1,1-Dichloroethylene
(Vinylidene Chloride)
Occurrence in
Drinking Water,
Food, and Air
Science and Technology Branch
Criteria and Standards Division
Office of Drinking Water
United States Environmental Protection Agency
JRB Associates
-------
OCCURRENCE OF 1,1-DICHLOROETHYLENE
IN
DRINKING WATER, FOOD AND AIR
Prepared by:
Frank Letkiewicz
Pauline Johnston, Ph.D.
Corinne Macaluso
Robert Elder, Ph.D.
William Yu
Carol Bason
JRB Associates
8400 Westpark Drive
McLean, Virginia 22102
EPA Contract No. 68-01-6388, Work Assignment 29
JRB Project No. 2-813-03-852-29
EPA Task Manager
Mr. William Coniglio
Movambef 18, 1983
-------
TABLE OF CONTENTS
Page
SUMMARY
INTRODUCTION 1
1. SOURCES OF 1,1—DICHLOROETHYLENE 3
1.1 NATURAL SOURCES 3
1.2 ANTHROPOGENIC SOURCES 3
1.2. 1 Production . . 3
1.2.2 Consumption 4
1.2.3 Releases 5
1.3 FATE AND TRANSPORT 6
2. OCCURRENCE IN DRINKING WATER 7
2.1 OVERVIEW AND QUALITY ASSURANCE ASSESSMENT
OF FEDERAL DRINKING WATER SURVEYS 10
2.2 GROUNDWATER 11
2.2.1 Federal Surveys 11
2.2.2 State Data 13
2.3 SURFACE WATER 18
2.3.1 Federal Surveys 18
2.3.2 State Data 20
2.4 PROJECTED NATIONAL OCCURRENCE OF
1,1—DICHLOROETHYLEME IN PUBLIC WATER SUPPLIES 21
2.4.1 Projected National Occurrence of
1,1—Dichloroethylene in Groundwater Supplies 22
2.4.2 Projected National Occurrence of
1,1—Dichloroethylene in Surface Water Systems 25
3. OCCURRENCE IN THE FOOD SUPPLY 33
4. OCCURRENCE INAMBIENTAIR 34
5. HUMAN EXPOSURE FROM DRINKING WATER, FOOD AND AIR 37
5.1 DRINKING WATER INTAKE 38
5.1.1 Population Exposed 38
5.1.2 Daily Intake of 1,1-Dichloroethylene
from Dri nk i ng Water . . . . . . . . . . . . 42
5.1.3 Population—Concentration and
Popul ation—Exposure Estimates 42
5.2 DIETARY INTAKE 49
5.3 RESPIRATORY INTAKE 49
5.4 RELATIVE SOURCE CONTRIBUTION 51
REFERENCES 55
APPENDIX A A—i
APPENDIX B B—i
APPENDIX C C-i
-------
ACKNOWLEDGEMENTS
The authors of this document express deep appreciation to Mr. William
Coniglio, EPA Office of Drinking Water, Science and Technology Branch, who
developed the basic concepts culminating in this series of reports on the
occurrence of volatile organic chemicals in drinking water, food, and air.
The authors also express appreciation to the following individuals at
EPA’s Office of Drinking Water for valuable input in developing the drinking
water occurrence data presented in this report: Hugh Hanson, Herb Brass,
Kitty Miller, David Schnare, Nancy Wentworth, Wayne Mello, Edward Glick,
Richard Redding, Jim Westrick, and Eugene Dotson.
Jack McGinnity, EPA Office of Air Quality Planning and Standards, is
acknowledged for his assistance in providing the information on occurrence in
air and respiratory intake.
Bruce Brower, Cornell University, is acknowledged for his assistance in
developing the data from the Rural Water Survey. David Taylor, Northern
Virginia Community College, is acknowledged for his assessments of the reli-
ability of the sampling and analysis methods used in the Federal surveys.
John Coleman, Elizabeth Jackson, and Karen Taylor of JRB Associates are
acknowledged for their contributions to a previous draft of this report.
A special acknowledgement is made to Diane Simmons for her painstaking
efforts in the word processing of all text and tables in this report.
-------
SUMMARY
This report presents information on the occurrence of 1,1—dichloro—
ethylene in public drinking water supplies and ambient air in the United
States and an estimate of human intake of 1,1—dichloroethylene from those
sources. There were no data available on the occurrence of 1,1—dichioro—
ethylene in food. Federal survey data on 1,1-dichloroethylene levels in
drinking water systems were combined and stratified according to water source
(surface and ground) and size based on population served. These data were
used to develop a national profile of 1,1—dichloroethylene levels in drinking
water that provides estimates of the number of systems of each source/size
category, and the total population served by them, at each exposure level.
The relative contribution of drinking water to total human intake of 1,1—di—
chloroethylene is estimated by analyzing data on levels of this chemical in
ambient air and estimating the likely range of 1,1—dichloroethylene combined
intake from drinking water and air for a 70 kg man and a 3.5 kg formula—fed
infant.
1,1-Dichloroethylene (CH 2 =CC1 2 ), also known as vinylidene chloride, is a
colorless liquid. It has a high vapor pressure (495 mm Hg at 20°C) and a low
water solubility (0.035 g/100 g at 25°C).
Production of 1,1-dichioroethylene in 1978 was estimated to be 144,200
kkg. This figure included 1,1—dichloroethylene captively produced for the
production of methyl chloroform. However, 1,1—dichloroethylene appears no
longer to be used in methyl chloroform production. In addition to its direct
manufacture, 1,1—dichloroethylene may be produced indirectly during the pro-
duction of other chlorinated chemicals.
The major uses of 1,1—dichioroethylene in 1978 were the production of
methyl chloroform and the production of copolymers for use in resins, coating
latex, and the manufacture of modacrylics. A minor application was its use in
the production of chloroacetyl chloride, a component of mace and tear gas.
Total releases of 1,1-dichloroethylene from production and use processes
in 1978 were estimated to be 2,000 kkg to air, 0 kkg to land, and 2 kkg to
water. However, the production and use of methyl chloroform was estimated to
account for 1,300 of the 2,000 kkg of 1,1—dichioroethylene released to air in
1978, and methyl chloroform no longer appears to be produced by this pro—
-------
cess. The production of copolyrners is currently estimated to account for
almost all environmental releases of 1,1—dichloroethylene. Quantities of
1,1-dichloroethylene entering water are expected to be volatilized and to
undergo rapid degradation in the troposphere.
Two Federal surveys were used to estimate levels of 1,1—dichioroethylene
in the nation’s public drinking water supplies: the National Screening
Program for Organics in Drinking Water (NSP) and the Groundwater Supply Survey
(GWSS). Additional state data are also reported,. but were not used in
developing the national estimates. These data came from only a few states and
were not well—characterized with respect to water type and system size
sampled.
There are approximately 60,000 public water supplies in the United States
serving approximately 214 million people. These supplies fall into two cate-
gories with respect to water source (surface water and groundwater) and into
five major size categories (with twelve size subcategories) according to the
number of individuals served. The data obtained from the two Federal studies
was combined and sorted by source and size category in order to develop esti-
mates of the number of systems nationally in each source/size category con-
taining 1,1-dichloroethylene within various concentration ranges. The
national estimate of systems was used to calculate the number of persons
exposed to public drinking water containing 1,1—dichloroethylene levels in
those ranges.
Using the combined survey data and the multinomial approach for estimat-
ing national occurrence, it was calculated that about 98% of the groundwater
systems of all sizes contain either no 1,1—dichloroethylene or levels less
than 0.2 ugh. It is not possible, however, to estimate how many of these
systems contain 1,1—dichloroethylene at low levels and how many are free of
1,1—dichloroethylene contamination. Of the estimated 858 systems expected to
have levels higher than 0.2 ug/l, 81 (0.2% of total groundwater systems) are
projected to have concentrations > S ug/l; none are expected to have levels
> 10 ug/l. The state data, however, indicate that there may be some supplies
with levels substantially higher than 10 ug/l
For surface water supplies, it is estimated that about 99.7% will have
either no 1,1-dichioroethylene present or levels ( 0.2 ugh. It is estimated
that 35 surface water systems have levels > 0.2 ug/l (0.3% of total surface
water systems); none are estimated to have 1,1-dichloroethylene above S ugh.
11
-------
It is important to note that some of the data used in computing the
national estimates are from samples held for a prolonged period of time prior
to analysis, with possible biodegradation of 1,1—dichioroethylene. Therefore,
these projections of national occurrence may underestimate actual contaminant
levels.
Using combined data from surface water and groundwater supplies, it was
estimated that 209,630,000 persons (97.8% of the population served by public
drinking water systems) are receiving water with no 1,1—dichioroethylene or
levels less than 0.2 ugh. Of the 4,789,000 persons (2.2%) receiving water
containing 1,1—dichloroethylene levels > 0.2 ug/l, an estimated 52,000
(< 0.1%) are exposed to levels > 5 ug/l. Flo individuals are estimated to be
exposed to levels > 10 ugh. Of the approximately 4.7 million people esti-
mated to be exposed to levels ranging from 0.2 to 5 ugh , 52% obtain water
from surface water supplies. All exposure to ].,1—dichloroethylene in drinking
water at levels above 5 ug/l is expected to be from groundwater sources. It
was also estimated that 40,000 formula-fed infants are exposed to 1,1-di-
chloroethylene levels greater than 0.2 ug/l and 440 formula-fed infants are
exposed to levels in excess of 5 ug/l. An adjustment of drinking water levels
for formula-fed infants was made to account for 1,1—dichloroethylene removal
from drinking water that is boiled prior to adding it to the formula.
The daily intake of 1,1-dichloroethylene from drinking water has also
been estimated. The majority of persons using public drinking water supplies
would be exposed to intake levels at or below 0.0057 ug/kg/day for adults and
0.048 ug/kg/day for formula-fed infants. Those individuals exposed to drink-
ing water containing higher levels of 1,1-dichloroethylene (> 5 ug/l) would
receive an intake of 1,1-dichloroethylene greater than 0.14 ug/kg/day for
adults and 1.2 ug/kg/day for infants.
No data were obtained on levels of 1,1-dichloroethylene in foods. There-
fore, the daily intake of 1,1-dichloroethylene could not be estimated.
Data on levels of 1,1—dichloroethylene in ambient air in the United
States were used to determine the respiratory intake of 1,1-dichioroethylene.
It was estimated that rural/remote, urban/suburban, source dominated, and
maximum levels of 1,1-dichloroethylene in ambient air would approximate 0.0,
0.020, 14, and 27 ug/rn 3 , respectively. Using these data, respiratory intake
for the adult male was estimated to vary between 0 and 8.9 ug/kg/day. Respi-
ratory intake for formula-fed infants would vary between 0 and 6.2 ug/kg/day.
lii
-------
The total daily multimedia intake of 1,1—dichioroethylene was estimated
using a range of low to high intake values for air and drinking water expo-
sure. Total intake was estimated to range from 0—9.2 ug/kg/day for the adult
male and from 0—86 u /kg/day for the formula-fed infant. At an air level of
0.020 ug/m 3 , drinking water will be the predominant source of 1,1—dichloro-
ethylene exposure in the adult male at drinking water levels of 0.2 ug/l and
above. In contrast, for the formula—fed infant exposed to the same ambient
air level (0.020 ug/m 3 ), drinking water will be •the predominant source of
1,1—dichioroethylene exposure at levels above 0.02 ugh.
Population—exposure estimates for 1,1—dichioroethylene in drinking water
were calculated to range from 0.024-1.3 x io8 ug/day x persons.
iv
-------
I NTRODUCTION
Monitoring studies of drinking water supplies in recent years have
uncovered evidence of contamination by volatile organic chemicals (VOC’s). On
March 4, 1982, the Environmental Protection Agency (EPA) published an Advance
Notice of Proposed Rulemaking that discussed regulatory and nonregulatory
approaches for limiting human exposure to 14 specific VOC’s identified to be
of particular concern as drinking water contaminants (47 FR 9350). 1,1-Di-
chloroethylene, the subject of this report, is one of those 14 chemicals.
This report presents information on the occurrence of 1,1—dichloro-
ethylene in drinking water, food, and air. This information is intended to
provide support for EPA’s assessment of 1,1—dichloroethylene in two principal
areas. As input to the health risk assessment, this report provides an esti-
mate of the number of individuals in the United States exposed to various
levels of 1,1-dichloroethylene in drinking water from public water supplies.
Information on dietary intake and respiratory intake from ambient air is
provided for perspective and is used to estimate the relative contributions of
the three sources, particularly of drinking water, to the total dose received
by individuals. While it is recognized that some individuals may be exposed
to 1,1—dichloroethylene from other sources, such as occupational settings or
the use of particular consumer products, this analysis is limited to drinking
water, food, and air because these are the major exposure routes common to all
individuals.
In addition to serving as input to the health assessment, this report
supports EPA efforts to estimate the economic impact of the regulatory and
treatment alternatives being considered. To aid in that effort, estimates are
provided of the number of public water supplies of various water source and
system size characteristics having 1,1-dichloroethylene present, and the
distribution of 1,1-dichloroethylene levels in those water supplies.
An extensive body of published and unpublished information is available
on 1 ,1—dichloroethylene relevant to its occurrence in the environment. This
report is based on information published since about 1965 and, where appro-
priate, utilizes EPA— and government—sponsored studies on the occurrence of
and human exposure to 1,1—dichioroethylene. This report is not intended to be
an exhaustive, comprehensive review of all existing data on 1,1—dichloro-
1
-------
ethylene. It does, however, present the most current and representative
information available for understanding the occurrence of 1,1-dichloroethylene
in food, air, and drinking water, and for assessing the importance of drinking
water as a route of human exposure.
A previous version of this report was prepared on September 2, 1982 and
was distributed at EPA for review. In this version, the authors have
attempted to reflect the many valuable comments and suggestions made by the
revi ewers.
2
-------
1. SOURCES OF 1,1-DICHLOROETHYLENE
1,1—Dichloroethylene, also known as vinylidene chloride (VDC), is a
colorless liquid with a low boiling point (31.6°C), very high vapor pressure
(495 mm Hg at 20°C), and low solubility in water (0.035 g/100 g at 25°C)
(Farmer et al. 1980). 1,1-Dichioroethylene is used in the manufacture of
copolymers that are incorporated in a variety of fibers and packing films. It
is also used to a lesser extent in the manufacture of chioroacetyl chloride, a
component of mace and tear gas.
1.1 NATURAL SOURCES
The limited monitoring data on 1,1-dichloroethylene do not indicate any
natural production of the chemical . No natural processes that generate
1,1-dichloroethylene have been identified.
1.2 ANTHROPOGENIC SOURCES
1.2.1 Production
Two manufacturers with a total of three plants (i.e., Dow Chemical in
Freeport, Texas and Plaquemine, Louisiana, and PPG Industries in Lake Charles,
Louisiana) produced most of the domestic 1,1-dichioroethylene in 1978. Pro—
duction data for 1,1—dichloroethylene are not readily available; the most
recent figure is 144,200 kkg for 1978 (Farmer et al. 1980). This figure
includes, the 1,1—dichioroethylene that was captively consumed in the produc-
tion of methyl chloroform. However, 1,1—dichioroethylene appears no longer to
be used in methyl chloroform production.
The most common process for production of 1,1—dichloroethylene involves
the dehydrochlorination of 1,1,2 —trichloroethane by dilute sodium hydroxide.
1,1-Dichioroethylene can also be produced by the chlorination of ethane or
ethylene when market conditions make this process more favorable (Farmer et
al. 1980).
1,1-Dichioroethylene can also be produced indirectly as a byproduct
during several industrial processes (Farmer et al. 1980):
1) Manufacture of 1,2-dichioroethane by the oxychlorination of
ethylene;
3
-------
2) Production of vinyl chloride monomer from 1,2-dichioroethane;
3) Oxychiorinatian and chlorination processes for the production of
tetrachioroethylene and trichloroethylene; and
4) Production of chioroprene.
Estimates for direct and indirect production of 1,1—dichioroethylene and
envi ronrnental release data are presented in Table 1.
Table 1. 1978 Production and Associated Releases of 1,1—Dichloroethylene
Releases to environment (kkg )
Process Pmount (kkg) Air Land Water
Direct Production
Production from 144,200 10
1,1, 2-tn chi oroethane
mdi rect Production
Production of 100 100
1, 2—di chi oroethane
Production of vinyl chloride Unknown
from 1, 2-di chioroethane
Contaminant of vinyl 20 < 2
chloride
Production of tetra— and 112 8
t ri chi oroethyl ene vi a
oxychiorination of
1, 2-di chi oroethane
Production of tetrachloro- 146 91
ethylene via chlorination
of 1,2-dichioroethane
Source: Farmer et al. 1980
1.2.2 Consumption
In 1978, an estimated 63,200 kkg of 1,1—dichioroethylene were captively
consumed in the production of methyl chloroform, with an additional 9 kkg
present in the end product as an impurity (Farmer et al. 1980). This use
accounted for nearly 45% of the total 1,1-dichioroethylene produced. Of the
4
-------
remaining 81,000 kkg, an estimated 78,200 kkg were used in the production of
polyvinylidene chloride (PVDC) for use in resins, coating latex, and in the
manufacture of rnodacrylics. Other uses included exports and the production of
chioroacetyl chloride. Quantitative estimates of the end uses, based on 1978
production figures, are found in Table 2.
Table 2. End Uses of 1,1—Dichloroethylene and
Associated Environmental Releases in 1978
Uses
Ouantity (kkg)
Releases to environment (kkg)
Air
Land
Water
Methyl chloroform
production
63,200
1,264
--
--
Use of methyl chloroform
(as
9
impurity)
9
Neg
Neg
Production of PVDC speciality
latexes for rug backing
4,690
66
-—
Meg
Production of modacrylics
10,005
Neg
——
--
Production of PVDC
emul si on/suspension latex
63,470
254
--
2
Other PVDC uses
< 5
< 5
Meg
Neg
Production of chioroacetyl
chloride
2,835
0
0
0
Exports
Neg
—-
—-
--
Source: Farmer et al . 1980
1.2.3 Releases
The principal estimated environmental releases associated with production
and use of 1,1—dichioroethylene in 1978 are shown in Tables 1 and 2. Major
releases are to the atmosphere. The manufacture of 1,1-dichloroethylene
results in a relatively small release to the atmosphere. Of 995 kkg estimated
to be released from reactor and distillation vents in this process in 1978,
only 1% (10 kkg) was emitted to the atmosphere; the remainder was incinerated
as gaseous waste (Farmer et al. 1980).
5
-------
Although 1,1-dichloroethylene is no longer used in the production of
methyl chloroform, in 1978 this process accounted for nearly 70% of the esti-
mated air releases of 1,1—dichioroethylene. Any 1,1—dichloroethylene retained
in methyl chloroform as an impurity would be released to the environment when
the product was used as a vapor degreaser, and in adhesives, aerosols, paints,
inks, textiles, and drain cleaners. As a result, negligible amounts of
1,1—dichloroethylene were estimated to be released to water (drain cleaners)
and land (landfill of aerosol cans) from the manufacture and use of methyl
chloroform (Farmer et al. 1980).
Most current releases of 1,1-dichloroethylene are expected to result from
other uses, among them production of PVDC and its use in the manufacture of
other products. The largest estimated amount of 1,1—dichloroethylene released
in this manner in 1978 came from the production of PVDC emulsion/suspension
latex (254 kkg to air, with 2 kkg to water). Lesser amounts were also
released during the manufacture of specialty latexes, and the use of PVDC
resin ift Saran, coating latexes, and solvent coatings. No releases of 1,1—di-
chloroethylene were reported from the production of chloroacetyl chloride, nor
is it present as an impurity. Releases from exports were negligible (Farmer
et al. 1980).
1.3 FATE AND TRANSPORT
Large amounts of 1,1—dichloroethylene enter the troposphere through air
emissions. In addition, volatilization appears to be the major transport
process for removal from the aquatic environment (Callahan et al. 1979). In
the troposphere, 1,1—dichloroethylene is photooxidized, resulting in produc-
tion of chloroacetyl chloride, phosgene, formic acid, hydrochloric acid,
carbon monoxide, and formaldehyde. The tropospheric lifetime is estimated to
be less than 1 day (Callahan et al. 1979).
Bloaccumulation of 1,1—dichloroethylene in the aquatic environment is not
thought to be significant (Callahan et al. 1979). No data on biotransforma-
tion and biodegradation are available.
6
-------
2. OCCURRENCE IN DRINKING WATER
1,1-Dichloroethylene is a synthetic organic chemical whose major routes
of entry to drinking water are a consequence of industrial activity. Quanti-
ties of 1,1-dichloroethylene in water may result from industrial discharge,
atmospheric fallout, or the release of quantities remaining as impurities in
products.
It is currently estimated that there are 59,660 public water supplies in
the United States (FRDS 1983). As shown in Table 3, 48,458 of the supplies
use groundwater and 11,202 use surface water as their primary source. Table 3
also shows the distribution of groundwater and surface water supplies among
the five major size categories and twelve subcategories based on the popula-
tion served.
There are two major types of data available that are potentially useful
for describing the occurrence of 1,1—dichioroethylene in the nation’s public
drinking water supplies. First, there are several Federal surveys in which a
number of public water supplies from throughout the U.S. were selected for
analysis of chemical contamination, including 1,1—dichloroethylene. Second,
data are available from state surveys and from state investigations of speci-
fic incidents of known or suspected contamination of a supply. For accom-
plishing the basic objectives of this study, namely to estimate the number of
public water supplies nationally within the various source and size categories
contaminated with 1,1—dichloroethylene, the distribution of 1,1—dichloro-
ethylene concentrations in those supplies, and the number of individuals
exposed to those concentrations, it was determined that the Federal survey
data provides the most suitable data base. The state data tend to be poorly
described with respect to the source and size categories of the supplies
examined and the sampling and analysis methods used for determining contami-
nant levels. The lack of source and system size information precludes using
the data for estimating levels in public water supplies of similar character-
istics. The absence of details on sampling and analysis methods precludes
evaluating those data for their qualitative and quantitative reliability.
Also, because much of the state data are from investigations in response to
incidents of known or suspected contamination (e.g., spills), they were judged
to be not representative of contaminant levels in the nation’s water supplies
in general. Although they are not used with the Federal data for the purpose
7
-------
of estimating contamination levels nationally, the available state data are
presented here to provide some additional perspective on 1,1-dichloroethylene
occurrence in drinking water.
This section of the report, which contains information on the occurrence
of 1,1—dichloroethylene in drinking water supplies in the United States, is
divided into four major parts. Section 2.1 presents an overview of the
Federal surveys providing data on 1,1—dichloroethylene. The methods of
analysis used in -each survey are discussed and an evaluation of the qualita-
tive and quantitative acceptability of the data is reported.
Sections 2.2 and 2.3 present the data available from the Federal and
state sources on the occurrence of 1,1—dichloroethylene in groundwater and
surface water supplies, respectively. Data are presented only on drinking
water samples taken from a consumer’s tap (i.e., distribution water samples)
or on treated water samples taken at the water supply (i.e., finished water
samples) because these are considered to be most representative of the water
consumed by the public. No data on raw (i.e., untreated) water are pre-
sented. It is recognized that for some groundwater supplies where no treat-
ment of the water occurs, samples identified as raw may be representative of
water consumed by the users of the supply. However, it was generally not
possible to differentiate between those groundwater supplies that do and those
that do not treat raw water from the available survey data.
Section 2.4 presents projections, based on the data from the Federal
surveys, of the occurrence of 1,1-dichioroethylene in the approximately 60,000
public water supplies in the nation.
8
-------
Table 3. Number of Systems and Population Served
by Primary Water Supply Source
(By Population Category)
Surface
Groundwater
System size No. of
(Population served) systems
Population
(thousands)
Average
population
served
No. of
systems
Population
(thousands)
Average
population
served
( 25-100 1,525
Very small s
101-500 2,412
501-1,000 1,377
Small 1,001—2,500 1,945
2,501-3,300 495
1 3,301-5,000 749
Medium
5,001-10,000 930
10,001—25,000 915 a
86
690
1,051
3,295
1,445
3,096
6,763
15 , 595 b
56
286
763
1,700
2,900
4,100
7,300
17,000
19,125
15,674
4,877
4,400
891
1,065
1,168
835 a
1,031
3,814
3,590
7,047
2,583
4,370
8,404
12 , 275 b
54
243
736
1,600
2,900
4,100
7,200
15,000
25,001-50,000 400 a
Large
50,001-75,000 155
13 , 945 b
9,483
35,000
61,000
290 a
64
10 , 977 b
3,911
38,000
61,000
75,001-100,000 82
7,131
87,000
14
1,184
85,000
Very large >100,000 217
78,366
360,000
55
14,286
260,000
TOTALSC: 11,202
140,948
48,458
73,475
aKuzmack 1983, as updated by David Schnare, Office of Drinking Water, U.S. Environ-
mental Protection Agency, in a personal communication with Frank Letkiewicz, JRB
Associates, May 25, 1983.
bEstimated by JRB Associates (see Appendix A).
cpopulations do not add to total due to rounding.
Source: FRDS 1983 (except as noted).
9
-------
2.1 OVERVIEW AND QUALITY ASSURANCE ASSESSMENT OF FEDERAL DRINKING WATER
SURVEYS
Two Federal drinking water surveys provide data on 1,1-dichioro-
ethylene: the National Screening Program for Organics in Drinking Water (NSP)
and the Groundwater Supply Survey (GWSS). The scope of the surveys, the
analytical methods used, and a description of the files accessed to obtain the
data from these surveys are presented below. The information available on the
analytical methods used in these surveys generally did not include precise
definitions of quantification limit, minimum quantifiable •concentration, or
similar terms used in reporting the results of the analyses. The terms used
in this report are those used in the individual surveys, recognizing that they
may not always correspond to strict technical definitions.
The National Screening Program for Organics in Drinking Water (NSP),
conducted by SRI International from June 1977 to March 1981, examined both raw
and finished drinking water samples from 166 water systems in 33 states for 51
organic chemical contaminants. Data are available for 1,1-dichloroethylene on
finished water samples from 12 groundwater and 103 surface water supplies.
The USP data were extracted from Boland (1981) and coded to create an SAS
computer file. Analyses were carried out using gas chromatography with an
electrolytic conductivity detector. Single values were reported for finished
water samples from most supplies in the FLSP. For those where multiple sample
data were reported, the values were averaged as described in Appendix A. The
quantification limit for 1,1-dichloroethylene in the NSP was reported to be
0.1 ug/l. The quantification limit was based on repeated analysis of formu-
lated standards (Boland 1981); however, the confidence intervals used to
establish the quantification limits were not reported.
The Groundwater Supply Survey (GWSS) was conducted from December 1980 to
December 1981 to develop additional data on the occurrence of volatile organic
chemicals in the nation’s groundwater supplies (Westrick et al. 1983). It was
hoped that this study would stimulate state efforts toward the detection and
control of groundwater contamination and the identi fication of potential
chemical “hot spots.” A total of 945 systems were sampled, of which 466 were
chosen at random. The remaining 479 systems were chosen ronrandomly based on
information from states encouraged to identify locations believed to have a
higher than normal probability of VOC contamination (e.g., locations near
landfills or industrial activity).
10
-------
The GWSS data were made available from EPA ’s Office of Drinking Water,
Technical Support Division in the form of an SAS computer file. The file
provided a single analytical result for each supply sampled. One sample of
finished water was collected from each supply at a point near the entrance to
the distribution system; mercuric chloride was added as a preservative to
prevent biodegradation of arornatics. Analyses were made for a total of 34
organic chemicals using purge and trap gas chromatography with an electrolytic
conductivity detector for halocarbons. The minimum quantitation limit for
1,1-dichloroethylene was 0.2 ug/l.
Each of the drinking water surveys was evaluated with respect to the
validity of the reported occurrence data for a number of organic chemicals,
including ],1—dichloroethylene. The evaluations were carried out by analyzing
information about the procedures used for collection and analysis of samples
as well as the quality control protocols used. The analyzed compounds dealt
with in each study were assigned one of three possible ratings: quantita-
tively acceptable, qualitatively acceptable (i.e., the substance measured was
1,1-dichloroethylene), and totally unacceptable. In the case of 1,1—dichioro-
ethylene, a qualitatively acceptable rating was given for data from the r s
because of suspected biodegradation of the samples, which were held unre-
frigerated for prolonged periods before analysis. 1,1-Dichioroethylene values
in excess of the quantitation limit reported for some samples in these studies
are qualitatively valid and can be taken as minimum values, representative of
samples which probably originally contained 1,1—dichioroethylene at higher
concentrations. In the case of the GWS.S, all data were rated both quantita-
tively and qualitatively acceptable.
2.2 GROUNDWATER
2.2.1 Federal Surveys
The National Screening Program for Organics in Drinking Water ( SP) and
and the Groundwater Supply Survey (GWSS) both contain data concerning the
levels of 1,1—dichloroethylene in groundwater supplies from across the
country.
Twelve groundwater supplies were tested for 1,1-dichloroethylene contami-
nation in the USP. Of these 12 systems, only one was found to be contaminated
11
-------
with 1,1-dichioroethylene at a level of 0.2 ugh (Table 4). The quantifica-
tion limit for 1,1—dichioroethylene was 0.1 ugh.
In the GWSS, 9 of the 456 randomly chosen water systems serving 25 or
more individuals were contaminated with 1,1-dichloroethylene, at concentra-
tions ranging from 0.22-6.3 ugh. The three systems with the highest values
were contaminated at 2.1, 2.2, and 6.3 ugh. Of the 9 positive systems, 5
were from systems serving populations in excess of 10,000 people (Table 5).
The average for all randomly chosen systems was 1.4 ugh with a standard
deviation of 2.0 ug/l; the median was 0.28 ugh. Of the 473 nonrandom loca-
tions sampled serving 25 or more individuals, 15 were contaminated with
1,1—dichloroethylene, at concentrations between 0.22-3 ug/l , the highest
values being 0.04, 1.2, and 3 ug/l. Of the 15 positive samples, 10 were from
systems serving populations in excess of 10,000 people (Table 6). The average
1,1—dichioroethylene level for the nonrandorn systems was 0.59 ug/l with a
standard deviation of 0.71 ug/l ; the median value was 0.35 ug/l . The minimum
quantitation limit for 1,1—dichloroethylene was 0.2 ug/l.
12
-------
Table 4. Frequency of Groundwater Contamination by 1,1-Dichloroethylene
National Screening Program for Organics in Drinking Water (NSP)
No. of No. of No. of
systems positive Positive negative
sampled systemsa systems systemsa
o o 0 0
o o 0 0
o 0 0 0
o 0 0 0
1 1 100 0
1 0 0 1
0 0 0 0
0 0 0 0
0 0 0 0
1 0 0 1
— -
12 1 8 11
System size
(population_served)
25-100
101- 500
501-1,000
1,001-2,500
2,501-3,300
3, 301-5, 000
5,001-10,000
10,001-50,000
50,001-7 5,000
7 5,001-100,000
> 100,000
Totals
Numbè of systems with measured
concentrations (ug/l) of:
< 0.1 0.1-5 > 5____
0 0 0
0 0 0
0 0 0
0 0 0
0 1 0
0 0 0
0 0 0
0 0 0
0 0 0
0 0 0
0 1 0
aQuant fication limit = 0.1 ugh.
-------
Table 5. Frequency of Groundwater Contamination by 1,1-Dichloroethylene
Groundwater Supply Survey (GWSS) -- Random Sites
System
(population
size
served)
No. of
systems
sampled
No. of
positive
systemsa
%
Positive
systems
No. of
negative
systemsa
Nuiiber ô 7 ii with measured
concentrations (ug/l) of:
<
0.2
0.2-5
>5—10
>10
25-100 70 b 1 1 69 0 1 0 0
101-500 88 1 1 87 0 1 0 0
501-1,000 28 1 4 27 0 0 1 0
1,001-2,500 31 0 0 31 0 0 0 0
2,501-3,300 17 0 0 17 0 0 0 0
3,301-5,000 19 1 5 18 0 1 0 0
5,001-10,000 17 0 0 17 0 0 0 0
10,001-50,000 153 4 3 149 0 4 0 0
50,001-75,000 15 0 0 15 0 0 0 0
75,001-100,000 5 0 0 5 0 0 0 0
> 100,000 _ _i I
Totals 456 9 2 447 0 8 1 0
a 1injmum quantitation limit = 0.2 ugh.
bjiie GWSS also reported 10 systems serving fewer than 25 people with no quantifiable 1,1-dichloro?thylene.
-------
Table 6. Frequency of Groundwater Contamination by 1,1-Dichioroethylene
Groundwater Supply Survey (GWSS) -— Nonrandom Sites
No. of No. of No. of Number ol systems with measureT
System size systems positive Positive negative concentrations (ug/l) of :
(population served) sampled systemsa systems systemsa < 0.2 0.2-5 > 5
25-100 29 b 0 0 29 0 0 0
101-500 43 0 0 43 0 0 0
501-1,000 34 2 6 32 0 2 0
1,001-2,500 71 0 0 71 0 0 0
2,501-3,300 21 0 0 21 0 0 0
3,301-5,000 42 2 5 40 0 2 0
5,001-10,000 75 1 1 74 0 1 0
10,001-50,000 118 8 7 110 0 8 0
50,001-75,000 17 2 12 15 0 2 0
75,001-100,000 3 0 0 3 0 0 0
>100,000 - 0
Totals 473 15 3 458 0 15 0
aMinimurn quantitation limit = 0.2 ug/l.
bihe GWSS also reported 6 systems serving fewer than 25 people with no quantifiable 1,1-dichloroethylene.
-------
2.2.2 State Data
Three states (California, Massachusetts, and New Jersey) provided the
U.S. Environmental Protection Agency with information concerning 1,1-dichloro—
ethylene contamination in groundwater supplies. Analytical results for
samples from three locations in California ranged from undetectable to 50
ug/l. 1,1—Dichloroethylene levels ranged from undetectable to 261 ugh in 22
samples from six Massachusetts cities. New Jersey provided data from 19
samples from Fair Lawn and Mahwah; 12 of the samples contained undetectable
1,1-dichloroethylene while the other seven samples ranged from 2.7-3.5 ug/l at
Mahwah and 0.9—27 ug/l at Fair Lawn (Table 7).
16
-------
Table 7. State Data on 1,1-Dichioroethylene in Groundwater
Total
Location
Water
type
Mean
(ugh
)
Range
(ugh )
No. of
samples
Reference
CALIFORNIA
Morada
N/S
1.4
1
CA-Ui
Aerojet General
Corp.
N/S
.
3-50
4
CA-12
Unspecified
N/S
ID
3
CA-Ui
MASSACHUSETTS
Acton
F
8.5
2.6-19.5
4
MA-09
Belchertown
F
12.7
2.0-41.5
12
(8 ND)
MA-Ui
Dartmouth
F
68.4
2
(1 ND)
MA-09
Lunenburg
D
261
1
MA-18
Rowley
F
ND
1
MA-09
Wilmington
F
ND
2
MA-Ol
NEW JERSEY
Fair Lawn
MIS,
F
13.4
0.9-27
13
(9 MD)
NJ-Oi
Mahwah
N/S.
0
3.1
2.7—3.5
6
(3 MD)
NJ-Ui
N/S = not specified, F = finished, D = distribution, ND = not detected
17
-------
2.3 SURFACE WATER
2.3.1 Federal Surveys
Only one federal survey, the National Screening Program (lISP), contains
data concerning 1,1-.dichloroethylene levels in surface water supplies. During
this survey, 106 drinking water systems were analyzed for 1,1-dichloroethylene
between June 1977 and March 1981. Of these, two systems contained detectable
levels of 1,1-dichloroethylene, with concentrations of 0.2 and 0.51 ug/l
(Table 8).
18
-------
Table 8. Frequency of Surface Water Contamination by 1,1-Dichloroethylerie
National Screening Program for Organics in Drinking Water (NSP)
lb. of No. of No. of Number of syst i with measur
System size systems positive Positive negative concentrations (ugh) of :
(population served) sanipled systemsa systems systemsa < 0.1 0.1—5 > 5
Unspecified 3 0 0 3 0 0 0
25-100 0 0 0 0 0 0 0
101-500 0 - 0 0 0 0 0 0
501-1,000 0 0 0 0 0 0 0
1,001-2,500 0 0 0 0 0 0 0
2,501-3,300 0 0 0 0 0 0 0
3,301-5,000 1 0 0 1 0 0 0
5,001-10,000 0 0 0 0 0 0 0
10,001-50,000 4 0 0 4 0 0 0
50,001-75,000 4 0 0 4 0 0 0
75,001-100,000 11 0 0 11 0 0 0
100,000 - - - - -
Totals 106 2 2 101 0 2 0
aQuantification limit = 0.1 ugh.
-------
2.3.2 State Data
Data from two surface water samples were reported from Niagara Falls, New
York. Of the two finished water samples, one contained no detectable 1,1 —di—
chloroethylene. The other sample was contaminated with 1,1—dichloroethylene
at 0.22 ugh.
20
-------
2.4 PROJECTED NATIONAL OCCURRENCE OF 1,1—DICHLOROETHYLENE IN PUBLIC WATER
SUPPLIES
There are approximately 60,000 public water supplies in the United
States. As shown in Table 3, these systems fall into two major categories
with respect to water source (surface water and groundwater) and into five
size categories and twelve subcategories according to the number of indivi-
duals served. Sections 2.2 and 2.3 presented the available data from Federal
surveys and state reports on the occurrence of 1,1-dichloroethylene in drink-
ing water supplies. This section of the report presents estimates of both the
number of drinking water supplies nationally within each of the source/size
categories expected to have 1,1-dichioroethylene present, and of the concen-
tration of 1,1-dichloroethylene expected to be present in those supplies.
The Appendices to this report describe the methodology used and assump-
tions made to develop the national estimates. The key features are summarized
here. The estimates are based on the data from the Federal surveys only. The
state data were not included for several reasons. Generally, these data are
from a few states and were not considered to be geographically representa-
tive. There was also a general lack of data on the population served by
systems measured, the type of water sampled, and the methodologies used to
sample, identify, and measure 1,1—dichioroethylene.
The Federal survey data from the NSP and GWSS were pooled together for
developing the national projections. It was assumed in combining these
surveys that the resulting data base would be representative of the nation’s
water supplies. In the case of the GWSS data, both the random and nonrandom
samples were included in the projections because, as described in Appendix A
(A.2.4), a statistical test of the GWSS data showed no statistical1 signifi-
cant difference in the frequency of occurrence of positive values or the mean
of the positive values of vinyl chloride between the random and nonrandom
samples.
Ideally, adequate survey data would be available to develop the national
projections separately for each of the twelve system size categories within
the groundwater and surface water groups; however, the available data were too
limited for this. It was, therefore, necessary to consolidate some of the
size categories to have sufficient data for developing the projections. In
consolidating data from various size categories, consideration was given to
21
-------
the potential for there being statistically significant differences in the
frequency of occurrence of 1,1-dichloroethylene as a function of system
size. The consolidation of size categories therefore involved a balancing of
the need to group size categories together to have an adequate data base for
developing the national projections against the need to treat size categories
separately in order to preserve the influence of system size as a determinant
of contamination potential. The consolidation of size categories also took
into account EPA’s classification of systems into the five major groups as
very small (25-500), small (5O1- ,3O0), medium (3,301—10,000), large (10,001-
100,000), and very large (> 100,000) (Kuzniack 1983).
Once the data were consolidated, statistical models for extrapolating to
the national level were tested and an appropriate model selected. In the case
of 1 ,1—dichloroethylene, the multinominal method was used. The frequency of
contamination of groundwater and surface water systems at various concentra-
tions was determined for each consolidated size category. For completing the
national estimates, it was assumed that the frequency of contamination
observed for each consolidated category was directly applicable to each of the
system sizes comprising it.
It is important to note that some of the data used in computing the
national estimates are from samples held for a prolonged period of time prior
to analysis, with possible biodegradation of 1,1-dichloroethylene. Therefore,
these projections of national occurrence may underestimate actual contaminant
level S.
2.4.1 Projected National Occurrence of 1,1-Dichloroethylene in Groundwater
Supplies
The combined 1,1-dichloroethylene groundwater data from the NSP and GWSS
surveys are presented in Table 9. It should be noted that the total number of
systems sampled indicated in Table 9 is less than the sum of the systems
sampled in each individual survey since some supplies were examined in two or
more surveys. For those supplies, an average value was computed based on the
results of the individual surveys.
As indicated in Table 9, data are available for a total of 938 supplies
from the combined surveys. Of these, 25 supplies were reported to have
1,1 —dichioroethylene present, at concentrations ranging from 0.2 ug/l to 6.3
ug/l.
22
-------
Based on the overall distribution of positive values and maximum possible
values for those supplies in which 1,1—dichioroethylene was not found, 0.2
ug/l was selected as the common minimum quantifiable concentration for the
combined survey data (see Appendix A.2.5). That is, quantitative projections
are made of supplies at several concentration ranges > 0.2 ug/l , while only a
total number for supplies expected to have either no 1,1-dichioroethylene or
levels below 0.2 ug/l can be determined.
When the twelve size categories were consolidated into the five major EPA
groupings, there was an apparent trend in the frequency of values > 0.2 ugh
as a function of size:
Very small 0.9% (2/230)
Srnafl 2.0% (4/203)
Medium 2.6% (4/154)
Large 4.5% (14/312)
Very large 2.6% ( 1/39 )
Overall 2.7% (25/938)
A test for statistical significance revealed that at the = 0.05 level,
the difference among the very small, small, and medium categories was not
significant; nor was the difference between the large and very large size
categories. However, the combined very small, small, and medium categories
and the combined large and very large categories were found to be different.
Therefore, two consolidated categories were selected for developing the
national estimates:
Very small/small/medium (25—10,000)
Large/very large (> 10,000)
As noted previously, the frequency of occurrence of 1,1-dichioroethylene at
various concentrations was determined for the two consolidated groups and then
applied to the number of supplies nationally within each of the size cate-
gories comprising each group.
Table 10 presents the estimates of groundwater supplies occurring
nationally within various 1,1—dichloroethylene concentration ranges: Table 11
23
-------
provides estimates of the cumulative number of groundwater supplies exceeding
a given value of 1,1—dichioroethylene and includes bounds based on the 95%
confidence intervals for the totals across all size categories. As indicated
by Table 11, 858 groundwater supplies (range of 366—1,349), approximately 1.8%
of the total groundwater supplies in the United States, are expected to have
1,1—dichloroethylene at levels of > 0.2 ug/l ; the remaining 47,600 supplies
have either no 1,1—dichloroethylene or levels < 0.2 ug/l.
It is estimated that 81 supplies (range of 0-237) are expected to have
1,1—dichioroethylene levels > 5 ug/l ; while no supplies are expected to have
levels > 10 ug/l. As shown by the data in Tables 10 and 11, most of the
supplies with 1,1-dichioroethylene levels > 0.2 are expected to be in the
smaller size categories. Although, as noted previously, the frequency of
1,1—dichioroethylene occurrence appears to increase with increasing system
size, the number of systems affected nationally is greater for the small sizes
because there are many more small systems in existence.
24
-------
Table 9. Reported Occurrence of 1,1-Dichloroethylene in Groundwater Systems —- Combined Federal Data
(NPS and GWSS Surveys)
r.)
(11
System size
(population served)
No. of
systems
sampled
Flo. of
positive
systemsa
%
Positive
systems
No. of
negativg
systems
Number of systems
concentrations
with measured
(ugh) of:
<
0.2
0.2-5
> 5-10 >10
25-100
99
1
1
98
0
1
0 0
101-500
131
1
1
130
0
1
0 0
501-1,000
62
3
5
59
0
2
1 0
1,001-2,500
102
0
0
102
0
0
0 0
2,501-3,300
39
1
3
38
0
1
0 0
3,301-5,000
62
3
5
59
0
3
0 0
5,001-10,000
92
1
1
91
0
1
0 0
10,001-25,000
25,001-50,000
182
89
8
4
4
4
174
85
0
0
8
4
0 0
0 0
50,001-75,000
32
2
6
30
0
2
0 0
75,001-100,000
9
0
0
9
0
0.
0 0
> 100,000
39
1
3
38
0
1
0 0
Totals
938
25
3
913
0
24
1 0
apositive systems are those with measured levels of 1,1—dichioroethylene.
bi egative systems are those in which 1,1-dichioroethylene was not observed. The maximum
based on minimum quantifiable limits of the analyses done, ranged from 0.1 to 0.2 ugh.
possible levels,
-------
Table 10. Estimated Number of Groundwater Systems in Each Size Category
with 1,1-Dichloroethylene in the Indicated Concentration Ranges (ug/l)
System size No. of ________________________________________
(population systems
served) in_U.S. < 02 a ____________________________
25-100 19,125 18,799
101-500 •15,674 15,407
501-1,000 4,877 4,794
1,001-2,500 4,400 4,325
2,501-3,300 891 876
3,301-5,000 1,065 1,047
5,001-10,000 1,168 1,148
10,001-25,000 835 799
25,001-50,000 290 278
50,001-75,000 64 61
75,001-100,000 14 13
>100,000 55 53
Total 48,458 47,600
Estimated number ot systems witri
concentrations (ug/l) of:
0.2-5 >5-10 >10
293 33 0
240 27 0
75 8 0
67 7 0
14 2 0
16 2 0
18 2 0
36 0 0
12 0 0
3 0 0
1 0 0
___ 2 0 0
777 81 0
acalculated as the difference between the systems expected to
and the total number of systems in that size category. This
those having no 1,1-dichloroethylene contamination and those
ugh.
have > 0.2 ugh
group includes
with levels < 0.2
26
-------
Table 11. Estimated Cumulative Number of Groundwater Systems
in Each Size Category with 1,1—Dichioroethylene
Exceeding the Indicated Concentration (ug/l)
System size
(population
No. of
systems
Est
imated
with
cumulative number of
concentrations (ugh)
sys
of:
tems
served)
in U.S.
>
0.2
> 5
>
10
25—100 19,125 326 33 0
101-500 15,674 267 27 0
501-1,000 4,877 83 8 0
1,001-2,500 4,400 75 7 0
2,501-3,300 891 15 2 0
3,301-5,000 1,065 18 2 0
5,001-10,000 1,168 20 2 0
10,001-25,000 835 36 0 0
25,001-50,000 290 12 0 0
50,001-75,000 64 3 0 0
75,001-100,000 14 1 0 0
>100,000 55 2 0 0
Total 48,458 858 81 0
Lower bOUflda 366 0 0
Upper bOUflda 1,349 237 0
aFrom 95% confidence intervals.
27
-------
2.4.2 Projected National Occurrence of 1,1—Dichioroethylene in Surface Water
Systems
Table 12 presents the 1,1—dichioroethylene surface water data from the
NSP survey. As indicated in Table 12, data are available for a total of 103
surface water supplies. Of these, 2 supplies were reported to have 1,1-di-
chloroethylene present at concentrations of 0.2 ugh and 0.5 ugh.
Table 12 indicates that all but one of the 103 surface water supplies
sampled in the NSP fall in the large and very large size cate.gories (i.e.,
serving > 10,000 people). Consequently, it was not possible to evaluate the
frequency of occurrence of 1,1—dichioroethylene as a function of system size
for the very small, small, and medium size categories. In the large size
category (serving 10,001-100,000), none of the 19 supplies were found to have
1,1-dichloroethylene present, while 2 of 83 in the very large category (serv-
ing > 100,000) had 1,1—dichioroethylene present. The difference in the fre-
quency of occurrence between the large and very large groups was not statisti-
cally significant at the 0.05 level. For the purpose of the national
estimates, the groundwater supplies were consolidated intO two groups:
Very small/small/medium (25-10,000)
Large/very large (> 10,000)
Again, the frequency of occurrence of 1,1—dichioroethylene at various
concentrations was determined for the two consolidated groups and then applied
to the number of supplies nationally within each of the size categories com-
prising each group.
Table 13 presents the national estimates of surface water supplies within
various 1,1—dichioroethylene concentration ranges; Table 14 provides estimates
of the cumulative number of surface water supplies exceeding a given value of
1,1—dichloroethylene and includes bounds based on the 95% confidence intervals
for the totals across all size categories. As indicated by Table 14, 35
surface water supplies (range of 0—81), approximately 0.3% of the total sur-
face water systems in the United States, are expected to have 1,1-dichloro-
ethylene at levels > 0.2 ugh ; the remaining 11,167 supplies have either no
1,1-.dichloroethylene or levels < 0.2 ugh. It is estimated that no surface
water supplies will have levels > 5 ug/l. Note that all of the supplies with
28
-------
levels in the 0.2-5 ugh range are in the large and very large category. The
estimate of no occurrence in the smaller categories is based on only one
sample and is probably not a reliable estimate.
29
-------
Table 12. Reported Occurr nce of 1,1-Dichioroethylene in Surface Water Systems
(N SP)
No. of No. of No. of Number of systems with measured
System size systems positive Positive negativg concentrations (ugh) of :
(population served) sampled systemsd systems systems < 0.2 0.2-5 > 5
25-100 0 0 0 0 0 0 0
101-500 0 0 0 0 0 0 0
501-1,000 0 0 0 0 0 0 0
1,001-2,500 0 0 0 0 0 0 0
2,501-3,300 0 0 0 0 0 0 0
3,301-5,000 1 0 0 1 0 0 0
5,001-10,000 0 0 0 0 0 0 0
10,001-25,000 2 0 0 2 0 0 0
25,001-50,000 2 0 0 2 0 0 0
50,001-75,000 4 0 0 4 0 0 0
75,001-100,000 11 0 0 11 0 0 0
>100,000 — -
Totals 103 2 0 101 0 2 0
apositive systems are those with measured levels of 1,1—dichloroethylene.
bNegative systems are those in which 1,1-dichloroethylene was not observed. The maximum possible level,
based on minimum quantifiable limits of the analyses done, was 0.1 ug/l.
-------
Table 13. Estimated Number of
with 1,1—Dichioroethylene in
Surface Water Systems in Each Size Category
the Indicated Concentration Ranges (ug/l)
Estimated number of systems
with concentrations (ugh) of:
< 02 a 0.2-5 >5
1,525 0 0
2,412 0 0
1,377 0 0
1,945 0 0
495 0 0
749 0 0
930 0 0
897 18 0
392 8 0
152 3 0
80 2 0
213 4 0
11,167 35 0
àcalculated as the difference between the systems expected to
and the total number of systems in that size category. This
those having no 1,1—dichloroethylene contamination and those
ugh.
have > 0.2 ug/l
group includes
with levels < 0.2
System size
(population
served)
2 5-100
101- 500
501-1,000
1, 001-2, 500
2, 501-3, 300
3,301-5,000
5,001-10,000
10,001-25,000
25,001-50,000
50, 001-75,000
75, 001-100, 000
>100,000
Total
No. of
systems
in U.S .
1,525
2,412
1,377
1,945
495
749
930
915
400
155
82
217
11,202
31
-------
Table 14. Estimated Cumulative Number of Surface Water Systems
in Each Size Category with 1,1-Dichioroethylene
Exceeding the Indicated Concentration (ugh)
System size
(population
No. of
systems
Estimated
with
cumulative number of
concentrations (ugh)
systems
of:
served)
in U.S.
> 0.2
> 5
25-100 1,525 0 0
101—500 2,412 0 0
501-1,000 1,377 0 0
1,001—2,500 1,945 0 0
2,501-3,300 495 0 0
3,301-5,000 749 0 0
5,001-10,000 930 0 0
10,001—25,000 915 18 0
25,001-50,000 400 8 0
50,001-75,000 155 3 0
75,001-100,000 82 2 0
>100,000 217 4 0
Total 11,202 35 0
Lower bOUflda 0 0
Upper bounda 81 0
aFrom 95% confidence intervals.
-------
3. OCCURRENCE IN THE FOOD SUPPLY
No information on the occurrence of 1,1—dichioroethylene in food consumed
in the United States was uncovered in our literature search.
33
-------
4. OCCURRENCE IN AMBIENT AIR
Data on levels of 1,1—dichioroethylene in ambient air, taken from a
report by Brodzinsky and Singh (1982), are presented in Table 15. Brodzinsky
and Singh have compiled available published and unpublished atmospheric data
on 1,1—dichloroethylene into a master file, and have subjected these data to
statistical analysis. An assessment of the quality, reliability, and repre—
sentativeness of the data was also performed. Each data point was then given
a “quality code” based on either a quantitative assessment or on the authors’
own experience and judgement. Quality codes were: 1 = Excellent; 2 = Good;
3 = Acceptable; 4 = Questionable; and Z = Unable to assign quality. These
quality codes are further explained in Table 15.
Using data for quality codes 1—3 in Table 15, Brodzinsky and Singh (1982)
calculated median concentrations of 20 ng/m 3 for urban/suburban areas and
14,000 ng/m 3 for source-dominated areas. No data for rural/remote areas were
reported under quality codes 1-3; two data points for quality code 4 showed
0.0 ng/m 3 of 1,1-dichloroethylene in these areas. As might be expected, the
majority of mean and median values reported were below 20 ng/m 3 . The maximum
value reported for quality codes 1-3 was 27,000 ng/m 3 .
34
-------
Table 15. Gcurrence of 1,1—Dichioroethylene in P4nblent Airm
Number
Number
Average
of
below
Standard
of
data
Location points
detection
limit
Mean
(ng/m 3 )
deviation
(ngfm 3 )
quality MinimumC MedianC
codosb (ng/m 3 ) (ng/m 3 )
MaximurnC
(hg/rn 3 ) Date/timed References
California
87 0 26
12 3.0
5.6 26 56 800102/1130 — 800712/1452
Singh et al. 1980
Colorado 85 28 26 84 3.0
5.6 560 800616/2012 — 800626/2359 Slngh ot ai. 1980
Michigan
2 0 24,000
24 2.0 24,000
0.0 24,000 770322/0949 — 770323/0941
Going and Splgarelll 1977
North Carolina
6 0 11,000
0.0 4.0 11,000 11,000 11,000
800600 — 800600
Wallace 1981
Louisiana
0.0
1
0
4,400
280
2.0
4,400
4,400
4,400
770104/0747 — 770203/1020
Going and
Spigareili 1977
11
7
120
0.0
4.0
120
120
120
Pellizzarl
at al. 1979
5
0
27,000
290
2.0
27,000
27,000
21,000
770125/0800 — 770126/0824
Going and
Splgarelli 1977
10
9
20
0.0
4.0
20
20
20
770301 — 770301
Peilizzari
1918a
25
25
0.0
0.0
4.0
0.0
0.0
0.0
770303/1150 — 770520/0900
Peiilzzari
et ml. 1979
j
Missouri
78
I
14
6.8
3.0
0.0
14
26
800530/0911 — 800608/1614
Singh et al. 1980
U,
New Jersey
46
46
0.0
0.0
4.0
0.0
0.0
0.0
790115 — 791229
Bozzeili ot ai. 1980
48
50
29
54
23
I
42
48
50
29
54
23
I
42
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
4.0
4.0
4.0
4.0
4.0
4.0
4.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
790127 — 791229
790121 — 791229
760324/1247 — 760701/1625
790115 — 791229
790403 — 791024
770922/1030 — 770922/1130
790226 — 791229
Bozzeiii at ml. 1980
Bozzelli et ai. 1980
Pelilzzari at al. 1979
Bozzeiii et ml. 1980
BozzellI at al. 1980
Pellizzarl and Bunch 1979
Bozzelil at al. 1980
Oklahoma
2
1
2
2
1
2
0.0
0.0
0.0
0.0
0.0
0.0
4.0
4.0
4.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
770711/0600 — 170921/0900
770921/1400 — 770921/1700
770710/0355 — 770921/0900
Pelllzzari 1978b
Pelilzzari 1978b
Peiiizzari 1978b
Pennsylvania
1
1
1
1
I
I
0.0
0.0
0.0
0.0
0.0
0.0
4.0
4.0
4.0
0.0
0.0
0..O
0.0
0.0
0.0
0.0
0.0
0.0
770821/1220 — 770821/1320
170822/1100 — 770822/1200
770819/1315 — 770819/1415
Peliizzari and Bunch 1979
Pellizzari and Bunch 1979
Pelllzzarl and Bunch 1979
-------
Table 15. Occurrence of 1,1—Dichioroethylene In Ambient Aira
(continued)
Number
Number
Average
of
below
Standard
of
Location
data
points
detection
limit
Moan
(ng/m 3 )
deviation
(ng/m 3 )
qualit
codes
MinimumC MedianC
(ng/m 3 ) (ng/m 3 )
MaximumC
(ng/m 3 )
Date/time 1
References
Texas
75
15
1
Il
3
0
15
0
0
2
64
0.0
520
48,000
140
60
0.0
0.0
0.0
250
3.0
4.0
4.0
4.0
4.0
10 56
0.0 0.0
520 0.0
48,000 48,000
0.0 0.0
220
0.0
520
48,000
440
770628/0745 — 800524/1906
760809/1348 — 760809/1555
800304 — 800304
770522/0945 — 771020/1510
Singh et al. 1980
Peliizzari et ai.
Peiiizzari et ai.
Wai lace 1981
Pellizzari et al.
1979
1979
1979
Virginia
ii
15
200
210
4.0
0.0 0.0
400
170929/1040 — 771028/0850
Pelllzzari 1978b
West Virginia
4
6
6
6
2
4
4
6
6
6
2
4
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
4.0
4.0
4.0
4.0
4.0
4.0
0.0 0.0
0.0 0.0
0.0 0.0
0.0 0.0
0.0 0.0
0.0 0.0
0.0
0.0
0.0
0.0
0.0
0.0
770921/1202 — 771025/1147
770927/1115 — 771118/1310
770927/1349 — 77i118/1241
710927/1400 — 771118/1250
771117/1235 — 711118/1210
770927/1020 — 771120/0930
Peilizzari 1978b
Pellizzari i978b
Peilizzari 1978b
Peiiizzarl 1978b
Peiiizzari 1978b
Pelllzzari 1978b
8 Valuos originaliy reported as ppt were converted to ng/m 3 Cl ppt 4.0 ng/m 3 ).
b 9 0 iit codes wore assigned to the data as foi lows;
• 1: c i = 1.25 (excellent)
• 2; a 1.50 (good)
• 3: = 2.00 (acceptabie)
• 4: > 2.00 (questionabie)
• Z; unabie to assign quailty
whore a true measurement 9 lies between 9/a and aQ. Where two lines of data are listed, line 1 contains data for quality codes 1—3; line 2 for 4.
cIhese values do not necessarily represent the absolute minimum, median, and maximum values presented in the reports cited. Some of the data were
preprocessed to facilitate the hand entry of the data into a computerized data base. Within a given reference, the data were averaged into daily
averages for a given location, whenever possible. For locations where data was averaged, the minimum and maximum values reported above are generally
higher and io ier, respectively, than those reported in the references cited (Brodzinsky and Singh 1982; personal communication between Richard
i3rodzlnsky, IBM, San Jose, California and Pauline Johnston, JRD Associates, April 7, 1983).
dFt is YYMHDO/TIME.
Source: Brodzinsky and Slngh 1982
-------
5. HUMAN EXPOSURE FROM DRINKING WATER, FOOD, AND AIR
Sections 2, 3, and 4 presented information on the occurrence of 1,1—di—
chioroethylene in drinking water, food, and air, respectively, focusing on
estimates of the levels of 1,1-dichloroethylene found in those three media.
This section uses the data presented there to estimate the extent of human
exposure, both in terms of individual intake and the size of the affected
population.
As mentioned in the Introduction, some individuals may be exposed to
1,1—dichloroethylene from sources other than the three considered here,
notably in occupational settings and from the use of consumer products con-
taining 1,1-dichloroethylene. This analysis, however, is limited to drinking
water, food, and air, since these media are considered to be general sources
common to all individuals. Even in limiting the analysis to these three
sources, it must be recognized that individual exposure will vary widely based
on many personal choices and several factors over which there is little
control. Where one lives, works, and travels, what one eats, and physiologic
characteristics related to age, sex, and health status can all profoundly
affect daily exposure and intake. Individuals living in the same neighborhood
or even in the same household can experience vastly different exposure
patterns.
Unfortunately, data and methods to estimate exposure of identifiable
population subgroups from all sources simultaneously have not yet been
developed.. To the extent possible, estimates are provided of the number of
individuals exposed to each medium at various 1,1—dichioroethylene concentra-
tions. For estimating intake, two specific subpopulations are considered:
adult male and newborn formula—fed infant. The adult male is typically used
for computing human intake when risk is extrapolated from animal models. Data
for the newborn formula-fed infant are included because this subpopulation, as
a result of differences in fluid consumption and respiratory rates, exper-
iences a significantly different intake (on a body weight basis)* than the
*For computing intake in units of mass/body weight/day, standard body weights
of 70 kg for an adult male and 3.5 kg for a newborn are used. These are
taken from the Report of the Task Group on Reference Man (ICRP 1975).
37
-------
adult male, even though the amounts of 1,1—dichioroethylene present in the
exposure media are the same for both groups. Generally, the intake from
drinking water for the newborn formula—fed infant is substantially higher than
that of the adult male, while the intake from air for the newborn formula—fed
infant is lower.
No data were obtained on regional variations in the concentration of
1,1—dichloroethylene in drinking water. The highest concentrations are
expected to occur near sites of production and use of 1,1-dichloroethylene.
5.1 DRINKING WATER INTAKE
5.1.1 Population Exposed
Estimates of the size of U.S. populations exposed to various 1,1-di-
chioroethylene levels in drinking water from public drinking water systems are
presented in Tables 16—20. Table 14 shows the population exposed to 1,1-di-
chloroethylene from groundwater supplies at various concentration ranges;
Table 17 shows the cumulative number of people exposed to 1,1-dichloroethylene
exceeding various levels in groundwater supplies. Tables 18 and 19 provide
similar population exposed information for surface water supplies. The cumu-
lative population exposed to concentrations exceeding various levels from both
sources are shown in Table 20. The values in these tables were obtained using
Federal Reporting Data Systems data on populations served by primary water
supply systems and the estimated number of these water systems that contain a
given level of 1,1-dichloroethylene (see Section 2.4). An estimated 4,789,000
individuals (2.2% of the population of 214,419,000 using public water
supplies) are exposed to levels of 1,1-dichloroethylene in drinking water at
or above 0.2 ugh, while 52,000 individuals (< 0.1%) are exposed to levels
above 5 ugh . It is estimated that no individuals are exposed to levels
greater than 10 ugh. Of the approximately 4.7 million people exposed to
levels ranging from 0.2 to 5 ug/l , 2.4 million (52%) obtain water from surface
water supplies. All exposure to 1,1—dichloroethylene in drinking water at
levels above 5 ug/l is expected to be from groundwater sources.
As noted earlier, newborn formula-fed infants are a special subpopulation
whose fluid consumption rate generally results in a higher intake on a body
weight basis than adult males exposed to drinking water at the same contami-
38
-------
nant level. To arrive at an estimate of the size of the infant population
potentially exposed to 1,1—dichioroethylene from drinking water and the intake
received by them, several assumptions were necessary.
Newborn infants are defined here as children up to 1 month old, corre-
sponding to a body weight of 3.5 kg. The U.S. Bureau of the Census reported
that 3,533,044 children less than 1 year of age lived in the United States in
1980 comprising 1.6% of the total 1980 U.S. population of 226,504,852
persons.* The percentaye of newborn infants up to one month of age at a given
point in time could be estimated as 1/12 of the number of infants under one
year of age, or (1/12) x (1.6%) of the U.S. population. In the course of a
year, however, all infants under one year of age, or 1.6% of the U.S. popula-
tion, would have been under one month of age at some time. For the purposes
of this report, the percentage of children in the U.S. under one year of age
was used to represent all children who would have been under one month of age
during a given year.
An estimated fluid intake level of 850 ml/day for the newborn infant was
derived by Coniglio** based on information presented in the Report of the Task
Group on Reference Man (ICRP 1975). The drinking water intake level of 2
liters per day used for estimating exposure for the adult male includes all
fluid intake (tap water, coffee, juices, beer, etc). In the case of newborn
infants, many obtain most if not all of their fluid from breast milk, which is
deemed inappropriate to include in the estimate of drinking water intake.
Martinez and Dodd (1983) have presented the results of a recent survey on milk
feeding patterns of infants in the United States. Based on their data, it is
estimated that at one month of age, 50% receive formula, 44% receive breast
milk, and 6% receive breast milk with supplemental formula. (Less than 1% are
estimated to receive cow’s milk or evaporated milk.) No information was given
on the amount of formula received by those breast—fed infants receiving sup-
plemental formula, so it is assumed that at one month half of these infants
*personal communication between Catherine O’Brien, U.S. Bureau of the Census,
and Pauline Johnston, JRB Associates, June 25, 1982.
**personal communication between William Coniglio, Office of Drinking Water,
U.S. Environmental Protection Agency, and Frank Letkiewicz, JRB Associates,
January 1982.
39
-------
receive the majority of their fluid from formula. No information was avail-
able on the amount of tap water consumed directly as water by one-month-old
infants. Consumption of juice among newborns is considered negligible. It
was, therefore, assumed that breast milk or formula constitute the predominant
fluid sources for newborns infants. Based on the Martinez and Dodd (1983)
data, it is estimated that 47% of the newborn infant population receive effec-
tively all of their fluid from breast milk and, therefore, are not exposed to
drinking water. The 53% that are formula-fed have drinking water as their
source of fluid intake. It is therefore estimated •that there are 1,818,000
formula—fed infants obtaining drinking water from public water supplies
(214,419,000 x 0.016 x 0.53).
It should be noted that the population of infants exposed to 1,1—di-
chioroethylene in drinking water from formula feeding would increase with age
up to approximately six months. Martinez and Dodd (1983) reported that, in
1981, 48% of newborn infants in the hospital received formula. Between the
ages of 2-6 months, the percentage of infants receiving formula had increased
to approximately 60-70% of the total infants surveyed. The percentage then
declined to 14% by age 12 months. As the infants grew older, breast feeding
and formula feeding were gradually replaced by feeding with cow’s milk or
evaporated milk. However, other sources of drinking water exposure (e.g., tap
water, juices) are expected to come into greater use as the child ages.
Infant formula is available in three primary forms: ready-to-feed,
liquid concentrate, and powder. Patzer* has provided information indicating
that, over the first year of life, the average formula—fed infant will consume
53% of the total amount of water consumed from formula from the liquid concen-
trate, 38% from ready-to-feed, and 9% from powder. Insufficient information
was available on processing techniques used to prepare ready-to—feed and
liquid concentrate formulas to determine whether VOCs would be removed from
the process water. In the absence of this information, it was assumed that
little, if any, of the VOC present in the water would be removed. (A similar
*personal communication between Emmons Patzer, Mead Johnson, Evansville, IN,
and Corinne Macaluso, JRB Associates, June 9, 1983.
40
-------
assumption is made for purchased beverages included in the drinking water
intake for adults.)
According to Patzer*, 78% of the formula users sterilize the tap water
added to the liquid concentrate and powder formulas. Glick** has indicated
that some data are available suggesting that a 10—minute boiling time of one
gallon of water removes 80—100% of VOC’s present, and that the rapid boiling
of about one quart followed by immediate removal from heating removes 50—80%
of VOCs. In estimating the newborn formula—fed infant pop jlation exposed to
various levels of 1,1-dichloroethylene from drinking water, the drinking water
concentration is adjusted for the portion of infants whose water is boiled
prior to adding to the formula.
The percentage of the original 1,1-dichloroethylene level in the water
received by the formula-fed infant after boiling of the added water is calcu-
lated as follows (using the formula distribution by type provided by Patzer*,
an assumption of no tap water added to the ready—to-feed, 50% dilution of the
liquid concentrate, 100% tap water in the powder, and an assumption of 80%
removal of 1,1—dichloroethylene by boiling tap water):
Liquid concentrate Powder
Ready-to-feed (53% of water, (9% of water,
(38% of water, 50% with no VOC removed, 80% of VOC
no VOC removed) 50% with 80% VOC removed) removed )
(0.38) (1.0) + (0.53) (0.5) (1.0) + (0.53) (0.5) (0.2) + (0.09) (0.2) = 0.72
Therefore, it is estimated that 78% of formula—fed infants receive water with
the equivalent of only 72% of the original VOC level in their water supplies
while the remaining 22% of the formula—fed infants receive water at the level
estimated to be present in their water supplies.
The population of newborn formula—fed infants exposed to the drinking
water levels shown in Table 21 (and later in Table 24) has been adjusted to
*personal communication between Emmons Patzer, Mead Johnson, Evansville, IPI,
and Corinne Macaluso, JRB Associates, June 9, 1983.
**personal communication between Edward Glick, Office of Drinking Water, U.S.
Environmental Protection Agency, and Dave Taylor, consultant to JRB
Associates, April 13, 1983.
41
-------
account for the boiling of added water. The population of infants whose water
is boiled (78%) is subtracted from the estimated population of newborn
formula-fed infants associated with various drinking water concentrations
prior to boiling (based on the data in Tables 16 and 18 and the assumptions
described above). This population is then added to the population at an
appropriate lower cohcentration. The lower concentration is calculated as the
value of 72% of the midpoint of the original concentration range.
From data on the estimated populations exposed to various levels of
1,1—dichioroethylene in drinking water, it was estimated that 40,000 formula-
fed infants are exposed to drinking water containing 1,1—dichloroethylene
levels at or exceeding 0.2 ug/l, while 440 formula—fed infants are exposed to
levels greater than 5 ugh.
5.1.2 Daily Intake of 1,1-Dichloroethylene from Drinking Water
Daily intake levels of 1,1-dichioroethylene from drinking water were
estimated using various exposure levels and the assumptions presented in Table
21. The data in the table suggest that the majority of the persons using
public drinking water supplies would be exposed to intake levels below 0.0057
ug/kg/day for adults and 0.048 ug/kg/day for infants.
5.1.3 Population-Concentration and Population-Exposure Estimates
An indication of the overall exposure of the total population to 1,1-di-
chioroethylene can be obtained through the calculation of population-concen-
tration values. These values are a summation of the individual levels of
1,1—dichioroethylene to which each member of the population is exposed. An
explanation of the derivation of these values is presented in Appendix C.
Population—concentration estimates for 1,1-dichioroethylene in drinking water
were obtained using the values presented in Tables 16 and 18. The estimates
are 1.2 x io6 ugh x persons (best case), 1.3 x iO ug/l x persons (mean best
case), 5.3 x ug/l x persons (mean worst case), and 6.4 x io ug/l x
persons (worst case).
Assuming a consumption rate of 2 liters of water/day, population-exposure
values of 2.4 x ug/day x persons (best case), 2.6 x iO ug/day x persons
(mean best case), 1.1 x io8 ug/day x persons (mean worst case), and 1.3 x i0
ug/day x persons (worst case) were derived.
42
-------
Table 16. Estimated Population (in Thousands) Exposed to
1,1-Dichloroethylene in Drinking Water at the Indicated
Concentration Ranges from Groundwater Systems in Each Size Categorya
System size
(population
served)
25- 100
101-500
501-1, 000
1, 001-2, 500
2, 501-3, 300
3,301-5,000
5,001-10,000
10,001-25,000
25,001-50,000
50, 001-75, 000
7 5,001-100,000
>100,000
Total
Number of
people served
in U.S.
(thousands)
1,031
3,814
3,590
7,047
2,583
4,370
8,404
12,276
10,977
3,911
1,184
14,286
73,473
1,031.4
3,749.0
3,528.8
6,926.9
2,539.0
4,295.6
8,260.8
11,751.4
10,507.9
3,743.9
1,133.4
13,675.5
71,125.6
Population (thousands) exposed to
concentrations (ug/l) of:
__________ 0.2-5 >5—10 >10
15.8 1.8 0.0
58.5 6.5 0.0
55.0 6.1 0.0
108.0 12.0 0.0
39.6 4.4 0.0
67.0 7.4 0.0
128.9 14.3 0.0
524.6 0.0 0.0
469.1 0.0 0.0
167.1 0.0 0.0
50.6 0.0 0.0
________ 610.5 0.0 0.0
2,294.8 52.5 0.0
apopulations may not add to total due to rounding.
blncludes individuals exposed to no 1,1—dichloroethylene
those exposed to levels < 0.2 ug/l.
in drinking water and
,1 )
_1 J
-------
Table 17. Estimated Cumulative Number of People (in Thousands)
Exposed to 1,1-Dichloroethylene in Groundwater Systems
Exceeding the Indicated Concentration (ug/l)
System size
(population
Number of
people served
in U.S.
Cumulat
exposed
ive
to
population (t
concentrations
housand
(ugh)
s)
of:
served)
(thousands)
> 0.2
> S
>
10
25—100 1,031 17.6 1.8 0.0
101-500 3,814 65.0 6.5 0.0
501—1,000 3,590 61.2 6.1 0.0
1,001-2,500 7,047 120.1 12.0 0.0
2,501-3,300 2,583 44.0 4.4 0.0
3,301-5,000 4,370 74.4 7.4 0.0
5,001-10,000 8,404 143.2 14.3 0.0
10,001-25,000 12,276 524.6 0.0 0.0
25,001-50,000 10,977 469.1 0.0 0.0
50,001-75,000 3,911 167.1 0.0 0.0
75,001-100,000 1,184 50.6 0.0 0.0
>100,000 14,286 610.5 0.0 0.0
Thtal 73,473 2,347.3 52.5 0.0
Lower bOUflda 1,515.5 0.0 0.0
Upper bOUflda 3,179.2 154.9 0.0
aFrom 95% confidence intervals.
44
-------
Table 18. Estimated Population (in Thousands) Exposed to
1,1 —Dichioroethylene in Drinking Water at the Indicated Concentration
Ranges from Surface Water Systems in Each Size Category
System size
(population
served)
Number of
people served
in U.S.
(thousands)
P
opulation (thousands) exposed
to concentrations (ug/l) of:
< 0.2
0.2—5
> 5
25-100
86
86.0
0.0
•
0.0
101-500
690
690.0
0.0
0.0
501—1,000
1,051
1,051.0
0.0
0.0
1,001-2,500
3,295
3,295.0
0.0
0.0
2,501-3,300
1,445
1,445.0
0.0
0.0
3,301.-5,000
3,096
3,096.0
0.0
0.0
5,001-10,000
6,763
6,763.0
0.0
0.0
10,001-25,000
15,595
15,289.2
305.8
0.0
25,001-50,000
13,945
13,671.6
273.4
0.0
50,001-75,000
9,483
9,297.1
185.9
0.0
75,001-100,000
7,131
6,991.2
139.8
0.0
>100,000
78,366
76,829.4
1,536.6
0.0
Total
140,946
138,504.4
2,441.6
0.0
apopulations may not add to total due to rounding.
blncludes individuals exposed to no 1,1—dichioroethylene in drinking water and
those exposed to levels < 0.2 ugh.
45
-------
Table 19. Estimated Cumulative Number of People (in Thousands)
Exposed to 1,1—Dichioroethylene from Surface Water Systems
Exceeding the Indicated Concentration (ugh)
System size peo
(population
served) (t
Number of
ple served
in U.S.
housands)
Cumulative population (thousands)
to concentrations (ug/l)
exposed
of:
>0.2 >5
25-100
86
0.0 0.0
101-500
690
0.0 0.0
501-1,000
1,051
0.0 0.0
1,001—2,500
3,295
0.0 0.0
2,501-3,300
1,445
0.0 0.0
3,301-5,000
3,096
0.0 0.0
5,001-10,000
6,763
0.0 0.0
10,001—25,000
15,595
305.8 0.0
25,001-50,000
13,945
273.4 0.0
50,001-75,000
9,483
185.9 0.0
75,001-100,000
7,131
139.8 0.0
>100,000
78,366
1,536.6 0.0
Total
140,946
2,441.6 0.0
Lower bounda
0.0 0.0
Upper bOUflda
5,711.0 0.0
aFrom 95% confidence intervals.
46
-------
Table 20. Total Estimated Cumulative Population (in Thousands)
Exposed to 1,1-Dichioroethylene in Drinking Water
Exceeding the Indicated Concentration
System type
Number of
people served
in U.S.
(thousands)
Cumulative
to
population (thousands)
concentrations (ug/l)
exposed
of:
> 0.2
> 5
> 10
Groundwater
73,473.
2,347
52
0
Surface water
140,946
2,442
0
0
Total
214,419
4,789
52
0
(% of tota’)
(100%)
(2.2%)
(<0.1%)
(0.0%)
47
-------
Table 21. Estimated Drinking Water Intake of 1,1—Dichioroethylene
by Adults and Formula-Fed Infants
Persons using supplies
exposed to indicated levels Intake (ug/kg/day )
, ,0
Exposure level Total Formula—fed % of Formula- Formula-fed
(ug/l) Population population infants fed infants Adult infant
>0.2 4,789,000 2.2% 40,000 2.2% >0.0057 >0.048
>5.0 52,000 < 0.1% 440 < 0.1% >0.14 >1.2
>10 0 0.0% 0 0.0% >0.29 >2.4
Assumptions: 70—kg man, 3.5—kg infant; 2 liters of water/day (man), 0.85 liters of
water/day (formula-fed infant).
48
-------
5.2 DIETARY INTAKE
No data were obtained on levels of 1,1-dichioroethylene in foods. There-
fore, the intake of 1,1—dichioroethylene from the diet could not be estimated.
5.3 RESPIRATORY INTAKE
Exposure to 1,1—dichioroethylene in the atmosphere varies from one loca-
tion to another. High levels, averaging greater than 20,000 ng/m 3 (20 ug/m 3 ),
have been detected in several areas. Normal levels, however, are lower.
Using the data for quality codes 1-3 presented in Table 15, Brodzinsky and
Singh (1982) calculated median air levels of 1,1-dichloroethylene for urban!
suburban areas and source dominated areas of 20 ng/m 3 (0.020 ug/m 3 ) and 14,000
ng/m 3 (14 ug/m 3 ), respectively. Two data points for 1,1—dichioroethylene in
rural/remote areas showed 0.0 ng/m 3 (quality code 4). The highest level of
1,1—dichloroethylene in the atmosphere reported under quality codes 1-3 was
27,000 ng/ni 3 (27 ug/m 3 ) (Going and Spigarelli 1977 cited in Brodzinsky and
Singh 1982).
The monitoring data avail able are not suffi ci ent to determine regi onal
variations in exposure levels for 1,1-dichloroethylene. However, urban and
industrial areas appear to contain higher levels, as expected.
The daily respiratory intake of 1,1-dichloroethylene from air was esti-
mated using the assumptions presented in Table 22 and the median and maximum
levels for 1,1—dichloroethylene reported above. The estimates in Table 22
indicate that the daily 1,1-dichioroethylene intake from air for adults in
source dominated areas is approximately 4.6 ug/kg/day. The intake calculated
using the maximum level reported is 8.9 ug/kg/day. The values presented do
not account for variances in individual exposure or uncertainties in the
assumptions used to estimate exposure.
49
-------
Table 22. Estimated Respiratory Intake of 1,1—Dichloroethylene
by Infants and Adults
Exposure (ug/ni 3 )
Intake
(ug/kg/day)
Adult
Infant
Rural/remote (0.0)
0.0
0.0
Urban/suburban (0.020)
0.0066
0.0046
Source dominated (14)
4.6
3.2
Iviaxinium (27)
8.9
6.2
Assumpti ons:
70-kg man, 3.5—kg infant; 23 m 3 of air inhaled/day (man), 0.8 ni 3
of air inhaled/day (infant) (ICRP 1975).
50
-------
5.4 RELATIVE SOURCE CONTRIBUTION
This section of the report considers the relative contribution of drink-
ing water and ambient air to total human exposure to 1,1—dichloroethylene from
these two sources. (Since no data were obtained on levels of 1,1—dichloro—
ethylene in foods in the United States, the contribution of 1,1-dichioro—
ethylene in the diet to. total 1,1—dichloroethylene exposure could not be
assessed.) The data presented here indicate the potential total exposure to
1,1—dichloroethylene which could occur if a population was exposed to specific
combinations of 1,1—dichloroethylene concentrations in drinking water and
ambient air.
Table 23 presents a general view of the total amount of 1,1-dichloro-
ethylene received by an adult male from air and drinking water. Table 24 pre-
sents similar conditions for a formula—fed infant. Four separate exposure
levels in air and four exposure levels in drinking water are shown in the
tables. The data presented represent possible exposures based on the occur-
rence data in Sections 2 through 4 and the estimated intakes in Sections 5.1
through 5.3.
Brodzinsky and Singh (1982) calculated a median urban/suburban air level
of 1,1—dichloroethylene of 0.020 ug/ni 3 based on air monitoring data (Section
4). Assuming an ir level of 0.020 ug/m 3 , drinking water would be the pre-
dominant source of 1,1—dichloroethylene exposure in the adult male at drinking
water levels above 0.2 ug/l. However, for formula—fed infants exposed to air
levels of 0.020 ug/rn 3 , intake of 1,1—dichioroethylene from drinking water is
estimated to be the predominant source of exposure at levels above 0.02
ug/l. An accurate assessment of the number of individuals for which drinking
water is the predominant source of exposure cannot be determined from the data
since specific locations containing high concentrations of 1,1-dichloro-
ethylene in drinking water and low concentrations of 1,1—dichloroethylene in
ambient air and food are unknown.
The data presented have been selected from an infinite number of possible
combinations of exposure from each of the two sources. The actual exposures
encountered would represent some finite subset of this infinite series of
combinations. Whether exposure occurs at any specific combination of levels
is not known; nor is it possible to determine the number of persons that would
be exposed to 1,1—dichloroethylene at any of the combined exposure levels.
51
-------
The relative source contribution data are based on estimated intake and
do not account for a possible differential absorption rate for 1,1-dichloro-.
ethylene by route of exposure. The relative dose received may vary from the
relative intake. In addition, the relative effects of the chemical on the
body may vary by different routes of exposure.
52
-------
Table 23.
from the
Estimated Intake of 1,1—Dichioroethylene
Environment by Adult Males in ug/kg/day
(% from Drinking Water)
Concentration in
drinking water
(ug/l)
Concentration in
air
Rural/remote
(0.0 ug/m 3 )
Urban/suburban
(0.020 ug/m 3 )
Source
(14
dom nated
ug/m )
Maximum
(27 ug/rn 3 )
0
0.0 (--)
0.0066 (0%)
4.6
(0%)
8.9
(0%)
0.28
0.0057 (100%)
0.012 (48%)
4.6
(0.1%)
8.9
(0.06%)
50 b
0.14 (100%)
0.15 (93%)
4.7
(3.0%)
9.0
(1.6%)
10 c
0.29 (100%)
0.30 (97%)
4.9
(5.9%)
9.2
(3.2%)
Intake from each source (see Sections 5.1-5.3) :
Food: Not included
a 4 , 789 , 000 individuals using public drinking water systems are estimated to be
exposed to levels > 0.2 ug/l (2.2% of population using public water supplies).
b 521000 individuals using public drinking water systems are estimated to be
exposed to levels > 5.0 ug/l (< 0.1% of population using public water
supplies).
cNo individuals using public drinking water systems are estimated to be
exposed to levels > 10 ug/l.
Water:
0.2 ug/l:
0.0057 ug/kg/day
5.0 ug/l:
0.14 ug/kg/day
10 ug/l:
0.29 ug/kg/day
Air:
.
0.0 ug/m 3 :
0.020 ug/m 3 :
14 ug/m 3 :
27 ug/m 3 :
0.0 ug/kg/day
0.0066 ug/kg/day
4.6 ug/kg/day
8.9 ug/kg/day
53
-------
Table 24. Estimated Intake of 1,1-Dichioroethylene
from the Environment by Newborn Formula-Fed Infants in ug/kg/day
(% from Drinking Water)
Concentration in
drinking water
(ugh)
Concentration in
air
Rural/reni te
(0.0 ug/m )
Urban/subur an
(0.020 ug/m )
Source
(14
dom nated
ug/m )
Maximum 3
(27 ug/m
)
0
0.0 (-—)
0.0046 (0%)
3.2
(0%)
6.2
(0%)
0 , 2 a
0.048 (100%
0.053 (91%)
3.2
(1.5%)
6.2
(0.8%)
50 b
1.2 (100%)
1.2 (100%)
4.4
(27%)
7.4
(16%)
10 C
2.4 (100%)
2.4 (100%)
5.6
(43%)
8.6
(28%)
Intake from each source (see Sections 5.1-5.3) :
Water: 0.2 ugh : 0.048 ug/kg/day
5.0 ugh: 1.2 ug/kg/day
10 ugh : 2.4 ug/kg/day
Air: 0.0 ug/rn 3 : 0.0 ug/kg/day
0.020 ug/n1 3 : 0.0046 ug/kg/day
14 ug/m 3 : 3.2 ug/kg/day
27 ug/m 3 : 6.2 ug/kg/day
Food: Negligible
a 40 , 000 formula-fed infants using public drinking water systems are expected
to be exposed to levels > 0.2 ug/l (2.2% of formula-fed infant population
using public water supplier).
b 440 formula—fed infants using public drinking water systems are expected to
be exposed to levels > 5.0 ug/l (< 0.1% of formula-fed infant population using
public water supplies).
CNO formula—fed infants using public drinking water systems are expected to be
exposed to levels > 10 ug/l.
54
-------
t I — QQ
&QCJ Ak/YSt&.. REFERENCES
Aitchison J, Brown JAC. 1957. The lognorinal distribution. London: Cambridge
University Press. p. 95.
Boland PA. 1981. National Screening Program for Organics in Drinking
Water. Prepared by SRI International for Office of Drinking Water, U.S.
Environmental Protection Agency, Washington, DC. Contract No. 68-01-4666.
Bozzelli JW, Kebbekus 8, Greenberg A. 1980. Analysis of selected toxic and
carcinogenic substances in ambient air in New Jersey. Prepared by New Jersey
Institute of Technology, Newark, NJ, for New Jersey Department of Environ-
mental Protection. Cited in Brodzinsky and Singh 1982.
Brass HJ, Feige MA, Halloran T, Mello JW, Munch 0, Thomas RF. 1977. The
National Organic Monitoring Survey: A sampling and analysis for purgeable
organic compounds. In: Drinking water quality enhancement through source
protection (Pojasek RB, ed.). Ann Arbor, MI: Ann Arbor Science. pp. 393-
416.
Brodzinsky R, Singh HB. 1982. Volatile organic chemicals in the atmosphere:
An assessment of available data. Prepared by SRI International, Menlo Park,
CA, for Environmental Sciences Research Laboratory, Office of Research and
Development, U.S. Environmental Protection Agency, Research Triangle Park,
MC. Contract No. 68-02-3452.
Callahan MA, Slimak MW, Gabel NW, May IP, Fowler CF. Freed JR. Jennings P.
Durfee RL, Whitmore FC, Maestri B, Mabey WR, Holt BR, Gould C. 1979. Water
related environmental fate of 129 priority polluants. Volume II. Washington,
DC: Office of Water Planning and Standards, U.S. Environmental Protection
Agency. EPA-.440/4-79-029b.
California State Data
CA-Ol Memorandum (with attached table of data): Analysis of California
Department of Health Services well water samples, January 20, 1980.
CA-12 1) Raw data tables from California Analytical Laboratories, Inc.,
3/29/80, 1/3/80, 2/8/80, 8/23/79, 8/22/79.
2) Raw data tables from Anlab, 2/12/80, 1/28/80, 12/31/79.
3) Raw data tables from California Department of Health, 10/23/80,
9/28/79.
Cochran WG. 1963. Sampling techniques. 2nd ed. New York, NY: John Wiley
arid Sons. pp. 106-107.
Farmer 1, Hodge V, Bryson H, Ney J, Oakley G, Slimak K. 1980. Materials
balance —— Chlorinated solvents. Final draft report. Prepared by JRB
Associates for Office of Pesticides and Toxic Substances, U.S. Environmental
Protection Agency. Contract No. 68—01-5793.
55
-------
FRDS. 1983. Federal Reporting Data System. Facilities and population served
by primary water supply source (FRDSO7), April 19, 1983. U.S. Environmental
Protection Agency, Washington, DC.
Going JE, Spigarelli JL. 1977. Environmental monitoring near industrial
sites: Vinylidene chloride. Prepared by Midwest Research Institute, Kansas
City, MO, for U.S. Environmental Protection Agency. EPP -56O/6—77—O26. Cited
in Brodzinsky and Singh 1982.
ICRP. 1975. International Commission on Radiological Protection. Report of
the task group on reference man. ICRP Publication 23. New York: Perg mon
Press.
Johnson ML. 1949. Systems of frequency curves generated by methods of
translation. Biometrika 36:149-176.
Johnson ML, Kotz S. 1969. Discrete distributions. New York, NY: John Wiley
and Sons. pp. 288-289.
Johnson NL, Kotz S. 1970. Continuous univariate distributions. Mew York,
NY: John Wiley and Sons.
JRB Associates. 1982. Evaluation of drinking water data in EPA studies.
Draft report. Prepared for Office of Drinking Water, U.S. Environmental
Protection Agency, Washington, DC. Contract No. 68-01-6185.
Kuzinack AM. 1983. Characterization of the water supply industry (FY 82).
M emorandum of May 16, 1983. Office of Water, U.S. Environmental Protection
Agency, Washington, DC.
Martinez GA, Dodd DA. 1983. 1981 milk feeding patterns in the United States
during the first 12 months of life. Pediatrics 71(2):166-170.
Massachusetts State Data
MA—Ol Tables presenting sampling and analysis data for Wilmington and
Belchertown. Massachusetts Department of Environmental Quality
Engineering, undated.
MA—09 Tables presenting sampling and analysis data for Acton, Assabet,
Rowley, Dartmouth, and Bedford. Massachusetts Department of
Environmental Quality Engineering, undated.
MA-18 Memorandum: Lunenburg update. August 10, 1980.
New Jersey State Data
NJ-O1 Two memorandums (with attached table of data): Fact sheet on
contaminated wells in Mahwah, Mew Jersey, January 24, 1979 and Well
contamination in Fairlawn, New Jersey, January 19, 1979.
Ostle B. 1963. Statistics in research. 2nd ed. Ames, IA: The Iowa State
University Press. p. 132.
56
-------
Pellizzari ED. 1978a. Measurement of carcinogenic vapors in ambient atmo-
spheres. Prepared by Research Triangle Institute, Research Triangle Park, NC,
for U.S. Environmental Protection Agency. EPA-600/7—78—062. Cited in
Brodzinsky and Singh 1982.
Pellizzari ED. 1978b. Quantification of chlorinated hydrocarbons in
previously collected air samples. Prepared by Research Triangle Institute,
Research Triangle Park, NC, for U.S. Environmental Protection Agency. EPA
450/3-78-112. Cited in Brodzinsky and Singh 1982.
Pellizzari ED, Bunch JE. 1979. Ambient air carcinogenic vapors —— Improved
sampling and analytical techniques and field studies. Prepared by Research
Triangle Institute, Research Triangle Park, NC, for U.S. Environmental Protec-
tion Agency. EPA-600/2-79—081. Cited in Brodzinsky and Singh 1982.
Pellizzari ED, Erickson MD, Zweidinger RA. 1979. Formulation of a prelimi-
nary assessment of halogenated organic compounds in man and environmental
media. Prepared by Research Triangle Institute, Research Triangle Park, NC,
for U.S. Environmental Protection Agency. EPA—560/13-179-006. Cited in
Brodzinsky and Singh 1982.
Singh HB, Salas U, Stiles R, Shigeishi H. 1980. Atmospheric measurements of
selected hazardous organic chemicals. Second year interim report. Prepared
by SRI International, Menlo Park, CA, for U.S. Environmental Protection
Agency. Grant No. 805990. Cited in Brodzinsky and Singh 1982.
Snedecor GW, Cochran WG. 1967. Statistical methods. 6th ed. Ames, IA: The
Iowa State University Press. pp. 213—219.
Stephens MA. 1974. EDF statistics for goodness of fit and some comparisons.
J. Amer. Stat. Assoc. 69:730-737.
Symons JM, Bellar TA, Carswell JK et al. 1975. National Organics
Reconnaissance Survey for Halogenated Organics. J. Amer. Water Works Assoc.
667(11):634-647.
Wallace L. 1981. Measurements of volatile organic compounds in breathing-
zone air, drinking water and exhaled breath. Preliminary draft. Cited in
Brodzinsky and Singh 1982.
Westrick JJ, Mello JW, Thomas RF. 1983. The Ground Water Supply Survey
summary of volatile organic contaminant occurrence data. EPA Technical
Support Division, Office of Drinking Water, Cincinnati, Ohio.
57
-------
APPENDIX A
Methodology for Estimating the National Occurrence of
Volatile Synthetic Organic Chemicals in Public Drinking Water Supplies
and the Size of the Exposed Population
The U.S. Environmental Protection Agency (EPA) is currently considering
the proposal of national revised primary drinking water regulations under the
Safe Drinking Water Act, as well as non-federal regulatory approaches, to
limit human exposure to high levels of certain volatile synthetic organic
chemicals (VOCs) that have been detected in drinking water (Advanced Notice of
Proposed Rulemaking, 47 FR 9350, March 4, 1982). The specific VOCs of imme-
diate interest to EPA are:
• Trichioroethylene
• Tetrachloroethylene
• Carbon tetrachioride
• 1,1 ,l-Trichloroethane
• 1 ,2-Dichloroethane
• Vinyl chloride
• Dichioromethane
• Benzene
• Chlorobenzene
• Di chi orobenzene(s)
• Trichioroberizene(s)
• 1 ,l-Dichloroethylene
• 1 ,2—Dichloroethylene (cis and trans)
The objective of the analysis of the occurrence of the VOCs listed above
is to support EPA ’s consideration of the need and alternatives for controlling
VOCs in public water supplies in two principal areas:
• As input to the health risk assessments for the VOCs, the analysis
provides estimates of the number of individuals in the United States
exposed to various levels of VOCs in drinking water from public water
supplies.
A-i
-------
• As input to the assessment of the economic impact of the regulatory
and treatment alternatives being considered, the analysis provides
estimates of the number of public water supplies of various source
(i.e., groundwater and surface water) and size (based on population
served) categories having VOCs present, and the distribution of VOC
levels in those water supplies.
The methodology for preparing these estimates involved the creation of a
data base drawing from the results of several Federal surveys on the measured
occurrence of the VOCs in public water supplies as a function of water source
and supply size. Statistical models were then used to extrapolate from the
observed frequency of occurrence of the VOCs in the supplies sampled in the
Federal surveys to the universe of public water supplies having similar source
and size characteristics. A separate report has been prepared for each of the
VOCs listed. Appendix A is included in each report to provide detail on the
sources of data and the methodology used to obtain these estimates. Some
specifics presented in Appendix A may not be applicable to the VOC that is the
subject of this report. Appendix B provides additional detail on the selec-
tion of an appropriate model for estimating the national occurrence of the VOC
that is addressed in this report.
A.i SOURCES OF DATA
A.1.i Number of Public Drinking Water Supplies in the United States and Size
of Populations Served
It is currently estimated that there are approximately 60,000 public
drinking water supplies in the United States serving approximately 214 million
people. Table A-i summarizes the estimated number of surface water and
groundwater systems of various sizes and the associated populations served by
them. These data, which correspond to the “FY 82 Characterization of the
Water Supply Industry” presented by Kuzmack (1983), as updated by Schnare*,
were derived from the Federal Reporting Data Systems (EROS) for FY 1982 (FRDS
1983).
*personal communication between David Schnare, Office of Drinking Water, U.S.
Environmental Protection Agency, and Frank Letkiewicz, JRB Associates, May
25, 1983.
A- 2
-------
Table A-i. Number of Systems and Population Served
by Primary Water Supply Source
(By Population Category)
Surface
Groundwater
Average
Average
System size No. of
(Population served) systems
Population
(thousands)
population
served
No. of
systems
Population
(thousands)
population
served
1 25-100 1,525
Very small ‘
L 101—500 2,412
86
690
56
286
19,125
15,674
1,031
3,814
54
243
501-1,000 1,377
Small 1,001—2,500 1,945
L 2,501—3,300 495
1,051
3,295
1,445
763
1,700
2,900
4,877
4,400
891
3,590
7,047
2,583
736
1,600
2,900
1 3,301—5,000 749
Medium (
5,001-10,000 930
3,096
6,763
4,100
7,300
1,065
1,168
4,370
8,404
4,100
7,200
10,001—25,000 9158
15 , 595 b
17,000
8358
12 , 276 b
15,000
25,001—50,000 400 a
La r g
13945 b
35,000
290 a
10977 b
38,000
50,001-75,000 155
9,483
61,000
64
3,911
61,000
75,001-100,000 82
7,131
87,000
14
1,184
85,000
Very large >100,000 217
78,366
360,000
55
14,286
260,000
TOTALSC: 11,202
140,948
48,458
73,475
aKuzmack 1983, as updated by David Schnare, Office of Drinking Water, U.S. Environ-
mental Protection Agency, in a personal communication with Frank Letkiewicz, JRB
Associates, May 25, 1983.
bEstimated by JRB Associates (see Table A-2).
Cpopulations do not add to total due to rounding.
Source: FRDS 1983 (except as noted).
A- 3
-------
It should be noted that FRDS (1983) does not provide a breakdown of the
number of systems or of the population served for the 10,001-25,000 and
25,001-50,000 size categories, but rather for 10,001-50,000 as a single cate-
gory. The estimated number of systems in the 10,001-25,000 and 25,001-50,000
size categories, as presented in Kuzmack (1983), were estimated by Dr. David
Schnare of the Office of Program Development and Evaluation, EPA Office of
Drinking Water from additional FRDS data. (Data for these additional
categories are needed for the economic impact analysis.) The population
served by systems in these size categories were estimated by JRB as shown in
Table A-2.
Table A—2. Analysis for Estimating the Population Served
by Surface Water and Groundwater Supplies in the
10,001-25,000 and 25,001-50,000 Population Size Categories
System size
(population
served)
Source
Total
Surface
water
Groundwater
10,001-25,000
w
x
27 , 87 O
25,001-50,000
y
24 , 920 a
Total
29 , 540 b
23 , 250 b
52,790
aEstimated total population for surface water and groundwater provided by
Kuzmack (1983).
bpopulation served by surface water and groundwater supplies in the 10,001-
50,000 category from FRDS (1983).
= 29,540 x 27,870 = 15,595
5 ,790
= 23,250 x 27,870 = 12,276
52,790
= 29,540 x 24,920 = 13,945
52,790
= 23,250 x 24,920 = 10,977
52, 190
A-4
-------
A.l.2 Data on Measured Occurrence of VOCs in Public Water Supplies
There are three primary sources of information available on the occur-
rence of VOCs in drinking water supplies that were considered for use in
preparing the national projections:
• Federal surveys,
• State data, and
• Miscellaneous data.
Only the Federal survey data were ultimately used for the national pro-
jections. While a substantial amount of state and miscellaneous published
data were available for some of the VOCs examined, these data usually did not
provide adequate details on the water source, size of the population served,
or type of sample taken (raw, finished, or distribution), which precluded
incorporating them into the analysis. Furthermore, there was usually no
detailed information available on the sampling and analysis methodology used
to obtain the data, which precluded their being subjected to a quality assur-
ance review (performed as a separate task) (JRB Associates 1982). In addi-
tion, much of the state and miscellaneous data appeared to have been obtained
in response to spills, citizen complaints, or other evidence of contamination
and, therefore, were not considered to be representative data for preparing
the national projections. While the state and miscellaneous published data
were not used for deriving the national projections, these data are presented
and discussed in the individual VOC occurrence reports.
The Federal survey data generally provided the information on water
source, population served, and sample type studied that was necessary to
perform the analyses in this report. These surveys also provided sufficient
information on the sampling and analysis methods to be subjected to the
quality assurance review. The following six Federal surveys were used for the
national projections:
• National Organics Reconnaissance Survey (NORS)
The National Organics Reconnaissance Survey (NORS) was conducted
early in 1975 for the purpose of determining levels of four trihalo-
methanes (chloroform, bromodichioromethane, dibromochloromethane, and
brornoform), carbon tetrachloride, and 1,2-dichioroethane in finished
water supplies from 80 cities across the country (Symons et al.
1975). A population base of 36 million individuals was covered during
A-S
-------
the study. Analysis of samples was performed by the Water Supply
Research Laboratory of EPA in Cincinnati using purge and trap 98S
chromatography with an electrolytic conductivity detector.
• National Organic Monitoring Survey (NOMS)
The National Organic Monitoring Survey (NOMS) was conducted to
determine the frequency of occurrence of specific organic chemicals in
finished water supplies of 113 cities across the country (Brass et al.
1977). Among the chemicals surveyed were trihalomethanes, 1,2-di—
chloroetharie, carbon tetrachloride, trichioroethylene, benzene, vinyl
chloride, dichlorobenzene, and trichlorobenzene. Data from three
phases (referred to as MOMS I, MOMS 11, and MOMS III) of the study
were collected over an eleven month period (March 1976 to January
1977) to reflect any long-term or seasonal variations. The analytical
treatment of the samples was similar to that for the NORS samples.
(Gas chromatography/mass spectronetry analyses were done for benzene.)
• National Screening Program for Organics In Drinking Water (NSP)
SRI International conducted a study from June 1977 to March 1981,
entitled National Screening Program for Organics in Drinking Water
(NSP), in which raw and finished drinking water samples were collected
from 166 water facilities located in 33 states (Boland 1981). The
compounds sampled were 23 halocarbons, 6 aromatics, and 22 pesticides,
phenols, and acids. The methods used for analysis included gas chrom-
atography with electron capture detection for purgeable halocarbons
and the base/neutral extraction fraction, and gas chromatography with
flame ionization detection for purgeable arornatics.
• Comunity Water Supply Survey (CWSS)
The Community Water Supply Survey (CWSS) examined 106 surface
water supplies, 330 groundwater supplies, and 16 supplies with mixed
water or purchased sources in 1978. Trihalomethanes and other vola-
tile organic chemicals, including carbon tetrachloride, chlorobenzene,
1,2—dichloroethane, cis- and trans—1,2—dichloroethylene, tetrachloro-
ethylene, 1,1,1-trichloroethane, trichloroethylene, benzene, toluene,
and xylenes were measurecj. One to five samples were collected from
each system, including raw, finished, and/or distribution water. Gas
chromatography with an electrolytic conductivity detector was used for
halocarbons and a flame ionization detector for aromatic analyses.
• Rural Water Survey (RWS)
The Rural Water Survey, conducted in 1978, was carried out in
response to Section 3 of the Safe Drinking Water Act, which mandated
that EPA “conduct a survey of the quantity, quality, and availability
of rural drinking water supplies.” Samples collected from 855 house-
holds in rural areas from across the United States were analyzed for
trihalomethanes and for carbon tetrachloride, 1,2-dichioroethane, cis-
and trans-i ,2-dichloroethylene, tetrachloroethylene, 1 ,l ,l-trichloro-
ethane, and trichloroethylene using gas chromatography with an elec-
trolytic conductivity detector. The majority of the 855 samples were
A-6
-------
from households using private wells or small supplies serving fewer
than 25 people. Using information provided by Dr. Bruce Brower at
Cornell University, Department of Rural Sociology, it was determined
that the RWS file had data for up to 207 groundwater and up to 45
surface water supplies serving more than 25 people.
• Groundwater Supply Survey (GWSS)
The Groundwater Supply Survey (GWSS), conducted between December
1980 and December 1981, involved the national sampling of 945 public
water supply systems using groundwater sources for 5 trihalomethanes
and 29 other organic chemicals. Analyses were done using purge and
trap gas chromatography with an electrolytic conductivity detector for
halocarbons and a non-destructive photoionization detector for aroma-
tics. There were 466 randomly selected supplies and 479 selected with
state and EPA regional input based on the likelihood of finding some
VOC contamination.
Table A-3 indicates which of the VOCs listed earlier were examined in
each of the six Federal surveys used for the national projections. (Addi-
tional details on those surveys providing data on the VOCs addressed in this
report are presented in Chapter 2.)
Table A-3. VOCs of Interest Examined in the Six Federal Surveys
Used for National Projections
NOMS NSP CWSS RWS GWSS
Trichioroethylene X X X X X
Tetrachloroethylene X X X X X
Carbon tetrachioride X X X X X X
1,1 ,l-Trichloroethane X X X X X
1 ,2-Dichloroethane X X X X X X
Vinyl chloride X X X
Dichloromethane X X
Benzene X X X X
Chlorobenzene X X X
Dichlorobenzene(s) X X X
Trichlorobenzene(s) X X
l,l-Dichloroethylene X X
l,2-Dichloroethylene X X X X
(cis and trans)
aDichloromethane data for the GWSS were not used due to a sample contamination
probl em.
A- 7
-------
A.2 GENERAL METHODOLOGY
A.2.1 Development of Survey Files
To facilitate the handling of data for preparing the national projec-
tions, suitable machine-readable files were developed for each of the six
Federal surveys. JRB was provided access to existing computer files for the
CWSS, RWS, and GWSS through the EPA Office of Drinking Water’s Technical
Support Division (ODW/TSD) in Cincinnati, Ohio. JRB used the published
results of the NORS (Synions et al. 1975) and the NSP (Boland 1981), and
printed results for the NOMS provided by ODW/TSD to create machine readable
files of those surveys. Three separate files were created for NOMS, one for
each of the three phases. (In effect, NOMS I, II, and III were treated as
three separate surveys.) The final files for all chemicals were in SAS for-
mat. All computer efforts for this project utilized EPA’s NCC-IBM (IBM 370)
computer at Research Triangle Park, North Carolina.
It was necessary to prepare working versions of each survey file contain-
ing the following minimum information for each of the sampled water supplies:
• Location of the supply (state and city);
• Population served by the supply;
• Water source (groundwater, surface water, mixed, purchased, etc.); and
• A single concentration value for each VOC sampled.
With the exception of the RWS and NSP, the existing files and printed
sources provided adequate information on the location of the supply sampled.
The RWS design involved the collection of drinking water samples from house-
holds in rural areas of the United States. With the assistance of Dr. Bruce
Brower at Cornell University’s Department of Rural Sociology (responsible for
the preparation of a detailed analysis of RWS results on inorganics, pesti-
cides, and other parameters), it was possible to determine which of the 855
households for which VOC analyses were done obtained water from public water
supplies. However, because of confidentiality restrictions on the RWS data,
it was only possible to determine the location of the household and the public
water supplies sampled at the state and county level, but not at the city or
town level.
A- 8
-------
For NSP, the locations were not reported in Boland (1981); however, the
Office of Drinking Water, Science and Technology Branch (ODW/STB) was able to
provide copies of data sheets on the supplies sampled in NSP which provided
information on location.
The existing files of the CWSS and GWSS each provided data on the size of
the population served by the supplies sampled. For NORS, it was necessary to
estimate the population served using information presented in Symons et al.
(.1975) on the location of the supply and recent population data for those
areas from other sources. It should be noted that most of the supplies
sampled in the NORS fall into the large and very large size categories. Con-
sequently, errors in the precise number used for persons served by those
systems would not alter their classification or their impact on the national
projections.
For HUMS, data on the populations served by supplies were provided by
ODW/TSD. In the case of NSP, the population served by each supply was not
reported in Boland (1981). Again, those data were obtained from the data
sheets provided by ODW/STB. (There were three NSP locations for which the
population was not specified.)
For RWS, data on the size of the population served by supplies were not
collected. However, data were obtained on the number of service connections
for each supply. With the assistance of Dr. Bruce Brower at Cornell
University, it was possible to estimate the population served by each supply
from the service connection data and data on the average number of individuals
per household observed in the survey (3.034).
The identification of water source as groundwater, surface water, mixed,
etc. was clearly designated in the CWSS, RWS, and GWSS files. For HORS, the
source was determined from the descriptive information in Symons et al.
(1975). For HUMS, source designations were provided by ODW/TSD. For NSP,
source information was given on the data sheets provided by ODW/STB.
Some public water supplies use a mixture of groundwater and surface water
sources. Although these “mixed” supplies are counted as groundwater or sur-
face water among the 60,000 supplies in FRDS, based on the predominant source
used, such supplies were excluded from the survey data for developing the
national projections because the predominant source was rarely indicated in
A- 9
-------
the survey file. Similarly, water supplies identified as purchasing water
from another, usually unspecified, supply were excluded from the survey data
for the national projections.
A.2.2 Computing Average Values for VOCs in Each Survey
In order to prepare the national projections, it was necessary that a
single value be obtained for each VOC in each supply sampled. This require-
ment presented certain difficulties for several of the files where multiple
sample results were reported for the supplies. NORS and NSP provided data on
raw (i.e., untreated water sampled at the supply) and finished (i.e., treated
water sampled at the supply) samples; CWSS provided data on raw, finished, and
distribution (i.e., water sampled at a user’s faucet) samples; MOMS and GWSS
used finished water only; and RWS used distribution water only. In order that
the national projections be derived from data on drinking water samples most
representative of what people actually consume, all data on raw water were
excluded from this analysis. Initially, consideration was given to excluding
finished water sample data in the CWSS for supplies also having distribution
samples. However, ODW/TSD staff indicated that inconsistencies in coding data
for the CWSS resulted in some errors in designating water as finished or
distribution. Therefore, all CWSS finished and distribution samples were
included in the analysis as being equally representative of the water to which
consumers are exposed.
The NORS, NOMS, and GWSS provided a single analytical result for each VOC
in finished water. For NSP and RWS, there was generally only a single value
reported for each VOC, although a few systems had multiple samples. In CWSS,
most supplies had multiple finished and/or distribution samples for each
VOC. Where multiple samples occurred, a single “supply value” was computed
for each VOC using the following rules:
• If positive values were reported for all samples, the supply value was
computed as the arithmetic mean.
• If both positive values and values below the minimum quantifiable
concentration were reported, the supply value was computed as the mean
of the positive values and the minimum quantifiable concentration
values.
A- 10
-------
• If values below the minimum quantifiable concentration were reported
for all samples, the supply value was computed as the mean of the
minimum quantifiable limits reported for each sample. (These means
were recorded as Thegative” values to indicate that the VOC was not
observed.)
Computing a single value for supplies where the value for one or more of
the samples was reported to be below the minimum quantifiable concentration
was problematic. These “negative” values imply that the analyte in question
may or may not be present and, if present, is so at a concentration below that
measurable by the analytical method. In other words, the actual value is
greater than or equal to zero and less than the minimum quantifiable concen-
trati on.
Where a supply was reported to have samples with both positive and nega-
tive values, two major alternatives were considered for treating the negative
values. The first was to ignore or eliminate the negative values from the
computation of the mean. This was considered unacceptable because it implies
that the negative data are less valid than the positive data, which is not the
case. This alternative would also necessarily result in a higher average
value for that system than would be the case if the actual value for the
rregative data were known.
The second alternative was to assign the negative data a specific value
for computing the supply average. Three possibilities were considered: 0, the
minimum quantifiable concentration, and the midpoint between 0 and the minimum
quantifiable concentration. Assigning the negative samples a value equal to
the minimum quantifiable concentration was selected since this gives the most
conservative estimate of the supply value. That is, if the analyte was in
fact present, the maximum possible concentration it could have in that sample
would be approximately (actually slightly less than) the minimum quantifiable
concentration. For example, if a supply was reported to have one sample with
a VOC present at 0.3 ug/l and another sample in which the VOC was not observed
at a minimum quantifiable concentration of 0.1 ug/l, a supply value of 0.2
ug/l was recorded in the working file. Using the minimum quantifiable concen-
tration with other actual positive values to compute the mean results is the
most conservative estimate of the supply value utilizing all sample data.
The treatment of supplies having only negative values reported derives
from the treatment of those with negatives and positives described above. If,
A-li
-------
for example, a VOC was not observed in two samples from a given system at
minimum quantifiable concentrations of 0.1 ugh and 0.3 ug/l, respectively,
the system value retained in the working file was a “negative” 0.2 ugh. That
is, the VOC was not observed, but if it had been present in both samples, the
maximum possible average concentration for that supply would have been 0.2
ugh.
A.2.3 Combining Data from the Federal Surveys
Once the multiple samples for each VOC were averaged to obtain a single
value for each supply sampled, tables on the frequency of occurrence of each
chemical were prepared for each survey as presented in Sections 2.2.1 and
2.3.1. In addition, the mean, median, range, and other statistics were com-
puted for the positive values in each survey.
The next step in developing the national projections was to combine the
results of all of the surveys together. In doing this, it was necessary to
identify those supplies that had been sampled in more than one survey and
compute an average supply value for each VOC. (The rules for averaging
samples within a survey described in Section A.2.2 applied to computing
averages across surveys.) It should be noted that supplies sampled in the RWS
could not be matched against the other surveys since the RWS locations could
only be determined at the state and county level, as previously described.
Table 1—4 presents a list of those systems which were duplicated across
the Federal surveys. When a system was sampled in two or more surveys, the
population used for that system in the combined survey file was the one
reported in the most recent survey, represented by the following chronological
order (most recent first): GWSS, NSP, CWSS, NOMS, and NORS.
1-12
-------
Table A-4. Systems Sampled in More Than One Survey
Location Source
Tucson, AZ G
Fresno, CA G
Jacksonville, FL G
Idaho Falls, ID G
Rockford, IL G
Campbellsburg, IN G
South Bend, IN G
Baton Rouge, LA G
Hammond, LA G
Lafayette, LA G
Senath, MO G
Webb City, MO G
Greenville, MS G
Kearney, NE G
Lincoln, NE G
Alburquerque, NM G
Baldwinsville, MY G
Dayton, OH G
Aliquippa, PA G
Sioux Falls, SD G
Memphis, TN G
San Antonio, TX G
Provo, UT C
Marysville, WA C
Spokane, WA G
Chippewa Falls, WI G
Madison, WI C
Powell, WY C
Birmingham, AL S
Camden, AR 5
Little Rock, AR S
Phoenix, AZ S
Concord, CA S
Surveys in whichthe sy f&niwas siniplid
GWSS, NORS
GWSS, NOMS
MOMS, NORS
GWSS, MORS
GWSS, MOMS
GWSS, CWSS
GWSS, MSP
NSP, MOMS
GWSS, CWSS
GWSS, CWSS
GWSS, CWSS
GWSS, CWSS
NOMS, NORS
GWSS, CWSS
MOMS, NORS
MOMS, NORS
GWSS, CWSS
NSP, MOMS, NORS
GWSS, CWSS
GWSS, MOMS
MOMS, NORS
GWSS, MOMS, NORS
CWSS, MOMS
GWSS, CWSS
GWSS, NSP, NOMS
GWSS, CWSS
NSP, NOMS
GWSS, CWSS
NSP, MOMS
MOMS, NORS
NSP, NOMS
NSP, NORS
MOMS, NORS
A- 13
-------
Table A-4. Systems Sampled in More Than One Survey
(conti nued)
Waterbury, CT
Washington, DC
Atlanta, GA
Davenport, IA
Chicago, IL
Fort Wayne, IN
Indianapolis, IN
Whiting, IN
Topeka, KS
Louisville, KY
Boston, MA
Lawrence, MA
Baltimore, MD
Portland, ME
Detroit, MI
Grand Rapids, MI
Mount Clemens, MI
St. Paul, MN
Cape Girardeau, MO
Kansas City, MO
St. Louis, MO
Jackson, MS
Charlotte, NC
Bismarck, ND
Surveys in which the
MOMS, NORS
NSP, MOMS
NSP, MOMS
MOMS, NORS
NO?vIS, NORS
NSP, MOMS, NORS
MOMS, NORS
NSP, MOMS
MOMS, NORS
NSP, MOMS, NORS
NSP, MOMS, NORS
NSP, MOMS, NORS
MSP, MOMS, NORS
NSP, MOMS
NSP, MOMS, NORS
MOMS, NORS
NSP, MOMS, NORS
NSP, MOMS
NSP, MOMS, NORS
NSP, NORS
NSP, MOMS, NORS
NSP, MOMS
MSP, MOMS, NORS
NSP, MOMS
MOMS, NORS
MOMS, NORS
MOMS, NORS
NSP, MOMS, NORS
NSP, MORS
NSP, MOMS
NSP, NOMS
CWSS, MOMS
Location
Los Angeles, CA
Oakland, CA
Sacramento, CA
San Diego, CA
San Francisco, CA
Denver, CO
Pueblo, CO
New Haven, CT
system was sampled
Source
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
A- 14
-------
Table A-4. Systems Sampled in More Than One Survey
(conti nued)
Location
Source Surveys in which the
system
was sampled
Omaha, NE
S
NSP, MOMS
Manchester, NH
S
NSP, MOMS
Elizabeth, NJ
S
NSP, NOMS
Passaic, NJ
S
NSP, NOMS,
NORS
Buffalo, NY
S
MOMS, NORS
Poughkeepsie, NY
S
CWSS, MOMS
Cincinnati, OH
S
NSP, NORS
Cleveland, OH
S
NSP, NOMS,
NORS
Columbus, OH
S
NSP, NOMS,
NORS
Toledo, OH
S
NSP, MOMS
Oklahoma City, OK
S
NSP, MOMS,
NORS
Tulsa, OK
S
NSP, MOMS
Corvallis, OR
S
NSP, CWSS,
MOMS,
NORS
Eugene, OR
S
MSP, CWSS,
NOMS
Portland, OR
S
NSP, NOMS
Harrisburg, PA
S
NSP, CWSS
Philadelphia, PA
S
MSP, NORS
Pittsburgh, PA
S
NSP, NORS
Newport, RI
S
MOMS, NORS
Providence, RI
S
NSP, NOMS
Charleston, SC
S
NSP, MOMS,
NORS
Huron, SD
S
MOMS, NORS
Chattanooga, TN
S
NSP, NOMS,
NORS
Nashville, TN
S
NSP, MOMS,
NORS
Brownsville, TX
S
NOMS, NORS
Dallas, TX
S
MOMS, NORS
Fort Worth, TX
S
NSP, MOMS
Houston, TX
S
NSP, NOMS
Salt Lake City, UT
S
NSP, MOMS,
MORS
Annandale, VA
S
NOMS, NORS
Richmond, VA
S
NSP, MOMS
Illwaco, WA
S
MOMS, NORS
A- 15
-------
Table A—4. Systems Sampled in More Than One Survey
(continued)
Location
Source Surveys in which the
system was sampled
Seattle, WA
S
NSP,
NORS
Milwaukee, WI
S
NSP,
NOMS,
NORS
Huntington, WV
S
NSP,
NOMS,
NORS
Wheelings WV
S
MOMS,
NORS
G = Groundwater
S = Surface water
A- 16
-------
A.2.4 Testing for Biases in the Data
There was concern that the selection of sample sites in some of the
surveys was biased toward VOC contamination, which, when combined with other
random survey data, would bias the national projections toward a higher esti-
mated frequency of occurrence and/or mean concentration. Purposeful selection
of sites having a higher than average probability of contamination with VOCs
was in fact the design of the nonrandom portion of the GWSS. It has been
suggested that the NORS, NOMS, and NSP surveys also may have involved a bias
towards systems known or suspected to be contaminated; however, this could not
be confirmed.
Initially, consideration was given to excluding the GWSS nonrandom data
for all VOCs when combining the survey data because the analysis presented by
Westrick et al. (1983) showed that 21.2% of the random supplies had at least
one VOC above its quantitation limit, whereas 27.3% of the nonrandom supplies
showed contamination. The nonrandom portion of the GWSS was reported to have
a higher frequency of occurrence of VOCs at all concentration levels for both
large and small systems. At higher concentrations, there was a two to four
times higher frequency of occurrence of VOCs in the nonrandom sample. It is
important to note that these comparisons are based on the combined results for
all VOCs. There was some concern, however, that sites selected because of
suspected contamination with one or another specific VOC would not necessarily
be biased for other VOCs. Westrick et al. (1983) did not present a comparison
of the random and nonrandom data on a chemical-by-chemical basis, so it was
not clear whether it was appropriate to exclude the GWSS nonrandom data for
each VOC. It was decided, therefore, that an analysis would be performed on
each VOC to evaluate whether there was a statistically significant difference
in the frequency of occurrence and in the mean of the positive values observed
in the random and nonrandoni portions of the GWSS.
To test the difference in frequency of occurrence, the results for each
VOC at each site were classified as positive or negative and summarized in a
two-way table as shown below:
A-17
-------
Portion of
GWSS
Results
Total
Positive
Negative
Random
nil
n 12
n 1
Nonrandorn
‘2i
“22
“2.
Total
n 1
n 2
n
Comparisons of relative frequencies of positive results in the random and
nonrandom segments of the GWSS were based on thex 2 statistic,
2 ll 22 - n 12 n 21 ) 2 n
x “2. .l “.2
with n’s defined in the table above. Under the null hypothesis that the
relative frequencies for the random and nonrandom segments are equal , the test
statistic has an approximate chi-square distribution with 1 degree of freedom;
this distribution was used to compute the P values for each VOC of interest
that was examined in the GWSS.
The mean values of the positive samples in the random and nonrandorn
portions of the GWSS were compared using the “t” test, which tests the null
hypothesis that the two means are the same (i.e., u 1 = u 2 ). The statistic for
testing the equality of two population means u 1 and u 2 using independent
samples from each are as follows:
t = - x 2 ) jJs2 (1/ni + 1/n 2 )
for and n 2 observations with a pooled variance ( 2) of:
= [ (n 1 - 1) s + (n 2 - 1) s ] / (n 1 + n 2 - 2)
where and are the observed sample means of the two groups and s and
are the corresponding variances. An underlying assumption for use of the “t”
test is that the variables are normally and independently distributed in each
group. The normal distribution is not considered an acceptable model for the
values of the positive samples because of the positive skewness and large co-
efficient of variation of their distribution. Therefore, the tests were based
on natural logarithms of the concentrations to make the normality assumption
more reasonable.
A- 18
-------
Table A-5 presents the results of the comparison of the frequency of the
positives arid of the means of the positives for each VOC in the random and
nonrandom portion of the GWSS. The value indicates the significance level
for evaluating the null hypothesis that the means of the random and nonrandom
sample are equal. This value represents the probability that the null hypo-
thesis (i.e., the population means are eqqal) has been rejected on the basis
of the sample means when it is actually true. For example, the results for
benzene shown in Table A—S indicate that there is a 69% probability of being
in error if one were to reject the hypothesis that the population means for
randomly and nonrandomly selected sites are the same based on the sample means
of 4.1 ug/l for the nonrandom and 6.2 ug/l for the random sample observed.
Similarly, the Px2 value represents the level of significance for evaluating
the null hypothesis that the frequency of positives are the same. Again,
referring to benzene, the Px2 value indicates that there is a 16% probability
of being in error by rejecting the hypothesis that the frequency of occurrence
of benzene in samples selected randomly and nonrandomly in the GWSS are the
same based on the observed frequencies of 1.7% for the nonrandorn and 0.7% for
the random samples.
The critical P value selected for this analysis was 0.01, implying the
acceptance of no more than a 1% probability of being in error by rejecting the
null hypothesis of equal means or equal frequencies. Using a critical value
of 0.01, it can be assumed that, for those P values which are less than 0.01,
the null hypothesis should be rejected (i.e., for a null hypothesis u 1 = u 2 ,
the alternative hypothesis, u 1 u 2 , should be c iosen). On the other hand,
for values that are greater than 0.01, it cannot be directly assumed that the
null hypothesis is true. It is only known that the two values are not signi-
ficantly different; it is not known if they are statistically the same.
However, assuming the null hypothesis is testing for equality between two
values, the P values obtained from the test can be used as a general guide to
determine how similar the two values are. A higher significance level (e.g.,
P = 0.70) would denote a greater similarity; a lower level (e.g., P = 0.05)
would denote a greater difference.
In evaluating the null hypothesis based on the P value, the size of the
sample on which the test of significance is made is also important. For small
samples, the null hypothesis is likely to only be rejected if it is very
A-i 9
-------
Table A-5. Results of Comparative Tests for Random and Nonrandom GWSS Data
Number of
Standard
positiv
samples°
Chemical NR/Ra (n)
Percentb
positive
Mean
(x)
(ugh)
deviation
(s)
(ugh)
Range
(ugh) P.tc Px2
Benzene NR 8 1.7 4.1 4.9 0.50-12 0.69 015 d
R 3 0.7 6.2 7.7 0.61-15
Carbon NR 15 3.2 2.2 4.2 0.20-15
tetrachioride R 15 3.3 1.7 4.0 0.20-16 0.54 0.92
Chlorobenzene MR 1 0.2 2.7 033 d
R 0 0 —- --
o-Dichloro— MR 2 0.4 2.5 0.35 2.2—2.7 016 d
benzene R 0 0 —— - - - - ——
m-Dichloro- MR 0 0
benzene R 0 0
p-Dichloro- MR 4 0.9 0.77 0.09 0.70-0.90 0.71 070 d
benzene R 5 1.1 0.75 0.31 0.52-1.3
1,2-Dichioro- NR 7 1.5 3.4 2.9 1.1-9.8 0.01 022 d
ethane R 3 0.7 0.68 0.23 0.53-0.95
1,1-Dichioro- NR 15 3.2 0.59 0.71 0.22-3.0 0.28 0.25
ethylene R 9 2.0 1.4 2.0 0.22—6.3
cis,trans-1,2-Di- MR 38 8.0 8.5 21 0.21—120
chloroethylene R 16 3.5 1.0 0.56 0.21—2.0 0.06 0.003
Tetrachloro- NR 43 9.1 4.7 11 0.22-69 0.04 0.37
ethylene R 34 7.5 1.5 4.0 0.21—23
1,1,1-Trichloro— MR 50 10.6 2.1 3.3 0.20—21 0.17 0.01
ethane R 27 5.9 1.7 3.6 0.20-18
Trichioro- MR 61 12.9 9.0 22 0.20-130 0.90 0.001
ethylene R 30 6.6 8.1 17 0.24-78
Vinyl chloride NR 6 1.3 4.4 2.9 1.4-8.4 0.22 006 d
R 1 0.2 1.1 — - --
aNR = nonrandom; R = random
bBased on 456 random and 473 nonrandom samples from supplies serving 25 or more
people.
CBased on the natural logarithms of the positive concentrations (see text).
dData may be unreliable due to the small number of positive samples (i.e., < 5 for
random and/or nonrandom data).
A-20
-------
wrong. Consequently, a hypothesis of equality may be accepted when it is
wrong because the test for being significantly different at the 0.01 level is
more stringent. In contrast, with a large sample small departures from the
null hypothesis can be determined to be statistically significant, even though
they are, quite unimportant in practice.
As can be seen from the results shown in Table A-5, there is an apparent
general trend of higher frequencies of occurrence and higher means in the
nonrandom samples. However, there are few cases where the differences are
statistically significant at the 0.01 level. With respect to means of posi-
tives, only 1,2—dichioroethane is found to have a statistically significant
difference (actually, borderline) in the means (the nonrandom being higher, as
would be expected). For the comparison of frequency of occurrence, statis-
tically significant differences were found for only three of the VOCs:
cis/trans—1,2-dichloroethylene, 1,1,1-trichloroethane, and trichioroethylene.
Again, the frequencies were higher in the nonrandorn samples as would be
expected. (Note that the small number of positive samples for several of the
VOCs makes the analysis of the frequencies of questionable validity.)
Based on these comparisons, it was decided that in preparing the national
projections, the GWSS nonrandorn data would be included exce t for the four
VOCs found to have statistically significant differences in the mean of the
positives or frequency of positives.
P.2.5 Establishing a Common Minimum Quantifiable Concentration for the
Combined Survey Data
As noted in the discussion on computing averages for VOCs in supplies,
the minimum quantifiable concentration for an analytical technique defines the
level below which it cannot be determined whether the VOC is present and, if
so, at what concentration. If a common minimum quantifiable concentration was
used across all surveys, then the national projections would provide estimates
of the number of systems at various ranges of contamination levels above the
minimum quantifiable concentration. Below that concentration, only a total
number of systems would be estimated for the range of 0 ugh to the maximum
possible concentration, i.e., just below the minimum quantifiable concentra-
tion. The number of systems having no contamination and the distribution of
systems at various levels below the minimum quantifiable concentration could
not be determined.
A-21
-------
In many cases, different minimuni quantifiable concentrations were used in
the different surveys and for different analyses in the same survey. Conse-
quently, some supplies in the combined data set were reported as having no
measured VOC (i.e., negative systems) with a maximum possible concentration
greater than quantified levels reported for other supplies analyzed with more
sensitive methods.
In some cases, it was reasonable to use the highest “negative” value as
the common minimum quantifiable concentration above which national estimates
of the distribution of systems containing various VOC levels are made and
below which only a total number of systems is given. Any positive values
below this level were used as evidence that some supplies in this latter group
are contaminated, i.e., all negative values are not necessarily 0 ug/l. These
positive values were not included with positive values above the established
common minimum quantifiable concentration used for selecting the appropriate
model, etc. for estimating national occurrence.
In several instances, it happened that the selection of the highest
negative value resulted in many, and sometimes most, 0 f the measured positive
values falling below that level, leaving too few positives above that level
for completing the national projections. Three major alternatives were con-
sidered for handling high negative values in these cases. The first alterna-
tive was to establish the common minimum quantifiable concentration at some
lower level that did not exceed substantial numbers of positive data points,
and treat the high negative values as though they were positives at their
maximum possible concentration. This would be generally consistent with the
rules for averaging samples described earlier. It was observed, ho iever, that
in a large number of cases, this would result in a substantial number of the
resulting total positive values being contributed from data where the VOC was
not actually observed; in some cases this contribution would exceed the number
of actual positive values. This was determined to be an improper use of the
data.
The second alternative was to simply discard the high negative data.
This was also considered inappropriate since it meant essentially eliminating
valid data because it could not be made to fit the projection methodology.
Furthermore, its elimination would artificially raise the computed frequency
of positives by lowering the total number of systems sampled.
A-22
-------
The third alternative, and the one selected for the national projections,
was to retain the data, but treat them as though they were negative data below
the lower common minimum quantifiable concentration selected for the combined
survey data. While it is true that VOCs could be present in those supplies at
a higher concentration, the other data suggested that the probability is
greater that they are not. This alternative also avoids the elimination of
valid data from the analysis. (It should be noted that, while this third
alternative is used for the reported national projections, projections were
also calculated using the first alt rnative to give the more conservative
estimate; these results are noted in the text).
A.2.6 Model Selection
In developing methods for estimating the numbers of drinking water
systems and people affected by different pollutant levels, it was necessary to
select a statistical model for the distribution of positive concentrations for
each data set. The following considerations guided the selection of a model:
• The same model should be applicable to data from different system size
groups for a given pollutant (to facilitate comparisons and evaluation
of estimation error).
• A continuous distributional model should be used where appropriate (to
smooth out random variations in relative frequency among concentration
intervals).
• The appropriateness of any continuous model should be checked through
goodness—of-fit tests.
• Estimates from continuous models should provide upper bounds on the
upper tail of the observed distribution (to avoid underestimating the
number of systems with high-level contamination).
Three types of continuous models were investigated:
• Statistical distributions: e.g., the lognormal (Aitchison and Brown
1957) and garmia (Johnson and Kotz 1970) distributions;
• Transformations: the Johnson (1949) system of transformation to
normality; and
• Empirical models: fitting the cumulative frequency with a polynomial
function of concentration.
The adequacy of different distributional models was tested by three goodness-
of-fit tests: Kolmogorov—Sniirnov, Cramer—von Mises, and Anderson-Darling
A-23
-------
(Stephens 1974). A model was considered unsatisfactory if it failed more than
one of these tests at the 0.05 significance level. None of the three types of
models worked consistently for different pollutant size-group data sets.
Of the models investigated, the delta distribution based on the logriormal
distribution (Aitchison and Brown 1957) seemed the most appropriate a
priori . This model has been used successfully in a wide variety of water
contamination problems. It allows for the positive probability of zero (not
detected) results and the skewed distribution of positive results that
generally characterize the drinking water data. However, in many cases this
model failed the goodness-of-fit tests for data from at least one size
group. In other cases there were insufficient positive results to test
whether the model was appropriate.
Based on the evaluations described above, it finally was decided that no
continuous model could be identified that would be useful for the drinking
water data. Therefore, a discrete model had to be employed. - The model
adopted was the multinomial distribution (Johnson and Kotz 1969), in which the
proportion of the distribution in a specified concentration interval is esti-
mated by the observed relative frequency for that interval. The intervals
used were the ones of interest in the evaluation ( 100 ugh).
• Establishing System Size Groupings
Consideration was given to developing estimates from data grouped by
system size because it was thought that contamination might be more likely in
larger systems located in more populous and probably more industrialized
areas. The five system size groups shown below formed the starting point for
grouping data by system size:
Size range
Group ( number of people served )
1 (very small) 1-500
2 (small) 501-3,300
3 (medium) 3,301-10,000
4 (large) 10,001-100,000
5 (very large) > 100,000
*MQC is the Minimum Quantifiable Concentration set for the combined survey
data (see A.2.5, above).
A-24
-------
Frequency tables showing the number of systems sampled and the numbers of
positive and negative results for each of these five size groups were produced
for groundwater and surface water systems for each pollutant. It was con-
sidered desirable to consolidate these groups as much as possible (consistent
with the data) because relatively few systems were sampled in some of the
groups.
The extent of further consolidation possible was evaluated by comparing
the relative frequencies of positives in different groups through a statis-
tical test procedure. Group relative frequencies were compared in the follow-
ing order, and the groups were combined when no significant difference in
relative frequency of positives was found:
1. Groups 1 and 2.
2. Groups 4 and 5.
3. Group 3 with groups 1 and 2 combined.
4. Groups 1, 2, and 3 combined with groups 4 and 5 combined.
Step 3 was done only if groups 1 and 2 could be combined as a result of step
1; step 4 was done only if combinations in previous steps were possible. The
order of the comparisons was chosen based on the possibility of a relationship
(trend) between system size and the percentage of positive systems.
In performing the statistical test for equal percentages of positives in
two size groups, the first step was to form a 2 x 2 summary table as illu-
strated below for groups 1 and 2:
Test
results
Size
group
Nega
tive
Po
sit
ive
Total
1
a
b
a+b
2
c
d
c+d
Total
a+c
b+
d
n
where
a = the number of negative results for group 1.
b = the number of positive results for group 1.
a + b = the number of systems sampled in group 1.
n = a + b + c + d, the number of systems sampled in both groups, etc.
A-25
-------
The test statistic was:
2 - ( lad - bcl - 1/2n) 2 n
X (a + b) (a + c) (b + d) (c + d)
When this statistic exceeded the critical value 3.84 (the 95th percentile of
the chi-squared distribution with 1 degree of freedom), the hypothesis of
equal relative frequencies of positives in the two groups was rejected, and
the groups were not combined. The chi-squared test for homogeneity of per-
centages in two populations is discussed by Snedecor and Cochran (1967).
In some cases, the expected number of positive systems was too small for
the x 2 test to be used (e.g., (a + b)(b + d)/n < 5 for group 1 under the
hypothesis of equal proportions positive). In such cases, the usual alterna-
tive procedure, Fisher’s Exact Test, was used to test equality of proportions
positive. The application of Fisher’s test is described in Ostle (1963).
• Projections of National Occurrence
After the final system size groups were selected for a pollutant, the
national projections were computed. First the proportion of systems in each
concentration interval of interest was estimated for each size group by the
observed relative frequency for the sampled systems. The proportions of
systems above the different specified concentrations also were estimated for
each size group (again by the observed relative frequency for sampled
systems). For example, if 100 systems were sampled in a given size group and
three were found in the range 40-50 ug/l, the estimated percentage of systems
in that range was 3%; if ten out of the 100 systems sampled were above 40
ug/l , the estimated percentage of systems above 40 ugh was 10%.
The number of systems above each concentration limit for a given size
group was estimated by multiplying the observed relative frequency (p ) by the
total number of systems (N,) in that group. Then the total number of systems
of all sizes above a given concentration was estimated by the sum of estimates
for the k individual system size groups:
m = i 1 Ni
A- 26
-------
It can be shown based on the multinomial model that m is an unbiased estimator
of the total number of systems above a specified concentration (M), and that m
has variance
k 2 - .\ P. (1-P.)
Var(m) = i 1 N ( _ ‘ ) n
where is the number of systems sampled out of the N 1 systems in group i and
P 1 is the true percentage of systems above the specified. concentration in
group i. Following Cochran (1963), the term P (l-P 1 )/nj in Var(rn) was re-
placed with its unbiased estimator, (‘ - p 1 )/(n - 1), to estimate Var(m).
Then an approximate 95% confidence interval on M was calculated from
A
in ± 1.96 [ Var(m)] 1/2
The accuracy of this interval improves with increasing sample size (n 1 ) and is
better when P 1 values are not close to zero. When confidence limits obtained
from the approximation were outside feasible M values (i.e., less than zero or
greater than N = Ni) , the limits were reset to the nearest feasible value.
Projections for numbers of people exposed to concentrations above speci-
fied limits were computed in the manner described above, letting N represent
the number of people served by systems in the ith system size group.
A-27
-------
APPENDIX B
System Size Grouping and Selection of Model for Estimating
National Occurrence of i,i—Dichloroethylene in Drinking Water Supplies
As indicated in Appendix A (A.2.6), an initial step in the estimation
process was to group the data based on system size. Based on results of
statistical tests comparing percentages positive in five initial size groups,
groundwater data was collapsed into two groups:
• < 10,000 persons served
• > 10,000 persons served
(Summary statistics for initial and final groundwater groups are given in
Table B-i. Results of the tests are shown in Table 8-3.) For surface water,
data was collapsed into two groups:
• < 10,000 persons served
• > 10,000 persons served.
(Summary statistics and test results are given in Tables B-2 and 8—3, respec-
tively.) Only one surface water system was sampled in groups 1-3; therefore,
these groups were not combined with 4 and 5 because of insufficient data to
test for differences in proportions positive.
For both groundwater and surface water data, the next step was to fit a
delta distribution to each system size grouping with sufficient positive
data. The delta distribution has cumulative distribution function:
1.0 ,x<0
P(X < x) = , x 0
— + (1 - )F(x), x > 0
with
p(lOgex -
F(x) = f(z)dz,
where f(z) is the standard normal probability density function. The mean and
standard deviation of loge_transformed data were used to estimate the para-
B-i
-------
Table B-i. Frequency of Positive Systems by Size Groups
for Groundwater Systems
System size
Number
negative
Number
positive
Number
sampled
Percent
positive
Initial, groups
228
2
230
0.87
(1) 1-500
(2) 501-3,300
199
4
203
1.97
(3) 3,301-10,000
150
4
154
2.60
(4) 10,001-100,000
298
14
312
4.49
(5) > 100,000
38
1
39
2.56
All
913
25
938
2.67
Final groups
(1,2,3) < 10,000
577
10
587
1.70
(4,5) > 10,000
336
15
351
4.27
All
913
25
938
2.67
B—2
-------
Table B-2. Frequency of Positive Systems by Size Groups
for Surface Water Systems
System size
Number
negative
Number
positive
T umber
sampled
Percent
positive
Initial groups
—-
--
0
--
(1) 1 500
(2) 501-3,300
--
—-
0
--
(3) 3,301—10,000
1
0
1
0.00
(4) 10,001-100,000
19
0
19
0.00
(5) > 100,000
81
2
83
2.41
All
101
2
103
1.94
Final groups
(1,2,3) < 10,000
1
0
1
0.00
(4,5) > 10,000
100
2
102
1.96
All
101
2
103
1.94
B- 3
-------
Table B-3. Chi-Squared Tests Comparing
System Size Group Proportions Positive
System Groundwater Surface water
size groups No. of Test No. of Test
compared systems stat. Decisiona systems stat. Decisiona
1 vs. 2 433 b Pass 0 Not done
4 vs. 5 351 b Pass 102 b
1,2 vs. 3 587 b Pass 0 Not done
1,2,3 vs. 4,5 938 4.65 Fail 103 Not done
acritical value for x 2 test (a 0.05) is 3.84.
bExpected number of positives too small for x 2 test. Used Fisher’s Exact Test
(Ostle 1963).
B-4
-------
meters and a; 6 was estimated by the observed proportion of negative
values. The goodness-of-fit of the delta distribution was tested, and the
model was adopted if no more than one of the three tests fafled for each size
group. The results in Table B-4 show that the model failed for groundwater,
and there was insufficient data to test the model for surface water. There-
fore, the rnultinomial model was used as the basis for estimation.
B-5
-------
Table B-4. Goodness-of-Fit Tests for Delta Distribution
1,1-Dichi oroethylene
System type
System size
(population
served)
Parameter
estimatesa
Anderson_Darlingb Cramer—von MisesC Kolmogorov_SmirnOvd
Test
stat.
Test
Decision stat.
Decision
Test
stat.
Decision
N
‘
‘ ‘
‘ ‘
Groundwater
< 10,000
587
0.983
—0.42
4.04
(Insufficient
data for test)
> 10,000
351
0.957
-0.91
4.27
1.010
Fail 0.152
Fail
0.965
Fail
Surface water
< 10,000
1
1.000
--
--
(Insufficient
data for test)
> 10,000
102
0.980
-1.14
1.74
(InsuffIcient
data •for test)
= number of systems sampled;
bcritical value for 0.05 level of
CCritical value for 0.05 level of
value for 0.05 level of
and are estimates of the delta distribution parameters.
significance = 0.787.
significance = 0.126.
significance = 0.895.
-------
APPENDIX C
Calculation of Population—Concentration Values
Chapter 5 presents population-concentration values for intake from air
and drinking water. The population—concentration is calculated as the sum of
the population exposed (P) times concentration (C) at various concentration
levels, or
n
population—concentration = C. x P 1 .
1 =1
For example, if 10,000 persons were exposed to a chemical in drinking water at
5 ug/l and 4,000 additional persons were exposed to the same chemical at 10
ugh, the population-concentration calculated would be:
cn = (10,000 persons x 5 ug/l) + (4,000 persons x 10 ug/l) =
90,000 ug/l x persons.
These calculations would be straightforward if the number of individuals
exposed to specific concentrations of the chemical were known. However, it is
not possible to determine the specific concentration to which each individual
or subpopulation is exposed. Instead, the calculations are based on the
estimate of the number of individuals exposed to concentrations within various
intervals (e.g., >5—10 ug/l). An example of a typical set of estimates of the
population exposed to a chemical in groundwater and surface water systems is
presented in Table C-i.
Table C—i. Total Estimated Population (in Thousands) Exposed to a
Chemical in Drinking Water at the Indicated Concentration Ranges
System type
Votal served
in U.S.
(thousands)
Population (thousands) exposed
concentrations (ug/l ) of:
to
< 0.2 0.2-5
>5-10
>10
Groundwater
69,000
67,000 1,700
73
0.0
Surface water
TOTAL
130,000
200,000
130,000
0.0
73
0.0
0.0
200,000
(% of total)
(100%)
(100%)
(0.04%)
(0.0%)
C-i
-------
The groundwater and surface water data in Table C-i represent the two
types of projections of the populations exposed to concentrations of a chemi-
cal within certain intervals that could be made from the available data. In
the first case, represented by groundwater in Table C-i, it was possible to
estimate the size of the population exposed within several discrete concentra-
tion ranges. In the second case, represented by surface water in Table C-i,
it was only possible to estimate an overall concentration range within which
the entire population is expected to be exposed. As described below, the
approach to the calculation of the population-concentration values differed
for these two cases.
C.1 CALCULATION OF POPULATION-CONCENTRATION ESTIMATES WHERE DATA ARE AVAIL-
ABLE FOR SEVERAL DISCRETE INTERVALS
For this situation, a best case, mean best case, mean worst case, and
worst case estimate were calculated. The best—case is calculated by using the
lower concentration bound for each interval to multiply by the population
exposed within that interval. Using the groundwater data in Table C—i, the
best case is calculated as:
c ?n (best case) = (67,000,000 persons x 0 ug/l) +
(1,700,000 persons x 0.2 ugh) + (73,000 persons x 5.01 ugh) =
710,000 ug/l x persons
For the worst case, the upper limit in each concentration interval is used.
(worst case) = (67,000,000 persons x 0.19 ug/l) +
(1,700,000 persons x 5 ugh) + (73,000 persons x 10 ug/l) =
22,000,000 ug/l x persons
These values give the absolute lowest and absolute highest values which can be
obtained using the data presented.
Two intermediate values are also calculated -— the mean best case and the
mean worst case. These values are calculated by using the mean value within
each concentration interval (i.e., the midpoint of that interval) except for
the lowest interval. For the mean best case, 0 ugh is used for the lowest
interval; for the mean worst case the upper bound of the lowest interval is
C-2
-------
used. The lower (i.e., 0 ugh) and upper bounds of the lowest concentration
interval are used rather than the midpoint value because of the considerable
uncertainty in the distribution of concentrations in the lowest interval given
that most of the sample data for that interval are reported as “undetected.”
Examples of these calculations based on the groundwater data in Table C-i
follow:
c n (mean best case) = (67,000,000 persons x 0 ug/l) +
(1,700,000 persons x 2.6 ug/l) + (73,000 persons x 7.5 ugh) =
5,000,000 ug/l x persons
c n (mean worst case) = (67,000,000 persons x 0.19 ug/l) +
(1,700,000 persons x 2.6 ug/l) + (73,000 persons x 7.5 ug/l) =
18,000,000 ug/l x persons
In some instances, it was projected that a portion of the population was
exposed to concentrations greater than 100 ugfl, with no estimate of the upper
bound to the concentration interval. In those cases, the concentration inter-
vals used were chosen to fit the actual sampling data. For example, if the
data above 100 ug/l for a chemical were 130 and 160 ug/l, the concentration
interval assigned would be > 120—160 ug/l.
C.2 CALCULATION OF POPULATION-CONCENTRATION WHERE DATA ARE AVAILABLE ONLY FOR
AN OVERALL CONCENTRATION RANGE OF EXPOSURE
In cases where only an overall range applicable to the entire population
could be determined, best case, median case, and worst case population-
concentration estimates were calculated as follows. The best case was always
equal to 0 ugh x persons (total population using systems x 0 ugh), while the
worst case was calculated as the total population using the systems times the
upper limit of the concentration range (130,000,000 persons x 5 ug/l =
650,000,000 ugh x persons for in the surface water data in Table C—i). For
the median case, the median of the positive values reported for that chemical
in the water supplies sampled was used, and was multiplied by the total popu-
lation. For the surface water data in Table C-i, the median positive value
was 0.36 ugh, giving a population-concentration estimate of 130,000,000
persons x 0.36 ug/l, or 47,000,000 ugh x persons.
C- 3
-------
C.3 CALCULATION OF TOTAL POPULATION-CONCENTRATION ESTIMATES
Once the population-concentration estimates were calculated for each
chemical for both groundwater and surface water systems, these values were
added to obtain total population—concentration estimates for each chemical.
In cases where data were available on discrete intervals for one water source
and for an overall interval for the other source, it was necessary to add
separately the mean best case and mean worst case estimates for the former to
median case estimates for the latter. This procedure was necessary to obtain
mid-range estimates of population-concentration for the chemicals.
C-4
------- |