BASELINE HUMAN HEALTH RISK ASSESSMENT
VASQUEZ BOULEVARD AND 1-70 SUPERFUND SITE
DENVER, CO
August, 2001
Produced by:
US Environmental Protection Agency, Region VIII
999 18th Street, Suite 500
Denver CO 80202
With technical assistance from:
Syracuse Research Corporation.
999 18th Street, Suite 1975
Denver Co 80202
-------
APPROVAL PAGE
This risk assessment has been prepared in accord with USEPA national and regional guidelines and is
approved for release without condition.
USEPA RemedialkProject Manager
Bonita Lavelle
Office of Ecosystems Protection and Remediation
Date /
Technical Approval
Christopher Weis, PhD, DABT
USEPA Regional Toxicologist
Office of Ecosystems Protection and Remediation
Date
-------
TABLE OF CONTENTS
EXECUTIVE SUMMARY ES-1
SECTION 1
INTRODUCTION 1
1.1 SITE DESCRIPTION 1
1.2 BASIS FOR POTENTIAL CONCERN 1
1.3 PURPOSE AND SCOPE OF THIS DOCUMENT 3
1.4 ORGANIZATION OF THIS DOCUMENT 3
SECTION 2
SUMMARY OF SITE DATA AND
SELECTION OF CHEMICALS OF POTENTIAL CONCERN 5
2.1 PHASE I/PHASE II GRAB SAMPLE INVESTIGATION 5
2.2 PHASE II CONFIRMATORY SAMPLING AND SOIL REMOVAL 7
2.3 RISK-BASED SAMPLING PROGRAM 7
2.3.1 Spatial Patterns of Contamination 7
2.3.2 Contaminant Levels in Other Environmental Media 8
2.3.3 Phase II Biomonitoring 12
2.4 PHYSICAL-CHEMICAL CHARACTERIZATION 12
2.4.1 Concentration in Sieved and Un-Sieved Soil Samples 12
2.4.2 Speciation of Arsenic and Lead 14
2.5 SELECTION OF CHEMICALS OF POTENTIAL CONCERN 18
2.5.1 Data Used to Select COPCs 18
2.5.2 COPC Selection Process 20
2.5.3 Summary: Chemicals Selected as COPCs at VBI70 23
2.6 PHASE III INVESTIGATION 23
2.6.1 Residential Soil Sampling 25
2.6.2 Residential Dust Sampling 29
2.6.3 Residential Garden Sampling 31
2.6.4 Sampling at Schools and Parks 33
2.6.5 Phase III Biomonitoring Program 36
2.7 DATA SELECTED FOR USE IN THIS RISK ASSESSMENT 36
SECTION 3
EXPOSURE ASSESSMENT 38
3.1 CONCEPTUAL SITE MODEL 38
3.1.1 Potential Sources 38
3.1.2 Migration Pathways 40
3.1.3 Exposed Populations and Potential Exposure Scenarios 40
3.2 PATHWAY SCREENING 40
3.2.1 Residential Exposures 40
ii
-------
3.2.2 Workplace Exposures 42
3.2.3 Exposures at Schools and Parks 42
3.3 SUMMARY OF PATHWAYS OF PRINCIPAL CONCERN 43
SECTION 4
QUANTIFICATION OF EXPOSURE AND RISK FROM ARSENIC 44
4.1 OVERVIEW 44
4.2 QUANTIFICATION OF EXPOSURE 44
4.2.1 Basic Equation 44
4.2.2 Variability and Uncertainty in Exposure Calculations 46
4.2.3 Derivation of the Concentration Term 46
4.2.4 Source of Exposure Parameters 47
4.2.5 Quantification of Exposure of Residents to Soil 49
4.2.5.1 Long-Term (Chronic and Lifetime) Exposure 49
4.2.5.2 Sub-Chronic Exposure 52
4.2.5.3 Acute Pica Exposure 56
4.2.6 Quantification of Exposure of Residents to Home-Grown Vegetables . . 58
4.3.1 Overview 62
4.3.2 Toxicity Summary for Arsenic 64
4.3.3 Adjustments For Relative Bioavailability 67
4.4 RISK CHARACTERIZATION FOR ARSENIC 67
4.4.1 Basic Approach 67
4.4.2 Risks from Soil and Dust 69
4.4.2.1 Cancer Risk 69
4.4.2.2 Chronic Noncancer Risks 72
4.4.2.3 Subchronic Noncancer Risks 72
4.4.2.4 Noncancer Risks from Acute Pica Behavior 72
4.4.3 Risks from Home-Grown Vegetables 75
4.4.4 Combined Risks from Soil and Home-Grown Vegetables 79
4.5 UNCERTAINTIES IN ARSENIC RISK ASSESSMENT 80
SECTION 5
EXPOSURE AND RISK FROM LEAD 89
5.1 OVERVIEW 89
5.2 IEUBK MODEL FOR ASSESSING LEAD RISK 89
5.3 RISK CHARACTERIZATION FOR LEAD 91
5.3.1 Risks from Lead in Soil and Dust 91
5.3.2 Risks from Lead in Garden Vegetables 94
5.4 UNCERTAINTIES IN LEAD RISK EVALUATION 95
SECTION 6
REFERENCES 107
in
-------
LIST OF APPENDICES
Appendix Title
A Garden Vegetable and Soil Data
B Screening Level Evaluation of Relative Risk from Arsenic via Inhalation of Dust
or Dermal Contact with Soil Compared to Soil Ingestion
C Risk-Based Concentration Values for Workers
D Monte Carlo Calculations of Cancer Risk
E Detailed Data and Calculations (electronic file)
IV
-------
LIST OF TABLES
Table 2-1 Biomonitoring Data for Residents at Phase II Removal Properties 13
Table 2-2 Data Used to Select Chemicals of Potential Concern 19
Table 2-3 Comparison of Maximum Values in Soil to Soil Screening Levels 21
Table 2-4 Comparison of Past and Present Data for Thallium in Soil 24
Table 2-5 Property Mean Summary Statistics for Phase III Soil Samples 26
Table 2-6 Phase III Soil Data for Schools and Parks 35
Table 2-7 Biomonitoring Data for Residents at Phase III Removal Properties 37
Table 4-1 Estimated Cancer Risk from Arsenic in Soil and Dust 70
Table 4-2 Estimated Chronic Noncancer Risk from Arsenic in Soil and Dust 73
Table 4-3 Estimated Subchronic Noncancer Risks from Arsenic in Soil 74
Table 4-4 Estimated Acute Noncancer Risks from Pica Behavior 76
Table 4-5 Estimated Cancer and Noncancer Risk from Arsenic in Garden Vegetables 77
Table 4-6 Estimated Total Cancer Risks from Soil and Vegetables 81
Table 5-1 IEUBK Model Inputs 92
Table 5-2 Estimated Risks to Children from Lead in Soils and Dust 93
Table 5-3 ISE Model Inputs 98
Table 5-4 Comparison of State Blood Lead Data to National Statistics 104
LIST OF FIGURES
Figure 1-1 Site Map 2
Figure 2-1 Phase I/Phase II Soil Grab Sample Data 6
Figure 2-2 Spatial Distribution of Contaminants - Property 1 9
Figure 2-3 Spatial Distribution of Contaminants - Property 2 10
Figure 2-4 Comparison of Concentration in Bulk and Fine Soil 15
Figure 2-5 Chemical Forms of Arsenic in Site Soils 16
Figure 2-6 Chemical Forms of Lead in Site Soils 17
Figure 2-7 Distribution of Property Mean Concentrations in Bulk Soils 27
Figure 2-8 Correlation between Lead and Arsenic 28
Figure 2-9 Relation between Concentration in Indoor Dust and Bulk Yard Soil 30
Figure 2-10 Relation between Total Arsenic in Garden Vegetables and Garden Soil 32
Figure 2-11 Relation between Contaminants in Garden Soil and Yard Soil 34
Figure 3-1 Conceptual Site Model for Operable Unit 1 39
Figure 4-1 Comparison of the UCL based on Composites to the Mean of Grab Samples 48
Figure 4-2 Coefficient of Variation in Yard Soil Grab Samples 54
Figure 4-3 Distribution of Arsenic Values in Phase III Soils 71
Figure 5-1 State Blood Lead Analysis Results 106
-------
LIST OF ACRONYMS AND ABBREVIATIONS
CDC Centers for Disease Control
CDPHE Colorado Department of Public Health and Environment
cm centimeter
COPC chemical of potential concern
CTE central tendency exposure
DI daily intake
DL detection limit
dL deciliter (0.1 liter)
dw dry weight
EC01 concentration in water that results in a 1% increase in excess lifetime cancer risk
EMPA electron microprobe analysis
EPC exposure point concentration
g gram
GFAA graphite furnace atomic absorption
GSD geometric standard deviation
HIF human intake factor
HQ hazard quotient
TCP inductively coupled plasma spectroscopy
m meter
m3 cubic meter
MS mass spectrometry
IEUBK Integrated Exposure, Uptake, and Biokinetic Model
IRIS Integrated Risk Information System
ISE Integrated Stochastic Exposure Model
kg kilogram
LOAEL lowest observed adverse effect level
mg milligram
ng nanogram
NOAEL no observed adverse effect level
NPL National Priorities List
PbB blood lead level
PDF probability density function
ppm parts per million
RBA relative bioavailability
RBC risk-based concentration
RfD reference dose
RME reasonable maximum exposure
SF slope factor
TAL Target Analyte List
UCL upper confidence limit
ug microgram
USEPA U.S. Environmental Protection Agency
VBI70 Vasquez Boulevard and 1-70 Site
ww wet weight
XRF X-ray fluorescence
VI
-------
EXECUTIVE SUMMARY
1.0 BACKGROUND
1.1 Site Description
The Vasquez Boulevard and 1-70 (VBI70) Superfund Site is an area of approximately four
square miles located in the north-central section of Denver, Colorado. The site is composed of a
number of neighborhoods that are largely residential, including Swansea, Elyria, Clayton, Cole,
and portions of Globeville. Most residences at the site are single family dwellings, but there are
also some multi-family homes and apartment buildings. The site also contains a number of
schools, parks, and playgrounds, as well as a number of commercial and industrial properties.
Figure ES-1 is a map which displays the site.
1.2 Basis For Potential Concern
The site came to the attention of the U.S. Environmental Protection Agency (USEPA) because
studies directed by the Colorado Department of Public Health and Environment (CDPHE) at a
nearby site (Globe Smelter) indicated that elevated concentrations of arsenic and/or lead
occurred in the soil of some residential properties in the Swansea/Elyria area. The source of
these elevated levels is not known, but a priori, it is considered plausible that the contamination
is associated with releases either from the Globe facility and/or from one or both of two other
smelters which previously existed in the area (the Argo Smelter and the Omaha and Grant
Smelter). The locations of these three smelters in relation to the VBI70 site are also shown in
Figure ES-1. Alternative potential sources include the historic application of arsenic- or lead-
containing lawn care products, and/or (for lead) anthropogenic sources such as automobile
exhaust, leaded paint, etc.
Based on the results of several rounds of soil sampling, USEPA concluded that the VBI70 site
contained multiple residences where the concentration of arsenic and/or lead in yard soil could
be above a level of potential human health concern. On this basis, USEPA proposed the VBI70
site for inclusion on the Superfund National Priorities List (NPL) in January, 1999, and the site
was added to the NPL on July 22, 1999.
The process of evaluating the nature and extent of environmental contamination at the site and of
estimating the potential risks to human and ecological receptors has been divided into several
sub-projects, or "Operable Units". This risk assessment focuses on risks associated with soil
contamination in residential areas of the site. This is referred to as Operable Unit 1 (OU1).
ES-1
-------
Figure ES-1 Site Map
N ^F^
MIT* '
...JPrs
««Q5
3GT ^
44Q4
• IMF. •:;:•-
Omaha and Grant
Smelter Site
Martn Luffiei King Btvd
)FEET
3RTH)
44Q2
Location of VB.170 Site
City and County of Denver
^^ Washington
V,.*r"
Legend
Study Area Boundary
Neighborhood
I I CLAYTON
I I COLE
I I ELYRIA
I I GLOBEVILLE
I I SWANSEA
Map Base:
USGS7.5' Quadrangle
Commerce City
0 500 1,000 2,000 3,000
Feet
300 150 0 300
^^^^^H
Meters
600
Vasquez Boulevard /1-70
Operable Unit 1
Figure ES-1
Site Location
Project No: RAC 6S-W7-0039 WA 004-RICO-089R
ES-2
-------
2.0 SUMMARY OF SITE DATA AND SELECTION OF CHEMICALS OF
POTENTIAL CONCERN
2.1 Initial Studies
Phase I/Phase II
Once investigations at the nearby Globe site began to suggest that elevated levels of arsenic
and/or lead might exist in soils at residential properties within the area of the VBI70 site,
CDPHE requested assistance from USEPA Region VIII in characterizing the nature and extent of
the contamination. In response, USEPA Region VIII undertook a study designed to identify
properties that had levels of arsenic or lead that were sufficiently high that time-critical action
(soil removal and replacement) might be warranted. Most of these samples were collected
during the initial round of sampling (referred to as Phase I), with the remainder being obtained in
a subsequent sampling effort (Phase II). In the majority of cases, two surface samples and one
subsurface sample were collected per property, with additional surface samples at some locations
(depending on the size of the property).
The action levels selected for time-critical soil removal were 450 ppm for arsenic and 2,000 ppm
for lead. For arsenic, a majority of properties sampled (927 out of 1390) had maximum values
that were below the limit of detection (average detection limit in Phase I/II = 51 ppm).
However, arsenic was detected in one or more surface soil samples at a number of properties,
with 40 of these properties having one or more samples above 450 ppm. For lead, most
properties (1,153 out of 1,390) had concentration values in surface soil that were below 400
ppm, but 238 properties had one or more values above 400 ppm. Of these, six properties had one
or more lead value above 2,000 ppm.
In order to help confirm the identity of properties which warranted time-critical soil removal
actions, USEPA collected two or more composite samples (each consisting of five sub-samples)
of surface soil from residential properties where one or more grab samples were above the
removal level for arsenic. Based on the results of this composite sampling program, a total of 21
residences were identified where one or more composites confirmed that arsenic levels were
above the action level. Using the authority provided under CERCLA 104, EPA performed soil
removal and replacement at 18 of these properties in the fall of 1998. The owners of the other
three properties refused permission for the removal. No properties were identified where lead
levels in composite soil samples were high enough to warrant a time-critical soil removal action.
Risk-Based Sampling Program
One of the striking findings that emerged from the Phase I/Phase II sampling programs was that
arsenic-affected properties did not appear to occur in a clear spatial pattern. That is, the
occurrence of high arsenic levels in soil did not appear to be associated with proximity to one or
more of the current or historic smelters, and properties with elevated levels of arsenic often
occurred immediately adjacent to one or more residences that were not apparently affected.
ES-2
-------
In order to obtain additional information on the spatial pattern of contamination both within and
between yards, USEPA selected eight properties to undergo detailed soil sampling. Five of the
yards were locations where Phase I/Phase II sampling indicated the arsenic concentrations were
above the removal level, while three of the properties had arsenic concentrations below the
removal level. At each property, a high-density grid was established on 5-foot centers, and soil
samples were collected wherever the grid node did not fall on a driveway, patio, etc. In addition,
whenever access could be obtained, the sampling grid was extended 10-15 feet into adjacent
properties in order to determine if there was a clear difference in contamination levels between
adjacent properties.
The results for one property are shown in Figure ES-2. As seen at this location, there is a fairly
clear boundary between the property of concern and the adjacent properties. Similar patterns are
observed at other properties, although there are some locations where the contamination may
extend somewhat into the adjacent property.
Other activities conducted under the Risk-Based Sampling Program included collection of a
number of environmental samples (dust, water, paint, vegetables) at the eighteen properties
selected for soil removal. Arsenic and lead levels in indoor dust were found to have no apparent
relationship to levels in yard soil, suggesting that soil was not a predominant source of
contaminant levels in indoor dust. Lead levels in tap water were all below the current USEPA
action level for lead in drinking water (15 ug/L), suggesting that tap water is not likely to be a
significant source of exposure. Lead was detected in paint at most locations, with 130 out of 144
samples having values above 1 mg/cm2. These data suggest that interior and/or exterior leaded
paint might be a source of lead exposure in area children, either directly (by paint chip
ingestion), or indirectly (by ingestion of dust or soil containing paint chips). Only one of the 18
properties scheduled for soil removal had a vegetable garden. At this location, concentrations of
arsenic and lead were below the level of detection in two vegetable samples. Because so few
vegetable samples were obtained, no conclusions can be drawn from this data set.
In addition to environmental sampling, a number of biological samples (hair, urine, blood) were
also collected from residents in the properties selected for soil removal. A total of 15 individuals
residing at six of the properties scheduled for soil removal volunteered to participate in the
program. None of the samples collected exceeded the normal range for lead or arsenic.
Although this data set is too small to draw firm conclusions, the results provide no indication
that exposures at these locations were of immediate health concern.
Physical- Chemical Characterization
USEPA also undertook two studies to characterize the physical chemical attributes of the lead
and arsenic contamination in residential site soils. These studies found that arsenic in site soils
occurs mainly as arsenic trioxide, with a smaller but significant contribution from lead arsenic
predominant form accounting for elevated lead levels in yard soils. Levels of lead phosphate and
lead manganese oxide also tend to increase as total lead concentrations increase, but these phases
may be secondary weathering products derived from the lead arsenic oxide.
ES-4
-------
Figure ES-2 Spatial Distribution of Contaminants-Property 1
Surface Soil - Arsenic (ppm)
t
iAs(<=70) • As (71-150) As (151-450) As (451-1000) • As (>1000)
Scale isapprox'mate
Surface Soil - Lead (ppm)
t
Pb(<=400) »Pb (401-1000) Pb (1001-1500) Pb (1500-2000) »Pb(>2000)
Scale is approximate
ES-5
-------
In addition, the concentration of metals in bulk (unsieved) soil samples were compared to that in
fine (sieved) samples. The slope of the best fit regression line through the paired data set was
close to 1.0 for zinc, but was slightly higher for arsenic (slope = 1.21), lead (slope = 1.09) and
cadmium (slope = 1.13). In all cases, these slopes were statistically different from 1.0 (p <
0.001). This indicates that the concentration of arsenic, lead and cadmium is about 10-20%
higher in fines than in bulk samples of soil.
2.2 Selection of Chemicals of Potential Concern
Chemicals of potential concern (COPCs) are chemicals which a) are present at a site, b) occur at
concentrations which are or might be of health concern to exposed humans, and c) are or might
be due to releases from a Superfund site. USEPA assumes that any chemical detected at a site is
a candidate for selection as a COPC, but identifies a number of methods that may be used for
determining when a chemical is not of concern and may be eliminated from further
consideration. Each risk assessment may choose to apply some or all of the methods identified
by USEPA to select COPCs, as appropriate.
At this site, COPCs were selected based on available data from full-suite analyses of soil
samples for the 23 metals included on USEPA's Target Analyte List (TAL). In accord with
standard methods identified in USEPA risk assessment guidance, chemicals were eliminated if:
a) the maximum value was below a level of health concern, b) the chemical is a beneficial
mineral that is required for good health, and c) if the risk contributed is minor compared to other
chemicals that will be retained. Based on these selection procedures, the COPCs selected for
quantitative evaluation at the VBI70 site are arsenic and lead. All other chemicals measured in
soil are either not of concern or are present at levels which contribute minimal risk compared to
arsenic and lead.
2.3 Phase III Investigation
Because of the absence of any clear spatial pattern of soil contamination, USEPA concluded that
the identity and location of properties with elevated levels of arsenic and/or lead could not be
reliably predicted using traditional approaches, and that sampling of every yard was necessary.
For this reason, USEPA undertook a large-scale sampling program designed to obtain data that
would help evaluate health risks to residents in the area. This program is referred to as the Phase
III investigation. The investigation consisted of four main parts:
Sampling of residential yard soils
Sampling of indoor dust at residences
Sampling of residential vegetable gardens (vegetables and soil)
Supplemental sampling of soil at local schools and parks
Phase III was implemented in two parts. The first part, referred to as Phase Ilia, focused mainly
on properties (including residences, schools, and parks) which had not been investigated in
Phases I or II, including a large portion of both the Cole and Clayton neighborhoods. The
ES-6
-------
second part, referred to as Phase Illb, consisted of re-sampling at properties that had previously
been sampled in Phase I or II, but for which the data were judged to be too limited to support
clear risk-management decision making. This risk assessment is based on the combined data
from Phase Ilia and Illb.
Residential Soil Sampling
A total of 30 surface soil (0-2 in.) grab samples were collected from each property where access
was granted. These 30 samples were combined into three composite samples, each containing
10 grab samples. The composites were prepared by combining every third grab sample, such
that each composite represents an independent estimate of the yard-wide mean concentration.
All composite samples were dried and mixed, and then analyzed for arsenic and lead by XRF.
The total number of properties sampled in Phase III was 2,986. Summary statistics, based on
average values at each property and stratified by neighborhood, are summarized in Table ES-1.
For arsenic, most properties (2,471 out of 2,986 = 83%) have average concentrations of 50 ppm
or less, with 258 properties (9%) between 50-100 ppm, 183 (6%) between 100-200 ppm, and 74
(2%) above 200 ppm. For lead, 2,712 (91%) properties have mean lead concentrations lower
than 400 ppm, with 266 (9%) between 400-800 ppm and 8 (0.3%) higher than 800 ppm. There is
only a weak correlation between the occurrence of elevated lead and elevated arsenic in soil,
suggesting that the main sources of lead and the main sources of arsenic in yard soil are not
likely to be the same.
Residential Dust Sampling
In accord with the initial results obtained during the Risk-Based sampling program, only a weak
correlation was detected between the level of either arsenic or lead in paired soil and dust
samples (R2 = 0.14 to 0.18, respectively). Nevertheless, the slopes of both regression lines are
statistically different from zero (p < 0.01), with best estimate parameter values as follows:
Arsenic: Cdust = 0.06-Csoil + 11
Lead: Cdust = 0.34-Csoil + 150
These equations were used to estimate the concentration of arsenic and lead in dust at each
property based on the measured values in soil.
USEPA collected 72 samples of different types of garden vegetables from 19 different properties
around the site. Each vegetable sample was washed in de-ionized water to minimize the amount
of adhering soil. Vegetables were not peeled before analysis. At each location where a
vegetable sample was collected, a co-located sample of garden soil was also collected.
ES-7
-------
Table ES-1 Property Mean Summary Statistics for Phase III Soil Samples
Residential Garden Sampling
ARSENIC
Neighborhood
Clayton
Cole
Elyria
Globeville
Swansea
ALL
Total
Properties
902
796
59
63
1166
2986
Distribution of Yard Average Concentration Values for Arsenic (ppm) (a)
5th
5.5
5.5
5.5
5.5
5.5
5.5
25th
5.5
7.7
8.5
8.5
5.5
5.5
50th
8.7
11.8
12.3
13.8
9.7
10.5
75th
38.3
24.8
22.3
22.3
30.6
30.3
95th
168.0
142.1
97.2
123.3
128.3
144.9
Maximum
758
660
431
297
604
758
LEAD
Neighborhood
Clayton
Cole
Elyria
Globeville
Swansea
ALL
Total
Properties
902
796
59
63
1166
2986
Distribution of Yard Average Concentration Values for Lead (ppm) (a)
5th
76
135
181
171
76
81
25th
106
221
299
257
119
127
50th
140
288
372
332
164
188
75th
193
371
438
482
250
292
95th
337
538
601
633
410
465
Maximum
1131
1130
922
835
776
1131
(a) Yard average is the mean of composites collected from the yard
ES-8
-------
For arsenic, the mean concentration in vegetables (averaged across all samples) was 0.043 mg/kg
wet weight (43 ng/g ww). One data point (an onion sample from property 6) appears to be
substantially higher than expected based on the other samples. The basis for this apparently high
value is not known, but might be attributable to incomplete removal of soil from the sample prior
to analysis. If that sample is considered to be an outlier and is excluded, then the mean
concentration of arsenic in vegetables is 30 ng/g wet weight. The slope of the best-fit regression
line through the data (outlier excluded) is quite low (0.0014 mg/kg wet weight per mg/kg in
soil), but the slope is statistically different from zero (p <0.001).
For lead, the mean concentration across all samples was 0.15 mg/kg wet weight (150 ng/g ww).
Again, one data point (a garlic sample from property 11) appears to be substantially higher than
expected based on the other samples. If that sample is considered to be an outlier and is
excluded, then the mean concentration of lead in vegetables is 62 ng/g wet weight. The slope of
the best-fit regression line through the data (outlier excluded) is not statistically different from
zero (p > 0.5).
There is only a weak relationship between the concentration of arsenic in yard soil and in garden
soil (slope = 0.066, R2 = 0.265), although the slope is statistically different from zero (p < 0.01).
For lead, both the slope (0.60) and the correlation (R2 = 0.410) are somewhat higher that for
arsenic, but the correlation is still rather weak. These results indicate that garden soil is not
equivalent to yard soil, with levels of arsenic and lead tending to be lower in the gardens than in
the yard. This might be because the garden soil is prepared by amending yard soil with clean
soil, peat moss, or other additives that dilute the yard soil contaminant level, or because the
source(s) that have affected the yard did not equally affect the gardens.
Sampling at Schools and Parks
Samples of surface soil were collected at 10 schools and one park. Concentrations of arsenic are
generally low, with average values ranging from 11-14 ppm, and maximum values less than 25
ppm. An exception to this pattern occurred at one school property where two values
significantly higher than expected were detected (1,517 ppm and 70 ppm). These values
occurred adjacent to each other, and were surrounded by values of 17-23 ppm, indicating the
presence of a small "hot spot". Even though no children were exposed at this area, EPA Region
VIII has worked with the property owner to address this area of contamination.
2.4 Data Selected For Use in This Risk Assessment
The data from the Phase III sampling program were selected for use in this risk assessment
because 1) all Phase III data were collected in accordance with project plans that were developed
with careful consideration of the Data Quality Objectives (DQOs) needed to support risk
assessment calculations, and 2) all data collected during Phase III are accompanied by Quality
Assurance (QA) samples that allow detailed evaluation of the reliability of the data. These
ES-9
-------
quality assurance data (USEPA 2000e) reveal that the data collected are of high quality, with
adequate accuracy and precision to support a reliable evaluation of human health risk.
Data collected during Phase I/Phase II were not used because they were collected only with the
intent of identifying locations that exceeded the removal action levels, and were not intended to
support detailed risk calculations or remedial decision making. More specifically, data from
Phase I/Phase II were not used because 1) many samples had elevated detection limits for arsenic
(average = 51 ppm, range = 44 to 800 ppm), 2) the sampling density at each property was
sometimes too low to ensure representativeness, and/or 3) exact sampling locations within a
property were not always clear. However, despite these limitations, it is clear that the data from
Phase I/Phase II and from Phase III are generally similar, each indicating the occurrence of
scattered properties with elevated levels of lead and/or arsenic.
3.0 EXPOSURE ASSESSMENT
Figure ES-3 presents a conceptual model showing the main pathways by which contaminants
present in surface soil may come into contact with area residents. This conceptual model was
developed in consultation with local community groups as well as representatives from the City
and County of Denver, the Colorado Department of Public Health and Environment, and the
Agency for Toxic Substances and Disease Registry. Exposure scenarios that are considered
most likely to be of concern are shown by boxes containing a solid circle, and greatest attention
is focused on these pathways. Pathways which are judged to contribute only occasional and
minor exposures are shown by boxes with an open circle. Incomplete pathways (i.e., those
which are not thought to occur) are shown by open boxes. Based on this conceptual model, the
following pathways are judged to be of sufficient potential concern to warrant quantitative
exposure and risk analysis for this Operable Unit (OU1):
Exposure Pathways of Potential Concern for Quantitative Risk Analysis
Population
Resident
Medium and Exposure Route
Incidental ingestion of soil and dust in and about the
home and yard
Ingestion of home-grown vegetables
Other exposure pathways are judged to be sufficiently minor that further quantitative evaluation
i
-------
Figure ES-3 Conceptual Site Model for Operable Unit 1
Exposure to Off-Facility Soils
Revision 2
/IN 1/\1V111N/\1N 1
SOURCE
Products
^ Stack
1 • T- • • ~
r,,rr,ntnr V^ EllliSMOilS
Historic — |
L^- Solid Wastes -
| | = Pathway is not complete
| (~) | = Pathway is or might be
complete, but is judged to
minor; qualitative evaluat
iKAiNsrurvi r/\inw/\i a /VLNIJ EXPOSURE
CONTAMINATED MEDIA ROUTE
Wind .
Deposition
W
Bui
Tra
Disposal w
be
ion
Off-Facility
Soils (OU1)
>
ind,
kSoil
nsport
k,
On-Facility
Soils
^ VpprfnhlRs ^ inec;;iion |
- uirect Contact Ingestion
^~ Dermal ^—t
. ,, ^I^Deposition
Adherence to T Tn action !
kunoii Surface Water ^ Ingestion :
^~ and Sediments ~ Dermal ^—t
Leaching Ingestion ;
^ ^ Dermal ;
*
To be evaluated as a separate operable unit.
EXPOSED
POPULATION
RESIDENT WORKER
^1 •
^ •
> O
o
o
> o
o
^. 0
> o
> o
0
0
0
*> o
> o
o
o
> o
> o
o
o
| ^ | = Pathway is or might be complete
and could be significant;
quantitative evaluation
ES-11
-------
4.0 QUANTIFICATION OF EXPOSURE AND RISK FROM ARSENIC
4.1 Quantification of Exposure
It is expected that different individuals who live in the VBI70 site will have a range of different
exposure levels to arsenic. This is because they have different intake rates of soil, dust and
vegetables, and live in areas of differing arsenic concentration. The risk assessment estimated
the exposure for two different types of resident: a resident with average exposure, and one at the
high end of the exposure distribution. These two cases are referred to as Central Tendency
Exposure (CTE) and Reasonable Maximum Exposure (RME). Estimates of exposure for the
CTE and RME cases were calculated for three different exposure scenarios: long-term
(chronic/lifetime), short-term (subchronic), and acute (pica). Standard exposure equations
identified in USEPA risk assessment guidance were used in all cases. When applicable, EPA
defaults were used for exposure parameter input values. In accord with Agency guidelines,
when reliable site-specific exposure data were available, these data were used in place of default
exposure assumptions. All concentration values in soil, dust and garden vegetables were based
on site-specific measurements.
4.2 Toxicity Assessment
The toxic effects of arsenic have been reasonably well established, based mainly on studies of
humans exposed to elevated levels of arsenic from a variety of sources. The main effects are
summarized below.
Acute Noncancer Effects
Very high doses of arsenic may cause acute lethality, but such exposures from environmental
sources are very unlikely. Oral exposure to non-lethal but high acute doses of arsenic produces
marked irritation of the gastrointestinal tract, leading to nausea and vomiting. Other signs may
include neuritis and vascular effects.
Subchronic Noncancer Effects
Symptoms resulting from sub-chronic ingestion of lower doses of arsenic often begin with a
vague weakness and nausea. As exposure continues, symptoms become more characteristic and
may include signs such as diarrhea, vomiting, anemia, injury to blood vessels, damage to kidney
and liver, and impaired nerve function that leads to "pins and needles" sensations in the hands
and feet.
Chronic Noncancer Effects
Chronic exposure to arsenic is associated with all of the effects noted above. In addition, after
exposure continues for a sufficient period of time, an unusual pattern of skin abnormalities,
including dark and white spots and a pattern of small "corns" may occur, especially on the palms
and soles.
ES-12
-------
Carcinogenic Effects
There is strong evidence from a number of human studies that oral exposure to arsenic increases
the risk of skin cancer. The most common type of cancer is squamous cell carcinoma, which
appears to develop from some skin corns. In addition, basal cell carcinoma may also occur,
typically arising from cells not associated with the corns. Although these cancers may be easily
removed, they can be painful and disfiguring and can be fatal if left untreated. More recent data
indicate that chronic oral arsenic exposure also increases the risk of several types of internal
cancer, including cancer of the bladder and lung.
Toxicity Factors for Arsenic
Based on the available toxicity data for arsenic, the USEPA has established a series of Reference
Doses (RfDs) for evaluating risk of non-cancer effects, and a cancer slope factor for quantifying
the risk of cancer. These values are summarized below.
USEPA Arsenic Toxicity Factors Utilized in the Risk Assessment
Toxicity Factor
Acute RfD
Subchronic RfD
Chronic RfD
Oral Slope Factor
Value
0.015 mg/kg-day
0.006 mg/kg-day
0.0003 mg/kg-day
1.5 (mg/kg-day)-1
Source
USEPA 200 If
USEPA 1995b
IRIS 2000
IRIS 2000
Because the oral RfD values and the oral SF for arsenic are based on studies of humans exposed
to arsenic either in drinking water or in other readily absorbable forms, solid forms of arsenic in
site soils may be less well-absorbed and require adjustments in the toxicity factors to derive
appropriate estimates of toxicity. In order to investigate the relative bioavailability (RBA) of
arsenic in site soils, USEPA performed a study in which five separate samples were fed to swine
for 12 days. The study found that arsenic in site soils was less well absorbed than a readily
soluble form of arsenic (sodium arsenate), with RBA values for individual samples ranging from
about 0.18 to 0.45, with a mean value of 0.31 for all site samples. In order to be conservative,
exposure calculations were based on the upper confidence limit of the RBA for arsenic in site
soils (0.42).
4.3 Risk Characterization for Arsenic
Risks from Soil and Dust
Cancer Risk
Cancer risks from exposure of residents to arsenic in yard soil and indoor house dust were
calculated for each property using the basic equations recommended by USEPA. The risk
ES-13
-------
estimates are expressed as the probability that an individual exposed to arsenic at the site will
develop a cancer by the age of 70 that would not otherwise have occurred. For example, a
cancer risk of 2E-05 means that the probability is 2 out of 10s (2 out of 100,000) that the exposed
individual might develop a tumor from site-related exposures. The results of these calculations
are shown in Table ES-2.
For CTE exposure conditions, most properties have estimated excess cancer risks for exposures
due to arsenic in soil plus dust that range from 1E-06 to IE-OS (5th to 95th percentiles), with a
maximum value of 9E-05. For RME exposure conditions, most properties have risks that range
from 9E-06 to 1E-04 (5th to 95th percentiles), with 92 properties having risks of 2E-04 or
higher. The highest RME risk value was 8E-04. The spatial pattern of properties with arsenic
RME cancer risk levels of 2E-04 or higher is approximately uniform across the site, with a
frequency of about l%-4% in each neighborhood.
In interpreting these risk estimates, it is important to recognize that arsenic is a naturally
occurring element in soil. Based on an analysis of the distribution of concentration values
observed in Phase III soil samples, it is estimated that background levels are well-characterized
as a lognormal distribution with a mean of 8 ppm and a standard deviation of 3.6 ppm. Based on
this, background levels may range up to about 15 ppm or slightly higher. If so, lifetime cancer
risks from naturally occurring levels of arsenic probably range from about 1E-06 for an average
(CTE) person up to about 1E-05 for an upper-bound (RME) individual.
Chronic Noncancer Risks
In accord with standard EPA methods, the risk of non-cancer effects is expressed as the ratio of
the dose resulting from exposure to site media compared to a dose that is believed to be without
risk of effects, even in sensitive individuals. This ratio is called the Hazard Quotient (HQ). If
the value of HQ is equal to or less than one (1E+00), it is believed there is no significant risk of
noncancer effects. If the HQ exceeds one, then there is a chance that noncancer effects may
occur, with the probability tending to increase as the value of HQ increases.
Estimated risks of non-cancer health effects from chronic exposure to arsenic in soil and dust are
shown in Table ES-3. For individuals with CTE exposure, risks at most properties fall between
2E-02 and 2E-01 (5th to 95th percentile), while individuals with RME exposure have risks that
lie mainly between 5E-02 and 6E-01. These results indicate that risk of noncancer effects from
chronic exposure is below a level of concern for most individuals at most locations. However, a
total of 20 properties have RME HQ values of 2E+00 or higher, with a maximum value of
4E+00. These locations where noncancer risks enter a range of concern (HQ > 1E+00) are also
above the usual level of concern (1E-04) for cancer.
Sub chronic Noncancer Risks
Estimated risks of non-cancer health effects from sub-chronic exposure of area children to
arsenic in soil are shown in Table ES-4. As seen, the incidence of properties with subchronic
HQ values above 1E+00 is relatively low (2 out of 2,986 = 0.07% for CTE individuals, 53 out of
ES-14
-------
Table ES-2 Estimated Cancer Risk from Arsenic in Soil and Dust
Neighborhood
Clayton
Cole
Elyria
Globeville
Swansea
All Neighborhoods
Number of
Properties
Evaluated
902
796
59
63
1166
2986
Number and Percent of Properties Within the Specified Risk Range
CTE Cancer Risk
<=1E-05
858
95%
772
97%
58
98%
61
97%
1132
97%
2881
96%
2E-05 - 1E-04
44
5%
24
3%
1
2%
2
3%
34
3%
105
4%
2E-04 - 1E-03
> 2E-03
RME Cancer Risk
<=1E-05
479
53%
344
43%
17
29%
25
40%
610
52%
1475
49%
2E-05 - 1E-04
385
43%
429
54%
41
69%
36
57%
528
45%
1419
48%
2E-04 - 1E-03
38
4%
23
3%
1
2%
2
3%
28
2%
92
3%
> 2E-03
CTE=Central Tendency Estimate
RME=Reasonable Maximum Exposure
ES-15
-------
Table ES-3 Estimated Chronic Noncancer Risk from Arsenic in Soil and Dust
Neighborhood
Clayton
Cole
Elyria
Globeville
Swansea
All Neighborhoods
Number of
Properties
Evaluated
902
796
59
63
1166
2986
Number and Percent of Properties Within the Specified Risk Range
CTE Hazard Quotient
<=1
901
100%
796
100%
59
100%
63
100%
1166
100%
2985
100%
2-5
1
0.1%
0
0%
0
0%
0
0%
0
0%
1
0%
6-10
—
—
—
—
—
..
—
—
—
..
—
—
> = 11
—
—
—
—
—
..
—
—
—
..
—
—
RME Hazard Quotient
<=1
895
99%
786
99%
59
100%
63
100%
1163
100%
2966
99%
2-5
7
0.8%
10
1.3%
0
0%
0
0%
3
0.3%
20
0.7%
6-10
—
—
—
—
—
..
—
—
—
..
—
—
> = 11
—
—
—
—
—
..
—
—
—
..
—
—
CTE=Central Tendency Estimate
RME=Reasonable Maximum Exposure
ES-16
-------
Table ES-4 Estimated Subchronic Noncancer Risks from Arsenic in Soil
Neighborhood
Clayton
Cole
Elyria
Globeville
Swansea
All
Number of
Properties
Evaluated
902
796
59
63
1166
2986
Number and Percent of Properties Within the Specified Risk Range
CTE Hazard Quotient
<=1
900
100%
796
100%
59
100%
63
100%
1166
100%
2984
100%
2-5
2
0.2%
0
0%
0
0%
0
0%
0
0%
2
0.1%
6-10
> = 11
RME Hazard Quotient
<=1
881
98%
777
98%
58
98%
62
98%
1155
99%
2933
98%
2-5
19
2%
19
2%
1
2%
1
2%
11
1%
51
2%
6-10
2
0.2%
0
0.0%
0
0.0%
0
0.0%
0
0.0%
2
0.1%
> = 11
ES-17
-------
2,986 = 1.8% for RME individuals). The maximum RME HQ value was 7E+00. All of the
locations where subchronic noncancer risks enter a range of concern (HQ > 1E+00) are also
above the usual level of concern (1E-04) for cancer.
Noncancer Risks from Acute Pica Behavior
Because of the substantial uncertainty which exists in most of the input parameters for the acute
pica scenario, it is not possible to specify a single set of inputs that are "best". Rather, a range of
HQ values were calculated for two different combinations of soil intake and RfD values:
Alternative Pica Exposure and Toxicity Values
Variable
Soil intake (mg/day)
Acute RfD (mg/kg-d)
Casel
CTE
5000
RME
10000
0.005
Case 2
CTE
2000
RME
5000
0.015
Case 1: RfD = 0.005 mg/kg; Pica soil intake = 10,000 mg/event
Case 2: RfD = 0.015 mg/kg; Pica soil intake = 5,000 mg/event
It should be understood that these cases represent an uncertainty range, and that the "true" acute
risk from pica behavior could lie anywhere in the interval. Indeed, it is quite possible that the
true value even lies outside the range, since the actual distribution of pica soil intakes is not
known.
The results are summarized in Table ES-5. As seen, the screening calculations above suggest
that a large number of properties (ranging from 662 to 1841, depending on which set of input
assumptions is deemed to be most appropriate) are of potential concern for the RME acute pica
scenario.
Because data are so sparse on the actual magnitude and frequency of soil pica behavior, and
considering that discussions continue to occur nationally on the most appropriate acute RfD for
arsenic, and it is difficult to judge which (if any) of these properties should be considered to be
an authentic acute health risk to children. In this regard, it should be noted that even though
many people are exposed to arsenic levels in soil that are predicted to be of acute concern, both
within the VBI70 site and elsewhere across the country and around the world, to the best of
USEPA's knowledge, there has never been a single case of acute arsenic toxicity reported in
humans that was attributable to arsenic in soil. Thus, these results for the acute pica scenario are
considered to be especially uncertain, since they predict a very substantial risk for which there is
no corroborating medical or epidemiological evidence.
ES-18
-------
Table ES-5 Estimated Acute Noncancer Risk from Pica Behavior
Exposure
Assumptions
Casel
Case 2
Number and Percent of Properties Within the Specified Risk Range
CTE Hazard Quotient
<=1
1475
49%
2692
90%
2-5
949
32%
268
9%
6-20
432
14%
26
1%
>20
130
4%
0
0%
Total > 1
1511
51%
294
10%
RME Hazard Quotient
<=1
1145
38%
2324
78%
2-5
580
19%
487
16%
6-20
328
11%
162
5%
>20
933
31%
13
0%
Total > 1
1841
62%
662
22%
Case 1: RfD = 0.005 mg/kg; Pica intake rate = 10,000 mg
Case 2: RfD = 0.015 mg/kg; Pica intake rate = 5,000 mg
ES-19
-------
Risks from Home-Grown Vegetables
A total of 72 different samples of garden vegetables were collected from 19 different properties
across the site. At each property, the 95% upper confidence limit (UCL) of the mean
concentration of arsenic was calculated, and this value (or the maximum, whichever was lower)
was used to estimate risks to residents. For individuals whose intake of home-grown garden
vegetables is average (CTE) for the western United States, neither non-cancer nor cancer risks
enter a range of concern at any property tested. For individuals whose intake is at the upper-
bound (RME) of the distribution of garden vegetable consumption, cancer and non-cancer risks
do enter a range of potential concern for two properties. However, these risks were driven either
by a single value that appeared to be anomalous, or by the margin of safety introduced by use of
the 95% UCL. Overall, it appeared that while risks from arsenic in garden vegetables could not
be entirely excluded, the risks were likely to be low. This is supported by noting that the intake
of arsenic from home-grown vegetables is predicted to be well within the normal dietary range
observed in the United States.
Total Risks for Ingestion of Soil and Home-Grown Vegetables
As noted above, data on arsenic levels in soil are available for all 2,986 properties investigated in
Phase III, but data on arsenic levels in gardens and vegetables were collected only at 19 of these
properties. Therefore, in order to calculate total risk at all properties, it was necessary to
estimate the concentration of arsenic in garden vegetables using site-specific data on the
relationship between arsenic in yard soil and in garden soil, and between arsenic in garden soil
and in vegetable tissues.
Because exposure and risk from soil ingestion and vegetable ingestion are both distributions,
care must be taken in the summation process. In the case of the non-cancer or cancer risk to an
individual who has average exposure to both soil and vegetables, the total risk is simply the sum
of the two pathway-specific risks:
CTE(total) = CTE(soil) + CTE(vegetables)
In the case of an individual who has RME exposure to soil or to vegetables, the estimate of RME
total risk is not the simple sum of the RME risk estimates, because the two pathways are
independent of each other, and an individual with RME soil intake is not likely to also have
RME vegetable intake (and vice versa). Thus, the estimate of RME total risk is calculated either
as:
1: RME(total) = RME(soil) + CTE(vegetables)
2: RME(total) = CTE(soil) + RME(vegetables)
The results are shown in Table ES-6. As seen, based on the site-specific relationships between
arsenic in yard soil and garden soil and between arsenic in garden soil and garden vegetables,
individuals with CTE exposure to garden vegetables are predicted to have excess cancer risks
that are less than or equal to 1E-05, while individuals that have RME intake of garden vegetables
ES-20
-------
Table ES-6 Estimated Total Cancer Risks from Soil and Vegetables
Statistic
CTE
Risk
RME
Risk
Pathway
Soil alone
Vegetables alone
CTE Soil + CTE vegetables
Soil alone
Vegetables alone
RME Soil + CTE vegetables3
CTE Soil3 + RME vegetables
Number of Properties
<= IE-OS
2881
2986
1475
933
2E-05 - 1E-04
105
2921
1419
2979
1954
2921
2E-04 - 1E-03
65
92
7
99
65
Adjusted to account for RME exposure duration (30 years)
ES-21
-------
are expected to have risks mainly between 2E-05 and 1E-04, with only a few properties having
risks that exceed 1E-04. When CTE risks are combined across pathways, there are 65 properties
where total risk exceeds 1E-04. When RME risks are combined across pathways, the highest
risks occur for case 1 (RME soil intake plus CTE vegetable intake). Based on this scenario,
there are 99 properties where total RME risks exceed 1E-04.
4.4 Uncertainties in Arsenic Risk Assessment
It is important to recognize that the calculations of short-term and long term exposure and risk
from arsenic ingestion in soil are based on a number of assumptions and estimates, and that these
introduce uncertainty into the risk results. The most important of the sources of uncertainty in the
calculations are summarized below.
Uncertainty in Yard-Wide Average Concentration
The concentration term that is appropriate for calculating chronic exposure and risk from
ingestion exposure to arsenic is the true mean concentration in the medium of concern (soil, dust,
vegetables), averaged over the area and time interval (averaging time) of concern. There are two
important sources of uncertainty in this value. First, because the true mean cannot be calculated
from a limited set of sample results, the USEPA utilizes the 95% upper confidence limit of the
mean as a conservative (high end) estimate of the true mean. This approach helps ensure that the
exposure and risk estimates that are derived are more likely to overestimate than underestimate
the actual risk. Second, the basic exposure unit selected for evaluation in this risk assessment is
the residential property. Using the UCL of the mean for a property is equal to assuming that an
individual residing at that location does not ingest soil or dust from any other location, even over
a time period of up to 30 years. While this might be true for a small sub-set of residents, it is
believed that most residents are sufficiently mobile that exposures will occur over a wider area
than just their own yard. This, in turn will result in lower exposures for people residing in homes
with affected soils, and their true risks will be lower than calculated.
Uncertainty in Concentration Values at Sublocations
As noted earlier, the sampling and analysis design for Phase III was based on a set of three
composite samples from each property. Consequently, there are no data that allow a direct
estimation of the concentration value at any specific sub-location of the yard (these are needed to
address risks from subchronic and acute exposures). To address this data limitation, the
distribution of concentration values within a property was modeled by assuming a lognormal
distribution, and the standard deviation within each property was estimated from a site-wide
average coefficient of variation. Since the mean at each property was estimated using the 95%
UCL or the maximum composite value, both the mean and the standard deviation are more likely
to be high than low at each property. Thus, the values estimated for evaluation of subchronic
and acute exposures are also more likely to be high than low.
ES-22
-------
Uncertainty in Intake Rates
Data on the amount of soil ingested by humans are very limited. Measurements are difficult to
perform, and results vary significantly from study to study and from method to method. In
addition, data are based mainly on short term studies, so estimates of long-term average intake
rates are especially uncertain. Moreover, intake rates are likely to vary from site to site and
property to property, depending on things such as climate, socioeconomic status, yard condition,
etc., so the default intake rates used in these calculations may not reflect the true intake rates at
the site. Because of the limitations in the data, the default values recommended by USEPA are
intended to be on the high side (i.e., are more likely to overestimate than underestimate actual
soil ingestion).
This is illustrated by comparing the default soil intake rates used by USEPA to data on soil
intake rates measured in a group of 64 children in Anaconda, Montana (Stanek and Calabrese
2000). This study, which utilizes the latest and most refined analytical and statistical methods
for estimating soil ingestion by children, estimated that the average (CTE) 7-day intake by
children is about 31 mg/day (compared to the default of 100 mg/day), and that the 95th
percentile intake for 7 days and 365 days are 133 and 106 mg/day, respectively (compared to the
default assumption of 200 mg/day). If these values from the Anaconda site were judged to be a
more reliable basis for estimation of risk from soil ingestion than the current default values, and
if adult soil intake is assumed to be about /^ that of children, then there are only 23 properties
(rather than 92 properties) in the VBI70 site where RME cancer risks from soil ingestion exceed
a level of 1E-04.
Uncertainty in the Fraction of Total Intake that is Soil
One of the variables used to calculate risks from ingestion of soil plus dust is the fraction of the
total intake that is soil (fs). The EPA default value for this variable (45%) is based mainly on
measurements in a set of 64 preschool children, but due to the difficulty in making these
measurements, as well as potential differences between children and between sites, this value
should be considered uncertain. It is not known whether the true value at the VBI70 site is more
likely to be higher or lower than the default values. If the true site-specific value of fs were
lower (e.g., 20% rather than 45%), risks would be about 12% lower than calculated. Conversely,
if the true site-specific value were higher (e.g., 70% rather than 45%), then the risks would be
about 12% higher than calculated.
Uncertainty in Exposure Duration
Cancer risk calculations depend on the duration of exposure. Default exposure durations used in
the risk assessment are not site-specific, and are estimated from data on the length of time that
people own a particular residence. Thus, actual exposure durations of residents at the site may
not be the same as the assumed exposure durations assumed, and might be either longer or
shorter than assumed. For example, if the exposure duration were assumed to be 45 years (6
years as a child and 39 years as an adult) rather than the default value of 30 years, the estimated
excess cancer risk level from soil ingestion would be about 19% higher than the values reported.
ES-23
-------
In addition, all of the exposure calculations presented here assume that exposure begins during
childhood, when intake rates are higher than during adulthood. Thus, risks to individuals who
move to the site after they are children will be lower than estimated. For example, risks to an
individual exposed for 30 years as an adult are only 37% of the risks to an individual exposed for
6 years as a child and 24 years as an adult.
Uncertainty in RME Exposures
In the default point estimate approach for estimating exposure and risk to an RME individual,
two exposure parameters (intake rate and exposure duration) are both assumed to be at their 95th
percentile values. In reality, because these two exposure parameters are independent of each
other, it is very unlikely that an individual with RME soil intake will also have RME exposure
duration. Therefore, an individual with both RME soil intake and RME exposure duration
represents not the 95th percentile of the risk distribution, but some significantly higher
percentile. One way to estimate what the percentile of the default RME individual is, as well as
the actual 95th percentile value, is through Monte Carlo modeling. Screening level calculations
performed with this approach suggest that the RME risk estimate derived by the point estimate
approach is about twice the Monte Carlo estimate of the 95th percentile value, and is located at
approximately at or above the 99th percentile of the risk distribution. This supports the
conclusion that RME point estimates of risk provide a substantial margin of safety.
Uncertainty in Toxicity Factors
One of the largest sources of uncertainty in most risk assessments stems from uncertainty in the
toxicity factors used to predict responses from the calculated doses. In the case of arsenic, dose-
response data are derived from studies in humans, which significantly reduces the degree of
uncertainty compared to extrapolations based on animal data. However, a significant degree of
uncertainty still remains in both the oral cancer slope factor and the chronic RfD. One of the
most important sources of this uncertainty is lack of reliable data on actual arsenic ingestion
rates by the human population used to quantify risk. There are also still large uncertainties in
how to extrapolate the dose-response curve from relatively high exposure levels to lower
exposure levels. For example, arsenic does not appear to cause cancer by a direct genotoxic
mechanism (USEPA 200Id), suggesting that a sub-linear (and perhaps even a threshold) model
might be reasonable. However, in the absence of information on the actual mode of action, an
assumption of linearity is still deemed to be necessary and appropriate (USEPA 200 Id). If the
dose response curve is sub-linear, current risk estimates would be too high. Further, there is
uncertainty in the importance of cultural, ethnic, dietary, and socioeconomic differences between
different study populations. While little is known about the relative importance of these factors,
it is likely that there are differences between people in their sensitivity to ingested arsenic, and it
is for this reason that USEPA seeks to ensure an adequate margin of safety in the derivation of
the RfD and the slope factor.
ES-24
-------
Uncertainty in Bioavailability
In order to cause an adverse response, arsenic that is ingested must be absorbed into the body.
Measurements of the arsenic relative bioavailability have been performed for five soils from the
VBI70 site. While measurements based on site soils significantly reduces uncertainty in this
exposure parameter, uncertainty still remains. For example, variability was observed between
different site soils, and a conservative estimate of the mean value was employed to represent the
site-wide average absorption. This approach is expected to result in an over-estimate of true
absorption. Another source of uncertainty is in the extrapolation of data from test animals to
humans. The test animals (swine) were selected because they are believed to have a
gastrointestinal system similar to that in humans, but it is also expected that absorption in
humans may vary as a function of age, stomach contents, nutritional status, etc. Thus, the
measurements in animals should be viewed as uncertain estimates of the true values in humans.
The RBA measured for soil was also assumed to apply to dust. This assumption is uncertain
because the size distribution of arsenic-containing particles in dust may be different than for soil,
and particle size might be one factor that influences RBA. If dust contains smaller particles than
soil, and if this size difference tends to increase RBA, then the use of the soil RBA could
underestimate the absorption of arsenic from dust. However, it should be remembered that the
RBA value for soil was measured using only the fine fraction of soil (only particles smaller than
250 micrometers in diameter), so the difference in particle size distribution between dust and soil
is not expected to be large. In addition, because arsenic concentrations in dust tend to be lower
than in soil, the dose contributed by dust ingestion is relatively small compared to that for soil,
so uncertainty in the absorption fraction for dust results in only a small uncertainty in the total
absorbed dose.
Uncertainty in Pica Exposure and Risks
As noted above, screening-level calculations suggest that acute high-dose exposures to arsenic in
soil (i.e., pica exposure) might be of concern at a number of properties within the site. However,
data on the amount of soil ingested during pica behavior are very sparse. Based mainly on one
study that observed an intake of 5-8 g/day by a single child, (Calabrese et al. 1989), USEPA has
indicated that 5-10 grams might be a reasonable estimate. If this intake rate is correct, and if
arsenic absorption from this mass of soil is similar to that estimated in site-specific studies
(42%), then anywhere from 22% to 62% of all properties within the VBI70 site (and perhaps
outside the site as well) could have arsenic levels above a level of acute concern. USEPA feels
this conclusion is especially uncertain, since the Agency is not aware of any reported cases of
acute arsenic toxicity attributable to ingestion of arsenic in soil. A more recent study of soil
intake did not observe intake rates above 700 mg/day in a group of 64 children, suggesting that
values of 5-10 grams might be unrealistically high. In addition, limited data on urinary arsenic
levels in residents of the VBI70 area and the nearby Globe neighborhood do not reveal the
occurrence of high soil intakes by children. These considerations suggest that arsenic risk from
soil pica may not be as significant as the calculations suggest. On the other hand, if this type of
exposure were to occur, it is possible the symptoms (transient upset stomach and general
malaise) would not be recognized as being arsenic-related, and could easily go un-detected or
ES-25
-------
un-reported. In addition, if pica behavior is assumed to occur only infrequently during
childhood, then the chances of observing the behavior in a study could be quite low. Because of
the high uncertainty regarding the magnitude and frequency of soil pica behavior, more reliable
risk estimates for this scenario will not be possible until better data are collected on pica intakes,
along with direct measures of soil-related exposures to arsenic in soil.
Summary of Uncertainties in Arsenic Risk Characterization
Because of the uncertainties summarized above, none of the exposure and risk calculations for
arsenic should be interpreted as accurate measures of the true risk, rather, all values should be
interpreted as uncertain estimates. Because a majority of the approaches for dealing with
uncertainty are more likely to overestimate than underestimate true risk, the final risk values
above should be thought of as more likely to be higher than lower than the actual risks.
5.0 EXPOSURE AND RISK FROM LEAD
5.1 Overview
Risks from lead are evaluated using a somewhat different approach than for most other metals.
First, emphasis is placed on evaluation of risks to young children because they are more likely to
be exposed and are more susceptible to the effects of lead than adults. Second, risks are
expressed as the probability that a child will have a blood lead value greater than 10 ug/dL. A
blood lead of 10 ug/dL is a value identified by EPA as the level at which effects that warrant
avoidance begin to occur, and EPA has set as a goal that there should be no more than a 5%
chance that any child will have a blood lead value above 10 ug/dL.
5.2 IEUBK Model for Assessing Lead Risk
Risks from Soil and Dust
The USEPA has developed an Integrated Exposure Uptake Biokinetic (IEUBK) model for
predicting the likely range of blood lead levels in a population of young children (age 0-6 years)
exposed to a specified set of environmental lead levels. The IEUBK model was used to predict
risks at each property that was sampled during Phase III, using the mean of the three composite
values from each property as the best estimate of the average bulk lead concentration in soil at
each property. This value was adjusted by a factor of 1.09 to estimate the concentration in fine
soil. Other input parameters for the IEUBK model were the defaults recommended by EPA
except for two site specific inputs: 1) the concentration of lead in dust as a function of the
concentration in bulk soil, which were based on site-specific measurements, and 2) the relative
bioavailability of lead, which was based on a test of site soils in an animal study. The site
specific RBA was 84%, higher than the default assumption of 60%.
ES-26
-------
The IEUBK model was used to calculate the expected blood lead distribution for children (age 0-
84 months) for each property. The results, characterized in terms of the probability of a random
child exceeding a blood lead value of 10 ug/dL (this is referred to as "P10"), are shown in Table
ES-7. As seen, a total of 1,655 out of 2,986 homes are predicted to have P10 values at or below
the health-based goal of 5%, while 1,331 (45%) are predicted to exceed the health-based goal.
Approximately 610 properties are predicted to have P10 values of 5-10%, slightly above the
heath-based goal. However, about 518 properties would be expected to have PI0 values
between 10-20%, and 203 homes are predicted to have P10 values greater than 20%
(substantially above the health-based goal). It should be noted that 1,057 of the 1,331 properties
(79%) with P10 values above 5% have mean bulk lead concentrations lower than 400 ppm (the
USEPA default level of concern). This is mainly because the site-specific KB A for lead (84%) is
higher than the default value (60%), and also because of the use of the concentration value in
the fine fraction rather than the bulk fraction in the risk calculations.
Although homes with elevated soil lead are found in all neighborhoods, the density of homes
with P10 values greater than 5% tends to be higher in the central and western part of the site than
in areas on the eastern side of the site.
In interpreting these risk estimates, it is important to recognize that lead is a naturally occurring
element in soil, and that there are many current and historic anthropogenic sources of lead (e.g.,
automobile exhaust, leaded paint, generalized industrial emissions, etc.). Based on the extensive
soil data set collected during Phase III, levels of lead in bulk soils at the VBI70 site range from
below the detection limit (about 52 ppm) up to a maximum of more than 1,000 ppm. If it is
assumed that the upper range of the lead from natural and area-wide anthropogenic sources is
about 400 ppm, then the mean of all samples that are less than 400 ppm is about 195 ppm. Using
this value (195 ppm in bulk soil) as a rough estimate of the mean concentration in urban
background samples, and assuming the same site-specific input values described above, the
IEUBK model predicts that blood lead levels attributable to urban background levels of lead
probably average about 4.4 ug/dL for a typical (median) child, and might be as high as 9.5 ug/dL
for a child with above-average (95th percentile) exposure to soil or dust.
Risks from Lead in Garden Vegetables
As noted previously, site-specific data show there is essentially no detectable uptake of lead
from soil into garden vegetables at this site. On this basis, it is concluded that exposure to lead
from ingestion of home grown garden vegetables is not of concern.
5.3 Uncertainties in Lead Risk Evaluation
It is important to stress that lead risk predictions based on the IEUBK model are uncertain. This
uncertainty arises from a number of factors. First, there is inherent difficulty in providing the
model with reliable estimates of human exposure to lead-contaminated media. For example,
exposure to soil and dust is difficult to quantify because human intake of these media is likely to
be highly variable, and it is very difficult to derive accurate measurements of actual intake rates.
Likewise, site-specific data on exposure to lead through the diet are generally not available, and
ES-27
-------
Table ES-7 Estimated Risks to Children from Lead in Soil and Dust
Neighborhood
Clayton
Cole
Elyria
Globeville
Swansea
All
Total Number of
Properties
902
100%
796
100%
59
100%
63
100%
1166
100%
2986
100%
Number and Percent of Properties Within Specified Risk Range
P10 <= 5%
712
79%
169
21%
6
10%
7
11%
761
65%
1655
55%
P10 > 5% and <= 10%
119
13%
248
31%
9
15%
18
29%
216
19%
610
20%
P10 > 10% and <= 20%
52
6%
273
34%
28
47%
21
33%
144
12%
518
17%
P10 > 20%
19
2%
106
13%
16
27%
17
27%
45
4%
203
7%
Total P10>5%
190
21%
627
79%
53
90%
56
89%
405
35%
1331
45%
P10=Prediced Risk of Exceeding Blood Lead of 10 ug/dL
ES-28
-------
because dietary lead levels have been decreasing over time, the default data used in the model
may no longer be accurate. Second, it is often difficult to obtain reliable estimates of key
pharmacokinetic parameters in humans (e.g., absorption fraction, distribution and clearance
rates, etc.), since direct observations in humans are limited. Finally, the absorption, distribution
and clearance of lead in the human body is an extremely complicated process, and any
mathematical model intended to simulate the actual processes is likely to be an over-
simplification. Consequently, IEUBK model calculations and predictions should not be thought
of as being identical to actual risk.
One way to help characterize the uncertainty that may exist in the IEUBK model calculations is
to investigate the effect of alternative (non-default) model inputs for some of the more uncertain
parameters. Especially important is the GSD value, which has a very powerful effect on the
number of properties of concern. Studies at other sites have shown that the GSD value may
often be lower than the default of 1.6, and if that were to be the case at this site, risks to children
from lead could be substantially overestimated. Another parameter that is uncertain is the soil
intake rate, and if data from the most recent study of soil intake in children were used in place of
the default soil intake values, risks from lead would be below a level of concern at most
locations.
Another way that may sometimes help assess whether the IEUBK model is yielding reliable
results at a particular site is to compare the IEUBK model predictions with actual observations of
blood lead levels in the population of children currently living at the site. At the VBI70 site,
only very limited blood lead data are available, with values from only 21 individuals available.
In this group of individuals, the maximum blood lead concentration observed was 5 ug/dL.
While this the data set is much too limited to support the conclusion that risks are absent, neither
do the results signal any cause for alarm. Data from several blood lead surveillance programs
conducted by the State suggest that lead in soil does contribute to blood lead in area children, but
that soil lead is not the primary reason for blood lead concentrations greater than 10 ug/dL.
6.0 CONCLUSION
Arsenic
Some residential properties at the VBI70 site contain arsenic at concentrations substantially
higher than the expected natural levels. Properties with elevated levels of arsenic occur at
widely scattered locations across the site, with no clear spatial pattern. At an affected property,
the contamination appears to be distributed across the yard area, with a fairly clear boundary
between the affected property and the adjacent properties. The chemical form of the arsenic is
predominantly arsenic trioxide.
In some cases, levels of arsenic in yard soil is sufficiently elevated to pose an RME excess
lifetime cancer risk that is above a level of 1E-04. Based on current data, about 3% of all
properties fall into this category. Chronic and subchronic non-cancer risks from arsenic are also
ES-29
-------
above a level of human health concern at some properties, mainly at the same locations where
cancer risks are above 1E-04.
Screening level calculations suggest that acute high level (pica) intake of soil by children might
be of acute non-cancer concern at a large number of properties at the site, but this finding is
judged to be especially uncertain due to lack of reliable information on the magnitude and
frequency of pica soil ingestion and on the most appropriate acute oral RfD value.
Lead
Lead also occurs at elevated levels in soil at some residential properties. Elevations occur in all
neighborhoods of the site, but levels tend to be higher on the western part of the site than the
eastern part. Using EPA's IEUBK model to evaluate the risk to children, it is estimated that
about 45% of residences have levels that exceed EPA's health-based goal (no more than a 5%
chance that a child will have a blood lead value above 10 ug/dL). Of these, many (about 79%)
have mean lead concentrations lower than 400 ppm (the USEPA default level of concern). This
is mainly because the site-specific RBA for lead (84%) is higher than the default value (60%).
ES-30
-------
SECTION 1
INTRODUCTION
1.1 SITE DESCRIPTION
The Vasquez Boulevard and 1-70 (VBI70) Superfund Site is an area of approximately four
square miles located in the north-central section of Denver, Colorado. The site is composed of a
number of neighborhoods that are largely residential, including Swansea, Elyria, Clayton, Cole,
and portions of Globeville. Most residences at the site are single family dwellings, but there are
also some multi-family homes and apartment buildings. The site also contains a number of
schools, parks, and playgrounds, as well as a number of commercial and industrial properties.
Figure 1-1 is a map which displays the site.
The site is largely flat in topography, sloping gently towards the Platte River, which flows in a
northeasterly direction through the site. Other than the Platte River, there are no other major
surface water bodies within the site.
The climate of the site is generally typical of Colorado's semiarid eastern plains. Temperatures
are moderate throughout the year, with monthly averages ranging from 30° F in January to 73° F
in July. Annual rainfall measures 16 inches, 60% of which falls during the spring and summer.
The rainiest month is May, with an average rainfall of 2.6 inches. Snowfall totals in the Denver
Metro area average 60 inches, with March usually receiving the most snow (12.5 inches). The
Rocky Mountain foothills, about 20 miles west of the site, help create a predominantly southern
wind flow at the site, with an annual average velocity of about 8.5 mph. Peak winds can reach
velocities of 30-50 mph, with the highest winds tending to be from the north-northwest
(Colorado Climate Center 2000).
1.2 BASIS FOR POTENTIAL CONCERN
The site came to the attention of the U.S. Environmental Protection Agency (USEPA) because
studies directed by the Colorado Department of Public Health and Environment (CDPFffi) at a
nearby site (Globe Smelter) indicated that elevated concentrations of arsenic and/or lead
occurred in the soil of some residential properties in the Swansea/Elyria area. The source of
these elevated levels is not known, but a priori, it was considered plausible that the
contamination could be associated with releases either from the Globe facility and/or from one or
both of two other smelters which previously existed in the area (the Argo Smelter and the Omaha
and Grant Smelter). The locations of these three smelters in relation to the VBI70 site are also
shown in Figure 1-1. Alternative potential sources include the historic application of arsenic- or
lead-containing lawn care products, and/or (for lead) anthropogenic sources such as automobile
exhaust, leaded paint, etc.
Based on the results of several rounds of soil sampling (see Section 2.0), USEPA concluded that
the VBI70 site contained multiple residences where the concentration of arsenic and/or lead in
1
-------
Figure 1-1 Site Map
m
-------
yard soil could be above a level of potential human health concern. On this basis, USEPA
proposed the VBI70 site for inclusion on the Superfund National Priorities List (NPL) in
January, 1999, and the site was added to the NPL on July 22, 1999.
1.3 PURPOSE AND SCOPE OF THIS DOCUMENT
This document is a baseline human health risk assessment. The purpose of the assessment is to
characterize the nature and magnitude of any risk to humans that may be attributable to
contamination of site media, assuming that no steps are taken to remediate the environment or to
reduce human contact with contaminated environmental media. More specifically, this
assessment focuses on the direct and indirect risks to humans from contamination that is present
in soils in current residential and commercial (non-smelter) areas of the site. This is referred to
as the "Off Smelter Facility Operable Unit" (Operable Unit 1). The potential human health risks
from exposure to other potentially contaminated environmental media (e.g., surface water,
groundwater) and on-site soils (i.e., soils at former smelter areas) will be investigated and
evaluated as separate Operable Units.
The results of this baseline risk assessment are intended to help inform risk managers and the
public about the level of health risk which is attributable to contamination in site soils, to help
determine the need for remedial action at the site, and to provide a basis for determining the
levels of chemicals that can remain in site soils and still be adequately protective of public health
(USEPA 1989).
The methods used to evaluate risks to humans and the environment employed in this assessment
are consistent with current guidelines provided by the USEPA for use at Superfund sites
(USEPA 1989, 199 la, 1991b, 1991c, 1992a, 1992b, 1993).
1.4 ORGANIZATION OF THIS DOCUMENT
In addition to this introduction, this report is organized into the following sections:
Section 2 This section provides a summary of the available data on the levels of chemical
contaminants (metals) in site soils, and identifies which of these chemicals are of
potential health concern to area residents or workers.
Section 3 This section discusses how residents and other people (workers, children at
schools or playgrounds) may be exposed to site-related chemicals, now or in the
future, and identifies exposure scenarios that are considered to be of potential
concern.
Section 4 This section assesses the level of exposure and risk to humans from arsenic in site
soils. This includes 1) a description of methods used to quantify exposure to
arsenic, 2) data on the toxicity of arsenic to humans, 3) calculation of the level of
noncancer and cancer risk that may occur as a result of exposure to arsenic in site
-------
soils, and 4) a discussion of the uncertainties which limit confidence in the
assessment.
Section 5 This section assesses the level of exposure and risk to area residents from lead in
site soils. This includes 1) a description of the toxic effects of lead, 2) a summary
of the method used by USEPA to evaluate risks from lead, 3) a summary of the
estimated risks at this site attributable to lead in site soils, and 4) a discussion of
the uncertainties which limit confidence in the assessment.
Section 6 This section provides full citations for USEPA guidance documents, site-specific
studies, and scientific publications referenced in the risk assessment.
-------
SECTION 2
SUMMARY OF SITE DATA AND
SELECTION OF CHEMICALS OF POTENTIAL CONCERN
Data on the level of arsenic, lead, and other metals which might have been released from area
smelters into site soils have been collected in a phased series of investigations. A detailed
summary and evaluation of these studies are presented in the Remedial Investigation/Feasibility
Study (RI/FS) report for this site (USEPA 2001e). Each of these investigations is described
briefly below, along with a summary of the key data collected during each phase.
2.1 PHASE I/PHASE II GRAB SAMPLE INVESTIGATION
Residential Soil Samples
Once investigations at the nearby Globe site began to suggest that elevated levels of arsenic
and/or lead might exist in soils at residential properties within the area of the VBI70 site,
CDPHE requested assistance from USEPA Region VIII in characterizing the nature and extent of
the contamination. In response, USEPA Region VIII undertook a study designed to identify
properties that had levels of arsenic or lead that were sufficiently high that time-critical action
(soil removal and replacement) might be warranted. The action levels selected for time-critical
soil removal were 450 parts per million (ppm) for arsenic and 2,000 ppm for lead (USEPA
1998a).
Details of the study are presented in UOS (1998a, 1998b). In brief, grab samples of surface soil
and subsurface soil were collected from 1390 residential properties in the area of potential
concern. Most of these samples were collected during the initial round of sampling (referred to
as Phase I), with the remainder being obtained in a subsequent sampling effort (Phase II). In the
majority of cases, two surface samples and one subsurface sample were collected per property,
with additional surface samples at some locations (depending on the size of the property). All
samples were analyzed for arsenic, lead, cadmium and zinc using X-ray fluorescence (XRF).
The results for arsenic in surface soil are summarized in Figure 2-1 (upper panel). As seen, a
majority of properties sampled (927 out of 1390) had maximum arsenic values that were below
the limit of detection (average detection limit = 51 ppm). However, arsenic was detected in one
or more surface soil samples at a number of properties, with 40 of these properties having one or
more samples above 450 ppm. Arsenic concentrations in subsurface samples were generally
somewhat lower than the concentrations in surface soil, with an average ratio of subsurface to
surface soil of about 0.8.
For lead (lower panel), most properties (1153 out of 1390) had maximum concentration values in
surface soil that were below 400 ppm, but 238 properties had one or more values above 400
-------
Figure 2-1 Phase I/Phase II Soil Grab Sample Data
1000
8 800
450
Maximum Concentration of Arsenic (ppm)
1400 -i
1200
| 1000
2000
Maximum Concentration of Lead (ppm)
Notes:
ND = Not Detected
Average detection limit for arsenic = 51 ppm (range = 44 to 800)
Average detection limit for lead = 32 ppm (range = 28 to 38)
-------
ppm. Of these, 6 properties had one or more lead values above 2,000 ppm. Lead levels in
subsurface soil tended to be lower than in surface soil, with an average ratio of subsurface to
surface soil of about 0.7.
Any property with one or more arsenic values above 450 ppm and/or one or more lead values
above 2,000 ppm were identified as candidates for soil removal, pending collection and analysis
of composite soil samples to better characterize the true level of contamination (see below).
2.2 PHASE II CONFIRMATORY SAMPLING AND SOIL REMOVAL
In order to help confirm the identity of properties which warranted time-critical soil removal
actions, USEPA collected two or more composite samples (each consisting of five sub-samples)
of surface soil from residential properties where one or more grab samples were above the
removal level for arsenic. This approach was employed because composites samples are judged
to provide a more reliable and representative characterization of a yard than a single grab
sample.
Based on the results of this composite sampling program, a total of 21 residences were identified
where one or more composites confirmed that arsenic levels were above the action level. Of
these, 18 underwent soil removal and replacement in the fall of 1998, while the owners of the
other three properties refused permission for the removal. No properties were identified where
lead levels in composite soil samples were high enough to warrant a time-critical soil removal
action.
2.3 RISK-BASED SAMPLING PROGRAM
Following completion of the Phase I/Phase II sampling programs, USEPA undertook a number
of additional studies in order to provide information that would help support long-term risk-
based decision making at the site. One of these studies, referred to as the Risk-Based Sampling
Program, collected more detailed data on metal contamination and exposure at the 18 properties
that had been identified as requiring time-critical soil removal. Key elements of the program
included: 1) detailed soil sampling to reveal the spatial pattern of contamination at some of the
affected properties; 2) measurement of arsenic and lead levels in indoor dust, attic dust, and
garden vegetables, as well as lead levels in paint and tap water; and 3) measurement of
biomarkers of lead and/or arsenic exposure in residents at those locations. The details of the
risk-based study design are presented in USEPA (1998b), and the results are detailed in USEPA
(200 le). The main findings of this program are summarized below.
2.3.1 Spatial Patterns of Contamination
One of the striking findings that emerged from the Phase I/Phase II sampling programs was that
properties that were affected by arsenic did not appear to occur in a clear spatial pattern. That is,
the occurrence of high arsenic levels in soil did not appear to be associated with proximity to one
or more of the smelters, and properties with elevated levels of arsenic often occurred
immediately adjacent to one or more residences that were not apparently affected.
7
-------
In order to obtain additional information on the spatial pattern of contamination both within and
between yards, USEPA selected eight properties to undergo detailed soil sampling. Five of the
yards were locations where Phase I/Phase II sampling indicated the arsenic concentrations were
above the removal level, while three of the properties had arsenic concentrations below the
removal level.
At each property, a high-density grid was established on 5-foot centers, and soil samples were
collected wherever the grid node did not fall on a driveway, patio, etc. In addition, whenever
access could be obtained, the sampling grid was extended 10-15 feet into adjacent properties in
order to determine if there was a clear difference in contamination levels between adjacent
properties. All samples were analyzed by XRF for arsenic, lead, cadmium, and zinc.
Diagrams which show the results for all four metals at all eight properties are presented in
USEPA (200 le). Diagrams from this report that show the spatial patterns of arsenic and lead at
two properties with high levels of arsenic contamination are shown in Figures 2-2 and 2-3. In
both cases, arsenic levels vary from location to location, but are elevated across most of the yard.
At property 1 (Figure 2-2), there is a fairly clear boundary between the property of concern and
the adjacent properties. A similar pattern is observed at property 2 (Figure 2-3), although there
are some locations where the contamination may extend somewhat into the adjacent property.
The pattern of lead contaminations at these properties also showed a similar boundary effect. No
clear boundary effect was observed for cadmium or zinc.
2.3.2 Contaminant Levels in Other Environmental Media
Samples of other environmental media were obtained at each removal property where access was
granted. The results are summarized below.
Indoor Dust
Dust from interior living spaces were collected at 15 properties, while attic dust was collected at
9 properties. Summary statistics are presented below.
Arsenic and Lead in Dust Samples from the Risk-Based Sampling Program
Medium
Interior dust (ppm)
Attic dust (ppm)
Arsenic
Detection
Frequency
14/15
7/9
Mean
(ppm)
107
230
Max
(ppm)
172
499
Lead
Detection
Frequency
15/15
9/9
Mean
(ppm)
243
1414
Max
(ppm)
1145
4106
Regression analysis between measured levels of arsenic and lead in indoor dust compared to the
mean of the two five-point yard soil composites that were collected in Phase II revealed very
little correlation between the concentration of either arsenic or lead in interior dust compared to
that in outdoor soil, and the slopes of the best fit regression lines were not different from zero:
-------
Figure 2-2 Spatial Distribution of Contaminants - Property 1
Surface Soil - Arsenic (ppm)
20 -,
18 -
t
i As (<=70) »As (71-150) As (151-450) As (451-1000) • As (>1000)
Scale is approxima te
Surface Soil - Lead (ppm)
Pb(<=400) »Pb (401-1000) Pb (1001-1500) • Pb (1500-2000) • Pb (>2000)
Scale is approxima te
-------
Figure 2-3 Spatial Distribution of Contaminants - Property 2
Surface Soil - Arsenic (ppm)
•• •
30
Garage
Driveway
Garden
•••
• •
•
•
•
••
£
3
o
ffi
35
iAs (<=70) »As (71-150) As (151-450) As (451-1000) • As (>1000)
Scale is approximate
40
Surface Soil - Lead (ppm)
>$••• •
••••••
House
••• 9999999999999 999
o
10
15
20
25
Garage
Driveway
Garden
30
35
40
Pb(<=400) »Pb (401-1000) Pb (1001-1500) • Pb (1500-2000) • Pb (>2000)
Scale is approxima te
10
-------
Correlation Between Yard Soil and Indoor Dust
Analyte
Arsenic
Lead
N
15
15
R2
0.003
0.001
Slope
-0.004
-0.0014
P Value
>0.8
>0.9
Although this data set is too small to draw definite conclusions, the results suggest that outdoor
soil is not a major determinant of arsenic or lead levels in indoor dust. There was also no
significant correlation for arsenic or lead between the concentration in indoor dust and in attic
dust. This suggests attic dust is not serving as an important source of indoor dust at this site.
Tap Water
Twelve properties allowed sampling and analysis of tap water for lead. Two types of water
samples were collected: first flush and post-flush. Summary statistics are presented below.
Occurrence of Lead in Residential Water Samples
Medium
First-flush tap water
Post flush tap water
Detect. Freq.
5/12
3/12
Mean (ug/L)
3.2
2.5
Max (ug/L)
11.4
6.0
All of these values are below the current USEPA action level for lead in drinking water (15
ug/L), and are sufficiently low that tap water is not likely to be a significant source of lead
exposure, at least in the 12 homes sampled.
Paint
Sixteen properties authorized analysis of lead levels in paint. Concentrations were measured by
XRF at multiple locations on both interior and exterior surfaces. Summary statistics are
presented below:
Occurrence of Lead in Residential Paint
Location
Interior
Exterior
N
89
55
Mean (mg/cm2)
4.2
4.8
Range (mg/cm2)
0.3 - 19
0.4 - 14
A total of 130 out of 144 samples had values above 1 mg/cm2, the national default screening
level for leaded paint (HUD 1995). These data suggest that interior and/or exterior leaded paint
might be a source of lead exposure in area children, either directly (by paint chip ingestion), or
indirectly (by ingestion of dust or soil containing paint-derived lead). In this regard, there is a
weak but significant correlation between the concentration of lead in exterior leaded paint and
the concentration of lead in yard soil (R2 = 0.283, p < 0.03, n = 16), suggesting that some of the
11
-------
lead in soil at the properties sampled may be attributable to exterior leaded paint. No significant
correlation was detected between lead levels in interior paint and in indoor dust (R2 = 0.016, p >
0.5, n= 12).
Garden Vegetables
Only one of the 18 properties scheduled for soil removal had a vegetable garden. At this
location, one sample of potato and one sample of mint were collected. Concentrations of arsenic
and lead were below the level of detection in both samples. Because so few samples were
obtained, no conclusions can be drawn from this data set.
2.3.3 Phase II Biomonitoring
During Phase II, a total of 15 individuals residing at properties scheduled for soil removal (i.e.,
arsenic concentration above 450 ppm, or lead above 2,000 ppm) volunteered to have samples of
hair, urine and/or blood analyzed for arsenic or lead The results are summarized in Table 2-1.
For convenience, reference values indicating the typical and upper end of the normal range are
also presented.
As seen, there were no cases where individuals living at the properties scheduled for soil
removal had arsenic or lead levels that exceeded the "background" range typically seen in
members of the general population, although one individual had a hair arsenic at the high end of
the normal range. Although this data set is too small to draw firm conclusions, the results
provide no indication that exposures at these locations were of immediate health concern.
2.4 PHYSICAL-CHEMICAL CHARACTERIZATION
In addition to the Risk-Based Sampling Program described above, USEPA also undertook two
studies to characterize the physical and chemical attributes of the metal contamination in
residential site soils, and to determine whether concentration estimates based on bulk (unsieved)
soil samples were representative of concentrations in fine (sieved) samples. The design of these
projects is presented in USEPA (1998c) and USEPA (1999e), and the results are detailed in
USEPA (1998d) and USEPA (200 le). The main findings are summarized below.
2.4.1 Concentration in Sieved and Un-Sieved Soil Samples
As discussed in greater detail in Section 3, the main pathway by which humans are likely to be
exposed to contaminants in soil is by incidental ingestion of soil particles adhering to the hand.
Although data are limited, it is generally expected that small soil particles are more likely to
adhere to the hands than coarse particles, and it is for this reason that USEPA Region VIII
recommends that measurements of contaminant concentrations in soil generally be performed on
samples that have been sieved to isolate the smaller particles (< 250 um). This sieved fraction is
generally referred to as the "fine" fraction. Soil that has not been fine sieved but only coarse
sieved (to remove particles larger than about 2 mm) is referred to as the "bulk" sample. Studies
12
-------
Table 2-1 Biomonitoring Data for Residents at Phase II removal Properties
Demogra
Index
Number
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
phic Data
Age (years)
3
7
9
13
16
17
22
43
43
47
51
56
58
65
70
Biomonitoring Data
Blood Lead
Value (ug/dL) Qual.
2
2
2
2
1 U
2
1
3
2
2
3
4
3
2
2
Hair Arsenic
Value (ug/g) Qual.
0.43 U
1.32 U
0.39 U
0.39 U
0.3 U
0.45 U
0.41
0.28 U
0.29 U
1.16 U
0.41 U
0.26 U
0.32 U
0.91 U
0.38 U
Urinary Inorganic Arsenic
Value (ug/L) Qual.
20 U
20 U
20 U
10 U
20 U
10 U
10 U
20 U
10 U
20 U
20 U
20 U
20 U
20 U
20 U
Summary Statistics
Blood Lead
Age (years)
1-5
>=6
All
Site Data (a)
N
1
14
15
Detect. Freq.
1/1
13/14
14/15
Geo. Mean
(ug/dL)
2.0
2.1
2.1
Min
(ug/dL)
2.0
1.0
1.0
Max
(ug/dL)
2.0
4.0
4.0
Reference (b, c)
Typical
(ug/dL)
2.5-4.1
1.5-4.0
2.3-2.8
High End
(ug/dL)
>10
>10
>10
Hair Arsenic
Age (years)
0-6
>6
All
Site Data (a)
N
1
14
15
Detect. Freq.
0/1
1/14
1/15
Mean
(ug/g)
0.4
0.5
0.5
Min
(ug/g)
0.4
0.3
0.3
Max
(ug/g)
0.4
1.3
1.3
Reference (d)
Typical
(ug/g)
0.2
High End
(ug/g)
1.0
Urinary Inorganic Arsenic
Age (years)
0-6
>6
All
Site Data (a)
N
1
14
15
Detect. Freq.
0/1
0/14
0/15
Mean
(ug/L)
20
17
17
Min
(ug/L)
20
10
10
Max
(ug/L)
20
20
20
Reference (d)
Typical
(ug/L)
<10
High End
(ug/L)
20
a Summary statistics calculated using unadjusted values for non-detects
b Brody et al 1994
c Pirkleetal 1998
d NRC 1999
ug = microgram
dL = deciliter (0.1 L)
g = gram
L = liter
13
-------
at other sites have shown that concentrations of metals in the fine fraction can sometimes be
somewhat higher (e.g, 10-30%) than in the bulk sample.
Because all of the samples collected during Phase I and Phase II were bulk samples, an
investigation was performed to determine if the concentration values obtained for the bulk
samples were likely to have values significantly different than if the samples had been sieved.
During the Physical-Chemical Characterization study, a total of 120 samples were selected for a
paired comparison of the concentration in bulk and fine samples, being sure to include samples
with a wide range of arsenic and lead concentrations. All of these samples were analyzed for
arsenic, lead, cadmium and zinc by XRF. For each analyte, only data pairs in which the analyte
was above the detection limit in both the bulk and fine samples were used for correlation
analysis.
The results are shown in Figure 2-4. As seen, the slope of the best fit regression line through the
paired data set is close to 1.0 for lead and zinc, but is slightly higher for arsenic (slope = 1.21)
and cadmium (slope = 1.13). This indicates that the concentration of at least some of the metals
is about 10-20% higher in fines than in bulk samples of soil.
2.4.2 Speciation of Arsenic and Lead
Most metals, including arsenic and lead, can occur in a variety of different chemical and physical
forms. These differences are of potential significance not only because they may help identify
the source of contamination, but also because the toxicity of the metals may differ between
different chemical forms. Therefore, USEPA undertook a study to obtain preliminary data on
the chemical forms of arsenic and lead present in site soils. The details of the sample preparation
and analysis methods are presented in USEPA (1998c). In brief, samples of site soil were
chosen for analysis to span a range of arsenic and lead concentration values. Each sample was
analyzed by electron microprobe analysis (EMPA), and the number and size of different
chemical forms ("phases") of arsenic and lead-bearing particles were measured. From these
data, the fraction of the total mass of arsenic and lead present in each phase was calculated.
Samples evaluated in this way included a set of 22 residential soils evaluated under the Physical
Chemical Characterization Study (USEPA 1998c), plus an additional 20 residential soils
evaluated as part of the Soil Pilot-Scale Characterization Study (USEPA 1999e).
The results are shown in Figures 2-5 and 2-6. As seen, arsenic (Figure 2-5) occurs mainly as
arsenic trioxide (As2O3), with a smaller but significant contribution from lead arsenic oxide
(PbAsO) and a trace of arsenic antimony oxide (AsSbO). In most samples, the majority of all
arsenic-bearing particles are 5-50 um in diameter. Lead (Figure 2-6) occurs in several phases,
including lead arsenic oxide (PbAsO), lead phosphate (Pb Phosphate), and lead manganese oxide
(PbMnO). The concentration of lead in lead arsenic oxide increases dramatically as total lead
concentration increases, suggesting this is the predominant form accounting for elevated lead
levels in yard soils. Levels of lead phosphate and lead manganese oxide also tend to increase as
total lead concentrations increase, but these phases may be secondary weathering products
derived from the lead arsenic oxide. In most samples, the majority of lead-bearing particles are
5-100 um in diameter.
14
-------
4500 -,
4000 -
3500 -
3000 -
2500 -
2000 -
1500 -
1000 -
500 -
0 -
Figure 2-4 Comparison of Concentration in Bulk and Fine Soil
Arsenic
y = 1.2078x
R2 = 0.9096
N=98
Line of identity
0 500 1000 1500 2000 2500
As in Bulk Samples [<2 mm] (ppm)
3000
3500
60
50 -
40 -
I 30
D.
20 -
10 -
Cadmium
y = 1.1273x
R2 = 0.8235
N=95
Line of identity
10 15 20 25 30
Cd in Bulk Samples [<2 mm] (ppm)
35
40
5000
4500 -
4000 -
3500 -
3000 -
2500 -
2000 -
1500 -
1000 -
500 -
Lead
y = 1.0128x
R2 = 0.9515
N= 120
Line of identity
1000 2000 3000 4000
Fb in Bulk Samples [<2 mm] (ppm)
5000
2500
2000 -
§. 1500
-------
FIGURE 2-5 CHEMICAL FORMS OF ARSENIC IN SITE SOILS
Total Arsenic
Concentration (ppm)
^3500
-3000
-2500
-2000
1500
1000
500
s
o.
-------
FIGURE 2-6 CHEMICAL FORMS OF LEAD IN SITE SOILS
2000-,
1800-
§1600-
Total Lead (ppm)
Phase
17
-------
2.5 SELECTION OF CHEMICALS OF POTENTIAL CONCERN
Chemicals of potential concern (COPCs) are chemicals which a) are present at a site, b) occur at
concentrations which are or might be of health concern to exposed humans, and c) are or might
be due to releases from a Superfund site. USEPA has derived a standard method for selecting
COPCs at a site, as detailed in Risk Assessment Guidance for Superfund: Human Health
Evaluation Manual (Part A) (USEPA 1989). In brief, USEPA assumes that any chemical
detected at a site is a candidate for selection as a COPC, but identifies a number of methods that
may be used for determining when a chemical is not of concern and may be eliminated from
further consideration. Each risk assessment may choose to apply some or all of the methods
identified by USEPA to select COPCs, as appropriate.
Data collected during Phase I and Phase II clearly indicated that arsenic and lead were both
chemicals of potential concern at the VBI70 site. However, at that time no systematic evaluation
had been performed to determine whether or not any other chemicals might also be of potential
concern. For this reason, a careful review of the available data was undertaken to determine if
other chemicals should be added to the list (USEPA 1999d). This review is summarized below.
2.5.1 Data Used to Select COPCs
As discussed above, most soil samples collected from the site were analyzed by XRF for only a
few contaminants (mainly arsenic and lead). However, a sub-set of samples were analyzed by
EPA Method 6010 (inductively coupled plasma atomic emission spectroscopy) (ICP) for the full
suite of 23 metals included on USEPA's Target Analyte List (TAL), and these data are the basis
of the COPC selection procedure. The data consist of two sub-sets:
• During Phase I, a total of 44 samples of soil were selected at random for ICP TAL
analysis. The chief purpose of the analysis was to assess the accuracy of the XRF
measurements for arsenic and lead. Because these samples were selected a priori and
without regard to the level of contamination, there are only 9 of these samples that
contain concentrations of arsenic above 100 ppm, with the maximum value being 1,200
ppm. Thus, these samples are helpful in the
COPC selection procedure, but may not necessarily represent the chemicals of concern at
the most contaminated properties.
• During the Risk-Based Sampling Program, USEPA performed an intensive study of
arsenic and lead levels at 8 residential properties in the study area, including 5 properties
with clearly elevated arsenic levels. Two samples from each of these five properties were
selected for ICP TAL analysis, since these samples all contain high levels of arsenic
(6,000 to 12,000 ppm) and are likely to reflect the contaminants most likely to be of
concern.
These data are summarized in Table 2-2. In the case of copper, there is one sample whose
analytical value (14,000 ppm) appears to be clearly inconsistent with all of the other 53 values
(average = 37 ppm, max = 71 ppm). On this basis, the one extreme value for copper was
18
-------
Table 2-2 Data Used to Select Chemicals of Potential Concern
Analyte
ALUMINUM
ANTIMONY
ARSENIC
BARIUM
BERYLLIUM
CADMIUM
CHROMIUM
COBALT
COPPER (a)
LEAD
MANGANESE
MERCURY
NICKEL
SELENIUM
SILVER
THALLIUM
VANADIUM
ZINC
CALCIUM
IRON
MAGNESIUM
POTASSIUM
SODIUM
N
54
54
54
54
54
54
54
54
53
54
54
54
54
54
54
54
54
54
54
54
54
54
54
Detection
Frequency
100%
22%
93%
100%
98%
100%
100%
98%
100%
100%
100%
93%
100%
19%
69%
89%
100%
100%
100%
100%
100%
100%
5%
Summary Statistics
Min (ppm)
4900
2.2
5
91
0.3
0.9
7.2
1.0
12
36
160
0.1
5.9
0.3
0.3
0.2
13
84
1900
7900
1400
1400
300
Max (ppm)
15000
54
9940
1000
1.1
19
99
7.0
71
3550
560
11
96
10
3
19
42
3680
41000
26000
4100
4100
440
Mean (ppm)
8761
6.8
543
251
0.7
5.9
22
4.6
37
712
323
1.0
11
9
0.7
11
21
499
6757
13405
2400
2350
304
(a) Excludes one value (14,000 ppm) that is considered anomalous
19
-------
excluded as an outlier, and screening was based on the remaining samples. All other data values
were used. Non-detects were evaluated using the reported detection limit.
2.5.2 COPC Selection Process
Step 1: Eliminate Chemicals Whose Maximum Value Is Below a Level of Concern
This step involves comparing the maximum detected value in a medium to an appropriate Risk-
Based Concentration (RBC). If the maximum value is less than the RBC, the chemical does not
pose an unacceptable risk and can be eliminated.
The RBCs used in this evaluation were taken from USEPA's Region III Risk-Based
Concentration (RBC) table for residential soil (USEPA 1999c). The value of each RBC depends
on the specified Target Risk level. The Target Risk levels used in this evaluation are 1E-06 for
carcinogenic chemicals and a hazard quotient (HQ) of 1.0 for noncarcinogenic chemicals.
Table 2-3 lists the Region III RBCs for each chemical and identifies those which can and cannot
be eliminated at this step. Based on this screening step, the following chemicals were
eliminated:
• Aluminum • Manganese
• Barium • Mercury
• Beryllium • Nickel
Cadmium • Selenium
• Chromium • Silver
Cobalt • Vanadium
• Copper • Zinc
Step 2. Eliminate Beneficial Minerals
In accord with USEPA (1989), chemicals that are normal constituents of the body and the diet
and are required for good health may be eliminated unless there is evidence that site-specific
releases have elevated concentrations into a range where intakes would be potentially toxic. At
this site, there is no reason to suspect this is the case, so the following chemicals were eliminated
on this basis:
• Calcium
• Magnesium
Potassium
Sodium
Iron was also eliminated on this basis, since the average concentration of iron (13,400 ppm) is
well below the screening level of 23,000 ppm. Additionally, only 1 of 54 samples exceeds the
RBC for iron, and this only by a small amount (26,000 vs. 23,000 ppm).
20
-------
Table 2-3 Comparison of Maximum Values in Soil to Soil Screening Levels
Analyte
Maximum
Concentration (ppm)
Region III Soil
Screening Level
Potential COCP
ALUMINUM
ANTIMONY
ARSENIC
BARIUM
BERYLLIUM
CADMIUM
CALCIUM
CHROMIUM
COBALT
COPPER (b)
IRON
LEAD
MAGNESIUM
MANGANESE
MERCURY
NICKEL
POTASSIUM
SELENIUM
SILVER
SODIUM
VANADIUM
ZINC
15000
54
9940
1000
1.1
19
41000
99
7.0
71
26000
3550
4100
560
11
96
4100
10
3
440
78400
31
0.43
5500
160
78
230
4700
3100
23000
400
1600
23
1600
390
390
no
yes
yes
no
no
no
no
no
no
no
yes
yes
no
no
no
no
no
no
no
no
yes
no
no
COPC=Chemical of Potential Concern
(a) USEPA(1999c)
(b) Excludes one value (14,000 ppm) considered to be anomolous
21
-------
Step 3. Eliminate Chemicals Whose Contribution is Minor Compared to Others
Following Steps 1 and 2, the list of chemicals remaining as potential COPCs was:
• Arsenic
• Antimony
• Lead
• Thallium
Antimony (a non-carcinogenic chemical) was eliminated because the magnitude of the non-
cancer risk which it poses is very small compared to that posed by arsenic. For example, in the
10 samples most contaminated with arsenic, the average non-cancer risk contributed by
antimony is less than 1% of that contributed by arsenic. That is, if antimony were retained and
the non-cancer risk were quantified, the risk would be less than 1% larger than if antimony were
not included. Because an increment of 1% is well within the uncertainty range of the risk
assessment procedure, inclusion of antimony would not change any risk interpretations and
therefore is judged to be unnecessary.
Step 4. Special Investigation for Thallium
Data on thallium available from the existing TAL analyses are internally inconsistent, as shown
below:
Thallium Data from TAL Analyses
Parameter
Method
Mean (ppm)
Max (ppm)
Detection Limit (ppm)
Data Set 1
ICP-Trace
13.5
19
10
Data Set 2
ICP-MS
0.45
0.68
0.1
The basis for this internal inconsistency is not clear. One possibility is that differences in
analytical methods are responsible. Data in Set 1 (collected during Phase I) utilized an analytical
method (ICP-Trace, USEPA Method 6010) that had a relatively high detection limit, and most of
the reported values were near that detection limit. In the second data set (collected during the
risk-based sampling), thallium was analyzed by USEPA Method 6020 (ICP-MS), which has a
much lower detection limit for thallium. In general, the results of the second analysis are
thought to be more reliable, and are in accord with expected thallium levels in background soils
(0.3-0.7 ppm) (ATSDR 1992). However, because it is not certain that the results from the
second analysis are actually more reliable than from the first, a special study was performed in
which thallium levels were measured in 10 site soils, including 6 samples from Set 1 (previously
analyzed by ICP-Trace) and 4 samples from Set 2 (previously analyzed by ICP-MS). Each of
the samples were analyzed for thallium by three analytical methods:
22
-------
• Inductively Coupled Plasma Atomic Emission Spectroscopy [TCP-trace]
(EPA SW-846 Method 601 OB)
• Inductively Coupled Plasma-Mass Spectrometry [ICP-MS]
(EPA SW-846 Method 6020)
Graphite Furnace Atomic Absorption Spectroscopy [GFAA]
(EPA SW-846 Method 7841)
The results of this analysis are provided in Table 2-4. A comparison of thallium levels in site
soils as reported in past and present studies clearly indicate that results contained in the Phase I
Investigation report (UOS 1998a) are biased high and are not reliable, with all of the 10 present
site soil measurements having thallium values lower than 1 ppm. Based on the Region III (EPA
1999c) risk-based concentration for thallium in soil (5.5 ppm), it is concluded that thallium is not
in a range of potential concern, and therefore it was eliminated as a COPC.
2.5.3 Summary: Chemicals Selected as COPCs at VBI70
Based on the methods and data detailed above, the COPCs selected for quantitative evaluation at
the VBI70 site are arsenic and lead. All other chemicals are either not of concern or are present
at levels which contribute minimal risk compared to arsenic.
2.6 PHASE III INVESTIGATION
Results from the Phase I/Phase II sampling programs, supplemented with the data and findings
from the Risk-Based Sampling Program and the Physical Chemical Characterization Program,
indicated that there are properties present in the VBI70 site where arsenic and/or lead could be in
a range of health concern to exposed humans. However, because of the absence of any clear
spatial pattern of soil contamination, the identity and location of such properties could not be
reliably predicted using traditional approaches. For this reason, USEPA undertook a large-scale
sampling program designed to obtain data that would help evaluate health risks to residents in
the area. This program is referred to as the Phase III investigation. The investigation consisted
of four main parts:
Sampling of residential yard soils
Sampling of indoor dust at residences
Sampling of residential vegetable gardens (vegetables and soil)
• Supplemental sampling of soil at local schools and parks
The details of the Phase III sampling program are presented in USEPA (1999d).
Phase III was implemented in two parts. The first part, referred to as Phase Ilia, focused mainly
on properties (including residences, schools, and parks) which had not been investigated in
Phases I or II. The second part, referred to as Phase Illb, consisted mainly of re-sampling at
properties that had previously been sampled in Phase I or II, but for which the data were judged
to be too limited to support clear risk-management decision making. The results of both Ilia and
Illb are summarized bellow.
23
-------
Table 2-4 Comparison of Past and Present Data for Thallium in Soil
Sample ID
C4690CYB-064
C4690CYB-046E
C4711THF-001
C4771VIN-001
D4145FIB10
D4715GYF10
D4050FIB10
D4701JOS10
D4780CBB10
D4785CLF10
Thallium Concentration (ppm)
Past Results
ICP-MS
0.63
0.20
0.33
0.33
ICP-Trace
12
17
11
10 U
16
15
Present Study
ICP-Trace
10 U
10 U
10 U
10 U
10 U
10 U
10 U
10 U
10 U
10 U
ICP-MS
0.70
0.10
0.30
0.30
0.20 U
0.30
0.20
0.10 U
0.50
0.20
GFAA
0.50 U
0.50 U
0.50 U
0.50 U
0.50 U
0.50 U
0.50 U
0.50 U
0.80
0.50 U
U = not detected
24
-------
2.6.1 Residential Soil Sampling
A total of 2,986 residential properties granted EPA access to collect soil samples during the
Phase III program. At each of these properties, 30 surface soil (0-2 inch) grab samples were
collected and combined into three composites samples, each containing 10 grab samples. The
composites were prepared by combining every third grab sample, such that each composite
represents an independent estimate of the yard-wide mean concentration. All composite samples
were dried and thoroughly mixed, and then sieved through a coarse sieve (2 mm) to isolate the
"bulk" fraction. A subset of samples were also sieved through a 250 um screen to isolate the
"fine" fraction (see Section 2.4.1 above). All samples were analyzed for arsenic and lead by
XRF.
Summary statistics for bulk soil samples, based on average values at each property and stratified
by neighborhood, are summarized in Table 2-5. The distributions of arsenic and lead
concentrations across the entire site are shown graphically in Figure 2-7. For arsenic, most
properties (2,471 out of 2,986 = 83%) have average bulk soil concentrations of 50 ppm or less,
with 258 properties (9%) between 50-100 ppm, 183 (6%) between 100-200 ppm, and 74 (2%)
above 200 ppm. For lead, 2,712 (91%) properties have mean lead concentrations lower than 400
ppm, with 266 (9%) between 400-800 ppm and 8 (0.3%) higher than 800 ppm.
The relationship between the concentration of lead and arsenic in residential yard bulk soil
samples is shown in Figure 2-8. As seen, there is a weak correlation between the concentration
of lead and arsenic in soil, with a slope of about 0.6 ppm of lead per ppm of arsenic. However,
this correlation accounts for only a small fraction of the variability in the lead concentration (R2
= 0.089), and inspection of the figure indicates that samples with lead values above 400-600 ppm
occur over a wide range of arsenic values, and are not associated predominantly with those
where arsenic is above 100-200 ppm. This indicates that the main source of lead and the main
source of arsenic in yard soil are not likely to be the same at most yards.
As noted earlier, data collected during one of the physical-chemical characterization studies
(USEPA 1998d) indicated that both arsenic and lead might be slightly enriched in the fine
fraction compared to bulk soil samples. In order to investigate this further, an additional set of
68 residential soil samples collected during the Phase III study were analyzed for lead and
arsenic in both the bulk and fine soil fractions. The results of these 68 samples were combined
with the results from the previous study (see Figure 2-4). The slope of the best fit linear
regression line through the combined data set was 1.21 for arsenic, and 1.09 for lead. In both
cases, these slopes were statistically different from 1.0 (p < 0.001). This confirms the earlier
indication that at this site the concentration of metals are about 10-20% higher in fines than in
bulk samples of soil.
25
-------
Table 2-5 Property Mean Summary Statistics for Phase III Soil Samples
Residential Garden Sampling
ARSENIC
Neighborhood
Clayton
Cole
Elyria
Globeville
Swansea
ALL
Total
Properties
902
796
59
63
1166
2986
Distribution of Yard Average Concentration Values for Arsenic (ppm) (a)
5th
5.5
5.5
5.5
5.5
5.5
5.5
25th
5.5
7.7
8.5
8.5
5.5
5.5
50th
8.7
11.8
12.3
13.8
9.7
10.5
75th
38.3
24.8
22.3
22.3
30.6
30.3
95th
168.0
142.1
97.2
123.3
128.3
144.9
Maximum
758
660
431
297
604
758
LEAD
Neighborhood
Clayton
Cole
Elyria
Globeville
Swansea
ALL
Total
Properties
902
796
59
63
1166
2986
Distribution of Yard Average Concentration Values for Lead (ppm) (a)
5th
76
135
181
171
76
81
25th
106
221
299
257
119
127
50th
140
288
372
332
164
188
75th
193
371
438
482
250
292
95th
337
538
601
633
410
465
Maximum
1131
1130
922
835
776
1131
(a) Yard average is the mean of composites collected from the yard
26
-------
Figure 2-7 Distribution of Property Mean Concentrations in Bulk Soils
2500
2000
in
-------
Figure 2-8 Correlation between Lead and Arsenic
1200
y = 0.6105X + 203.14
R2 = 0.0884
100 200 300 400 500 600
Mean Arsenic Concentration (ppm)
700
800
28
-------
2.6.2 Residential Dust Sampling
As discussed in greater detail in Section 3, one pathway by which residents may be exposed to
contaminants in soil is by transport of outdoor soil into the house where it combines with other
sources to form house dust. When data are absent, USEPA often assumes that the concentration
of contaminants in house dust is the same as in yard soil. However, studies at other sites have
shown that dust levels of metals are often lower in indoor dust than in outdoor soil. Therefore,
USEPA Region VIII undertook a study to define the relationship between arsenic and lead levels
in soil and dust at this site. The details of the sampling and analysis plan are presented in the
Phase III Project Plan (USEPA 1999d). In brief, dust samples were collected from 74 properties.
The locations of these properties were selected to span a range of arsenic and lead levels in soil,
and to provide for spatial representativeness across the site. One composite sample was
collected from each residence by vacuuming dust from 8-14 different living areas within the
house, focusing on those areas judged to be most likely to be a source of dust exposure (e.g.,
bedroom, family room, kitchen, etc.). Samples were collected in October and November, 19991.
The results are shown in Figure 2-9. In the case of lead, two dust samples were excluded as
outliers because they contained lead at concentration values (2,000 ppm and 9,900 ppm) that
were much higher than that observed in yard soil (268 ppm and 320 ppm, respectively). The
source of the high dust lead at these two locations is not known, but could be associated with
releases from indoor leaded paint. Individuals living in these two homes were referred to the
City and County of Denver's Department of Environmental Health to discuss the possible source
of lead in the dust in their home, and the USEPA offered free blood lead testing to all family
members.
As seen, there is only a weak correlation between the level of either arsenic or lead in paired soil
and dust samples (R2 = 0.14 to 0.18, respectively). Nevertheless, the slopes of both regression
lines are statistically different from zero (p < 0.01), with best estimate parameter values as
follows:
Arsenic: Cdust = 0.06-Csoil + 11
Lead: Cdust = 0.34-Csoil + 150
These slope values are somewhat higher than were observed in the Risk-Based sampling data
(arsenic = -0.004 ppm per ppm, lead = -0.014 ppm per ppm) (see Section 2.3.2), perhaps because
of the larger number of samples or perhaps because of differences in sampling and analysis
methods for soil and dust. These slope values are within the range of values that have been
observed at other sites investigated in Region VIII, as shown below:
1 It is not known if dust concentrations at the site vary seasonally, but maximum impact from
yard soil is suspected to occur in the late summer.
29
-------
Figure 2-9 Relation between Concentration in Indoor Dust and Bulk Yard Soil
Arsenic
700
600
500
i. 400
Q.
3
Q
300
y = 0.0577X+ 11.179
R2= 0.1414
100 200 300 400 500
Mean Yard Soil (ppm)
600
700
900
800
700
600
I- 500
Q.
Q
400
300
200
100
0
Lead
Two outliers
excluded
y = 0.3369X+ 150.06
R2= 0.1827
100 200 300 400 500 600 700
Mean Yard Soil (ppm)
800 900
30
-------
Soil-Dust Relationships at Other USEPA Region VIII Sites
Site
Anaconda
Bingham Creek
Butte
Deer Lodge
East Helena
Flagstaff/Davenport
MidvaleOUl
Leadville
Murray Smelter
Sandy City
Sharon Steel
Slope (ppm in dust per ppm in yard soil)
Arsenic
0.31
0.001
0.03
0.10
0.17
Lead
0.43
0.24
-0.01
0.88
0.06
0.04
0.33
0.19
0.13
0.76
2.6.3 Residential Garden Sampling
Another pathway by which residents might be exposed to soil-related contaminants is ingestion
of vegetables grown in home gardens that contain contaminated soil. In order to obtain site-
specific data on this potential exposure route, USEPA Region VIII collected 72 samples of
different types of garden vegetables from 19 different properties around the site. As detailed in
the sampling plan (USEPA 1999d), each vegetable sample was washed in de-ionized water to
minimize the amount of adhering soil. Vegetables were not peeled before analysis. At each
location where a vegetable sample was collected, a co-located sample of garden soil was also
collected. The detailed results for arsenic and lead levels in garden vegetables and soil are
presented in Appendix A.
For arsenic, the mean concentration in vegetables (averaged across all samples) was 0.043 mg/kg
wet weight (43 ng/g ww). A graph showing the relationship between the concentration of
arsenic in garden soil and the corresponding concentration in garden vegetables is shown in
Figure 2-10 (upper panel). As seen, one data point (an onion sample from property 6) appears to
be somewhat higher than expected based on the other samples. The basis for this apparently
high value is not known, but might be attributable to incomplete removal of soil from the sample
prior to analysis, or to an uptake of arsenic into the outer skin of the onion. If that sample is
considered to be un-representative of what would typically be ingested from home-grown garden
vegetables (either because the vegetables would be more thoroughly washed and/or peeled
before being eaten), then the mean concentration of arsenic in vegetables is 30 ng/g wet weight.
The slope of the best-fit regression line through the data (outlier excluded) is quite low (0.0014
mg/kg wet weight per mg/kg in soil), but the slope is statistically different from zero (p <0.001,
R2 = 0.292).
11
-------
Figure 2-10 Relation between Total Arsenic in Garden Vegetables and Garden Soil
05
.*:
"DJ
.
_
-Q
£
c
o
*=
ro
o
O
JD
D.
4 -
2 -
LEAD
Outlier
y = -4E-05x+ 0.0697
R2 = 0.0018
100
200 300 400
Pb Concentration in Bulk Garden Soil (mg/kg)
500
600
32
-------
For lead, the mean concentration across all samples was 0.15 mg/kg wet weight (150 ng/g ww).
A graph showing the relationship between the concentration of lead in garden soil and the
corresponding concentration in garden vegetables is shown in Figure 2-10 (lower panel). As
seen, one data point (a garlic sample from property 11) appears to be substantially higher than
expected based on the other samples. As above, the basis for this apparently high value is not
known, but might be attributable to incomplete removal of soil from the sample prior to analysis.
If that sample is considered to be an outlier and is excluded, then the mean concentration of lead
in vegetables is 62 ng/g wet weight. The slope of the best-fit regression line through the data
(outlier excluded) is -4E-05 mg/kg wet weight per mg/kg in soil, which is not statistically
different from zero (p > 0.5).The relationship between the concentration of arsenic and lead in
garden soil and yard soil is shown in Figure 2-11. For lead (lower panel), the data are based on
the mean garden soil values for the 19 gardens sampled collected during the garden vegetable
sampling effort described above. For arsenic (upper panel), the data set includes the 19
properties described above, plus an additional 17 composite garden soil samples that were
collected following the completion of the Phase III effort. These 17 samples were specifically
selected to include properties with yard soil concentrations of arsenic greater than 100 ppm. As
seen, there is only a weak correlation between arsenic levels in yard soil and garden soil (slope =
0.066, R2 = 0.265), although the slope is statistically different from zero (p < 0.01). For lead,
both the slope (0.60) and the correlation (R2 = 0.410) are somewhat higher than for arsenic, but
the correlation is still rather weak. These results indicate that garden soil is not equivalent to
yard soil, with levels of arsenic and lead tending to be lower in the gardens than in the yards.
This might be because the garden soil is prepared by amending yard soil with clean soil, peat
moss, or other additives that dilute the yard soil contaminant level, or because the source(s) that
have affected the yard did not equally affect the gardens.
2.6.4 Sampling at Schools and Parks
As noted above, data on the levels of arsenic and lead in surface soil were collected at a number
of schools and parks during the Phase I investigation. However, in most cases only a few
samples were collected from each location, and not all schools and parks were sampled.
Therefore, the Phase III Sampling and Analysis Plan included collection of 15-30 supplemental
surface soil grab samples from each school and park within the site where access was granted.
Samples were collected from a total of 7 parks or playgrounds and 15 schools. The results are
shown in Table 2-6.
As seen, concentrations of lead are generally low, with average values ranging from 67-240 ppm.
Mean concentrations of arsenic are also low in most locations (ranging from 11-14 ppm) and
most maximum values are less than 25 ppm. An exception to this pattern occurred at one
property owned by a school (location code S8). At this property, arsenic concentrations in two
soil samples were significantly higher than the other samples (1517 ppm and 70 ppm)2. These
values occur adjacent to each other near a sidewalk, and are surrounded by samples with arsenic
2 These two samples were re-analyzed in triplicate to confirm the data. The mean values for the re-
analyzed samples were 978 ppm and 114 ppm, respectively.
33
-------
Figure 2-11 Relation between Contaminants in Garden Soil and Yard Soil
AC\C\ -,
I 300 -
Q.
'5
£=
0)
ro 200 -
CD
o
0)
I 100-
n
c
ARSENIC
y = 0.0659x+ 15.006
R2 = 0.2648
w ^-
D 100 200 300 400 500 600 7(
Arsenic in Yard Soil (ppm)
DO
700
600 -
I. 500 -
400 -
CD 300 -
200 -
100 -
LEAD
y=0.5982x +32.668
R2 = 0.4099
• •
...-V"
100 200 300 400 500
Lead in Yard Soil (ppm)
600
700
34
-------
Table 2-6 Phase III soil Data for Schools and Parks
Category
School
Park
Code
SI
S2
S3
S4
S5
S6
S7
S8
S9
S10
Sll
S12
S13
S14
S15
PI
P2
P3
P4
P5
P6
P7
N
30
30
30
15
30
15
30
30
30
30
30
30
30
30
30
30
30
30
30
30
30
30
Arsenic (ppm)
Mean (a)
11
12
11
11
11
11
11
67
11
12
11
11
11
11
11
14
12
11
12
11
11
12
Max
12
19
11
13
11
11
12
1517
18
19
17
13
11
11
17
21
18
17
21
12
15
19
Min
11
11
11
11
11
11
11
11
11
11
11
11
11
11
11
11
11
11
11
11
11
11
Lead (ppm)
Mean
95
200
67
83
72
69
104
310
223
235
136
70
120
172
119
215
134
134
218
91
144
240
Max
164
628
126
102
255
95
245
1811
567
359
901
159
354
316
352
398
290
308
294
153
299
614
Min
52
55
52
57
52
52
52
88
61
127
52
52
52
100
52
52
52
52
110
52
67
52
(a) Non-detects evaluated without adjustment
35
-------
concentrations of 17-23 ppm. This suggests there might be a small arsenic "hot spot" at this
location. The property was being developed for use as a school, but no children were present at
the site at the time of sampling, so this was not a source of immediate concern. However, EPA
Region VIII worked with the property owner to ensure that this location was re-landscaped and
covered with a layer of clean topsoil during development so that future exposures would not be
of concern.
2.6.5 Phase III Biomonitoring Program
In keeping with the approach established during Phase II, properties identified during Phase III
which had an arsenic concentration above 400 ppm or a lead concentration above 2,000 ppm
were scheduled for soil removal and replacement3. All residents at such properties were
encouraged to participate in a voluntary biomonitoring program to evaluate if excess exposure
was occurring to arsenic and/or lead. However, only seven individuals chose to participate.
Summary statistics for those individuals are shown in Table 2-7. For convenience, reference
values indicating the typical and upper end of the normal range are also presented.
As seen, similar to the observations obtained during Phase II, there were no cases where
individuals living at the properties scheduled for soil removal had arsenic or lead levels that
exceeded the "background" range typically seen in members of the general population.
Although this data set is too small to draw firm conclusions, the results provide no indication
that exposures at these locations were of immediate health concern.
2.7 DATA SELECTED FOR USE IN THIS RISK ASSESSMENT
The data from the Phase III sampling program were selected for use in this risk assessment
because 1) all Phase III data were collected in accordance with project plans that were developed
with careful consideration of the Data Quality Objectives (DQOs) needed to support risk
assessment calculations, and 2) all data collected during Phase III are accompanied by Quality
Assurance (QA) samples that allow detailed evaluation of the reliability of the data. A detailed
review of these quality assurance data (USEPA 2000e) reveal that the data collected are of high
quality, with adequate accuracy and precision to support a reliable evaluation of human health
risk.
Data collected during Phase I/Phase II were not used because they were collected only with the
intent of identifying locations that exceeded the removal action levels, and were not intended to
support detailed risk calculations or remedial decision making. More specifically, data from
Phase I/Phase II were not used because 1) many samples had elevated detection limits for arsenic
(average = 51 ppm, range = 44 to 800 ppm), 2) the sampling density at each property was
sometimes too low to ensure representativeness, and/or 3) exact sampling locations within a
property were not always clear. However, despite these limitations, it is clear that the data from
Phase I/Phase II and from Phase III are generally similar, each indicating the occurrence of
scattered properties with elevated levels of lead and/or arsenic.
3 The concentration of arsenic that triggered an immediate cleanup during the Phase III program (400 ppm)
was based on the lower limit of the range of concern identified by the USEPA Region VIII toxicologist.
36
-------
Table 2-7 Biomonitoring Data for Residents at Phase III Removal Properties
Demographic Data
Index
Number
1
2
3
4
5
6
7
Age (years)
2
35
36
42
59
75
ND (adult)
Biomonitoring Data
Blood Lead
Value (ug/dL)
1
1
1
5
3
2
NA
Qual.
U
U
U
Hair Arsenic
Value (ug/g)
0.75
0.2
0.35
0.26
0.28
0.41
NA
Qual.
U
U
U
U
U
Urinary Inorganic Arsenic
Value (ug/L)
10
20
10
10
20
10
10
Qual.
U
U
U
U
U
U
U
U=Target analyte not detected
Summary Statistics
Blood Lead
Age (years)
1-5
>=6
All
Site Data (a)
N
1
5
6
Detect. Freq.
0/1
3/5
3/6
Geo. Mean
(ug/dL)
1.0
2.0
1.8
Min
(ug/dL)
1
1
1
Max
(ug/dL)
1
5
5
Reference (b, c)
Typical
(ug/dL)
2.5-4.1
1.5-4.0
2.3-2.8
High End
(ug/dL)
>10
> 10
> 10
Hair Arsenic
Age (years)
0-6
>6
All
Site Data (a)
N
1
5
6
Detect. Freq.
1/1
0/5
1/6
Mean
(ug/g)
0.75
0.3
Min
(ug/g)
0.75
0.20
0.20
Max
(ug/g)
0.75
0.41
0.75
Reference (d)
Typical
(ug/g)
<0.2
High End
(ug/g)
1.0
Urinary Inorc
Age (years)
0-6
>6
All
janic Arsenic
Site Data (a)
N
1
6
7
Detect. Freq.
0/2
0/6
0/7
Mean
(ug/L)
10.0
13.3
12.9
Min
(ug/L)
10
10
10
Max
(ug/L)
10
20
20
Reference (d)
Typical
(ug/L)
<10
High End
(ug/L)
20
a Summary statistics calculated using unadjusted values for non-detects
b Brodyetal 1994
c Pirkleetal1998
d NRC1999
ug = microgram
dL = deciliter (0.1 L)
g = gram
L = liter
37
-------
SECTION 3
EXPOSURE ASSESSMENT
Exposure is the process by which humans come into contact with chemicals in the environment.
In general, humans can be exposed to chemicals in a variety of environmental media (e.g., soil,
dust, water, air, food), and these exposures can occur through one or more of several pathways
(ingestion, dermal contact, inhalation). Section 3.1 provides a discussion of possible pathways
by which area residents and workers might come into contact with contaminants present in
outdoor soil. Section 3.2 describes the basic methods used to estimate the amount of chemical
exposure which humans may receive from direct and indirect contact with contaminants derived
from outdoor soil.
3.1 CONCEPTUAL SITE MODEL
Figure 3-1 presents a conceptual site model for Operable Unit 1 (off-facility soils), showing the
main pathways by which contaminants from current or former smelter activities and other
sources might have reach off-facility soils, and the pathways by which people who live or work
within the VBI70 site boundary might come into contact with those contaminants. This
conceptual model was developed in consultation with local community groups as well as
representatives from the City and County of Denver (CCOD), the Colorado Department of
Public Health and Environment (CDPHE), and the Agency for Toxic Substances and Disease
Registry (ATSDR). Exposure scenarios that are considered most likely to be of concern are
shown in Figure 3-1 by boxes containing a solid circle, while pathways which are judged to
contribute only minor exposures are shown by boxes with an open circle. Incomplete pathways
(i.e., those which are not thought to occur) are shown by open boxes.
The following sections present a more detailed description of each of the exposure scenarios
which are potentially relevant to the risk assessment for Operable Unit 1, and presents the basis
for selecting the pathways that are of sufficient concern that quantitative evaluation is
appropriate.
3.1.1 Potential Sources
The source of soil contamination at off-facility soils at the VBI70 site is not yet established.
Two alternative hypotheses (which are not mutually exclusive) are that the contamination
observed in off-facility soils (mainly residential soils) is due to 1) smelter-related releases (either
airborne fallout from historic operations and/or bulk transport of contaminated waste material),
or 2) application of some sort of pesticide or lawn care product (e.g., various herbicides or
pesticides) that contained arsenic and/or lead. (Such products were commercially available and
widely used in the period from the 1950s to the early 1970s). Studies are currently underway to
obtain data that may help distinguish between these alternatives (USEPA 1999e). However, it is
not necessary to know the source of the contamination in order to evaluate the potential human
health risks from the contamination.
38
-------
Figure 3-1 Conceptual Site Model for Operable Unit 1
Revision 2
/IN IrtlVlllNrtlN 1
SOURCE
Products
w Stack
Tin-rent nr Emissions
Smelters
L-^- Solid Wastes -
| | = Pathway is not complete
1 C*) \ = Pathway is or might be
complete, but is judged to
minor; qualitative evaluat
iRrtiN^r
CONT
UR.1 rrtinwrti 0 rtlNlJ FYPHQTTRF FYPOWD
AMINATED MEDIA ™jf pQ^ULATON
RESIDENT WORKER
~ Off-Facility
Wmd ^ Soils (OUI)
Deposition ^"
>
Wind,
Bulk Soil
Transport
^ So
k
icility *
ils
^^^^ y pp-p-j-oj-^ I pt; ^^^^ m&~ uiiLiii | ^p
- iJirect contact Ingestion ^
^ Dermal ^
4»TV dtnn
Adherence lo T Tn^^sti^n w
Runoff Surface Water ^ Ingestion ^
•^ and Sediments ™ Dermal ^
'^ ^" Dermal ^
*
To be evaluated as a separate operable unit.
be
on
0
•
o
o
o
o
o
0
o
o
o
o
o
o
o
o
o
o
o
| ^ | = Pathway is or might be complete
and could be significant;
quantitative evaluation
39
-------
3.1.2 Migration Pathways
Metals in soil tend to have relatively low mobility. Metals are not volatile, but may enter air
attached to dust particles that are eroded from the yard soil into air by wind or mechanical forces.
This is one pathway by which yard soil may enter a house and contribute to indoor dust.
Another pathway by which yard soil may contribute to indoor dust is by bulk transport of soil
adhering to shoes, clothing, pets, etc. Metals in soil can also leach downward toward
groundwater, and can migrate as a function of surface water erosion. Finally, metals in soil can
be taken up into home-grown garden vegetables.
3.1.3 Exposed Populations and Potential Exposure Scenarios
There are a number of different groups or populations of humans who may directly or indirectly
come into contact with contaminants in area soils. This includes area residents and workers, as
well as individuals who may be exposed at area schools or parks. The following text describes
the scenarios which are considered plausible for each population, and identifies which are likely
to be most important and which are sufficiently minor that they need not be evaluated
quantitatively.
3.2 PATHWAY SCREENING
3.2.1 Residential Exposures
Incidental Ingestion of Soil
Few people intentionally ingest soil. However, it is believed that most people (especially
children) do ingest small amounts of soil that adhere to the hands or other objects placed in the
mouth. In addition, outdoor soil can enter the home and mix with indoor dust, which may also
be ingested during meals or during hand-to-mouth activities. This exposure pathway is often one
of the most important routes of human intake, so it was selected for quantitative evaluation.
Dermal Contact with Soil
Residents can get contaminated soil on their skin while working or playing in their yard. Even
though information is limited on the rate and extent of dermal absorption of metals in soil across
the skin, most scientists consider that this pathway is likely to be minor in comparison to the
amount of exposure that occurs by soil and dust ingestion. This view is based on the following
concepts: 1) most people do not have extensive and frequent direct contact with soil, 2) most
metals tend to bind to soils, reducing the likelihood that they would dissociate from the soil and
cross the skin, and 3) ionic species such as metals have a relatively low tendency to cross the
skin even when contact does occur. These presumptions are supported by screening level
calculations which indicate that dermal exposure of most metals is likely to be no larger (and
probably much lower) than absorption due to soil ingestion (see Appendix B). Based on these
considerations, along with a lack of data to allow reliable estimation of dermal uptake of metals
40
-------
from soil, Region VIII generally recommends that dermal exposure to metals in soils not be
evaluated quantitatively (USEPA 1995c). Therefore, this pathway was not evaluated
quantitatively in this risk assessment.
Inhalation of Soil/Dust in Air
Particles of contaminated soil or dust become resuspended in air, and residents may breathe
those particles both inside and outside their house. However, screening level calculations
(presented in Appendix B) indicate that inhalation of soil particles released to air by wind
erosion is likely to be a small source of risk (less than 0.2%) compared to the risk from
incidental ingestion of soil. Likewise, monitoring data from a large construction project on the
site indicate that mechanical erosion of soil into air is also likely to be of minimal concern.
Based on this, it was concluded that inhalation exposure from airborne particulate matter is a
sufficiently minor contributor to exposure and that it need not be included in the quantitative
evaluation of residential exposure.
Ingestion of Home-Grown Vegetables
If a resident raises vegetables or fruits in a home garden that contains contaminated soil, some
contamination may be taken up from the soil into the vegetable. If so, the resident would be
exposed when those vegetables were consumed. Therefore, this pathway was selected for
quantitative evaluation.
Contact with Surface Water and Sediment
There are no permanent surface water bodies within the VBI70 OU1 site boundary other than the
Platte River. Although it is possible that site-related contaminants may be transported via
surface water runoff and/or groundwater migration to the Platte, it is considered likely that
human exposure levels to site-related contaminants would be relatively low at locations along
the Platte. This is because human contact with surface water and sediments in the river is likely
to be infrequent and relatively low in magnitude, at least compared with the level of exposure to
residential yard soils and indoor dust. On this basis, exposure of residents to surface water and
sediment is considered to be sufficiently minor that quantitative assessment is not warranted for
Operable Unit 1.
Contact with Contaminated Groundwater
At present, there are no data to establish that metals in off-facility soils are a significant source of
groundwater contamination. To the contrary, because the mass of contamination in the soil at
any off-facility location is relatively small, it is not considered likely that off-facility soils are a
significant source to groundwater. In addition, there are no known cases of area residents using
a well for drinking water (the area is supplied with municipal water). On this basis, exposure to
groundwater was not evaluated in this risk assessment.
41
-------
3.2.2 Workplace Exposures
Workers at commercial or industrial locations within the site boundary may be exposed to soil
while working in outdoor locations, so incidental ingestion, inhalation of particulates and/or
dermal contact may occur. As is the case with residents, ingestion exposure is the most
important of these exposure routes. Although only one soil sample has been collected from
commercial properties at the VBI70 site4, extensive sampling has been performed at commercial
properties in the vicinity of the Globe plant (EnviroGroup 2000). This sampling has revealed
that even the highest values detected during the sampling are below a level of potential health
concern for workers, as shown below:
Summary of Soil Data from Commercial Properties in the Vicinity of the Globe Plant
Parameter
Number of commercial properties sampled
Average concentration (ppm)
Highest concentration (ppm) (average across property)
Risk-based concentration for workers (ppm) (see Appendix C)
Arsenic
345
20
96
454
Lead
345
145
1064
1104
Because there is no known reason why commercial properties in the vicinity of the Globe site
should be less contaminated than commercial properties within Operable Unit 1 of the VBI70
site, these data are assumed to be representative of what would be obtained if sampling were to
proceed at commercial properties within OU1. On this basis, it is concluded that sampling at
commercial properties and detailed quantitative risk calculations for workers are not needed at
OU1 of the VBI70 site. Therefore, the worker population is not evaluated further in this risk
assessment.
3.2.3 Exposures at Schools and Parks
Area residents could also be exposed to contaminants in soil at community areas such as schools
or parks. The pathway of primary concern for this scenario is direct ingestion of surface soil. As
above, dermal contact and inhalation of airborne particles may occur, but these pathways are
believed to be minor compared to the ingestion pathway.
As discussed above in Section 2.6.4 (see Table 2-6), concentrations of lead in surface soils from
VBI70 schools and parks are generally low (67-240 ppm) and are below the EPA screening level
(400 ppm) for health concern (USEPA 1994b). Mean concentrations of arsenic in soils from
4 This sample was collected from a location that is currently commercial but is scheduled to be
converted to a school.
42
-------
schools and parks are also low in most locations (ranging from 11-14 ppm) and most maximum
values are less than 25 ppm. An exception to this pattern occurred at one small location on one
property owned by a school, but this apparent hot spot has been addressed by EPA and the
property owner. On this basis, it is concluded that risks to children or other area residents are
not of concern at area schools and parks, and further quantitative evaluation is not needed for
this scenario.
3.3 SUMMARY OF PATHWAYS OF PRINCIPAL CONCERN
Based on the evaluations above, the following exposure scenarios are judged to be of sufficient
potential concern to warrant quantitative exposure and risk analysis:
Exposure Scenarios of Potential Concern for Operable Unit 1
Population
Resident
Exposure Location
Residences
Medium and Exposure
Route
Incidental ingestion of soil
and dust in and about the
home and yard
Ingestion of home-grown
vegetables
Other exposure pathways are judged to be sufficiently minor that further quantitative evaluation
is not warranted
is not warranted.
43
-------
SECTION 4
QUANTIFICATION OF EXPOSURE AND RISK FROM ARSENIC
4.1 OVERVIEW
The USEPA has established standard methods for estimating the level of exposure and risk to
residents from a variety of chemical contaminants in soil. These methods are employed below to
estimate the exposure and risk to residents at the VBI70 site from arsenic in soil. Whenever
possible, site specific data are used in preference to non-site specific default assumptions.
Because the approach used to evaluate exposure and risk from lead is somewhat different than
that used for arsenic, the assessment of lead risks is presented separately in Section 5.
4.2 QUANTIFICATION OF EXPOSURE
4.2.1 Basic Equation
The amount of a chemical which is ingested, inhaled, or taken up across the skin is referred to as
"intake" or "dose", and is usually calculated using an equation of the following general form:
DI = C-(IR/BW)-(EF-ED/AT)
where:
DI = Daily intake of chemical (mg of chemical per kg of body weight per day)
C = Concentration of the chemical in the contaminated environmental medium
(soil, dust, etc.) to which the person is exposed. The units are mg of
chemical per unit of environmental medium (e.g., mg/kg for soil, food,
etc.).
IR= Intake rate of the contaminated environmental medium. The units are
usually kg/day for solid media (soil, dust, food).
BW = Body weight of the exposed person (kg).
EF = Exposure frequency (days/year). This describes how often a person is
likely to be exposed to the contaminated medium over the course of a
typical year.
44
-------
ED = Exposure duration (years). This describes the exposure interval of
concern (how long a person is likely to be exposed to the contaminated
medium).
AT = Averaging time (days). This term specifies the length of time over which
the average dose will be calculated. Usually, two different averaging
times are considered:
• "Chronic" exposure includes averaging times on the scale
of years (typically ranging from 7 years to 70 years). This
exposure duration is used when assessing the non-cancer
risks from chemicals of potential concern.
• "Lifetime" exposure employs an averaging time of 70
years. This exposure interval is selected when evaluating
cancer risks.
In some cases (when the concentration of contaminants is sufficiently high
that short-term exposures might be of concern), a separate evaluation of
"subchronic" exposure (typically from several months to several years), or
"acute" (single dose) exposure may also be performed.
Note that the last three factors (EF, ED, AT) combine to yield a factor between zero and one.
Values near 1.0 indicate that exposure is nearly continuous over the specified averaging period,
while values near zero indicate that exposure occurs only rarely.
For mathematical convenience, the general equation for calculating dose is often written as:
DI = C-HIF
where:
HIF = Human Intake Factor. This term describes the average amount of an
environmental medium contacted by the exposed person each day. The value of
HIF is typically given by:
HIF = (IR/BW)-(EF-ED/AT)
The units of HIF are kg/kg-day for solid media such as soil, dust, and food.
45
-------
4.2.2 Variability and Uncertainty in Exposure Calculations
For every exposure pathway of potential concern, it is expected that there will be differences
between different individuals in the concentration of chemical to which they are exposed, as well
as differences in intake rates, body weights, exposure frequencies and exposure durations. Thus,
there is normally a wide range of average daily intakes between different members of an exposed
population. Because of this, all daily intake calculations must specify what part of the range of
doses is being estimated. Typically, attention is focused on two different parts of the exposure
distribution:
Average or "Central Tendency" Exposure (CTE) is either the arithmetic mean or the
median exposure. It is calculated using the average values for all of the exposure
parameters.
Reasonable Maximum Exposure (RME) is the highest exposure that is reasonably
expected to occur at a site. The intent of the RME is to estimate a conservative exposure
case that is still within the range of possible exposures. This is done by using a
combination of upper-bound estimates for some exposure parameters and average
estimates for some exposure parameters.
This variability in exposure between different members of the population should not be confused
with the difficulties that are often encountered in attempting to estimate either CTE or RME
daily chemical intake levels. These difficulties arise because there are usually insufficient data
to accurately define key exposure parameters such as typical and upper bound intake rates,
exposure frequencies and exposure durations. Thus, the choice of values for average and upper-
bound intakes are often rather uncertain.
4.2.3 Derivation of the Concentration Term
When people are exposed to a chemical in a medium such as soil, the level of exposure and risk
is proportional to the average concentration in the area where exposure occurs. The location
where exposure occurs (e.g., a specific residential yard or house) is usually referred to as the
Exposure Unit (EU), and the average concentration within the EU is referred to as the Exposure
Point Concentration (EPC). Typically, the EPC is estimated based on a set of measured values
of the medium collected from the EU. However, the simple average of the measured values is
only an estimate of the true mean, and the actual value could be either higher or lower. Because
of this uncertainty, the USEPA typically recommends that, for chemicals such as arsenic, the
EPC that is used to calculate exposure and risk be based on either the 95% upper confidence
limit (UCL) of the mean concentration or the maximum concentration (whichever is lower)
(USEPA 1989). Note that this approach is used for both the CTE and the RME exposure
scenarios (USEPA 1992a). The equation used to calculate the UCL depends on what is known
about the underlying distribution of values. In most cases, it is assumed the distribution is right-
skewed, and the equation for a lognormal distribution is used (USEPA 1992a). However, when
the data are described by a distribution that is more nearly symmetric, then the equation for a t-
46
-------
distribution is used (USEPA 1992a). Samples that are below the detection limit are evaluated
using a value equal to one-half the detection limit.
As discussed in the Project Plan for the Phase III sampling (USEPA 1999d), preliminary data
from the site indicated that although the distribution of grab samples from a property was likely
to be approximately lognormally distributed, concentration values in 10-point composite samples
drawn at random from a property were likely to be distributed approximately normally,
indicating that the 95% UCL for a property could be calculated from the mean and standard
deviation of the composite values using the t-equation. In most cases, data would not be
available to test the validity of an assumption of this type. However, at this site, the assumption
can be evaluated based on a supplemental set of data that were collected at 119 of the Phase III
properties. At each of these properties (selected because it was suspected they might be of
concern for short-term noncancer effects), a repeat set of 30 grab samples were collected, and the
samples were analyzed individually rather than being composited. Because of the relatively
large number of grab samples (30) collected at each property, the mean of the 30 samples at each
property may be assumed to be relatively close to true mean at the property. Accepting the mean
of the grab samples as a reliable estimate of the true mean, if the method used for calculating the
95% UCL for each property has worked correctly (i.e., if the assumption of normality of
composite values is correct), approximately 95% of the UCL values should be larger than the
mean of the 30 grab samples. The results of this check are shown in Figure 4-1, expressed as the
frequency distribution of the ratio of the 95% UCL of the composites to the mean of the grab
samples. The expected result is that approximately 95% of the ratios should have a ratio greater
than 1.0. As seen, most samples (83%) had 95% UCL values greater than the mean of the grab
samples (i.e., a ration greater than 1.0), but a total of 20 out of 119 (17%) had 95% UCL values
less than the mean of the grab samples. When this same test was performed using the EPC rather
than the 95% UCL, 75% of the samples had a ratio greater than 1.0. These results suggests that
the approach used to calculate the UCL and the EPC from the composite samples may be slightly
less conservative than intended. However, the actual number of UCL and EPC values that are
not higher than the true mean may be somewhat less than estimated by the ratio test, since some
of the sample means based on the grab samples may be significantly higher than the true means,
even though the N value is 30. Further, in those cases where the 95% UCL or the EPC is not
higher than the mean of the grab samples, the magnitude of the difference is relatively small (less
than a factor of 2 in early all cases). On this basis, it is concluded that use of the t-equation to
calculate 95% UCL and EPC values from the composite samples is reasonable and provides an
adequate margin of safety.
4.2.4 Source of Exposure Parameters
The USEPA has collected a wide variety of data and has performed a number of studies to help
establish reasonable values for many human exposure parameters. The chief sources of these
standard default values are the following documents:
47
-------
Figure 4-1 Comparison of the UCL based on Composites to the Mean of Grab Samples
25 n
20
.CD
| 15
CL
"5
£ 10
E
5
o
21
16
I _n 15
10
9 ^
1
0
| |
^^^
^_^_^
13
1 1
w n i i i i i i i
LOOLOOLOOLOO
OOOt-T-T-T-CM
7
LO
CM
CM
3 3
O LO
LO r-
CM CM
5
o
o
CO
222 2
1
0 0 0
I I I I I I II II
LO O LO O LO O LO
CM LO ["- O CM LO ["-
CO CO CO ^" ^" ^" ^"
Ratio (95% UCL vs Mean of Grabs)
48
-------
1. Risk Assessment Guidance for Superfund (RAGS). Volume I. Human Health
Evaluation Manual (Part A). USEPA 1989.
2. Human Health Evaluation Manual, Supplemental Guidance: " Standard Default
Exposure Factors". USEPA 199 la.
3. Superfund's Standard Default Exposure Factors for the Central Tendency and
Reasonable Maximum Exposure. Draft. USEPA 1993.
4. Exposure Factors Handbook. Volumes I to III. USEPA 1997.
However, for some parameters, there is no guidance and there are few or no data to support the
selection of CTE or RME values, so professional judgement and input from community members
were utilized in some cases.
4.2.5 Quantification of Exposure of Residents to Soil
4.2.5.1 Long-Term (Chronic and Lifetime) Exposure
Basic Equation
Based on the assumption that the concentration of contaminants is approximately equal in
outdoor yard soil and indoor house dust, the USEPA usually evaluates long-term average
residential exposure to soil and dust in a single step. The basic equation is as follows:
IR
DI , = EPC
,
sd
BW AT
Both chronic and lifetime average intake rates are time-weighted to account for the possibility
that an exposed individual may begin exposure as a child (USEPA 1989, 199 la, 1993), as
follows:
IRQ EFQ ED,
TWA-DI^ = EPC J • — - +
LCJ - JiJ-^ J
sa sai BWc AT BWa AT
where:
TWA-DIsd = Time-weighted Daily Intake from ingestion of soil and dust (mg/kg-d)
EPCsd = Exposure Point Concentration of chemical in soil and dust (mg/kg)
IR = Intake rate of soil and dust (kg/day) when a child (IRJ or an adult (IRa)
BW = Body weight (kg) when a child (BWC) or an adult (BWa)
EF = Exposure frequency (days/yr) when a child (EFC) or an adult (EFa)
ED = Exposure duration (years) when a child (EDC) or an adult (EDa)
AT = Averaging time (days)
49
-------
Default Exposure Parameters
Default values and assumptions recommended by USEPA (1989, 1991a, 1993, 1997) for
evaluation of chronic and lifetime residential exposure to soil and dust are listed below:
USEPA Default Parameters for Long-Term Residential Exposure to Soil and Dust
Exposure Parameter
IR as child
IR as adult
BW as child
BW as adult
EF as child or adult
ED as child
ED as adult
AT (noncancer effects)
AT (cancer effects)
Unit
kg/day
kg/day
kg
kg
days/yr
years
years
days
days
CTE
1E-04
5E-05
15
70
234
2
7
9-365
70-365
RME
2E-04
1E-04
15
70
350
6
24
30-365
70-365
CTE=Central Tendency Exposure
RME=Reasonable Maximum Exposure
Based on the exposure parameters above, the time-weighted HIFs for chronic and lifetime
exposure of residents to soil and dust are as follows:
Human Intake Factors (HIFs) for Long-Term
Residential Exposure to Soil and Dust
Residential Exposure
to Soil plus Dust
TWA-chronic (non-cancer)
TWA-lifetime (cancer)
HIF,rt (kg/kg-d)
CTE
1.3E-06
1.7E-07
RME
3.7E-06
1.6E-06
TWA = Time Weighted Average
HIFsd = Human Intake Factor for soil and dust
Adjustment for Unequal Concentrations in Soil and Dust
As noted in Section 2, studies at a number of sites have revealed that the concentration of metals
such as lead and arsenic is often not as high in indoor dust as in outdoor soil. In this situation, it
is necessary and appropriate to evaluate exposure to soil and dust separately, as follows:
50
-------
DIsd = EPCS-HIFS + EPCd-HIFd
where:
EPC = Exposure Point Concentration in soil (EPCS) or in dust (EPCd)
HIF = Human Intake Factor for soil (HIFS) or dust (HIFd)
Derivation of the EPCS Term
As noted above, the EPC for soil is the 95% UCL or the maximum detected value at an exposure
area, whichever is lower. Because measurements of arsenic concentration in soil are based on
bulk soil samples, and exposure is suspected to be associated mainly with the fine fraction, the
value for EPCS is adjusted to account for the enrichment of arsenic in the fine fraction compared
to the bulk fraction as follows (see Section 2.6.1):
ECP,= 1.21-EPC(bulk)
Derivation of the EPCd Term
In general, the concentration of contaminants in dust can be expressed as a function of the
concentration in outdoor bulk soil using the following equation:
EPCd = DO + ksd-EPC(bulk soil)
where:
DO = Concentration in dust (ppm) that is not attributable to yard soil
ksd = Fraction of indoor dust that is derived from outdoor soil
As discussed in Section 2.6.2, in order to derive a reliable site-specific estimate of the relation
between yard soil and indoor dust, paired samples of yard soil and indoor dust were collected at
74 properties at the site. These data are presented in Figure 2-9. For arsenic, the best estimate of
the relation between soil and dust is given by the equation:
EPCd = 0.06-EPCS + 11
That is, DO = 11 ppm and ksd = 0.06.
Estimation of HIFs and HIFd
If fs is defined as the fraction of total intake that is soil, then the HIF for soil and dust intake
(combined) may be separated into its two component parts, as follows:
HIFs = fs-HIFsd
HIFd = (l-fs)-HIF
sd
51
-------
Data are sparse on the relative amounts of soil and dust ingestion by residents, but limited data
support the view that total intake is composed of about 45% soil and 55% dust in children
(Stanek and Calabrese 1992, USEPA 1994a). By extrapolation, this ratio is also assumed to
apply to resident adults. Thus:
fs = 0.45
Combined Final Equation
Combining all the relationships above yields the following final equation:
DIS d = 1.21 • EPC(bulk) • fs • HIF§d + [DO + ksd • EPC(bulk)] • (1 - fs) • HIF§d
Substituting the exposure parameters above and simplifying yields the following:
Equations for Calculating Long-term Average Daily Intake of Arsenic from Soil (mg/kg-d)
Exposure Duration
Chronic (noncancer)
Lifetime (cancer)
CTE
7.5E-07-EPC(bulk) + 7.9E-06
9.8E-08-EPC(bulk) + l.OE-06
RME
2.1E-06-EPC(bulk) + 2.2E-05
9.2E-07-EPC(bulk) + 9.7E-06
4.2.5.2 Sub-Chronic Exposure
In most cases, if chronic noncancer and cancer risks from arsenic are below a level of concern,
risks from shorter term exposures will also be below a level of concern. However, there are
some cases where this may not be so. For example, a child playing in the yard during the
summer months might have soil intakes that are higher than the long-term average, and exposure
might occur preferentially at a sub-area of the yard with arsenic levels that are higher than the
yard-wide average. This is the scenario that is evaluated below.
Basic Equation
The basic equation used to evaluate noncancer risk from this type of scenario is the same as
described previously, except that only soil exposure is considered (not dust). Thus, the basic
equation is:
DIs(sub-chronic) = EPCS • (TR/BW) • (EF / AT)
Each of the inputs are discussed below.
It is assumed that during a relatively short exposure interval (e.g., a period of 1-3 months over
the course of a summer), a child might play in a particular sub-location of the yard where soil
52
-------
concentrations of arsenic are higher than the yard-wide average. Ideally, if data were available
from multiple sampling points within each yard, it would be possible to calculate the average
concentration for multiple sub-areas within the yard, and then evaluate exposure at a selected
sub-location that is elevated compared to the average. The choice of which sub-area to assume
for this exposure is a matter of judgement, but the ninth highest out often subareas (i.e., the 90th
percentile of subarea means) seems reasonable. However, because the soil samples collected
from each residential property during Phase III were all 10-point composites (each representing
the yard wide average) rather than a set of discrete grab samples, these data are not suited for
direct estimation of the distribution of mean concentration values in sub-areas of the yard.
However, a conservative estimate of the 90th percentile sub-area mean may be derived as
follows:
a) The distribution of grab sample values at a property can be estimated by extrapolation
from the detailed grab sampling data collected during the Risk-Based sampling program
and the Phase III grab sampling program. These data are shown in Figure 4-2. As seen,
the standard deviation of the grab samples within a yard tends to increase in proportion to
the mean value in the yard, and the ratio of the standard deviation compared to the mean
(the coefficient of variation, or CV) may be estimated from the slope of the best-fit
regression line through the data:
CV = s / m = 1 .02 (where s = standard deviation and m = mean)
Thus, at any yard where the mean concentration is known (as is the case for all properties
sampled during Phase III), then the standard deviation for a set of grab samples from
within that yard may be estimated using the equation above. Assuming the soil samples
within the yard are distributed approximately lognormally, then the concentration
corresponding to any specified percentile of all yard samples can be calculated.
b) Because the distribution of sub-area means will be narrower than the distribution of
individual grab sample values, use of the 90th percentile of the underlying grab sample
distribution will be a conservative estimate of the 90th percentile of the sub-area means.
The 90th percentile is calculated as:
EPCs(sub-chronic) = C(90th percentile) = GM-GSD
1 282
where:
m
GSD = exp
In
2 :
m + s
2
m
53
-------
Figure 4-2 Coefficient of Variation in Yard Soil Grab Samples
3000
500 1000 1500 2000
Mean Concentration of Arsenic (ppm)
2500
54
-------
Solving and simplifying yields:
EPCs(subchronic) = 2.07-Mean
As above, the EPC (the 95% UCL of the mean or the maximum value) is used as a conservative
estimate of the mean, and this value is adjusted by a factor of 1.21 to account for potential
enrichment of arsenic in the fine fraction compared to the bulk. Thus, the final equation for
estimating the sub-chronic EPC in soil is:
EPCs(subchronic) =1.21- 2.07 • EPC(bulk) = 2.50-EPC(bulk)
Body Weight
The age at which soil ingestion by a child is most likely to occur is not known. Based on
professional judgement, it is suspected that children ages 1-2 years are most at risk, so a body
weight of 12.3 kg (the mean for boys and girls age 1-2) is assumed (USEPA 1997).
Soil Intake Rate
The average amount of soil ingested per day by a child during a short-time exposure is not
known. As noted above, USEPA typically assumes a long-term (six year) average intake of 100
mg/day for a typical (CTE) child and a long-term average intake of 200 mg/day for an RME
child. In the absence of data, it was assumed that the average intake over a period of several
months (sub-chronic) might be about twice as high as the long-term average, so CTE and RME
values of 200 mg/day and 400 mg/day were assumed, respectively. These values are consistent
with the recommendations of USEPA (1997) which identifies 400 mg/day as an upper percentile
for short term soil intake by children.
Exposure Frequency and Averaging Time
For the purposes of this evaluation, the sub-chronic exposure interval of chief concern is
assumed to be the summer months when the child frequently plays outdoors and the soil is not
frozen or snow-covered. In the absence of any site-specific data or USEPA guidance, the
exposure frequency is assumed to be 15 days per month for the CTE child and 25 days per
month for the RME child.
55
-------
Summary of Sub-Chronic Exposure Assumptions
Sub-Chronic Exposure Assumptions
Variable
EPC
Intake rate (mg/day)
Body weight (kg)
Exposure Frequency (days per month)
Averaging Time (days)
HIF (kg/kg-day)
CTE
2.50-EPC(bulk)
200
12.3
15
30
8.1E-06
RME
2.50-EPC(bulk)
400
12.3
25
30
2.7E-05
Combined Final Equation
Substituting the exposure parameters above and simplifying yields the following:
CTE Dl(subchronic) = 2.03E-05-EPC(bulk)
RME Dl(subchronic) = 6.77E-05-EPC(bulk)
4.2.5.3 Acute Pica Exposure
Pica behavior is the intentional ingestion of non-food items, and this may include ingestion of
soil. In this scenario, a child is envisioned as going to some location in the yard and ingesting a
relatively large amount of soil over a short time period. The prevalence of soil pica behavior is
not known, but is assumed to be low in the general population. However, it is plausible that
many children exhibit some pica behavior if studied for long periods of time (USEPA 1997).
For the purposes of this evaluation, the acute pica scenario focuses on the risks from a single
event in which a child ingests a large mass of soil from a small location within a yard.
Basic equation
The basic equation used to evaluate risk from pica behavior is as follows:
DIs(pica) = EPCS-IR/BW
Each of the inputs are discussed below.
Because exposure could occur at any location in the yard, the concentration value used as input
could be the value from any sampling location where a child might play. In order to be
conservative, it is assumed that this concentration could be a high value such as the 95th
56
-------
percentile of the samples within the yard. As noted above, the composite samples collected
during Phase III are not suited to estimating the 95th percentile directly, so the 95th percentile is
estimated assuming a lognormal distribution and a CV of 1.02 (see Figure 4-2). The 95th
percentile value is given by:
1.645
EPCs(acute) = C(95th percentile) = GM-GSD
Solving and simplifying yields:
EPCs(acute) = 2.81-Mean
As above, the EPC for the yard is used as a conservative estimate of the mean. However, in the
case of pica behavior, it is assumed that bulk soil rather than fine soil is ingested. Thus, the EPC
for acute exposure is calculated as follows:
EPCs(acute)= 2.81-EPC(bulk)
Body Weight
The age at which pica behavior is most likely to occur is not known. Based on professional
judgement, it is suspected that children ages 1-2 years are more likely to engage in soil pica that
either younger or older children, so a body weight of 12.3 kg (the mean for boys and girls age 1-
2) is assumed (USEPA 1997).
Soil Intake Rate
Data on soil pica are very sparse. Based on the limited information that is available, USEPA
(1997) has identified 10 grams as a reasonable value for use in an acute exposure assessment.
However, this estimate is based on observations of only one child in one study, so this rate is
considered to be especially uncertain. Because of this uncertainty, two alternative assumptions
were evaluated in this risk assessment:
Assumed Pica Soil Intake (mg/event)
Case
1
2
CTE
5,000
2,000
RME
10,000
5,000
57
-------
Summary of Acute Pica Exposure Assumptions
Acute Pica Exposure Assumptions
Variable
EPC
Intake rate (mg/day)
Case 1
Case 2
Body weight (kg)
CTE
2.81-EPC(bulk)
5000
2000
12.3
RME
2.81-EPC(bulk)
10000
5000
12.3
Combined Final Equation
Substituting the exposure parameters above and simplifying yields the following:
Equations for Calculating Acute Pica Intake of Arsenic from Soil (nig/kg)
Scenario
Case 1
Case 2
CTE
1.14E-03-EPC(bulk)
4.57E-04-EPC(bulk)
RME
2.28E-03-EPC(bulk)
1.14E-03-EPC(bulk)
4.2.6 Quantification of Exposure of Residents to Home-Grown Vegetables
Basic Approach
Two basic options are available for evaluating exposure from home-grown garden vegetables. In
the first approach, data are collected on the concentration of chemical in each type of vegetable
grown at each garden, and these concentrations are multiplied by the intake rate appropriate for
that specific type of vegetable. The second approach is to evaluate the intake from a garden as a
whole, averaging concentration values across all vegetables from that garden, and multiplying by
the estimated total intake rate for home-grown garden vegetables.
Each of these approaches has advantages and limitations. The strength of the first approach is
that it can account for differences in concentration between vegetable types. However, this
approach requires multiple measurements of concentration in each type of vegetable harvested
from each garden, which is usually not possible when samples are collected at a single time
point. In addition, the current or future resident might change the types of crops grown in a
garden, invalidating the type-specific calculations for that garden. The advantage of the second
approach is that it is less sensitive to the specific types of vegetables that happen to be present
when samples are collected, but can be misleading if there are significant variations between
vegetable types. After consideration of both options, the second approach was selected for use
in this risk assessment, since there were not enough data for each vegetable type in each garden
to support reliable type-specific dose calculations. Based on this approach, the equation for
58
-------
evaluation of exposure from ingestion of home-grown vegetables or native vegetation is as
follows:
DI=EPC -IR •
EF
AT )
where:
DI^ = Average daily intake of chemical from home-grown garden vegetables (mg/kg-
day)
= Concentration in garden vegetables (mg/kg wet weight), averaged across types
= Average total intake rate of home-grown garden vegetables (kg wet weight per kg
body weight per day)
EF^ = Exposure frequency to home-grown garden vegetables (days/yr)
ED = Exposure duration (years)
AT = Averaging time (days)
Calculation of EPC^
At each of the 19 properties where garden vegetables samples were collected, the EPC for the
garden vegetables was calculated as the 95% UCL of the mean concentration or the maximum
detected value, whichever was smaller. The 95% UCL was calculated based on the assumption
that the sample values were distributed lognormally, and non-detects were evaluated using an
assumed concentration equal to !/2 the detection limit.
At properties where no garden vegetables samples were collected, the concentration of arsenic in
garden vegetables was estimated using site-specific data on the relationship between arsenic in
yard soil and in garden soil, and between arsenic in garden soil and in vegetable tissues. These
site-specific relationships have been presented previously (see Section 2.6.3), and are
summarized below:
C(garden) = 0.066- C(bulk yard soil) + 15.01
C(vegetable) = 0.0014- C(garden) + 0.0054
Thus, given C(yard soil) at a property (i.e., the bulk soil EPC for the property), the concentration
of arsenic in garden vegetables may be calculated using the equations above.
Adjustment for Organic Content
It is important to recognize that EPA measured the total arsenic content of each vegetable
sample. However, some of the arsenic in vegetables occurs in an organic form that is believed to
be substantially less hazardous than inorganic arsenic. The fraction of total arsenic that is
inorganic varies from vegetable to vegetable, but a mean value is approximately 60% (Schoof et
al. 1999). Thus, for arsenic ingestion in garden vegetables, the following adjustment is used:
59
-------
EPCg^inorganic) = EPC^total) • 0.6
Intake Rates
A number of studies on the intake of homegrown garden vegetables are summarized in the
Exposure Factors Handbook (USEPA 1997). Intake rates vary as a function of several
parameters, including vegetable type, geographic region and age. For this evaluation, the intake
rates were based on the seasonally-adjusted lifetime mean value of home-grown garden
vegetable intakes by people living in the western region of the United States (USEPA 1997).
These intake rates are summarized below:
Home Grown Vegetable Intake (kg wet weight/kg body wt/day)
Percentile
50th (CTE)
95th (RME)
Value
4.92E-04
5.04E-03
In this case, time-weighted averaging of intakes across childhood and adulthood is not needed
since the lifetime average values above are essentially identical to the calculated time-weighted
average values.
These intake rates are based on "household consumption", which reflects the amount of each
type of food item purchased at the store. Thus, these rates do not account for loss of vegetable
material during preparation. Therefore, USEPA (1997) recommends adjusting the intake rates
above to account for the preparation loss, as follows:
IR(adj) = IR(unadjusted) • Loss Factor
The mean preparation loss across multiple vegetable types is 14% (USEPA 1997), so the
adjustment factor is 0.86.
Summary of Exposure Assumptions
These exposure parameters used to evaluate residential exposure from garden vegetable
ingestion are summarized below:
60
-------
Exposure Parameters for Residential Ingestion of Garden Vegetables
Parameter
EPC (inorganic)
IR (kg wet weight/kg body wt/day)
Loss factor
EF (days/yr)
ED (years)
AT (noncancer effects) (days)
AT (cancer effects) (days)
CTE
0.6-EPC(total)
4.92E-04
0.86
350
9
9-365
70-365
RME
0.6-EPC(total)
5.04E-03
0.86
350
30
30-365
70-365
Based on these exposure parameters, the HIF values for exposure of residents to home-grown
vegetables are as follows:
Human Intake Factors for Exposure of Residents to Home-Grown Garden Vegetables
Residential Exposure
to Home-Grown Garden Vegetables
Chronic (non-cancer)
Lifetime (cancer)
HIFPV (kg ww/kg-d)
CTE
4.1E-04
5.2E-05
RME
4.2E-03
1.8E-03
Final Equations
Substituting the exposure parameters above and simplifying yields the following:
Equations for Calculating Exposure from Garden Vegetables
Effect
Non-cancer
Cancer
CTE (mg/kg-day)
2.43E-04-EPCgv(total)
3.13E-05-EPCgv(total)
RME (mg/kg-day)
2.49E-03-EPCgv(total)
1.07E-03-EPCgv(total)
Using the equations above to relate the EPC^total) to bulk yard soil yields the following:
Equations for Calculating Exposure from Garden Vegetables Based on Bulk Yard Soil
Effect
Non-cancer
Cancer
CTE (mg/kg-day)
2.25E-08*EPC(bulk soil) + 6.43E-06
2.89E-09*EPC(bulk soil) + 8.27E-07
RME (mg/kg-day)
2.30*EPC(bulk soil) + 6.59E-05
9.88E-08*EPC(bulk soil) + 2.82E-05
61
-------
4.3 TOXICITY ASSESSMENT
4.3.1 Overview
The objective of a toxicity assessment is to identify what adverse health effects a chemical
causes, and how the appearance of these adverse effects depends on dose. In addition, the toxic
effects of a chemical frequently depend on the route of exposure (oral, inhalation, dermal) and
the duration of exposure (subchronic, chronic or lifetime). Thus, a full description of the toxic
effects of a chemical includes a listing of what adverse health effects the chemical may cause,
and how the occurrence of these effects depends upon dose, route, and duration of exposure.
The toxicity assessment process is usually divided into two parts: the first characterizes and
quantifies the non-cancer effects of the chemical, while the second addresses the cancer effects
of the chemical. This two-part approach is employed because there are typically major
differences in the time-course of action and the shape of the dose-response curve for cancer and
non-cancer effects.
Non-Cancer Effects
Essentially all chemicals can cause adverse health effects if given at a high enough dose.
However, when the dose is sufficiently low, typically no adverse effect is observed. Thus, in
characterizing the non-cancer effects of a chemical, the key parameter is the threshold dose at
which an adverse effect first becomes evident. Doses below the threshold are considered to be
safe, while doses above the threshold may cause an effect.
The threshold dose is typically estimated from toxicological data (derived from studies of
humans and/or animals) by finding the highest dose that does not produce an observable adverse
effect, and the lowest dose which does produce an effect, following some specified duration of
exposure. These are referred to as the "No-observed-adverse-effect-level" (NOAEL) and the
"Lowest-observed-adverse-effect-level" (LOAEL), respectively. The threshold is presumed to
lie in the interval between the NOAEL and the LOAEL. However, in order to be conservative
(protective), non-cancer risk evaluations are not based directly on the threshold exposure level,
but on a value referred to as the Reference Dose (RfD). The RfD is a duration-specific estimate
(with uncertainty spanning perhaps an order of magnitude) of a daily exposure to the human
population (including sensitive subgroups) that is likely to be without an appreciable risk of
deleterious effects, even in sensitive individuals.
The RfD is derived from the NOAEL (or the LOAEL if a reliable NOAEL is not available) by
dividing by an "uncertainty factor". If the data are from studies in humans, and if the
observations are considered to be very reliable, the uncertainty factor may be as small as 1.0.
However, the uncertainty factor is normally at least 10, and can be much higher if the data are
limited. The effect of dividing the NOAEL or the LOAEL by an uncertainty factor is to ensure
that the RfD is not higher than the threshold level for adverse effects. Thus, there is always a
"margin of safety" built into an RfD, and doses equal to or less than the RfD are nearly certain to
be without any risk of adverse effect. Doses higher than the RfD may carry some risk, but
62
-------
because of the margin of safety, a dose above the RfD does not mean that an effect will
necessarily occur.
Cancer Effects
For cancer effects, the toxicity assessment process has two components. The first is a qualitative
evaluation of the weight of evidence that the chemical does or does not cause cancer in humans.
Typically, this evaluation is performed by the USEPA, using the system summarized in the table
below:
Cancer Weight of Evidence Categories
Category
A
Bl
B2
C
D
Meaning
Known human carcinogen
Probable human
carcinogen
Probable human
carcinogen
Possible human
carcinogen
Cannot be evaluated
Description
Sufficient evidence of cancer in humans.
Suggestive evidence of cancer incidence in humans.
Sufficient evidence of cancer in animals, but lack of data
or insufficient data from humans.
Suggestive evidence of carcinogenicity in animals.
No evidence or inadequate evidence of cancer in
animals or humans.
For chemicals which are classified in Group A, Bl, B2, or C, the second part of the toxicity
assessment is to describe the carcinogenic potency of the chemical. This is done by quantifying
how the number of cancers observed in exposed animals or humans increases as the dose
increases. Typically, it is assumed that the dose response curve for cancer has no threshold,
arising from the origin and increasing linearly until high doses are reached. Thus, the most
convenient descriptor of cancer potency is the slope of the dose-response curve at low dose
(where the slope is still linear). This is referred to as the Slope Factor (SF), which has
dimensions of risk of cancer per unit dose.
Estimating the cancer Slope Factor is often complicated by the fact that observable increases in
cancer incidence usually occur only at relatively high doses, frequently in the part of the dose-
response curve that is no longer linear. Thus, it is necessary to use mathematical models to
extrapolate from the observed high dose data to the desired (but unmeasurable) slope at low
dose. In order to account for the uncertainty in this extrapolation process, USEPA typically
chooses to employ the upper 95th confidence limit of the slope as the Slope Factor. That is,
there is a 95% probability that the true cancer potency is lower than the value chosen for the
Slope Factor. This approach ensures that there is a margin of safety in cancer risk estimates.
63
-------
4.3.2 Toxicity Summary for Arsenic
The toxic effects of arsenic have been reasonably well established, based mainly on studies of
humans exposed to elevated levels of arsenic from a variety of sources. The findings from these
studies are summarized briefly below.
Acute Noncancer Effects
The estimated LD50 (the dose that causes 50% lethality) from arsenic ingestion is about 1-4
mg/kg in humans (USEPA 200Id). Oral exposure to non-lethal but high acute doses of arsenic
produces marked irritation of the gastrointestinal tract, leading to nausea and vomiting. Other
signs may include neuritis and vascular effects (USEPA 200Id). Incidents of acute arsenic
toxicity are generally associated with accidental exposures, but may sometimes occur from
ingestion of herbal medicines.
USEPA has reviewed available data on the acute and short-term toxicity of arsenic (USEPA
200If), and has concluded that a large cross-sectional study of arsenic-induced skin lesions in
children (Mazumder et al.1998) identifies aNOAEL of 0.015 mg/kg/day. Because this value is
based on observations in a large number of individuals, including those who are likely to be
sensitive, an uncertainty factor of 1 is recommended, yielding an RfD of 0.015 mg/kg-day
(USEPA 2001f). Likewise, ATSDR has reviewed the available data, and noted that a study by
Mizuta et al. (1956) reported multiple signs of acute arsenic toxicity in people exposed to
arsenic-contaminated soy sauce. The exposure level causing the effects was estimated to be
about 3 mg/day, which corresponds to a dose of about 0.05 mg/kg-day. Based on this study,
ATSDR (2000) derived an acute oral MRL of 0.005 mg/kg/day by using a safety factor of 10 to
extrapolate from the LOAEL to a NOAEL. ATSDR recommends use of this MRL as a
screening value.
Sub chronic Noncancer Effects
Symptoms resulting from sub-chronic ingestion of lower doses of arsenic often begin with a
vague weakness and nausea. As exposure continues, symptoms become more characteristic and
may include signs such as diarrhea, vomiting, anemia, injury to blood vessels, damage to kidney
and liver, and impaired nerve function that leads to "pins and needles" sensations in the hands
and feet. The USEPA has developed a subchronic oral RfD for arsenic of 6E-03 mg/kg-d
(USEPA 1995b). This value is based on an estimated LOAEL of 0.06 mg/kg-day in humans
(both children and adults) exposed to arsenic for periods of time from six months up to about 15
years. An uncertainty factor of 10 is used to account for extrapolation from a LOAEL to a
NOAEL.
Chronic Noncancer Effects
Chronic exposure to arsenic is associated with all of the effects noted above. In addition, after
exposure continues for a sufficient period of time, an unusual pattern of skin abnormalities,
64
-------
including dark and white spots and a pattern of small "corns" may occur, especially on the palms
and soles (ATSDR 2000, USEPA 200Id).
The average daily intake of arsenic that produces these skin effects varies from person to person.
In a large epidemiological study in Taiwan, Tseng et al. (1968) reported skin lesions in humans
exposed to chronic oral doses of 0.014 mg/kg-day or higher. Intake was through the drinking
water. These effects were not observed in a control population ingesting 0.0008 mg/kg-day.
The USEPA used the NOAEL of 0.0008 mg/kg/day for skin and vascular lesions (Tseng et al.
1968) to derive a chronic oral RfD of 3.0E-04 mg/kg/day (IRIS 2000). The NOAEL was divided
by an uncertainty factor of 3 to account for both the lack of data to preclude reproductive toxicity
as a critical effect and to account for some uncertainty in whether the NOAEL of the critical
study accounts for all sensitive individuals (IRIS 2000). Confidence in the RfD is rated medium.
A higher rating was not given due to uncertainties in dose estimates and other problems in the
epidemiological data base (IRIS 2000).
Cancer Effects
There is strong evidence from a number of human studies that oral exposure to arsenic increases
the risk of skin cancer (USEPA 1988, NRC 1999, ATSDR 2000, USEPA 200Id). The most
common type of cancer is squamous cell carcinoma, which appears to develop from some skin
corns. In addition, basal cell carcinoma may also occur, typically arising from cells not
associated with the corns. Although these cancers may be easily removed, they can be painful
and disfiguring and can be fatal if left untreated. More recent data indicate that chronic oral
arsenic exposure may also increase the risk of internal cancers, including cancer of the bladder
and lung (NRC 1999, USEPA 200Id).
Based on a study of skin cancer incidence in Taiwanese residents exposed mostly to arsenic in
drinking water (Tseng et al. 1968), the USEPA has calculated a unit risk of 5E-5 (jig/L)"1
corresponding to an oral slope factor of 1.5 (mg/kg/day)"1 (IRIS 2000). Assuming a water intake
of 2 L/day by a 70-kg person, a concentration of 10 ug/L corresponds to a lifetime excess cancer
risk of about 4E-04. The NRC (1999) has reviewed a number of alternative approaches for
quantification of cancer risk at low doses, and noted that the risk estimates depend heavily on the
mathematical approach employed as well as the cancer data set utilized. Based on the incidence
of bladder cancer in males in Taiwan, several different methods yield estimates of the EC01 (the
concentration in water that results in a 1% increase in excess lifetime cancer risk) of about 400-
450 ug/L. If the dose response curve is assumed to be linear and to have no threshold, this
corresponds to an oral slope factor of about 0.8-0.9 (mg/kg-day)"1, slightly lower than the value
based on skin cancer. Assuming a water intake of 2 L/day by a 70-kg person, this slope factor
would correspond to a risk of about 2E-04 at an exposure concentration of 10 ug/L.
More recently, Morales et al. (2000) used a number of alternative risk models to analyze the
incidence of bladder and lung cancer in the Taiwanese population exposed to arsenic in drinking
water. USEPA (200Id) reviewed these results and, after consultation with the authors,
concluded that a model without a reference population was most appropriate, since the available
65
-------
reference population (urban residents in Taiwan) are not considered to be a good control group
for the rural workers exposed to the high arsenic levels. For people exposed at a concentration
of 10 ug/L, the risk model preferred by USEPA yielded estimates of excess cancer risk of 0.6E-
04 to 3.0E-04 for an average individual, and from 1.3E-04 to 6.1E-04 for an individual at the
90th percentile of the risk distribution. These risk estimates are similar to the risk estimates
derived previously by USEPA and by NRC (1999). This indicates that the slope factor of 1.5
(mg/kg-day)"1 based on the incidence of skin cancer in the Taiwanese population is likely to be
generally appropriate for estimation of risks from other cancer types as well. This is probably
because most individuals who develop arsenic-induced bladder or lung cancer also develop skin
cancer, and so the total number of people with any type of arsenic induced cancer is similar to
the number with skin cancer.
Potential Beneficial Effects
Several studies in animals suggest that low levels of arsenic in the diet may be beneficial for
reproduction and normal postnatal development. The USEPA (1988) reviewed the evidence and
concluded that the essentiality of low levels of arsenic in animals has not been established, but is
plausible. The NRC (1999) also reviewed the evidence and noted that studies to date do
establish that arsenic supplementation of low-arsenic semi-synthetic diets prevents the
occurrence of abnormal reproductive or decreased growth in animals, but that there is no proof
that arsenic is an essential element in humans or that it is required for any biochemical process.
If arsenic is beneficial or essential in animals, it is also likely to be so for humans. Based on the
animal data, the estimated beneficial dose for humans would be approximately 10 to 50 jig/day
(USEPA 1988). This level of arsenic intake is usually provided in a normal diet, and no cases of
arsenic deficiency in humans have been reported (NRC 1999, ATSDR 2000).
Summary of Toxicity Values for Arsenic
Based on the information reviewed above, this risk assessment utilized the following toxicity
factors for ingested arsenic:
Arsenic Toxicity Factors Utilized in the Risk Assessment
Toxicity Factor
Acute RfD
Subchronic RfD
Chronic RfD
Oral Slope Factor
Value
0.015 mk/kg-day
0.060 mg/kg-day
0.0003 mg/kg-day
1.5 (mg/kg-day)-1
Source
USEPA (200 If)
USEPA (1995b)
IRIS 2000
IRIS 2000
RfD=Reference Dose
66
-------
4.3.3 Adjustments For Relative Unavailability
As discussed in USEPA (1989), most oral RfD and SF values developed by USEPA are based on
the empirical relationship between the occurrence of toxic effects and the amount of chemical
ingested, and the amount of chemical that is actually absorbed into the body is not explicitly
considered. Thus, if it is expected that the absorption of a chemical from an on-site medium is
significantly different than from the medium used in the study supporting the RfD or SF, then it
is appropriate to adjust the RfD or SF to account for this difference in absorption. This
adjustment increases the accuracy of the subsequent risk calculations while still being protective
of public health.
The ratio of the absorption fraction for a chemical in site medium compared to the medium used
in the key toxicity studies is referred to as the Relative Bioavailability (RBA). If reliable
estimates of RBA are available for chemicals of potential concern in site media, these can be
used to adjust the default RfD and SF values as follows:
RfDadj = RfDdefault/RBA
SFadj = SFdefault • RBA
In the case of arsenic, all of the oral RfDs as well as the oral SF are based on studies of humans
exposed to arsenic either in drinking water or in other readily absorbable forms. Thus, solid
forms of arsenic in site soils may be less well-absorbed and require adjustments in the toxicity
factors to derive appropriate estimates of toxicity.
In order to investigate the relative bioavailability of arsenic in site soils, USEPA performed a
study in which five separate samples were fed to swine for 12 days. Swine were selected as the
test species because it is believed the gastrointestinal system (and hence the behavior of ingested
arsenic) in swine is similar to that in humans. The details of the study design and of the findings
are presented in a separate report (USEPA 200Ib). In brief, the study found that arsenic in site
soils was less well absorbed than a readily soluble form of arsenic (sodium arsenate), with RBA
values for individual samples of site soil ranging from about 0.18 to 0.45. Because it is believed
that these differences in RBA reflect mainly experimental variation, a single site-wide RBA
value was derived by calculating the 95% upper confidence limit of the mean RBA for all of the
site soils tested. The resulting value was 0.42.
4.4 RISK CHARACTERIZATION FOR ARSENIC
4.4.1 Basic Approach
Cancer Risk
The risk of cancer from exposure to a chemical such as arsenic is described in terms of the
probability that an exposed individual will develop cancer because of that exposure by age 70.
For each chemical of concern, this value is calculated from the daily intake of the chemical from
67
-------
the site, averaged over a lifetime (DIL), and the slope factor (SF) for the chemical, as follows
(USEPA 1989):
Cancer Risk = 1 - exp(-DIL • SF)
In most cases (except when the product of DIL-SF is larger than about 0.01), this equation may be
accurately approximated by the following:
Cancer Risk = DIL-SF
Because of the uncertainty in both the exposure term and the slope factor term, USEPA guidance
recommends that all cancer risk estimates be expressed to only one significant figure (USEPA
1989).
The level of total cancer risk that is of concern is a matter of personal, community and regulatory
judgement. In general, it is the policy of the USEPA that remedial action is not warranted where
excess cancer risks to the RME individual do not exceed a level of 1E-04 (USEPA 1991b).
Noncancer Risk
The potential for noncancer effects from exposure to a chemical is evaluated by comparing the
estimated daily intake of the chemical over a specific time period (chronic, sub-chronic, acute)
with the RfD for that chemical derived for the corresponding exposure period. This comparison
results in a noncancer Hazard Quotient, as follows (USEPA 1989):
HQ = DI / RfD
where:
HQ = Hazard Quotient
DI = Daily Intake (mg/kg-day)
RfD = Reference Dose (mg/kg-day)
Because of the uncertainty in both the exposure term and the reference dose term, USEPA
guidance recommends that all HQ values be expressed to only one significant figure (USEPA
1989). If the HQ for a chemical is equal to or less than one (1E+00), it is believed that there is
no appreciable risk that noncancer health effects will occur, even in sensitive individuals. If an
HQ exceeds 1E+00, there is some possibility that noncancer effects may occur, although an HQ
above 1E+00 does not indicate an effect will definitely occur. This is because of the margin of
safety inherent in the derivation of exposure estimates and RfD values. However, the larger the
HQ value, the more likely it is that an adverse effect may occur.
68
-------
4.4.2 Risks from Soil and Dust
4.4.2.1 Cancer Risk
Cancer risks from exposure of residents to arsenic in yard soil and indoor house dust were
calculated for each property using the basic equations described above. The exposure point
concentration (EPC) for soil at each property was the 95% UCL of the mean value of the three
10-point composite values or the maximum composite value (whichever was lower). The 95%
UCL of the mean was calculated based on an assumption that the distribution of 10-point
composite values at a property is likely to be approximately normally distributed (USEPA
1999d). Non-detects were evaluated by assuming a value equal to one-half the detection limit.
The concentration in dust was calculated from the soil exposure point concentration as described
in Section 4.2.5.1 (above). The resulting risk estimates are presented in Table 4-1.
For CTE exposure conditions, most properties have estimated excess cancer risks for exposures
due to arsenic in soil plus dust that range from 1E-06 to 1E-05 (5th to 95th percentiles), with a
maximum value of 9E-05. For RME exposure conditions, most properties have risks that range
from 9E-06 to 1E-04 (5th to 95th percentiles), with 92 properties having risks of 2E-04 or
higher. The highest RME risk value was 8E-04. As shown in Table 4-1, the spatial pattern of
properties with arsenic RME cancer risk levels of 2E-04 or higher is approximately uniform
across the site, with a frequency of about l%-4% in each neighborhood.
In interpreting these risk estimates, it is important to recognize that arsenic is a naturally
occurring element in soil. Figure 4-3 presents the distribution of mean arsenic concentrations in
residential properties sampled during Phase III. As seen, the distribution is fairly well-
characterized as the sum of two different lognormal distributions with the following statistics5:
Parameters of the Best Fit Lognormal Distributions
Statistic
GM (ppm)
GSD
AM (ppm)
Stdev (ppm)
95th (ppm)
Distribution 1
7.3
1.5
8.0
3.6
15
Distribution 2
28.4
3.5
62.5
123
224
In order to estimate the parameters of the two lognormal distributions, the data were log-transformed and
fit to an equation of the following form: Cone ~ k-N(a,b) + (l-k)-N(c,d), where a and b are the log-mean and log-
standard deviation of distribution 1, and c and d are the log-mean and log-standard deviation of distribution 2. The
parameter k is the mixing fraction of the two distributions. Parameter values (a, b, c, d, and k) were derived using
least square regression on order statistics.
69
-------
Table 4-1 Estimated Cancer Risk from Arsenic in Soil and Dust
Neighborhood
Clayton
Cole
Elyria
Globeville
Swansea
All Neighborhoods
Number of
Properties
Evaluated
902
796
59
63
1166
2986
Number and Percent of Properties Within the Specified Risk Range
CTE Cancer Risk
<=1E-05
858
95%
772
97%
58
98%
61
97%
1132
97%
2881
96%
2E-05 - 1E-04
44
5%
24
3%
1
2%
2
3%
34
3%
105
4%
2E-04 - 1E-03
> 2E-03
RME Cancer Risk
<=1E-05
479
53%
344
43%
17
29%
25
40%
610
52%
1475
49%
2E-05 - 1E-04
385
43%
429
54%
41
69%
36
57%
528
45%
1419
48%
2E-04 - 1E-03
38
4%
23
3%
1
2%
2
3%
28
2%
92
3%
> 2E-03
CTE=Central Tendency Estimate
RME=Reasonable Maximum Exposure
70
-------
Figure 4-3 Distribution of Arsenic Values in Phase III Soils
Combined Distribution
Background Distribution
Histogram of Observed
Values
Distribution of
Elevated Levels
Natural Logarithm of Arsenic Concentration
71
-------
Although this analysis cannot reveal the basis for the two-component nature of the distribution,
the most straightforward interpretation is that the first distribution represents background, and
the second distribution represents an extra amount of arsenic present in some yards due to some
other (non-background) source. If so, the best estimate of the average background level of
arsenic is about 8 ppm, although some background levels may range up to about 15 ppm or
higher. Based on this, risks from naturally occurring levels of arsenic probably range from about
1E-06 for an average (CTE) person up to about IE-OS for an upper-bound (RME) individual.
4.4.2.2 Chronic Noncancer Risks
Estimated risks of non-cancer health effects from chronic exposure to arsenic in soil and dust are
shown in Table 4-2. For individuals with CTE exposure, risks at most properties fall between
2E-02 and 2E-01 (5th to 95th percentile), while individuals with RME exposure have risks that
lie mainly between 5E-02 and 6E-01. These results indicate that risk of noncancer effects from
chronic exposure is below a level of concern for most individuals at most locations. However, a
total of 20 properties have RME HQ values of 2E+00 or higher, with a maximum value of
4E+00. These locations where noncancer risks enter a range of concern (HQ > 1E+00) are also
above the usual level of concern (1E-04) for cancer.
4.4.2.3 Sub chronic Noncancer Risks
Estimated risks of non-cancer health effects from sub-chronic exposure of area children to
arsenic in soil are shown in Table 4-3. As seen, the incidence of properties with subchronic HQ
values above 1E+00 is relatively low (2 out of 2,986 = 0.07% for CTE individuals, 53 out of
2986 = 1.8% for RME individuals). The maximum RME HQ value was 7E+00. All of the
locations where subchronic noncancer risks enter a range of concern (HQ > 1E+00) are also
above the usual level of concern (1E-04) for cancer.
4.4.2.4 Noncancer Risks from Acute Pica Behavior
Because of the substantial uncertainty which exists in most of the input parameters for the acute
pica scenario, it is not possible to specify a single set of inputs that are "best". Rather, a range of
HQ values were calculated for two different combinations of soil intake and RfD values:
Soil Intake and Arsenic Toxicity Factors for Calculating Non-Cancer Risks
for two Acute Pica Scenarios
Variable
Soil intake (mg/day)
Acute RfD (mg/kg-d)
Casel
CTE
5000
RME
10000
0.005
Case 2
CTE
2000
RME
5000
0.015
72
-------
Table 4-2 Estimated Chronic Noncancer Risk from Arsenic in Soil and Dust
Neighborhood
Clayton
Cole
Elyria
Globeville
Swansea
All Neighborhoods
Number of
Properties
Evaluated
902
796
59
63
1166
2986
Number and Percent of Properties Within the Specified Risk Range
CTE Hazard Quotient
<=1
901
100%
796
100%
59
100%
63
100%
1166
100%
2985
100%
2-5
1
0.1%
0
0%
0
0%
0
0%
0
0%
1
0%
6-10
—
—
—
—
—
..
—
—
—
..
—
—
> = 11
—
—
—
—
—
..
—
—
—
..
—
—
RME Hazard Quotient
<=1
895
99%
786
99%
59
100%
63
100%
1163
100%
2966
99%
2-5
7
0.8%
10
1.3%
0
0%
0
0%
3
0.3%
20
0.7%
6-10
—
—
—
—
—
..
—
—
—
..
—
—
> = 11
—
—
—
—
—
..
—
—
—
..
—
—
CTE=Central Tendency Estimate
RME=Reasonable Maximum Exposure
73
-------
Table 4-3 Estimated Subchronic Noncancer Risks from Arsenic in Soil
Neighborhood
Clayton
Cole
Elyria
Globeville
Swansea
All
Number of
Properties
Evaluated
902
796
59
63
1166
2986
Number and Percent of Properties Within the Specified Risk Range
CTE Hazard Quotient
<=1
900
100%
796
100%
59
100%
63
100%
1166
100%
2984
100%
2-5
2
0.2%
0
0%
0
0%
0
0%
0
0%
2
0.1%
6-10
> = 11
RME Hazard Quotient
<=1
881
98%
777
98%
58
98%
62
98%
1155
99%
2933
98%
2-5
19
2%
19
2%
1
2%
1
2%
11
1%
51
2%
6-10
2
0.2%
0
0.0%
0
0.0%
0
0.0%
0
0.0%
2
0.1%
> = 11
74
-------
It should be understood that these cases represent an uncertainty range, and that the "true" acute
risk from pica behavior could lie anywhere in the interval. Indeed, it is quite possible that the
true value even lies outside the range, since the actual distribution of pica soil intakes is not
known.
The results are summarized in Table 4-4. As seen, the screening calculations above suggest that
a large number of properties (ranging from 662 to 1841, depending on which set of input
assumptions is deemed to be most appropriate) are of potential concern for the RME acute pica
scenario. In the absence of reliable data on the magnitude and frequency of soil pica intake, and
considering that national decisions continue on the most appropriate acute RfD for arsenic, it is
difficult to judge which (if any) of these properties should be considered to be an authentic acute
health risk to children. In this regard, it should be noted that even though many people are
exposed to arsenic levels in soil that are predicted to be of acute concern, both within the VBI70
site and elsewhere across the country and around the world, to the best of USEPA's knowledge,
there has never been a single case of acute arsenic toxicity reported in humans that was
attributable to arsenic in soil. Thus, these results for the acute pica scenario are considered to be
especially uncertain, since they predict a very substantial risk for which there is no corroborating
evidence.
4.4.3 Risks from Home-Grown Vegetables
As discussed previously (see Section 2.6.3), a total of 72 different samples of home-grown
garden vegetables were collected from 19 different properties across the site. At each property,
the 95% UCL of the mean concentration of arsenic averaged across all vegetables samples from
the garden was calculated using an assumption of lognormality. Non-detects were evaluated by
assuming a value equal to one-half the detection limit. The EPC was then the 95% UCL or the
maximum detected value (whichever was lower). As noted above, the concentration of inorganic
arsenic was assumed to be 60% of total arsenic concentration.
Cancer and non-cancer risks from ingestion of home-grown vegetables at each of the 19
properties sampled were calculated by combining the EPC value with the estimated intake of
garden vegetables described in Section 4.2.6, and calculating HI values and excess cancer risks
as described in Section 4.4.1. The results are summarized in Table 4-5.
As seen, for individuals whose intake of home-grown garden vegetables is average (CTE) for the
western United States, neither non-cancer nor cancer risks enter a range of concern at any
property tested. For individuals whose intake is at the upper-bound (RME) of the distribution of
garden vegetable consumption, cancer and non-cancer risks do enter a range of potential concern
for two properties, as discussed below:
• At Property 6, a number of vegetables had arsenic concentration values that were higher
than in samples from most other properties. The concentrations of arsenic in the garden
75
-------
Table 4-4 Estimated Acute Noncancer Risk from Pica Behavior
Exposure
Assumptions
Casel
Case 2
Number and Percent of Properties Within the Specified Risk Range
CTE Hazard Quotient
<=1
1475
49%
2692
90%
2-5
949
32%
268
9%
6-20
432
14%
26
1%
>20
130
4%
0
0%
Total > 1
1511
51%
294
10%
RME Hazard Quotient
<=1
1145
38%
2324
78%
2-5
580
19%
487
16%
6-20
328
11%
162
5%
>20
933
31%
13
0%
Total > 1
1841
62%
662
22%
Case 1: RfD = 0.005 mg/kg; Pica intake rate = 10,000 mg
Case 2: RfD = 0.015 mg/kg; Pica intake rate = 5,000 mg
76
-------
Table 4-5 Estimated Cancer and Noncancer Risk from Arsenic in Garden Vegetables
Property
Number
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
Neighborhood
CLAYTON
CLAYTON
CLAYTON
CLAYTON
CLAYTON
CLAYTON
CLAYTON
COLE
COLE
COLE
COLE
COLE
COLE
COLE
COLE
COLE
SWANSEA/EL YRIA
SWANSEA/EL YRIA
SWANSEA/EL YRIA
DF
1/10
0/1
0/1
3/6
1/2
12/12
11/11 (b)
0/2
1/2
1/2
1/1
4/6
4/4
3/9
3/3
0/4
1/1
0/2
1/1
1/3
EPC (based on
inorganic
arsenic) (a)
3.2E-03
2.5E-03
2.6E-02
3.3E-02
1.2E-02
3.3E-01
1.3E-01
9.6E-03
4.0E-02
1.1E-03
1.2E-03
1.2E-01
4.4E-02
2.0E-02
1.2E-02
1.9E-02
1.2E-02
2.0E-03
8.7E-04
2.9E-03
Chronic Noncancer Risk
CTE RME
4E-03 4E-02
3E-03 3E-02
4E-02 4E-01
4E-02 5E-01
2E-02 2E-01
4E-01 5E+00
2E-01 2E+00
1E-02 1E-01
5E-02 6E-01
1E-03 1E-02
2E-03 2E-02
2E-01 2E+00
6E-02 6E-01
3E-02 3E-01
2E-02 2E-01
3E-02 3E-01
2E-02 2E-01
3E-03 3E-02
1E-03 1E-02
4E-03 4E-02
Lifetime Cancer Risk
CTE RME
2E-07 8E-06
2E-07 7E-06
2E-06 7E-05
3E-06 9E-05
9E-07 3E-05
3E-05 9E-04
1E-05 3E-04
7E-07 3E-05
3E-06 1E-04
8E-08 3E-06
9E-08 3E-06
1E-05 3E-04
3E-06 1E-04
2E-06 5E-05
9E-07 3E-05
1E-06 5E-05
9E-07 3E-05
2E-07 5E-06
7E-08 2E-06
2E-07 8E-06
EPC=Eposure Point Concentration
CTE=Central Tendency Estimate
RME=Reasonable Maximum Exposure
Notes:
Shading indicates that vegetable concentration and resulting risk may exceed protective levels.
(a) Units are mg arsenic per kg wet weight of vegetable. Inorganic arsenic is assumed to be 60% of the total arsenic content.
(b) Outlier excluded
77
-------
soil samples at this location were also somewhat higher (mean = 51 ppm) than for most
other gardens (average =11 ppm, range = 6 to 23 ppm), suggesting the elevated values in
vegetables were likely attributable to the elevated soil levels6. One vegetable sample (an
onion) from this property was especially high in arsenic (see Figure 2-10, upper panel),
possibly because of soil adhering to the sample or because of uptake from soil into the
outer layer of the onion. If this one sample is judged to be un-representative of what a
resident is likely to ingest (either because of washing and/or peeling before ingestion)
and is excluded from the risk calculations, the estimates of noncancer and cancer risks
are both reduced, but are still slightly above the usual USEPA level of concern. These
results indicate that ingestion of garden vegetable samples from this location could be of
potential concern for an RME (but not a typical) consumer, but it is possible these risks
are not real, being attributable either to an anomalous analytical result and/or to the extra
safety margin introduced by use of the EPC rather than the simple mean.
At Property 11, the RME cancer risk estimate of 3E-04 is attributable to a single
vegetable sample (garlic) that was significantly higher than the remainder of the samples
from this location. This caused the 95% UCL of the mean to exceed the maximum value
(the garlic sample), so the risk calculation was based on the garlic sample. Because this
value seemed to be questionable compared to other samples from the garden, USEPA
returned to the property and collected a second sample of garlic. This sample yielded a
lower concentration for arsenic (0.2 ppm vs 1.24 ppm dry weight), suggesting the first
result may have been anomalous. This is supported by the observation that soil arsenic
concentrations at this location are quite low (mean = 12 ppm), and elevated
concentrations in vegetables are not expected at such low soil levels. Even if the
concentration in the one garlic sample were considered to be reliable, because the
average mass of garlic ingested per day is relatively small compared to other vegetable
types, risks from garden vegetables at this location are not likely to be of concern.
An alternative approach for evaluating the potential health risks from arsenic in home grown
garden vegetables from the site is to compare the average daily intakes of arsenic in site
vegetables to intakes that occur in the average United States diet. These data are summarized
below:
6 The property owner was not aware of any additions or treatments of the garden that would
account for the moderately elevated arsenic levels in garden soil.
78
-------
Comparison of Average Daily Intakes of Arsenic in the Average United States Diet
and from Vegetables at VBI70 Properties
Parameter
Typical dietary intake of arsenic
Gunderson 1995
Yostetal. 1998
Estimated Average Intake at VBI70 Properties
Property 6 (including outlier)
Property 6 (excluding outlier)
Property 11
All other properties
Value (ug/kg-day)
Total Arsenic
0.36-0.81
0.75
0.07
0.04
0.02
0.01
Inorganic Arsenic
0.20
0.04
0.02
0.01
0.004
As seen, even at Property 6, predicted mean intake of arsenic from site vegetables is a relatively
small fraction of the normal average intake of arsenic from the diet, both for total and inorganic
forms. This supports the conclusion that ingestion of home-grown vegetables from the site is not
likely to cause doses that are outside the normal dietary range.
Overall, the data and calculations above indicate that ingestion of arsenic in home-grown
vegetables is not likely to be a source of significant exposure or risk to most area residents. A
limitation to this conclusion is that garden vegetable samples were not obtained from gardens
with soil arsenic levels higher than about 90 ppm. As noted above, it appears that arsenic
concentrations in garden soils are only weakly correlated with and are substantially lower than
arsenic levels in yard soils (see Figure 2-11), even at yard soil concentrations up to 600 ppm. On
this basis, it is considered that arsenic levels substantially above 90 ppm are not likely to occur
in garden soil, even when yard soils are much higher. However, if vegetables were to be grown
in garden (or yard) soil with high arsenic concentrations, then uptake into vegetables might be
higher than in the samples evaluated.
4.4.4 Combined Risks from Soil and Home-Grown Vegetables
Residents may be exposed to contaminants in soil both by incidental ingestion of soil and by
ingestion of home-grown garden vegetables. Thus, the total risk attributable to contaminants in
soil is the sum of these two pathways:
Risk(total) = Risk(soil) + Risk (vegetables)
Data on arsenic levels in soil are available for all 2,986 properties investigated in Phase III, but
data on arsenic levels in vegetables were collected only at 19 of these properties. Therefore, in
order to calculate total risk at all properties, the concentration of arsenic in garden vegetables
was estimated at each property as described in Section 4.2.6.
79
-------
Because exposure and risk from soil ingestion and vegetable ingestion are both distributions,
care must be taken in the summation process. In the case of the risk to an individual who has
average exposure to both soil and vegetables, the total risk is simply the sum of the two pathway
specific risks:
CTE(total) = CTE(soil) + CTE(vegetables)
In the case of an individual who has RME exposure to soil or to vegetables, the estimate of RME
total risk is not the simple sum of the RME risk estimates, because the two pathways are
independent of each other, and an individual with RME soil intake is not likely to also have
RME vegetable intake (and vice versa). Thus, the estimate of RME total risk is calculated either
as:
1: RME(total) = RME(soil) + CTE(vegetables)
2: RME(total) = CTE(soil) + RME(vegetables)
However, because the RME individual is assumed to have 30 years of exposure to soil, it is also
necessary to assume the individual has 30 years of exposure to garden vegetables (rather than 9
years, which is the usual CTE exposure duration). To account for this, the equations above are
modified as follows:
1': RME(total) = RME(soil) + (30/9)*CTE(vegetables)
2': RME(total) = (30/9)*CTE(soil) + RME(vegetables)
The results are shown in Table 4-6. As seen, based on the site-specific relationships between
arsenic in yard soil and garden soil and between arsenic in garden soil and garden vegetables,
individuals with CTE exposure to garden vegetables are predicted to have excess cancer risks
that are less than or equal to IE-OS, while individuals that have RME intake of garden vegetables
are expected to have risks mainly between 2E-05 and 1E-04, with only a few properties having
risks that exceed 1E-04. When CTE risks are combined across pathways, there are 65 properties
where total risk exceeds 1E-04. When RME risks are combined across pathways, the highest
risks occur for case 1 (RME soil intake plus CTE vegetable intake). Based on this scenario,
there are 99 properties where total RME risks exceed 1E-04.
4.5 UNCERTAINTIES IN ARSENIC RISK ASSESSMENT
It is important to recognize that the calculations of short-term and long term exposure and risk
from arsenic ingestion in soil are based on a number of assumptions and estimates, and that these
introduce uncertainty into the risk results. The most important of the sources of uncertainty in
the calculations are summarized below.
Uncertainty in Average Concentration Terms
The concentration term that is appropriate for calculating chronic exposure and risk from
ingestion exposure to arsenic is the true mean concentration in the medium of concern (soil, dust,
80
-------
Table 4-6 Estimated Total Cancer Risks from Soil and Vegetables
Statistic
CTE
Risk
RME
Risk
Pathway
Soil alone
Vegetables alone
CTE Soil + CTE vegetables
Soil alone
Vegetables alone
RME Soil + CTE vegetables3
CTE Soil3 + RME vegetables
Number of Properties
<= IE-OS
2881
2986
1475
933
2E-05 - 1E-04
105
2921
1419
2979
1954
2921
2E-04 - IE-OS
65
92
7
99
65
Adjusted to account for RME exposure duration (30 years)
81
-------
vegetables), averaged over the area and time interval (averaging time) of concern. There are
two important sources of uncertainty in this value. First, because the true mean cannot be
calculated from a limited set of sample results, the USEPA utilizes the 95% upper confidence
limit of the mean as a conservative estimate of the true mean. This approach helps ensure that
the exposure and risk estimates that are derived are more likely to overestimate than
underestimate the actual risk. Second, the basic exposure unit selected for evaluation in this risk
assessment is the residential property. Using the UCL of the mean for a property is equal to
assuming that an individual residing at that location does not ingest soil or dust from any other
location, even over a time period of up to 30 years. While this might be true for a small sub-set
of residents, it is believed that most residents are sufficiently mobile that exposures will occur
over a wider area than just their own yard. This, in turn will result in lower exposures for people
residing in homes with affected soils, and their true risks will be lower than calculated.
Uncertainty in Concentration Values for Sublocations
As noted earlier, the sampling and analysis design for Phase III was based on a set of three
composite samples from each property. Consequently, there are no data that allow a direct
estimation of the concentration value at any specific sub-location of the yard (these are needed to
address risks from subchronic and acute exposures). To address this data limitation, the
distribution of concentration values within a property was modeled by assuming a lognormal
distribution, and the standard deviation within each property was estimated from the mean value
by multiplying by a site-specific average coefficient of variation of 1.02. This approach should
be considered to yield only approximate values, but since the mean at each property was
estimated using the 95% UCL or the maximum composite value, both the mean and the standard
deviation are more likely to be high than low at each property. Thus, the values estimated for
evaluation of subchronic and acute exposures are also more likely to be high than low.
This expectation is supported by a comparison of the estimated and actual sub-location
concentrations at the eight intensively samples properties from the Risk-Based sampling
program. This comparison was performed as follows. First, at each of these eight properties the
yard was divided into 16-20 sub-areas, and the mean concentration in each sub-area was
calculated based on the values of the grab samples that fell within the sub-area (typically 5-20).
Second, the mean values for these 16-20 sub-areas were rank ordered and used to estimate the
mean at the 90th and 95th percentile sub-area. Third, the 90th and 95th percentile values of the
underlying distribution of grab samples was calculated using the estimation method described
above. In order to complete step 3, it was necessary to estimate what the EPC at these properties
would have been if only three composites had been collected instead of a large number of grab
samples. Based on the Phase III data, the typical ratio of the EPC to the mean is about 1.4.
Thus, an EPC value equal to 1.4-times the mean was assumed for each of the eight properties.
Based on this, the calculated values of the 90th and 95th percentile values of the underlying
distribution of grab samples were, on average, about 2-times higher than the 90th and 95th
percentile values for subarea means. These results support the conclusion that the method used
to estimate EPC values for the subchronic and acute risk calculations is conservative (more
likely to be high than low).
82
-------
Uncertainty in Intake Rates
Data on the amount of soil ingested by humans are very limited. Measurements are difficult to
perform, and results vary significantly from study to study and from method to method. In
addition, data are based mainly on short term studies, so estimates of long-term average intake
rates are especially uncertain. Moreover, intake rates are likely to vary from site to site and
property to property, depending on things such as climate, socioeconomic status, yard condition,
etc., so the default intake rates used in these calculations may not reflect the true intake rates at
the site. Because of the limitations in the data, the default values recommended by USEPA are
intended to be on the high side (i.e., are more likely to overestimate than underestimate actual
soil ingestion).
This is illustrated by comparing the default soil intake rates used by USEPA to data on soil
intake rates measured in a group of 64 children in Anaconda, Montana (Stanek and Calabrese
2000). This study, which utilizes the latest and most refined analytical and statistical methods
for estimating soil ingestion by children, estimated that the average (CTE) 7-day intake by
children is about 31 mg/day (compared to the default of 100 mg/day), and that the 95th
percentile intake for 7 days and 365 days are 133 and 106 mg/day, respectively (compared to the
default assumption of 200 mg/day). If these values from the Anaconda site were judged to be a
more reliable basis for estimation of risk from soil ingestion than the current default values, and
if adult soil intake is assumed to be about /^ that of children, then there are only 23 properties
(rather than 92 properties) in the VBI70 site where RME cancer risks from soil ingestion exceed
a level of 1E-04.
Uncertainty in the Fraction of Total Intake that is Soil
One of the variables used to calculate risks from ingestion of soil plus dust is the fraction of the
total intake that is soil (fs). When concentrations of a contaminant in dust are similar to the
concentration in yard soil, the exact value of fs has very little impact on the calculated risks.
However, at this site, concentrations of arsenic in dust are substantially lower than soil, so the
value for fs is important (the larger the value, the higher the risk). The EPA default value for this
variable (45%) is based mainly on measurements in a set of 64 preschool children (Stanek and
Calabrese 1992). However, due to the difficulty in making these measurements, as well as
potential differences between children and between sites, this value should be considered to be
uncertain. It is not known whether the true value at the VBI70 site is more likely to be higher or
lower than the default values. If the true site-specific value of fs were lower (e.g., 20% rather
than 45%), risks would be about 12% lower than calculated. Conversely, if the true site-specific
value were higher (e.g., 70% rather than 45%), then the risks would be about 12% higher than
calculated.
Uncertainty in Exposure Duration
Cancer risk calculations depend on the duration of exposure. Default exposure durations used in
the risk assessment are not site-specific, and are estimated from data on the length of time that
83
-------
people own a particular residence. Thus, actual exposure durations of residents at the site may
not be the same as the assumed exposure durations assumed, and might be either longer or
shorter than assumed. For example, preliminary data collected at the Globe site suggest that 10-
15% of the residents have lived in their homes for more than 30 years (Mitchell 2001a).
Likewise, analysis of available demographic data by an contractor for ATSDR indicates that
about 13-20% of the residents in the VBI70 area may have resided in their home for more than
30 years (Claritas 2000). These data suggest that an assumed 95th percentile exposure duration
of 30 years may be somewhat too low. However, the data from the Globe site (Mitchell 2001a)
suggest that of the residents who have lived at the site for more than 30 years, only a fraction
resided at the home as a child (when exposure rates are assumed to be highest).
If the RME exposure duration were assumed to be 45 years (6 years as a child and 39 years as an
adult) rather than the default value of 30 years, the estimated excess cancer risk level from soil
ingestion would be about 19% higher than the values reported. In addition, all of the exposure
calculations presented here assume that exposure begins during childhood, when intake rates are
higher than during adulthood. Thus, risks to individuals who move to the site after they are
children will be lower than estimated. For example, risks to an individual exposed for 30 years
as an adult are only 37% of the risks to an individual exposed for 6 years as a child and 24 years
as an adult.
Uncertainty in RME Exposures
In the default point estimate approach for estimating exposure and risk to an RME individual,
two exposure parameters (intake rate and exposure duration) are both assumed to be at their 95th
percentile values. In reality, because these two exposure parameters are independent of each
other, it is very unlikely that an individual with RME soil intake will also have RME exposure
duration. Therefore, an individual with both RME soil intake and RME exposure duration
represents not the 95th percentile of the risk distribution, but some significantly higher
percentile. One way to estimate what the percentile of the default RME individual is, as well as
the actual 95th percentile value, is through Monte Carlo modeling. These calculations
(described in detail in Appendix D) characterize the variability in risk to different individuals in
a hypothetical population of people exposed at a specified exposure location. For an arbitrary
exposure point concentration of 200 ppm arsenic in fine soil (165 ppm in bulk soil), the results of
the point estimate calculation and the Monte Carlo calculations are as shown below:
84
-------
Cancer Risk Estimates for 200 ppm Arsenic in Fine Soil
Method
Point Estimate
Monte Carlo (a)
(see Appendix D)
Statistic
RME cancer risk
90th percentile
95th percentile
99th percentile
99.9th percentile
Soil Alone
1E-04
1E-05 to 4E-05
2E-05 to 6E-05
5E-05 to 1E-04
1E-04 to 2E-04
Vegetables Alone
7E-05
9E-06
IE-OS
3E-05
8E-05
Total Risk
1E-04
2E-05 to 5E-05
3E-05 to 7E-05
6E-05 to 1E-04
1E-04 to 2E-04
(a) Range is based on two alternative PDFs for soil intake rate (see Appendix D)
As seen, the RME risk estimate derived by the point estimate approach is about twice the Monte
Carlo estimate of the 95th percentile value, and is located at or above the 99th percentile of the
risk distribution. This supports the conclusion that the RME point estimate of risk is
conservative and will provide protection to nearly all individuals in the exposed population.
Uncertainty in Toxicity Factors
One of the largest sources of uncertainty in most risk assessments stems from uncertainty in the
toxicity factors used to predict responses from the calculated doses. In the case of arsenic, dose-
response data are derived from studies in humans, which significantly reduces the degree of
uncertainty compared to extrapolations based on animal data. However, a significant degree of
uncertainty still remains in both the slope factor and the chronic RfD. One of the most important
sources of this uncertainty is lack of reliable data on actual arsenic ingestion rates by the
Taiwanese population used to quantify risk. For example, dose-response curves in the key
studies are based on village-based estimates of the concentration of arsenic in well water, rather
than individual specific intake rates (USEPA 200Id). This type of approach, referred to as an
ecological study, is well-known to have a number of limitations, and might either overestimate
or underestimate the true dose-response relationship. In addition, exposures to arsenic through
the diet are believed to be significant, but the magnitude of this contribution can only be
estimated. There are also still large uncertainties in how to extrapolate the dose-response curve
from relatively high exposure levels to lower exposure levels. For example, arsenic does not
appear to cause cancer by a direct genotoxic mechanism (USEPA 200Id), suggesting that a sub-
linear (and perhaps even a threshold) model might be reasonable. However, in the absence of
information on the actual mode of action, an assumption of linearity is still deemed to be
necessary and appropriate (USEPA 200Id). If the dose response curve is sub-linear, current risk
estimates would be too high. Further, there is uncertainty in the importance of cultural and
ethnic differences between different study populations. These differences could include factors
such as inherent differences in the level and capacity of liver enzymes to methylate (and hence
detoxify) ingested inorganic arsenic. Likewise, because methylation requires an adequate supply
of the methyl group (usually derived from dietary methionine), it is plausible that people with
poor diets (especially diets that are low in methionine) might have decreased ability to methylate
arsenic. Differences in diet might also influence the relative amount of arsenic that is absorbed
85
-------
from the gastrointestinal tract into the blood. While little is known about the relative importance
of these factors, it is likely that there are differences between people in their sensitivity to
ingested arsenic, and it is for this reason that USEPA seeks to ensure an adequate margin of
safety in the derivation of the RfD and the slope factor.
Uncertainty in Bioavailability
In order to cause an adverse response, arsenic that is ingested must be absorbed into the body.
As detailed in USEPA (200 Ib), measurements of the arsenic relative bioavailability have been
performed for five soils from the VBI70 site. While measurements based on site soils
significantly reduces uncertainty in this exposure parameter, uncertainty still remains. For
example, variability was observed between different site soils, and a conservative estimate of the
mean value was employed to represent the site-wide average absorption. This approach is
expected to result in an over-estimate of true absorption. Another source of uncertainty is in the
extrapolation of data from test animals to humans. The test animals (swine) were selected
because they are believed to have a gastrointestinal system similar to that in humans, but it is
also possible that absorption in humans might vary as a function of age, stomach contents,
nutritional status, etc. Thus, the RBA value measured in the site-specific study should be viewed
as an approximation of the true RBA value in humans.
The RBA measured for soil was also assumed to apply to dust. This assumption is uncertain
because the size distribution of arsenic-containing particles in dust may be different than for soil,
and particle size might be one factor that influences RBA. If dust contains smaller particles than
soil, and if this size difference tends to increase RBA, then the use of the soil RBA could
underestimate the absorption of arsenic from dust. However, it should be remembered that the
RBA value for soil was measured using only the fine fraction of soil (only particles smaller than
250 micrometers in diameter), so the difference in particle size distribution between dust and soil
is not expected to be large. In addition, because arsenic concentrations in dust tend to be lower
than in soil, the dose contributed by dust ingestion is relatively small compared to that for soil,
so uncertainty in the absorption fraction for dust results in only a small uncertainty in the total
absorbed dose.
Uncertainty Due to Potential Chemical Interactions
All of the risk calculations presented in Section 4 predict the health effects of arsenic acting
alone. However, most people are exposed to many different chemicals in air, water, soil and the
diet, and the possibility exists that some of these chemicals might either increase or decrease the
toxicity of ingested arsenic. Very few data are available on toxicokinetic or toxicologic
interactions of arsenic with other chemicals, although some epidemiological studies suggest that
lead and arsenic might both be associated with behavioral deficits (Moon et al. 1985). This lack
of detailed knowledge on chemical interactions is a general source of uncertainty, but it is not
considered likely that risk is significantly underestimated as a result of any such (hypothetical)
interactions.
86
-------
Uncertainty in Risks Combined Across Exposure Pathways
When exposure of an individual occurs by more than one exposure pathway (e.g., ingestion of
soil and ingestion of home-grown produce), the total exposure and risk is given by the sum
across the pathways. However, calculation of this sum is difficult in the standard point estimate
approach, especially for the RME individual, and the value must be estimated by assuming CTE
intake of one pathway plus RME intake of the other pathway. Appendix D presents an
evaluation of total risks calculated using Monte Carlo modeling which demonstrates that the
point estimate of RME risk that is calculated by summing the RME soil risk with the CTE
vegetable risk will result in a risk estimate that exceeds the 95th percentile of the combined
Monte Carlo distribution. This demonstrates that the RME point estimate is likely to be
conservative and will protect more than 99% of the exposed population.
Uncertainty in Pica Exposure and Risks
As noted earlier, screening-level calculations suggest that acute high-dose exposures to arsenic
in soil (i.e., pica exposure) might be of concern at a number of properties within the site (see
Section 4.4.2). However, data on the amount of soil ingested during pica behavior are very
sparse. Based mainly on one study that observed an intake of 5-8 g/day by a single child
(Calabrese et al. 1989), USEPA has indicated that 5-10 grams might be a reasonable estimate.
If this intake rate is correct, and if arsenic absorption from this mass of soil is similar to that
estimated in site-specific studies (42%), then anywhere from 22% to 62% of all properties within
the VBI70 site (and perhaps outside the site as well) could have arsenic levels above a level of
acute concern. USEPA feels this conclusion is especially uncertain, since the Agency is not
aware of any reported cases of acute arsenic toxicity attributable to ingestion of arsenic in soil.
The most recent study of soil intake by children (Stanek and Calabrese 2000) did not observe
intake rates above 700 mg/day in a group of 64 children, suggesting that values of 5-10 grams
might be unrealistically high. In addition, limited data on urinary arsenic levels in residents of
the VBI70 area and the nearby Globe neighborhood do not reveal the occurrence of high soil
intakes by children (Mitchell 200 Ib). For example, two children from the VBI70 area who were
exposed to high soil arsenic levels (above 400 ppm) both had urinary arsenic levels below the
limit of detection (see Table 2-1 and Table 2-7). In the Globe area, 7 out of 62 children exposed
to soil arsenic concentrations of 5-200 ppm had urinary arsenic levels that were above the
detection limit, but the maximum concentration value was only about 15 ug/L. This
concentration may be contrasted to a value of 100-1000 ug/L which is what would be expected
to occur in a child who ingested 5,000 to 10,000 mg of soil at a location that contained 50 mg/kg
arsenic in soil.
These considerations suggest that arsenic risk from soil pica may not be as significant as the
calculations suggest. On the other hand, if this type of exposure were to occur, it is possible the
symptoms (transient upset stomach and general malaise) would not be recognized as being
arsenic-related, and could easily go un-detected or un-reported. In addition, if pica behavior is
assumed to occur only infrequently during childhood (e.g., 1 day out of 500-1000), then the
chances of observing the behavior in a study of only a few hundred children could be quite low.
That is, it is possible that exposure to arsenic via pica ingestion of soil might be occurring in the
87
-------
children evaluated in these studies, but that the exposure was not detected because it is an
infrequent event. Because of the high uncertainty regarding the magnitude and frequency of soil
pica behavior, more reliable risk estimates for this scenario will not be possible until better data
are collected on pica intakes, along with direct measures of soil-related exposures to arsenic in
soil.
Summary
Because of the uncertainties summarized above, none of the exposure and risk calculations for
arsenic presented above should be interpreted as accurate measures of the true risk. Rather, all
values should be interpreted as uncertain estimates. Because most of the approaches for dealing
with uncertainty are intended to be conservative (i.e., are more likely to overestimate than
underestimate), the risk values above should generally be thought of as high-end estimates of the
true risk, and actual risks are more likely to be lower than the calculated values.
88
-------
SECTION 5
EXPOSURE AND RISK FROM LEAD
5.1 OVERVIEW
As noted earlier, risks from lead are evaluated using a somewhat different approach than for
most other metals. First, because lead is wide-spread in the environment, exposure can occur by
many different pathways. Thus, lead risks are usually based on consideration of total exposure
(all pathways) rather than just to site-related exposures. Second, because studies of lead
exposures and resultant health effects in humans have traditionally been described in terms of
blood lead level7, lead exposures and risks are typically assessed using an uptake-biokinetic
model rather than calculating an estimated dose and comparing that dose to an appropriate
reference dose (RfD). Therefore, calculating the level of exposure and risk from lead in soil also
requires assumptions about the level of lead in other media, and also requires use of
pharmacokinetic parameters and assumptions that are not needed in traditional methods.
For residential land use, the sub-population of chief concern is young children. This is because
young children 1) tend to have higher exposures to lead in soil, dust and paint, 2) tend to have a
higher absorption fraction for ingested lead, and 3) are more sensitive to the toxic effects of lead
than are older children or adults.
It is currently difficult to identify what degree of lead exposure, if any, can be considered safe in
young children. Some studies report subtle signs of lead-induced neurobehavioral effects in
children beginning at blood lead levels around 10 ug/dL or even lower, with population effects
becoming clearer and more definite in the range of 30-40 ug/dL (CDC 1991, ATSDR 1999). On
the other hand, some researchers and clinicians believe the effects that occur in children at low
blood lead levels are so minor that they need not be cause for concern. After a thorough review
of all the data, the USEPA has identified 10 ug/dL as the blood lead level at which effects that
warrant avoidance begin to occur, and has set as a goal that there should be no more than a 5%
chance that any child will have a blood lead value above 10 ug/dL (USEPA 1994a, 1994b). This
approach focuses on the risks to a child at the upper bound (about the 95th percentile) of the
exposure distribution, very much the same way that the approach used for other chemicals
focuses on risks to the RME individual. The Centers for Disease Control (CDC) has also
established a guideline of 10 ug/dL in preschool children which is believed to prevent or
minimize lead-associated cognitive deficits (CDC 1991).
5.2 IEUBK MODEL FOR ASSESSING LEAD RISK
The USEPA has developed an Integrated Exposure Uptake Biokinetic (IEUBK) model for
predicting the likely range of blood lead levels in a population of young children (age 0-6 years)
7 The concentration of lead in the blood is usually abbreviated "PbB", and is expressed in units of
micrograms of lead per deciliter of blood (ug/dL). One dL is equal to 100 mL.
89
-------
exposed to a specified set of environmental lead levels (USEPA 1994b). This model requires as
input data on the levels of lead in soil, dust, water, air, and diet at a particular location, and on
the amount of these media ingested or inhaled by a child living at that location. All of these
inputs to the IEUBK model are central tendency point estimates. These point estimates are used
to calculate an estimate of the central tendency (the geometric mean) of the distribution of blood
lead values that might occur in a population of children exposed to the specified conditions.
Assuming the distribution is lognormal, and given (as input) an estimate of the variability
between different children (this is specified by the geometric standard deviation or GSD), the
model calculates the expected distribution of blood lead values, and estimates the probability that
any random child might have a blood lead value over 10 ug/dL.
If all of the IEUBK model exposure levels and intake rates are set at their default values, and if
the concentration of lead in dust is assumed to be 70% of that in soil (the default assumption),
then the IEUBK model predicts that a soil lead level of about 350 ppm corresponds to the target
risk level (no more than a 5% chance of exceeding a blood lead level of 10 ug/dL) for children
age 0-84 months. If default estimates of dietary intake are adjusted downwards by a factor of 0.7
to partially account for the lower lead levels in the current food supply (Bolger et al. 1996,
Gunderson et al. 1995, Griffin et al. 1999) than are assumed in the default IEUBK model, then
the soil lead level that corresponds to the target risk level is about 400 ppm. Based in part on
these results, USEPA has established a national policy that soil lead levels below 400 ppm may
be assumed to be below a level of health concern (USEPA 1994a). Soil lead levels above 400
ppm may or may not be of concern, depending on site-specific factors. At 400 ppm in soil, the
IEUBK model predicts that exposure from soil (including ingestion of both soil and dust)
accounts for 73-78% of the total absorbed dose of lead, with even larger relative contributions at
higher soil lead levels. Of the non-soil exposure, food is about three times larger than water, and
intake from air is negligible.
Whenever reliable site-specific data are available on any of the IEUBK model input parameters,
these are used in preference to the assumptions employed in the default case. At this site, three
types of site-specific data are available, as follows:
Adjustment for Lead Enrichment in the Fine Fraction
As discussed in Section 2.4.1, it is suspected that exposure to soil occurs mainly via ingestion of
the fine fraction. Since Phase III data on the concentration of lead in soil are based on the
concentration in bulk soil, the mean concentration in bulk soil at each property was adjusted to
account for the enrichment of lead in the fine fraction, as follows (see Section 2.6.1):
Cs(fme)=1.09-Cs(bulk)
Soil-Dust Relationship
The site-specific relationship between lead in bulk yard soil and lead in indoor dust was
presented earlier in Figure 2-9. As shown in this figure, the average relationship is described by
an equation of the form:
Cd = 0.34-Cs(bulk) + 150
90
-------
Lead Bioavailability
In order to investigate the relative bioavailability of lead in site soils, USEPA Region VIII
performed a study in which two separate samples of site soil were fed to swine for 15 days.
Swine were selected as the test species because it is believed the gastrointestinal system (and
hence the behavior of ingested lead) in swine is similar to that in humans. The details of the
study design and of the findings are presented in a separate report (USEPA 200 Ic). In brief, the
study found that lead in site soils was absorbed by swine about 81-87% (mean = 84%) as well as
a readily soluble form of lead (lead acetate). This in vivo estimate is supported by the
bioaccessability measured in vitro:
In Vivo Bioavailability and In Vitro Bioaccessability Measurements of VBI70 Site Soils
Test
Material
Sample 1
Sample 2
Sample
Location
Eastern part of site
Western part of site
In Vivo Relative
Bioavailability (%)
87%
81%
In Vitro
Bioaccessability (%)
86%
85%*
* Mean of duplicate analyses
This RBA value is somewhat higher than the typical USEPA default value of 60%, suggesting
that the lead in site soils is in a form that can be readily absorbed. Based on this site-specific
finding, an RBA of 0.84 was used in the evaluation of lead risks. Based on a default absorption
fraction of 50% for lead in water and food, this RBA corresponds to an absolute bioavailability
(ABA) of 42% (0.42).
These adjustments to the model, along with the other model inputs, are summarized in Table 5-1.
This site-specific adjusted model was used to evaluate risks to children from lead in soil and
dust, as described below.
5.3 RISK CHARACTERIZATION FOR LEAD
5.3.1 Risks from Lead in Soil and Dust
The expected blood lead distribution for children (age 0-84 months) was calculated for each
property using lEUBKwin vl.O (build 241). The soil value at each property was the estimated
concentration in fine soil (1.09 times the mean bulk concentration), and the dust lead
concentration was predicted using the equation above. The results, characterized in terms of the
probability of a random child exceeding a blood lead value of 10 ug/dL (this is referred to as
"P10"), are shown in Table 5-2.
As seen, a total of 1,655 out of 2,986 homes are predicted to have P10 values at or below the
health-based goal of 5%, while 1,331 (45%) are predicted to exceed the health-based goal.
Approximately 610 properties are predicted to have P10 values of 5-10%, slightly above the
91
-------
Table 5-1IEUBK Model Inputs
SOIL/DUST INPUTS
Csoil = 1.09 • Bulk property-specific average (ppm) (a)
Cdust = 0.34 • Csoil + 150 (ppm) (a)
CONSTANTS
PARAMETER
Air concentration (ug/m3)
Indoor air concentration
Drinking water concentration (ug/L)
Absorption Fractions:
Air
Diet
Water
Soil/Dust (a)
Fraction soil
GSD
VALUE
0.10
30% of outdoors
4.0
32%
50%
50%
42%
45%
1.6
AGE DEPENDENT
Age
0-1
1-2
2-3
3-4
4-5
5-6
6-7
AIR
Time
Outdoors
(hrs)
1.0
2.0
3.0
4.0
4.0
4.0
4.0
Vent. Rate
(m3/day)
2.0
3.0
5.0
5.0
5.0
7.0
7.0
DIET
Dietary intake
(ug/day)
3.87
4.05
4.54
4.37
4.21
4.44
4.90
WATER
Intake
(L/day)
0.20
0.50
0.52
0.53
0.55
0.58
0.59
SOIL
Intake
(mg/day)
85
135
135
135
100
90
85
(a) Values based on site-specific data
92
-------
Table 5-2 Estimated Risks to Children from Lead in Soil and Dust
Neighborhood
Clayton
Cole
Elyria
Globeville
Swansea
All
Total Number of
Properties
902
100%
796
100%
59
100%
63
100%
1166
100%
2986
100%
Number and Percent of Properties Within Specified Risk Range
P10 <= 5%
712
79%
169
21%
6
10%
7
11%
761
65%
1655
55%
P10 > 5% and <= 10%
119
13%
248
31%
9
15%
18
29%
216
19%
610
20%
P10 > 10% and <= 20%
52
6%
273
34%
28
47%
21
33%
144
12%
518
17%
P10 > 20%
19
2%
106
13%
16
27%
17
27%
45
4%
203
7%
Total P10>5%
190
21%
627
79%
53
90%
56
89%
405
35%
1331
45%
P10=Prediced Risk of Exceeding Blood Lead of 10 ug/dL
93
-------
heath-based goal. However, about 518 properties would be expected to have P10 values
between 10-20%, and 203 homes are predicted to have P10 values greater than 20%
(substantially above the health-based goal). It should be noted that 1,057 of the 1,331 properties
(79%) with P10 values above 5% have mean bulk lead concentrations lower than 400 ppm (the
USEPA default level of concern). This is mainly because the site-specific KB A for lead (84%) is
higher than the default value (60%), and also because of the use of the concentration value in
the fine fraction rather than the bulk fraction in the risk calculations.
Although homes with elevated soil lead are found in all neighborhoods, the density of homes
with P10 values greater than 5% tends to be higher in the central and western part of the site than
in areas on the eastern side of the site. This is illustrated the following table:
Count of Properties
Location
Western (a)
Eastern (b)
Total Number
918
2068
Number with P10 > 5%
736
595
%
80%
29%
(a) Western = Cole, Elyria, Globeville
(b) Eastern = Clayton and Swansea
In interpreting these risk estimates, it is important to recognize that lead is a naturally occurring
element in soil, and that there are many current and historic anthropogenic sources of lead (e.g.,
automobile exhaust, leaded paint, generalized industrial emissions, etc.). As noted earlier (see
Figure 2-7), levels of lead in bulk soils at the VBI70 site range from below the detection limit
(about 52 ppm) up to a maximum of more than 1,000 ppm. In contrast to the situation that was
found for arsenic (see Figure 4-3), analysis of this distribution does not reveal the presence of
two distinct components, so the boundary between the values that are "background" (including
both natural and area-wide anthropogenic sources) and those that are elevated due to site-specific
sources is difficult to judge. If it is assumed that the upper range of the typical urban
background levels is about 400 ppm, then the mean of all samples that are less than 400 ppm is
about 195 ppm. Using this value (195 ppm in bulk soil) as a rough estimate of the mean
concentration in urban background samples, and assuming the same site-specific input values as
shown in Table 5-1, the IEUBK model predicts that blood lead levels attributable to urban
background levels of lead probably average about 4.4 ug/dL for a typical (median) child, and
might be as high as 9.5 ug/dL for a child with above-average (95th percentile) exposure to soil or
dust.
5.3.2 Risks from Lead in Garden Vegetables
As shown previously (see Figure 2-10), there is essentially no uptake of lead from soil into
garden vegetables at this site. On this basis, it is concluded that exposure to lead from ingestion
of home grown garden vegetables is not of concern.
94
-------
5.4 UNCERTAINTIES IN LEAD RISK EVALUATION
It is important to stress that lead risk predictions based on the IEUBK model are uncertain. This
uncertainty arises from a number of factors. First, there is inherent difficulty in providing the
model with reliable estimates of human exposure to lead-contaminated media. For example,
exposure to soil and dust is difficult to quantify because human intake of these media is likely to
be highly variable, and it is very difficult to derive accurate measurements of actual intake rates.
Likewise, site-specific data on exposure to lead through the diet is generally not available, and
because dietary lead levels have been decreasing over time, the default data used in the model
may no longer be accurate. Second, it is often difficult to obtain reliable estimates of key
pharmacokinetic parameters in humans (e.g., absorption fraction, distribution and clearance
rates, etc.), since direct observations in humans are limited. Finally, the absorption, distribution
and clearance of lead in the human body is an extremely complicated process, and any
mathematical model intended to simulate the actual processes is likely to be an over-
simplification. Consequently, IEUBK model calculations and predictions should not be thought
of as being identical to actual risk.
Alternative IEUBK Model Runs
In order to investigate some of these sources of uncertainty in the IEUBK model predictions, a
series of three alternative IEUBK model runs were performed using several alternative model
input values, including the following:
a) Dietary lead intake values based on the latest market-basket study by the FDA (Bolger
et al. 1996, Gunderson et al. 1995, Griffin et al. 1999). These values are listed below:
FDA Dietary Lead Intake Values
Age
6-11 months
1 year
2 years
3 years
4 years
5 years
6 years
Dietary Intake (ug/d)
1.82
1.90
1.87
1.80
1.73
1.83
2.02
b) A series of alternative GSD values ranging from 1.2 to 1.5. The GSD is the most
sensitive input parameter in the IEUBK model, and a small change in the GSD can result
in a large change in the calculated P10 value. As discussed below, there is some reason
to think that the default GSD value of 1.6 used by the IEUBK model might be somewhat
95
-------
too high, so these runs were performed to investigate how the results would change if the
GSD were indeed smaller than the default.
c) A mean soil intake value based on the soil intake study by Stanek and Calabrese
(2000). In this study the estimate of the long-term average soil intake rate was 31
mg/day. Age-specific intake values were estimated by multiplying the mean value (31
mg/day) by the ratios of the IEUBK age-specific intake rates (see Table 5-1) compared to
the IEUBK age-averaged intake rate (109 mg/day).
The results of these alternative IEUBK model runs are shown below:
Uncertainty Analysis Results for Alternative IEUBK Model Inputs
Model Run (a)
Default (see Table 5-2)
Revised dietary intakes (see above)
GSD= 1.5
GSD= 1.4
Revised dietary intakes (see above) and GSD=1.4
GSD= 1.3
Revised dietary intakes (see above) and GSD=1.3
GSD= 1.2(b)
Revised dietary intakes (see above) and GSD=1.2 (b)
Soil intake based on Stanek and Calabrese (2000)
P 10 Value (%)
<5%
1655
1937
2058
2413
2572
2728
2801
2911
2931
2986
5-10%
610
507
450
315
229
134
91
37
30
0
10-20%
518
402
345
171
118
67
59
19
12
0
> 20%
203
140
133
87
67
57
35
19
13
0
Total
with
P10>5%
1331
1049
928
573
414
258
185
75
55
0
(a) All runs include site-specific adjustments for lead enrichment in the fine fraction (1.09), RBA
(0.84), and for soil-dust relationships.
(b) Calculations performed using the DOS version (0.99d) of the IEUBK model
These calculations help illustrate the range of potential uncertainty in risk estimates for lead that
may be associated with uncertainty in the IEUBK model inputs, especially the dietary intake of
lead, the soil/dust intake rate, and the GSD.
ISE Model Predictions
Another approach for assessing hazard from lead in soil is currently under development by
USEPA Region VIII. This approach, referred to as the Integrated Stochastic Exposure (ISE)
Model for Lead, uses the same basic equations and algorithms for calculating exposure and
96
-------
blood lead values as the IEUBK model, except that it uses probability distribution functions
(PDFs) rather than point estimates as inputs for a number of exposure parameters. These
distributions are combined using Monte Carlo simulation techniques to yield a predicted
distribution of absorbed lead doses (ug/day) for different members of the exposed population.
These doses are then used as input to the biokinetic portion of the IEUBK model in order to
generate the predicted distribution of blood lead values in the population. Thus, the variability
between children is evaluated in the ISE model based on the variability in environmental and
exposure parameters, rather than by application of an assumed or estimated GSD value as in the
IEUBK model. Because this model has not yet undergone peer review or validation, it is
considered to be only an investigative tool.
The input distributions used in the ISE model runs are summarized in Table 5-3. The
distribution for soil ingestion is based on reliable data and a well-characterized empirical
distribution function (EDF) reported by Stanek and Calabrese (1995). The mean soil intake
value assumed by the IEUBK model (about 109 mg/day) is located between the 75th and 80th
percentile of the EDF reported by Stanek and Calabrese (1995). Variability in the RBA term is
based on the observed inter-individual variability in response in the animal study used to develop
the RBA. In this study, the mean coefficient of variation (standard deviation divided by the
mean response) across dose groups was about 0.2. based on the logic that variability is likely to
be higher in a group of children than in a group of test animals, a coefficient of variation of 0.3
was assumed. Thus, given a mean RBA of 84% and a mean absolute absorption fraction of 42%,
the standard deviation was assumed to be 12.6%. The basis of the other distributions is provided
in Goodrum et al. (1996). It is important to note that these other distributions are screening-level
only. In most cases a distribution is assumed to be lognormal, even though the true shape is not
known. Likewise, the mean value of the distribution is selected to match the mean value used by
the IEUBK model, but the estimate of the standard deviation is often an estimate based mainly
on professional judgement.
The results of a risk evaluation based on the ISE model compared to the predictions of the
IEUBK model are presented below:
Comparison of ISE and IEUBK Model Predictions
Model
IEUBK Model
ISE Model
P10 Value (%)
<5%
1655
2986
5-10%
610
0
10-20%
518
0
> 20%
203
0
Total
with P10>5
1331
0
97
-------
Table 5-3 ISE Model Inputs
AT/EF:
Exposure Frequency
Averaging Time
SOIL:
Point
Point
365.00
365.00
days/yr
days/yr
C_soil (soil Pb cone) Point 600 ug Pb/g
IRsd (soil+dust IR) PDF-Cumulative mg/day
Number: 8 Min: 0 Max: 7000
Values: {0,10,45,88,186,208,225,7000}
Percen: {0,0.25,0.5,0.75,0.9,0.95,0.99,1}
Age: 0-1 IR scale factor
Age: 1-2 IR scale factor
Age: 2-3 IR scale factor
Age: 3-4 IR scale factor
Age: 4-5 IR scale factor
Age: 5-6 IR scale factor
Age: 6-7 IR scale factor
Fs (frac ingest as soil)
DUST:
C_dust (dust Pb cone)
Regression Variable A
Regression Variable B
WATER:
C_water (water Pb Cone)
Age: 0-1 IR Water
Age: 1-2 IR Water
Age: 2-3 IR Water
Age: 3-4 IR Water
Age: 4-5 IR Water
Age: 5-6 IR Water
Age: 6-7 IR Water
Point 0.6296
Point 1
Point 1
Point 1
Point 0.7407
Point 0.6666
Point 0.6296
PDF-Triangular (0.1,0.45,0.8)
PDF-Log Normal
PDF-Log Normal
PDF-Log Normal
(315,307)
(150, 50)
(0.34, 0.2)
ug Pb/g soil
C_dust=A+B*C_soil
C dust=A+B*C soil
PDF-Log
PDF-Log
PDF-Log
PDF-Log
PDF-Log
PDF-Log
PDF-Log
PDF-Log
Normal
Normal
Normal
Normal
Normal
Normal
Normal
Normal
(4,
(o.
(o.
(o.
(o.
(o.
(o.
(o.
,3)
2, 0.2)
5, 0.4)
52,
53,
55,
58,
59,
0
0
0
0
0
•4)
•4)
•4)
•4)
•4)
ugPb/L
L/day
L/day
L/day
L/day
L/day
L/day
L/day
98
-------
DIET:
Table 5-3 (Continued)
Age:
Age:
Age:
Age:
Age:
Age:
Age:
0-1
1-2
2-3
3-4
4-5
5-6
6-7
Diet
Diet
Diet
Diet
Diet
Diet
Diet
Intake
Intake
Intake
Intake
Intake
Intake
Intake
PDF-Log
PDF-Log
PDF-Log
PDF-Log
PDF-Log
PDF-Log
PDF-Log
Normal
Normal
Normal
Normal
Normal
Normal
Normal
(3
(4
(4
(4
(4
(4
(4
.87
.05
.54
.37
.21
.44
.9,
,2)
,2)
,2)
,2)
,2)
,2)
2)
ug Pb/day
ug Pb/day
ug Pb/day
ug Pb/day
ug Pb/day
ug Pb/day
ug Pb/day
OTHER:
Age: 0-1 Other
Age: 1-2 Other
Age: 2-3 Other
Age: 3-4 Other
Age: 4-5 Other
Age: 5-6 Other
Age: 6-7 Other
ABSORPTION:
Intake
Intake
Intake
Intake
Intake
Intake
Intake
Point
Point
Point
Point
Point
Point
Point
0
0
0
0
0
0
0
ug Pb/day
ug Pb/day
ug Pb/day
ug Pb/day
ug Pb/day
ug Pb/day
ug Pb/day
Soil: % accessible
Dust: % accessible
Water: % accessible
Diet: % accessible
Other: % accessible
Passive Fraction
Half Saturation Level
AIR:
AirPb
Age: 0-
Age: 1-
Age: 2-
Age: 3-
Age: 4-
Age: 5-
Cone Outdoors
-1 Ventilation Rate
-2 Ventilation Rate
-3 Ventilation Rate
-4 Ventilation Rate
-5 Ventilation Rate
-6 Ventilation Rate
PDF-Log Normal
PDF-Log Normal
PDF-Log Normal
PDF-Log Normal
Point 30
Point
Point 100
PDF-Log Normal
PDF-Log Normal
PDF-Log Normal
PDF-Log Normal
PDF-Log Normal
PDF-Log Normal
PDF-Log Normal
(42,12.6,100,10) percent
(42,12.6,100,10) percent
(50,20) percent
(50,20) percent
percent
0.2
ug/day
(0.1,0.05)
(2, 1.2)
(3, 1.4)
(5, 2.4)
(5, 2.4)
(5, 2.4)
(7,3.4)
ug Pb/m3 air
m3 air/day
m3 air/day
m3 air/day
m3 air/day
m3 air/day
m3 air/day
99
-------
Table 5-3 (Continued)
Age: 6-7 Ventilation Rate
Indoor Cone (% of Outdoor)
Age: 0-1 Time Outdoors
Age: 1-2 Time Outdoors
Age: 2-3 Time Outdoors
Age: 3-4 Time Outdoors
Age: 4-5 Time Outdoors
Age: 5-6 Time Outdoors
Age: 6-7 Time Outdoors
Lung Absorption Age 0-1
Lung Absorption Age 1-2
Lung Absorption Age 2-3
Lung Absorption Age 3-4
Lung Absorption Age 4-5
Lung Absorption Age 5-6
Lung Absorption Age 6-7
PDF-Log Normal
Point 30
Point 1
Point 2
Point 3
Point 4
Point 4
Point 4
Point 4
Point 32
Point 32
Point 32
Point 32
Point 32
Point 32
Point 32
(7, 3.4) m3 air/day
percent
hr/day
hr/day
hr/day
hr/day
hr/day
hr/day
hr/day
percent
percent
percent
percent
percent
percent
percent
100
-------
As seen, the ISE model predicts that there are no homes above the level of health concern. This
is in marked contrast to the IEUBK model, which predicts that there are 1,331 homes of concern.
The main reason for this difference is that the estimate of long-term average inter-individual
variability generated by Monte Carlo simulation (GSD = 1.2) is substantially lower than the
assumed variability in the IEUBK model (GSD = 1.6). If the variability between individuals is
examined over shorter time scales (e.g., 24-36 months rather than 1-84 months), the GSD
predicted by the ISE model approaches that assumed by the IEUBK model:
ISE Model GSD Values Calculated Over Various Averaging Times
Model
IEUBK
ISE
GSD as a Function of Averaging Time
Months 1-84
1.6
1.2
Months 24-36
1.6
1.4
Month 24
1.6
1.5
These results highlight the sensitivity of both models to the degree of inter-individual variability
(as reflected in the assumed or calculated GSD), and suggests that the GSD value used by the
IEUBK model may be more nearly appropriate for short-term (one-month) exposure intervals
than for estimating variability in long-term average blood lead values.
Another factor which may contribute to the apparent difference between the models is that the
blood lead point estimate calculated by the IEUBK model is not likely to be equivalent to the
true geometric mean of the distribution of values among members of the exposed population.
This is because the input point estimates used by the model are usually more likely to be
arithmetic means than geometric means. If all of the inputs were arithmetic means, then the
expected value of the IEUBK point estimate would be closer to the arithmetic mean blood lead
than the geometric mean. Because inter-individual variability in blood leads is represented by a
lognormal distribution in the IEUBK model, the arithmetic mean will always be greater than the
geometric mean, so treating the IEUBK point estimate as the geometric mean may tend to shift
the distribution to the right, thereby tending to increase the percent of the distribution above the
health-based level of concern (10 ug/dL).
Comparison of IEUBK Re suits to Observed Blood Lead Values
Another way to help determine whether the IEUBK model is yielding reliable results at a
particular site is to compare the IEUBK model predictions with actual observations of blood lead
levels in the population of children currently living at the site. This approach has been used at a
number of other sites in the Rocky Mountain west (e.g., Aspen, Leadville, Midvale), and it is
often found that the observed incidence of elevated blood lead values is not as high as predicted
by the model. There are a number of reasons why this might be so, including potential
limitations in the blood lead study itself. However, the consistency of this pattern across sites
suggests that, on average, the default IEUBK model may tend to be somewhat over-conservative.
If so, this would presumably stem from imprecision in one or more of the model inputs (e.g., soil
101
-------
or dust intake rates, biokinetic factors, GSD, etc.), but the actual basis of the apparent
discrepancies between predicted and observed blood lead values remains uncertain and
controversial.
At the VBI70 site, biomonitoring programs offered by the USEPA have resulted in collection of
only very limited blood lead data. These data were derived by recruiting individuals living at
homes selected for soil removal as part of the Phase II and Phase III programs to allow sampling
of hair and urine to assess arsenic exposure, and sampling of blood to assess lead exposure. A
total of 21 individuals participated. The results for blood lead are summarized below:
Blood Lead and Residential Soil Lead Levels at VBI70 Removal Properties
Age category
Child (0-6 years)
Adult (>6 Years)
All
Number of
Participants
2
19
21
Lead in Bulk Soil (ppm)
Mean
263
459
439
Maximum
499
1700
1700
PbB (ug/dL)
Geometric
mean
1.0
1.8
1.7
Maximum
2.0
5.0
5.0
This data set is much too limited to support any strong conclusion, especially because the
number of children participating was so low, and because many of the properties had lead levels
in soil that were only moderately elevated. However, the data do not provide an indication that
lead exposures are above a level of concern.
Another source of potentially relevant blood data is from three different blood lead testing
programs sponsored by the State. The Colorado Department of Public Health and Environment
consolidated the data from these studies and provided the results to USEPA for evaluation in this
risk assessment. A brief description of these three studies is provided below:
1. Denver Childhood Lead Survey. In this study, children age 0-3 years were tested
from targeted neighborhoods where the risk of finding elevated blood lead levels was
thought to be highest. Testing was conducted from June through September 1995.
2. Globe Medical Monitoring Program. In this study, children age 6 years and under
have been tested through the Globe Medical Monitoring Program. The majority of the
children were tested in the spring of 1994 at the Globe field office. These children were
recruited via door-to-door outreach. Four additional children were tested at local clinics
held in the south Globeville neighborhood in April of 2000.
3. State Lead Surveillance Program. This study includes blood lead results for children
tested between 1995 and 2000. Most of the data collected between 1995 and 1999 were
reported to the State as part of mandatory reporting of elevated blood lead levels (> 10
ug/dL) by state laboratories, and most values below 10 ug/dL were not reported.
102
-------
Beginning in 2000, the State requires the laboratories to report all data, not just values
above 10 ug/dL. Because of this important difference, the results for this study have been
divided into Part 3A (1995-1999) and Part 3B (2000). Test results from 2000 were
primarily collected at targeted clinics held specifically to recruit children living in the
VBI70 area.
Because blood lead data are confidential medical information, the only information provided to
USEPA on the study subjects besides their the blood lead level (ug/dL) was their age (years), the
soil lead level (mg/kg) at the child's residence, and whether the residence is located within or
outside the VBI70 study area.
The results are summarized in Table 5-4. Data from Phase 2 of the Third National Health and
Nutrition Examination Survey (NHANES III) are also shown to provide a frame of reference
(Pirkle et al. 1998). Inspection of this table reveals the following main points:
a) Within a study, there is no consistent pattern of difference in blood lead values for
children living within VBI70 and those residing outside of the study area boundary. This
suggests that residents living within the site do not have a substantially higher risk of lead
exposure than people living in locations adjacent to the site.
b) For children age 0-5 residing within the VBI70 area, geometric mean blood lead
levels observed in Study 1 (5.7 ug/dL) and Study 3A (15.6 ug/dL) are clearly higher than
the national average for children age 1-5 (2.7 ug/dL). However, an elevation over
average may be expected in these cases because the children in these studies do not
represent a random set of children but a set selected for study because they were believed
to have high risk of exposure (study 1) or were included in the study specifically because
they have elevated blood lead levels (study 3 A). The results for study 2 and study 3B
(these studies are more nearly random than the other studies) suggest that blood lead
levels for children age 0-5 residing within VBI70 (GM = 3.2 to 4.6) are somewhat higher
than the national average for children age 1-5 (2.7 ug/dL), but are not clearly distinct
from values seen elsewhere in the nation for children age 1-5 residing in old housing
(GM = 3.8 ug/dL) or in families with low income (GM = 3.8 ug/dL) (Pirkle et al. 1998).
c) Geometric standard deviations within the different studies for children age 0-5 within
the VBI70 area are range from 1.5 to 2.4. These values tend to be somewhat higher than
the default GSD value of 1.6 assumed in the IEUBK model, but this is not considered to
be evidence that the IEUBK default GSD value is too low. Rather, GSD values measured
in most blood lead studies are expected to be higher than the true GSD for two main
reasons: 1) the observed GSD includes variability in blood lead attributable to variability
in environmental levels as well as variability in childhood contact with those media,
while the desired value includes only variability in contact parameters; and 2) the
variability in blood lead values between children is based on a single measurement in
each child, rather than the long term average value in each child. As noted above,
103
-------
Table 5-4 Comparison of State Blood Lead Data to National Statistics
STATE BLOOD LEAD DATA
Denver Survey
(1)
Globe Program
(2)
State Surveillance
Program (ALL)
(3)
State Surveillance
Program (Prior to 2000) a
(3A)
State Surveillance
Program (2000 Results)
(3B)
Age (yrs)
0-4
0-5
6-11
0-5
6-11
0-5
6-11
0-5
6-11
Within VBI70
N
83
32
6
156
17
47
-
99
17
Geomean
(ug/dL)
5.7
3.2
3.4
6.6
3.9
15.6
-
4.6
3.9
GSD
1.8
1.7
1.5
2.4
2.0
1.5
-
2.2
2.0
Outside VBI70
N
83.0
69
17
99
8
46
4
53
4
Geomean
(ug/dL)
6.0
3.3
2.8
7.1
6.7
10.3
12.7
5.1
3.5
GSD
2.0
1.8
1.3
2.4
2.9
2.0
1.5
2.4
3.2
a This data set excludes 10 samples collected prior to 2000 with PbB < 10 ug/dL.
NATIONAL GEOMETRIC MEAN BLOOD LEAD LEVELS
Demographic Variable
ALL
HOUSING
INCOME
Pre 1946
1946-1973
Post 1973
Low
Middle
High
Age
(years)
1-5
6-11
1-5
1-5
Geomean
(ug/dL)
2.7
1.9
3.8
2.8
2.0
3.8
2.3
1.9
NHANES III, Phase 2: 1991-1994. (Pirkle et al., 1998)
104
-------
between-children variability in instantaneous measurements of blood lead values will
always be larger than variability in long-term average values, and this results in an
overestimate of the GSD. Thus, without more detailed data (e.g., repeated blood lead
measurements in each child, data on the level of lead exposure in several environmental
media for each child), it is considered that these data do not provide a way to estimate a
reliable site-specific GSD value.
Figure 5-1 (upper panel) plots the blood lead levels across all three studies as a function of the
mean soil lead level (based on data collected during the Phase III Program) at the child's
residence. As seen, there is only a low degree of correlation (R2 = 0.019), with high blood lead
values occurring at low soil lead concentrations, and low blood lead values occurring at high soil
lead concentrations. This observation establishes that soil lead is not the only source of lead
exposure in children, and that soil lead is likely to explain only a small amount of the variability
in blood lead levels between different children. Although the slope of the line that relates blood
lead to soil lead (0.0075 ug/dL per ppm) is not statistically significant (p = 0.07), the slope is
similar to that predicted by the IEUBK model, supporting the conclusion that soil lead probably
does contribute to childhood lead exposures at the site.
Figure 5-1 (lower panel) compares the blood lead values predicted by the IEUBK model with
those actually observed in study participants. As seen, there is only a weak correlation (R2 =
0.059), with the IEUBK model tending to over-predict the lower blood lead values and under-
predict the higher blood lead values. This suggests that the IEUBK model may be over-
estimating the contribution of the common sources of lead exposure (soil, dust, water, diet), and
is not accounting for one or more large sources of lead exposure (most likely leaded paint
ingestion).
In conclusion, even though these blood lead studies were not designed or intended to support risk
assessment purposes, they do support the following broad conclusions: a) elevated blood lead
levels do occur in children residing within the site, b) soil is not likely to be the main source of
elevated blood lead levels, and c) the elevations are not clearly different from areas outside the
site.
Summary of Uncertainties
As discussed above, there are a number of sources of uncertainty in any evaluation of lead risks
to children. When mathematical modeling is used to evaluate risks, the most important sources
of uncertainty are in average soil ingestion rates, and in the degree of variation between the
exposure rates of different children. As shown, the range of results across different sets of input
values and different models can be quite large. When direct observation of blood lead values is
used as the basis for evaluating risk, the main source of uncertainty is whether the study
population is sufficiently large and sufficiently representative to allow correct interpretation. At
this site, the available blood lead data set is clearly too small to provide a basis for any firm
conclusions, but the data do not reveal any large hazard.
105
-------
Figure 5-1 State Blood Lead Analysis Results
T3
^
C
O
"ro
"c
0)
o
o
O
O
JD
CO
Blood Lead versus Soil Lead
VBI70 Properties-All 3 studies combined
y = 0.0075X + 5.8268
R2 = 0.0192
Soil Lead Concentration (mg/kg)
Observed vs Predicted Blood Lead
(all 3 studies combined)
30
O
O^
m
0)
"o
25
20-
15-
0)
Dl 10
m
^ 5
Theoretical
\,'X'
over-predict
y = 0.056x + 4.3983
R2 = 0.0586
Measured Blood Lead (ug/dL)
106
-------
SECTION 6
REFERENCES
ATSDR. 1992. Agency for Toxic Substances and Disease Registry. Toxicological Profile for
Thallium. U.S. Department of Health and Human Services, Agency for Toxic Substances and
Disease Registry. July 1992.
ATSDR. 1999. Agency for Toxic Substances and Disease Registry. Toxicological Profile for
Lead (Update). U.S. Department of Health and Human Services, Agency for Toxic Substances
and Disease Registry.
ATSDR. 2000. Agency for Toxic Substances and Disease Registry. Toxicological Profile for
Arsenic (Update). U.S. Department of Health and Human Services, Agency for Toxic
Substances and Disease Registry.
Bolger PM, Yess NJ, Gunderson EL, Troxell TC, Carrington CD. 1996. Identification and
Reduction of Sources of Dietary Lead in the United States. Food Additives and Contaminants
13:53-60.
Brody DJ, Pirkle JL, Kramer RA, Flegal KM, Matte TD, Gunter EW, Paschal DC. 1994. Blood
Lead Levels in the US Population. Phase 1 of the Third National Health and Nutrition
Examination Survey (NHANES III, 1988 to 1991). JAMA 272:277-283.
Calabrese EJ, Barnes R, Stanek EJ, Pastides H, Gilbert CE, Veneman P, Wanf X, Lasztity A,
Kostecki PT. 1989. How much soil do young children ingest: an epidemiologic study.
Regulatory Toxicol Pharmacol. 10:123-137.
Claritas. 2000. Demographic data analysis performed for ATSDR by Claritas. Partial data
summary provided to Bonita Lavelle, USEPA Region VIII, by David Mellard, ATSDR.
Colorado Climate Center. 2000. Data downloaded from on-line meteorological database
maintained by the Colorado Climate Center, Colorado State University.
http://ulysses.atmos.colostate.edu.
CDC. 1991. Preventing Lead Poisoning in Young Children. US Department of Health and
Human Services, Centers for Disease Control.
CDOT. 2000a. Air Quality Data Report. 1-70 and Brighton Boulevard Interchange
Modifications. January-March 2000. Report prepared for the Colorado Department of
Transportation by Spectrum Services, Inc, Myers Environmental Services, Inc, and Asoian
Associates LLC. May 2000.
107
-------
CDOT. 2000b. Air Quality Data Report. 1-70 and Brighton Boulevard Interchange
Modifications. April-June 2000. Report prepared for the Colorado Department of
Transportation by Spectrum Services, Inc, Myers Environmental Services, Inc, and Asoian
Associates LLC. August 2000.
CDOT. 2000b. Air Quality Data Report. 1-70 and Brighton Boulevard Interchange
Modifications.
July-September 2000. Report prepared for the Colorado Department of Transportation by
Spectrum Services, Inc, Myers Environmental Services, Inc, and Asoian Associates LLC.
November 2000.
EnviroGroup. 2000. Commercial Property Data, Asarco Globe Plant Site. Memo (with
attachments) from Dave Folkes, EnviroGroup, to Bill Brattin, ISSI, dated February 10, 2000.
Goodrum PE, Diamond GL, Hassett JM, Johnson DL. 1996. Monte Carlo Modeling of
Childhood Lead Exposure: Development of a Probabilistic Methodology for Use with the
USEPA IEUBK Model for Lead in Children. Human and Ecological Risk Assessment 2: 681-
708.
Griffin S, Goodrum PE, Diamond GL, Meylan W, Brattin WJ, Hassett JM. 1999. Application of
a Probabilistic Risk Assessment Methodology to a Lead Smelter Site. Human and Ecological
Risk Assessment 5:845-868.
Gunderson EL. 1995. FDA Total Diet Study, July 1986-April 1991, Dietary Intakes of
Pesticides, Selected Elements, and Other Chemicals. Journal of AOAC International:78:1353-
1363.
HUD. 1995. Guidelines for the Evaluation and Control of Lead-Based Paint Hazards in
Housing. U.S. Department of Housing and Urban Development, Washington DC. June 1995.
ISSI. 1999. Selection of Chemicals of Potential Human Health Concern at the Vasquez
Boulevard and 1-70 Site. Memorandum to USEPA. June 18, 1999.
IRIS. 2000. Retrieval from USEPAs Integrated Risk Information System (IRIS). February,
2000.
Mazumder DNG, Haque R, Ghosh N, De BK, Santra A, Chakraborty D, and Smith AH. 1998.
Arsenic levels in drinking water and the prevalence of skin lesions in West Bengal, India. Int. J.
Epidem. 27:871-877.
Mitchell J. 200 la. Personal communication from Jane Mitchell, Colorado Department of Public
Health and Environment, to William Brattin, Syracuse Research Corporation. July 27, 2001.
108
-------
Mitchell J. 2001b. Draft Fact Sheet: Results of Globe Medical Monitoring Urinary Arsenic
Testing. Personal communication from Jane Mitchell, Colorado Department of Public Health
and Environment, to Bonita Lavelle, USEPA Region VIII. Dated 05/25/01.
Mizuta N, Mizuta M, Ito F, Ito T, Uchida H, Watanabe Y, Akama H, Murakami T, Hayashi F,
Nakamura K, Yamaguchi T, Mizuta W, Oishi S, Matsumura H. 1956. An Outbreak of Acute
Arsenic Poisoning Caused by Arsenic Contaminated Soy-Sauce (Shoyu): A Clinical Report of
220 Cases. Bull. Yamaguchi Med. School 4:131-149.
Moon C, Marlowe M, Stellern J, Errera J. 1985. Main and Interaction Effects of Metallic
Pollutants on Cognitive Functioning. J. Learning Disabilities 18:217-221.
Morales KH, Ryan L, Kuo TL, Wu MM, Chen CJ. 2000. Risk of internal cancers from arsenic
in drinking water. Environ. Health Perspectives 108:655-661.
NRC. 1999. Arsenic in Drinking Water. National Research Council, National Academy of
Sciences. National Academy Press, Washington DC.
Pirkle JL, Kaufmann RB, Brody DJ, Hickman T, Gunter EW, Paschal DC. 1998. Exposure of
the U.S. Population to Lead, 1991-1994. Environ. Health Perspectives 106:745-750.
Schoof RA, Yost LJ, Eickhoff J, Crecelius EA, Cragin DW, Meacher DM, Menzel DB. 1999. A
Market Basket Survey of Inorganic Arsenic in Food. Food Chem. Toxicol. 37:839-846.
Stanek EJ, Calabrese EJ. 1992. Soil Ingestion in Children: Outdoor Soil or Indoor Dust? J.
SoilContam. 1:1-28.
Stanek EJ, Calabrese EJ. 1995. Daily Estimates of Soil Ingestion in Children. Environ. Health
Perspect. 103:276-285.
Stanek EJ, Calabrese EJ. 1999. Development of Exposure Distribution Parameters for Use in
Monte Carlo Risk Assessment of Exposure Due to Soil Ingestion. Final Report. December
1999. Department of Biostatistics and Epidemiology. Department of Environmental Health
University of Massachusetts at Amherst Amherst, Massachusetts. Sponsor: USEPA R8
Ecosystems Protection&Remediation. Contract: LOR056 1998 T 08L, TIN:043-16-7352.
Stanek EJ, Calabrese EJ. 2000. Daily Soil Ingestion Estimates for Children at a Superfund Site.
Risk Analysis 20:627-635.
Tseng, W.P., Chu, H.M., How, S.W., et al. 1968. Prevalence of skin cancer in an endemic area
of chronic arsenism in Taiwan. J. Natl. Cancer Inst. 40:453-463.
UOS. 1998a. Final Sampling Activities Report for North Denver Residential Soils - Phase I.
June 1998.
109
-------
UOS. 1998b. Sampling Analysis Report - Phase II Sampling for Removal Site Assessment.
Vasquez Boulevard/Interstate 70 Site. September 21, 1998.
USEPA. 1988. Special Report on Ingested Inorganic Arsenic: Skin Cancer; Nutritional
Essentiality. U.S. Environmental Protection Agency, Office of Health and Environmental
Assessment. EPA/625/3-87/013.
USEPA. 1989. Risk Assessment Guidance for Superfund. Volume I. Human Health
Evaluation Manual (Part A). U.S. Environmental Protection Agency, Office of Emergency and
Remedial Response. EPA/540/1-89/002.
USEPA. 199la. Human Health Evaluation Manual, Supplemental Guidance: "Standard Default
Exposure Factors". U.S. Environmental Protection Agency, Office of Solid Waste and
Emergency Response. OSWER Directive 9285.6-03.
USEPA. 1991b. Role of the Baseline Risk Assessment in Superfund Remedy Selection
Decisions. Memo from Don R. Clay, Assistant Administrator, dated April 22, 1991. U.S.
Environmental Protection Agency, Office of Solid Waste and Emergency Response. OSWER
Directive 9355.0-30.
USEPA. 1991c. Risk Assessment Guidance for Superfund. Volume I. Human Health
Evaluation Manual (Part B, Development of Risk-Based Preliminary Remediation Goals). U.S.
Environmental Protection Agency, Office of Emergency and Remedial Response. EPA/540/R-
92/003.
USEPA. 1992a. Supplemental Guidance to RAGS: Calculating the Concentration Term. U.S.
Environmental Protection Agency, Office of Solid Waste and Emergency Response. Publication
9285.7-081.
USEPA. 1992b. Dermal Exposure Assessment: Principles and Applications. Interim Report.
U.S. Environmental Protection Agency, Office of Research and Development. EPA/600/8-
91/011B.
USEPA. 1993. Superfund's Standard Default Exposure Factors for the Central Tendency and
Reasonable Maximum Exposure. Draft, dated 11/04/93. U.S. Environmental Protection
Agency.
USEPA. 1994a. Guidance Manual for the Integrated Exposure Uptake Biokinetic Model for
Lead in Children. U.S. Environmental Protection Agency, Office of Emergency and Remedial
Response. Publication Number 9285.7-15-1. EPA/540/R-93/081.
USEPA. 1994b. Revised Interim Soil Lead Guidance for CERCLA Sites and RCRA Corrective
Action Facilities. U.S. Environmental Protection Agency, Office of Solid Waste and Emergency
Response. Elliot P. Laws, Assistant Administrator. July 14, 1994. OSWER Directive 9355.4-
12.
110
-------
USEPA. 1995a. Urban Soil Lead Abatement Demonstration Project. Review Draft. U.S.
Environmental Protection Agency, National Center for Environmental Assessment. EPA/600/R-
95/139.
USEPA. 1995b. Subchronic Reference Dose for Arsenic. Memo from Robert Benson, Ph.D., to
Chris Weis, Ph.D., dated 9/12/95.
USEPA. 1995c. USEPA Region VIII Standard Operating Procedure. Dermal Absorption from
Water or Soil. Draft Document prepared for Region VIII by Roy F. Weston, Inc.
USEPA. 1996. Soil Screening Guidance: Technical background Document. U.S.
Environmental protection Agency, Office of Solid Waste and Emergency Response.
EPA/540/R95/128. May 1996.
USEPA. 1997. Exposure Factors Handbook, Volumes I, II, and III. U.S. Environmental
Protection Agency, Office of Office of Research and Development. EPA/600/P-95/002Fa.
USEPA. 1998a. Exposure to Elevated Levels of Arsenic and Lead in Residential Soils in the
Vicinity of Vasquez Blvd. and 1-70 May Pose an Imminent and Substantial Endangerment to
Public Health. Memo from Christopher P. Weis, Regional Toxicologist, to Peter Stevenson, On-
Scene Coordinator. Dated July 27, 1998.
USEPA. 1998b. Project Plan for the Vasquez Boulevard and 1-70 Residential Risk-Based
Sampling Stage I Investigation. Report prepared by IS SI Consulting Group, Inc. for USEPA
Region VIII. August 1998.
USEPA. 1998c. Project Plan for the Vasquez Boulevard and 1-70 Residential Soils
Supplemental Investigation: Physico-Chemical Characterization of Soils. Report prepared by
ISSI Consulting Group, Inc. for USEPA Region VIII. August 1998.
USEPA. 1998d. Data Report for the Vasquez Boulevard and 1-70 Residential Soils
Supplemental Investigation: Physico-Chemical Characterization of Soils. Report prepared by
ISSI Consulting Group, Inc. for USEPA Region VIII. November, 1998.
USEPA. 1999a. Draft Report for the Vasquez Boulevard and 1-70 Site Residential Risk-Based
Sampling Stage I Investigation. Report prepared by ISSI Consulting Group, Inc. for USEPA
Region VIII. October 1999.
USEPA. 1999b. Selection of Chemicals of Potential Concern at the Vasquez Boulevard and I-
70 Site. Report prepared by ISSI Consulting Group, Inc. for USEPA Region VIII. June 18,
1999.
Ill
-------
USEPA. 1999c. U.S. Environmental Protection Agency. Region III Risk-Based Concentration
Table. Philadelphia, PA. April 12.
USEPA. 1999d. Phase III Project Plan for the Vasquez Boulevard & 1-70 Site, Denver CO.
Phase III Investigation. Report prepared by IS SI Consulting Group, Inc., for U.S.
Environmental Protection Agency, Region VIII. August 4, 1999.
USEPA. 1999e. Project Plan for the Vasquez Boulevard & 1-70 Site, Denver CO. Pilot-Scale
Soil Characterization Study. Report prepared by ISSI Consulting Group, Inc., for U.S.
Environmental Protection Agency, Region VIII. September 9, 1999.
USEPA. 2000a. Data Quality Assessment, Phase IIIA Sampling Program, Vasquez
Boulevard/I-70 Remedial Investigation.. Report prepared by Washington Group International,
Inc. for USEPA Region VIII. July 21, 2000.
USEPA. 2000b. Data Quality Assessment, Phase IIIB Sampling Program, Vasquez
Boulevard and 1-70 Remedial Investigation. Report prepared by Washington Group
International, Inc. for USEPA Region VIII. November 21, 2000.
USEPA. 200la. Data Quality Assessment, Phase IIIB Grab Sampling Program, Vasquez
Boulevard and 1-70 Remedial Investigation. Report prepared by Washington Group
International, Inc. for USEPA Region VIII. February 9, 2001.
USEPA. 2001b. Relative Bioavailability of Arsenic in Soils from the VBI70 Site. Report
prepared for USEPA Region VIII by Syracuse Research Corporation. January 2001.
USEPA. 200 Ic. Relative Bioavailability of Lead in Soils from the VBI70 Site. Report
prepared for USEPA Region VIII by Syracuse Research Corporation. February 2001.
USEPA. 200Id. National Primary Drinking Water Regulations; Arsenic and Clarifications to
Compliance and New Source Contaminants Monitoring. Federal Register 66:6976-7066.
January 22, 2001.
USEPA. 200 le. Remedial Investigation Report. Vasquez Boulevard/I-70 Site. Operable Unit
1. Report prepared by Washington Group International for USEPA Region VIII. July, 2001.
USEPA. 200 If. Derivation of Acute and Subchronic Oral Reference Doses for Inorganic
Arsenic. U.S. Environmental Protection Agency, Office of Solid Waste and Emergency
Response. February 2001.
Yost LJ, Schoof RA, Aucoin R. 1998. Intake of Inorganic Arsenic in the North American Diet.
Human Ecological Risk Assessment 4:137-152.
112
-------
APPENDIX A
GARDEN VEGETABLE AND SOIL DATA
-------
APPENDIX A GARDEN VEGETABLE AND SOIL DATA
Property
ID
1
1
1
1
1
1
1
1
1
1
2
3
4
4
4
4
4
4
5
5
6
6
6
6
6
6
6
6
6
6
6
6
7
7
8
8
9
9
10
11
11
11
11
11
11
strPropAddress
3340 MONROE ST
3340 MONROE ST
3340 MONROE ST
3340 MONROE ST
3340 MONROE ST
3340 MONROE ST
3340 MONROE ST
3340 MONROE ST
3340 MONROE ST
3340 MONROE ST
341 2 CLAYTON ST
3510 SAINT PAUL ST
3534 COLUMBINE ST
3534 COLUMBINE ST
3534 COLUMBINE ST
3534 COLUMBINE ST
3534 COLUMBINE ST
3534 COLUMBINE ST
3546 HARRISON ST
3546 HARRISON ST
3604 BRUCE RANDOLPH AVE
3604 BRUCE RANDOLPH AVE
3604 BRUCE RANDOLPH AVE
3604 BRUCE RANDOLPH AVE
3604 BRUCE RANDOLPH AVE
3604 BRUCE RANDOLPH AVE
3604 BRUCE RANDOLPH AVE
3604 BRUCE RANDOLPH AVE
3604 BRUCE RANDOLPH AVE
3604 BRUCE RANDOLPH AVE
3604 BRUCE RANDOLPH AVE
3604 BRUCE RANDOLPH AVE
3650 COOK ST
3650 COOK ST
3223 RACE ST
3223 RACE ST
3244 VINE ST
3244 VINE ST
3250 HIGH ST
3310 VINE ST
3310 VINE ST
3310 VINE ST
3310 VINE ST
3310 VINE ST
3310 VINE ST
Garden Vegetables
Sample ID
3-041 56-B
3-041 59-B
3-041 51 -B
3-041 66-B
3-041 69-B
3-041 57-B
3-041 58-B
3-041 54-B
3-041 55-B
3-041 62-B
3-04602-B
3-04600-B
3-04620-B
3-0461 8-B
3-0461 5-B
3-0461 7-B
3-0461 9-B
3-0461 4-B
3-04625-B
3-04623-B
3-04749-B
3-04768-B
3-04758-B
3-04755-B
3-04753-B
3-04762-B
3-04756-B
3-04757-B
3-04745-B
3-04743-B
3-04748-B
3-04769-B
3-05234-B
3-05225-B
3-05239-B
3-05240-B
3-05237-B
3-05238-B
3-04585-B
3-04792-B
3-05226-B
3-05222-B
3-05230-B
3-04791 -B
3-04799-B
Vegetable Type
Rhubarb
Chard
Peppers
Squash
Squash
Eggplant
Cabbage
Cauliflower
Tomatoes
Squash
Tomatillo
Collard Greens
Lettuce
Carrots
Beets
Turnip Greens
Rutabaga
Collard Greens
Collard Greens
Peppers
Onions
Carrots
Beets
Turnips
Cauliflower
Collard Greens
Collard Greens
Cucumbers
Zucchini
Squash
Tomatoes
Cabbage
Cabbage
Tomatillo
Beets
Turnips
Tomatoes
Tomatoes
Tomatoes
Garlic
Chard
Onions
Collard Greens
Collard Greens
Cucumbers
Dry Wt. Cone (mg/kg dw)
Total As Pb
0.05
0.10
0.05
0.06
0.05
0.08
0.05
0.05
0.05
0.05
U
J
U
J
U
J
U
U
U
U
0.61
0.57
0.11
0.20
0.05
0.05
0.05
0.05
0.05
0.07
J
U
U
U
U
U
J
0.05 U 0.05 U
0.34 0.24
0.10
0.06
0.05
0.08
0.17
0.32
0.16
0.05
6.30
0.50
1.09
3.45
0.46
0.37
0.63
2.92
1.63
0.63
0.08
0.31
0.08
0.05
0.34
0.52
0.05
0.05
J
J
U
J
U
J
J
U
U
U
2.20
0.96
0.94
0.68
0.80
0.20
0.50
0.05
1.78
1.34
1.13
1.21
0.50
0.12
0.11
0.27
0.11
0.08
0.05
0.05
U
J
J
J
J
U
U
0.08 J
0.06 J
2.32
0.98
0.05 U
0.05 U
0.05 U 0.62
1.24
0.14
0.10
0.07
0.12
0.67
J
J
J
J
38.60
4.31
1.67
0.53
0.16
0.18
% Solid
8.70
6.52
10.60
3.64
13.70
10.10
10.10
9.98
7.71
4.26
16.50
12.90
10.50
13.20
12.70
13.60
11.40
17.10
12.10
15.00
15.60
13.10
13.90
6.11
10.00
15.50
16.20
5.92
9.54
9.70
13.30
9.95
19.90
16.30
19.70
7.29
7.07
6.54
7.67
16.50
13.20
12.40
13.80
13.60
6.02
^dj. Wet Wt. Cone (mg/kg ww
Total As Pb
2.2E-03 5.31 E-02
6.5E-03
2.7E-03
2.18E-03
3.43E-03
8.08E-03
2.53E-03
2.50E-03
1.93E-03
1.07E-03
3.72E-02
1.17E-02
7.28E-03
3.43E-03
2.53E-03
2.53E-03
2.50E-03
1 .93E-03
2.98E-03
4.13E-03 4.13E-03
4.39E-02 3.10E-02
1.05E-02
7.92E-03
3.18E-03
1 .09E-02
1 .94E-02
5.47E-02
2.31 E-01
1.27E-01
1.19E-01
9.25E-02
9.12E-02
3.42E-02
1 .94E-02 6.05E-02
3.75E-03 3.75E-03
9.83E-01
6.55E-02
1 .52E-01
2.11E-01
4.60E-02
5.74E-02
1 .02E-01
1 .73E-01
1 .56E-01
6.11 E-02
1 .06E-02
3.08E-02
1 .59E-02
4.08E-03
6.70E-02
3.79E-02
1.77E-03
1.64E-03
2.78E-01
1 76E-01
1 .57E-01
7.39E-02
5.00E-02
1 .86E-02
1 .78E-02
1 .60E-02
1 .05E-02
7.76E-03
3.33E-03
2.49E-03
1 .59E-02
9.78E-03
4.57E-01
7.14E-02
1.77E-03
1 .64E-03
1.92E-03 4.76E-02
2.05E-01 6.37E+00
1.85E-02 5.69E-01
1.24E-02 2.07E-01
9.66E-03 7.31 E-02
1 .63E-02 ~1 8E-02
4.03E-02 1 .08E-02
Garden Soils (mg/kg)
Sample
Number
3-041 56-B
3-041 59-B
3-041 51 -B
3-041 66-B
3-041 69-B
3-041 57-B
3-041 58-B
3-041 54-B
3-041 55-B
3-041 62-B
3-04602-B
3-04600-B
3-04620-B
3-0461 8-B
3-0461 5-B
3-0461 7-B
3-0461 9-B
3-0461 4-B
3-04625-B
3-04623-B
3-04749-B
3-04768-B
3-04758-B
3-04755-B
3-04753-B
3-04762-B
3-04756-B
3-04757-B
3-04745-B
3-04743-B
3-04748-B
3-04769-B
3-05234-B
3-05225-B
3-05239-B
3-05240-B
3-05237-B
3-05238-B
3-04585-B
3-04792-B
3-05226-B
3-05222-B
3-05230-B
3-04791 -B
3-04799-B
Raw Concentration
As Pb
11.0 U 122.9
11.0
11.0
15.0
11.0
11.0
11.0
11.0
11.0
11.0
U
U
U
U
U
U
U
U
110.2
152.2
248.8
100.8
127.3
111.2
100.8
104.8
222.6
11.0 U 114.0
11.0U 95.9
11.5
11.0
11.3
11.0
11.0
11.0
11.0
12.3
73.3
46.5
54.5
40.4
92.5
56.6
43.0
68.6
45.1
46.2
24.5
48.2
U
U
U
U
J
128.4
119.2
129.7
130.2
116.4
115.5
52.0
87.3
145.5
103.9
98.1
89.7
123.6
140.4
131.5
172.1
280.3
137.0
138.5
| 110.1
U
J
11.0U 65.8
11.0U 56.6
11.0U 236.0
19.5 260.5
11.0
37.1
U
137.2
170.6
18.4 314.1
11.0
12.0
11.0
17.7
16.4
16.7
U
U
270.6
147.5
250.1
184.9
212.7
259.1
Adjusted Cone
As Pb
5.5 122.9
5.5 110.2
5.5 152.2
15.0 248.8
5.5 100.8
5.5 127.3
5.5 111.2
5.5 100.8
5.5 104.8
5.5 222.6
5.5 114.0
5.5 95.9
11.5 128.4
5.5 119.2
11.3 129.7
11.0 130.2
5.5 116.4
5.5 115.5
5.5 26.0
12.3 87.3
73.3 145.5
46.5 103.9
54.5 98.1
40.4 89.7
92.5 123.6
56.6 140.4
43.0 131.5
68.6 172.1
45.1 280.3
46.2 137.0
24.5 138.5
48.2 110.1
5.5 65.8
5.5 56.6
5.5 236.0
19.5 260.5
5.5 137.2
37.1 170.6
18.4 314.1
5.5 270.6
12.0 147.5
5.5 250.1
17.7 184.9
16.4 212.7
16.7 259.1
Veg Data Final.xls
-------
Property
ID
12
12
12
12
13
13
13
13
13
13
13
13
13
14
14
14
15
15
15
15
16
17
17
18
19
19
19
strPropAddress
331 5 RACES!
331 5 RACES!
331 5 RACES!
331 5 RACES!
3322 VINE S!
3322 VINE S!
3322 VINE S!
3322 VINE S!
3322 VINE S!
3322 VINE S!
3322 VINE S!
3322 VINE S!
3322 VINE S!
3351 GAYLORD S!
3351 GAYLORD S!
3351 GAYLORD S!
351 1 LAFAYETTE S!
351 1 LAFAYETTE S!
351 1 LAFAYETTE S!
351 1 LAFAYETTE S!
3630 RACE S!
4300 S!EELE S!
4300 S!EELE S!
431 4 JOSEPHINES!
4755 GAYLORD S!
4755 GAYLORD S!
4755 GAYLORD S!
Garden Vegetables
Sample ID
3-04773-B
3-04765-B
3-04776-B
3-04775-B
3-04789-B
3-04798-B
3-04794-B
3-04779-B
3-04786-B
3-04782-B
3-04771 -B
3-04784-B
3-04781 -B
3-041 48-B
3-041 44-B
3-041 50-B
3-05249-B
3-05247-B
3-05248-B
3-05244-B
3-04608-B
3-04588-B
3-04589-B
3-04744-B
3-04597-B
3-04595-B
3-04592-B
Vegetable !ype
Carrots
Collard Greens
Collard Greens
Collard Greens
Onions
Celery
!urnips
Collard Greens
Squash
Peas
Cabbage
!omatoes
Cabbage
Onions
Peppers
Broccoli
Cucumbers
!omatoes
!omatillo
!omatoes
Peppers
!omatoes
Peppers
!omatoes
Beans
!omatillo
!omatoes
Dry Wt. Cone (mg/kg dw)
!otal As Pb
0.27
0.38
0.56
0.27
0.17
0.19
0.33
0.11
0.05
0.05
0.13
0.05
0.05
0.14
0.05
0.08
0.68
0.05
0.10
J
J
U
J
U
U
J
U
J
U
J
1.15
0.58
0.30
0.29
1.87
2.05
1.57
0.16
0.29
0.06
0.05
0.05
0.05
J
U
U
U
0.68
0.20
0.06 J
0.66
0.33
0.16
0.05 U 0.10 J
0.15 0.15 J
0.05 U 0.18
0.05 U
0.05 U
0.05 U
0.05 U
0.05 U
0.05
U
0.11 J
0.13 J
0.20
0.05 U
% Solid
11.00
12.00
13.00
11.00
14.80
8.25
10.20
13.70
6.17
22.50
11.60
9.96
9.24
13.80
9.68
12.20
4.55
5.84
7.37
6.63
13.10
5.90
13.60
5.77
19.00
6.47
6.05
^dj. Wet Wt. Cone (mg/kg ww
!otal As Pb
2.97E-02
4.56E-02
7.28E-02
2.97E-02
2.52E-02
1 .57E-02
3.37E-02
1.51E-02
3.09E-03
5.63E-03
1.51E-02
1.27E-01
6.96E-02
3.90E-02
3.19E-02
2.77E-01
1 .69E-01
1 .60E-01
2.19E-02
1.79E-02
1 .35E-02
2.90E-03
2.49E-03 2.49E-03
2.31 E-03 2.31 E-03
1 .93E-02 9.38E-02
2.42E-0~ 1.94E-02
9.76E-03 7.32E-03
3.09E-02 3.00E-02
1.46E-03 1 .93E-02
7.37E-03
1.66E-03
1.18E-02
6.63E-03
1 .97E-02 1 .97E-02
1 .48E-03 1 .06E-02
3.40E-03 3.40E-03
1.44E-03 6.35E-03
4.75E-03 2.47E-02
1 .62E-03 | T29E-02
1.51 E-03 1.51 E-03
Garden Soils (mg/kg)
Sample
Number
3-04773-B
3-04765-B
3-04776-B
3-04775-B
3-04789-B
3-04798-B
3-04794-B
3-04779-B
3-04786-B
3-04782-B
3-04771 -B
3-04784-B
3-04781 -B
3-041 48-B
3-041 44-B
3-041 50-B
3-05249-B
3-05247-B
3-05248-B
3-05244-B
3-04608-B
3-04588-B
3-04589-B
3-04744-B
3-04597-B
3-04595-B
3-04592-B
Raw Concentration
As Pb
26.0
17.5
22.8
25.1
16.9
11.0
11.0
25.4
11.0
11.0
11.5
11.0
11.0
11.6
15.3
21.3
11.0
11.0
11.0
11.0
15.2
11.0
11.0
13.7
11.0
16.0
11.0
140.1
224.9~
157.4 "
J 152.4
U
U
U
U
217.7
338.8
210.3
344.0
240.8
294.0
186.0
U 253.0
U] 195.5
162.2
U
U
U
U
171.6
183.5
369.3
570.0
335.1
381.1
| 79.5
U
U
U
U
52.0 U
61.2
572.9
408.3
236.1
260.8
Adjusted Cone
As Pb
26.0 140.1
17.5 224.9
22.8 157.4
25.1 152.4
16.9 217.7
5.5 338.8
5.5 210.3
25.4 344.0
5.5 240.8
5.5 294.0
11.5 186.0
5.5 253.0
5.5 195.5
11.6 162.2
15.3 171.6
21.3 183.5
5.5 369.3
5.5 570.0
5.5 335.1
5.5 381.1
15.2 79.5
5.5 26.0
5.5 61.2
13.7 572.9
5.5 408.3
16.0 236.1
5.5 260.8
U=Analyte not detected
J=Estimated
Veg Data Final.xls
-------
APPENDIX B
SCREENING LEVEL EVALUATION OF
RELATIVE RISK FROM ARSENIC VIA
INHALATION OF DUST OR DERMAL CONTACT WITH SOIL
COMPARED TO SOIL INGESTION
-------
APPENDIX B
SCREENING LEVEL EVALUATION OF
RELATIVE RISK FROM ARSENIC VIA
INHALATION OF DUST OR DERMAL CONTACT WITH SOIL
COMPARED TO SOIL INGESTION
1.0 INHALATION OF PARTICIPATES IN AIR
The basic equations recommended by USEPA (1989) for evaluation of risk from inhalation exposure
of soil particles in air and for incidental ingestion of soil are as follows:
Inhalation Exposure
Riskair = Ca-BRa-EF-ED/(BW-AT)-SFinh
Ingestion Exposure
Risksoil = C^-IR^-EF-ED/CBW-A^-SF^
where:
C = Concentration of contaminant in air (Ca, mg/m3) or soil (Csoib mg/kg)
BR = Breathing rate (mVday)
IRsoil = Ingestion rate for soil (kg/day)
EF = Exposure frequency (days/yr)
ED = Exposure duration (years)
BW = Body weight (kg)
AT = Averaging time (days)
SF = Cancer slope factor for inhalation or oral exposure
Assuming that the values of BW, EF, ED, and AT are all the same for inhalation and oral
exposure, the ratio of the risk from inhalation of particulates in air to that from ingestion of soil
is then:
Relative risk (inhalation/oral) = (Cair/Csoil)(BR/IR)(SFinhal/SForal)
Soil particles may be released from soil and enter air due either to wind-based erosion or
mechanical disturbance. A screening level evaluation of each type of scenario is presented
below.
B-l
-------
Exposure from Wind-Based Soil Erosion
The amount of soil released to air by wind is a complex function of wind speed, soil characteristics,
and the surface features of the site. The USEPA has developed a conservative screening level
approach for evaluating wind-based releases, as described in USEPA (1996). Screening level
defaults inputs for this equation are as follows:
• The ratio Cair/Csoil (ug/m3 per ug/kg) is given by the inverse of the Particulate
Emission Factor (PEF), calculated in accord with the equation and region-specific
intake values identified in USEPA (1996). The resulting value is 9.1E-10 kg/m3
(0.91 ug/m3).
The ratio of BR/IR for a resident is 20 m3/day / 1E-04 kg/day = 2E+05 m3/kg
(USEPA 1989, 1991b)
• For arsenic, the ratio of the inhalation slope factor to the oral slope factor is
15/1.5 = 10 (IRIS 2000).
Based on these values, the ratio of the risk from inhalation exposure to arsenic in airborne
soil particles compared to that from ingestion exposure is:
Relative risk = 9.1E-10 • 2E+05 • 10 = 0.0018 (0.18%)
As seen, the risk from inhaled arsenic is very small (< 0.2%) compared to that from ingested
soil, so this pathway is considered to be sufficiently minor that quantitative evaluation is not
required at this site.
Exposure from Mechanical Disturbances
The amount of soil which enters air as a result of mechanical disturbances (e.g., automobile
traffic on a dirt road, agricultural tilling of a field, etc) is a complex function of the type and
frequency of the disturbance. At the VBI70 site, data are available from a large highway
construction project being carried out by the Colorado Department of Transportation for the
Brighton Road Interchange on 1-70 (CDOT 2000a, 2000b, 2000c). These data include 79-82
samples collected at each of three different monitoring stations over the interval from January
through September, 2000. Each sample was analyzed for PM10 (particulate matter less than
10 um in diameter) and/or TSP (total suspended particulates). In addition, the levels of arsenic
and lead in PM10 and TSP were measured.
The average level of PM10 measured at one station was 65 ug/m3. This level is nearly two
orders of magnitude higher than the default level of 0.9 ug/m3 used to evaluate wind-erosion
(see above). If this airborne matter were all attributable to mechanical erosion of soil into air,
the relative cancer risk from inhalation compared to ingestion might be as large as about 13%
(still a relatively small fraction). However, it is important to note that not all PM10 particles
B-2
-------
in air are derived from soil. In support of this, average arsenic levels in PMlOs and/or in TSP
ranged from 2.4 to 3.2 ng/m3, a level that is lower than the average of 20-30 ng/m3 for urban
areas across the United States (ATSDR 2000). Likewise, the average level of lead was 28-37
ng/m3, lower than the default value of 100 ng/m3 used by USEPA in the IEUBK model. These
data indicate that even under conditions of mechanical disturbance, airborne levels of arsenic
and lead from soil are still quite low and are not a source of significant health concern.
2.0 DERMAL EXPOSURE VIA SOIL
The basic equations recommended for estimation of risk from dermal contact with soil and
ingestion of soil are as follows (USEPA 1989, 1992):
Dermal Exposure
Riskdemal = Cs-SA-AF-ABS-EF-ED/(BW-AT)-(SForal/AFo)
Oral Exposure
Risksoil = Cs-IRsoil-EF-ED/(BW-AT)-SForal
where:
Cs = concentration of chemical in soil (mg/kg)
SA = surface area in contact with soil (cm2)
AF = soil adherence factor (kg/cm2)
ABS = dermal absorption fraction (unitless)
AFo = oral absorption fraction
IRsoii = ingestion rate for soil (kg/day)
BW = body weight (kg)
EF = exposure frequency (days/yr)
ED = exposure duration (years)
AT = averaging time (days)
SForal = cancer slope factor for oral exposure
Thus, assuming the values of BW, ED, and AT are the same for dermal and oral exposure, the ratio
of the risk for dermal contact compared to that for soil ingestion is given by:
Relative risk (dermal/oral) = (SA-AF-EFdennal-ABS)/(IR-EForal-AFo)
Screening level inputs for this equation are as follows:
SA = 10% of whole body = 2,000 cm2 (USEPA 1991b).
AF = 1E-06 kg/cm2 (USEPA 1992)
EFdemal = 50 days/yr (assumed)
• ABS is not known for arsenic, but is likely to be no higher than 0.01 (USEPA 1992)
IR = 1E-04 kg/day (USEPA 1989, 1991b)
B-3
-------
EForal = 350 days/yr (USEPA 1989, 1991b)
• AFo =1.0 for arsenic (assumed)
Based on these inputs, the estimated ratio of dermal risk to ingestion risk for arsenic in soil is:
Relative Risk = (2E+03-1E-06-50-0.01)/(1E-04-350-1.0) = 0.029 (2.9%)
Thus, the relative risk from dermal contact with arsenic in soil compared to ingestion exposure is
likely to be no more than about 3%, and could be less if the frequency or extent of dermal
contact is lower than assumed, or if the dermal absorption fraction for arsenic is lower than
0.01. On this basis, it is concluded that dermal absorption is a minor contributor of risk
compared to oral exposure, and that this pathway may be excluded from quantitative evaluation.
3.0 REFERENCES
USEPA. 1989. Risk Assessment Guidance for Superfund. Volume I: Human Health Evaluation
Manual Part A. Interim Final. Office of Solid Waste and Emergency Response
(OSWER), Washington, DC. OSWER Directive 9285.701 A.
USEPA. 1991a. Risk Assessment Guidance for Superfund. Volume I: Human Health
Evaluation Manual (Part B, Development of Risk-Based Preliminary remediation
Goals). Interim. Office of Research and Development, Washington, DC. EPA/540/R-
92-003.
USEPA. 1991b. "Standard Default Exposure Factors." Supplemental Guidance for Risk
Assessment Guidance for Superfund, Volume I: Human Health Evaluation Manual.
OERR, Washington, DC. OSWER Directive 9285.6-03.
USEPA. 1992. Dermal Exposure Assessment: Principles and Applications. Interim Report.
Office of Research and Development, Washington, DC. EPA/600/8-91/01 IB.
USEPA. 1996. Soil Screening Guidance: User's Guide. Office of Solid Waste and Emergency
Response, Washington DC. Publication 9355.4-23. July 1996.
IRIS. 2000. Retrieval from USEPA's Integrated Risk Information System (IRIS). February, 2000.
B-4
-------
APPENDIX C
RISK-BASED CONCENTRATION VALUES
FOR WORKERS
-------
APPENDIX C
RISK-BASED CONCENTRATION VALUES
FOR WORKERS
1.0 OVERVIEW
A Risk-Based Concentration (RBC) is a concentration of a chemical in a medium that is not of
health concern to a specified population under a specified set of exposure assumptions. RBC
values are derived by reversing the risk assessment process, solving for the concentration of a
chemical that corresponds to a specified target risk value. This Appendix calculates the RBC
values for exposure of workers to arsenic and lead in soil. These values may then be used to
assess whether there is a need for quantitative evaluation of risk to this population.
2.0 RBC FOR ARSENIC
The basic equation used to calculate the RBC for exposure of workers to arsenic in soil is:
RBC =
Target Risk
IR
Input values applicable to worker exposure to soil are listed below, along with the resulting RBC
value.
Parameter
Target Risk
IR (kg/day)
BW (kg)
EF (days/yr)
ED (years)
AT (years)
RBA
oSF (mg/kg-d)'1
RBC (mg/kg)
Default Value
1E-04
1E-04
70
250
25
70
0.42
1.5
454
Source
USEPA 1991b
USEPA 199 la
USEPA 199 la
USEPA 199 la
USEPA 199 la
USEPA 199 la
USEPA 200 Ib
IRIS 2000
Calculated
C-l
-------
3.0 RBC FOR LEAD
The EPA has not established a default soil action level for lead for protection of workers.
However, the EPA has developed an interim method for calculating the risk to workers from lead
in soil (USEPA 1996). The basic equation is:
GM PbB = PbBO + PbS-BKSF-IRs-AFs-EFs/AT
where:
GM PbB = Geometric mean blood lead (ug/dL) in a population of workers
PbBO = Baseline geometric mean blood lead value (ug/dL) in the workers in the
absence of occupational exposure
BKSF = Biokinetic slope factor (ug/dL increase in blood lead per ug/day of lead
absorbed)
PbS = Concentration of lead in soil (ug/g)
IRj, = Intake rate of soil (g/day)
AFS = Absorption fraction for lead from soil. This value is given by:
AFS= AFfood-RBAsoil
EFS = Exposure frequency to soil (days/yr)
AT = Averaging time (days)
Given the GM PbB, and assuming the distribution of PbB values is lognormal with a geometric
standard deviation of GSD, the 95th percentile of the distribution is given by:
1.645
95th = GM-GSD
The subpopulation of primary concern for protection of workers from excessive lead exposure is
pregnant females. The goal is to ensure that there is no more than a 5% chance that the blood
lead level of the fetus will exceed 10 ug/dL. The ratio between the blood lead concentration in
the mother and the fetus is given by:
R(fetal/maternal) = PbB(fetus) / PbB(mother)
Default input values recommended by USEPA for each of these parameters are summarized in
Table C-l. Using these inputs, the concentration of lead in soil which yields a 95th percentile
value of 10 ug/dL in the blood of the fetus may be calculated. This value is 1,545 ppm.
C-2
-------
TABLE C-l DEFAULT INPUT PARAMETERS
ADULT WORKERS LEAD EXPOSURE MODEL
INPUTS
PbBO 2.0 ug/dL
BKSF 0.4 ug/dL per ug/day
IRsoil 0.05 g/day
EFsoil 219 days/yr
AT 365 days/yr
AFfood 0.2
RBAsoil (a) 0.84
R(fetal/maternal) 0.9
GSD 1.8
CALCULATED VALUES
Target 95th (maternal) 11.1 ug/dL
Target GM (maternal) 4.23 ug/dL
AFsoil 0.17
RESULT
RBC 1104 ug/g
(a) Site-specific value estimated from studies in animals (USEPA 200 Ic)
-------
APPENDIX D
MONTE CARLO MODELING OF EXPOSURE AND RISK
FROM ARSENIC IN SOIL AT THE VBI70 SITE
-------
APPENDIX D
SCREENING LEVEL MONTE CARLO MODELING
OF EXPOSURE AND RISK FROM ARSENIC IN SOIL
AT THE VBI70 SITE
1.0 INTRODUCTION
Monte Carlo modeling is a computer-based mathematical technique that may be used for
calculating exposure and risk where input terms are characterized as Probability Density
Functions (PDFs) rather than point estimates. This approach has the advantage that the full
distribution of exposure and risk may be predicted (as opposed to two point estimates, the CTE
and RME values), and that the percentiles of those estimates may be quantified. In addition, the
Monte Carlo approach helps guard against "compounding conservatism", whereby a series of
conservative assumptions are combined into a single but unlikely scenario.
2.0 BASIC EQUATIONS
The basic equations used to calculate risk using the Monte Carlo approach are identical to those
used in the point estimate approach. These equations are presented in Section 4.2 of the main
risk assessment.
3.0 SELECTION OF INPUT VARIABLES
In concept, every term used in the point estimate equation is a variable, and could be modeled as
a probability density function (PDF). However, for simplicity, it is generally not necessary to
evaluate every term as a PDF. Rather, only those terms that are the most variable and which are
the primary sources of variability in the output (exposure and risk) need be modeled as PDFs.
For this screening level evaluation, the following inputs are judged to be the chief sources of
variability in exposure and risk among individuals:
Exposure frequency (EF)
Exposure duration (ED)
Intake rate for soil and dust (IRsd)
Fraction of intake that is soil (Fs)
Vegetable intake rate (IRveg)
The distribution functions selected to model each of these variables are described below.
D-l
-------
Exposure Frequency (EF)
Exposure frequency is the average number of days per year spent at home. No data were located
on the distribution this variable, so a triangular distribution was selected, as follows:
EF ~ TRI(200, 234, 365)
The central tendency value of 234 days/yr is based on the default CTE value recommended by
EPA, while the upper bound would represent the case where a person was at home continuously.
This distribution yields a average value of 266 days per year (somewhat higher than the EPA
default of 234 days/year for the CTE resident), and a 95th percentile value of 332 days per year
(slightly lower than the EPA default of 350 days/year for the RME receptor).
Exposure Duration (ED)
Data on the length of time that people live in a specific residence are available in the Exposure
Factors Handbook (USEPA 1997) (see Table 15-167). The empiric cumulative distribution
based on data from 500,000 individuals is shown in Table D-l.
Soil and Dust Intake Rate (IRsd)
Two alternative distributions were used to evaluate soil and dust intake by children. The first is
a lognormal distribution selected to match the USEPA default values of 100 and 200 mg/day for
the CTE and RME child. The parameters of this distribution (mean and standard deviation) are
as follows:
IR(soil,dust)child~LN(100,53)
The second distribution is an empiric cumulative distribution based on the recent study by
Stanek and Calabrese (1999). The study included observations on 64 children for a period of 2-7
days (a total of 331 child-days). The parameters of this distribution are shown in Table D-l.
Fraction Soil (Fs)
Data on the fraction of total intake of soil plus dust that is soil are very limited. Stanek and
Calabrese (1992) analyzed data from 64 pre-school children over a 2-week period. The data
ranged from a minimum of zero percent up to a maximum of 100%, and the cumulative
distribution was very nearly equal to a straight line. On this basis, Fs was modeled as a uniform
distribution with parameters (0,1).
D-2
-------
Vegetable Intake Rate
Data on seasonally adjusted consumer-only intake of home grown vegetables, stratified by
region, are provided in the Exposure Factors Handbook (Table 13-33). The empiric cumulative
distribution function is shown in Table D-l.
Other Inputs
All other exposure and risk model terms were the same as used in the point estimate calculations.
4.0 RESULTS
Table D-2 shows the results of a Monte Carlo simulation at an exposure point where the
concentration of arsenic in soil (fine fraction) is assumed to be 200 ppm. Similar results are
obtained at other soil concentrations.
Figure D-l plots the distribution of cancer risks from ingestion of soil and dust at this location
(concentration in fines = 200 ppm). The two curves shown in the figure represent the results for
the two different PDFs assumed for soil intake (see above). Inspection of this figure reveals the
following main points:
1. The distribution of risks based on the soil intakes reported by Stanek and Calabrese
(2000) are substantially lower than the values based on the EPA default intake
parameters
2. Compared to the distribution that assumes default EPA intake rates, the CTE point
estimate is lower than the mean of the distribution, and corresponds to the 56th
percentile. The RME point estimate is substantially higher than the 95th percentile of the
distribution, and corresponds to a value above the 99th percentile.
3. Compared to the distribution that assumes the soil intake data of Stanek and Calabrese
(2000), the CTE point estimate corresponds to the 86th percentile, while the RME point
estimate corresponds to a value well above the 99.9th percentile.
These results indicate that RME point estimates of risk are likely to be conservative (i.e., will
provide protection to more than 95% of the exposed population), especially if soil intake is
actually closer to the data of Stanek and Calabrese (2000) than to the EPA defaults.
Figure D-2 compares point estimates and Monte Carlo estimates of total risk from arsenic (the
sum of exposure via vegetable intake and soil/dust intake) across a range of soil concentrations.
In all cases, the Monte Carlo calculations assume a soil intake that is lognormal and the
parameters are matched to the EPA defaults. The upper panel compares the CTE point estimate
of risk (CTE soil + CTE vegetable) with the mean of the Monte Carlo simulation. As noted
above, at any specified soil level, the point estimate of CTE risk is below the mean value of the
D-3
-------
MCA. The lower panel compares the 95th percentile of the MCA with three alternative
estimates of the total RME risk:
Method 1 = RME soil + CTE vegetables
Method 2 = CTE soil + RME vegetables
Method 3 = RME soil + RME vegetables
As seen, the 95th percentile of total risk calculated by MCA is lower than the point calculations
of RME total based on Method 1 (used in this risk assessment) at all soil levels. As expected,
Method 3 (RME soil + RME vegetable) yields a result much higher than Method 1 or the MCA
value. These results provide assurance that the estimates of total risk calculated across pathways
calculated using Method 1 are likely to be conservative (higher than actual).
D-4
-------
TABLE D-l EMPIRIC DISTRIBUTION FUNCTIONS
USED IN MONTE CARLO MODELING
Exposure Duration
Years
EFH Table 15-167
1
1.9
2
3
9
16
26
33
41
47
55
59
87
0.00
0.05
0.10
0.25
0.50
0.75
0.90
0.95
0.98
0.99
0.998
0.999
1.000
Soil Intake
mg/day
(Stanek and Calabrese 1999)
0
2
9
16
21
24.5
29
35
53
75
91
137
173
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
0.80
0.90
0.95
0.99
1.00
Veg Intake
kg ww/kg bw/day
EFH Table 13-33
1.80E-03
1.91E-02
3.83E-02
1.14E-01
4.92E-01
1.46E+00
2.99E+00
5.04E+00
8.91 E+00
1.12E+01
0.00
0.05
0.10
0.25
0.50
0.75
0.90
0.95
0.99
1.00
D-5
-------
TABLE D-2 MONTE CARLO RESULTS
Soil Cone = 200 ppm in fines
Soil IR
LN(100,53)
Empiric
(see Table D-1)
Percentile
0.050
0.250
0.500
0.750
0.900
0.950
0.990
0.999
0.050
0.250
0.500
0.750
0.900
0.950
0.990
0.999
Risk(s+d)
1E-06
4E-06
9E-06
2E-05
4E-05
6E-05
1E-04
2E-04
7E-08
6E-07
2E-06
6E-06
1E-05
2E-05
5E-05
1E-04
Risk(veg)
2E-08
2E-07
9E-07
3E-06
9E-06
1E-05
3E-05
8E-05
2E-08
2E-07
9E-07
3E-06
8E-06
1E-05
4E-05
7E-05
Risk(total)
1E-06
5E-06
1E-05
3E-05
5E-05
7E-05
1E-04
2E-04
3E-07
1E-06
4E-06
1E-05
2E-05
3E-05
6E-05
1E-04
D-6
-------
FIGURE D-1
MONTE CARLO RESULTS FOR EXPOSURE TO ARSENIC IN SOIL/DUST
Concentration in Fine Fraction = 200 ppm
1.0
0.8
I 0.6
o
CD
E
D
O
0.4
0.2 -
0.0
Soil intake modeled
with data from
Stanek and
Calabrese 2000
Soil intake
modeled as
lognormal and
parameters
matched to EPA
default values
CTE Point
Estimate of
Cancer Risk
RME Point
Estimate of
Cancer Risk
1E-07
1E-06
1E-05
Risk
1E-04
1E-03
D-7
-------
FIGURE D-2
COMPARISON OF POINT ESTIMATE AND MONTE CARLO ESTIMATES
OF TOTAL RISK ACROSS A RANGE OF ARSENIC CONCENTRATIONS IN SOIL
CTE (Mean)
5E-05
w 4E-05
hi
s_
CD
O
co 3E-05
O
LU
2E-05
to
OE+00
CTE Point Est
•MCA Mean
100 200 300 400 500
Concentration in Soil (fine fraction) (ppm)
600
RME (95th)
4E-04
t 3E-04
CD
O
co
O
g 2E-04
or
T3
JD
"ro
E 1E-04 H
LU
OE+00
RME Point Estimate 1
RME Point Estimate 2
RME Point Estimate 3
MCA 95th
100 200 300 400 500
Concentration in Soil (fine fraction) (ppm)
600
Monte Carlo evaluation assumes soil intake is distributed lognormally with a mean of 100 mg/day
and a standard deviation of 53 mg/day (95th percentile = 200 mg/day)
D-8
------- |