculated by taking the lowest and highest incidence rates for
a source category and multiplying it by 240 (1986 U.S. population in millions). The total for
this column is the summation of the low end of the range and the sum of the high end of the
range.
k The numbers in this column present the results of the reduction analyses. In most instances, a
point estimate of nationwide annual cancer incidence was derived for each pollutant/source
category combination. In some instances, a point estimate could not be reasonably derived. For
these instances, as narrow a range as possible of nationwide annual cancer incidence was
estimated, and such ranges appear in this column. The text discusses how these point estimates
and ranges were derived.
0 Estimate for this source category was assumed to be included in the "other" source category.
See text for explanation.
B-12
-------
Modeled Estimates. The estimates of cancer cases based on
modeling provide for a wider range of estimates. One difficulty in
assembling Table B-2 was ensuring that the categories are mutually
exclusive. The most important' examples in terms of estimated cancer
cases are the categories "Heating" and "Other/Point Sources." The
"heating" category, which was used in the lEMP-Kanawha Valley study and
the 35-County study, was assumed to be the same as or a subcategory to
the "Coal and Oil Combustion" category of the Coal and Oil Combustion
study,3 as were the "utility boilers" and "oil combustion" categories of
i
the 5-City study.
There are four source categories in Table B-2 for which more than
one study estimated cancer cases. For "coal and oil combustion/
heating," the lEMP-Kanawha Valley and the 35-County studies, which used
"heating" to describe the source category, both estimated nearly
identical cancer cases per year per million population rates (0.18
versus 0.16, respectively). .These estimates are higher than the coal
and oil combustion estimates in the Coal and Oil Combustion and the
5-City studies, both of which calculated approximately 0.022 cancer
cases per year per million population. One reason for this difference
in annual cancer incidences per million population appears to be the use
of different emission factors. (Both sets of annual cancer incidences
per million population already have been "corrected" for unit risk
factors.) the 5-City study notes that the coal and oil combustion
emission factors for arsenic, chromium, formaldehyde, and nickel were
revised from those of the original 5-City study using more recent test
3 The 35-County study identified heating as being composed of
commercial, industrial,.and residential heating by fuel type, i.e., coal
and oil. This is the same type of breakdown as in the Coal and Oil
Combustion study.
B-13
-------
data, whereas the 35-County study notes that newer source factors for
area sources such as heating and waste oil burning were not included
within the updated 35-County study. Thus, it appears that a best
estimate of nationwide risk from this source category would be based on
the annual cancer incidences per million population from the Coal and
Oil Combustion and 5-City studies, which yield an estimated 5 cancer
cases per year.
Three studies included "Waste oil combustion" as a source
category. Although a fairly wide range of cancer cases per year per
million population is shown (0.0004 to 0.014), total annual cancer cases
are relatively small (less than 4 per year at the highest annual cancer
incidence per million population). The higher estimate is from the 35-
County study, and as noted above, the 35-County study apparently did not
incorporate newer source factors for waste oil burning. Although
emission factors between the studies could not be compared, as a
specific national study on waste oil burning was available, the estimate
from that study (0.5 annual cancer cases) was selected as the best
estimate.
Two studies included "non-ferrous smelters" sources - the
NESHAP/ATERIS data base and the 5-City study. Because of the specific
locations of the non-ferrous facilities and the national scope of the
study, the NESHAP/ATERIS data base estimate was selected as the best
estimate.
Three studies include a source category for "other" sources. For
the 35-County study, the cancer risk associated with this source
category is large, 34 cancer cases per year, or approximately 0.72
cancer cases per year per million population. It is unclear as to what
sources were modeled to obtain this estimate, although municipal
B-14
-------
incinerators and steel mills (coke ovens) were included. For the
5-City study, the "other" source category shows a cancer rate of
approximately 0.05 cancer cases per year per million population. Again,
the sources included in this category are unspecified, although it is
clear that "iron and steel" is not included. The 5-City study did not
report any annual cancer risk due to arsenic emissions from iron and
steel mills. Finally, the Southeast Chicago study showed a relatively
small annual cancer incidence per million population for the "other"
category (0.004 cancer cases per year per million population)," but a
more significant one for steel mills (0.05 cancer cases per year per
million population). For the "other" and "iron and steel/steel mills"
source categories, a combined range of 0.05 to 0.72 cancer cases per
year per million population can be created. Based on the 35-County and
the Southeast Chicago studies, steel mills appear to be the largest
contributor to this annual cancer incidence per million population. To
apply the annual cancer incidences per million population from these two
studies for these two source categories to the total U.S. population
would result in an estimated 13 to 173 cancer cases per year. One
difficulty with this is that steel mills are site-specific sources that
cannot easily be extrapolated to national estimates. For example, the
Southeast Chicago study modeled four steel mills and the 35-County study
selected counties that, in part, were khown to have sources emitting the
pollutants being studied, in this case arsenic. Thus, it is unlikely
that applying the annual cancer incidences per million population from
these two studies to the entire U.S. population is appropriate.
Because of the uncertainty associated with the sources included in
the "other" source categories and with extrapolating risk from the
"steel mills/iron steel" source category, it is extremely difficult to
B-15
-------
narrow the range of risks. Excluding these two source categories, a
total of approximately 8 cancer cases per year nationwide is estimated.
If the cancer risk from the 35-County study is primarily attributable to
steel mills/iron and steel, then the nationwide estimate could be
increased to 42 (8 plus 34), keeping in mind that not all steel mills
may be located in these 35 counties.
Point Estimate. The best estimate of nationwide cancer cases
using ambient measured data appears to be 68 cancer cases per year. For
modeled estimates, a range of 8 to 222 annual cancer cases was
developed. For reasons noted above, the upper end is a likely
overestimation. Because of the uncertainties in trying to narrow the
modeling range (which was narrowed to 8 to 42 annual cancer cases) and
the relative extensiveness of the ambient data, the Ambient Air Quality
study's estimate of 68 annual cancer cases was preferred. Thus, a total
of 68 cancer cases nationwide due to exposure to arsenic is estimated.
B-16
-------
Asbestos. Annual cancer cases as a result of asbestos emissions have
been estimated for point sources, such as fabrication, milling,
renovation, and demolition, and from motor vehicles (see Table B-3).
One study examined point sources and three studies examined motor
vehicles. All four studies used models to estimate cancer risk. Annual
cancer cases due to asbestos emissions from point sources were estimated
to be approximately 82 per year under current compliance conditions with
the current asbestos standards. If full compliance with current
regulations were being met, annual cancer cases from point sources would
be less than 1 per year.
Using a range of unit risk factors derived from the National
Academy of Sciences (NAS), the Mobile Source study estimated urban
cancer cases due to asbestos from motor vehicles to be 0.41 cases per
year based on an emission rate of 4 ^g/mile and 113.4 cases per year
based on an emission rate of 27 //g/mile. The emission rate factors of 4
//g/mile and 28 ^g/mile were estimated to result in maximum annual
average asbestos levels in a central city area of approximately 2.5 x
"4
10" fjg/m and 1.75 x 10
~3
, respectively. Adjusting the estimated
annual incidences to a unit risk factor of 7.6 x 10"3 (^g/m3)"1 (as
listed in Table 2-6) results in a narrower range of estimated
incidence -- 4.7 to 33 cancer cases per year. According to the report,
the 4 //g/mile emission rate may be a better overall estimate than the 28
//g/mile. Using 4 ^g/mile, the 4~7 cancer cases per year translates into
approximately 0.026 cases per year per urban million population (urban
population equal 180 million).
The 5-City study shows cancer rates for motor vehicles ranging
from 0.0013 to 0.012 cancer cases per year per million population, with
a five-city average of 0.008 per million population. Without knowing
B-17
-------
TABLE B-3
ESTIMATED ANNUAL CANCER CASES FROM ASBESTOS BY SOURCE CATEGORY
SOURCE
CATEGORY
Hilling
Manufacturing
Fabricating
Renovation
Demolition
Motor Vehicles
TOTALS
MODELED
INDIVIDUAL STUDY
Mobile Southeast
Asbestos Sources Five City Chicago
0.004-0.005
(<0. 00021)
0.3-0.7
(0.0013-0.003)
0.05-0.2
(0.0002-0.0008)
0.41 0.00057C
(0.0017) (0.0015)
80.5
(0.335)
4.7-33d 0.13 0.0014
(0.026-0.183) (0.008) (0.004)
81.2-81.8 4.7-33 0.13 0.002
(0.34) (0.026-0.183) (0.008) (0.005).
NATIONWIDE
POINT
RANGE8 ESTIMATE"
0.004-0.005 0.005
0.3-0.7 0.5
0.05-0.2 0.13
0.41 0.41
80.5 80.5
1-44 6.24e
82-126 87.8
NOTE: Numbers in parentheses are annual cancer cases per million population. For nationwide
studies, annual cancer cases were divided by 240 million, unless otherwise noted. For
studies with smaller geographic scopes, the annual cancer cases were divided by the study's
population.
a The numbers in this column were calculated by taking the lowest and highest incidence rates for
a source category and multiplying it by 240 (1986 U.S. population in millions). The total for
this column is the summation of the low end of the range and the sum of the high end of the
range.
For
The numbers in this column present the results of the reduction analyses.1 In most instances, a
point estimate of nationwide annual cancer incidence was derived for each pollutant/source
category combination. In some instances, a point estimate could not be reasonably derived.
these instances, as narrow a range as possible of nationwide annual cancer incidence was
estimated, and such ranges appear in this column. The text discusses how these point estimates
and ranges were derived.
Includes "demolition." Based on data contained in the background information document to
support the Asbestos NESHAP. Scaled national estimates of asbestos emissions based on number of
households.
Risk estimates adjusted from original study by using a unit risk factor of 2.3 x 10
(fibers/ml) . Estimate is for urban population only (180 x 10° population). Original risk
estimates were 0.41 to 113.4 cancer cases per year, which reflect the use of an NAS-derived
range of unit risks.
Reflects applying the urban incidence rate from motor vehicles of 0.026 annual cancer cases per
urban million population to total U.S. population.
B-18
-------
the emission factors used in the 5-City Study, it is difficult to
determine the differences in the annual cancer incidences per million
population estimated for these two studies.
The Southeast Chicago study used an asbestos emission factor of
4 ^g/mile and a unit risk factor of 8.1 x 10"3 (Aig/m3)"1 to estimate
cancer incidence from mobile sources. The Southeast Chicago study
estimated approximately 0.0014 cancer cases per year, or 0.004 cancer
cases per year per million population. This annual cancer incidence per
million population falls within the range created in the 5-City study
for the five individual cities, but is approximately 6.5 times smaller
than the annual cancer incidence per million population from the Mobile
Source study based upon the same emission factor of 4 /zg/mile.
Different models were used in estimating risk between the two studies,
and this difference may explain the different annual cancer incidences
per million population. However, the information available is
insufficient to resolve this difference.
Point Estimate. For point sources, the best nationwide estimate
is 82 cancer cases per year. For motor vehicles, a range between 1 and
44 annual cancer cases can be created. The lower estimate applies the
Southeast Chicago study's annual cancer incidence per million population
to the total U.S. population and the upper estimate applies the upper
annual cancer incidence per million population from the Mobile Source
study to total U.S. population. Since the differences between the
Mobile Source study and the 5-City and the Southeast Chicago studies
cannot be resolved at this time, the results of the Mobile Source study
were selected as the best estimate for calculating nationwide incidence.
As the 4 ^g/mile emission rate appears to be a better overall estimate
than 28 ^g/mile, the upper end of the range may be closer to 6 cancer
B-19
-------
cases per year (0.026 cancer cases per year per urban million times 240
million) than to 44 cancer cases per year. Although this applies urban
data to rural populations, the difference in total annual incidence is
small (5 vs. 6 cancer cases per year). In light of these
considerations, the best estimate of cancer risk from motor vehicles is
selected as approximately 6 cancer cases per year. Combining the
estimates, a total of 88 cancer cases per year nationwide from exposure
to asbestos is estimated.
B-20
-------
Benzene. Benzene emissions occur from a multitude of sources, both
point and area sources. Most of the emissions are associated with
gasoline and other fuel combustion (such as motor vehicles) and
marketing (such as service stations). Fifteen studies included benzene
as a pollutant of concern, covering approximately 20 source categories
(see Table B-4).
Several of the studies estimated cancer risk using ambient
measurements or compared ambient measurements with modeled ambient
concentrations. In studies that compared ambient measurements with
modeled ambient concentrations (e.g.,the South Coast study, the
Southeast Chicago study, and the lEMP-Philadelphia study), amb,ient
measured concentrations were generally about two times higher than the
modeled concentrations. This is considered to be a fairly reasonable
agreement.
Ambient Estimates. The Ambient Air Quality study was the only
study to rely solely on ambient measurements to estimate the risk from
benzene. Ambient concentrations ranged from approximately 3 /ig/m3 to
15.5 /ig/m3 for individual city (urban) averages. The national urban
population weighted average concentration was 8.07 ng/m3 and the
national rural population average concentration was 0.6 ^g/m3. Based on
these average concentrations, annual cancer incidences were estimated to
be 181 per year, with a cancer rate of 0.75 cancer cases per year per
million population.
The other studies that included ambient measured concentrations
included the South Coast study, the Mobile Source study, the Southeast
Chicago study, the lEMP-Baltimore study, and the LEMP-Philadelphia
B-21
-------
,_
Ul
g
>-
CO
UJ
UJ
S!
£
U.
>
g
§
_J
i
s
i
K-
Ul
S
CO
1
CO
ro co
£
* 2
UJ -M
O3
X
4-*
o
u.
£ "*" 0
O 0
M
0> U>
4-> 3
u.
w
» u
x co
if
W C.
O SE
a. co
V) Ul
!«.£
rl
SOURCE
CATEGORY
ss
o o o ro
°.°. 00
o o *~*
N- N- eo ro CM
o « ^ « ro o
c\j o r\j o CM ro
o o.-» CD o -o o
fs_ v *r- « <\J
«^ o co ro
»- ir»
°S "ro "S
c
r~ fC
« o
0 0
O CO
0 0
o
O CO
o
CVI
12
1 1
U"\ <}
^^
CO
CO CM
CJ
o
0 ^
*4- to ro ro co -* o
in*- N-o S-o CMO oo oo
**OOOOO -0 ii OO
«o -o CD^ mrvj «^
ro o o o o o o
\s O O O O O
\^ v> N^ O O ^^
O
QJ
(/)D)«
_. (0 (D O D> >
r- 4-» 1 O C C C
_* <1) **- QJ -M QJ O '- t-
O^£.^ E N N 4->CO
« c_ » o; croc w «t,4->
ro ro 3 jz o D a; co ro u o
uzco u mu-ca UJ n CD z:
T *>O CO *O
ro o o o CM o
0* O O O O
x
2
ro o
o
0) O CD
S c c i? 2
CO CO O P O
CO O C- CJ
C <- O O) C
C d) Q. CO
O M- .E CO 3 O
£ CD 4J > (U C
_, Q; O UJ CO "-
B-22
-------
I
1
UJ
§
UJ
0
i
CO
z
UJ
m
g
CK
u_
CO
Ul
u
1
»
3
I
a
Ul
g
i
CO
Ul
^
o
1
CO
<
3
INDIVID
ro
3E CD
Si C -^
* >
o
3E
UJ +»
at
ca
+J
u
CD
u_
b£
1 O
* «
^ £_
O O
+J
(/) -O
3 O
<->
u-
o
CO
1
CO
(D 0)
o
o o
1. CO
o>
- 4J
O.S
CO L.
fU (D
ei z:
a. .
C 4->
(D
fl> c ^
OJ «r- ro
(/) (D 7T> "U QJ
i c g3 '££ '?, If I
3 = Q- CQECOQ. 5S"<
Jp ^*J
o -*
C3 O
O CM
v^>
CM in
CO O
»s
CM C3
5§
o o
o o
o
0 0
o
~
35
0°.
o
CM
o in in
* o in NO
ii *
in ** o
o ^*
CO
03 CM
O
^^
CM CM
°! °
^O C3
^^
«- m
OO N-
"~ O
O 1
UJ =E
_J UJ
UJ
8 1
E <
CO
_1
o
1
B-23
-------
UJ
o
1
1
^
8
to
1
i
Is
si
O CO
0. HI
CO
UJ
u
i
0
» 4-*
CO CO
0) 01
CO C
CD -
S ,i
C/J *^
(A
u
JT
g
CO
4-f
W O
ro en
sd~ ^
co5of\J«-o'5>cou-i
nj O^
sQ 0 0 0 0 -
D) 5"
E " S
CJ
i o "^-ooo-^'TT
. co o f^ ro
v CJ . D) *
U> 0 C
S- M
* 0-
^
CM ^ ro oo
00 O O
o o o o
o> C o oj £J rj
o >o c\j o r*- CM
o o o o «;";
° <3 0 S ^
CM C>
ii
CD O
o>
" « .8 o
co en <- T; i^
c ^ 2 " S
o £ E iS" ^ -g TJ
c »- o ro (U > c
*-4-*H- o c-»-c c ro
_*4) .^ O+JtU ^'^t _
O^^i E h4«^-N 4-»CO ^
COL. ' S
0 O
o *^
U CJ
»- en co
c c -S i-
CO O J O
i- o ro i-
(U t_ CJ CU
C U 0 D) C
.- CD 5. ro -
4- J= CO 30
m 4-> >
-------
11
UJ
o
t
z
1
=3
O
Z
UJ
i
H- z:
O in
a. ui
CD
UJ
CD
z
<
a:
CD
CD CA
T3 t-
3 O
*-*
cn eo
CD O
D) C
CD
2 U
CD C
cn -*
en
(D
o
.c
§
CO
Southeasi
Chicago
(0
Q- O
UJ
uj o:
CJ O
BUJ
f
en <
o
roooooro o o o
*"~ *"~
i * _
O O T Si" >O
10 i o ro o * co
^~ o in o *
o o o
o
co in
ro
ro
o
o . CD
-Q ^ C\J ^2 ** ^O
co c\j T- ro
D) CD CD TO O O.
- O CD O O
D) . o D) O O
CD. 0^0^
i/^ CO f*- i/n NO ^^ **^ CNJ 1^"
coo CDO oo T-O oo
oo o
oo ««oo «ooo
x* O O *-^ O ^M* ^«^
\f\ ^N
Kl CO
o o
O 0
0 0
O 0
0) O
C D)
cj IP- ro
*-"*-»(/) (fl
cn CD ra "D o
to .0 * o
*U 3(DCJO>.r-U t-
** J ' 73 M- OJ t- 4-» '^-C O
E C C 4J C O OO CO
co « «- £_ v) 4-J QL 4^ ti (^
O £--MT3-f-»CO tOTOQ Q>
O CU O DOJCOQ) C-*-»CO t_
_
ro «-
^
1 1
CM *
in co
o
o o
0 O
o o
o o in ru
r^- xf
co tn
o «- «-
^ N-,
T m
o o
ro oj in o
f\l OJ CO
o
o *- o
in eg ^N
ro o oj ;r-
« o N-
CO T
O CO
o t
1 UJ
UJ _
<
CO
1
B-25
-------
35
o*. -
^
w
o
u
I
to
ca
1
f
ca
cu
jc
C
"S
I
ca
II
"D **-
£ o
ca _
CT) CU
CD JC
S
8
io'E
u cu
!*
CO C
o o
o
Ol O
.r- CO
3 4J
U I
e .
§u
ragc
88-
."(U£-
3 C. CO
4J CO
CO 4-*
C O CO
C CD
H- O)
O 13 C
* CO
0) 3 t-
4-1 O
CO U J=
J« 3
4-» 4-» CO
SS-o
CO »^" CO
.8.1
CA 4->
CD CO CO
CO CO CO
4-1 CU CO
CO O X
:.£§
CA
t-.
CO
CO «- O
1.1
"co o
C *- CO
e « S-p
.I-55!
4J "E CU 'cl
a gS-S
§
o - o
CB O) 4-> 3
J= CU CO
4-» 4J C CO
CO CU
H- O S- D)
O O C
CU CO
ca o cu t-
^
C. 4J O CO
§Q- CU
4->
go o co
a. c cu
ca co
cu jc f- *>
0 x
cu cu
4-> C
CO CU
ca co
o cu
u c-
o v>
V «
CO
CO
u
cu
U
U 2
2 a "3a£ gS
4-> O
cu g
c o
3 cu
O CU 4J
cu
CO _> 0
de rived
CU
cu
£
I
I
B-26
-------
study. Except for the Mobile Source study, all of the ambient
concentrations reported in the other studies were within the range of
concentrations for urban areas in the Ambient Air Quality study.
For example, the concentrations in the South Coast study were between
7.9 and 15.4 /^g/m3; in the Southeast Chicago study, between 3.6 and
5.1 /ig/m3; and in the lEMP-Philadelphia study, 6 /tg/m3. The ambient
data in the lEMP-Baltimore study was considered marginal in that study.
The Mobile Source study used "old" national average ambient measured
concentration to estimate the mobile source contribution to benzene
risk. This method has been updated to reflect the "new" national
average ambient concentrations in the Ambient Air Quality study.
For the national estimate for cancer risk from ambient measured
concentrations of benzene, the annual cancer incidence estimated in the
Ambient Air Quality study would represent the best estimate. The other
studies illustrate the geographic variation that can occur and by
themselves are not the best estimates from which to extrapolate
nationwide risk from benzene.
Modeled Estimates. Except for the Ambient Air Quality study and
the lEMP-Baltimore study, all of the studies estimated cancer risk from
modeled ambient concentrations of benzene. Approximately twenty source
categories were identified as benzene emission sources. About one-half
of the source categories were overlapping between the studies. Of
these, only the motor vehicle and the iron and steel source categories
appear to potentially contribute more than 10 cancer cases per year. Of
the other source categories, industrial solvent coatings in the IEMP-
Santa Clara study and the unspecified stationary source category in the
35-County and the South Coast studies have annual cancer incidences per
B-27
-------
million population that would result in about 10 or more annual cancer
cases nationwide if applied to the total U.S. population.
The cancer rate from stationary sources ranges from approximately
0.15 to 0.28 cancer, cases per year per million population, excluding the
35-County's and the South Coast's unspecified stationary source
category. The South Coast's unspecified stationary source annual cancer
incidence per million population is between 25 and 130 percent higher,
being 0.35 cancer cases per year per million population. This is
consistent with the higher measured ambient concentration in the South
Coast study of 12 /tg/m3, which is approximately 44 percent higher than
the national urban average of 8.35 /tg/m3 found in the Ambient Air
Quality study. Thus, the South Coast study's estimate is probably a
very geographic-specific annual cancer incidence per million population
that one can not reasonably use to extrapolate nationwide risks. The
35-County's unspecified stationary source annual cancer incidence per
million population is close to the lower end of this range (0.13 vs.
0.15 cancer cases per year per million population). It is known that
this source category in the 35-County study contains iron and steel
sources (coke ovens), which can contribute a significant portion of this
risk from benzene emissions from stationary sources. This is
illustrated by the results of the Southeast Chicago study, where iron
and steel sources contributed to approximately one-half (0.08) of the
total annual -cancer incidence per million population in that study.
Lacking more specific information on the specific stationary sources,
this source category in the 35-County study was considered duplicative
of the stationary sources in the other studies.
Of the individual source categories with "overlapping" estimates*
only the motor vehicle category will be discussed in detail. As seen in
B-28
-------
Table B-4, seven studies estimated cancer risk from motor vehicles.
Five of these seven studies have very similar cancer rates, ranging
between 0.32 and 0.43 cancer cases per year per million population. The
lEMP-Santa Clara study's model was identified in that study as
underestimating benzene emission levels by 2 to 3 times. Increasing
modeled emission levels two to three times would increase the cancer
rate in the IEMP:Santa Clara study to between 0.25 and 0.38 cancer cases
per year per million population. This is certainly in line with the
other studies. The Southeast Chicago study also noted that its modeled
estimates appeared to underestimate measured ambient concentrations by
two to three times. Using the measured ambient concentrations increases
the Southeast Chicago estimated cancer rate to between 0.14 to 0.21
cancer cases per year per million population. While this is below the
other annual cancer incidence per million populations, it is consistent
with the lower measured ambient concentrations in the Southeast Chicago
study, which were between 3.6 and 5.1 ^g/m3. Thus, what we are seeing
are differences in the modeling techniques as well as geographic
variations.
Point Estimate. The best estimate from modeled concentrations
appears to be about 143 cancer cases per year nationwide (0.6 cancer
cases per year per million population) and about 181 cancer cases per
year nationwide (or approximately 0.75 cancer cases per million
population) from ambient measured concentrations. As noted above, two
of the studies discussed how the models underestimated benzene
concentrations (by a factor of 2 to 3). These underestimations could be
simply due to incomplete emission inventories in those studies and the
narrower underestimation (0.6 vs. 0.75) in the present study due to a
more complete accounting created by examining more studies and source
B-29
-------
categories. Based on these considerations, the ambient-based estimate
of 181 cancer cases per year (based on 0.75 cancer cases per year per
million population) was selected as the estimate of nationwide annual
cancer incidence due to exposure to benzene.
B-30
-------
1,3-Butadiene.. Eight studies examined 1,3-butadiene (see Table B-5).
Emission sources of 1,3-butadiene include point sources, primarily
synthetic rubber producers, and area sources (e.g., motor vehicles).
One study (the 35-County study) identified motor vehicle emissions of
1,3-butadiene occurring from both exhaust and tire wear. Seven studies
used modeled estimates to calculate cancer risk and one study, the
Ambient Air Quality study, used ambient measurements.
Modeled Estimates. Five of the seven studies included point
sources. Of these five, the NESHAP/ATERIS data base, the 5-City study,
and the TSDF study identified the specific types of sources; the 35-
County and the lEMP-Kanawha Valley studies did not. Thus, an effort was
made to determine whether any of the sources in the NESHAP/ATERIS data
base were included in the lEMP-Kanawha Valley and the 35-County studies.
Based on information in the lEMP-Kanawha Valley study report, the point
source facility is located in Institute, West Virginia. The
. NESHAP/ATERIS data base did not include a facility in Institute, WV,
although one in Washington, WV, was included. Thus, it appears that the
cancer risk from chemical manufacturing in the lEMP-Kanawha Valley is in
addition to that from the NESHAP/ATERIS data base.
In contrast, a comparison of city locations in the NESHAP/ATERIS
data base with the counties in the 35-County study showed an overlap of
geographic areas. For the 35-County study, it is likely that the two
risk estimates are not mutually exclusive, although to what extent there
is an overlap has not been determined. When the annual cancer cases
from the NESHAP/ATERIS data base is divided by the exposed population,
the cancer rates from both studies are the same - approximately 0.3
cancer cases per year per million population. This strongly suggests a
likelihood of much overlapping between these two studies. In the case
B-31
-------
UI
g
Ul
u
g
g
l/\ M
i «
CO r-
Ul £
UJ
a
*
1
I
o
z
JD
UI
0. CO
UJ
(0
1
Southeast
Chicago
(0 >.
1 J= 0
a. a -
ul c"ra
M
CJ
I
U.
ill
£ "*" S
co
ii_
Mobile
Sources
NESHAP/
ATERIS
w >.
g ,_£
!5 < "ra
< 0
111 0£
U O
a: u
u
3 o ^ 0
ru in oj o o ^ ! .
, o «- o °
? o ' -^ , a
Sin CM o o «- -* ou
0 ^ *" "" 0^0
D) «
O O
4-
K?
*- *"7 O «-
w O *~*
O *«^
0
CD »
ro co r*
V *o ro o
^ ^ oo
PJ O ^^
t^.
~°m °1 ~
|I *§
OJ O
o o
OJ O
o o
PJ
o o
4-
CM *^
£ OJ PJ OJ
inoj ojin «_S oo
o o ^ "^
Q) 03 D) C/)
C C C . (T3 (-(O^i ^2 C «
*j^ o ' roo >»4JjQ * in
O 0
OJ 0
O 0
** OJ
-3- 0
OJ
o
* OJ
OJ
Q 1
UJ =Z
_l UJ
Ul i
8 CD
i
I
h-
o
1
B-32
-------
1 CO
ft'
CD co
T3 rtl
..- CU
> .C
4~»
o ^
>x
CD -Q
L.
CD TJ
3 CD
TJ
CO «
SJ >
to
ro TJ
o
o
£_ I_
CD C
CO C
TJ CD
a£
C
o
*- CO
4_i CU
CD Q.
C O
<- CO
u- U
.4
g 2
O)
4-> O
CD CD
D)
CL I
O CD
E~CO
E
- to
'5
CD CO
.2!
S"B
CO 4->
ro to
u
t_.
<&" £
u
c
ro
CD O
c C
ro co
c. t_
ro CD
JC
CO 4->
CD O
to
CD CO
JC. C/3
4-1
££
4-> CD
. _c
4-»
D>
i°
0.1
4-t
E 4->
"g "g
CD CD
>. CD
I- D)
0 C
O) CO
CD £-
4J
CO CD
U JC
CD
CJ M-
I- O
g-a
CO C
CD
CD
M-
01
to jc
CD 4J
4J>
CO M-
I- 0
CD c
c .°
0 4J
TJ CO
'" 1
C 3
.- co
4-> CD
CO JC
CD 4->
r-
D> CO
JC "
C
C 3
CD '
O
4J CJ
CO
CD CO
3E
O JC
S 0
C CO
*~ 4J
xy O
CO 4->
4-<
-Q
CD
CO CO
C
3 O
O
"co --^
O -g
c_
CD
3 C
§""
C
o
O 4J
U CO
£1 .
4-» Q. CD"
O)
c c
- tv> to
to :zt
£_ CD
CD ^O _n*
.jQ CO 4-»
.
3 to
C JO CD
C CD O>
CO C C
o to
CD CO t-
TJ CO
CD JC
S'-S
O CD CO
ja
4-> TJ
CO 4J1 C
C 0 CD
c
°3"g
t» 3 4J
4-* O ffl
. (D '4!!
4J 4-» CO
CO CO CD
CD E
CO
4J> 4-> CD
C CO- jt
._ tD
O O
Q. 4-* O
C C
10 CD
.8.2
CO U
CD ro. c
c -""
ro co c.
4-* CU CD
S 2 £TJ
ro ro CD
4-> O >
4-> CO
CO C ' t-
I'" 3-S
c i E CD
- O CO C-
CA CD
CD jt
« 5 iH a,
CD jt CD
CO CZ O)
X o c
C ro
ro o 4-» c
c to
ro 4J c XI
ro c
c c M- to
o o
4-> "E CD CD
3 O JQ CO
1 it-'i-S
O Q CO
JC CD
4-f 4-< CO 4J
CO CO C
M- 0 -^
O CD O
CD O) Q.
CO 0 C
4-> C. CD CD
3 t- CO
3 O CD
CO CO CO JC
CD -» 4J
"t. g 3
CD CD t_ O
JC 4J c- JC
4J 3 tO
C CO
4J CD
C O CO CO
CD Q. CO CO
CO 3
t- U CO CO
Q. co CD
CD O "O
I CO 4->
5 o 4J x
' M- CO CD
O C 4J
CD CD
CO > CD JC
- -p- CO 1
JC C- CD
c**c
- to t- E
ro o 3
to 3 u
t_ o
CD -G
(DOC.
J= C CD C
ro
1
ro
in
in'
CD
£_
ro
£_
8.
CO
CD
to
ro
o
i_
CD
U
C
ro
CJ
R
2
06
g
CO
I
8.
c
1
3
C
O
_J
E
1
CO
CD
CO
to
U
t-
o ,
u
c
CO
u
c
* 4J
to
I- 1
CO CL
CD
CD
t_ t_
8.H
CO C
CD O
CO
ro '
0
fc1
2 0
co CL
CJ
CO
r- CD
* CO
CM CD
CJ
.. t_
C CD
jS C
c- ro
r> o
CO O c-
r- «Ji ID
1 4J CO
CD CD
JC CO
4J tO
CD U
r CD t
E 4-» ro
o to D
»*- >. c
o 4-» ro
o o co
ro m
vt- po *
CD
C CO
CD -^ 4J
C TJ C
CD "D CD
-5 ra B
to c to
i * to
ro *-*
- O CD
-.T. S
CD E J=
O> --x. 4-»
ro a>
£_ N-T C
CU .E
ro ' 3
$
.C CD «-
O) CL
.,- o
ro
ro t- cu
D) 4-»
c ro
0 CO E
TJ 0 *X
CD O O
CO l~
o ro
to 4->
CD C
woo
CD CO
O CD CO
CD tO
£_ . .r-
CU CD E
o cD
C -4-
CO O CD
° C.S
' O >
ro 4J - ro
D O TJ M
C CO CM
C s- TJ
ro » o
C D 4J
CO O O
C\J - 3 CM
c
TJ E O >-
4-* CU CD
ro c TJ !-»
CO CJ O X
CD CD »- O
' 4J t_
XM- ro CL
"D CD c CL
3 c_ ro
+-* to
tO O C CD
4-» O .Q
X
4-« TJ CO O
C Q) CO 4-»
D -M *-
O to E TJ
O ^00
1 > 4--
in TJ 4-» ro
ro to to E
CD CO i *-»
.G ro TJ to
i s ro CD
^
CD
c
CO
u
to
3
C
c
CD
r^
r^
"S
4-»
E
4J
to
CD
Ti
H
CO
to
£_
O
4-»
O
ro
s-
to
to
g
o
1
"S
3
t
CO
CD*
e
CD
CO
O
CD
O
of
jj
O ro
§__.
0 g.
o a
M- C
0 O
L. ^
o *
81
s- i_
CU
c a.
o
.,- t_
co ro
to CD
X
CD
vQ O
CO
2: S
CO
ro co
u
4-1
(J t_
CO
TJ ro
ro co
I t 3
O CD O
Q. 0 M-
o ro CD
M O -C
ro ro
o
o ~- CD" t-
CD -Q O
' ro c
co o 4-» ro
m o
^.f r- C_
. CU XD '
CD > jc ro
4-» ZJ
to TJ O C
ro ro c c
o ro ro
H- CO E CD
- ro O
4-» a: u CD
C H
CD 4-» TJ
TJ t- (0 CD
"- O C 4-*
t_ C. '£ <~-
QJ ro 4-»
3E O t. «
JC 4-* CD
CD 4J C
S o
M- CU -
CU IS 4-»
' to o o ro
CD i_ ar
CO 3 ZJ
to o o
CJ CO *
co ro
C. CD CD O
CD C -»
u cu u to
ro TJ ^ ^
u ro cu i
4-> >
CD -Cl TJ
is i ro TJ
c ro o cu
C - t- 4J
co r- ro
o
« L 4-* ZJ
ro CD o
4-< x: CD
O 4-» ZI tO
t o -a o
CD
*E
^
D>
ro
o
X
o
CD
4-*
X
o
t
&
CD
O
o
o
to
g
CO
CO
CO
4-*
'c
c
ro
TJ
CD
CO
f
4-»
CO
O
D)
ro
u
-G
o
4-»
CO
ro
cu
ZJ
o
CD
4->
ro
£_
CD
TJ
o
C
CO
g
CJ
M- C
r- CD
CJ T!
8.-0
CO C
CD "~
JZ. CD
4-> TJ
H- '5
O C
o
ro
4J C
ro
4-*
C 4-»
CD ro
to E
s *
TJ O
ZJ C
(/)
Is
.. CD
ro -Q
ro
ro
to
c o
CO 4-*
^
i"8
s: E
UJ D
i-t CO
co
CD CO
r-
4-» tO
t- >-
^*~ ^
CD -M
4-» CO
ro
t- >s
o u
C i
cu in
.r-
-------
of the 5-City study, all of the estimated stationary source cancer cases
were from two facilities in one city. These two facilities are included
in the NESHAP/ATERIS data base. Thus, the cancer estimates of the
5-City study and the NESHAP/ATERIS data base are duplicative of each
other.
Based on the above, point sources appear to result in between 24
and 38 cancer cases per year, depending on the overlap between the
35-County study and the NESHAP/ATERIS data base. It seems likely that a
more detailed comparison would show the estimate closer to 24 annual
cancer cases than to 38.
Five studies used modeled concentrations to estimate risk from
motor vehicles. In order to compare the risk estimates and annual
cancer incidences per million population among the studies, a consistent
emission factor was applied to four of the five studies. (The IEMP-
Kanawha Valley study was not included because the study did not identify
the emission factor used.) The emission factor selected to put the risk
estimates on a more common basis was the estimated 1986 emission factor
of 0.0089 to 0.0098 grams per mile (g/mile), which was taken from the
latest work by the Office of Mobile Sources.4 The range reflects the
presence and absence of an inspection/maintenance program, respectively.
After adjusting to a common emission factor (0.0089 to 0.0098 g/mile), a
cancer rate range of 0.25 to 1.02 cancer cases per year per million
population is created. Each of the four studies used a different model
to estimate risk. The modified CO NEM model used in the Mobile Source
study appears to generate higher risk estimates than the model used in
4 Carey, Penny M. and Joseph Somers. Air Toxics Emissions from Motor
Vehicles. Paper presented at 81st Meeting of APCA. Dallas, TX. June 19-
24, 1988.
B-34
-------An error occurred while trying to OCR this image.
-------
population. The Ambient Air Quality study was based on data from
California. Of the NESHAP/ATERIS point sources, two are located in
California, but not in cities that provided 1,3-butadiene ambient
measurements. It is not unreasonable to expect ambient measurements to -
be higher in cities with point sources. Thus, it is not unreasonable in
this instance that the ambient data and the modeled point source data
are mutually exclusive. On this basis, cancer incidence is estimated to
be 266 annual cancer cases nationwide (22 from point sources (including
TSDFs) and 244 from motor vehicles/ambient measured data) due to
exposure to 1,3-butadiene.
B-36
-------An error occurred while trying to OCR this image.
-------
1
lu
g
a
i
S
I
§"*
^
Ul u-
CO CO
< Ul
- g
_!
s
^
Q
I
CO
Ul
s
&
o
1
i
.c o
t- o
u
r~ O
o «
a, §
** .3
ea o
«
o
O O 4-*
fOS (0
"Q C.
c
low-
a.«|
)5 g W
^ B
31 o
CO
01 CJ
11
x co
Hazardous
Waste
Combustors
is
TJ *J
c w
4
§ u
u
NESHAP/
ATERIS
** >*
Is!
SOURCE
CATEGORY
^% *O K) CM CO -st »*
fo O* O^ -
-------
CJ
i
UJ
o
z
0
I
z
^
C/>
J
^
CD,
1
.d
UJ
1
2S
o £
a. en
UJ
CD
a
2
VI
0
u
J=
g
CO
4J
CA O
CO CD
OJ to
-C O
D .1=
o cj
CO
CO
a. o
UJ CO
""" c
CD
CD
0- 3 0)
£ CO
Ul C
CO CO
0)
t-
cx E
UJ 4-»
CO
CD
4-1
O
u.
UJ ***
o o
Q£ C3
j UJ
«S
CM **2 .J- *-
' fO >* CM O* ff\ O
CM >O CM N- *o « 1 fOOO
"OCOV-OQ CMi. . !
O O O O O O
,
-*.CDCM JOCM -ocoro
oiv »P-oo Oiroroo
"ocMCD^S^ o'e:!£iocD' '
»- o o o
CD O O CD
0 0
O «»
~* O X
CD CD
*- o « m «- ro
«- CM O 0 0 0
O O X O O O O
i O CD CD CD
O 0 ....
^ 0 O O O
X X
* * o o in in
CM C\J O O O O
O O CD CD CD CD
o o o o
.0 -0
o ^ o o o >-*
-
m o* fo CM
v* o ro o
«- 0 O 0
o o ° °
*-* 0 O
0)
* 4-» Cft
O O) O 01 Dl
0) (D D) O) C 4-»
4-»3 O*DT3I-"- C
C CO « C W a> 3 O £- 0) Q> (0
O ' t- ' O W t- JZ »»- » -M 3 C "U '
"*s, *»- "O TOOO) 13 O<1) M- W4-» C Q.4-* I- O4-» O4J> '^« 1- O X
C trt (0 -r- « Q) W TJ « OQ>a)CJCD4-'Oe04-*
3 O34^» a; 5 t_ dj 3 L. a> o en c M- CD *^c
4-»£ C «^-£ » 4-«.5 (04-iXI O Q-t_ 30 3:30 s: zjw w^-. aco rsia. o
i O
%o «-
xf £-
0
OJ «
co NO in
O 00 CO %fr
O* O -J- CD
00 00
oj m
«- o
^ -
o
O v^
-
T- 0
0 CD
i i i
sj- ON
0
O
0
OJ OJ
CD CD
O
. 0
O v^
O
-j- oj
0 0
o o"
o o
CO OJ
-------
>> co
t7 co *-
CU Q. CO
en
cu J= -
c. u co
Q. CO CU
cu o
CO
41
8.
o
c.
Q.
£
g
8
CD O C.
J= C CU
t- - Tl
B-40
-------
TABLE B-7.
ESTIMATED ANNUAL CANCER CASES FROM
CADMIUM EMISSIONS FROM MOTOR VEHICLES
ITEM
Original Estimate
(Annual Incidence
per million
population)
1.9xlO'6 g/mile
Without tire wear
Mobile
Study
0.001a
0.001
0.001
STUDY
35 County
0.009 to
0.013b
0.004 to
0.006
0.0012 to
0.0017
--
South
Coast
0.0398C
0.011
0.011
Southeast
Chicago
0.0036d
0.001
0.001
Assumes cars with
catalytic conver-
ters have zero
cadmium emissions
0.001 <(0.0012 to
0.0017)
<0.0012£
0.001
Exhaust emissions only; 1.9xlO"6 grams/mile emission factor; based on
1.9x10 g/mile for non-catalytic equipped cars and 0 g/mile for cars
with catalytic converters.
Exhaust and tire wear emissions; 9.0xlO"6 g/mile exhaust emission
factor and 4.85xlO"6 g/mile factor for tire; no distinction as to
catalytic or non-catalytic equipped cars.
Exhaust emissions only; emission factor not given; emission rate
assumes all cadmium in gas (0.02 mg/1) is exhausted from both
catalytic and non-catalytic equipped cars. At 0.02 mg/1, an emission
factor of 6,6xlO"6 g/mile is calculated.
d Based on emission factor of 6.7xlO"7 grams/mile.
Assumes 12 percent of fleet is non-catalytic equipped (same
assumption as in Mobile Source Study).
B-41
-------
cases per million population is used for the best estimate. This is
equivalent to 0.2 to 9.6 cancer cases per year nationwide.
Cancer risk due to heating was found to have a fairly wide range in
annual cancer incidences per million population, from approximately
0.005 to almost 0.04 cancer cases per year per million population. This
is equivalent to approximately 1 to 9 cancer cases per year nationwide.
The 35-County study indicated that the emission estimates for cadmium
were based on species apportionment factors used in the earlier
35-County study. The 5-City study indicated species apportionment
factors revised since then were used. Using the 5-City and Coal and Oil
Combustion studies' estimates, a narrow range of 1.1 to 2.2 annual cases
per year nationwide is obtained. Insufficient information was available
to determine why the lEMP-Kanawha Valley had a higher annual cancer
incidence per million population. It is possible that the urban nature
of the 5-City study may contribute to the higher annual cancer incidence
per million population than the one from the Coal and Oil Combustion
study, but the range of 1.1 to 2.2 estimated cancer cases per year
nationwide is retained.
The NESHAP/ATERIS data base and the Southeast Chicago study both
estimated risk from iron and steel plants. The higher annual cancer
incidence per million population in the Southeast Chicago study could be
attributable to two factors. First, the Southeast Chicago study area
could have a concentration of iron and steel plants that results in a
higher annual cancer incidence per million population, whereas the lower
annual cancer incidence per million population of the NESHAP/ATERIS data
base reflects the spreading of the annual cancer cases over the entire
U.S. population. As noted elsewhere, the Southeast Chicago study
modeled four steel mills. Second, the Southeast Chicago study's
B-42
-------
inventory was designed to estimate actual emissions assuming full
utilization of existing steel production facilities. This is apparently
different from the U.S. EPA's study on steel mill emissions. The
Southeast Chicago study notes that the U.S. EPA's revaluation of steel
mill, emissions conducted a review of the operating status of major units
at each of the steel mills in the study area. The inventory as of July
1987 is based on particulate matter emissions estimates contained in the
National Emissions Data System (NEDS), which reflects sometimes outdated
judgments of which units are operating and which units may be considered
permanently shutdown. Thus, the Southeast Chicago study's inventory
would contain a higher level of estimated emissions than under the
NESHAP/ATERIS data base. For this source category, the total annual
cancer cases associated with the NESHAP/ATERIS data base (0.06 per year)
was selected as the best estimate of nationwide cancer risk.
The NESHAP/ATERIS data base and the Sewage Sludge Incinerator
study both estimated cancer risk from sewage sludge incinerators. Since
the Sewage Sludge Incinerator study is a more recent estimate, the
cancer risk estimated in it was selected as the best estimate of- cancer
risk from cadmium from sewage sludge incinerators.
Even though there are similar wide ranges of annual cancer
incidences per million population for some of the other source
categories, total annual cancer cases from the remaining six source
categories are estimated to most likely be less than three. Combining
the best modeled estimates from all of the source categories yields an
estimated 7 to 15 annual cancer cases nationwide (or between 0.029 and
0.063 cancer cases per year per million population).
The above estimate does not include the "point sources" category
of the 35-County study. It was not possible to identify what overlap
B-43
-------
there might be with the other specified point source categories.
However, it is known that some of the iron and steel plants are located
in cities in counties that are included in the 35-County study (such as
Chicago in Cook County, IL). Thus, it might be reasonable to compare
the total annual cancer incidence per million population of the 35-
County study with the upper end of the above range (which includes the
higher annual cancer incidence per million population for the heating
category from the 35-County study). When this is done, there is
somewhat better agreement (0.10 vs. 0.06 cancer cases per year per
million population). By adjusting the motor vehicle contribution from
the 35-County study as described earlier, the total annual cancer
incidence per million population for the 35-County study is lowered
marginally. Considering that the counties in the 35-County study were
selected, in part, on the basis of sources known to emit the pollutants
being studied, it is not necessarily inconsistent that the resulting
annual cancer incidence per million population is higher than the
aggregate total from the nationwide studies.
Ambient Estimates. As noted earlier, four studies used ambient-
measured "concentrations to estimate risk. The ambient-measured
concentrations of cadmium for the lEMP-Baltimore study were all below
the minimum detectable limits of the analytical technique. Thus, the
range of cancer cases reflects zero to the detection limit (between
0.001 and 0.002 ng/m3) concentrations. The lEMP-Santa Clara used the
1985 Ambient Air Quality Study's concentrations (0.001 to 0.003 Aig/m3)
to estimate cancer cases.
The updated Ambient Air Quality study used an annual average
concentration of approximately 0.0016 ^g/m3 to estimate cancer cases.
This estimate was based on data from 164 counties. The South Coast
B-44
-------
study report showed a concentration range of 0.0014 to 0.0018
Both studies result in approximately the same annual cancer incidence
per million population - 0.042 to 0.045 cancer cases per year per
million population.
Point Estimate. The annual cancer incidences estimated from
ambient-measured concentrations lie within the range derived from the
modeled concentrations - 10-11 vs. 2-44 cancer cases per year. The best
estimate of total nationwide cancer cases based on modeled concentration
is estimated to between 6 and 16 per year. A single point estimate of
10 cases per year was selected based on the ambient data of the Ambient
Air Quality study.
B-45
-------
Carbon Tetrachloride. Carbon tetrachloride was included as a pollutant
of concern in thirteen studies (see Table B-8). Carbon tetrachloride
sources are primarily point sources. At least eleven source categories
were considered in the studies. The lEMP-Philadelphia and the South
Coast studies incorporated both ambient measured concentrations and
modeled ambient concentrations. Since carbon tetrachloride remains in
the atmosphere long after it is emitted, ambient-measured concentrations
are more likely to result in better risk estimates than those estimates
based on modeled ambient concentrations. Therefore, the analysis only
focuses on the ambient-measured cancer risk estimates.
Several comprehensive studies6 have identified a global background
concentration for carbon tetrachloride of approximately 0.8 /*g/m3. As
carbon tetrachloride is difficult to measure and as there are no known
"sinks" for carbon tetrachloride, any ambient-measured concentration
much below this level must be viewed as due to test method error. This
information is important in assessing the cancer risk for carbon
tetrachloride as reported in the studies.
Ambient Estimates. Six studies estimated risk based on ambient-
measured concentrations. The Ambient Air Quality study was based on
data from 24 counties. The Ambient Air Quality study estimated a
cancer rate of 0.15 cancer cases per year per million population, based
6 P.G. Simmonds et- a]_. "The Atmospheric Lifetime Experiment 6-
Results for Carbon Tetrachloride Based on 3 Years Data." The Journal of
Geophysical Research. Vol. 88, No. CIS., pp. 8427-8441. October 20,
1983.
H.B. Singh, L.J. Falas, R.E. Stiles. "Selected Manmade Halogenated
Chemicals in the Air and Oceanic Environment." The Journal of Geophysical
Research. Vol. 88, No. C6, pp. 3675-3683. April 20, 1983.
H.B. Singh et. a]_. Toxic Chemicals in the Environment: A Program
of Field Measurements. EPA/600/3-86-047. August 1986.
B-46
-------
TABLE B-8
ESTIMATED ANNUAL CANCER CASES FROM CARBON TETRACHLORIDE BY SOURCE CATEGORY
SOURCE
CATEGORY
Pesticide
Production
Pharmaceutical
Production
Chemical Users
and Producers
Aerators
POTU's
TSDF'S
Unspecified
Indirectd
Impacts
A--ea
TOTALS
MODELED.
AMBIENT
INDIVIDUAL STUDY
Ambient NESHAP/
Air ATERIS
Quality
0.42
(0.0018)
0.00013
(0.0000005)
0.19
(0.0008)
13. 9e
(0.058)
14.5
(0.06)
36
(0.15)
Drinking Thirty- IEMP-
Uater POTU's TSDF's five Five City Baltimore
Aerators County
0.01 X
(0.00063)
<0.0047
(<0. 00002)
0.03 0.13 X
(0.00013) (0.0027)
2.28
<0.0095)
0.44
(0.028)
<0.0047 0.03 2.28 0.13 0.46
(0.00002) (0.00013) (0.0095) (0.0027) (0.029)
0.3
(0.196)
B-47
-------
TABLE B-8 -- concluded
ESTIMATED ANNUAL CANCER CASES FROM CARBON TETRACHLORIDE BY SOURCE CATEGORY
SOURCE
CATEGORY
Pesticide
Production
Pharmaceutical
Production
Chemical Users
and, Producers
Aerators
POTU's
TSOF'S
Unspecified
Indirect
licpacts
Area
TOTALS
MODELED
AMBIENT
INDIVIDUAL STUDY
IEHP- IEHP- IEHP- Southeast
Konawha Philadelphia Santa Clara Chicago South Coast
Valley
-.
X
0.0086-0.021
(0.086-0.214)
0.00014
(0.00036)
negligible
0.064
(0.164)
0.0086-0.021 0.035 . 0.00014 XJ.0014
(0.086-0.214) (0.02) (0.00036) (0.00013)
0.64 0.2 0.064 1.41
(0.39) (0.14) (0.164) ;(Q-13)
NATIONWIDE
POINT ,
RANGE3 ESTIMATE0
0.42 0.42
0.00013 0.00013
0.16-0.19° 0.19
O.0047 <0.0047
0.03-0.65 0.03
2.3 2.3
6.7 6.7
_ _ - -
13.9 13.9
23.5-24.2 24
31.2-47f 419
B-48
-------
Footnotes to Table B-8.
NOTE: Numbers in parentheses are annual cancer cases per million population. For nationwide
studies, annual cancer cases were divided by 240 million, unless otherwise noted. For
studies with smaller geographic scopes, the annual cancer cases were divided by the study's
population.
a The numbers in this column were calculated by taking the lowest and highest incidence rates for
a source category and multiplying it by 240 (1986 U.S. population in millions). The total for
this column is the summation of the low end of the range and the sum of the high end of the
range.
The numbers in this column present the results of the reduction analyses. In most instances, a
point estimate of nationwide annual cancer incidence was derived for each pollutant/source
category combination. In some instances, a point estimate could not be reasonably derived. For
these instances, as narrow a range as possible of nationwide annual cancer incidence was
estimated, and such ranges appear in this column. The text discusses how these point estimates
and ranges were derived.
c Does not include extrapolating the incidence rate from the lEMP-Kanawha Valley study to a
nationwide risk estimate because of the unknown type of facilities.
"Indirect impacts" refers to carbon tetrachloride emissions that have been emitted in the past.
As noted in the text, carbon tetrachloride persists in the atmosphere long after it has been
emitted. Since the Southeast Chicago study used ambient-measured concentrations to estimate
risk from this source category, the results are reported under the "Ambient" total and not'under
the "Modeled" totals.
e Incudes solvent applications and grain fumigation.
Due to the minimal number of data points associated with the lEHP-Philadelphia study, the range
does not include extrapolating nationwide cancer risk from the incidence rate for that study.
" As discussed in the text, this is based on a global background concentration of 0.8 /tg/m .
B-49
-------
on population-weighted ambient concentrations of 0.79 /^g/nv5 for urban
areas and 0.4 ^g/m3 for rural areas. The range of concentrations
identified in the study was from 0 to 1.87 /^g/m3. The other five
studies were for specific locales - Baltimore, Philadelphia, the South
Coast Air Basin, Santa Clara, and Southeast Chicago. Of these five
studies, the lEMP-Philadelphia study appears to be based on a single
data point. Ambient-measured concentrations were 1.8 ^g/m3 for the
lEMP-Philadelphia study. The Southeast Chicago study had two sets of
sample data for two sample sites. One set measured 0.44 //g/m3 over 10
to 15 samples and the second set measures 2.7 /^g/m3 for 5 to 7 samples.
The IEMP-Baltimore study had at least 10 monitoring sites, and the South
Coast study had five monitoring sites with a combined total of over 100
samples. The 10 sites in the lEMP-Baltimore study showed a range of
average concentrations from 0.6 to 1.4 //g/m3 with a population weighted
average of 0.9 /^g/m3. The South Coast study showed annual (1985)
average concentrations ranging from 0.6 to 0.76 fjg/m3 for the five sites
and a population weighted average of 0.69 //g/m3. The Santa Clara study
reported monitored concentrations ranging from 0.2 to 1.2 ^g/m from a
single monitoring site.
As seen above, the various studies present a range of carbon
tetrachloride concentrations. A number of reported ambient-measured
concentrations were substantially below the expected background
concentration of 0.8 ^g/m3. Such instances are likely due to sampling
error. Based on a concentration of 0.8 //g/m3, a nationwide estimate of
approximately 41 cancer cases per year (0.17 cancer cases per year per
million population) is obtained. This result is very close to that
estimated by four of the five studies. Only the lEMP-Philadelphia study
shows a substantial deviation, with an estimated cancer rate of 0.39
B-50
-------
cancer cases per year per million population. This could be due to
sampling error, the small number of data points, geographical variation,
or a combination of any of these factors.
Point Estimate. It appears that the best estimate of nationwide
cancer risk from carbon tetrachloride is based on applying the
background concentration to the total U.S. population, which results in
an estimated cancer risk of 41 cancer cases per year. The studies
indicate that there can be locally high levels of concentrations to
which populations are exposed. This would increase the estimate of
nationwide cancer risk based only on the background concentration of
0.8 pg/m3. The magnitude of this potential increase, however, is
unknown. Thus, nationwide cancer risk from carbon tetrachloride is
estimated to be 41 cancer cases per year.
B-51
-------
Chloroform. Thirteen studies included chloroform as a pollutant of
concern (see Table B-9). Only a few specific source categories were
identified as chloroform emission sources in these thirteen studies.
One study, the South Coast study, found that ambient measured
concentrations of chloroform were much higher than the modeled ambient
concentrations, and suggested that this might be due to sources not
included in the emission inventory. Thus, as was for carbon
tetrachloride, risk estimates based on ambient measurements may yield
better estimates. In addition, the non-specificity of a, number of the
studies as to the specific source categories examined made it difficult
to sum across the estimates based on modeling.
Ambient Estimates. Three studies used ambient monitoring data to
estimate cancer risk - the lEMP-Baltimore study, the South Coast study,
and the Ambient Air Quality study. The lEMP-Baltimore study showed an
average cancer rate of 0.29 cancer cases per year per million
population, with a range for individual geographic areas within the city
from 0.07 to 1.54 cancer cases per year per million population. Ambient
measured concentrations ranged from 0.2 to 4.7 ng/m3, with an average
concentration of 1.7 ^g/m3 over the ten monitoring sites. The
population weighted average, however, was approximately one-half of
it-
that, being 0.88 ^g/m3.
In the South Coast study, ambient concentrations from five
monitoring sites ranged from 0.27 to 0.55 /ig/m3, for a population-
weighted annual average of approximately 0.38 /tg/m3. This concentration
is slightly less than one-half of Baltimore's population-weighted
average concentration of 0.88 ^g/m3. The resulting cancer rate in the
South Coast study was approximately 0.12 cancer cases per year per
million population.
B-52
-------
CD X
J= o
0) C
QL CO
O *-
I i
CO CO
in in
O
dS
N. »-
T- O
O O
S(XJ
«
o o
CO
SI-
CO O
Ox 0
ro o
d°.
0
t^ xQ CO
-o o xo eo «- o
v- O O O (M O
O O 1 O
O CM O
o o o
SE? SL
r- ,- CO
S3 -n
o o c
3D.
roc--
"D
rao
-
3O
4J
«(D
t-
(Ud)
OC
(0
30
-------
g
g
o.
I
m
UJ
I
s
s
3
1
|
«J
§
2
s
f|
J3
&i
o c/>
O. UJ
"u,
i
2
u
.£=
g
cn
M
tt O
01 tO
-C U
o o
«
g
o
o. 3
5 ro
" C
0
M
O
i!
tt t *u
*M 0}
JE
a.
tn (^
r ^
* ~ '
o OJ o T- o o i «-
i i i i CM
NOCNJPJOO °o"~
O
D> CT O ^
£ c c
OO LA «~ O
o o o o
00 00
c- M "r;
82S.Q.-2 8» 2
w 't! £ 30 "§0 "8 ^3
o o C ra 5 a t. -jj _c _«z
§3OL 'i ^" * n " 3 Q. OP
m C ^-* O T3 O 3 O 1 W ^ *^
5 ro 3 ^C«Sc O C .El-
0.3:0. cjroh-c/)»-« O- O CJQ
in
in
in
o «-
i ' i
CJ ON
(M
'"*
n~
"~ C3
W
O> N-
e\j o
0 O
o
oS
T- iQ
0 O
-------
(3\
en
eu
~Q
eo
t
o
CO
eu
4-1
o
c
*t
CD to
r- tU
> JZ
5 **
o ja
i_
STJ
CU
eo .-
to -~
cox.
0
t_ e_
eu eu
O 3t
co en
o cu
to
CO
eo o
3
Et_
CU
CD O
C
- CD
CO U
CU
73 CD
3 3
4-> C
eo c
CO
0
a cu
"- JC
3 4->
c
o -
to
4-» O
eo a.
C 0
o
c- to
o
U- O
.-g.
to CD
3 ro
S..2
O E
CO
-C
"E .
2
£_
tO
CL 01
CO "D
eo CD
eu jc
o
eo
"2 E
CO 3
4_f o
So
3 CO
O
O
4-> O
O>
^ 4^
CO O
£J
03 ^N
CO
35
eo -^
O '
eu 'i
eu
3 C
|s
O 4J
O CD
CO 3
o_
JC Q
4-> a. cu
- CO CO
to 3
t- CU
c ^ o
<0 O Tg
* «\J CU
C CD
CO CU
0 1
CD
CO X
3 ' CO
c J2 eu
CO C C
O CO
a> co c_
a co
~ CD JZ
§1- CJ
3
J5
4-» T3
eo 4-> c
C O CO
c
M- <
4-> O CO
eo o E
E -
^ CU 4->
4-» 4J tO
to eg eu
CO
4J 4-» eo
C CO 3
. CU
o cu
eo -^ eu
to o
CU CD C
CD to e-
4-» CU CU
eo o u
C C C T3
co co eu
4J O >
4-» tO "-
CO C t_
§ CO CU
3 73
cu c
c § c eu
i O CO t-
to cu
eu 3
C T3
S to
3 CU
eo co
>. 0 C
C - CO
eo o 4-» t_
C CO
CO 4J C T)
CO C
C C -4- CO
o o
4-» "E CD CU
O O 4J
3 O J3 CO
S >*'«>
t- £- tO 4-»
'O O CO
C. CO CU
' 3 c_ CO
3 o eu
co eo CD jc
eu *x 4->
e- 4-> 3
C 0 3
CD CO c~ O
JC 4-» t_ JZ
4-" 3 CO
c to
4-1 ' CU
So to to
a. CD co
eu JZ o
c- o to co
a. co cu
CD O "D
Jt CO 4-*
O 4-> X
H- tO CU
O C 4J
0 T3 .-
CU CD
co > CD JZ
r- .r- CO »
JZ t- CU
4-» CU JC
co t_ E
CD O 3
eo 3 u
e- o
CD CD O
"g c ~a co
C "D > JC
CD O t-
h - "U
JZ
o eo
3CO
03 <
O O£
CO ^
a> a.
4-* «£ C-
m 0= 0)
.- UJ 4J
4-» Z CD
CO O)
O CD O
.C 4-»
CO O
»- H- 2
t- 0 4J
O -M
C 4-»
CD 03 CO
u u E
CD ' CO
~V Q.
o to
c to co
O £-
4-J 4-J
CD CO
>^.C CD
4-» tO 0)
CO CD -
e t-
O CO CO
' CO «-
* CD O£
03 LU
> CD 1-
03 C *-s.
CO =C
C H- CO
03 O LU
^ z
Q_ C
31 CU CD
LU .jC
.c "^ *o
*J 4-*
O CU CD
( ' "O
M- cj C
8 i o
CO U) C
C- C -
CU CD tO
O <"* «^
cu eo ^>£
TJ CD eo
O 4-» t_
C ^ to
cu o §
jz eo 3
4-* H- tO
eo
O) CO
C H-
0 CD
CO 0 I?
QL eo
o x i-
Q.4J
CD *4-
t- O O
4-1 JZ
X 4J 73
cu c
o cu
-S~j=
3 (0 D)
CD -
« 4-*
C 4-» m
CO CD
SO 4-»
E to
o 3 oj
c.
4-*
o
g
M-
co
H-
o
CD
4J
CD
o
I
CO
t
to
^-
o
ir>
eu
c
0)
g
4-1
eo
CD
eo
1_
cu
JZ
CD
^
ej>
2
to
"-
to
4-*
CD
to
3
CD
1
c
CD
en eo
CD
to -
"~ 0
0 S?
O CO
u
0
en cu
C 0
CO C-
t- 3
o
~O CO
CU TJ
£- -r-
o o
- 8.
< to
B-55
-------
The Ambient Air Quality study used ambient measurement data from 22
geographic areas, the majority of which were located in California. The
annual average ambient concentrations ranged from 0 to 9.3 ^g/m3. The
California data has a much narrower range (0.13 to 1.81 ng/m3), and a
much lower average concentration than the non-California data. Using a
population weighted urban concentration of 1.86 ng/m3 and a rural
concentration of 0.1 jig/m3, the Ambient Air Quality study estimated
total annual cancer cases at 115, or equivalently 0.48 cancer cases per
year per million population. Of the three studies using ambient-
measured concentrations, the Ambient Air Quality study was selected as
providing the best estimate of nationwide annual incidence because of
its broader geographic coverage.
Modeled Estimates. Excluding the "chemical manufacturing" source
category in the lEMP-Kanawha Valley study for the moment, the modeled
estimates of cancer risk from chloroform are estimated to be between 4
and 10 cancer cases per year nationwide. (This supports the South Coast
study's finding that models may be "missing" chloroform emission sources
when compared to the cancer risk estimates based on ambient-measured
concentrations.) The "chemical manufacturing" source category in the
lEMP-Kanawha Valley study has a very high cancer rate - 0.58 to 1.9
cancer cases per year per million population. However, it does not
appear reasonable at this time to try to extrapolate nationwide cancer
risks from this annual cancer incidence per million population because
the types of facilities and their products that lead to these chloroform
emissions have not been identified. Thus, the representativeness of the
emission sources is unknown.
Point Estimate. As noted above, ambient measurements appear to
provide a more complete accounting of chloroform concentrations then do
B-56
-------
modeled estimates. The cancer risks based on ambient-measured
concentrations, therefore, were selected as being more likely
representative of actual risks. Of the three studies that estimated
cancer risk from ambient-measured concentrations, the Ambient Air
Quality study used data with a broad geographic coverage, including data
from areas covered by the two other studies. Therefore, the estimate of
115 cancer cases per year from the Ambient Air Quality study was
selected as estimate of nationwide cancer risk from exposure to
chloroform emissions.
B-57
-------
Chromium. Thirteen studies included chromium as a pollutant of concern
(see Table B-10 and B-ll). Ten of the studies estimated cancer risk
based on modeled ambient concentrations; four using ambient measured
concentrations. (The South Coast study used both modeled and monitored
ambient concentrations to estimate risk.) Of the fourteen plus source
categories identified, two are of primary concern - chrome platers and
cooling towers - for estimating cancer risk. For both modeled and
ambient measured concentrations, the percent of total chromium assumed
to be hexavalent is also important. For some source categories, such as
chrome platers, nearly 100 percent of total chromium emissions are
hexavalent; while for some other source categories, such as
incinerators, less than 1 percent of total chromium emissions are
hexavalent.
Ambient Estimates. Assuming 100 percent of the measured ambient
concentrations are hexavalent, the four studies resulted in estimated
cancer rates between 0.82 and 2.77 cancer cases per year per million
population (see Table B-10). On a nationwide basis, this is equal to
approximately 197 to 665 cancer cases per year.
Results from the 5-City study suggest that the ratio of hexavalent
to total chromium concentrations in model-predicted ambient levels range
from 0.085 to 0.815. Applying the appropriate ratios to the annual
cancer incidences per million population in Table B-10 to the IEMP-
Baltimore study and the South Coast study yields very similar cancer
rates - 0.8 vs. 0.67 cancer cases per year per million population. For
the five cities, an arithmetic average ratio of 0.4 and a population-
weighted average of 0.6 for hexavalent-to-total chromium emissions were
obtained. Applying these ratios to the Ambient Air Quality estimate of
283 cancer cases per year yield an estimated 113 and 175 cancer cases
B-58
-------
TABLE B-10
ESTIMATED ANNUAL CANCER CASES FROM CHROMIUM BY SOURCE CATEGORY
AMBIENT MEASURED CONCENTRATIONS
STUDY
Ambient Air Quality
100% Hexavalent
62% Hexavalent
40% Hexavalent
IEHP - Santa Clara
100% Hexavalent
IEMP - Baltimore
100% Hexavalent
29% Hexavalent
South Coast
100% Hexavalent
81.5% Hexavalent
Totals
100% Hexavalent
<100% Hexavalent
Concentration3 Incidence .
(j*g/m3) Annual Annual Per Million
Population
0.0069 283 1.18
0.0043 175 0.73
0.0028 113 0.47
0.0126-0.0138 3.02-3.3 2.16-2.37
0.016 4.24 2.77
0.8
0.0048 8.97 0.82
0.67
Nationwide
Incidence
283
175
,113
518-569
665
192 .
197
161
197-665
113-192
Point
Estimate
283
113
Total chromium.
_
B-59
-------
TABLE B-ll
ESTIMATED ANNUAL CANCER CASES FROM CHROMIUM - MODELED CONCENTRATIONS
SOURCE
CATEGORY
BOPF's
Chemicals
Chroraite
Cooling Towers
Ferrochromium
Production
Glass Furnaces
Municipal
Incinerators
Refractories
Sewage Sludge
Incinerators
Specialty Steel
Heating/
Combustion
Hazardous Waste
Combustors
Waste Oil
Conbustors
Chrome ploters
Unspecified
Other
Hotor Vehicles
TOTALS
INDIVIDUAL
STUDY
NESHAP/
ATERIS
Coal and Oil
Combustion
0.001
(0.000004)
0.22
(0.00091)
0.00045
(0.000002)
0.58
(0.0024)
0.062
(0.00026)
0.013
(0.000054)
0.00071
(0.000003)
0.016
(0.000067)
0.00016
(0.0000007)
0.25
(0.001)
330°
(1.38)
331
(1.38)
Hazardous
Waste
Combustion
Municipal
Incinerators
Waste Oil
Combustion
Thirty-
five
County
0.0018
(0.00004)
0.12
(0.0005)
0.20
(0.0008)
<0.25
(<0.001)
0.0012-0.0065
<<0.000027)
0.06
(0.0013)
0.01
(0.0002)
273
(5.77)
0.20
(0.0008)
<0.25
(<0.001)
0.12
(0.0005)
0.0012-0.0065
(<0.000027)
273.4
(5.78)
Five City
7.37
(0.464)
0 44
(i).028)
0.00089
(0.00006)
0.0052
(0.00033)
(0.5)
15.81
(1.00)
B-60
-------
TABLE B-ll -- concluded
ESTIMATED ANNUAL CANCER CASES FROM CHROMIUM - MODELED CONCENTRATIONS
SOURCE
CATEGORY
BOPF'S
Chemicals
Chronrite
Cooling Towers
Ferrochromium
Production
Glass Furnaces
Municipal
Incinerators
Refractories
Sewage Sludge
Incinerators
Specialty Steel
Heating/
Combustion
Hazardous Waste
Combustors
Waste Oil
Combustors
Chrome platers
Unspecified
Other
Motor Vehicles
TOTALS
INDIVIDUAL STUDY
Sewage Southeast
Sludge Chicago South Coast
Incinerators
X
<0.024
(0.062)
X
0.26
(0.0011)
0.00086
(0.0022)
X
0.186
(0.473)
6.97
(0.64)
1.17
(0.107)
0.26 0.211 8.14
(0,0011) (0.537) (0.75)
NATIONWIDE
. RANGE3
0.001
0.22
0.0045
0.01-111
0.062
0.013
0.00071-0.12
0.02-6.7
.00016-0.26
0.01-0.53
0.08-0.3
<0.25
0.0012-0.05
113-343°
154
330
26
140-4896
POINT
ESTIMATE
0.001
0.22
0.0045
0.01-111
0.062
0.013
0.12
0.02-6.7
0.26
0.01-0.53
0.2
<0.25
0.0012-0.05
120
--
--
26
147-265
B-61
-------
Footnotes to Table B-11.
NOTE: Numbers in parentheses are annual cancer cases per million population. For nationwide
studies, annual cancer cases were divided by 240 million, unless otherwise noted. For
studies with smaller geographic scopes, the annual cancer cases were divided by the study's
population.
NOTE: An "x" in a column indicates that the source category was considered in the study, but a
specific cancer risk for the source category was not indicated.
a The numbers in this column were calculated by taking the lowest and highest incidence rates for
a source category and multiplying it by 240 (1986 U.S. population in millions). The total for
this column is the summation of the low end of the range and the sum of the high end of the
range.
b The numbers in this column present the results of the reduction analyses. In most instances, a
point estimate of nationwide annual cancer incidence was derived for each pollutant/source
category combination. In some instances, a point estimate could not be reasonably derived. For
these instances, as narrow a range as possible of nationwide annual cancer incidence was
estimated, and such ranges appear in this column. The text discusses how these point estimates
and ranges were derived.
c Based on adjusting estimated incidence from 272 to 191 based on newer emission data for one city
(see text) and then extrapolating the 35-County study estimate nationwide according to
information provided in the 35-County study, which indicates 55% of chrome platers are in 35
counties.
d Includes nine source categories: chrome plating, refractory, chromium chemicals, steel
manufacturing, ferrochromium manufacturing, chromium ore manufacturing, sewage sludge
incineration, municipal refuse incineration, and cement manufacturing. Over 90 percent of the
annual cancer incidence is associated with refractory, chromium chemicals, chrome plating, and
steel manufacturing.
e Does not include risk estimates from "unspecified" and "other" source categories.
B-62
-------
per year respectively or, equivalently, cancer rates of approximately
0.47 and 0.73 cancer cases per year per million population.
Modeled Estimates. Summing across the various source categories
results in a range of nationwide cancer cases between 140 and 489 per
year. By far the majority of this range is due to the estimate for
chrome platers (113 to 343 per year) and secondarily to cooling towers
(0.01 to 111 per year).
The wide range of incidence from the chrome platers appears to be
mostly due to the estimate of total chromium emissions attributable to
chrome platers in one particular city. The 35-County study uses
emissions approximately 4.2 times that used in the 5-City study.
However, the higher level of emissions used in the 35-County study is
apparently out-of-date. The 5-City study's data are more recent and are
known to be in agreement with the local records for that city.
Adjusting the 35-County study's estimate to the lower emissions used in
the 5-City study, a new nationwide cancer rate of about 1.43 cancer
cases per year per million population is calculated. This is still
higher than the 5-City study's cancer rate of 0.5 cancer cases per year
per million population. The Southeast Chicago study shows an annual
cancer incidence per million population essentially equivalent to that
of the 5-City study (0.47 vs. 0.5). For the best estimate of risk from
chrome platers, the annual cancer incidence per million population from
the 5-City study and the Southeast Chicago study (0.5 cancer cases per
year per million population) was selected as the best estimate to
extrapolate to a nationwide estimate of annual incidence.
The 35-County study and the 5-City study also show significantly
different cancer rates for cooling towers (0.0004 vs. 0.46 cancer cases
B-63
-------
per year per million population). However, insufficient data are
available to understand why such a difference exists.
Point Estimate. The range of cancer risks from the studies based
on modeled ambient concentrations and those from the ambient measured
studies are similar (147 to 265 vs. 197 to 665 cancer cases per year).
Considering the ambient-based estimates only, the Ambient Air Quality
study, by virtue of its broader geographic scope, may better reflect
nationwide incidence. The Ambient Air Quality study would result in an
upper estimate of about 283 cancer cases per year (at 100% hexavalent).
By applying the results of the 5-City study as to estimated average
ratio'of hexavalent to total chromium to the Ambient Air Quality study's
result, total cancer cases would be estimated to be 113 per year.
Considering the modeled-based estimates, there does not seem to be
sufficient information to further narrow the range (147 to 265).
Because of the uncertainty of applying a nationwide ratio of
hexavalent to total chromium to ambient-measured data, the modeled
estimates' range of 147 to 265 cancer cases per year was selected as the
estimate of nationwide annual cancer incidence due to exposure to
hexavalent chromium emissions.
B-64
-------
Coke Oven Emissions. Three studies estimated cancer incidences from
coke oven emissions - the NESHAP/ATERIS data base, the 35-County study,
and the Southeast Chicago study (see Table B-12). Only the Southeast
Chicago study identified estimated concentrations of coke oven
emissions. The Southeast Chicago study estimated a range of
concentrations from approximately 0 ng/m3 to 6.1,/ig/m3. (That study
noted that the actual peak concentration for coke oven pollutants is
probably somewhat higher than the 6.1 /ig/m3:) The Southeast Chicago
study estimated approximately 0.35 cancer cases per year for its study
area. This is equivalent to an areawide average coke oven emissions
concentration of approximately 0.1 ^g/m3. The other two studies show
areawide average concentrations of approximately 0.005 /tg/m3 (35-County
study) and 0.0033 ^g/m3 (NESHAP/ATERIS data base).. This trend in
calculated concentrations is expected since the area covered by the
Southeast Chicago study is known to contain these emission sources and
the counties in the 35-County study were selected, in part, on the basis
of known emission sources. Thus, those two studies would be expected to
result in higher cancer rates and estimated coke oven concentrations.
The NESHAP/ATERIS data base is broadest in scope, including areas with
and without-eoke oven emission sources, and was therefore selected as
the estimate of cancer incidence from exposure to coke oven emissions
(approximately 7 cancer cases per year).
B-65
-------
TABLE B-12
ESTIMATED ANNUAL CANCER CASES FROM COKE OVEN EMISSIONS
SOURCE
CATEGORY
Iron and Steel
TOTALS
MODELED
INDIVIDUAL STUDY
NESHAP/
ATERIS
6.9
(0.029)
6.9
(0.029)
Thirty-
five
County
2.1
(0.044)
2.1
(0.044)
Southeast
Chicago
0.346
(0.88)
0.346
(0.88)
NATIONWIDE
POINT
RANGE3 ESTIMATE0
7-11c 7
7-11° 7
NOTE: Numbers in parentheses are annual cancer cases per million population. For nationwide
studies, annual cancer cases were divided by 240 million, unless otherwise noted. For
studies with smaller geographic scopes, the annual cancer cases were divided by the study's
population.
a The numbers in this column were calculated by taking the lowest and highest incidence rates for
a source category and multiplying it by 240 (1986 U.S. population in millions). The total for
this column is the summation of the low end of the range and the sum of the high end of the
range.
k The numbers in this column present the results of the reduction analyses. In most instances, a
point estimate of nationwide annual cancer incidence was derived for each pollutant/source
category combination. In some instances, a point estimate could not be reasonably derived. For
these instances, as narrow a range as possible of nationwide annual cancer incidence was
estimated, and such ranges appear in this column. The text discusses how these point estimates
and ranges were derived. _.
0 The range does not include the nationwide estimate that would be calculated using the incidence
rate from the Southeast Chicago study, because it was felt that the concentration of iron and
steel facilities in the Southeast Chicago study area was too non-representative of typical
nationwide conditions.
B-66
-------
1,2 Dichloropropane. Although two studies reported cancer risk for 1,2
dichloropropane, the lEMP-Baltimore study apparently applied the annual
cancer incidence per million population generated in the lEMP-
Philadelphia study (i.e., 0.067 cancer cases per year per million
population) to the Baltimore population to estimate cancer risk. In the
IEMP-Philadelphia study, the initial source of 1,2 dichloropropane is
from an unspecified chemical manufacturing plant. Thus, it would be
reasonable to apply the lEMP-Philadelphia annual cancer incidence per
million population only in those instances where a similar facility
exists. Since the type of facility is not reported, a reasonable
nationwide estimate can not be made. Obviously, the lEMP-Philadelphia
annual cancer incidence per million population could be applied to the
total U.S. population to yield a nationwide estimate of 16 cancer cases
per year. This estimate, however, has essentially no meaning. Thus,
the best that can be done is to say that there are possibly as little as
0,2 cancer cases per year (in Philadelphia and Baltimore).
B-67
-------
Dioxin. Five studies included dioxin as a pollutant of concern (see
Table B-13). The risk estimates in each study are highly uncertain.
The South Coast study used ambient data found in an article7 because
there were no currently available data on ambient concentrations of
dioxins and furans in the South Coast .study area. The South Coast study
report notes that "these data have limited usefulness because the vapor
phase concentrations of these pollutants were not measured."
Both the Southeast Chicago study and the Municipal Waste Combustor
study examined dioxin emissions from incinerators. Both studies noted
problems with estimating risk. For example, the Municipal Waste
Combustor study identified two problems with estimating risk from
dioxins. One problem dealt with the capture efficiency of the sampling
method used to estimate emissions of dioxin and the other problem dealt
with the methodology needed to extrapolate the risk from
tetrachlorinated dibenzodioxin (TCDD) to the other dioxin subspecies.
The TSDF study presents a very rough initial estimate of potential
risk from a large number of pollutants. By proportioning according to
projected emissions and unit risk factors, cancer risk for each
individual pollutant can be generated. When this is done, dioxin is
' calculated to contribute 91 of the estimated 140 cancer cases from
TSDFs. This estimate must be viewed as a very crude estimate. In fact,
there may be substantially less dioxin emitted from TSDFs so that the
actual risk is much lower.
Point Estimate. For dioxin, it is extremely difficult to identify
a point estimate because of the "limited usefulness" of the ambient data
7Czuczwa, J. and R.A. Hites, 1984. "Environmental Fate of Combustion
Generated Polychlorinated Dioxins and Furans," Environ. Sci. Technol.
18(6)-.444-50.
B-68
-------
TABLE B-13
ESTIMATED ANNUAL CANCER CASES FROM DIOXIN BY SOURCE CATEGORY
SOURCE
CATEGORY
TSDFs
Sewage Sludge
Incinerators
Municipal Waste
Combustors
TOTALS
MODELED
AMBIENT
INDIVIDUAL STUDY
Municipal Sewage Sludge Southeast
Waste Incinerator TSDFs Chicago South Coast
Combustors
91 0.00014
(0.38) (0.00036)
0.42
(0.0013)
1-20° 0.0029
(0.004-0.083) (0.0073)
1-20 0.42 91 0.003
(0.004-0.083) (0.0013) (0.38) (0.0076)
0.29-5.71
(0.026-0.52)
NATIONWIDE
POINT
RANGE3 ESTIMATE"
0.09-91 0.09-91
0.42 0.42
1-20 1-20
2-111 2-111
6-125 6-125
NOTE: Numbers in parentheses are annual cancer cases per million population. For nationwide
studies, annual cancer cases were divided by 240 million, unless otherwise noted. For
studies with smaller geographic scopes, the annual cancer cases were divided by the study's
population.
a The numbers in this column were calculated by taking the lowest and highest incidence rates for
a source category and multiplying it by 240 (1986 U.S. population in millions). The total for
this column is the summation of the low end of the range and the sum of the high end of the
range.
The numbers in this .column present the results of the reduction analyses. In most instances, a
point estimate of nationwide annual cancer incidence was derived for each pollutant/source
category combination. In some instances, a point estimate could not be reasonably derived. For
these instances, as narrow a range as possible of nationwide annual cancer incidence was
estimated, and such ranges appear in this column. The text discusses how these point estimates
and ranges were derived.
0 This estimate is based on an older assessment and is calculated by applying the ratio of the
newer total, risk estimate to the older total risk estimate. See the summary on municipal waste
combustors in Appendix C.
B-69
-------
and the great uncertainty associated with the modeled estimates.
Therefore, the nationwide estimate of annual cancer incidence from
exposure to dioxin is a range, from approximately 2 to 125 cancer cases
per year nationwide. Even this range remains very crude and this caveat
should be kept in mind.
B-70
-------
Ethvlene dibromide. Ten studies included ethylene dibromide (EDB) (1,2-
dibromoethane) as a pollutant of concern (see Table B-14). Sources of
$
EDB in these ten studies included motor vehicles, drinking water
aerators, gasoline marketing (service stations, refueling, bulk plants
and terminals), TSDFs, and EDB manufacturing and formulation facilities.
EDB is used in leaded gasoline as a "scavenger," and as leaded gasoline
is phased-out, EDB emissions will be reduced. Two studies used ambient
measured concentrations to estimate risk, and eight used modeled
concentrations. (The South Coast study used both ambient measured and
modeled concentrations.)
Modeled Estimates. As seen in Table 3-23, the estimate of annual
cancer cases per million population varies dramatically for the gasoline
marketing source category. There is a 1,000-fold difference (0.038 vs.
0.000033) in this estimate. This could be explained, in part, as the
NESHAP/ATERIS data base's estimate is based on a July 1978 report8,
whereas the Gasoline Marketing study is a more recent study.
Furthermore, the Gasoline Marketing study's estimate is based on a 33-
year projection period in which EDB emissions fall to zero for the last
20 to 23 years due to the projected complete phase-out of leaded
gasoline. Considering its focused subject area and the explicit
accounting of the projected phase out of EDB as a gasoline additive, the
estimate of cancer incidence from the Gasoline Marketing study was
selected as the best estimate of cancer risk from gasoline marketing for
EDB emissions.
For motor vehicles, the range of cancer incidence per million
population is narrower, from almost 0.002 to 0.011 annual cancer cases
8 Mara, Susan J, and Shonh S. Lee, Atmospheric Ethylene Dibromide:
A Source-Specific Assessment. SRI International, July 1978.
B-71
-------
TABLE B-14
ESTIMATED ANNUAL CANCER CASES FROM ETHYLENE DIBROMIDE BY SOURCE CATEGORY
SOURCE
CATEGORY
Gasoline
Marketing
Drinking Water
Aerators
Motor Vehicles
TSOFs
EOB Mfg. and
Formulation
TOTALS
MODELED
AMBIENT
INDIVIDUAL STUDY
Ambient NESHAP/ Drinking Gasoline Mobile IEMP- ^i^-..,
Air ATERIS Water Marketing Sources TSDFs Kanawha Santa Clara
Quality Aerators Valley
9.2 0.008 0.0003
CO. 038) (0.000033) (0.0028)
<0.0002C
(<0. 000001)
x 0.78 0.0011
(0.004) (0.01D
0.02
(0.0008)
X
11.5 <0.0002 0.008 0.78" 0.02 ' 0.0013 0.004
(0.048) (<0. 000001) (0.000033) (0.004) (0.0008) (0.013) (0.003)
£8
(0.28)
B-72
-------
TABLE B-14 -- concluded
ESTIMATED ANNUAL CANCER CASES FROM ETHYLENE DIBROMIDE BY SOURCE CATEGORY
SOURCE
CATEGORY
Gasoline
Marketing
Drinking Water
Aerators
Motor Vehicles
TSDFs
EDB Mfg. and
Formulation
TOTALS
MODELED
AMBIENT
INDIVIDUAL STUDY
Southeast
Chicago South Coast
0.00071 0.02
(0.0018) (0.002)
0.00071 . 0.02
(0.0018) (0.002)
1.13
(0.104)
NATIONWIDE
POINT
RANGE3 ESTIMATE13
0.008-9.2 0.008
<0.0002C <0.0002C
0.44-2.64 0.78
0.02 0.02
<2.3 1.5d
0.5-14 2.3
25-68 68
NOTE: Numbers in parentheses are annual cancer cases per million population. For nationwide
studies, annual cancer cases were divided,by 240 million, unless otherwise noted. For
studies with smaller geographic scopes, the annual cancer cases were divided by the study's
population.
NOTE: An "x" in a column indicates that the source category was considered in the study, but a
specific cancer risk for the source category was not indicated.
a The numbers in this column were calculated by taking the lowest and highest incidence rates for
a source category and multiplying it by 240 (1986 U.S. population in millions). The total for
this column is the summation of the low end of the range and the sum of the high end of the
range.
The numbers in this column present the results of the reduction analyses. In most instances, a
point estimate of nationwide annual cancer incidence was derived for each pollutant/source
category combination. In some instances, a point estimate could not be reasonably derived.
For these instances, as narrow a range as possible of nationwide annual cancer incidence was'
estimated, and such ranges appear in this column. The text discusses how these point estimates
and ranges were derived.
0 The Drinking Water Aerator study estimated 0.0002 annual cancer cases from EDB and
dibromochlbropropane combined. No separate estimate for EDB was given.
Assumes 0.003 cancer cases per year per million population is due to motor vehicles and the
remainder (0.0063) is from EDB manufacturing and formulation.
B-73
-------
per million population. These estimates translate into total nationwide
cancer cases of approximately 0.5 to 2.6 per year. Considering again
the specific source nature of the Mobile Source study, the best estimate
of cancer risk from mobile sources for EDB emissions was selected from
the Mobile Sources study (0.78 cancer cases per year nationwide).
Cancer cases from TSDFs and drinking water aerators appear to be
negligible. The estimates of cancer risk from EDB manufacturing and
formulation and from motor vehicles in the NESHAP/ATERIS data base could
not be "broken out" from the NESHAP/ATERIS data base's total.
When combined, the above data result in a potential cancer rate
range of 0.002 to 0.056 cancer cases per year per million population (or
0.5 to 14 annual cancer cases nationwide), with a best estimate of 2.4
cancer cases per year nationwide. The best estimate includes 1.5 cancer
cases per year nationwide from EDB manufacturing and formulation
facilities. This estimate was obtained by subtracting the best estimate
of cancer risk from motor vehicles (0.78 cancer cases per year) from the
NESHAP/ATERIS data base's estimated 2.3 cancer cases per year from motor
vehicles plus EDB manufacturing and formulation facilities.
Ambient Estimates. Two studies used ambient measured
concentrations to estimate risk from EDB - the South Coast study and the
Ambient Air Quality study. The South Coast study measured annual EDB
concentrations at five locales, ranging from 0.0154 to 0.0616 /ig/m3 in
1985, with a population weighted annual average concentration between
0.021 and 0.048 /ig/m3. These concentrations were substantially higher
than the modeled concentrations. The South Coast study suggested that
this discrepancy might be due to entrainment and out-gassing from the
ground, which-would increase the ambient measured concentrations
relative to the modeled ambient concentrations.
B-74
-------
The Ambient Air Quality study used ambient data available from 30
locations. In 29 of the 30 locations, concentrations ranged from 0.03
to 0.09 /ig/m3, with one location reporting a concentration of 0.2 ng/m3.
Most of the data was for California locations, which were measured in
1986 and 1987. Calculating cancer cases in California based on the
California ambient measured concentrations (0.03 to 0.1 ^g/m3) and
population and in the rest of the U.S. based on the non-California
ambient measured concentrations (0.04 to 0.02 ^g/m3) and population, the
Ambient Air Quality study estimated 68 annual cancer cases nationwide.
This is equivalent to the cancer risk calculated from population
weighted concentrations of 0.10 ^g/m3 for urban populations and 0.05
fig/m3 for rural populations.
Point Estimate. The consistency of the ambient measured
concentrations suggests that the studies that modeled EDB concentrations
did not fully account for all sources of EDB emissions, whether they
occur from entrainment or outgassing, as suggested in the South Coast
study, or for some other reason. The ambient measured concentrations,
thus, seem to be a preferable basis for estimating risk. The
concentrations measured in the South Coast Air Basin (0.021 to 0.048
jtg/m3) are very similar to the California data used in the Ambient Air
Quality study (0.03 to 0.1 /tg/m3). Given the broader geographic scope
of the Ambient Air Quality study, the results from that study (68 cancer
cases per year nationwide) were selected as the estimate of cancer
incidence from exposure to EDB emissions.
B-75
-------
Ethvlene dlchloride. Ethylene dichloride (EDC) (1,2-dichloroethane)
emissions come from both point and area sources. Point sources of EDC
include the production of EDC, vinyl chloride, methyl chloroform (CHC),
ethyl amines, trichloroethylene, perch!oroethylene, vinylidene chloride,
ethyl chloride, polysulfide rubber, and liquid pesticide. Area source
emissions include grain fumigation, leaded gasoline, paints, coatings,
adhesives, cleaning solvents, and waste treatment, storage, and disposal
facilities.9
Thirteen studies included EDC as a pollutant of concern (see Table
B-15). At least thirteen source categories were specified in these
studies. The cancer risk estimates in the Ambient Air Quality study and
the lEMP-Baltimore study were based on ambient measured concentrations.
The lEMP-Philadelphia study compared ambient measured concentrations
with modeled ambient concentrations. The lEMP-Philadelphia study and
the remaining ten other studies used models to estimate cancer risk.
Ambient Estimates. The Ambient Air Quality study used
an urban average concentration of 0.59 /ig/m3 and a rural average
concentration of 0.20 ^g/m3 to estimate cancer risk. EDC concentration
data from 17 locations were used,''with a range of concentration from
0.09 to 4.12 Mg/m3. The lEMP-Baltimore study measured annual average
ambient concentrations ranging from 0.2 to 2.6 /*g/m3, with a population
weighted annual average concentration of 0.26 ^g/m3. The lEMP-Baltimore
study data falls within the range used in the Ambient Air Quality study.
Since the higher ambient concentrations in the Ambient Air Quality study
seem to correspond to cities with known point sources of EDC emissions
9 U.S. EPA. Locating and Estimating Air Emissions from Sources of
Ethvlene Dichloride. EPA-450/4-84-007d. March 1984.
B-76
-------
TABLE B^15
ESTIMATED ANNUAL CANCER CASES FROM ETHYLENE DICHLORIDE BY SOURCE CATEGORY
SOURCE
CATEGORY
Pesticide
Production
POTWs .
Pharmaceutical
Manufacturing
EDC Production
CMC Users
Drinking Water
Aerators
Gasoline
Marketing
TSDFs
Unspecified
Point Sources
Chemical
Manufacturing
Refineries
Sewer
Volatilization
Delaware
River
Motor
Vehicles
TOTALS
MODELED
AMBIENT
INDIVIDUAL STUDY
Ambient NESHAP/ Drinking Gasoline
Air ATERIS Water Marketing POTWs
Quality Aerators
0.01
(0,00004)
0.09
(0.0004)
0.0029
(0.000012)
0.79
(0.0033)
0.0044
(0.000018)
X
0.01
(0.00004)
0.81 negligible 0.01 0.09
(0.0034) (0.00004) (0.0004)
45
(0.19)
TSDFs
5.37
(0.024)
5.37
(0.024)
B-77
-------
TABLE B-15 continued
ESTIMATED ANNUAL CANCER CASES FROM ETHYLENE DICHLORIDE BY SOURCE CATEGORY
SOURCE
CATEGORY
Pesticide
Production
.
Thirty-
five
County
IEMP- IEMP-
Five City Baltimore Kanawha
Valley
IEHP- Southeast
Philadelphia Chicago
POTWs
Pharmaceutical
Manufacturing
EOC Production
CHC Users
4.62
(0.098)
0.083
(0.05)
negligible
Drinking Water
Aerators
Gasol ine
Marketing
TSOFs
Unspecified
Point Sources
Chemical
Manufacturing
Refineries
Sewer
Volatilzation
Delaware
River
Motor
Vehicles
TOTALS
MODELED
AMBIENT
0.12 0.013 0.00035 0.00087
(0.003) (0.0008) (0.0035) (0.00052)
1.25 0.814 0.00002
(0.026) (0.051) (0.0002)
0.009 , 0.00004
(0.0005) (0.000024)
0.011
(0.0066)
0.019
(0.011)
0,022
(0.013)
5.99 0.83 0.00037 0.138
(0.127) (0.052) (0.0037) (0.083)
0.148
(0.097)
negligible
0.00071
(0.0018)
-
0.00071
(0.0018)
B-78
-------
TABLE B-15 -- concluded
ESTIMATED ANNUAL CANCER CASES FROM ETHYLENE DICHLORIDE BY SOURCE CATEGORY
SOURCE
CATEGORY
Pesticide
Prpduction
POTWs
Phar(naqe.utical
Manufacturing
EDC Production
CHC Users
Drinking Water
Aerators
Gasoline
Marketing
TSDpS
Unspecified
Point Sources
Chemical
Manufacturing
Refineries
Sewer
Volatilzation
Delaware
River
Motor
Vehicles
TOTALS
MODELED
AMBIENT
Sputh Coast
X
0.007
(0.0007)
NATIONWIDE
RANGE3
0.01
0.1-24,
0.0029
0.79
0,0044
negligible
0.01-0.84
5.37
0-12.2
0.007-0.12
1.92
3.36
3.84
0-0.17
15.4-52.5
23.3-45
POINT
ESTIMATE13
0.01
0.1
0.003
0.79
0.0044
negligible
0.01
5.4
0-12,2
0.1
1.92
3.36
3.84
<0.2
16-28
45
B-79
-------
Footnotes to Table B-15.
NOTE: Numbers in parentheses are annual cancer cases per million population. For nationwide
studies, annual cancer cases were divided by 240 million, unless otherwise noted. For
studies with smaller geographic scopes, the annual cancer cases were divided by the
study's population.
NOTE: An "x" in a column indicates that the source category was considered in the study, but a
specific cancer risk for the source category was not indicated.
a The numbers in this column were calculated by taking the lowest and highest incidence rates for
a source category and multiplying it by 240 (1986 U.S. population in millions). The total for
this column is the summation of the low end of the range and the sum of the high end of the
range.
The numbers in this column present the results of the reduction analyses. In most instances, a
point estimate of nationwide annual cancer incidence was derived for each pollutant/source
category combination. In some instances, a point estimate could not be reasonably derived.-
For these instances, as narrow a range as possible of nationwide annual cancer incidence was
estimated, and such ranges appear in this column. The text discusses how these point estimates
and ranges were derived.
B-80
-------
the lower concentrations in Baltimore could reflect the lack of such
sources.
On the basis of its wider geographic scope and its coverage of
cities with and without known point sources, the results from the
Ambient Air Quality study seem to provide a better estimate of
nationwide cancer risk from EDC than does the lEMP-Baltimore study.
Modeled Estimates. Unlike the results for many other compounds,
there appears to be very good agreement as to the risk from EDC between
the estimates based on modeled versus measured ambient concentrations.
As seen in Table B-15, when the various individual source categories are
summed, the range of nationwide risks is nearly identical to the range
based on the two studies using ambient measured concentrations.
The major difficulty in summing the source categories is the
"unspecified point source" source category in the 35-County study. If
this source category duplicates other specified source categories, then
the range of nationwide cancer cases decreases from 15 to 53 per year to
3 to 40 per year. It is interesting to note that the total cancer rate
of the 35-County study (0.127 cancer cases per year per million
population) falls within the range created by the lEMP-Baltimore and the
Ambient Air Quality studies (0.097 to 0.19 cancer cases per year per
million population).
Of the individual source categories, the nationwide cancer risk
associated with POTWs has the largest absolute difference. The POTW
study shows a much lower cancer rate (0.0004 annual cancer cases per
million population) than does either the lEMP-Philadelphia study (0.059
annual cancer cases per million population) or the 35-County study
(0.098 annual cancer cases per million population). The causes for this
wide difference are unknown. If the nationwide POTW study is assumed to
B-81
-------
more accurately reflect the exposure to EDC emissions from POTWs than
the two smaller geographic studies, estimates of nationwide cancer risk
would be between 15.4 and 28.6 cases per year. If the "unspecified
point source" source category is also eliminated (as discussed above),
the nationwide cancer cases decrease further, to 3.2 to 16.4 cases per
year.
Three of the source categories (refineries, sewer volatilization,
and Delaware River) are extrapolated from the lEMP-Philadelphia study to
obtain nationwide cancer risk estimates. Whether this is reasonable is
very uncertain. For example, while a large number of cities have
petroleum refineries, they are-better modeled on a site-specific basis
than by applying the results of one city with two refineries to the
nation as a whole.
Point Estimate. Overall, the results from the various studies are
fairly close. The Ambient Air Quality study's result, 45 cancer cases
per year, was selected as the estimate for nationwide cancer incidence
from exposure to EDC on the basis of its wider geographic scope and
greater likelihood of accounting for area-wide emission sources.
B-82
-------
Ethylene oxide. Six studies estimated cancer risk from ethylene oxide
(ETO) emissions (see Table B-16). The TSDF study also included ethylene
oxide as a pollutant of concern, but no emissions of ethylene oxide
where indicated and thus no risk was reported. Specific source
categories include ETO production and commercial sterilization. All of
the studies used modeled ambient concentrations to estimate risk.
Modeled Estimates. The six studies show a wide range of cancer
rates, from approximately 0.02 to 8.4 annual cancer cases per million
population. The lEMP-Kanawha Valley has the highest cancer rate, 3.5 or
8.4 cancer cases per year per million population, depending on which
model is used. The 8.4 cancer rate is likely to be an overestimate
because of the nature of the model. The sources of ethylene oxide in
the Kanawha Valley are particular chemical manufacturing facilities.
These facilities are not included in the NESHAP/ATERIS data base. In
addition, the document "Locating and Estimating Emissions from Sources
of Ethylene Oxide" (U.S. EPA, EPA-450/4-84-0071, September 1986) does
not list any source in West Virginia. The high cancer rate is due to
specific sources that may be unique to the Kanawha Valley. Even if not
unique to the Kanawha Valley, no information is available to extrapolate
to obtain a nationwide estimate.
Except for the 5-City study and assuming the ATERIS file is the
more accurate estimate of risk for commercial sterilizers under the
NESHAP/ATERIS data base, the remaining studies have estimated cancer
rates between 0.018 and 0.028 annual cancer cases per million
population. This results in a relative narrow absolute range when
extrapolated to nationwide cancer cases of 4 to 7 per year. Of these
four studies, three have even closer estimates, 0.018 to 0.02 annual
cancer cases per million population. The slightly higher 35-County
B-83
-------
>-
g
Ul
§
IU
o
8
>-
CO
a
X
o
i!
Ul
.0 g
^ as
m uj
a g
CQ 04
< U.
H*
OT
1
o:
1
g
g
o
Ul
i
£
Ul
Ul
o
>
«e
z
1
H-
(/>
1
0
~
-
gl
O 1-
0. CO
Ul
CD
1
oc
05 O
CO U>
<->
CJ
1
U_
1
t- > C
.C M- O
H- CJ
CL CO
CO til
LU |
z <
UJ O£
CJ O
^ ill
O 1
CO o NO
in ON m «~ CM T
. O r- O O 0
*~ CD CD 0 CD
vx *^ O
co
«- CM
ro o
«- CD
"
CO
o -a-
in o '^
i in
*- O CM O
T 1 "CD
4- T- *~ 0
CD N«X
-^ C
C O
O "~ D)
.r^ 4-1 T3 **_
*J U 5 £
(0 (D W 13 -
O -
-------
m
CU
IQ
CO
I
o
"co
CU
O
C
4->
U
O
OJ Q.
>. Q.
cu >»
2-3
> 4-*
CO ,
cu .c
£- 4-J
CU
:er cases
divided 1
C 0)
cn c-
o cu
CO Cfl
c cu
cu o
Cfl Q)
CU O
5 co
3 o
CO '
s
o
CD
O
0)
o
I_
o
CO
cu
£_
0
CO
£_
c.
o
o
c
o
'o
d)
D.
CO
ro
1
CO
o
x:
c
73
CU
f_
CU
CO
c
0
o
CO
§
>,
o
4J
o
1
CO
cu
4-»
ro
o
o
1
o
o
4->
- ro
c c
UJ
o cu
ro
o t-
OJ CU
J=
4-» O
_;?1
||
4J
_, CU
2 -c
E 4-»
73 M-
xl
t- CO
o
cn o
cu .c
4-1 4-*
O ~O
C
O CO
0
I- 01
3 D)
O C
cfl ro
£_
CU
cu
C_ r-
O 4J
M-
CO O
0)
01 3
0 O
C
CU
"U cu
"- .c
O 4-»
c
"~ "o
CO C
0) O
O) 4J
IE E
-n §
C CO
CU
CU
Cfl 4-J
3 CO
o
J-
JC 3
4-* '
0
O) O
c
cn
cu .E
>.~
"*-
g^
4-» CU
D 4-»
0
0)
CO .E
O r
o
cu >-*
11
Cfl
p- C
.E
4-*
c
c o
cn cu
§ 8.
c
CU C/}
t 13
o o
u_ _E
1
t- T3
0) 0) C
ro tl
O 01 O
o o
ro X cfl
c ro 4-»
ro c
o c
CU CO
T3 ro
CD t_
2 i- ro
c cu
4J ro
CU 4-»
c o co
C 0)
M- D)
O -0 C
' ro
0) 3 c_
4-> O
E: O
CU 3
4-» 4-« Cfl
co ro
0) E -0
4-* 4-1 CU
C CO
.- 0> »
Q.4-. CU
C 4J
co ro
Cfl 4-*
CU CO CO
0 0>
c
ro cfl co
4-* cu ro
g.0 3
CU O
4-> O
4-» CO C
CO C CU
§ "~ p
cu o
C E C
_ o
Cfl
£-
c o
CO P-I O
cu c
CO CU
>* p O
c
ro o
c ro
ro 4-> 3
ro c. TJ
§c c cu
i- ro >
.r- n .
4-> E CU C_
o o ~o cu
3 0 pp- "D
0 3
0) X C 0)
£» £_ O 1-
O "- 0)
0) D) 4-> 3
.E 0> CO
4-> 4-> C CO
ro cu
M- O >*- 0)
o o c
cu ro
CO 0 CU t-
4-> C '
P 3 ,Q TJ
D 0 C
Cfl CO Cfl CU
t- 4-« O CO
C O- CU
cu ro 4-*
SL 4-» co ro
4J D ro E
4_| i CL) 4J
co ro
0) .C C- 4-1
I- 0 C
Q- ro ro --
cu o
§<- i °-
3 O C- CU
>4- C- Cfl
o ro cu
o -o c ^r
CU 4-<
Cfl > Cfl
ro 3
-C C- O
"O cfl
C CU cfl
. WOO)
ro c cfl
cn 3 ro cn
C- 4-» 3
feu cn o
o c co
C 73 0)
CO 4J
0) O CU X
.E C J= 0)
t 4-* 4-»
O)
o
c
ro-
o
1
c
c
ro
ro
o
H-
o
4-»
86 percen
.If
4-*
3
t_
c
o
o
CO
ro
4-f
c
-
o
c_
ro
Cfl
cu cu
' H-
tl «
0)
Cfl UJ
1
- <
.2 ~ CO
"O ro
3 O
4-*
cfl c_
cu
a. o
< c
z ro
c/> o
UJ
ro
ro c
c c
(U
CD
C
C- O
O
TJ
§CU
Cfl
a. ro
3 .a
T3 CO
0) *p~
Cfl
ro cu
JD D)
c
co ro
r- £_
cu 3
D) O
C _J
ro
t- CO
&CO
ro
O 0
TJ *-
Cfl CU
ro 4-> cj
ro i
X cu in
4-* C_
c o cu
. .c
ro co 4J
I- " O)
0) 0) C
3 ro ro
s^i
ro
(U CU t_
CO 4-» 4->
cu e cu
o -
Jj CO ^2
CU
CO 73
CJ (I) CU
CO 73 C
fD -
O 2 (D
C 4V
t_ O .Q
CU - O
O 4-»
c ro
o ro
ro E
ro c 4~»
3 *- CO
c ro cu
C 4J
to _Q _:*;
O CO
H- -
O 0 (L.
M
fl) £
4J ^: cu
to o o
- JZ CO
4-» 3: O
CO
. a.
CU O)73
o c cu crt
C 4-«
cu 4-* ro :D
73 4-» E
o e 4-» to
C CU CO 4-»
cu o
CO 4-*
O) CU CO
C QJ CU
»- 4-» CO JZ
4-» -r- CO 4J
ro » o
O 0 £- +-
n ro .
CO ill - 73
ro
0
o
0
c_
3
O
cfl
s
o
C-
o
Cfl
2
M-
0
cu
Q.
CO
C
~§
TJ
c
4-1
i
CO
I
8-85
-------
cancer rate may be due to the selection of counties with known sources
of ethylene oxide emissions (e.g., commercial sterilizers). If the
higher estimate for commercial sterilizers from the NESHAP/ATERIS data
base is used, then the range is approximately 4 to 13 cancer cases per
year nationwide.
The 5-City study has a calculated cancer rate of 0.11 annual
cancer cases per million population. One city has a cancer rate of
0.144 cancer cases per year per million population, with the other four
cities having rates between 0.001 and 0.04. Without the one city, the
cancer rate for the remaining four cities is calculated to be 0.022
cancer cases per year per million population. This estimate is much
more in line with the other studies. The high cancer rate in the one
city, which is also located in one of the counties in the 35-County
study, appears to be attributable to an abundance of commercial
sterilizers. Based on information in the NESHAP/ATERIS data base, this
city has approximately 9 commercial sterilizers, whereas each of the
other four cities have between 0 and 2 commercial sterilizers.
Point Estimate. Excluding the lEMP-Kanawha Valley study and based
on the above considerations, a nationwide estimate based upon 0.02
annual cancer cases per million population appears to be a reasonable
estimate. This results in an estimate of approximately 4 to 5 cancer
cases per year. The sources covered in the lEMP-Kanawha Valley appear
to be independent of the other source categories. Thus, the
approximately 1 cancer case per year from that study can be added to the
4 to 5 cancer cases per year to result in 5 to 6 cancer cases per year.
B-86
-------
Formaldehyde. Ten studies considered formaldehyde in their estimate of
cancer risk from ambient air pollutants (see Table B-17). Numerous
chemical manufacturing production processes and other point sources
contribute to formaldehyde emissions. In addition, area sources, such
as motor vehicles, contribute to formaldehyde emissions. Finally, a
large portion of formaldehyde in the air is the result of secondary
formation. This source of formaldehyde is not typically accounted for
in modeling studies because there are no validated photochemical models
to estimate secondary formaldehyde production from VOC and other
precursors. Thus, assessments based on ambient monitoring data provide
a more complete accounting of actual exposure to formaldehyde than from
emission estimates alone.
Ambient Estimates. Average annual formaldehyde data used in the
Ambient Air Quality study ranged from 1.1 to 5.0 /tg/m3 for individual
locales. Estimates of cancer risk in the Ambient Air Quality study were
made using an average urban concentration of 3.16 ng/m3 and an average
rural concentration of 1.50 /tg/m3. The South Coast study used a
concentration of approximately 14.7 /ig/m3 to estimate cancer risk. In
the 5-City study, a single representative annual average formaldehyde
concentration was selected for each city, ranging from 3 /tg/m3 to
6.7 ;ig/m3. in the Southeast Chicago Study, an ambient-measured
concentration of 2.98 Mg/m3 was obtained at a single site. This
concentration reflects 16 samples collected for 24 hours every 12 days
from September 1987 to March 1988. However, the Southeast Chicago study
notes that "the absence of data from the summer, when photochemical
formation of formaldehyde is greatest, indicates that available data
probably understate the annual average formaldehyde concentration."
B-87
-------
TABLE B-17
ESTIMATED ANNUAL CANCER CASES FROM FORMALDEHYDE BY SOURCE
SOURCE
CATEGORY
Chemical
Manufacturing
Motor Vehicles
Heating/
Congestion
Municipal
Contbustors
Municipal Waste
Incinerators
TSOFs
Unspecified
Sources
Hor.ferrous
Smelters
Petroleum
Refining
Solvent Use
Uoodsmoke
Secondary
Formation
TOTALS
MODELED
AMBIENT
INDIVIDUAL STUDV
Ambient
Air
Quality
NESHAP/
ATERIS
Coal and Oil
Combustion
Mobile
Sources
Municipal
Incinerators
TSDFs
0.062C
CO.0003)
43d-48e
(0.18-0.2)
0.01
(0.00004)
0.009
(0.00004)
0.31
(0.0013)
1.81
(0.0075)
124
(0.52)
0.062-1.81
(0.0003-
0.0075)
0.01
(0.00004)
43d-43e
(0.18-0.2)
0.009
(0.00004)
0.31
(0.0013)
Thirty-
five
County
9-10
(0.2)
3.37
(0.07)
2.93
(0.06)
13.0
(0.27)
B-88
-------
TABLE B-17 -- concluded
ESTIMATED ANNUAL CANCER CASES FROM FORMALDEHYDE BY SOURCE
SOURCE
CATEGORY
Chemical
Manufacturing
Motor Vehicles
Heating/
Combustion
Municipal Waste
Combustors
Municipal
Incinerators
TSDFs
Unspecified
Sources
Nonferrous
Smelters
Petroleum
Refining
Solvent Use
Woodsmoke
Secondary
Formation
TOTALS
MODELED
AMBIENT
INDIVIDUAL STUDY
Five City
0.05
(0.003)
4.17
(0.26)
0.39
(0.025)
1.43
(0.009)
0.16
(0.01)
0.21
(0.013)
0.0004
(0.00002)
0.23
(0.014)
10.1
(0.64)
16.73
(1.05)
Southeast
Chicago South Coast
0.0186
(0.047)
0.0032
(0.008)
negligible
0.0028
(0.008)
0.174
(0.44)
0.024
(0.06)
0.194 28.6
(0.494) (2.62)
NATIONWIDE
RANGE3
0.07-0.72
11.3-62
0.01-16.8
0.009
0.31
1.9-21.6
2.4
3.1
0.005
3.4
106-154
23-110f
129-2649
119-629
POINT
ESTIMATE13
0.5
48
0.01
0.01
0.31
2-22
2.4
3.1
0.005
3.4
106-154
60-80f
166-2349
124
B-89
-------
Footnotes to Table B-17.
NOTE: Numbers in parentheses are annual cancer cases per million population. For nationwide
studies, annual cancer cases were divided by 240 million, unless otherwise noted. For
studies with smaller geographic scopes, the annual cancer cases were divided by the study's
population.
8 The numbers in this column were calculated by taking the lowest and highest incidence rates for
a source category and multiplying it by 240 (1986 U.S. population in millions). The total for
this column is the summation of the low end of the range and the sum of the high end of the
range.
b The numbers in this column present the results of the reduction analyses. In most instances, a
point estimate of nationwide annual cancer incidence was derived for each pollutant/source
category combination. In some instances, a point estimate could not be reasonably derived.
For these instances, as narrow a range as possible of nationwide annual cancer incidence was
estimated, and such ranges appear in this column. The text discusses how these point estimates
and ranges were derived.
c Source categories are: phenolic formaldehyde resins; urea formaldehyde; formaldehyde
production; melamine formaldehyde; 1,4-butanediol; 4,4-methylenedianol; hexamethlenetetramm;
pentaerylthiotol; phthalic anhydride; polyacetal resin, and trimethylolpropane.
d Assumes 35 percent of ambient concentrations are attributable to motor vehicles.
multiplying risk from Ambient Air Quality study by 35 percent.
c Based on modeling of direct formaldehyde emissions.
* Excludes cancer risk from "secondary formation" source category.
9 Includes cancer risk from "secondary formation" source category.
Calculated by
B-90
-------
Of these four studies, the Ambient Air Quality study contains the
most complete set of ambient-measured concentration data. Some of the
earlier data collected apparently were sampled using older sampling
techniques that are now known to bias the data, overestimating ambient
concentrations. Recently collected data, which are used in the Ambient
Air Quality study, show that ambient-measured concentrations may be
approximately one-half to one-third of the average concentrations
measured previously. Based upon the new set of formaldehyde
concentration data, the Ambient Air Quality study estimates 124 cancer
cases per year. The techniques used to obtain the samples and the
concentrations reported in the South Coast study were not identified in
the report. The data used in the South Coast study, however, came from
samples collected between 1980 and 1984. This suggests that some of
these data may have been collected using sampling techniques that are
now known to overestimate formaldehyde concentrations. The Ambient Air
Quality study's estimate of 124 cancer cases per year is selected as the
best nationwide estimate of risk from among the studies that based their
risk estimates on ambient-measured concentrations.
Modeled Estimates. As seen in Table B-17, total nationwide cancer
risk based on the modeled estimates is calculated to be between 23 and
110 cancer cases per year, with a best estimate range of 60 to 80 cancer
cases per year. Two studies, the 5-City study and the Southeast Chicago
study, calculated the difference between the cancer risks estimated
based on selected or measured concentrations and the cancer risks
estimated based on the modeled concentrations, and assigned the
difference to a "secondary formation" category. When the risk estimates
B-91
-------
for "secondary formation" are included, the total risk based on modeled
emissions range from 166 to 234 cancer cases per year.
Of the individual source categories, the largest discrepancy
occurs with estimates of risk from primary (direct) formaldehyde
emissions from motor vehicles. Three of the four studies estimate a
cancer rate of approximately 0.2 to-0.26 cancer cases per year per
million population. The Southeast Chicago study estimate is about one-
fifth (0.047 cancer cases per year per million population) of this
cancer rate. Part of this difference appears to be due to the
particular vehicle mix and/or average speed in the Southeast Chicago
area that led to lower average hydrocarbon emissions and to lower
formaldehyde emissions. It has been estimated based on information in
the Southeast Chicago study that a comparably based formaldehyde
emission factor of between 0.011 and 0/033 g/mile was used, being
approximately 25 to 75 percent lower than the emission factors used in
the Mobile Source study. Adjusting the Southeast Chicago cancer rate
for this difference in emission factors results in an adjusted cancer
rate between 0.06 and 0.19 cancer cases per year per million population.
Different models used in the two studies may explain the remaining
differences.
Point Estimate. As noted above, ambient-measured data can
directly account for formaldehyde that is the result of secondary
formation, whereas models can not. Thus, risk estimates based on
ambient-measured concentrations are to be preferred. Of the studies
that estimated risk using ambient-measured concentrations, the Ambient
Air Quality study had the broadest geographic data base, which is
preferred for nationwide estimates. (The three individual studies that
used ambient-measured concentrations more reasonably show the city-to-
B-92
-------
city variation that may be associated with formaldehyde.) Based on the
recently obtained data in the Ambient Air Quality study, the estimate of
nationwide cancer risk is estimated to be 124 cancer cases per year due
to exposure to formaldehyde.
B-93
-------
Gasoline Vapor. Eight studies examined risk from exposure to gasoline
vapors (see Table B-18). Sources of gasoline vapors were identified as
vapor displacement due to the refueling of motor vehicles, the transfer
of gasoline at bulk terminals, bulk plants, and refineries, and TSDFs.
One study, the Southeast Chicago study, also identified evaporative
gasoline vapors loss from motor vehicles. All of the risk estimates for
gasoline vapors are based on modeled ambient concentrations.
Modeled Estimates. The primary study on gasoline vapors is the
Gasoline Marketing study. Table B-19 shows the breakdown by sources
within the gasoline marketing source category as estimated in the
Gasoline Marketing study. Several of the studies (e.g., the Mobile
Source study) appear to have incorporated the results of the Gasoline
Marketing study. As seen in Table B-19, cancer risks are shown for both
total gas vapors'and for the "C6 and higher" fraction of gas vapors.
Some evidence suggests that it is the C6 and higher fraction of gas
vapors that is the carcinogenic portion. At this time, it is EPA's
policy to report both numbers with equal weight until further studies
suggest whether risks based on total gas vapors or on the C6+ fraction
are preferred.
As noted above, the Southeast Chicago study estimated risk from
evaporative gasoline vapor loss from motor vehicles. This risk was
estimated, in part, by treating evaporative emissions as equivalent to
gasoline vapors and estimated the risk using the cancer risk factor for
gasoline vapor. The Office of Mobile Sources, however, states that "the
composition of totally vaporized gasoline is markedly different from
evaporative emissions" and that "the majority of evaporative emissions
are C6 and lower." Thus, the estimate provided in the Southeast Chicago
B-94
-------
g
Q.
LU
O
O.
£
<
z.
>-
3
. 1
-J
s
a
0
S
,
-Q
LU
1 >
O
a. i
en
LU
ra
LU
0
or
4J
to o
ra o)
CU CO
x: u
0 CJ
CO
CD
c ra
to £-
CO CO
a. o
UJ
"~*
ra
a. »
gcu
^
M ra
r-
Q.
4-"
P
>
U-
1 ,
t! > c
!c ^ o
K- CJ
CO
U-
g
CO
o cu
o
- t-
J3 3
0 0
:E co
o>
c
" 4^
' CU
o ^
CO 1-
co to
C3 Z
>-
LU Of
C_> O
^5 LU
O 1
CJ
NO
S : d co
NO
§v-
ro o co
. CM
NO 0
to to t^ in
5S S§
do do
in
*-* in '^
-J- N- O IO
NJJ- c\j . ro
o o oo
o o o
NO in
N- 0 .
d d
NO IO
NO' d
N-'
o
*- o
CD O
0 O
in
^ «-
ro
o
* '
s
co f>-
NO O
1 1
« 0
CD
CO
u> cu
cu c to
c eu c-
f- 4J CU
' CU C- O CO C
o -i<: o u-
to t_ 4-> x: Q *t-
ra ra o cu co cu
0 Z E > I- OS
ff. 2:
ON CM
ro co"
\o ^^
sQ N-
O «
C> CD
gfe
o o
^
£>. ^y
ON *Q
O CD
CZ) 0
NO in
o o
-0 ro
\o o
V.X
O
NO 0
«- 0
O CD
CD CD
tn
CD
*-\ f~\
CO N-
CO CM N- CD
O *~ O
CO
£_
8.
<0
C/> CO
1 4-> CO +
< O (O \O
h- 1 O) O
O
C*
O
13
'i
o
OJ
>- c
£1 0
T3 4J
CU CO
T3
> i
=5 a
cu to
i|
CO 4-»
CD CO
CO
ra CD
o ^z
4-*
o &
c
CO "Q
U CD
T3
11
CO CU
c_
CO-S
cu
.t- to
n cu
3 to
CO O
CU t-
T3 CU
'3 c
C CD
o u
4J '
ra ra
C 3
c
t- c
O CD
CU
x:
4-*
C
o -
.t- to
ra Q.
* o
3 o
Q. CO
8.0
c !c
- CO
CO
.r- O
E cu
D)
S. fc
CO ^
cu ra
co E
ra co
CJ
.c
I_ 4-»
CD «^-
CJ 2
ra co
U CD
ll
C CO
ro i-
o
CU u-
£_
CD
CO TO
cu cu
CO 4->
xi c
S^ cu
to
£- >r-
ra 3
Q. t-
0)
c x:
*- 4-»
O
to
C- CO
cu to
-3 o
5 "c
.*o
3"g
M.C
O)
"2 c
CO CD
.C
>s4-»
I_
O H-
D) O
S e
ra §
O CO
CU
co x:
CU 4-*
x:
o
c
"E-2
CD 4->
CD
4-1 £
8 i
g ">
cu
CU 4->
x:
4-1 CO
0)
C C
CD
4J O
U
X? CO
li
ra t-
_ o
CJ
co "co
CD 4-»
£_
oo
i
o to
o c
o
CO
x:
4-1 »-
£
C
C
to
t- C
IP o
S CD
13
O Q.
1 Q.
V>
C
»
to o
3 CO Q.
CO
CU CU
0 - CO
c to cu
cu cu x;
"T3 o 4-»
c
U CO 3
C 4-> O
to x:
c
L- CO
CU CU
u cu to
C CO CO
CD CU 3
o x: CJ
4-* CO
CO t- T3
3 O
c x:
CO CU
4-*
T3 CU CU
> x:
3 1
Et_
cu
ra >» S
H- to O
o c o
o
O CO f)
4J tD -
£1-4-*
4J CU C
CO XI
cu
4-» O CD
c c cu
0-08:
G. * ro
D
CO O CO
CJ CD
» O)
CO 0) C
CD 4-» CO
CJ CO C-
ro - .c
4-» 4-» O
Cfl CO U
C CD CO
4-* ~O
4J C C
CO - CO
i s. .
c ra cu
§*** *"*
CO O 4-»
CD C CO
CO CO CD
' CO CO
S.£i
ra
CEO
o o c
*5 c"8
13 i (.}
c_ .
C I-
CU O -M C
ro ra
M- C O
o
ill
D C
co >- ra
CU '
c. o cu
cu o «
x: 4-i 3
4J ra c
u o
go +J T3
O CO tU
cu a
U O H- C-
Q. CO O CU
§4-1 CU
C 0
CO Xl C_
*J - CD
O D (A 3
to o Q. cu
Q. D)
x: co c
4J x: to to
u c-
C ra cu
- CU O) "O
co c- ra ra
i- O t_
CU H- co
-Q CO CU
g TI 4J
5 CU 3 CO
C > 0 E
CU t- t- 4-»
x: cu ra co
1 13 C O
o
4-*
cu
ra
ra
ra
to
ra
3
J
4-J
O
M- -
C C
.,- o
4-> CO
C CO
(U 3
r- O
U CO
co ra
c c
r- O
flj 4-*
..- -5
4-» TI
CO
CO
- c_
x: o
CO X
CD
4-J 4-*
CO
CO
4-*
"CD
M- CO
CD
CO
ra u
3
if
* c.
t- O
§-1
cu o
C t-
- H-
~O CO
CO O
n to
to
c cu
0 >
CO 4^
CO to
- t-
18.
ra
cu >
> cu
4-1 E
ra o
c_ t_
s.4-
CO CD
> 0
CD C
O "-
c c
o
4-1 t-
CO 0
o c
a. ra
O 4-
o
c
co ra
S to
t- 0)
CD
H- C
^ O
Q.
CD
.c ro
4-*
oj
o a.
B-95
-------
TABLE B-19
ESTIMATED ANNUAL CANCER CASES
FROM GASOLINE MARKETING
Facility
Category
Bulk Terminals
Bulk Plants
Service Stations
Community Exposure
Self-service
Occupational
TOTAL
Average Annual Incidence
Total Gas Vapor
3.5
1.4
13
33
17
68
C6 and Higher
0.9
0.4
3.3
8.3
4.3
17
B-96
-------
study is likely to be a very conservative (i.e., overestimate) estimate
of risk from evaporative emissions.
Point Estimate. Since the risks from motor vehicle evaporation
and petroleum refineries are exclusive from the gasoline marketing
source category, the cancer risks can be summed from each. However, due
to the differences in composition of evaporative emissions from gasoline
vapor, it was felt that, at this time, insufficient information was
available to include an estimate of cancer risk from evaporative
emissions as part of the best estimate. Thus, based on total gas
vapors, a nationwide cancer risk of approximately 76 cases per year is
calculated. .Assuming the risk comes only from the C6+ fraction, which
is approximately 25 percent of totally vaporized gas, nationwide cancer
cases are estimated to be approximately 19 per year. Extrapolation of
the refinery incidence rate from the lEMP-Philadelphia study to
nationwide incidence is uncertain due to the point source nature of
petroleum refineries. The effect of this extrapolation, however, is
likely to have a smaller effect on total cancer risk from gas vapors
than the total vapor vs. C6+ fraction question. Thus, a range of 19 to
76 cancer cases per year nationwide was selected as the estimate of
nationwide annual cancer incidence due to exposure to gasoline vapors.
B-97
-------
Methvlene chloride. Eleven studies included methylene chloride as a
pollutant of concern (see Table B-20). Two of the studies (the IEMP-
Kanawha Valley and the South Coast studies) had estimated annual
incidences per million population that would result in 10 or more cancer
cases per year if extrapolated to the total U.S. population. The
ambient concentrations used to calculate the cancer risk in these two
areas reflect geographic variation as seen in the ambient monitoring
data used in the Ambient Air Quality study. The lEMP-Kanawha Valley
study reported ambient concentrations ranging from 3.1 to 20.8 fjg/m and
the South Coast study from 7.7 to 17.3 /*g/m3. The Ambient Air Quality
study's data base showed ambient concentrations ranging from
approximately 0.5 to 10.0 /ig/m3. Thus, it is not reasonable to use
either of these two cancer rates to estimate nationwide cancer cases.
The Ambient Air Quality study's results are based on the. largest
data base. Based on a population weighted urban concentration of
approximately 4.0 /ig/m3 and a nonurban concentration of approximately
0.2 ^g/m3, the Ambient Air Quality study estimated approximately 5
cancer cases per year, or a cancer rate of 0.02 cancer cases per year
per million population. This cancer rate is essentially the same as
that obtained by summing individual source categories in the 35-County
and the 5-City studies. Total nationwide instances, in either case, are
approximately 5 cancer cases per year.
B-98
-------
TABLE B-20
ESTIMATED ANNUAL CANCER CASES FROM METHYLENE CHLORIDE BY SOURCE CATEGORY
SOURCE
CATEGORY
Pesticide
Production
Pharmaceutical
Mfg.
Paint and
Other
Stripping
Chemical Users
and Producers
POTWs
TSDFs
Unspec i f i ed
Solvent
Usage
Aerosol
Area
TOTALS
Modeled
Ambient
INDIVIDUAL STUDY
Ambient NESHAP/ATERIS POTUs TSDFs Thirty- Five City IEMP-
Air five Kanauha
Quality County Valley
0.0045
(0.00002)
0.04
(0.00017)
0.22 f
(0.0007)
0.14 0.0012-0.003
(0.00059) (0.012-0.03)
0.03
(0.0001)
0.07
(0.0003)
0.037
(0.00078)
0.85 0.33 0.0012
(0.018) (0.02) (0.012)
0-4 0.03 0.07 0.89 0.33 0.0024-0.0042
(0.0017) (0.0001) (0.0003) (0.019) (0.02) (0.024-0.042)
5
(0.02)
B-99
-------
TABLE B-20 -- concluded
ESTIMATED ANNUAL CANCER CASES FROM METHYLENE CHLORIDE BY SOURCE CATEGORY
SOURCE
CATEGORY
Pesticide
Production
Pharmaceutical
Mfg.
Paint and
Other
Stripping
Chemical Users
and Producers
POTWs
TSOFs
Unspecified
Solvent
Usage
Aerosol
Area
TOTALS
Modeled
Ambient
INDIVIDUAL STUDY
IEMP- lEMP-Santa Southeast South Coast
hiladelphia Clara Chicago
0.0016
(0.001)
i
X
negligible negligible
negligible
0.0013 x 0.00057 x
(0.00076) (0.00145)
0.0013 x
(0.00076)
x
0.0014
(0.0036) x
0.0066 0.0011 O.OOZ 0.386
(0.004) (0.0008) (0.005) (0.035)
0.92
(0.084)
NATIONWIDE
.RANGE3 POINT
ESTIMATE0
0.0045 0.0045
0.04-0.24 0.04-0.24
0.22 0.22
0.14° 0.14
0-0.03 0.03
0.07 0.07
0-0.35
2.9-4.8 2.9-4.8
""
0.9
4.3-6.8 3-5.5
5-20.5 5
B-100
-------
Footnotes to Table B-20.
NOTE: Numbers in parentheses are annual cancer cases per million population. For nationwide
studies, annual cancer cases were divided by 240 million, unless otherwise noted. For
studies with smaller geographic scopes, the annual cancer cases were divided by the study's
population.
NOTE: An "x" in a column indicates that the source category was considered in the study, but a
specific cancer risk for the source category was not indicated.
The numbers in this column were calculated by taking the lowest and highest incidence rates for
a source category and multiplying it by 240 (1986 U.S.'population in millions). The total for
this column is the summation of the low end of the range and the sum of the high end of the
range.
The numbers in this column present the results of the reduction analyses. In most instances, a
point estimate of nationwide annual cancer incidence was derived for each pollutant/source
category combination. In some instances, a point estimate could not be reasonably derived.
For these instances, as narrow a range as possible of nationwide annual cancer incidence was
estimated, and such ranges appear in this column. The text discusses how these point estimates
and ranges were derived.
0 Does not include extrapolating the incidence rate from the lEMP-Kanawha Valley study to
nationwide cancer risk estimate because of the uncertainty as to the type of facilities being
modeled.
B-101
-------
Perchloroethvlene. Fifteen studies included perch!oroethylene as a
pollutant of concern (see Tables B-21 and B-22). Several of the studies
examined both ambient measured and model predicted concentrations.
Within a study, ambient measured concentrations were in general higher
than those predicted by the models, but in general were in reasonably
good agreement.
Based on the modeled estimates (see Table B-21) available for
specific source categories, nationwide cancer cases are estimated to be
between approximately 4 and 11 per year. Based on the ambient-measured
data estimates (see Table B-22), nationwide incidences due to
perch!oroethylene appear to fall between approximately 6 and 13 cancer
cases per year. Although one of the studies (the South Coast study) has
a cancer rate that would extrapolate to a somewhat higher nationwide
incidence of 10 to 13 cancer cases per year, the cancer rate is due to
the geographic variability of perchloroethylene and it would not be
reasonable to extrapolate to the nationwide estimate.
Point Estimate. Risk from perchloroethylene seems to be highly
variable with geographic location, though overall risk appears to be
relatively small. The scope of the Ambient Air Quality study and its
data account for this geographic variability. Therefore, the result
from the Ambient Air Quality study, 6 cancer cases per year, was
selected as the estimate of nationwide annual cancer incidence due to
exposure to perchloroethylene.
B-102
-------
o
i
z
UJ
CJ
§
CJ
0
Ul
UJ
^
o
UJ
B
UJ
CJ
Z3
o
CO
CO
CM UJ
CO UJ
t
UJ >-
1 3C
~
§
CO
J
§
Q
z
C
.c »T o
1 O
t_
o o
4-» T-* 4~»
to o 3
3 ' 'i
«
s
CO
*"""
O) CO
TJ t_
cr> to
o <^x Q
«- o m o M -
f O O -* O D>
O O ...,._
1*^ O O ^ O '
^x \»^ D)
O
C
OJ
OJ
.
0)
c
coc?
CO O
CD CD
***
?
CD
O CP
f ^^
O O
0
o
o
"
N- OJ
ss
O 0
CD
0
V O
O 0
O 0
CD CD
i i
CM O
O O
O O
«2
^^
1
' L.
^ O> D) CO c/3 4-1
0 M- T3 t- 10
<1> 5E 3 O D> -3
C t-i c CO
TO ' c/3 (U -r- «^- D) .*- CO O> CO
CO [_ 4JCOOD) CI-D)Ct-
^^u coo c^*ca c f> *-» c o
CD n- CO O)C CO CUCUCU CUfr- c CO'« ^4-*
Op U. CO"- 3 >O)i-4->C CO D4-» CCO
WCU Q 3O 1 CO O) COC >sd) "OCO -t-t.
^x: c/? cue o ococu COD t- ' co £_a>
ECJ \~~ ) . Q. cnzjQ 3m ocj ~o Q<
in
O CM
* o
O CD
CM
(O M
in o
«- CD
V
'ra
»r
_*
CD
cu
c
^
§1
_ o
CD O
^
o
0 0
«- o
o CD
^^
Kl
N- 0
o o
o
o
o
N- CM
^d1 CO
CD O
O CD
CD
O
V O
^
Ov O
Q* O
O O
NO «-
CM CD
0 O
CD CD
O
O
O
TOTALS
Modeled
B-103
-------
I
S
Ul
UJ
o
3
§
i
I
H-
w
1
>
§
o
UJ
*- £
g £
CO
UJ
\
*J CO
3 CO
0 0
co o
4j
U) O
CO CD
0) CO
JC O
13 x:
o u
en
CO
t_
CD
1 *
0. U
UI CO
"I
tn
CO
^f
ui -a
""
a.
i J? X
z: eg
ui c *
H- CO CO
^_
UJ O£
~j UJ
u
N-
§ ^ ^ r, ~ . I
V OOOK1 OtO 0
8: - t;
? §°.?fro£.S
s o o g s K; ' o
^ « ^
o o
oj 0
h- in ^Q -Q
in -d- - *g:
§§ ^ r x x
O) O)
00 0 U
X XX XX,
CD
1 ll I! S5
.F- ^ ^ « ' '
^ o o o o
as o o ^ ^
c
o
CM CM
O O O
O H-
O
O ^-- D)
C
c
1 1_
w o> ,9 ?i
DO) CD M <" *:
OH- T3 I- > 3
n_, cnra .E^ cn.r-Mcnu)
_jfn t_ +JC/)OCT> Ct-C3JCt_
«(j OO C***C0 C n~*JC"-O
MCU Q 3O 1 CO CD O> <- >*-« £u C/53Q 3CO OO <(_) O
-------
>> >
"° 1i
tt) «
TJ
> JZ
. r- +J
«*
L_
$"8
TJ
CO -r-
CO
COTJ
CU
3:
cu co
Q. tt)
« "^
tt) 3
CO 4-«
ro CO
o
£_
t- o
tt) U-
o
c
to
| 8
to co
tt) 3:
CO tt)
« ^
tt) O
CO
0) CO
JZ CO
4-> tt)
c
£ S
c o c
CO ' 4->
a> 'i JS
log.
3 vd- o
3= CM Q.
..
o
O>
to
u
0)
o
£_
o
CO
tt)
JZ
4-*
L.
o
H-
co
L.
t^
CU
o
c
(0
u
o
H-
0
8.
CO
CO
3
o
1
CO
CD
JZ
4-»
C
I
tt)
TJ
CO
c
o
o
CO
to
*
£*
o
D)
tt)
4-»
to
o
tt)
0
t_
o
CO
tt)
JZ
CO
J=
CO
0)
to
u
TJ
c
ll
o to
u u
to TJ
c -
=x 8
c. to
cu
-Q ro
CO CU
O D-
<-* co ro
co ro
ECU CO
1- CU
C 05
CO CU C
J3 CO
cu c.
3 8 "S
O "D CO
ro o c
C U CO
>»- cu -
O 4-> TJ
co cu
CU E 4->
CO 4-* E
E CO
- CU 4J
4-> CO
CO 4J CU
cu c
t CO
4-» O CO
C Q. 3
O CO CU
Q. 0
ro si
^ o *-
CO C O
CU CO C
O 4->
C CO
CO C t-
co u
C 4) CTJ
- E ro cu s
0 0 >
M CO "-
CO L.
o c ro cu
E g TJ
c c cu
* ro c_
c cu
o cu 3:
CO 4-" *- CO
cu ro 3 tt)
' J3 -^ CO
C O CO
ro o c TJ
c >*H- ro
O t- O
- O CO
M U) CU CU
0 CU -W
TJ ro s
CU O CO T-
t~ CO 4->
CU 0 CO
cu o u. cu
J2 t_
4-» 3 CO 4-»
O CO C
M- CO -
o ^ o ro
ro c co
+> cu cu
C CO CO
cu t_ ro co
CO O D
OH- O
C- CO CO
Q.TJ O
CU O TJ
§> C
^ CO -M
CU CO CU
O TJ C 4->
o
CO CU
co ro cu jz
3: co t
J= CU
M CU JZ
O -tJ
c c c
cu i. E
co ~° £ -2
t- O O
cu c o
e "D co
C CU > j=
O - 4J
(1) C C_
JZ CO CU C
1 O TJ
4J
CO
o
cu
JZ
CO
4-*
o
cu
H-
cu
£_
cu
M
1
tt)
to
-M
H-
o
0
O)
£8
t- o
O <1>
cu "cu
4J CO
E tt)
- £_
4J tt)
s *
CO
tt) tt)
ag
c to
^- t_
CO
L.
4-» O
CO
JZ CO
3C 0)
4.1
o to
tO 4-»
(0 CO
to
.,- 0)
CO
(0
C CO
to
tt)
tt) CO
O 3E
o
CO JC
CO
3 CU
CO CO
tt) CO
L. U
u
tt) CO
4-» T3
O X
tt) tt)
_* 4-J
H-
tt) tt)
C
o to
o -
CO
-C O
cu
C 4J
CO *
(_ *-«
fto
cu
c
cu
tt) .M
1 »
ro co
E to
O TJ
4-» CJ CU
CO 4-r
tt) jz ro
tt) 3 "5
> O 0
.- GO >
ro tt) o
»- 4-» CO
4J -^
C TJ
cT*~
JZ D) O
4-* ro 4->
o
C "- (M
O J=
CJ «
^4J
C CO O
o ro t-
tt) M-
TJ JZ
QJ 4-» £_
CO D tO
ro o Q)
m > >.
-B-1Q5
-------
TABLE B-22
MEASURED AMBIENT CONCENTRATIONS OF PERCHLQROETHYLENE
STUDY CONCENTRATION
(jig/nO
Airbient Air Quality
Range 0.33-17.0
Urban 3.83
Rural 0.3
South Coast 6.8
Cancer Cases by Study Nationwide
Per Year Per Million Cancer Cases
Population Per Year
6 0.025 6
0.59 0.054 13
lEMP-Kanawha Valley 1.0-3.4
lEMP-Baltimore
Range 1.5-9.3
Average 5.51
lEHP-Philadelphia 4.7
Totals
Range
Point Estimate
0.06 0.038 .9
0.06 0.039 9
6-13
6
B-106
-------
PIC. "Products of Incomplete Combustion" (PIC) is a term used to refer
to a large number of organic particulate compounds that result from
incomplete combustion, such as may occur from gasoline- and diesel-
fueled motor vehicles. These organic particulate compounds consist
primarily of polynuclear organics, or, synonymously, polycyclic organic
matter (POM). POMs would therefore be considered a subset of the
compounds termed PIC.
Polycyclic organic matter, in turn, is a generic term that covers
hundreds of chemical substances that contain two or more ring
structures. Compounds covered by the term POM include: (1) compounds
composed only of carbon and hydrogen, which are known as polycyclic
aromatic hydrocarbons (PAHs); (2) compounds with a ring nitrogen (aza
and imino arenes); (3) oxygenated species; and (4) nitrated and
chlorinated POM, including dioxins and pesticides such as aldrin and
DDT.
Polycyclic aromatic hydrocarbons (PAH's) can be divided into three
compound categories: (1) naphthalene; (2) the anthracene groups; and (3)
the benzo(a)pyrene (BaP) group. The individual constituents of the last
group include BaP, acenaphthylene, benz(a)anthrancene,
benzo(k)fluoranthene, benz(g,h,i)perylene, and indeno(l,2,3-c,d)pyrene.
Risk Estimation. Twelve studies include risk estimates for PIC.
A total of four different risk estimation methodologies were used.
These studies and the risk estimation methodologies used in each study
are shown in Table B-23. The most frequently used methodology assumed
that all of the risk from PIC can be adequately represented by using BaP
emissions as a surrogate. This methodology uses measured or modeled BaP
emission concentrations and applies either (1) the BaP unit risk factor
or (2) the PIC unit risk factor to those concentrations to calculate
B-107
-------
S
u
CL
g
u.
O
w
Ul
1
g
g
111
£
g
P
j:
UJ
VARIOUS
CA O
CO O)
.
r JT 0)
jg CO -*
-l>
ej
>
u.
>-«Br
£;*; g
f CJ
H-
CA
S
1
t L.
S.0,2
J~ ^J fA
^- CO ^
Z CJ
cu w
i i?
£ O
ta
1 1
' C/l
"ca O ^
s s
CJ
4_( ^x
S4->
C-
i'5l
§
i
X
X
X
X
X
X
g
CO
t_
oS
4-* O
O CO
ra H-
H-
\y
emission
unit risl
EX CX
CO CO
CO CO
CO
X
X
X
X
I- O
O 4->
4J U
U CO
CO M-
emission
unit risl
CX CJ
CO
CD 0.
X
X
'
X
X
4-1
O- t- "-
CO O C
4-1 CO 4-1 3
c ~a ^ c o
0) C O 0 4->
o ra ex"cA 8.
Q. CA CA
e t_ t- - o
O O O E >
o *-* o "
CJ CA 4J
CX CO c- TJ CO C_
CO H- O CO C- O
CO 4-* 4-< CO 4-»
^ C O CO Q. CJ
s: o co - g co
O 1- O O H-
0. CA 0) U
^ CA ^^ Q. ^£
O CA CA TJ CA
CL CO tl D CO ^
X
X
X
CO
£_
o
4-*
0 t-
co o
S -2
CA ^
CA CA
4-*
|'l
IE 0-
< (U
Q_ CQ
G)
(0
'i
(0
M
8
0)
t_
0)
CO
o
u
(0
H-
s-
-------
risk. (Note: The 35-County study treated these two approaches as non-
duplicative and summed their results to give an estimate of total risk
from all P-IC compounds.) For example, suppose an ambient BaP
concentration of 1 //g/m3 is measured. A population of 100,000 people is
exposed to this concentration for 70 years. Applying the unit risk
factor for BaP of 1.7xlO"3 (//g/m3)"1 yields an estimated 170 cancer cases
over 70 years, or approximately 2.4 cancer cases per year. This
methodology assumes that all of the risk from PIC is attributable to
BaP. In other words, none of the other components have any cancer risk
associated with them. Suppose instead the PIC unit risk factor of
4.2xlO"1(from the Six-Month Study) was applied to this measured ambient
concentration. Estimated cancer cases from PIC in this example would be
42,000 over 70 years, or 600 cancer cases per year. The method used to
calculate the PIC unit risk factor reported in the Six-Month Study was
unusual and any risk estimate based on its use should be treated as a
very preliminary estimate.
A second variation involving BaP was to use specific PAH/POM
emission factors specific to a particular source category to estimate
concentration.levels of PIC and then apply the BaP unit risk factor to
estimate risk. This methodology, which is separate and distinct from
the first two identified, assumes that the average unit risk of all
components that make up the modeled concentration is the same as the
unit risk factor for BaP or that each component has the same risk value.
Another methodology uses individual PIC component emission factors
specific to a particular source category to estimate the concentrations
of the individual components within the PIC mixture and applies to those
concentrations the corresponding unit risk factors for those components.
B-109
-------
This technique allows for variation in the overall unit risk factor that
is estimated for specific source categories.
A similar methodology is known as the comparative potency factor
approach. This approach involves using an emission rate for particle-
associated organics (as an unspeciated mixture) and a unit risk factor
for these organics as an unspeciated mixture. This approach has been
used, for example, in estimating risk from diesel emissions. The unit
risk factor for a suspect human carcinogen (e.g., diesel emissions) for
which there are no epidemiological cancer data is estimated by
comparison to a known human carcinogen (e.g., coke oven emissions);
the risk associated with the known human carcinogen is multiplied by
bioassay potency of the suspect human carcinogen divided by the bioassay
potency of the known human carcinogen. (A variation on this methodology
is to use particulate emission factors and comparative potency unit risk
factors adjusted to reflect the particle-associated organic fraction.)
Table B-24 summarizes the risk estimates from the ten studies,
broken down by source category. Two of the ten studies estimated risk
based on ambient measurements; the others based their risk estimates on
modeled concentrations.
Ambient Estimates. The two studies that used ambient measurements
were the lEMP-Santa Clara study and the Ambient Air Quality study. The
lEMP-Santa Clara study estimated cancer risk by scaling other national
ambient concentration data for PAH's from similar urban areas to
estimate PAH concentrations. Using EPA's unit risk factor of 1.7 x
10"3 (jtg/m3)"1 for BaP, cancer rates between 0.004 and 0.49 annual cancer
cases per million population are calculated. These rates correspond to
a PAH (BaP group) concentration of 0.00016 /tg/m3 to 0.02 ^g/m3. These
B-110
-------
o
CD
111
\ ,
CJ
LU
CJ
0£
|
^_
m
o
rx
g .
Qi
U_
GO
LU
Sr
(j
o
1
ID
Z
^
2
1
h-
LU
>_
O
to
(
^
0
o
(0
i-S fr
X ra -T-
UJ C '
&
u
U»
w o 5"
^ ,- 3
CO
h
to
t t_
CD O
Q. CD, 4->
'o M 3
1-1
I O
'"
01 O
o
- t_
Xt 3
0 0
X c/>
CO
CO C_
3 O
CD CD &
' x r o
5 §
"2 '£
C CO
03 ^
"§
o ej
ej
S < "co
E 3
< a
III fV
o o
g>f O
s <
CD
Sr? 5 2 ? «*
§0 °.° .? g°.
oo °s -s.de
^ 0
C
OjS ^^ p
^3 CD LO hv CVJ fNJ ^^
d^ *"o J-oo1^: ^S
°- ~ ^
C\l ^ *U *"* ^3 C3
o v- ^_^ m «- co
_ _ V CNJ ^W 00 "f V
xx o i *- . o o ro
«- r- LO 0 CO «-
^ o tn T i o mo
s»x
c-i->Q.*j>co - coc-cao
° .0 3 C5 ° CD 3 0 0
CJP^U^: 300=3
in *-
CM O
0 O
o
o
0
^^
CM
O 0
o o
o o
So
o o
o
O 0
o
o
4-1
CO
CD
CO
CO C-
3 O
O V
"D co
CO t- 3
u- CD Xt
O ME
C/} CD O
i z: cj
B-m
-------
g
Ul
1
5
0
g
-g »
5 u
r- QU
I 5
« 8!
1 CO
UJ
«? is
CD
CC
111 UJ
Se g
_,
^
o
Ul
1
~
8
>-
1
CO
IVIDUAU
o
£
(U
cL-ifr
ass
CO CO
-£ >
O
* X
4J CU 4J
- «^« 3
JZ H- O
1- 0
U.
o
to
_. "
ra o
Q. O *-»
£ CJ
CO
O 01
'§1
Z CO
U)
CA C-
=> O
^ 2 3
CO CO Ji
N ca g
31 U
'5.1
"§ 1
_^ 5
CO O
o u
u
v >.
C 4-1
.2! .^ 2
t" < g
a
UJ O£
O O
O UJ
O
CM *O
%* CM O in
Kl O CO C3
o o o o
in co
*"
o o
"cu i CO >- O
tO CM
o to
0 O
0 O
in in
CM
o «-
in
i i
h^ ^~
^
in *-*
CM *
O 0
o
**
to
o
MD 0
O CJ
1 1
0 0
o
o o
o
o
>O CM
O
* i
i tn
«- CM
O
to «-
^
PJ
o o
C> C3
i i
O O
0
o
o
\o
v- O
o
^^
in
"^o to
CO N-
D) O
s
§ 5
S S
1 i
CO
-------
"B ^
a.
o
s §
<
E
I
CO
.a
UJ
1
1 <
Z :E
O i^
a. co
UJ
LU
O
3 CO
O S
1 <
< a:
z
O) CO
CO <0
0) CO
o a. o
UJ CO
*~* c
(D
uj or
CJ O
cog
-0
o
1 O
to «- %o
* o eg as
i <3 i
00 g
in ^ >o o
CO > O
0 C 3 CO .S -S
o ' t- jc * cy
T3 - CO O O O to
C 4-> Q. 4-> > to (1)
CO to -f- to (0 >~
13 O 3 t_ C3 p
"(0 ~E 'c S 2 i ,
O 5 3 O O
O CJ 3E O Z
00 >O
$
CD
in o
CM CM in
Q * c\i o
Q 'T' °. ±
i in o ?
o in o
in fo *- o
o «- - r\j
o Q o o
(P
__ ' pe
(o -r' a> to t.
4-> --, > 3 O
O O O) O> O O 4-*
*J C C * TI to
J2 0> in- W tO <- 3
3 we *j"n u. co^j
CO tOt- (OOQ N£
(03 0) O CO (0 O
3m z 3 i-i- z u
B-113
-------
I
S
o
CL
CO
UJ
C/3
ca
ui
S
UJ
a
i
i
i
»-«
§
POINT .
ESTIMATE0
CO
UI
1
11)
cn en
"§ t-
3 O
l_
11
CO
Southeast
Chicago
P
C.
03
i
t i -!-»
i
CO
LU C£
a2 111
0
in in
v ro
in «
-* CM "
ro in ro
O T-
rvj o
in o
^ o
4-> § t-
co 3 o
* 4->
10 03 QJ
§(0 - CO
20 J±
t. 0) C JJ
«- CO i O
CO O
CM CM JO
o T-* r^-
,i - - CO
,11 CM *- «- i
i C> 1 I ^*
>O CO
.» ro
ro -*
10 o ^o in
rj ^ ro ro CM \p
o ro o o i" r~-
s* « « i T « co
10 > co i i '
to ^O o in *~" ^O *^
CM >o ro ro
tfl CO CO
o ' '
C- 0 0
i '£ « t "2 -&
C/l O C > O
jj ^ 1- 0 ^ h-
.^ O '-M O
O +- 03 -M
0. 0 -M 0
U- jr flj
UJ 0 >
13 C3 Q. -P- 4->
C « Q. 4^ - t-
f Qu <0 TJ 03 C O
LL. 3 "O S- ^* *n
o; ^ C u- E O u-
D a<0 LU 5 C^
0 ^ U. S5 4-" (0
fttu. OS S t- 0 -
CO UJ 13 CV O Q. Qi
CM O
in o
v* O
O sf
o c>
i i
cn - x^
S ~
3 "J
UJ '"'
CO
1
o
B-114
-------
»n population. For nationwide studies, annual cancer cases were divided by
taller geographic scopes, the annual cancer cases were divided by the study's
«j t
.p- tO
' -C.
4-1
2
£_
-^^
?"
CO -^
u o
CD '
O
CD 1
Sc
§"~
c
o
O 4-»
U (0
2 U
-C 0
4-» Q. CU
D)
C C
'- e/3 ra
CO 13
f , (U
CU O TJ
JZ -* C
1 (M CU
on analyses. In most instances, a point estimate of nationwide annual
ry combination. In some instances, a point estimate could not be reasonably
of nationwide annual cancer incidence was estimated, and such ranges appear
and ranges were derived.
- O CO
4-> D> CU CU
O O ' 4-»
3 v & ro
*u ro - E
CU 0 CO
I- CO 4J
CU O CO
CU U Q. CU
.C t-
4J 3 0> 4J
o ro c
H- CO '-
O ^ CU O
4-1 D) CL
CO C C
4-* ro ro cu
' 4-» t_ CO
33 CU
co ra _c
CU ' 4->
i- O 3
O. O 2
o
CO CU
co co tu ^:
"~ 3E tO 1
The numbers in th
cancer incidence
derived. For the
in this column.
Mobile Source study only, the cancer rate was calculated based on an urban
t_
o
u_
t_
O
4-1
U
CO
M-
^
CO
t_
4-*
S
-Q
C
as
L.
o
4-*
O
ro
M-
c
o
CO
CO
Q_
Based on using Ba
population only.
CO
i
CU
4-4
ro
t_
c_
cu
u
c
ro
o
o
.c
1
f
4-*
CO
CU
u
£_
0
CU
!5
o
L.
o
u_
1
1
o
X
CM
M-
o
£_
o
u
ro
s-
CO
4-*
Si
oS
CD CD
o CL
CJ S.
CO
s-
c
g^
CO 3
CO
"- C
E ro
c
Q_ O
Based on using Bal
calculated based
t risk factors.
c
g
ro
CO
o
o
ra
g
emissii
4-*
C
CU
I
u
0.
ro
m
o
Q.
CL
C7>
C
CO
3
C
O
g
CO
CD
m
i
H-
o
8.'
CO
O
C
o
t_
§
CO
Industrial point :
c
o
4->
CO
t_
4-»
c
CU
U
c
o
o
Q_
CD
O
o
4-»
U
CO
CO
c.
ro
m
g
'>.
a
CD
g
"8
CO
ro
m
BaP concentration.
o
4.1
"S
D)
a.
^_
o
X
OJ
»4-
O
L_
O
O
ro
<+-
c
Cu
Based on applying
eated by assuming the 5-City study identifies all cancer cases attributed to
e range (5 cancer cases per year) is created by assuming total nationwide
per year per million population) times the total U.S. population (240
(_ _r-
O 4-»
CO **-
, o
e~H
CO CU
tu
>- £-
t O.
ill
to cu
tu jc
to t-
(0
u
t_ CO
CJ
co e
o
-sf tU
ro
-------
estimates were included in the lEMP-Santa Clara study to provide at
least a rough estimate of cancer risk rather .than ignore this pollutant
altogether.
The Ambient Air Quality study estimated risk from PIC based upon
1986 and 1987 ambient BaP concentrations. These data were used because
the 1982 through 1985 data were found to have a positive bias because of
some unknown contamination. The 1986 and 1987 data are higher than the
1977 through 1980 data, but are significantly lower than the 1981 and
1982 data. The 1985 Ambient Air Quality study on BaP and PIC used the
1977-1982 period. (Since the lEMP-Santa Clara report was published in
1986, it is possible that at least some of the ambient data used in that
study came from 1977-1982 period.) Using the BaP concentration
(approximately 0.0006 ^g/m3) as an estimate for PIC emissions and then
applying the Six Month Study's unit risk factor for PIC, 876 total
annual cancer cases due to PIC were calculated. Assuming all of the
cancer risk from PIC is due to BaP, the Ambient Air Quality Study
estimated 4 cancer cases per year nationwide.
Of the two ambient-based estimates, the Ambient Air Quality study
was selected as the better study from which to estimate nationwide risk
than the Santa Clara study. This selection was based on consideration
of the Ambient Air Quality Study's broader scope and use of more recent
and, presumably, better ambient data. In addition, applying the lEMP-
Santa Clara cancer rate to the national population would not be
appropriate as the estimated ambient concentrations were calculated
based on emission sources specific to Santa Clara. Based on the two
methodologies used in the Ambient Air Quality study, a range of 4 to 876
cancer cases per year due to PIC is estimated.
B-116
-------
Modeled Estimates. The other ten studies calculated ambient
concentrations using models. Nine specific source categories and one
"unspecified" source category were examined in these eight studies. Of
the nine specific source categories, the most important contributor to
cancer incidence is motor vehicles. Woodsmoke/woodstoves as part of the
"heating/woodstove" source category are the second largest contributor.
The other seven source categories appear to be relatively insignificant,
totalling less than 8 cancer cases per year.
As shown in Table B-23, these studies used a variety of methods
for estimating risk. In selecting estimates of cancer incidences per
year per million population with which nationwide estimates of cancer
incidence would be made, the cancer rates derived from unit risk factors
based on the carcinogenicity of the entire PIC mixture were favored over
those cancer rates derived .from either assuming the entire cancer risk
from PIC is attributable to BaP or using the Six-Month Study's unit risk
factor for PIC. This was done because it was felt that the unit risk
factors estimated for the PIC mixtures are an improvement over the other
two approaches. In any event, the reader is reminded that the unit risk
factors for specific PIC mixtures have not received the same level of
scrutiny as for other pollutants and that all cancer risk estimates for
PIC remain highly uncertain. The following paragraphs discuss the
source categories and their estimated risk from PIC.
As noted above, motor vehicles appear to be the most important of ,
the nine source categories associated with PIC. Five of the ten studies
estimated risk from motor vehicles. Table B-25 summarizes the unit risk
factors, annual cancer cases, and annual cancer cases per million
population for this source category in the five studies. The 5-City
study uses emission factor data provided by EPA's Office of Mobile
B-117
-------
TABLE B-25
ESTIMATED ANNUAL CANCER CASES FROM PIC FROM MOTOR VEHICLES
STUDY
35 -County
Mobile Sources
5-City
Southeast Chicago
lEHP-Kanawha Valley
Pollutant
Emission
Factor
POM
BaP
BaP
BaP
BaP
Particle
associated
organics
Particle
associated
organics
POM
BaP
Unit Risk Factor
Gasoline: 5.4E-04
Diesel: 6.6E-06
BaP: 3.3E-03
PIC: 4.2E-01
BaP: 3.3E-03
PIC: 4.2E-01
Gasoline: 2.5E-04
Diesel: 2.0E-05
to
10E-05
Gasoline: 1.2E-04
7.9E-04
Diesel: 3.0E-05
BaP: 1.7E-03
BaP: 3.3E-03
Annual Cancer Cases
Gasoline Diesel Total
56
0.8
102
1.3 -- 1.3
122 -- 122
163-176 -- 163-176
178-860 178-860
341-1,036
7.9 -- 7.9
11.3 11.3
19.2
0.053 -- 0.053
0.0028 -- 0.0028
Annual Cancer Cases
Per
Million Population
1.18
0.02
2.16
0.007a
0.68a
0.68-0.73b
0.74-3.58b
1.42-4.32
0.5
0.71
1.2
0.134
0.028
a Based on urban population only (180 million), as reported in the Mobile Source study.
b Based on urban (180 million) and rural (60 million) populations, as reported in the Mobile Source study.
B-118
-------
Sources, which were used in the Mobile Source study. The Mobile Source
study uses more recent PIC emission factors than used in the 35-County
study. However, due to different unit risk factors for gasoline and
diesel particulates, the 35-County and Mobile Source studies result in
nearly identical estimates of cancer incidences per year per million
urban population. In the Mobile Source study, the estimates for cancer
risk from organics associated with gasoline particulates using BaP
emissions and the 1985 Six-Month Study's PIC unit risk factor results in
an urban cancer rate of approximately 0.68 cancer cases per year per
million urban population, which is essentially identical to the urban
cancer rate calculated using emission factors for gasoline particle-
associated organics and a unit risk factor for these organics. This, in
turn, is the same as that found in the 35-County study, where the only
difference is in the emission rate.
The lEMP-Kanawha Valley and the Southeast Chicago study use data
more specific to their locales. In the case of the lEMP-Kanawha Valley,
information of the emission rate used to estimate BaP emissions was not
available. The Southeast Chicago study used an emission factor for POM,
which is approximately 55 times larger than the BaP emission factor used
in the Mobile Source Study. When the same emission factor and unit risk
factor are used, the resulting annual cancer incidence per million
population between the two studies are the same.
In summary, the results from the Mobile Source study seem to be
the best national estimate for risk from motor vehicles. The
differences between studies seem to lie mainly in the assumptions
concerning emission factors and unit risk factors, although different
models were used. Among the estimates of risk reported in the Mobile
Sources study, the best estimate of PIC risk from motor vehicles was
B-119
-------
selected as that estimated using the unit risk factors estimated
specifically for diesel particulates and'gasoline particulates. For ,
organics associated with gasoline particulates from motor vehicles, an .
estimate of 163 to 176 annual cancer cases is selected. For diesel
particulate, the range of 178 to 860 is selected, because of the
inability at this time to select a more likely unit risk factor from the
range reported in the Mobile Source study.
Woodsmoke/woodstoves were estimated to be the second largest
potential source of risk from PIC. Estimated cancer rates ranged from
0.018 to 1.01 cancer cases per year per million population, with
nationwide annual cancer cases ranging from 55 to 242. Four studies
estimated risk from this source category. Two of the studies, the 5-
City study and the 35-County study, estimated risk using unit risk
factors for the PIC mixture. The estimated cancer rates from these two
studies using these unit risk factors were 0.3 and 0.24 cancer cases per
year per million population, respectively. The Southeast Chicago study
estimated concentrations of the full class of POM compounds, and then
estimated risk by multiplying the POM concentrations by BaP unit risk
factor. (As that study noted: "While some POM compounds are probably
more carcinogenic and other POM compounds are less carcinogenic, this
approach in effect assumes that the average cancer potency of the full
range of POM compounds equals the cancer potency of benzo(a)pyrene.")
The Southeast Chicago study, using this approach, estimated a cancer
rate of 0.29 cancer cases per year per million population.
The 35-County study and lEMP-Kanawha Valley study estimated risk
by applying the BaP unit risk factor to BaP concentrations. This
resulted in similar estimates of cancer risk -- 0.013 and 0.018 cancer
cases per year per million population. The 35-County study also
B-120
-------
estimated cancer risk by applying the Six-Month Study's unit risk factor
for PIC to BaP concentrations. The resulting cancer rate was 1.01
cancer cases per year per million population.
As noted earlier, the approach favored in this study for
estimating risk from PIC is to use those estimates based on PIC unit
risk factors for specific mixtures. Both the 35-County study and the
5-City study used this approach. Their resulting estimates of cancer
rates were similar -- 0.24 and 0.3 cancer cases per year per million
population. Applying these rates to total U.S. population results in an
estimated 58 to 72 cancer cases per year. The Southeast Chicago study
used a slightly different approach, which resulted in an estimated
cancer rate of 0.29 cancer cases per year per million population (or,
when extrapolated nationwide, approximately 70 cancer cases per year
nationwide). Overall, it was felt that the 5-City study provided a
better accounting of this source category then either of the two
studies. Thus, its estimated cancer rate was used for calculating the
best estimate of nationwide cancer risk from PIC emissions from
woodsmoke/woodstoves.
For the remaining stationary source categories, there is little
individual risk or differences in estimates of that risk. Two studies
estimated risk from coal and oil combustion. The Coal and Oil
Combustion study estimated risk to be approximately 1.1 cancer cases per
year nationwide using the BaP unit risk factor applied to BaP emissions.
Using the cancer rate estimated in the 5-City study, nationwide risk was
-i
estimated to be approximately 0.43. cancer cases per year. The 5-City
study applied PIC unit risk factors that were specific to the source
category. These two studies created a range of 0.43 to 1.1 annual
cancer cases nationwide. Because it was based on the approach preferred
B-121
-------
in this study, the best estimate of nationwide risk was selected as 0.43
cancer cases per year. In either case, the relative magnitude is fairly
small.
Only one study, the 5-City study, estimated risk from the "iron
and steel" source category. The estimated cancer cases in that study
was 0.34 cancer cases per year, and resulted from just one of the five
cities studied. Since iron and steel facilities are not limited to that
one city, a nationwide estimate of 5 cancer cases per year was
calculated by applying the cancer rate of 0.022 cancer cases per year
per million population to the total U.S. population (240 million). The
estimate of 5 cancer cases per year is viewed as an upper limit.
Finally, the other remaining stationary source categories showed
very little annual incidence or were reported in only one study (e.g.,
sewage sludge incinerators). The analysis, therefore, did not try to
further refine these estimates.
In summary, the best estimates of annual cancer cases based on
modeled estimates were: 346 to 1,028 from motor vehicles; 72 from
woodsmoke/woodstoves; less than 5 from iron and steel sources; 1.5 from
sewage sludge incinerators; 13 from "other" sources; and 1.5 from the
other remaining source categories. The total cancer risk from PIC based
on the modeled estimates is thus estimated to be 438 to 1,120 cancer
cases per year.
Point Estimate. The estimates of risk from ambient-measured
concentrations in the studies examined were based on applying either the
BaP unit risk factor to BaP concentrations or the Six-Month Study's PIC
unit risk factor to BaP concentrations. Since it was felt that the
newer approaches that use unit risk factors estimated from PIC mixtures
from specific sources are an improvement over those two approaches, the
B-122
-------
estimates of cancer incidence from PIC were selected based on the
modeled estimates using the newer approaches. Thus, the estimate of
nationwide annual cancer incidence was selected to be 438 to 1,120
cancer cases per year. This range results from the inability at this
time to select a single unit risk factor for diesel particulates.
Further, these estimates in themselves remain highly uncertain.
B-123
-------
Trichloroethvlene. Emissions of trichloroethylene (TCE) have been
identified as coming from the production of trichloroethylene, ethylene
dichloride/vinyl chloride, polyvinyl chloride, and vinylidene chloride.
The majority of TCE is used as a solvent for degreasing operations, the
largest source of TCE emissions. Other sources include chemical
distributors, POTWs, and solvent usage in adhesives, paints, and
coatings.10
Fourteen studies estimated cancer risk from TCE (see Table B-26).
Three studies relied on ambient measured concentrations to estimate
risk; the others used modeled concentrations. As seen in,Table B-26,
the majority of TCE emission sources have been included in one or more
studies.
Ambient Estimates. The Ambient Air Quality study, the IEMP-
Baltimore study, and the South Coast study used ambient measured
concentrations to estimate cancer risk. The Ambient Air Quality study
used data from 25 locations to estimate risk. Average population
weighted annual TCE concentrations of 1.50 /zg/m3 and 0.2 //g/m3 for urban
and rural areas, respectively, were used to estimate risk. The IEMP-
Baltimore study used average annual ambient data from 10 locations. The
range of concentrations was from 0.2 to 3.9 //g/m3, with a population
weighted average of 0.71 //g/m3. The South Coast study showed a range of
concentrations from 0.53 to 2.12 //g/m3, and a weighted annual average
concentration of 1.7 //g/m3. The latter two studies are best viewed as
reflecting the potential geographic variation between urban areas. For
a nationwide estimate, the Ambient Air Quality study was selected as the
best estimate.
10
U.S. EPA. Survey of Trichloroethvlene Emission Sources.
450/3-85-021. July 1985.
B-124
EPA-
-------
^
1
t
g
UJ
o
o
t/t
^_
m
LU
LU
-J
3=
V
UJ
O
0
_l
t_J
t
5
a.
u_
UJ
GO
o
os
CJ
1
<
^
3!
2
a
\
S
j
LU
Q
f
CO
^
O
~
K
U
H)
u_
4-» O 4J
I- > C
-C H- O
1 O
^ §
0 -
s
4-1
o «-
CD «- 00 »- O 0
O O I- «- O Q
O " O O O
00 ' 0 ... 0
O >O O
O V
^ O
0)
in ~ ^> Ji
CD CM CM O .Q CO ^^
« CM O * - CO N*
o So o> ' . o
0 O Q t-
o1 °d » S
«
C*
!o
*5>
^3
f
^^
in
CO o
« o
CD
d
p
x^
o o
o
~
^
P m , o in * o
Kl O -O O 0 O
0 O O
°d MS d^
§___ " *^s O>
O ft) M
g ^Oj « CO CO 3E
fljH-eo .^-o 3'* OM
cocot- H* Q o)E-jo>a3 'cc
EZD«a; Q- i 3ocooo E i
in CM
55
o o
CD Q
CM O
01
a>
CO
-------
z
s
a
2
z
^
§
CO
<
~
2
>
o
.O.
UJ
t
a; 3E
a. in
UJ
ro
UJ
ca
z
a?
4-*
tn
(D
o
CJ
JC
i
CO
*»
CD O)
Oj m
.C 0
4-> .-
o o
to
(a
C,
03
a. cj
UJ CO
"g
CO
CO
1 Q.
UJ -o
»i CD
JE
O_
31 (0 *
'^ (5 5
H)
r
fc .£l
UI *j
1-1 m
m
ti 1 o£
CJ O
ce ca
8UJ
i-
CO <
U
3 "ro ^
0 O
c
o <^ ^>
^ O O C\J CD
D) O CD O O
"~ 00 CD
tn CD CD o ^
g
xxx x x
"O 0)
m ^^ ' ,.
ro PO ja p £
o o 'en o o
^ CD - ' '
. f O O
o o en ^
*"* c
ja T- T- ro ro
o «- o o
O> 0 O 0 O
'i- ^
o o o
O) *^ O *^
>. t»
§ ^ oi w « x:
C«.- tL-DL. 'CO
i'5o ° to ' c 10 o . ^
oo.ro 3 "- ^"B^1" ^'cc
^cm Q M (QoOOJJZ U(D
EO< Q- 1- 30COOCJ E-l
!n
ro o
T Nf
in
£
ro ro
-4- o
o o
^
ro N.
0 0
O CD
O 0
ro o.
CD «
o o
o
CD
O
Is- O
CD CD
CD O
*"^
^\
ro ro
3°.
o
^
OJ *
0 O
0 O
***
Q 1-
UJ Z
_J UJ
UJ » i
8 CO
I
CO
1
0
t
B-126
-------
>->-
-°3
D -M
CD CO
o
t (D
> JZ
+-
° x
CU XI
L-
CU TJ
2 CU
o
CO -r-
CU >
CO
COT!
CU
C- C_
CU gj
CD Q.
c o
o
e_ co
o
u- U
.g.
C CO
O i.
O)
4J 0
CD CU
' D)
Q. L.
O CU
Q. '
§1
.- CO
.c
£
2
E_
CU CO
Q. CU
CO "D
CU 3
en *-*
CD CO
u
c- o
CU "
o
c
CO >
° "S
CD O
C en
§~" cu
JZ
4-*
g-D
CD C
CD
>-
<- CU
O C3>
0> C
CU CO
4-> C-
CO
u cu
JT
CU 4->
0
t- H-
3 O
O
CO T3
CD CU
fe ?
H
CO 0)
o S
en
4-1
CO CU
cu .c
x: 4J
D)
CO
-C
CO 3
4J O
U) O
cu
2 CO
O "-
cu
JZ t-
4-> O
M-
D)
C
CO
CD O
X O
o j^
1
S
03 *"^
3 C
o o
O3 »
0
CD 'i
-
3 ' CO
cr ^2 cu
c5 S g
O O3
(U CO (_
T3 ro
^
tu u *-»
4-» O (D
.^- eu -M
4-» 4-J CO
CO 03 CU
cu E
.,- CO
+J 4J 03
C CO 3
. CD
O ,- CD
.8.5
CO U
CU OJ C
g .-
OJ CO t_
4-* (D dl
CO O U *
C C C TJ
-- «3 flj 01
4~» O >
+-» CO
CO C ' £-
E . CO '
i- S CO 4-»
O Q CO
(U CD Q. CD
JT CU
OJ OJ C
M- O *-
O CU O
CD O) Q.
CO U C
4-» £_ O3 (D
' 3 J- 0)
DO ^
C T3 > J=
(D 'tj *tl ^
J= C
Q_
CU
£_
CD
CO
0>
CO
CD
o
c_
u
c
O3
o
D)
C
CO
CO
CD
03
CO
4-J
C
J-
o
3
C
CO
Q_
CO
UJ
2
CD
C
L_
O »
CO
S S
-0 'co
CD CO
CO -r-
CD
-------
Modeled Estimates. Summing across source categories from the
studies, that used modeled concentrations yields a nearly identical range
of nationwide cancer estimates as that presented by the ambient
concentrations. The major significant risk source appears to be solvent
usage/degreasing, which is consistent with this source category being
identified as the major user of TCE. This source category shows a range
of nationwide risk between 2 and 10 cancer cases per year. The higher
risk estimate is from the 35-County study; the lower risk estimate is
from both the lEMP-Philadelphia and the Southeast Chicago studies. The
larger geographic scope of the 35-County study may suggest that its
cancer rate of 0.04 cancer cases per year per million is a more
reasonable rate to extrapolate to a national estimate. On the other
hand, the counties selected in the 35-County study were selected, in
part, for presence of known sources, and may be biased on the high side,
although this is less likely to occur for an area source such as
degreasing than for a point source. If the cancer rate for solvent
use/degreasing from the other four studies is used (i.e., approximately
0.01 cancer cases per year per million population), the range of
estimated cancer cases narrow to 4 to 6 per year. (Within the 5-City
study, individual cities had estimated cancer rates between 0.005 and
0.051 annual cancer cases per million population.)
Point Estimate. The range of estimated cancer incidence from both
ambient-measured and modeled concentrations is relatively narrow (4 to 9
and 5 to 13 cancer cases per year, respectively.) The wider range could
probably be accepted as is for a reasonable nationwide estimate. As
noted above, the range could be narrowed to 4 to 6 using the lower, but
consistent, cancer rate of 0.01 cancer cases per year per million
population for solvent use/degreasing. For ambient-measured estimates,
B-128
-------
the Ambient Air Quality study's estimate of 7 cancer cases per year is
considered the best estimate. Based on these considerations, the 7
cancer cases per year estimated by the Ambient Air Quality study is
selected as the estimate of nationwide annual cancer incidence due to
exposure to trichloroethylene.
B-129
-------
Vinvl Chloride. Nine studies included vinyl chloride as a pollutant of
concern (see Table B-27). Very few of the same source categories were
examined by more than one study. Further, four of the studies had
general, nonspecific source categories. Except for the Ambient Air
Quality study, modeled ambient concentrations were used to estimate
cancer risk.
Modeled Estimates. The range of nationwide risk has been
estimated to be between 6 and 25 cancer cases per year. The largest
reported risk estimate (19 cancer cases per year) is from the
NESHAP/ATERIS data base. This estimate reflects emissions estimated
from all source categories emitting vinyl chloride and not just from
those source for which regulations have been developed.11 The specific
source categories are not identified in the NESHAP/ATERIS data base,
other than for ethylene dichloride manufacturing. Since TSDFs and POTWs
are relatively "new" source categories, it is very likely that they are
not included in the NESHAP/ATERIS data base. It is unknown if sewage
sludge incinerators are included in the NESHAP/ATERIS data base for
vinyl chloride emissions. Thus, the best estimate of risk based on
modeled estimates is estimated to be 22 to 25 cancer cases per year
nationwide (the NESHAP/ATERIS data base estimate plus the estimates from
TSDFs, POTWs, and sewage sludge incinerators).
Ambient Estimates. The Ambient Air Quality study used test
results from 10 locations to estimate nationwide risk. These data are
summarized in Table B-28. For eight of the data points, the tests
actually did not indicate any vinyl chloride; only the Institute, W.V.
11 U.S. EPA. Estimation of the Public Health Risks Associated with
Exposures to Ambient Concentrations of 87 Substances. July 1984.
Appendix A. Public Health Risks Associated with Substances Listed Under
Section 112 of the Clean Air Act.
B-130
-------
TABLE B-27
ESTIMATED ANNUAL CANCER CASES FROM VINYL CHLORIDE BY SOURCE CATEGORY
SOURCE
CATEGORY
Chemical
Manufacturing
Aerators
Sewage Sludge
Incinerators
TSDFs
PVC and ED/VC
Manufacturing
Unspecified
POTW's
-
TOTALS
MODELED
AMBIENT
INDIVIDUAL STUDY
Ambient NESHAP/ Drinking Sewage Thirty-
Air , ATERIS Water Sludge TSDFs five
Quality Aerators Incinerators County
0.0051C
(0.000022)
negligible
2.7
(0.011)
0.023
(0.0001)
18.5
(0.077)
0.11
(0.0023)
0.68
(0.014)
0.0051-18.5 negligible 2.7 0.023 0.79
(0.000021-0.077) (0.011) (0.0001) (0.017)
13
(0.054)
B-131
-------
TABLE B-27 -- concluded
ESTIMATED ANNUAL CANCER CASES FROM VINYL CHLORIDE BY SOURCE CATEGORY
SOURCE
CATEGORY
Chemical
Manufacturing
Aerators
Sewage Sludge
Incinerators
TSOFs
PVC and ED/VC
Manufacturing
Unspecified
POTW's
INDIVIDUAL STUDY
IEMP- Southeast
Five City Kanawha Chicago
Valley
0.00037 negligible
(0.00002)
negligible
0.013 negligible
(0.0008)
NATIONWIDE
POINT
RANGE3 ESTIMATE
<0.0055
negligible negligible
2.7 2.7
0-0.023 0.023
18.5 18.5
0-0.6 0-0.6
3.49 3.5
TOTALS
MODELED
AMBIENT
0.0136
(0.0009)
negligible
negligible
6.2-24.7
13
25
13
NOTE: Numbers in parentheses are annual cancer cases per million population'. For nationwide
studies,
annual cancer cases were divided by 240 million, unless otherwise noted. For studies with
smaller geographic scopes, the annual cancer cases were divided by the study's population.
8 The numbers in this column were calculated by taking the lowest and highest incidence rates for
a source category and multiplying it by 240 (1986 U.S. population in millions). The total for
this column is the summation of the low end of the range and the sum of the high end of the
range.
The numbers in this column present the results of the reduction analyses. In most instances, a
point estimate of nationwide annual cancer incidence was derived for each pollutant/source
category combination. In some instances, a point estimate could not be reasonably derived. For
these instances, as narrow a range as possible of nationwide annual cancer incidence was
estimated, and such ranges appear in this column. The text discusses how these point estimates
and ranges were derived.
0 Ethylene dichloride (EDO manufacturing.
Lower estimate assumes sewage sludge incinerators are included in the NESHAP/ATERIS data base.
B-132
-------
and the Baton Rouge, LA, tests provided actual measured concentrations.
For six of the eight tests for which no concentrations were actually
measured, the Ambient Air Quality study assumed concentrations to be
one-half of the limit of detection (LOD) of the tests. In reality, the
actual concentrations could be from 0 /^g/m3 to approximately 2.5 Atg/m3.
For the other two tests for which no vinyl chloride was measured, no LOD
values were indicated. For these two locations, the Ambient Air Quality
study assigned a value of 0 ^g/m3. While specific point sources have
not been identified in California, there are at least two known point
sources in Baton Rouge, LA. This likely accounts for concentrations
being high enough to actually measure. On the other hand, the IEMP-
Kanawha Valley study identifies a single point source, located in Nitro,
WV, but no sources in Institute, WV. This appears to be at least
confusing with the ambient data in Table B-28, which shows the highest
concentration in Institute, WV. The lEMP-Kanawha Valley study did find
negligible concentrations in Nitro, which is consistent with the table
if the actual concentration is below the LOD of the test method.
Point Estimate. Given the paucity and suspect nature of the
ambient data, the risk estimated using modeled concentrations was
selected as the estimate of cancer cases nationwide. Thus, nationwide
risks from vinyl exposure to chloride emissions are estimated to be 25
cancer cases per year.
B-133
-------
TABLE B-28
VINYL CHLORIDE CONCENTRATIONS USED IN THE
AMBIENT AIR QUALITY STUDY
LOCATION
Fremont, CA
Mountain View, CA
Napa, CA
Redwood City, CA
San Leandro, CA
Vail e jo, CA
Las Vegas. NV
Institute, WV
Nitro, WV
Baton Rouge, LA
Concentration
Ug/m3)
1.278b
1.278b
1.278b
1.278b
1.278b
1.278b
Oc
2.442
Oc
1.41
NOBS3
1
1
1
1
1
1
2
1
5
1
a NOBS = number of site-years satisfying the minimum data
requirements of the Ambient Air Quality study.
b These values are based on one-half of the limit of
detection of the test method.
c Tests did not indicate any vinyl chloride. Limit of
detection for the test methods were not reported.
B-134
-------
Vinvlidene chloride. Emission sources of vinylidene chloride (VDC)
include the production of VDC, perchloroethylene and trichloroethylene,
1,1,1-trichloroethane, VDC polymers and copolymers, and chloroacetyl
chloride. In addition, VDC emissions occur from waste treatment,
storage, and disposal facilities.12 Five studies included VDC as a
pollutant of concern (see Table B-29). Four of the studies used modeled
ambient concentrations and one study, the Ambient Air Quality study,
used ambient measured concentrations.
Modeled Estimates. The three.specified source categories covered
by the three studies are a portion of the known sources of VDC
emissions, but are expected to be the major emitters. The two specified
source categories (i.e., VDC polymer and VDC monomer) under the
NESHAP/ATERIS data base are expected to be sources covered by the
"unspecified" source category. Thus, the most likely estimate for risk
from the NESHAP/ATERIS data base is 0.05 cancer cases per year. As the
facility modeled in the Kanawha Valley does not appear in the
NESHAP/ATERIS data base, the three modeled estimates can be summed. The
lEMP-Kanawha Valley facility is not known as to the type of
manufacturing process, and as it is a point rather than an area source,
one cannot reasonably extrapolate cancer risk to larger geographic
areas. Thus, it is more reasonable to add the cancer risk from the
study rather than apply its cancer rate (of 0.001 annual cancer cases
per year per million population) to the entire U.S. population in
estimating nationwide incidence. Given these considerations, a
nationwide cancer risk of approximately 0.5 cancer cases per year is
estimated from the modeled estimates.
U.S. EPA. Locating and Estimating Air Emissions from Sources of
Vinvlidene Chloride. EPA-450/4-84-007K. September 1985.
B-135
-------
TABLE B-29
ESTIMATED ANNUAL CANCER CASES FROM VINYLIDENE CHLORIDE BY SOURCE CATEGORY
SOURCE
CATEGORY
VOC Polymer
VOC Monomer
Chemical
Manufacturing
TSOF's
POTWs
Unspecified
TOTALS
MODELED
AMBIENT
INDIVIDUAL STUDY
Ambient NESHAP/ IEMP- Southeast
Air ATERIS TSDF's Kanawha Chicago
Quality Valley
0.017
(0.000071)
0.0023
(0.00001)
0.001
(0.01)
0.49 negligible
(0.002)
negligible
0.05d negligible
(0.0002)
0.019-0.05 0.49 0.001 negligible
(0.00008-0.0002) (0.002) (0.01)
10
(0.04)
NATIONWIDE
. POINT
RANGE3 ESTIMATE*3
--
--
0.001C 0.001
0-0.49 0-49
negligible negligible
0.05 0.05
0.05-0.5 0.5
10 10
NOTE: Numbers in parentheses are annual cancer cases per million population. For nationwide studies, annual
cancer cases were divided by 240 million, unless otherwise noted. For studies with smaller geographic
scopes, the annual cancer cases were divided by the study's population.
8 The numbers in this column were calculated by taking the lowest and highest incidence rates for a source
category and multiplying it by 240 (1986 U.S. population in millions). The total for this column is the
summation of the low end of the range and the sum of the high end of the range.
b The numbers in this column present the results of the reduction analyses. In most instances, a point
estimate of nationwide annual cancer incidence was derived for each pollutant/source category
combination. In some instances, a point estimate could not be reasonably derived. For these instances,
as narrow a range as possible of nationwide annual cancer incidence was estimated, and such ranges
appear in this column. The text discusses how these point estimates and ranges were derived.
c Due to unknown nature of chemical facility, the incidence rate from the lEMP-Kanawha Valley study was
not extrapolated nationwide.
d This number likely includes VDC polymer and VDC monomer sources as well as other unspecified sources.
B-136
-------
Ambient Estimates. The Ambient Air Quality study estimates 20
times this risk (10 cancer cases per year versus 0.5 per year). The
results of the Ambient Air Quality study are based upon ambient measured
data from ten locations. At least four of these locations have known
VDC emitters (Los Angeles, Chicago, Charleston, W.V., and Sacramento,
CA.). The ambient concentrations for these four cities were 0.02,
0.088, 0.03, and 0.27 /*g/m3, respectively, for a per city average of
0.10 ^g/m3. Based on locations identified in "Locating and Estimating
Air Emissions from Source of the Vinylidene chloride," (EPA-450/4-85-
007k), none of the other six cities have point sources of VDC. Ambient
concentrations in these seven other cities ranged from 0.036 to
0.124 ^g/m3, for a per city average of 0.066 /ig/m3. Given the known
locations of VDC point source emitters, it is not surprising that the
four-city average concentration is larger than the six-city average
concentration, although it is somewhat surprising that two of the four
cities with known VDC sources had the two lowest concentration reading
of all ten locations.
Point Estimate. Considering the above information-, a range of
cancer cases of between 0.5 and 10 per year nationwide is created.
Although more information on VDC sources and a broader data base would
be desirable, the Ambient Air Quality study's results (10 cancer cases
per year)'were selected as the estimate of cancer risk to total VDC
exposure at this time.
B-137
-------
-------
APPENDIX C
SUMMARIES OF POLLUTANT-SPECIFIC
AND SOURCE-SPECIFIC STUDIES
(INCLUDING NONCANCER HEALTH RISK PROJECT ON AIR TOXICS)
C-l
-------
This appendix presents summaries of ongoing EPA studies that are
related to specific pollutants and source categories. Most of these
studies are related to the development of national emission standards
for hazardous pollutants (NESHAPs). 'in addition, a summary of the EPA
study on noncancer health risks of air toxics is provided (Noncancer
Health Risk Project). An index to these studies is presented below.
Study
Page No.
1. Asbestos C-3
2. Coal and Oil Combustion C-6
3. Drinking Water Aerators C-12
4. Gasoline Marketing C-17
5. Hazardous Waste Combustors C-24
6. Municipal Waste Combustors C-34
7. Municipal Solid Waste Landfills C-41
8. Publicly Owned Treatment Works (POTWs) C-42
9. Radionuclides C-46
10. Sewage Sludge Incinerators C-50
11. Superfund Sites C-55
12. Treatment, Storage, and Disposal
Facilities for Hazardous Waste (TSDF) C-56
13. Waste Oil Combustors C-61
14. Woodstoves C-67
15. Noncancer Health Risk Project C-70
C-2
-------
Asbestos
The Office of Air and Radiation promulgated the initial asbestos
NESHAP in 1973 and revised the rule in 1975, 1978, and 1984. The
purpose of this asbestos project is to review the current NESHAP, assess
its effectiveness and revise the rule as necessary. The current NESHAP
covers asbestos milling, manufacturing and fabricating, removal of
.asbestos prior to renovation or demolition, the disposal of asbestos
waste, and the use of asbestos in spraying, insulation, and asphalt-
concrete for roadways.1 The standard was based on a qualitative
assessment of the risk from exposure to asbestos. With the development
of a unit risk estimate for asbestos, it is now possible to make a
quantitative assessment of risk.
The risk assessment has been performed to assess the risk from the
current asbestos emissions as well as the regulatory alternatives
(Reference 1). Table C-l presents the current risks and the minor
revisions alternative that would promote full compliance to the NESHAP.
Other alternatives (not presented) reduce risk to negligible levels.
Asbestos emissions from milling, manufacturing and fabricating and
waste disposal from these facilities were modeled using the point source
algorithm of the Human Exposure Model (HEM). Plant specific data were
obtained by Section 114 letters for the plants with the highest maximum
lifetime risk and annual incidence. Two of these plants were modeled
using ISCLT/LONGZ. The maximum individual lifetime risk reported in
Table C-l for manufacturing results from this more detailed modeling.
1 Due to the discontinued use of asbestos in spraying, insulation,
and asphalt concrete roadways, emissions and, therefore, risks are thought
to be negligible. The regulation does not address unpaved roadways
containing asbestos-contaminated gravel, which occurs naturally in some
areas. This was concluded a local problem and risk was not assessed.
C-3
-------
TABLE C-l
ESTIMATES OF INCIDENCE AND INDIVIDUAL RISK DUE TO
ASBESTOS EMITTED INTO THE AIRa
Source
Categories
Milling
Disposal
Manufacturing
Fabrication
Renovation,
Removal
Disposal
Demolition,
Removal
Disposal
TOTAL
Maximum Individual
Lifetime Risk
Full
Compliance
3 x If5
6 x 10'9
2 x 1CT3
2 x 10'4
3 x 10'7
6 x 10'8
2 x 10'5
1 x 10-5
Current
Compliance6
same
same
same
same
6 x 10'7
3 x 10"5
4 x 10'5
7 x 10"3
Estimated Excess Annual
Lung Cancer and Mesotheliomas
Full
Compliance
0.004 - 0.005
<0.0001
0.3 - 0.7
0.05 - 0.2
0.003
0.0007
0.3
0.1
0.7 - 1.2
Current
Compliance
same
same
same
same
0.0071
0.35
0.5
80
81.6
Source: Reference 1, pages A-28, A-32, A-35, and A-36.
a Please refer to footnote 1, page C-3, for a list of caveats and an
explanation of the methodology used to generate these results.
b The large number of sources and inadequate enforcement resources have
resulted in noncompliance with the demolition and renovation (including
waste disposal) standards. The Stationary Source Compliance Division
estimated compliance in 1985 at about 50 percent. The risk estimates in
parentheses were estimated under the assumption that only 50 percent of
the demolitions and 80 percent of the renovations were in compliance.
C-4
-------
Asbestos emissions from removal and waste disposal during
renovation and demolition activities were modeled using the area source
algorithm of HEM. Asbestos emissions were assigned to each county based
on the population of that county. This process generated the annual
incidence figures. The maximum individual lifetime risk was generated
assuming that emissions assigned to the county with the highest
population density were emitted from a single point source. This
technique overestimates risk.
The renovation and demolition source categories for asbestos are
unique because it is estimated that only 80 percent and 50 percent are
in compliance, respectively, to the current NESHAP. This makes baseline
risk different from full compliance to the current NESHAP.
Asbestos is a known human carcinogen. The unit risk estimate is
based on several human studies. The health data base for asbestos is
much better than most toxicant data bases. It is important to note,
however, that in order for asbestos to cause lung cancer or mesothelioma
the fibers must be respirable. Respirable refers to fibers small enough
to enter small airways. Lacking fiber size distribution information,
100 percent respirability was assumed. The larger nonrespirable
particles may comprise a large portion of the emissions. The emissions
of respirable asbestos and thus estimated risk could be greatly
overestimated.
References
1. U.S. Environmental Protection Agency, Emission Standards and
Engineering Division. National Emission Standards for Asbestos-
Background Information for Proposed Standards. Draft. March 5,
1987.
C-5
-------
Coal and Oil Combustion
The Environmental Protection Agency's Office of Air Quality
Planning and Standards has evaluated on a national scale toxic emissions
from utility, industrial, commercial and residential combustion units
(Reference 6). These four combustion sectors, briefly described in
Table C-2, are known to emit several carcinogenic compounds, of which 9
were specifically included in this effort. These 9 pollutants are:
acetaldehyde, acrolein, arsenic, beryllium, cadmium, hexavalent
chromium, polycyclic organic matter (POM), formaldehyde, and
radionuclides. Other pollutants were not evaluated because of a paucity
of emissions data. Because of the nature of the available emissions
data (national averages, of data with large variations), short-term
exposures were not specifically considered and long-term exposures (and
associated cancer risks) were given the most attention. The pre-
liminary cancer risk assessment estimates (see Tables C-3 and C-4)
indicate that the national cancer incidence is about 11 cases per year
and that the maximum individual risk for all sectors is less than 10"4.
However, these estimates are crude and at best are considered "order of
magnitude" values since the exposure techniques (described below) are
not based on site-specific analysis.
As seen in Table C-2, the number of combustion units is very large
and reasonably precludes site-specific analysis. However, for the
utility sector, a data base which contained basic stack parameters and
control technology status was available for a large majority of the
plants and was used for this study. The Human Exposure (computer) Model
was run for each plant in the data base in conjunction with an emissions
data base containing national average emissions factors and average
control efficiencies. Flat terrain was assumed for the air dispersion
C-6
-------
TABLE C-2
BACKGROUND INFORMATION ON THE COAL AND OIL COMBUSTION SECTORS
SECTOR
Utility
Industrial
Commercial
Residential
NO. OF UNITS
COAL
987b
51,000C
163,000°
430,000d
OIL
264b
190,000°
443,000°
13,000,000d
FUEL BURNED3
COAL OIL
1012 Btu/yr
(Millions of
tons/yr)
12,500a
(594)
2,500*
(105)
115a
(5)
77a
(3)
1012 Btu/yr
(Millions of
barrel s/yr)
l,600a
(250)
2,400a
(390)
840a
(138)
l,050a
(180)
a Reference 3.
b Reference 1.
c Reference 4.
d Reference 2.
C-7
-------
o
a
, 'H 'o _^ c _^
^oo 7300 BOO woo
3 -« o ec
CO
ru
d
-
d
in
o
ro
in,
(A
a
M
o
h-
I
g
S
O
t.
a>
2
f
S
o
C-8
-------
(/)OJ
^ en
CO ^
1-4 }
o: -
01
ui
LLl
X
«c
E
<
_J
V)
o
1
c
CO
3
o
a.
s
o
1
o
CO
oc.
i
"O
1
0
LL.
5
o
x .c
3;
3
1
o
§
£_
OJ
CO
0
c
CO
CU O
O O)
U 0)
o to
CO O
in ^o in in in in
0 O 0 O O O
XX XX XX .1
(M CM in « « <>
<,
'o 'o *o
X i X » X i ii
ro ro -*
o o o o o CD
XX XX XX »
*- «- nj eo ^- CM
co eo so r- ^ h-
O O O O CD O
XX. XX XX ii
(M «- CM fO in M
N- CO f»- t"1* N- ^»
'o 'o 'o !o 'o 'o
XX XX XX ii
r*- ro co >t KJ eo
N- S- *o *O f^ in
O O CD O O CD
XX XX XX IP
in in (MM -j- «-
O O O CD O O
XX XX XX I I
^ CM CM o ro in
in >o in "O *o in
ii ii ii
O O O CD O O
XX XX XX ii
"ra "(5 "-
>._ r_ o_ |_
t~
-------
analysis and no site-specific or geographic or seasonal adjustments were
made to the emission factors. As in the case for all sectors,
distinctions were made for three coal types (bituminous, anthracite, and
lignite) and two different oil types (distillate and residual). For the
industrial and commercial combustion units, a subset of all the boilers,
a stratified random sample from the National Emissions Data System, was
analyzed in a manner similar the utility sector. Because this sample of
boilers were representative of boilers greater than two million Btu's
per hour, an additional exposure analysis, which applied a simple area
source model, was used for these very small boilers. Toxic emissions
and long-term concentrations were estimated on a county-by-county basis.
For the last sector, residential heating, the same approach as that used
for the very small industrial/commercial boilers was applied.
For this project there are several uncertainties of note. Based
on a review of the emissions data, there is a very wide range of
emission factors found in the literature; however, this study assumed
that average or typical emission factors were applicable at boiler site.
Coal and oil combustion is known to emit a wide range of compounds, but
all the pollutants evaluated in this study (a total of 9) account for
less than 10 percent of the particulate matter and the volatile organic
compound emissions. Thus, there is a considerable fraction of the
combustion emissions of unknown toxicity. Lastly, and most important
because the estimated maximum and average concentrations are low, the
models by which public risks are calculated must extrapolate a health
data base established from high exposure levels to public exposure
levels which are several too many orders of magnitude lower.
C-10
-------
References
1. Peters, W.D., US EPA, Pollutant Assessment Branch. Coal and Oil
Combustion. July 25, 1988. 6 pages.
2. Utility Data Institute. Power Statistics Database. 1983. Developed
by: Edison Electric Institute.
3. U.S. Department of Commerce, Bureau of the Census. Statistical
Abstract of the United States - 1986. 106th Edition. Table 1315,
page 733.
4. U.S. Department of Energy, Energy Information Administration. State
Energy Data Report: Consumption Estimates. 1960-1982. May 1984.
5. U.S. Environmental Protection Agency, Industrial Environmental
Research Lab. Population and Characteristics of
Industrial/Commercial Boilers in the U.S. EPA-600-7/79-178a, August
1979.
6. U.S. Environmental Protection Agency, OAQPS. Coal and Oil
Combustion Study: Summary and Results. External Review Draft.
September 1986.
C-ll
-------
Aeration of Toxics from Drinking Water Treatment Facilities
In response to several requests from the Office of Drinking Water
(ODW), the Office of Air Quality Planning and Standards (OAQPS) has
assessed the cancer risk associated with aeration treatment of drinking
water in the U.S. When drinking water supplies contain volatile
compounds (VCs) that are toxic, the aeration process can be used very
effectively to remove the VCs from the water, but at the same time,
create VC emissions to the atmosphere. To date, OAQPS has evaluated 10
pollutants (listed in Table C-5) in three different studies. The
preliminary risk assessment results are summarized in Table C-6. These
results, which are based on a screening analysis described below,
provide crude estimates and are, at best, order of magnitude estimates.
In Study Number 1 (Reference 1), the first seven chemicals in Table
C-5 were evaluated from 22 existing sources with known contamination
levels that were either aerating or planning to aerate their water
supplies in the near future. The ODW supplied the necessary emissions
and stack data, but the exact locations of the facilities were unknown.
The facilities were assumed to be located in: (1) the center of the
cities to which units were supplying water, and (2) in areas of flat
terrain. The VC emission rates were based on actual site-specific data
and the assumption of 100 percent efficient aerators. The Human
Exposure Model (HEM) was used to estimate the air dispersion of the
emissions, the public exposure to the emissions, and the associated
cancer risks. As seen in Table C-6, risk projections were made based on
the thought that the 22 selected sites were typical operations and were
representative of as many as 200-500 facilities which were anticipated
to be built over the next ten years. It was assumed that the aggregate
C-12
-------
TABLE C-5
LIST OF POLLUTANTS EMITTED FROM AERATION OF DRINKING WATER
TREATMENTFACILITIES WHICH HAVE BEEN EVALUATED BY OAQPS
POLLUTANT
Trichloroethylene
Tetrachl oroethyl ene
1,1,1 Trichloroethane
1,2 Dichloroethane (EDC)
Carbon Tetrachl oride
1,1,2,2 Tetrachl oroethane
Vinyl chloride
Ethylene dibromide (EDB)
'
Dibromochloropropane (DBCP)
Radon
STUDY NO.
1
1
1
1
1
1
1
2
2
3
REFERENCE
NO.
1
1
1
1
1
1
-
1
2
2
3
NOTE: Study numbers refer to studies listed in Table C-6.
C-13
-------
TABLE C-6
RISK ASSESSMENT RESULTS FOR THE THREE
DRINKING WATER AERATION STUDIES
1
2
3
No. of
Plants
22
(existing)
200
500
7
20
Approx.
26000
No. of
Pollutants L
All but
EDB, DBCP,
Radon
All but
EDB, DBCP,
Radon
All but
EDB, DBCP,
Radon
EDB & DBCP
Radon
Radon
Max. Individual Annual
ifetime Risk Incidence
2 x 10~5 0.0047
2 x 10'5
2 x 10~5
3 x 10'6 0.0002
5 X 10~5 0.016
5 x 10~5
Projected
Annual Incidence
-
0.043
0.11
-
-
0.4a
a Assumes all facilities using water supplies with radon concentrations
> 200 pCi/L apply aeration as a control technique.
C-14
-------
population risks were proportional to the number of plants applying
aeration treatment.
In Study Number 2 (Reference 2), only two chemicals (see Table
C-6) from seven sites were evaluated. However, in this case, although
the analysis was conducted in a manner similar to Study Number 1, there
were no projections of national or future level of aggregate risks. The
site-specific contamination data were thought to be untypical of most
plants in the country, since these chemicals were not usually found in
drinking water supplies.
In Study Number 3 (Reference 3), the OAQPS, in conjunction with
the Office of Radiation Programs (ORP), estimated cancer risks
associated with potential radon emissions5 from the aeration process.
The ODW selected 19 sites that were thought to be typical of facilities
across the country plus one site that was known to have a very large
radon emission rate. Many of the facilities selected, in addition to
most facilities in the country, are not currently aerating their
drinking water; the goal of this study was to determine the potential
level of risks if many of the existing facilities would aerate their
water supplies. Because of the complicated mathematics that are
required to model air dispersion of radioactive emissions of both the
parent isotopes and progeny of the radioactive decay process, the HEM
cannot adequately estimate public exposure. So, the ORP computer models
that were specifically designed for radioactive emission exposure
(AIRDOS-EPA, RADRISK, DARTAB) were required. These computer models
estimate radionuclide concentrations in the air, rates of deposition on
the ground, and the amounts of radionuclides taken into the body via
inhalation of air and ingestion of meat, milk, and fresh produce. As in
C-15
-------
the case where the HEM was used, flat terrain was assumed when running
the ARDOS model.
In addition, using a technique like that used in the first study,
national risk estimates were projected based on the results of the 19
facilities.
References
1. 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, Science and Technology Branch. Risks Associated with Air
Emissions from Aeration of Drinking Water. November 18, 1985.
(Study Number 1)
2. Memorandum. W.D. Peters, US EPA, Pollutant Assessment Branch, to
S.W. Clark, US EPA, Science and Technology Branch. Aeration
Drinking Water Facilities - EDB and DBCP Emissions. February 18,
1986. (Study Number 2).
3. Memorandum. W.D. Peters, US EPA, Pollutant Assessment Branch, and
C.B. Nelson, US EPA, Office of Radiation Programs, to S.W. Clark, US
EPA, Technology Section, STB, CSD, ODW. Preliminary Risk Assessment
for Radon Emissions from Drinking Water Treatment Facilities. May
1988. (Study Number 3)
C-16
-------
Risk Assessment for the Gasoline Marketing Source Category
A cancer risk assessment was performed for the gasoline marketing
source category to determine risk from high exposure and cancer
incidences, due to exposures to gasoline vapor emissions. Pollutants of
concern were benzene, gasoline vapors (as a collection of all
components), ethylene dibromide (EDB) and ethylene dichloride (EDC).
The study evaluated uncontrolled and controlled emissions from bulk
gasoline terminals, bulk plants, storage tanks, and service stations.
This discussion presents a summary of these risk analyses. More
detailed discussion can be found in the EPA reports describing the
entire analysis (see References 2 and 3).
The purpose of the overall study was to evaluate environmental
impacts, costs, risks, and benefits associated with reducing emissions
at' gasoline marketing facilities. Many regulatory strategies were
analyzed in this study. However, risk assessments centered on the
evaluation of exposures for individuals living in the vicinity of
gasoline marketing facilities (community exposures). Risks for these
individuals were based upon emissions from bulk terminal and bulk plant
storage tank and tank truck loading operations, gasoline deliveries to
service stations (service station Stage I) and vehicle refueling
operations (service station Stage II). The vehicle refueling analyses
included not only community exposures but also self-service refueling
exposures to individuals refueling their own vehicle, and occupational
exposures to service station attendants.
The project is on-going and risk assessments have centered on
benzene and gasoline vapor exposures. EDB and EDC are components of
leaded gasoline only and were found to be very small, especially with
the decline of leaded gasoline usage. Gasoline vapor risk analyses were
C-17
-------
originally based upon studies of exposures to wholly vaporized gasoline,
since this was the basis of the animal exposure studies from which the
risk factors were calculated. Based on review by the Science Advisory
Board and on public comments received, there was some concern whether
all components in wholly vaporized gasoline were indicative of actual
gasoline vapor exposure. An estimate of components C6 and higher
(thought to be the components of concern in gasoline vapors) was
calculated. As a result, exposure to gasoline vapors was calculated and
expressed as due to total vapors and due to C6 and higher components in
gasoline vapors. Table C-7 contains a summary of unit risk factors used
in the gasoline marketing analysis.
Risk estimates for all gasoline marketing source categories proved
to be very difficult because of the large number of sources involved
(1500 terminals, 15,000 bulk plants, and 400,000 service stations).
Obviously risk assessments could not be conducted on each individual
source, so a scheme of model plants and representative locations was
developed.
The assessment methodology derived for bulk terminals and bulk
plants were similar. A series of model plants for each source category
were developed for both product storage and truck loading operations.
Since these facilities are usually clustered due to access to pipelines,
railways and barge transport points, clusters of complexes of facilities
were developed. Several cities, of varying population sizes and
densities, were selected to represent the country as a whole. The model
complex selected for use in each city was developed to represent the
city size (e.g., large terminals in larger cities, smaller terminals in
smaller cities). Model facility complexes were placed at coordinates of
C-18
-------
TABLE C-7
UNIT RISK FACTORS APPLICABLE TO THE GASOLINE
MARKETING SOURCE CATEGORY
Pollutant
Unit
Risk Factor3
Benzene
2.6 x 10"2
Gasoline Vapors
- Rat Studies
o PULb
o MLEC
3.1 x 10"3
2.0 x 1CT3
- Mice Studies
o PULb
o MLEC
Ethylene Dibromide
Ethylene Dichloride
2.1 x 10'3
1.4 x 10~3
4.2 x 10"1
2.8 x 10~2
SOURCE: Reference 2, pages 6-2 and 6-31, and Reference 3, page 2-61.
a Probability of cancer incidence from exposure to 1 ppm over a 70- '
year lifetime.
b PUL = Plausible Upper Limit.
c MLE = Maximum Likelihood Estimate.
d Risk factor used as basis for gasoline vapor risk estimates in latest
analysis.
C-19
-------
known bulk loading sites within each city selected. Emission rates and
heights were calculated and used as input to the Human Exposure Model
(HEM) to estimate cancer incidences for individuals living in the
vicinity of the model loading complex. Nationwide incidences were then
calculated based on incidences for each city evaluated and the
population distribution of each city size found in the country. For
example, the country was divided into seven population ranges. Several
cities were selected to represent each population range. The HEM
results for each city within a range were averaged, and the results used
to represent that range. The average exposure from each population
range was weighted by the percent nationwide population in that range to
obtain the nationwide exposure.
Because of the vast number of service stations and the ability to
locate them virtually anywhere within a metropolitan area, the method
for estimating incidences due to exposures to individuals living in the
vicinity of service stations could not be based on actual locations.
Several metropolitan areas around the country were selected to
represent population ranges for the nation. Within each metropolitan
area, the gasoline consumption was used to estimate total emissions from
service stations. These emissions were then assumed to be uniformally
spread over the metropolitan area and a uniform exposure concentration
was calculated. This uniform concentration was used as input to the HEM
model to determine cancer incidence estimates in each of the selected
metropolitan areas. Nationwide incidences were calculated by weighing
the results from each population range by the percent nationwide
population in that range, as was done for bulk terminals and bulk
plants. Risks from high exposure were based upon calculations of
exposures to individuals living near a model complex or service stations
C-20
-------
such as may be found at intersections with service stations on every
corner.
Vehicle refueling self-service and occupational exposures were
calculated based upon field studies to determine actual concentration
experienced in the breathing zone of individuals refueling their
vehicles (see Reference 1). These exposure concentrations, coupled with
the risk factors and known quantities of gasoline pumped nationwide at
self-service and full-service operations were then used to calculate
cancer incidences.
Table C-8 presents a summary of the nationwide average annual
baseline risks associated with exposures to benzene, gasoline vapors,
EDB, and EDC. Lifetime risks from high exposures are based upon
exposures to total gasoline vapors. Table C-9 summarizes the residual
risks and risk reductions associated with the regulatory strategies
revaluated in this analysis. Values for EDB and EDC are not included in
the summary since they had been dropped from consideration at the time
this analysis was conducted.
References «
1. -Clayton Environmental Consultants, Inc. Gasoline Exposure Study for
the American Petroleum Institute. Job No. 18629-15. Southfield,
MI. August 1983.
2. U.S. Environmental Protection Agency. Evaluation of Air Pollution
Regulatory Strategies for Gasoline Marketing Industry. EPA-450/3-
84-012a. July 1984.
3. U.S. Environmental Protection Agency. Draft Regulatory Impact
Analysis: Proposed Refueling Emissions Regulations for Gasoline-
Fueled Motor Vehicles - Volume I Analysis of Gasoline Marketing
Regulatory Strategies. EPA-450/3-87-001a. July 1987.
C-21
-------
TABLE C-8
SUMMARY OF ESTIMATED BASELINE CANCER RISKS FOR THE GASOLINE
MARKETING INDUSTRY3
Lifetime Risk
Facility From High
Category Exposure Benzene
Bulk Terminals 5.7 x 10~3 0.1
Bulk Plants 2 x 10~4 0.05
Service Stations
Community Exposure
- Stage I 6.7 x 10"* 0.1
- Stage II 1 x 10"* °-4
(Total) (1.6 x10"4) (0.5)
Self-Service 8 x 10~5 4.4
Total Public Incidence 5.1
Occupational 4 x 10~ 1.7
(Service Stations)
Total Incidence for 6.8
Gasoline Marketing
Source Category
Average Annual Cancer Cases
Gasoline Vapors EDB
Total C6
3.5 0.9 0.0005
1.4 0.4 0.0002
3 0.8 -c
10 2.5
(13) (3.3) (0.001)
33 8.3 0.006
51 13 0.008
17b 4.3
68 17 0.008
EDC
, 0.0006
0.0002
(0.001)
0.008
0.01
-
0.01
SOURCE: Reference 2, page 6-31 and Reference 3, page 2-63.
a Baseline risks are those projected throughout the study period (1988-2020) with no additional
controls.
b Based on plausible upper limit for total gasoline vapors.
c Hot calculated.
C-22
-------
o
CJ
CJ _1
ro
a: m
^
ro
Q
1
ro
CO
8.
5 S
o
CO ir-
ro 4J
^ §"
SCj
Cj
N O
C
cu
en
0
o cu
c o
CU I -
i° i >
O 01 CU
£Z C/3 CO
ro
§
-« g g
"ro oj 5
3 "> CO
^
cu
D) CO
ro -X S
*- "^ c
o 3 ro
> co _^
< a.
co
j^ ro
3-H
CO E
CU
t
> CU CO
t- N C
O X O
4-» CO CO -t-
03 CD 4->
4-» _C Q.
U CO 4-J E
CO C- *- CU
CD 4-» 3 x
tt CO ' UJ
sf ^O
o o
CO *t CO
>Q %Q ^
*N^ **^. ^L
^ N: NT
rs. rs. is.
CM CM CM
to to to
to to to
to «- to
in s* m
o o o
in
in CM in
CD O O
O O CD
CM
o o o
CU CO
1 1
o c
o
4-* CJ
CO
x. cu
>
1 «*-
CD 4-i
C -- CO
r- £_
' CD O
CD CO Q.
CO CO CO
CO 4-» >
CXI CO UJ
o
^r
CO
in
^
in
CM
o
CM
*-
5r
o
in
0
o
v^
^r
0
ro
(U
L.
^.
O)
ro
CO
^O
s*
*s.
CO
in
£
in
*^
CM
O
CM
"s.
T-
^X
^f
O
in
o
CD
^>
-*N.
*
CO
"o
L.
1$
Is- '4^
CM ro
:|
0)
ro
CO
CM
to
i
CJ
CM
st
si-
si-
vO
to
^
1
to
CM
o
Ss,
3
^
T
CM
CO
Ss.
ro
o
ss.
in
0
o
sj
o
T3
CU
tionwid
ro
x.
cu
01
ro
4-1
CO
u
CO 0 U
to ro in
co -i -i
cj in in
si-
si-
si-
ro CD o
; to CM
^*^ Ss, s^
to ro CM
CM
I C> CO
O s^ s^
^ 0 CD'
o
CM
*s. O
to «- o
CM «^ ^~-
CO CM C\T
r> ° °
o
^s. s^ s^
in in in
0 00
O 0 CD
s* s}- si-
O 00
CO
°« 2 -8
1 § n, %'%
c ° % v.°
° cu x i; *-
X; .> i * «>
ro *r^ o is.
1 o JS ' x
CO
L-
8.
CO
CO
ro
CO
O
U
O
T-,
O
0
O
£
c
ro
u
CO
C-
8.
CO
CO
CO
: ,
CO
o
I
CO
L.
8.
ro
CO
CO
O)
ro
4-1
o
L.
O
3
0
CO
CO
o
ro
.
0>
c
n
§
O
C_
o
cu
3
a
o
ro
X
cu
a.
3
T3
ro
4-1
o
c
>s
CO
4J
3
CO
cu
g
4J
g
C_
H-
O
C
E
ro
CO
c
o
CO
4?
O.
0
ju
u
.c
cu
e
«4-
co
c
o
4J
8
g
cu
u
c
cu
0
c
T3
"cu
CO
o
C-
§
u
cu
4^
ro
o
a.
ro
UJ
c ro
CD
s^ C.
cu a.
> a.
3
u cu
CD .C
c
"" T?
ro ro
o
4-* "O
C C
CU "-
u
L. CO
8.-
(U
CM CO
sQ C
t_
C O
§c
o
o
c ^
CU CO
t_
CO
CO CU
E h-
1
E ro
H-
O CO
CD
E '
CO 3
CO CU
0 .C
0.8-j!
CO C 4-«
c
C i CU
01 CU CO
CD CO CD
3 CO C.
4-» .C Q.
cy CL
^ CO
TJ C "~
M- «r- I_
CO Q" >
cu a> co
o H- a>
OJ 4-1
r- CD
O CO "0
M^ CO CO
s- c_
CU 4-*
^- 5 S
O 4J
.
O) t-
to o to
S Q. >
' CO
CO 4^ CD
> c 3
*t- E "co
o
CD t_ t_
CO O CU
C M- 3
03 C O
^ dj .
C-23
-------
Hazardous Waste Combustors
Wastes containing hazardous materials are commonly burned in
incinerators, boilers, and industrial furnaces. The U.S. Environmental
Protection Agency has estimated toxic emissions from hazardous waste
incinerators as part of regulations under the Resource Conservation and
Recovery Act (RCRA) and from the burning of hazardous wastes in boilers
or industrial furnaces, also as part of regulations under RCRA. Table
C-10 summarizes the toxic emissions being regulated from hazardous waste
combustors.
For incinerators, boilers, and industrial furnaces, EPA has
determined that risks from the burning of hazardous wastes in these
devices can be unacceptable under reasonable, worst-case circumstances.
»
For purposes of the rules, EPA defined unreasonable risk to be either:
(1) an exceedance of incremental lifetime cancer risk of greater than 1
x 10"5 to the potential maximum exposed individual (MEI) for toxic metal
and organic compound emissions and other carcinogens; or (2) an
exceedance at the MEI of Reference Air Concentrations for noncarcinogens
established at 25 percent of the Reference Dose.3
Risk Assessment. For hazardous waste incinerators, a risk assessment
was performed under existing baseline and post-compliance conditions for
82 incinerators. The risk assessment was performed for three
carcinogenic metals (arsenic, cadmium, and hexavalent chromium),
2 Components of the reasonable, worst-case circumstances included
concentrations of constituents in the incinerated waste, combustion
capacity or feed rate, partioning of metals to bottom ash, collection
efficiency of emission control equipment, and local terrain and
meteorological conditions.
3 Except for lead and hydrogen chloride. The exceedance for lead was
set at 10 percent of the national ambient air quality standard for lead,
and for hydrogen chloride the reference air concentration was based
directly on inhalation exposure studies. (Reference 2, page 13).
C-24
-------
TABLE C-10
TOXIC EMISSIONS SUBJECT TO
HAZARDOUS WASTE COMBUSTOR REGULATIONS
Pollutant
Antimony
Arsenic
Barium
Beryllium
Cadmium
Chromium (VI)
Hydrogen Chloride
Lead
Mercury
Principal organic hazardous
constituents (POHCs)
Products of incomplete
combustion (PIC)a
Silver
Thallium
SOURCE: Reference 3, Exhibit 7-3.
a Includes the following compounds:
benzene
perch!oroethylene
carbon tetrachloride
1,1,1-trichloroethane
1,1,2-trichloroethane
chloroform
trichloroethylene
1,2-dichloroethane
1,1-dichlorqethylene
1,1,2,2-tetrachloroethane
1,2-dichloroethylene
C-25
-------
principal organic hazardous compounds (POHCs), and products of
incomplete combustion (PIC), and for noncarcinogens (hydrogen chloride,
lead, barium, and mercury).4 For the three carcinogenic metals, both
lifetime cancer to the maximum exposed individual and the annual cancer
incidence attributable to all metals at each facility were estimated.
Emissions of the six metals from each facility were approximated
by using estimates of: (1) the quantity of hazardous waste combusted by
RCRA code, (2) the estimated fraction of metals in each RCRA code, (3)
the fraction of each metal segregated as bottom ash and stack emissions,
and (4) metal removal efficiencies for in-place air pollution control
devices. Maximum and area-wide ambient concentrations were predicted
using dispersion modeling for ten hypothetical facilities plus the
actual facility at 24 different sites. The unit cancer risk values were
obtained from EPA's Carcinogen Assessment Group. Population data for
estimating the number of exposed individuals was obtained from U.S.
Census data available from the Office of Toxic Substance's Graphical
Exposure Modeling System (GEMS).
For hazardous waste boilers and industrial furnaces, the risk
assessment performed also examined both existing baseline and post-
compliance conditions. The analysis predicted health risks from stack
releases and resulting atmospheric concentrations of POHCs, PICs,
metals, and hydrogen chloride. Both cancer and non-cancer health
effects were considered; .however, estimates were only made for the
aggregate number of cancer cases. Both maximum exposed individual risk
and aggregate cancer cases over 70 years were estimated.
4 Lead has since been designated as a B2 carcinogen, and is not
included in this report's estimate of cancer risk.
C-26
-------
In estimating cancer risk from boilers and industrial furnaces,
risks were calculated assuming two types of hazardous wastes being
burned (a base case waste and a high risk waste)5 and two levels of
control device performance (a base case and a pessimistic performance
level). The analysis assumed that all toxic compounds in the waste are
emitted unless destroyed or removed by air pollution control devices.
For metals, the risk calculations assume that all metals are present in
stack emissions and that none remain in the ash. Estimates of ambient
concentrations were made using the Industrial Source Complex Long-Term .
(ISCLT) Model. Site meteorology and population data were obtained from
GEMS.
Results. Table C-ll summarizes the estimated excess cancer cases from
incinerators burning hazardous wastes and Table C-12 for boilers and
furnaces burning hazardous wastes. Table C-13 summarizes the
distribution of MEI risk levels for boilers and furnaces.
Incinerators. The estimated annual baseline cancer incidence for
the three carcinogenic metals, aggregated across all 167 sites at which
EPA estimates such metals are contained in hazardous waste that are
incinerated, is approximately 0.03, or roughly 2 cases in 70 years for
the U.S. as a whole. Hexavalent chromium accounts for over half of the
predicted annual cancer incidence, with cadmium and arsenic contributing
approximately 34 percent and 13 percent, respectively. Twenty-two
"Base case" waste is not a "typical" waste in that it contains both
metals and organic constituents. It contains metals equal to the 50th
percentile values for wastes that contain metals (rather than the 50th
percentile values for all hazardous wastes, including those containing no
metals). Both POHC and chlorine content are higher than reported for a
large number of actual waste streams. Thus, the hypothetical base case
waste could result in greater risk when burned than many types of
hazardous waste that may be burned for fuel. "High risk" waste consists
of 90 percent organic constituents and the 90th percentile levels for
metals. (Reference 3, p. 5-6)
C-27
-------
TABLE C-ll
ESTIMATE OF EXCESS ANNUAL AND LIFETIME CANCER INCIDENCE FROM
HAZARDOUS WASTE COMBUSTORS - INCINERATORS
Pollutant
Arsenic
Cadmium
Chromium (VI)
Total
Annual Cases
Baseline After Comoliance3
0.005 0.003 (0.001)
0.012 0.007^ (0.004)
0.018 0.009 (0.005)
0.034 0.019 (0.011)
Cases
' Baseline
0.318
0.824
1.248
2.39
per 70 Years
After Cornel iancea
0.184
0.509
0.603
1.297
(0.103)
(0.299)
(0.368)
(0.771)
SOURCE: Reference 2, pages 128 and 132.
a Numbers not in ( ) represent compliance with the proposed rule that
would require controlling emissions such that a maximum individual
risk level of 1 x 10"5 is not exceeded at any individual facility.
Numbers in ( ) represent control of emissions such that a maximum
individual risk level of 1 x 10'6 is not exceeded at any individual
facility.
C-28
-------
TABLE C-12
ESTIMATE OF EXCESS CANCER CASES OVER 70 YEARS FROM HAZARDOUS WASTE
COMBUSTORS - BOILERS AND FURNACES
Pollutant
POHCs
PICs
Metal
Total
POHCs
PICs
Metals
Total
Type of
Waste0
Base Case
High Risk
Control Device Performance
Rasp fa<;pa
Baseline
1
1
16
18
25
4
582
611
After Regulation
0
0
15
15
2
0
- 292d
294d
Pp
-------
TABLE C-13
DISTRIBUTION OF MEI FROM HAZARDOUS
WASTE COMBUSTORS - BOILERS AND FURNACES
MEI
>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
Waste0
Base Case
'
High Risk
Control Device Performance
Base Case8
Basel ine
0"
0
10
61
103
778
952
0
19
100
167
198
468
952
After Regulation
0
0
6
48
56
650
759d
0
0
73
52
35
595
755d
Pessimistic0
Basel ine
0
C
10
65
101
777
952
0
21
102
167
207
456
953
After Reaulation
0
0
6
48
72
634
759d
0
0
73
58
36
585
752d
Mote: (lumbers in table indicate the numbers of hazardous waste combustors associated with each
maximum exposed individual risk level.
SOURCE: Reference 3, Exhibits 7-6, 7-9, 7-12, and 7-14.
8 "Base case" assumes "typical" removal efficiencies for control devices.
b "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.
0 See Footnote 5 on page C-28 for description of types of waste.
d Difference in total device due to some devices that discontinue burning due to the regulations.
C-30
-------
incinerators are estimated to pose a risk of 1'x 10"5 to the MEI under
baseline conditions.
After compliance with the proposed rule, which would require
control of emissions such that the maximum individual risk at any
facility is no greater than 1 x 10~5, EPA conservatively estimates that
the annual cancer incidence for these metals could be reduced from 0.03
to 0.02, or a reduction of approximately one lifetime cancer case in a
70-year period. The risk reduction may be understated as the actual
environmental protection afforded by the recommended control
technologies at each affected facility could be higher.
Boilers/Furnaces. Assuming base case waste composition and base case
control device performance, 18 excess cancer cases are estimated over
the next 70 years from the baseline annual level of burning. If all
devices were to burn high risk waste, baseline burning practices are
predicted to cause 611 excess cancer cases over the next 70 years. Most
of the cancer cases in both scenarios are attributable to metals
emissions. After compliance with the proposed rule, the estimated
excess cancer cases drop to 15 over 70 years for the base case wastes
and to 294 over 70 years for the high risk waste. All 15 excess cancer
cases after compliance are attributable to metals emissions, while 292
out of the 294 excess cancer cases are attributable to metal emissions
after compliance under the high risk waste scenario. The pessimistic
control device performance assumption has little effect on aggregate
cancer cases when base case waste are assumed to be burned, but a
slightly more pronounced effect when high risk wastes are assumed to be
burned.
Maximum exposed individual risks were also calculated (see Table
C-13). Of the 952 devices burning base case hazardous waste under
C-31
-------
baseline conditions, 10 are estimated to result in a MEI of 1 x 10"5, 61
in a MEI of 1 x 10"6, 103 in a MEI of 1 x 10"7, and the remaining 778
devices in a MEI of less than 1 x 10"7. Burning high risk waste
increases the MEI and the number of devices estimated to pose higher MEI
risk levels. For example, 19 devices are estimated to pose a 1 x 10"
MEI when burning a high risk waste. After compliance, the number of
devices for each of the MEI levels decreases with some control devices
projected to discontinue the burning of hazardous wastes due to the
regulations. The pessimistic control device performance assumption has
little effect on the distribution of devices among the various MEI
levels.
The analysis of human health risks from burning hazardous wastes
is very uncertain and suffers from several important limitations. The
major limits of the analysis include:
. The calculations suffer from the lack of information about key
toxics such as hazardous waste composition, cancer potencies,
and baseline control device performance. Some wastes being
burned as fuel in boilers and industrial furnaces may be less
contaminated than the base case waste and in other cases may be
more contaminated than the base case. Therefore, the base case
waste scenario will either overstate or understate risks for
specific facilities.
. The analysis does not consider possible effects of clustering
of devices in the same general locations. While such
clustering would not affect aggregate cancer case estimates,
the distribution of cancer risks across the population and to
the MEIs would be altered.
In calculating aggregate cancer cases from boilers and
industrial furnaces, it is assumed that wastes displaced from
burning under the rule will be burned in certain kilns and
industrial furnaces, and will present risks equal to the
average for these devices. Net reductions in cases may be over-
or understated depending on the accuracy of this assumption.
No adjustments are made to reflect risks from displaced wastes
when calculating the distribution of devices by MEI cancer
risks and threshold ratios.
C-32
-------
References
1. U.S. Environmental Protection Agency, Office of Solid Waste.
Burning of Hazardous Waste in Boilers and Industrial Furnace'
Proposed rule and request for comment. 52 FR 16982. May 6,
'1987.
2. U.S. Environmental Protection Agency, Office of Solid Waste. Draft
Preamble for Hazardous Waste Incinerator Regulation. June 14, 1988.
3. U.S. Environmental Protection Agency, Office of Solid Waste.
Regulatory Impact Analysis for Hazardous Waste Boilers and
Industrial Furnaces. Draft.
C-33
-------
Municipal Waste Combustors (MWCs)
(NOTE: The EPA is currently developing a revised New Source Performance
Standard (NSPS) for MWC emissions. The new NSPS, which was proposed on
December 20, 1989, does not contain estimates of cancer risk, although
work associated with it did revise cancer risk estimates from previous
efforts. The most recent risk estimates are shown in Table C-17 in
comparison with the previous estimates. The newer risk estimates do not
show the breakdown of risk by pollutant. For purposes of risk estimates
presented in Appendix B, the individual pollutant risks reported in
Table C-16 have been used, but have been cut in half to generally
reflect the overall decrease in estimated risk from MWCs.)
The Environmental Protection Agency's (EPA) Office of Air Quality
Planning and Standards has conducted a multipollutant risk assessment of
air emissions from existing and projected/new MWCs (incinerators).
Based on the results of this study, the Administrator determined that
EPA will regulate MWC emissions through the development of a revised new
source performance standards for municipal incinerators (Sections lll(b)
and lll(d) of the Clean Air Act). There are three major types of MWCs:
massburn, modular and refuse derived fuel (RDF). The number, type and
capacity of both existing and projected MWCs are summarized in Table
C-14. The pollutants evaluated in this risk assessment are summarized
in Table C-15. Other pollutants were not evaluated due to the lack of
emissions and health effects data. Since limited data were available on
short-term emissions, the risk analyses focused on long-term health
impacts. The estimated risks for existing MWC ranged from 2 to 40
cancer incidences per year with an estimated maximum individual risk
(MIR) of 1 in 1000 (1 x 10"3). The estimated risk for projected/new
sources ranged from 2 to 20 cancer incidences per year with an MIR of 1
in 10,000 (1 x 10'4).
As shown in Table C-14, the number and type of MWCs range from a
relatively large number of small modular facilities (average design
capacity of 100 tons/day) to a small number of large capacity
C-34
-------
TABLE C-14
NUMBER AND CAPACITY OF EXISTING AND PROJECTED MWCs IN THE U.S.
EXISTING MWCs
DESIGN TYPE
A.
B.
C.
D.
E.
MASSBURN
- No heat
recovery
- With heat
recovery
MODULAR
- No heat
recovery
- With heat
recovery
RDF
- With heat
recovery
UNKNOWN
TOTAL
NO.
MWCs
21
24
17
39
10
0
111
CAPACITY
(METRIC TONS/
DAY)
13,000
20,100
600
3,900
11,400
0
49,000
PROJECTED MWCs
CAPACITY
NO. (METRIC TONS/
MWCs DAY)
0 0
118 . 113,000
0 0
24 5,000
31 39,000
37 36,000
210 193,000
SOURCE: Reference 2, pages 15 and 18.
C-35
-------
TABLE C-15
POLLUTANTS EVALUATED IN MUNICIPAL WASTE COMBUSTOR
RISK ASSESSMENT
Arsenic
Beryllium
Cadmi urn
Chiorobenzenes
Chlorophenols
Chromium*6
Chlorinated Dibenzo-p-dioxins and Chlorinated Dibenzofurans (COD/CDF)9
Formaldehyde
Hydrogen chloride
Lead
Mercury
Polychlorinated biphenyls (PCB)
Polycyclic aromatic hydrocarbons (PAH)
aThe terms dioxins and dibenzofurans refer to a group of 75
chlorodibenzo-p-dioxin compounds and 135 chlorodibenzofuran compounds,
each having similar chemical and physical properties'.
C-36
-------
RDF facilities with an average design capacity of 1140 tons/day. The
projected/new MWC facilities are expected to be similar in size except
for the modular units which, on average, are expected to double in size.
Due to the limited number of existing facilities and because the data
base contained the stack parameters and control technology status
necessary for a risk analysis, a detailed risk assessment was conducted
for the existing sources. Model plant data were used to estimate risk
from the projected/new MWC facilities. The Human Exposure Model was run
for each existing facility and for the model plants using average
emissions factors based primarily on available U.S. emission test data.
Emission factors varied by design type and for existing and projected
facilities. The analysis considered the cancer risk impacts for
existing control levels and regulatory requirements (see Table C-16).
Annual incidence was estimated to be from 4 to 60 and maximum individual
risk levels from 10"3 to 10"4. The risks from MWCs are dominated by the
dioxin emissions. In most cases, over 90 percent of the estimated risk,
is from dioxin/furans.
There are significant uncertainties effecting this analysis.
There are a wide range of emissions data found in the MWC data base,
with average emission estimates used in this analysis. The feed
material are heterogeneous and vary from day to day, season to season.
The thirteen pollutants considered in this analysis are only a small
portion of the total air emissions from MWCs, therefore the risks from
this portion of emissions are not known and not represented in this risk
analysis. Also, there is significant uncertainty in dioxin emissions
due to variability in stack sample recovery results (from 10 to 100%
reported pollutant recovery) and homolog versus isomer specific
C-37
-------
CO
o
vo
I
c_>
CO
o
t-
a.
12 I?
- > ro
ro "5 "
^ 1o
i5
n ca ai
So 0
O » OJ
cu ro ~p
'i?l'i
Q. <
xg.2
UJ i C£
X
CO
O (LTD
3 li
go o
01
'!< c c
UJ «£ >
en
<0
O
O
O
CM
a
co a
c o
'x ^
o to
c
Tj CO
- -a
u
o
1
O
o
4-*
O
O
ro
0
0
,
0
*"*
^
o
o
o
CD
CO
O
O
O
"o
ro
o
o
o
CD
o
4-1
CD
0
O
o
to
°o
c
cu
f
o
.C
CJ
0
o
o
0
o
o
o
0
0
-0
CD
o
^
o
CM
CD
O
o
0
o
o
CO
N
C
o
CJ
N-
0
o
CO
0
CM
0
CD
CO
O
O
CD
CD
O
CD
JZ
m
CO
o
u~
in
o
o
N-
o
0
ro
o
4-1
in
o
o
in
CD
o
^_
I
O
*o
CD
O
0
0
C
o
t-
CD
8
D
O
4_i
g:
CO
U
u
"o
a_
CD 0
o o
O CO
1 1
o o
CVJ *
CD CD
in -*
CD O
o o
CO N-
1 1
CD O
CJ
CD CM
CD 0
CO
>^
c
0)
iL
!S
-o
ru
4-*
CO
c.
o
o
g 'E
>r S
S. $
CO
CD
r
0
4-»
,_
S
0
o
0
0
-0
'o
o
Q>
O
CM
0
O
E
^
*_
0
*~
o
J-
o
o
^.
1
o
CM
O
4-*
C
CU
1
ro
S
S
£
i_
CJ
0
*~
o
*"*
*o
1
0
o
CM
O
4-1
CM
ro
'o
o
V
o
o
o
4-"
CM
O
CO
4-*
O
t
1
O
Cf.
g"
Q.
CM"
0)
o
CL) CJ
0) CD E C
-C D Q
4-> C 4-» Q.
C
CO 4-> CO -r-
§O 4-1
cu co ca
*- CD a. c co
CO t_ X ' CD
co n QJ 3 c
^- D) 0
= .r- CO ' CD
**" co "o eo
C 4-» tO 4- CD
r- CO
CD t- "O
CO O £- 0> CD
CD ' CU O CO
O M- O C CO
C C 03 CU
a> c CD u t_
E- D) O CL
0) "- D) X
H- CO CO C CJ
£ 0 CD 'EL CO
p c o o
O X '
4-» CU 0) >.
O CO O > -M
cu cu 4-1 **- CD -
_!; .- o -a -*
» o
CO CO D 3 4J t_
r- 0 0 C - O-
£_£-'
cu co ro co to H- i
O O C I- -Q O
c a. o cj o «-
n o > L.
O t- 4-» CO Q. C CO
a. c o to
CU CD CU CU
p- C C 4-» 4-* CD CJ
4-» CD *- *- CO
CU CO CO 4-> CO
H- O) O CO CO C CD
' C 4-» CJ t_
^ .^_ -p -p o a.
4-> CJ CU CD C X
-* CO 3 C C O CD
3 X ^ «£ CO
"O <1> CJ CO CO C
- ^C CO CO O
"2 *"* .2 "~ '*" C '-M
F- C CO i *
!!~S<-
8. ^ *^ ro 3"-^> o
C-38
-------
analysis. The variability in the dioxin emissions are the primary
source of the range in risks frpm MWCs.
The evaluation of stack emissions from MWCs was limited to
pollutants for which emission test data were available and some
indication of public health or welfare were reported. Data were
sufficient for analysis of 13 pollutants or classes of pollutants as
summarized in Table C-15. On a total mass basis, the predominant
emissions are carbon monoxide, hydrogen chloride, nitrogen oxides, and
sulfur oxides.
As part of the original effort associated with the proposed MWC
NSPS for new facilities (lllb) and emission guidelines for existing
facilities (Hid), a risk assessment for baseline emissions from MWCs
was conducted. -This assessment used recently developed emissions data
in conjunction with 17 model plants representing existing MWC (Hid)
facilities and 10 model plants representing new (lllb) facilities to
estimate cancer risks from direct inhalation exposure. A comparison of
the baseline risks developed for the proposed NSPS and emission
guidelines and the previous 1987 study are presented in Table C-17. As
seen in Table C-17, the new estimates reduce estimated annual incidence
between 25 and 50 percent.
References
1. Morrison, R., U.S. Environmental Protection Agency, Pollutant
Assessment Branch. Municipal Waste Combustion fMWCs). September 7,
1988. 8 pages. ~
2. U.S. Environmental Protection Agency. Office of Solid Waste and
Emergency Response. Municipal Waste Combustion Study: Report to
Congress. EPA-530-SW-87-021a. June 1987.
C-39
-------
TABLE C-17
MUNICIPAL WASTE COMBUSTOR BASELINE RISK ESTIMATES
Old Estimate
(1987)a
New Estimate
(1989)a
New MWCs
Annual Incidence
MIR
2 to 20
10"4 to 10"6
1 to 5
10
-5
Existing MUCs
Annual Incidence
MIR
2 to 40
10'3 to 10"6
1 to 15
10
-4
SOURCE: Personal communication. Ray Morrison, U.S. EPA, Pollutant
Assessment Branch.
a Only direct inhalation.
C-40
-------
Municipal Solid Waste Landfills
Regulations are being proposed to control air emissions from
municipal solid waste landfills under the Clean Air Act. New source
performance standards are being developed under Section lll(b) for newly
constructed landfills. Emission guidelines are being developed under
Section lll(d) for existing landfills. The emission guidelines will be
implemented by the States through plans approved by EPA.
Municipal solid waste landfill emissions are a complex aggregate
of compounds. The gas that is generated from the decomposition of waste
consists of approximately 50 percent methane, 50 percent C02, and trace
constituents of non-methane organic compounds (NMOCs). Public health
and welfare concerns are from NMOCs--which are composed of volatile
organic compounds, some of which are toxic; and methane emissions which
contribute to global warming and can cause explosions at' or near
landfills. The proposed regulations would set an annual emission cutoff
for NMOCs, that when controlled at affected landfills, would reduce the
bulk of the NMOCs, toxics, and methane emissions.
Best Demonstrated Technology (BDT) consists of an active gas
collection system and an add-on control device as applied to landfills
emitting large quantities of emissions. The add-on control, device
required at a minimum is a flare. The regulation would also encourage
the use of energy recovery devices such as boilers, internal combustion
engines, and gas turbines.
A background information document for the proposal is being
revised and should be available by the end of the 1990. A copy of the
document can be obtained by contacting Alice Chow, EPA/OAQPS at FTS 629-
5626 or (919) 541-5626 or Mark Najarian, EPA/OAQPS at FTS 629-5393 or
(919) 541-5393.
C-41
-------
Publicly Owned Treatment Works fPOTWs)
Estimates of emissions from publicly owned treatment works (POTWs)
were developed as part of the NESHAP development program. POTWs were
identified as significant emitters of potentially hazardous air
pollutants (PHAPs) during the source assessment work for the individual
pollutants. Data collected by the Office of Water Regulations and
Standards were used to identify industries discharging PHAPs to POTWs.
Site-specific loadings and model plant loadings were combined to
generate the current industrial loadings at 1,621 POTWs, which treat 97
percent of all industrial wastewater. The TSDF aerated tank models were
incorporated into a computer program that estimated emissions at each of
the 1,621 POTWs. The Human Exposure Model was then used to develop risk
estimates.
Results. Risk estimates were estimated for seven pollutants (see
Table C-18). These pollutants are acrylonitrile, carbon tetrachloride,
chloroform, ethylene dichloride, methylene chloride, perch!oroethylene,
and trichloroethylene. Total annual cancer risk from POTWs was
estimated to be 1.5 cancer cases per year. Approximately one-quarter of
this total was attributed to acrylonitrile (0.4 cancer cases per year).
Three pollutants (trichloroethylene, methylene chloride, and chloroform)
each were estimated to contribute 20 percent of the total, or 0.3 cancer
cases per year for each pollutant. Maximum individual increased
incidence was estimated to be 4.5 x 10"2.
On a source category basis (see Table C-19), equipment
manufacturers and the organic chemicals, 'plastics, and synthetic fibers
industries were estimated to be the largest contributors to increased
incidence at 0.51 and 0.44 cancer cases per year, respectively. This is
approximately 63 percent of the total estimated cancer risk. The pulp
C-42
-------
TABLE C-18
SUMMARY OF CANCER INCIDENCE FROM AIR TOXICS
FROM POTWS
Pollutant
Trichloroethylene
Perchloroethylene
Methyl ene chloride
Chloroform
Acrylonitrile
Ethylene dichloride
Carbon tetrachloride
Total
Emissions
(Ma/vr)
4,840
3,230
2,130
. 439
182
102
47.9
10,971
Cancer Cases
oer vear
0.3
0.07
0.3
0.3
0.4
0.09
0.03
1.49
SOURCE: Reference 1,
_
C-43
-------
TABLE C-19
SUMMARY OF CANCER INCIDENCE FROM AIR
TOXICS FROM POTWs, BY SOURCE CATEGORY
Source
Category
Equipment Manufacturing and Assembly
Hazardous Waste Treaters
Pulp and Paper Manufacturing
Organic Chemicals, Plastics, and
Synthetic Fibers
Pharmaceutical Mfg.
Pesticides Mfg.
Electrical and Electronic Components Mfg.
Electroplating and Metal Finishing
Industrial Laundries
Textile Mills
Paint Manufacture and Formulation
Leather Tanning and Finishing
Petroleum Refining
Small Quantity Industrial Commercial, and
Residential
Number
of Sites
5,317
641
262
424
87
39
267
712
"1,000
1,411 .
518
150
45
24,177
Totals
Emissions,
Potentially
Hazardous
Air
Pollutants
8,710
312
254
248
179
92.3
34.7
33.4
31.1
11.9
10.3
1.61
1.35
1,060
10,980
Mg/yr
Total
Hazardous
Organics
19,200
1,676
965
3,970
680
138
798
73.5
404
48.8
35
85.3
331
6,570
34,975
Annual
Cancer
Cases
0.51
0.059
0.095
0.44
0.084
0.076
0.0026
0.002
0.0023
0.0158
0.0023
0.00032
0.001
0.019
1.48
SOURCE: Reference 1.
C-44
-------
and paper industry and the pharmaceutical industry were estimated to be
the next largest contributors to cancer risk from POTWs, each
contributing approximately 6 percent of the total risk.
References
1. 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.
C-45
-------
Radionuclides
Background. The EPA's Office of Radiation Programs (ORP) has evaluated
radionuclides as a hazardous pollutant, based on the widespread human
exposure to radionuclides in the ambient air, and on numerous studies
that document the incidence of cancer resulting from exposure to
ionizing radiation in many species of animals and human populations.
Subsequently, EPA has listed radionuclides as hazardous air pollutants
under section 112 of the Clean Air Act and has promulgated emission
standards or work practices for seven categories of sources: (1)
Department of Energy Facilities; (2) Nuclear Regulatory Commission-
Licensed Facilities and Non-DOE Federal Facilities; (3) Elemental
Phosphorous Plants; (4) Licensed Uranium Mill Tailings; (5) Underground
Uranium Mines; (6) Uranium Fuel Cycle Facilities; and (7) Phosphogypsum
Stacks.6 Exposure to indoor concentrations of radon due to radon in
soil gases entering homes through foundations and cellars was not
included in this rulemaking.
Results to Date. The most recent estimates available on cancer risk due
to exposure to radionuclide emissions to air are from a background
information document in support of rules for radionuclides emissions to
the air (see Reference 2). Table C-20 summarizes the cancer risk
estimates from radionuclides and Table C-21 summarizes those from radon.
As seen in these two tables, total estimated cancer incidence is
approximately 4 fatal cancer cases per year. Maximum individual risks
range from 7 x 10"6 to 4 x 10'3.
6 Other sources that can contain and emit radionuclides include coal
and oil combustion, drinking water aerators, municipal waste combustors,
publicly owned treatment works, sewage sludge incinerators, Superfund
sites, TSDFs, waste oil combustors, and woodstoves.
C-46
-------
TABLE C-20
CANCER RISKS FROM RADIATION SOURCES EXCLUDING RADON
Source
Category
NRC- Licensees
DOE Facilities
High-level Wastes'3
Uranium
Fuel Cycles
Elemental
Phos. Plants
Coal Fired
Boilers
TOTALS
SOURCE: References 2
a Maximum individual
individual risks.
and not individual
Maximum
Number of . Individual
Sources Risk3
6,000 2 x 10~4
27 2 x 10~4
0 *
136 2 x 10'4
8C 6 x 10'4
50,000d 7 x 10~6
1,200e 3 x 10"5
and 3.
risk is for one facility; other facilities
The maximum individual risk estimates for
boi lers'.
Fatal
Cancers/yr
0.2
0.3
*
0.1
0.07
0.4
0.4
1.5f
are estimated to
boilers are based
Population
w/in 80 km
240,000,000
67,000,000
*
240,000,000
1,800,000
240,000,000
have lower maximum
on typical boilers
There are no high-level waste disposal facilities operating in the U.S. (Reference 2, p. 5-1).
Of these 8, five are operating and three are closed. Risk estimates based on operating plants
only. Estimated maximum individual risk and fatal cancers per year for the three idle plants are
- 9 x 10 J and 0-04, respectively.
Industrial boilers (most of which are much smaller than utility boilers).
e Utility boilers.
Based on Reference 3, total cancer effects (fatal plus nonfatal) would be approximately 3 cancer
cases per year.
C-47
-------
TABLE C-21
CANCER RISKS FROM RADON SOURCES
Source
Category
Underground
Uranium Mines
Open-Pit
Uranium Mines
Uranium Mill
Tailings
(existing)
Disposal of
Uranium Mill
Tailings
Radon from
DOE Facilities
Phospho-
gypsum
Stacks
TOTALS
Maximum
Number of Individual
Sources Risk3
15b 4 x 10"3
(1 site)
1,300C 5 x 10~5
(.2 sites)
26d 3 x 10"5
<1 site)
50 3 x 10"4
5 1 x 10~3
(1 sites)
66e 9 x 10~5
(2 sites)
Fatal
Cancers/yr
0.8
0.03
0.0043
0.07
0.07
1.0
2.0f
Population
w/in 80 km
2,200,000
30,000,000
1,900,000
9,400,000
28,000,000
95,000,000
SOURCE: References 2 and 3.
8 The number of sites associated with the maximum individual risk is shown below the risk estimate
in parentheses. Other facilities are estimated to have lower maximum individual risks. Number
of sites for disposal of uranium mill tailings with this MIR was not identified.
b In 1982, there were 139 underground uranium mines in operation in the U.S., Currently, thirteen
are producing ore and two are on standby.
c Over 1,300 surface uranium mines have been identified in tlie U.S. The risks are based on 265
mines, which account for over 99 percent of all surface uranium ore production; 2 are operating
and the other 263 are closed or in varying states of reclamation.
d Of these 26, four are operating, eight are on stand-by, and 14 are being or have been
discontinued. Cancer risks based on the twelve operating and stand-by facilities for operating
and standby phases only.
e Of the 66 identifiable phosogypsum stacks, 63 are addressed in this assessment.
f Based on Reference 3, total cancer affects (fatal plus non-fatal) would be approximately 2.1
cancer cases per year.
C-48
-------
Sources of Uncertainty. Source term measurement errors are not
considered significant compared to other uncertainties.
Atmospheric dispersion models are a major source of uncertainty.
Studies have indicated that an uncertainty of approximately a factor of
about 2 for locations within 10 kilometers of the release point can be
expected for estimates of annual average concentrations.
Dose estimates based on unit concentrations of radionuclides are a
major source of uncertainty. Much of this uncertainty reflects real
differences in individual characteristics within the general population.
Dose estimates should be accurate within a factor of three or four.
Risk estimate uncertainties are believed to be within a factor of
three of the true value. Risk estimates are continually being re-
evaluated as new information becomes available.
References
1. U.S. EPA, Office of Radiation Programs. Risk Assessment
Methodology. Environmental Impact Statement. NESHAPs for
Radionuclides. Background Information Document - Volume 1.
520/1-89-005. September 1989.
EPA
U.S. EPA, Office of Radiation Programs. Risk Assessments.
Environmental Impact Statement. NESHAPs for Radionuclides.
Background Information Document - Volume 2. EPA 520/1-89-006-1
September 1989.
U.S. Environmental Protection Agency. National Emission Standards
for Hazardous Air Pollutants: Radionuclides. Final rule and
notice of reconsideration. 54 FR 51654. December 15, 1989.
C-49
-------
Sewage Sludge Incinerators
On an annual basis, approximately 1.7 million dry metric tons of
sludge are estimated to be incinerated in 282 sludge incinerators at 169
publicly owned treatment works (POTWs) in the United States. The
incineration of sewage sludge is regulated under the Clean Air Act, the
Resource Conservation and Recovery Act, the Clean Water Act, and the
Toxic Substances Control Act. Sewage siudge>incinerators use wet
scrubbing systems to control emissions. These systems have been
designed primarily to control particulate emissions to meet both Federal
and State requirements.
The Office of Water, U.S. EPA, proposed standards on February 6,
1989, (54 FR 5746) that would control seven toxic metals and total
hydrocarbons from sewage sludge incinerators (see Table C-22). As part
of this regulatory work, the Office of Water estimated both cancer and
noncancer risk. The unit risk values for cancer risk were estimated
based upon work completed by the U.S. EPA Carcinogen Assessment Group
(CAG). The risk assessment considers only exposure due to inhalation.
In brief, the methodology used to estimate risk combined site-
specific treatment plant data with air dispersion information for ten
sites that serve as model facilities. Each POTW was assigned to one of
the 10 model incinerators. Although these model facilities served as
the basis for the fate and transport modeling, individual
characteristics (e.g., volume of sludge incinerated daily) of each
incinerating POTW were used in the risk analysis. One facility in each
of the 10 groups of incinerators was modeled to determine its air
dispersion characteristics by using the Industrial Source Complex
Long-Term (ISCLT) model supplemented by LONGZ model and the COMPLEX I
model to account for terrain effects in urban and rural settings,
C-50
-------
TABLE C-22
POLLUTANTS FROM SEWAGE SLUDGE INCINERATORS
FOR WHICH STANDARDS HAVE BEEN PROPOSED
Pollutant
Arsenic
Beryllium
Cadmium
Chromium
Lead3
Mercury
Nickel
Total Hydrocarbons'5
Carcinogen (C)/Noncarcinogen (NC)
C
C
C
C
C
NC
C
C, NC
Lead has recently been designated as a B2 carcinogen.
Includes both carcinogens and noncarcinogens. Carcinogenic
hydrocarbons include such compounds as carbon tetrachloride, vinyl
chloride, and PCB's.
C-51
-------
respectively. Population data for each of the facilities were generated
from the Human Exposure Model (HEM).
The results of the risk analysis showed that, under current
conditions, exposure to seven metals and total hydrocarbon emissions
from sewage sludge incinerators results in a projected upper bound
estimate of 13 cancer cases per year and a maximum individual risk (MIR)
of 5 x 10"2 summed across all pollutants. Most of the annual cancer
incidence is projected to result from exposure to cadmium (see Table
C-23). However, adjusting the unit risk factors to those reported in
Table 2-6 of this report results in an estimated 37 cancer cases per
year, with most of the annual incidence attributed to vinyl chloride.
This occurs because the unit risk factor for vinyl chloride is
approximately 10 times larger than that used in the sewage sludge
incinerator study.
The estimates of risk from sewage sludge incinerators are
especially sensitive to the assumptions made concerning the metal
removal efficiencies of the scrubbers, and the percent of chromium
emissions that is hexavalent. Other factors affecting the risk
estimates include: (1) the assumption that all particulate emissions
remain airborne (thus maximizing their potential for inhalation by the
maximum exposed individual) and (2) the constituent concentrations in
the sewage sludge being incinerated. The constituent concentration data
used in the analysis are believed to underestimate the content of
organic pollutants and to overestimate the content of metal pollutants
in the sewage sludge. This uncertainty is due to the fact that the data
on the sewage sludge used in the risk assessment were collected prior to
the implementation of many pretreatment programs. Pretreatment programs
that are available for a limited number of metals may lower the
C-52
-------
TABLE C-23
ESTIMATED CANCER INCIDENCE BY POLLUTANT
FOR SEWAGE SLUDGE INCINERATORS
Pollutant
Cancer Cases Per Year
Arsenic
Beryllium
Cadmium
Chromium
Nickel
Total Hydrocarbons
Acrylonitrile
Aldrin
Benzene
Benzidine
Benzo(a)pyrene
Bis(chloromethyl)ether
Chlordane.
Chloroform
Chloromethane
Chloromethyl methyl ether
Dibenzo(a,b)anthracene
1,2-Dibromo-3-chloropropane
Dieldrin
Diethylstilbestrol
Heptachlor epoxide
2,3,7,8 Hexachloro-dibenzo-p-dioxin
3-Methylcholanthrene
2-Nitropropane
N-Nitrosodiethylamine
N-Nitrosodimethylamine
PCBs
2,3,7,8-Pentachlorodibenzo-p-dioxin
Reserpine
2 , 3 , 7, 8- Tet rach I orodi benzofuran
2,3,7,8-Tetrachlorodibenzo-p-dioxin
Other tetrachlorodibenzo-p-dioxin
Tetrachloroethylene
Vinyl chloride
0.17
<0.01
3.3
0.26
0.28
8.6
0.98
0..02
0.09
0-26
1.52
0.25
0.15
0.10
0.01
0.01
0.06
0.05
0.01
0.56
0.01
0.01
0.01
0,25
0.17
0.06
0.76
0.29
0.01
0,08
0,02
0.02
0.10
2.7a
TOTAL
13
SOURCE: Reference 1, p. 7-55.
a Adjusting this estimate to the unit risk factor reported in Table 2-6 of this
report results in an estimate of 27 cancer cases per year from vinyl chloride,
for a total of 37 cancer cases per year from sewage sludge incinerators.
C-53
-------
concentrations of these metals in sewage sludge. On the other hand, the
Domestic Sewage Sludge Exclusion of RCRA may channel more organic wastes
into municipal sewers as limits are imposed on the land disposal of
hazardous wastes, particularly liquid wastes, thereby increasing the
concentration of organic pollutants in municipal sewage sludge.
References
1. U.S. Environmental Protection Agency. Human Health Risk Assessment
for Municipal Sludge Disposal: Benefit of Alternative Regulatory
Options.
2. U.S. Environmental Protection Agency. Standards for the Disposal of
Sewage Sludge. Proposed Rule. 54 FR 5746. February 6, 1989.
C-54
-------
Super-Fund Sites
As of May 1988 there were 800 Superfund sites listed on the
National Priorities List (NPL). Approximately 20 percent of these sites
were placed on the NPL because of a high air score on the Hazard Ranking
System (HRS). This means that the site had observed air releases that
were significantly above background concentration. In addition, there
have been estimates that approximately 40 to 60 percent of the sites on
the NPL have a significant air component that must be considered either
as a result of disturbing the site to implement a remedy or implementing
the selected remedy itself (e.g., air stripping, incinerator, or soil
vapor extraction). For many of these sites, the air emissions would
include a variety of potentially toxic air pollutants.
Each Superfund site is unique as to the mix of air toxics that may
be released. This uniqueness is due to the fact that the types of air
toxics released depends on the type of hazardous materials located at
the site, which will vary from one site to the next. Most of the air
toxics data obtained has been the identification of the type of
hazardous materials at individual sites that may result in the release
of air toxics. Quantifying the levels of emissions has begun at a
number off-sites. Thus, there are no national estimates of cancer risk
from air toxics released from Superfund sites.
References
1. Memorandum. D. Dunbar, PEI Associates, Inc., to K. Meardon, Pacific
Environmental Services, Inc. Superfund Material for Update Six
Month Study. September 1, 1988. Attachment: Superfund Sites.
C-55
-------
Treatment, Storage, and Disposal Facilities for Hazardous Waste
Background. Regulations to control organic air emissions as a class
from hazardous waste treatment, storage, and disposal facilities (TSDFs)
under Section 3004(n) of the Resource Conservation and Recovery Act
(RCRA) were promulgated on June 21, 1990. These regulations apply to
process vents and equipment leaks at TSDFs. Proposal of regulations
that would apply to tanks, surface impoundments, and containers are
scheduled for late 1990. A longer term effort is planned to address
individual toxic constituent emissions as necessary to provide
additional health protection.
Nationwide cancer incidence has been estimated through summing the
results from a model that approximates the cancer incidence resulting
from each individual facility and maximum lifetime cancer risk and acute
and chronic non-cancer effects have been estimated using a model
facility. A draft Background Information Document (BID) dated March
1988 was developed to support the proposal of standards. The BID
provides a detailed review of the TSDF health risk assessment.
Results. The results of the health risk assessment for TSDF organic air
emissions indicate that there are about 140 cancer incidences per year
due to these emissions. Due to the large number of TSDF nationwide
(over 2000 facilities) and the lack of site-specific data about these
facilities, health risks have been estimated using models. Organic
emissions have been calculated for each TSDF individually through a
model that uses site-specific data where it is available and national
averages for missing information. Cancer incidences associated with
these emissions were calculated using a weighted average national cancer
potency estimate. The weighted average potency was developed by
weighing the national TSDF emissions of all non-carcinogens at a potency
C-56
-------
of zero with the national TSDF emissions of each carcinogen at its
specific potency. The results of this analysis are summarized in Table
C-24. While estimating site-specific emissions and potencies based on
national average parameters causes a high degree of uncertainty for
site-specific cancer incidence estimates, summing to a nationwide total
yields a reasonable estimate of total cancer incidence.
The TSDF health risk assessment further shows that the maximum
lifetime cancer risk to the most exposed individual is.2 in 100
(2 x 10~2). Because the emission estimates for individual facilities
that are calculated by the national emission model are highly uncertain,
another analysis was used to calculate maximum cancer risk. Two model
facilities were selected for analysis. The operation of these two
facilities were then characterized in terms of the facility layout of
waste management units and the types of wastes managed. Maximum risks
for the two facilities were calculated through emission models and
dispersion models identifying the maximum ambient concentration of
organics. The same average potency used for calculating cancer
incidence was used with the concentration to determine risk. The higher
of the risks calculated for the two facilities was used as the maximum
individual cancer risk. A major source of uncertainty in this
assessment is the selection and characterization of the facilities used
to estimate maximum risk and the use of the average potency.
The same two facilities that were used to calculate cancer risk
were also used to assess non-cancer risks. Both short-term (acute) and
long-term (chronic) non-cancer endpoints were compared to the ambient
concentrations predicted for the two model facilities. The short-term
concentration did not exceed any available level of concern and the
long-term concentrations did not exceed available acceptable daily
C-57
-------
TABLE C-24
EMISSIONS-WEIGHTED COMPOSITE UNIT RISK FACTOR (URF)
Chemical
name (carcinogen)
1 , 1-d ! ch 1 or oethy tone
1,2-dtphenyl hydrazine
l,2-d!bromoothane
1 , 2-d i bromo-3-ch 1 oropropane
1,2-dlch loroethane
l,4-d!ox»ne
2-nitropropane
acetaldehyde
acetonitr! le
aery 1 amide
crylonitri lo
aldrin
ally! chloride
an! line
benzene
benzotr} chloride
benzo(a)pyrene
benzo (b) f 1 uoranthene
benzy 1 ch 1 or i de
benz (a) anthracene
b i s (ch 1 o romethy 1 ) ethe r
bl s (2-ch 1 oroethy 1 ) ether
bis(2-othylhoxyl)phthalate
bromo-2-ch 1 oroethane
butadiene
carbazole
carbon tetrachloride
chlordane
chloroform
chloromethyl methyl ether
ch 1 oron 1 trobenzene
chrysene
creosote
fiftT
W 1
d ! benz (a , h) anthracene
dlchlorobonzene(l,4) (p)
dichloropropene
dimethoxy benzidlne, (3,3')
dimethyl phenol
dimethyl sulfate
dinitrotoluene
eplchlorohydrin
ethyl aery late
ethyl carbamate
LDR8 uncontrolled
emissions, Mg/yr
1,093
1
0
2
23,101
270
8
6,214
469.100
74
17,770
34
248.600
5,380
6164.000
21.653
2
1.219
289.800
0.230
374
0
338.062
10.310
115
46.760
16,920
8
4,586
0
2508.980
0.316
37.110
27
0.053
0.085
30.540
0.000
21.310
0.192
250.000
1,595
28.920
12.180
URF
6.0 x 10~5
2.2 x 10-4
2.2 x 10"4
5.0 x 10-3
2.6 x 10-B
1.0 x 10-6
3.0 x 10-3
2.2 x 10-6
1.0 x 10-3
6.8 x 10-5
4.9 x 10-3
1.0 x 10-5
8.0 x 10-6
1.7 x 10-3
8.9 x 10-4
3.3 X 10~4
.
2.8 x 10~4
1.5 x 10-5
3.7 x 10~4
2.3 X 10-5
2.7 x 10-3
3.0 x 10~4
1.4 x 10-2
8.8 x 10-5
1.2 x 10-6
-
URF x emissions for chemical
Total TSDF emissions
3.0 x 10-8
8.8 x 10-H
0
4.6 x 10-9.
3.3 x 10-7
1.6 x 10-10
1,4 x 10-8
7.4 x 10-9
4.0 x 10-8
6.6 'X 10-7
8.9 x 10-8
2.9 x 10-8
2.7 X 10-8
1.4 x 10-9
1.1 x 10-10
0
1.8 x 10"8
1.4 x 10-7
1.6 x 10-9
5.7 X 10-8
0
4.6 x 10-9
4.0 x 10-10
1.2 x 10-8
1.0 x 10-9
/ !_ l\
C-58
-------
TABLE C-24
EMISSIONS-WEIGHTED COMPOSITE UNIT RISK FACTORS (URF) (concluded)
Chemi ca t
name (carcinogen)
ethyl ene di bromide
ethylene imine (azaridine)
ethylene oxide
formaldehyde
gaso 1 t ne
heptach 1 or
hexach lorobenzene
hexach lorobutadiene
hexach loroethane
hydraz i ne
i ndeno (123-cd) pyrene
lead acetate
lead subacetate
1 indane
methyl chloride
methyl cholanthrene (3)
methyl hydraz ine
methyl iodide
methylene chloride
nitrobenzene
n i tro-o-to 1 u i d i ne
n-n i trosopy rro 1 i d i ne
n-n i troso-n~methy I urea
parathion
pentach 1 oroethane
pentach 1 oropheno 1
phony 1 ene diamine
polychlorinated biphenyls
propylene dichloride
sty rene
TCDD (tetrach lorodi benzo-p^dio)
tetrach loroethane(l ,1,1,2)
tetrach 1 o roe thy 1 ene
th i ourea
toluene diamine
toxaphene
tr i ch 1 oroethane (1,1,2)
tr i ch 1 oroethy 1 ene
trichlorophenol
v i ny 1 ch 1 or i de'
Total nationwide
uncontrolled emissions
LOR uncontrol led
emissions, mg/yr
10
51640.000
0.000
2,646
2,742
1
158
45780.000
1,553
238
0.033
1.901
0.000
9.5 x 10~5
58
5
8
0.000
16,676
5438.900
0.000
0.000
0.000
75.950
2458.000
27.630
1171.000
0.061
45.460
682.499
0.310
7,135
17,271
5
21.718
56
18,458
56,353
30
626
1,839,267
URF
2.2 x 10-4
1.0 x 10-4
1.3 x 10-5
6.6 x 10-7
1.3 x 10-3
4.9 x 10-4
2,2 x 10-E
4.0 x 10-6
2.9 x 10-3
3.8 x 10-4
3.0 x 10-3
4.7 x 10-7
6.1 x 10-4
8.6 x 10-2
33
5.8:x 10-5
5.8 x 10-7
5.5 x 10-4
3.2 x 10-3
1,6 x 10-5
1.7 x 10-6
5.7 x 10-6
4.1 x 10-6
URF x emissions for chemical
Total TSDF emissions
. 1.2 x 10-9
0.000 .
1.9 x 10-8
9.8 x 10-10
8.6 x 10-10
4.2 x 10-8
5.4 x 10r7
3.4 x 10-9
3.8 x 10-7
2.0 x 10-14
8.6 x 10-9
4.3 x 10-9
0
0
5.6 x 10-6
2.3 x 10-7
S.4 x 10-9
1.5 x 10-9
9.8 x 10-8
1.6 x 10-7
5.2 x 10-8
9.6 x 10-H
1.4 x 10-9
8.6 x 10-6
aLDR = Land disposal restrictions.
C-59
-------
intake benchmark levels. The major source of uncertainty, beyond the
uncertainty associated with the selection and characterization of the
facility as noted above under cancer risk, is the characterization of
the specific non-carcinogenic constituents and their concentrations.
References
1. U.S. EPA, OAQPS. Hazardous Waste TSDF - Background Information for
Proposed RCRA Air Emission Standards. Volume I - Chapters.
Preliminary Draft EIS. March 1988.
2. U.S. EPA, OAQPS. Hazardous Waste TSDF - Background Information for
Proposed RCRA Air Emission Standards. Volume II - Appendices.
Preliminary Draft EIS. March 1988.
C-60
-------
Waste Oil Combustors
In 1985 estimates supplied by the Office of Solid Waste, about 638
million gallons of waste oil were combusted. Based on the available
data, there are nine hazardous constituents in waste oil known to be
emitted from combustion sources. Cancer risks were estimated for nine
pollutants based on the availability of cancer potency estimates, while
maximum concentration estimates were also estimated for the lead
emissions (See Table C-25). Other known hazardous emissions such as
formaldehyde are emitted from all oil combustion, but this study was
designed to focus only on those pollutants that would be emitted at
rates exceeding those found in virgin fuels. Therefore, the pollutants
included in this study were those contaminants that were not normally
found in virgin oils or those with waste oil concentrations higher than
those typically found in virgin oils.
The emission factors for the ten pollutants studied were
calculated from the typical level of these pollutants found in waste oil
(Table C-26). These values were based on several sampling studies.
Although virgin oil combustion generates very little or no bottom ash,
typical waste oils will generate some bottom ash because not all the
constituents in waste oil can be burned. Thus, assuming that 100
percent of the contaminants entering the boiler in the waste oil feed
are emitted in the flue gas would overestimate emissions. Earlier
studies provided enough data to estimate the partitioning and/or
destruction efficiency for each pollutant of concern. Since waste oil
is mostly burned in virgin oil combustion devices which are typically-
uncontrolled, it was assumed that no air pollution control devices were
being used at any facility.
C-61
-------
TABLE C-25
CANCER RISK ASSESSMENT RESULTS - WASTE OIL COMBUSTION
METHOD 1
(based on 633 residual
oil facilities)
Maximum
Individual Annual
METHOD 2
(based on 70 facilities
known to burn waste oil)
Maximum
Individual Annual
Risk Incidence
KOI lutani
Arsenic
Cadmi urn
Chromium (+6)
Trichloroethylene
Tetrachl oroethyl ene
Benzene
Benzo(a)pyrene
Polychlorinated
biphenyls
Total
Maximum Long-Term
Lead Concentration
[\ 1 3N XII
2.4 x 10"6
3.2 x 10'7
3.2 x 10"8
6.2 x 10'11
4.7 x 10"11
8.3 x 10'10
2.3 x 10"9
3.8 x 10"9
2.7 x 10'6
0.0047
0.48
0.064
0.0065
0.000013
0.0000096
0.00017
0.00047
0.00077
0.56
/iQ/m3
1.6 x 10'4
2.1 x 10"5
2.1 x 10"6
4.1 x 10"9
3.1 x ID'9
5.4 x 10'8
1.5 x 10"7
2.5 x 10'7
1.8 x 10'4
0.087
0.012
0.0012
0.0000022
0.0000017
0.000030
0.000085
0.00014
0.10
0.33 /ig/m3
SOURCE: Reference 1, page 3.
C-62
-------
ID
CM
CO
1 t t
1
CO
}
LU
j
CO h-
^g III
UJ CO
1C CO
t-.-m 1 1 1
^*~ LJUJ
CO
-=Z CO
> H ^f
Q ;*£
LU CO
CO l-H
~ ^ ^y*
<£ ^
r-^LU
r-
CO
o
r~
O
CO
CO
si
4_>
-
^3
I-"
03
-W
3
r
i
O
Q-
^_>
CQ
\j
c
o
1
CO
_a
- '
^^
CO
-t->
+->
r
E
LU
5§
^Z
+->
LO
l-~.
f-
0?
CO
(D
o
4_)
O
03
i i
LO
^D
CD 00 i i O
>f 0 0
CO l-H l-H
CO
55 55
55 55 i-H ^^
CD CD CO ID
r--. «*
CD
CO
i l ^^.
^* cn t^» »*
cr> ^- LO "^
i i t i
en CM CD CD
o o
__>
. ^ . S T-H
. ~ CD t I CM
LO rrH 1 CO
en to CM o
co cn
tO N1 CVJ
CD CD CD
i i i i i i E ^
C T g=
0) E 0-0
CO TD S- 03
S- OS _C CO
^
t
[ ^
CO
o
o
r
-£=
O
^
1
l-H
LO
55
,
CD
CD
O
CM
«s
CO
1
1 1
V
1_-
CD
00
r-.
cn
CM
cn
N.
i
CD
i t
X
00
LO
CO
c
CO
^
c~
^_3
CO
o
s-
o
[^
o
03
5-
co
1
CM
ID
5S
.
CD
CD
CD
CD
»»
LO
LO
CO
V
^
CD
CD
CM
cn
CM
I 1
**
1 1
*o
CD
I 1
X
CO
CO
o>
c
CO
M
O)
CO
*i- oo
o i
55 55
. ,
0 0
CD
in
i i
!>«. **
i 1 CO
t 1
I-H CD
V
S . %
CM CD
. i CO
OO CO
CO
s
ro to
CD CD
l-H l 1
X X
1 CO
i-H l 1
CO
c~
CO
i-
>-J
Q.
"o?
^~*
O
M CO
C OQ
CO O
CQ Q_
.
ro
E
cn
a.
i i
O
+-1
CO
i-
3
CO
o
Q.
X
CO
CO
lr-
+->
»*- CO
CO !jl
cn ,
03
°- S-
co
Q.
1 1
vx
CO CO
0 .,_
CO *"
S c
co .£:
c^.
CO CO
f*** J_
O3
CO
LU +J
co .T-
ra -§
o
CO ro -
.
^-
o
r
^_
.,_
^
S-
co
«->
CO
^
O3
pi
CO
s_
co
4^
£_
.^
Q
C-63
-------
Two approaches were used in this study. The first approach used
to assess waste oil combustion risks directly incorporated the results
of the residual oil combustion analysis. This approach assumes each
U.S. residual oil burner also combusts a proportionate amount of waste
oil. Thus, OAQPS could reuse the existing dimensionless HEM residual
oil results by making simple modifications to the Boiler Computer Model
(BCM) to accommodate the different fuel rates and emission factors
required to evaluate waste oil combustion emissions.
The second approach focused on those specific boilers that are
known to burn waste oil. These combustors were identified by searching
through an OAQPS data base" (the National Emissions Data Systems or NEDS)
to obtain a list of boilers burning waste oil. With this approach, each
boiler required a HEM analysis (using the dimensionless emissions rates)
and a BCM analysis to convert the HEM results to risk values. Since not
all of the waste oil being burned in the U.S. was accounted for in those
waste oil burners identified by OAQPS, the annual incidence results were
scaled to national estimates by the ratio of sample waste oil use to
national waste oil use. The estimated maximum individual risk
associated with this sample of boilers was assumed to be representative
of the entire population of waste oil units.
As can be noted in Table C-25, there is approximately two orders
of magnitude difference between maximum individual risk estimates for
the two approaches. Given that Approach 2 is based on actual reported
quantities .of waste oil burned, and that this risk statistic depends
largely on site-specific characteristics, we believe that the results
from Approach 2 are to be preferred. On the other hand, given the
inherent uncertainties of the risk assessment process, the two
approaches produce aggregate risk estimates that are quite similar (0.56
C-64
-------
versus 0.10 cases/year). Because the NEDs data base is biased towards
larger capacity units, which tend to produce lower estimates of risk per
unit of fuel, the aggregate risk estimates from the first approach
probably reflect more accurate estimates of national exposure.
Due to the nature of the risk assessment methodologies, an
understanding of the uncertainties within the analysis is as important
as the results. A brief summary of these uncertainties follows:
1. There are significant data gaps in our knowledge of the number
and toxicity of pollutants being emitted from waste oil combustion.
This study has included only those pollutants that have been measured in
waste oil and are known as probable carcinogens.
2. Site-specific emissions and fuel contaminant level data were
not available; average values were used.
3. The study assumes a steady-state condition in which fuel use
patterns, control technology, and population remain constant over a
period long enough to evaluate the cancer risk (a 70-year lifetime). It
is certainly reasonable to expect that one or more of these parameters
will change substantially over the study period.
4. The uncertainty in the estimates of carcinogenic potency is
considerable. For the most part, unit risk estimates represent
plausible upper bounds of the cancer risk. The estimates have been
derived from studies of workers or test animals exposed to levels of the
substances much higher than those found or modeled in the ambient air.
It is not clear how applicable these exposures are to the lower
concentrations of trace constituents present in the atmosphere due to
waste oil combustion emissions. The aggregate cancer incidence
estimates reflect the exposure of large numbers of people to low
pollutant concentrations. In addition, cancer risks were evaluated for
C-65
-------
each pollutant and total risks were calculated by adding individual
pollutant risks. Synergistic or antagonistic effects that may be
associated with complex mixtures of pollutants have not be calculated.
5. Only cancer risks have been estimated. Although this
represents an important portion of the health concern about such
mixtures, the possibility of other health effects was not examined.
6. There is a potential for exposure by routes other than direct
inhalation. Although few data are available to estimate the relative
contribution to exposure from the deposition and subsequent re-
suspension or ingestion of emitted compounds, some affect on soil and
water levels and subsequent exposure would be expected.
7. There are numerous simplifying assumptions in exposure
assessment. The estimation of human exposure requires simplifying
assumptions about the dispersion of the pollutants such as assumptions
about terrain features (assumed flat terrain) at each boiler site and
the use of the nearest meteorological data site as representative of the
study area. Maximum individual lifetime risk is particularly sensitive
to changes in such assumptions. Further, exposures beyond 50 kilometers
were not examined.
Based on the above uncertainties within this analysis, we believe
that this study does not provide accurate, absolute estimates of public
risk. The study results must be viewed as rough estimates with error
bands in the range of orders of magnitude.
References
1. Peters, W.D. and Fegley, R. U.S. Environmental Protection Agency.
Waste Oil Combustion Cancer Risk Assessment. Technical Staff Paper.
October 1987.
C-66
-------
Woodstoves
Many studies on woodstove emissions have been undertaken as a part
of EPA's Integrated Air Cancer Project (IACP). The goals of the IACP
are to (1) identify the principal airborne carcinogens, (2) determine
which emission sources are major contributors of carcinogens to ambient
air, and (3) improve the estimate of human exposure and comparative
human cancer risk from specific air pollution emission sources
(Reference 1). The initial phases of the IACP were aimed at quantifying
carcinogens emitted from residential woodstoves and motor vehicles
because data at that time indicated that these two sources made a
significant contribution to the mutagenic activity of ambient air
samples.
Initiaf IACP studies were conducted in Raleigh, NC, and
Albuquerque, NM, two communities with relatively simple airsheds where a
significant percentage of the homes use wood as the major heating fuel.
These initial IACP studies emphasized field and laboratory evaluation to
select sampling and analysis methodologies for a major field study
initiated in Boise, ID, in the winter of 1986-87. These field studies
were designed to simultaneously sample and characterize.the emissions at
the source, in the ambient air near to specific sources, and in the
ambient air distant from the sources. Sampling indoors and outdoors of
homes both with and without woodstoves was conducted to provide an
indication of total human exposure.
Other studies on woodstove emissions conducted under the IACP or
s
elsewhere have included examining (1) the mutagenicity of woodsmoke
(References 2 and 7), (2) the chemical characteristics of respirable
particulate matter (Reference 7), (3) the effect of photooxidation
C-67
-------
reactions and aging by itself on the mutagenic activity of woodsmoke
(References 3 and 4), (4) the relative mutagenic activity
of the gas-phase reactants and products compared to that of the
particulate phase (Reference 4), and (5) the relative mutagenicity and
carcinogenicity of particle bound organics in woodsmoke with a variety
of other sources of incomplete combustion products (References 5 and 6).
One study has estimated a unit risk factor of 1.0 x 10'5 for
woodstove emissions (Reference 6). Combining relative potencies with
the percentage organic extractable matter and particle emission rates
suggests that woodstove emissions make a significantly greater
contribution to ambient hazardous organics than the use of residential
fuel oil on a mutagenic emission rate per joule of energy basis. More
research is being undertaken to understand the relationship between the
emissions, potential atmospheric transformations, human exposure,
dosimetry, and final cancer risk from woodstove emissions as well as
other sources of products of incomplete combustion.
References
1. Cupitt, L. and Joel 1 en Lewtas. EPA's Integrated Air Cancer Program.
Proceedings of US-Dutch Expert Workshop on Air Toxics, Amersfoort,
The Netherlands, May 16-18, 1988, in press.
2. Claxton, L.D., et. al. The Mutagenicity of Ambient and Source
Samples from Woodsmoke-Impacted Air Sheds. Proceedings of the 1987
EPA/APCA Symposium on Measurement of Toxic and Related Air
Pollutants, APCA, Pittsburgh, PA, pp. 591-596.1987.
3. Kamens, R.M., et. al. "Mutagenic Changes in Dilute Wood Smoke as It
Ages and Reacts with Ozone and Nitrogen Dioxide: An Outdoor Chamber
Study," in Environmental Science Technology, Vol. 18, No. 7, 1984.
pp. 523-530.
4. Kleindienst, T.E., et. al. "Woodsmoke: Measurement of the Mutagenic
Activities of Its Gas- and Particulate-Phase Photooxidation
Products," in Environmental Science Technology, Vol. 20, No. 5,
1986, pp. 493-501.
C-68
-------
5.
6.
7.
Lewta's, J. "Comparison of the Mutagenic and Potentially Carcinogenic
Activity of Particle Bound Organics from Woodstoves, Residential Oil
Furnaces, and Other Combustion Sources." 1981 International
Conference on Residential Solids Fuels, pp. 606-619. 1982.
Lewtas, J. "Genotoxicity of Complex Mixtures: Strategies for the
Identification and Comparative Assessment of Airborne Mutagens and
Carcinogens from Combustion Sources," in Fundamental and Applied
Toxicology. Vol. 10, No. 4, May 1988. pp. 571-589.
Watts, R.R., et. al. "Wood Smoke Impacted Air: Mutagenicity and
Chemical Analysis of Ambient Air in a Residential Area of Juneau,
Alaska," in AP'CA Journal. Vol. 38, No. 5, pp, 652-660. July 1988.
C-69
-------
NONCANCER HEALTH RISK PROJECT
(Note: The following is the Executive Summary taken from Toxic Air
Pollutants and Noncancer Health Risks: Screening Studies (U.S.
Environmental Protection Agency, Office of Air Quality Planning and
Standards, Research Triangle Park, NC, 27711) Final External Review
Draft. September, 1990. For additional information on the noncancev*
health risk project, contact Beth Hassett-Sipple, Pollutant Assessment
Branch, (919)-541-5346.)
Greater than 2,000 man-made chemicals have been detected in
ambient air. Many of these chemicals are known to cause adverse health
effects in exposed humans or laboratory animals. Historically, the
evaluation of risks associated with exposure to toxic air pollutants has
focused on the potential for a carcinogenic response. In a recent
Agency-wide comparison of environmental risks, noncancer risks
associated with exposure to toxic air pollutant were among the Agency's
highest concerns. To better understand the potential for the occurrence
of adverse noncancer health effects as a result of exposure to routine
emissions of toxic air pollutants, the Environmental Protection Agency
(EPA) Office of Air Quality Planning and Standards (OAQPS) conducted the
subject screening studies.
Approach
The screening studies represent approaches taken to characterize
the potential noncancer risks associated with exposure to toxic air
pollutants, each looking at slightly different aspects of the question.
The initial phase included review of case reports; State, local, and
Federal agencies' experiences; health effects literature; and exposure
data (i.e., modeled and monitored ambient concentrations). From this
information, two assessments were conducted by OAQPS: (1) an evaluation
of the potential nationwide noncancer problem, and (2) a more complete
analysis of a typical industrialized urban area.
C-70
-------
Nature of the Problem
Several data sources were evaluated in the screening studies.
These included: incidences of noncancer diseases in the United States;
reports of noncancer effects linked with nonoccupational exposures to
industrial air releases; and experiences of State, local, and Federal
agencies involved in the regulation of toxic air pollutants. The data
support the finding that adverse noncancer effects are an important
public health concern and that environmental factors may play an
important role in disease incidence. A survey of State, territorial,
and local agencies indicated that a number of air releases, with the
potential to result in serious noncancer health effects in the exposed
population, are likely to occur each year. Many State and local air
pollution control agencies have required additional air pollution
control equipment for sources emitting toxic air pollutants specifically
to reduce potential noncancer effects.
Available Exposure Data
An evaluation of available exposure data for toxic air pollutants
revealed that air releases of these pollutants are widespread, but
neither a comprehensive monitoring or modeling data base nor a complete
toxicity data base exists. Biological indicators studied (e.g., human
adipose and other tissue samples) revealed that many chemicals found in
the atmosphere also have been detected in human tissues. Although other
exposure pathways besides inhalation are expected to contribute to the
presence of these chemicals in human tissue samples, air exposures can
not be ignored.
OAQPS Analyses - Broad Screening and Urban County Studies
To examine the potential association between noncancer health
s
effects and exposure to toxic air pollutants, two studies were
C-71
-------
undertaken by OAQPS. In both cases, data limitations precluded
quantification of the magnitude of noncancer risks. However, the data
do indicate ambient air concentrations of many pollutants may
significantly contribute to potential noncancer health risks associated
with environmental exposure.
The assessments were conducted by comparing modeled and/or
monitored ambient concentrations to health reference levels and lowest-
observed-adverse-effect levels (LOAELs).7 The Broad Screening Study
examined exposure to individual or multiple pollutants in ambient air
based on exposure data from many areas of the country. Exposure data
included ambient concentrations for approximately 325 pollutants
monitored throughout the United States and annual averaged ambient
modeled concentrations for approximately 40 pollutants emitted from more
than 3,500 facilities. Health data and quantitative exposure data were
available for only about 150 pollutants, less than ten percent of the
chemicals which have been detected in ambient air. For those few
chemicals with both health and exposure data, noncancer health risks
appeared to be of concern. For approximately half of these chemicals,
modeled and/or monitored levels exceeded health reference levels at
numerous sites through the country. Ambient levels for approximately
one-third of these chemicals exceeded the health reference level at more
than 25 percent of the sites studies. Less than 5 percent of sites and
7 Health reference level .- The LOAEL divided by appropriate
uncertainty factors to account for inter- and intra-species variability
and identifies their LOAEL versus a NOAEL (no-observed-adverse-effect
level).
LOAEL - 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 unexposed control group. The study
LOAEL was converted to an human equivalent level for comparison with
exposure levels in the analyses conducted.
C-72
-------
chemicals indicated ambient concentrations exceeding LOAELs. These
exceedances were seen with short-term and long-term ambient monitored
concentrations. Modeling of short-term emissions was not performed
because of data limitations.
The simultaneous presence of several pollutants in ambient air is
a frequent occurrence. When considering the potential impact of
exposure to chemical mixtures, combined exposures were of concern for
several types of health endpoints (e.g., reproductive/developmental
toxicity, respiratory toxicity, etc.) in many geographical areas. The
impacts of chemical mixtures were frequently dominated by a small subset
of chemicals. For example, 15 chemicals associated with neurotoxic'
effects may have been monitored at one location though only two or three
chemicals were monitored at concentrations that contributed
significantly to health reference level exceedances.
The second analysis involved a more detailed evaluation of a
midwestern industrialized urban county. This analysis expanded the
number of chemicals evaluated in the Broad Screening Study and assessed
the combined impact of multiple emission sources versus the impact of
sources independently. Approximately 200 chemicals from 122 point
sources plus 9 area sources were evaluated. Health reference levels and
LOAELs were compared to modeled pollutant concentrations in three
independent modeling exercises. Results suggested that a larger number
of pollutants exceeded health reference levels for short-term modeled
concentrations than for long-term modeled concentrations. Ambient
concentrations were estimated to exceed health reference levels for
long-term concentrations predicted by the Industrial Source Complex-Long
Term model and the Human Exposure Model (4 and 8 percent of pollutants
respectively) and short-term (24-hour) concentrations predicted by the
C-73
-------
SCREEN model (22 percent of pollutants). Estimated long-term
concentrations did not exceed any pollutant LOAELs. Estimated short-
term (24-hour) concentrations exceeded LOAELs for approximately 2
percent of the pollutants assessed. In general, proximity to individual
sources was a significant factor in determining degree of potential
exposure. Another important finding of this study was that the additive
contribution for a single pollutant emitted from a variety of sources
resulted in health reference level exceedances over a broad geographic
area.
Conclusions
Based upon analysis of the available data, it is clear that
environmental exposures to toxic air pollutants have the potential to
adversely impact public health. Although the magnitude of such
noncancer risks can not be estimated from the available data, the broad
implications of this study suggest that public health risks resulting
from exposure to toxic air pollutants are not limited to carcinogenicity
which has traditionally been the focus of regulatory programs. For
certain pollutants, the combined impact of multiple sources may result
in substantial exposure to many people. This finding suggests that the
problem may not be limited to large point sources, but that smaller
point sources and area sources that are numerous in populated areas can
not be ignored. Similarly, exposure to chemical mixtures may result in
adverse noncancer health risks that might not be predicted if only
impacts of individual pollutants are considered.
The sparseness of available data represents the principal
limitation of the screening studies. Few data were available to aid in
the prediction of ambient concentrations and the derivation of health
benchmarks. Despite the limitations, however, the studies support a
C-74
-------
finding that toxic air pollutants represent a potential noncancer health
risk that warrants routine evaluation.
In recent Congressional activity to amend the Clean Air Act, the
importance of adverse noncancer effects is emphasized. Many provisions
of the proposed legislation focus on better understanding potential
noncancer public health risks and controlling emissions of toxic air
pollutants in order to reduce these risks.
C-75
-------
------- |