United States
Environmental
Protection Agency
Office of Air Quality EPA-450/2-89-011
Planning and Standards ^ Jurte 1989
Research Tnangle Park NC 27711
O'EPA REVIEW OF THE NATIONAL AMBIENT
AIR QUALITY STANDARDS FOR LEAD:
EXPOSURE ANALYSIS METHODOLOGY
AND VALIDATION
OAQPS STAFF REPORT
-------
This report has been reviewed by the Office of A1r Quality Planning and
Standards, U.S. Environmental Protection Agency, and approved for publication.
%
Mention of trade names or commercial products 1s not tnteided to constitute
endorsement or reconnendation for use.
-------
REVIEW OF THE NATIONAL AMBIENT AIR QUALITY STANDARDS FUR LEAD:
EXPOSURE ANALYSIS METHODOLOGY AND VALIDATION
OAQPS STAFF REPORT
Air Quality Manayement Division
Office of Air Quality Planniny and Standards
U.S. Environmental Protection Ayency
Research Trianyle Park., N.C. 27711
-------
Acknowledgments
This staff paper is the product of the Office of Air Quality
Planning and Standards. The principal author is Jeff Cohen, with
assistance from Allan Marcus, Donna Sledge, Dave McLamb, Gail
Brion, and John Haines. Technical contributions were provided by
New York University Medical Center, PEI Associates, Inc., and
Alliance Technologies Corporation. The report incorporates
comments from OAQPS, the Office of Research and Development, the
Office of Drinking Water, the Office of Policy, Planning, and
Evaluation, and the Office of General Counsel within EPA and was
formally reviewed by a subcommittee of the Clean Air Scientific
Advisory Committee.
Helpful comments were also submitted by a number of
independent scientists and by representatives of the U.S. Centers
for Disease Control, the Lead Industries Association, the
Canadian Department of Health and Welfare, the Canadian Ministry
of the Environment, and the California Department of Health
Services.
The authors wish to thank Tricia Holland, Teresa demons,
and Virginia Wyatt for excellence in word processing and numerous
other tasks and Dick Atherton for graphics assistance.
-------
iii
TABLE OF CONTENTS
Page
List of Figures v
List of Tables v
Executive Summary vii
I. Introduction 1-1
II. Lead Exposure: Multimedia Considerations 11-1
A. Airborne Lead 11-3
B. Lead in Soil 11-5
C. Lead in Dust II-9
D. Lead in the Diet 11-12
E. Lead in Water 11-13
F. Lead in Paint 11-14
III. Estimating Lead Exposures and Blood Lead Levels III-l
A. Alternative Modeling Approaches 111-1
B. Calculating Blood Lead Distributions Around Predicted
Means 111-2
IV. Integrated Lead Uptake/Biokinetic Model IV-1
A. Estimates of Lead Uptake IV-3
B. Uptake and Blood Lead Concentrations IV-6
V. Statistical Relationships Between Blood Lead and Airborne
Lead: Use in Disaggregate and Aygreyate Models V-l
A. Disaggregate Model V-3
B. Aggregate Model V-14
VI. Validation of Integrated Lead Uptake/Biokinetic Exposure
Model VI-1
A. Validation Using East Helena Data VI-2
B. Application of Uptake/Biokinetic Model to Other Data
Sets V1-12
C. Conclusions VI-17
1
-------
1v
TABLE OF CONTENTS (Continued)
Paye
VII. Conclusions and Applications of Lead Exposure
Modeling V11-1
Appendix A. Estimates of Lead Uptake A-l
Appendix B. Estimation of the Relationship Between Soil, Dust,
and AmDient Lead B-l
Appendix C. Estimating 1990 Baseline Blood Lead Averayes C-l
Appendix D. CASAC Closure Report D-l
References
i
-------
V
LIST OF FIGURES
Number
2-1
2-2
4-1
6-1
6-2
Title Page
Principal Pathways of Human Exposure to Lead
and Subsequent Physiological Distribution 11-2
Trends in Maximum Quarterly Lead Concentrations for
Various Monitor Types 11-6
Summary of Relationships Between Daily Lead Uptake
and Blood Lead for Different Aye Groups IV-17
Comparison of Distribution of Measured Blood Lead Levels
in Children, 1-5 years of Age, Living Within 2.2b miles
of E.Helena Smelter vs. Levels Predicted by Uptake/Biokinetic
Model (Run B) V1-8
Comparison of Distribution of Measured Blood Lead Levels
in Children, 1-5 years of Age, Living Within 2.25 miles
of E.Helena Smelter vs. Levels Predicted by Uptake/Biokinetic
Model (Run C) V1-9
LIST OF TABLES
Number T itle Page
2-1 Typical Lead Concentrations in Various Exposure
Media . . . 11-2
2-2 Frequency Distributions of Maximum Quarterly Lead
Concentrations by Type of Site 11-6
3-1 Comparison of Estimated GSDs Across Several Studies . . . 111-b
3-2 Illustrative Blood Lead Distributions Assuming Different
Mean Blood Leads and Geometric Standard Deviations . . . Ill-8
4-1 Estimated 1990 Average Lead Intake and Uptake in 2-Year
Children Under Average Air Lead Levels IV-4
-------
vi
4-2 Predicted Equilibrated Blood Lead Levels Over Time Among
Children with Constant Lead Uptakes IV-15
5-1 Disaggregate Model of Contributions from Various Media to 1990
Median Blood Lead Levels of U.S. Children (2 Years of Age):
Background Levels and Incremental Contributions from Air. . . V-9
5-2 Blood Lead/Air Lead Slopes in Children for Different Ayes
and Exposures V-lb
6-1 Comparison of Integrated Lead Uptake/Biokinetic Model
Predictions to 1983 Measurements in East Helena V1-6
6-2 Children's Blood Lead Levels Measured in Omaha, 1971-1977
Vs. Integrated Uptake/Biokinetic Model Predictions VI-14
6-3 Children's Blood Lead Levels Measured in Silver Valley,
Idaho 1974-1975 Vs. Integrated Uptake/Biokinetic Model
Predictions V I-16
A-l Age-Specific Estimate of Total Dietary Lead Intake for
1990-1996 A-5
A-2 Total Dietary Lead Intake for 1978-1983 for Five Age
Groups of Children A-5
A-3 Summary of Environmental Lead Measurements from Various
Point Source Locations A-8
A-4 Long-Term Relationships Between Dust, Soil .and Ambient
Air Lead. A-13
A-5 Estimating Short-Term Responses in Equilibrium Soil and
Dust Lead Levels from Changes in Ambient Air Lead A-13
B-l Environmental Lead Concentrations in Omaha, Nebraska. . . . ts-b
B-2 Linear Model Parameter Estimates for Air: Soil and Oust
Lead Relationships B-8
B-3 Average Dust and Soil Lead vs. Air Lead Relationships
% by Location B-8
C-l Selected 1978 Blood Lead Levels from NHANES II C-2
C-2 Predicted Changes in Average Blood Lead Levels Associated
with Gasoline Lead Trend C-6
C-3 Trends in Dietary Lead Intake: Total Intake (and
Atmospheric Contributions) C-8
C-4 Adjustments to 1978 Mean Blood Lead Levels for Dietary
Lead Reductions C-l 1
C-5 Derivation of 1990 "Baseline" Average Blood Lead Levels . . C-14
C-6 Maternal:Child Blood Lead Concentrations and Ratios .... C-16
-------
EXECUTIVE SUMMARY
For its review of the National Ambient Air Quality Standards (NAAQS) for
lead, EPA is assessing health risks associated with lead exposure, particularly
near stationary sources of atmospheric lead emissions (e.g., lead smelters),
under alternative regulatory scenarios. A critical element in this
process will be an exposure analysis whereby blood lead levels are estimated
among populations exposed under alternative lead regulations in the
future. This report summarizes relevant information on lead exposure and
presents the modeling methodologies that the staff believes should be
considered for the lead NAAQS exposure analysis. Results of validating
one of these methodologies are also presented.
This report does not discuss health effects associated with lead, or
the implications of different levels of lead exposure. That information
is reviewed in the criteria document and will be integrated into a separate
staff paper along with results of exposure analyses that employ the methodologies
described here. The following discussion highlights the major points and
conclusions discussed in this paper.
Multiple Pathways of Exposure
Human exposure to lead occurs through multiple pathways, and can be
traced primarily to lead in paint pigments, solder in canned foods and
plumbing, and atmospheric emissions from motor vehicles and stationary lead
sources. Although airborne lead is a principal starting point of environmental
contamination, oral intake of deposited atmospheric lead is often the
primary identifiable factor predicting the blood lead of young children.
The principal pathways of human exposure can be seen in Figure 1.
-------
viii
Figure 1. PRINCIPAL PATHWAYS OF HUMAN EXPOSURE TO LEAD AND SUBSEQUENT
PHYSIOLOGICAL DISTRIBUTION.
Downward Trends 1n lead Exposure
lead contamination of the human environment has been occurring for
thousands of years since its many uses began to be discovered. Widespread,
global dispersion of lead was greatly accelerated beginning in the 19201s
with its introduction as an anti-knock compound in gasoline. Up until
<•
recently, approximately 90% of airborne lead in the U.S. was due to
automotive emissions. Use of lead in gasoline has declined by aDout 90%
since 1978; this trend will continue as compliance with the 1986
gasoline phasedown standard is completed and as the fleet of lead-burning
cars shrinks. This trend has been paralleled by a major reduction in
ambient air lead levels, which now average between 0.1 and 0.3 yg/m3 in
most U.S. cities without lead point sources.
The reduction in atmospheric lead, combined with the gradual phaseout
of lead solder in canned food manufacturing and the ban of lead solder in
-------
plumbing has resulted in a significant decline in dietary lead intake.
For example, it is estimated that from 1978 to 1985, average dietary lead
intake in a 2-year child dropped from 52.9 micrograms (pg) per day to
10.4 gg/day. This trend is expected to continue.
These recent changes in nationwide exposure and expected future
trends, as well as contributions from the other air and non-air sources
of lead exposure (e.g., drinking water, contaminated soils) are important
components of the methodologies described in this report. The air lead
contributions include both direct inhalation of lead particles and indirect
exposure through ingestion of lead in dust and in soil.
Lead-based paint, however, will continue to be the major source of
high-dose lead exposure and symptomatic lead poisoning for children in
the U.S. for years to come. High exposures to lead in older housing with
deteriorating or even intact lead-painted surfaces (via chalking or
weathering) will not be significantly influenced by any changes in atmospheric
lead emissions. Because children excessively exposed to lead-based paint
cannot be expected to be protected by any lead NAAQS, no matter how stringent,
they will not be included in the exposure analyses described in this
report. Available data are presented to demonstrate how such estimates
could be generated. Comprehensivew pians for safe and effective removal
or containment of lead painted surfaces from old housing need to be
developed and implemented.
Focus on Point Sources
The focus of the exposure modeling approaches described in this
report, and the overall lead NAAQS review, is on stationary lead sources
such as primary and secondary lead smelters and battery plants. Although
such sources have contributed little to the overall pollution load across
-------
X
large, regional areas, fallout from these sources can be severe on a
local scale. Ambient lead levels around such point sources vary depending
on distance, topography, wind speed and direction, plant operating parameters,
emissions controls, particle sizes and deposition rates, etc.
Data Limitations Preclude Quantitative Estimates for All Sensitive Sub-Groups
To assess the health risks associated with alternative air lead
standards, it is necessary to estimate the blood lead (PbB) levels that
would be distributed in the populations of concern under various air
lead concentrations. Young children, pregnant women (as exposure surrogates
for the fetus), and middle-aged men, are identified as particularly
susceptible to lead-related risks. Quantitative estimates of blood lead
distributions under alternative standards will be made only for young
children and middle-aged men.
Fetal exposure to lead, in most cases, can be expected to be dominated
by maternal bone lead stores from past exposures. Therefore, accurately
predicting future changes in fetal blood lead levels would require estimates
of maternal lead burden and information on how bone lead stores would be
mobilized and transferred across the placenta. Although it is likely
that there is extensive mobilization of lead, like calcium, during periods
of physiological stress.such as pregnancy, there are no biokinetic data
to quantify this dynamic process. In the absence of such data, no attempt
will be made to estimate fetal lead exposures associated with maternal
PbB levels under alternative standards. Given the sensitivity of the
fetus and neonate, however, potential risks associated with pre- and neo-
natal exposures will be of major emphasis in the overall lead NAAQS
assessment.
-------
Older women may also be at increased risk due to release of
bone lead after menopause due to the processes of osteoporosis. A lack
of biokinetic information on post-menopausal demineralization precludes
quantitative estimation of blood lead changes under alternative lead
NAAQS for these individuals as well.
Average Estimates Predicted by Exposure Models Will be Combined With
Empirical Estimates of Variance to Calculate Blood Lead Distributions
This paper presents three approaches that can be used to estimate
the impact of inhaled and ingested lead aerosols and compounds on PbB
levels distributed in populations living near lead point sources. Because
most of the data input to these models are generally average estimates,
all three models first estimate PbB levels in terms of population mean
levels. Given the significant behavioral and biological variability
within populations, and the importance of highly exposed individuals within
the population at risk, it is necessary to determine the entire blood
lead distribution corresponding to a given mean PbB. This allows, for
example, percentages of the population expected to exceed PbB levels of
concern to be ascertained. Because.PbB levels are typically distributed
lognormally, it is possible to calculate percentiles of a blood lead
distribution around a given mean PbB level by using the geometric standard
deviation (GSD) of that population's blood lead distribution.
Several epidemiological studies have indicated that the log values
of measured individual PbB levels in a uniformly exposed population are
normally distributed with a variation, including analytical variation,
ranging between 1.3 and 1.4. For purposes of predicting future exposures,
a range of GSD values of 1.30-l.b3 has been documented for children
living near lead point sources, and the midpoint of 1.42 will be assumed
-------
xii
as a reasonable best estimate. A GSD value of 1.37 derived from the
NHANES II survey will be used for adults.
The three models are summarized below.
Uptake/Biokinetic Model: Applicable to Children from Birth
The first modeling approach presented is the uptake/biokinetic
model which uses measured rates of absorption, or "uptake" of lead through
different pathways (e.g., inhalation, ingestion) from experimental studies
together with available mathematical ("biokinetic") models from lead
balance studies to project either total body burden or the amount of lead
in any of the presumed "physiological" kinetic compartments (e.g., blood,
soft tissue, bone) at any time. The model is thereby capable of reflecting
"non-equilibrium" lead exposures. It has been developed at this time only
for young children.
The model is based on age-specific air, soil, dust, food, and water
lead intakes, and on age-specific absorption factors. Blood lead and
tissue lead distribution from absorbed lead is described by a linear
pharmacokinetic model whose parameters change with aye. The air lead
concentrations around a point source can be estimated from historical data
on emissions and site-specific information about meteorological dispersion
parameters. To address situations when this information is not known, as
in the case of assessing future NAAQS scenarios, a methodology was developed
to estimate the associated soil and household dust lead concentrations
using relationships derived from available data from a wide range of
point source areas. This approach allows estimation of effects on soil
and dust lead of historical changes or proposed changes in air lead levels.
Dietary lead intake estimates are derived from historical and current
data and future projections of the Multiple Source Food Model, originally
-------
xiii
developed in the 1986 criteria document, and updated to include more
recent information.
Because the biokinetic model assumes linear rates of lead transfer
between tissues, it is limited to predicting low-moderate level exposures
(i.e., PbB < 25-30 gg/deciliter, or pg/dl). The model's mathematical
assumptions and numerical parameters, however, combine plausible biological
hypotheses, animal experimental data, and results of observational studies.
Further, it allows explicit projections of future lead concentrations in
various media and in turn can estimate the impacts of these different
changes on different age groups of children. It is this flexibility that
makes the integrated uptake/biokinetic model adaptable for a wide range
of predictive exposure assessments, and why it was the focus of the validation
exercises described in this paper.
Aggregate Air Lead Model: Applicable to Youny Children
The second approach, referred to as an "aggregate" model, uses a
directly applied mathematical relationship between air lead and blood
lead derived from community epidemiological studies that includes both
direct inhalation exposure and indirect exposures via secondary deposition
processes. This approach requires an implicit assumption that dust and soil
lead concentrations are approximately in equilibrium with air lead levels,
and that the present air lead exposures reflect historical levels. This
model also requires estimated contributions to children's blood from
non-air sources of lead exposure, which are derived for current and
future years based on such nationwide data as gasoline usage, dietary
patterns, dietary lead concentrations, and blood lead surveys. Because
aggregated air lead:blood lead relationships are not available for adults,
and since the importance of indirect oral exposures to deposited atmospheric
-------
xiv
lead is relatively minor among adults, this approach is limited to young
children.
This type of model is best used on a site specific basis due
to the large number of confounders and covariates such as industrial
exposure, age of housing, and social conditions. A summary of key studies
useful in deriving aggregate blood lead/air lead slopes for children is
included, along with a plausible range for these slopes (3-5 pg lead/dl
blood per pg lead/cubic meter of air).
Disagyreyate Air Lead Model; Applicable to Children and Adults
The third approach, referred to as a "disaggregate" model, is a
hybrid of the first two that combines separate empirical relationships
from multiple regression models between 6lood lead and lead intake from
air, food, water, dust, and soil lead sources. This model.was presented
in the 1986 criteria document and is updated here to include more recent
data. A number of experimental exposure and epidemiological studies, in
which such data were available are discussed.
In this model, there is an implicit assumption that the soil and
dust lead levels measured at some time reflect earlier exposures, so that
the PbB levels estimated are essentially at equilibrium and not just
reflecting some recent change or event.
The most plausible disaggregate relationship, or "slope" between
inhaled air lead and children's blood lead, derived from analysis of 3
point source studies, is 1.97 py/dl per py/m3. The average inhalation
slope for adults derived from the most reliable experimental and epidemiology
studies is 1.4 jjy/dl per pg/m3. A factor of 1.3 is applied to account
for the resorption of lead from bone in adults and a slope of 1.8 py/dl
per pg/m3 is derived. The slope values used in the "disayyregate" model
-------
XV
are In general, consistent with those derived from the "uptake/ biokinetic"
model.
Uptake/Biokinetic Model Validations Successful; Other Applications
Should Proceed with Caution
Results of several validation exercises utilizing the uptake/biokinetic
model comparing predicted and observed blood lead levels are presented
in this paper. The most detailed analysis was performed with data gathered
around the smelter in East Helena, Montana. Two types of validation efforts
were undertaken with this data set: 1} in the first effort, the best
historical data regarding such model parameters as measured air, yard soil,
and household dust lead exposure estimates were used; 2) in the second
validation effort, predicted levels of air (using dispersion modeliny that
accounted for fugitive emissions and background contributions), and soil
and dust (using generalized relationships derived from empirical analyses
of a wide range of point source data) were used to estimate PbB levels.
The latter work was undertaken to determine how well the model behaved
when actual measurements of necessary input data are not available. This
was necessary so that the reliability of the model could be assessed for
policy analysis purposes when less than full information is available to
use with the model. Results of the different East Helena validations
indicate good concordance between observed and predicted average PbB
levels in children living near the smelter. The differences observed
were not statistically significant.
Less extensive validation exercises on other published data are reported
in this paper to test the uptake/biokinetic model at other U.S. locations
and times including Omaha, Nebraska, from 1971 to 1977 and Silver Valley,
Idaho, in 1974-1975. In these other validations, the model performed
-------
xvi
reasonably well in predicting average PbB levels below 25-35 pg/dl. As
expected at higher exposure levels, the linear model underestimated actual
exposures.
Given the many uncertainties in the input data and the biological
variability that cannot be incorporated, the results of the validation
exercises presented reveal that the lead uptake/biokinetic model performs
quite well in predicting mean blood lead concentrations in children
living near point sources of lead at exposure levels of current relevance.
Using this methodology, along with the approach to estimate lognormal
blood lead distributions using empirically-derived GSDs, exposure analyses
of childhood populations living near stationary lead sources (i.e., "case
studies") will be prepared. Blood lead distributions for middle-aged men
in these areas will also be estimated using the disaggregate approach.
Results of these analyses will be incorporated into the staff paper to
better inform the Administrator on the impacts of alternative lead NAAQS.
The uptake/biokinetic model'can also be a useful tool in estimating
PbB levels in children living with different lead hazards, such as heavily
contaminated soils from historical deposition near major urban roadways
or closed smelters or mines. Ongoing regulatory efforts by different
components of EPA to control concentrations of lead in air, water and
soil have created a significant need to model blood lead concentrations
that delineates specific routes of multi-media lead exposure. The
"uptake/biokinetic" model adequately provides such specificity for
young children. Further refinements are underway to include non-linearities
in absorption rates and biokinetics. Until then, the model is limited to
estimating relatively low to moderate exposures and should be used with
caution in other applications.
-------
REVIEW OF THE NATIONAL AMR IENT AIR QUALITY STANDARDS FOR LEAD:
Exposure Analysis Methodology and Validation
I. INTRODUCTION
EPA is assessing health risks associated with lead exposure, especially
near stationary sources of atmospheric lead emissions (e.g., lead smelters),
under alternative regulatory scenarios for its review of the National
Ambient Air Quality Standards (NAAQS) for lead. The scientific and
technical information on lead has been reviewed in EPA's "Air Quality
Criteria for Lead" (EPA, 1986). A "staff paper", in final preparation,
will evaluate and interpret the most relevant of that information to help
the Administrator in selecting the averaging times, forms, and levels for
the primary and secondary standards.* A critical element in this process
will be an exposure analysis whereby blood lead levels are estimated
among populations exposed under alternative lead NAAQS in the future.
This report summarizes relevant information on lead exposure and
presents the modeling methodologies that the staff believes should be con-
sidered for the lead NAAQS exposure analysis. Results of validating one
of these methodologies are also presented. Using methodologies described
and validated in this report, exposure analyses of populations living near
stationary lead sources (i.e., "case studies") are being prepared. Results
of these analyses will be incorporated into the staff paper along with the
overall health risk assessment. The staff will use results of the case
studies to better inform the Administrator on the impacts of alternative
lead NAAQS.
*The current primary standard for lead (to protect public health) is 1.5
micrograms per cubic meter (iJg/m^), maximum arithmetic mean averaged over
a calendar quarter. The secondary standard (to protect public welfare)
is identical to the primary standard.
-------
1-2
Normally, material of this type would be incorporated into the staff
paper. Due to the complex nature of lead exposure analysis, we chose to
place this material in a separate report to achieve a more thorough
review and also to facilitate other applications of lead exposure analyses
unrelated to the NAAQS review. This report should then be considered as
a supplement to the OAQPS staff paper. It is intended for those readers
familiar with the technical information contained in the criteria document
(hereafter referred to as "CO"). This report does not discuss health
effects associated with lead or the implications of different levels of
lead exposure. That information is reviewed in the criteria document and
will be integrated in the staff paper with results of the exposure analysis.
This paper, as will be done with the staff paper, has been circulated for
review by the Clean Air Science Advisory Committee (CASAC) and the public.
A copy of the closure letter from the CASAC sub-committee that reviewed
the August 1988 draft report is contained in-Appendix D.
The approach used in this paper is to assess and integrate exposure-
related information derived from the criteria document review that the
staff believes should be considered in the review of the primary lead
NAAQS. Section II presents relevant features of human exposure to
atmospheric and non-atmospheric sources of lead through various pathways.
Section III introduces different approaches to estimate the impact of
alternative air lead levels on lead body burdens among different populations.
Sections IV and V discuss the methodologies of these different approaches
in detail. Section VI presents results of validation exercises using the
uptake/biokinetic model. Section VII discusses some applications of the
modeling approaches described in this report in addition to their role in
the review of the lead NAAQS.
-------
II. LEAD EXPOSURE: MULTIMEDIA CONSIDERATIONS
To assess the risks associated with alternative lead NAAQS, it is necessary
to understand the influence exerted by atmospheric lead on the total lead
exposure of the population(s) of concern through various exposure pathways.
Relevant data on these pathways and the relationships between air lead and
lead in other media are surranarized in this section. The goal of the exposure
assessments presented in this report is to address aggregate lead absorption
from many pathways for populations living near stationary lead sources that
will be affected by alternative lead NAAQS. Much of the methodoloyy presented
can also be applied to other populations with different exposure profiles.
Exposure situations (other than operating stationary lead facilities) dominated
by single, intense, and/or intermittent sources are not included for lack of
adequate data, and/or because any lead NAAQS would have little, if any, pro-
tective effect for those situations. They include most significantly, high paint
lead exposure (see Subsection II.F below) and occupational exposures. (Indirect
exposures of children to occupational dusts tracked home by working parents
are implicitly included in the modeling). Other miscellaneous intermittent
sources not addressed here include various consumer products (e.g., leaching
of lead-glazed pottery), and household activities or hobbies, such as stained
glass construction and removal of lead paint solderiny. See Section 7.3.2 of
the CD for a more complete discussion of these kinds of exposure sources.
The sources and pathways of human lead exposure are diagrammed in
Figure 2-1 and typical levels of lead in different media to which U.S.
populations are exposed are presented in Table 2-1. Human exposure to lead
can be traced primarily to lead in paint pigments, solder in canned foods and
plumbing, as well as to atmospheric lead. Although airborne lead is a
principal starting point of environmental contamination, oral intake of
deposited atmospheric lead is often the primary identifiable factor predicting
-------
II—2
AUTO
(MISSIONS
industrial
(MISSIONS
KIONCV
*
f«cu uumi
SOIL
SUIWACS ANO
WOUMOWATU
Figure 2-1. PRINCIPAL PATHWAYS OF HUMAN EXPOSURE TO LEAD AND SUBSEQUENT
PHYSIOLOGICAL DISTRIBUTION. ADAPTED FROM CD, FIGURE 7-1.
TABLE 2-1. TYPICAL LEAD CONCENTRATIONS IN VARIOUS EXPOSURE MEDIA
Median Rural Area Urban Area Near Point Source(s)a References
Ambient Air (ug/m^)'1
0.1
0.1 - 0.3
0.3 - 3.0
Faoro, 1988
Indoor Air (yg/o^)
0.03 - 0.08
0.03 - 0.2
0.2 - 2.4
See Footnote c
Soil (ppn)
5-30
30 - 4500
150 - 15,000
CD, Tahle 7-11; Mielke
et al.; 1983; LaBelle
et al.. 1987; See Table A-3
Street Dust (ppn)d
80 - 130
(90)
100 - 5,000
(1500)
(25,000)
Nriagu, 1978; CD,
Table 7-26
House Dust (ppm)d
SO - 500
(300)
SO - 3,000
(1000)
100 - 20,000
(10,000)
U.S. EPA, 1977;
Landrigan et al., 1975;
Morse et al., 1979;
Anqle and Mclntire, 1979
Typical Foods (dph)
0.002 • 0.8
Same
Same
Fleqel et al.. 1988
Hater (yg/1)
5 - 75 pg/L
Same
Same
Briskin, 1988
Painte
<1 - >5 mg/cm^
Same
Sane
Billick and Gray, 1978
'Within 2-5 km of sources including primary and secondary lead smelters, battery plants.
Represents quarterly averages Monitored in 1986.
cRange of indoor/outdoor ratios used (0.3 - 0.8) from CD, Table 7-6 except near point sources Wiere large particles pre-
dominate and infiltration into hones is low, ratio appears to be closer to 0.3 (Cohen and Cohen, 1980).
^Values in parentheses represent estimates provided in CD (Tables 7-18 and 7-20) as typical averages.
'Since there may be several layers of lead-based paint on a given surface, absolute concentration of lead is less
useful than mg/cm^. Surveys by HUD in Pittsburgh showed that more than 701 of pre-1940 dvelling units and 20* of post-
1960 units had at least one surface with more than 1.5 mq/cm? lead paint (NAS, 1980).
-------
11-3
the blood lead of young children. High correlations have been found between
children's blood lead and the lead content of household dust, playground
dust, mouthing behavior, lead on children's hands, and household cleanliness
(Sayre et al., 1974; Baker et al., 1977; Charney et al., 1980; Roels et
al., 1980, Yankel et al., 1977; Angle et al., 1979; Walter et al., 1980;
Brunekreef et al., 1981; Quah et al., 1982; Bellinger et al., 1986; Bornschein
et al., 1986; Schilling and Bain, 1988). Large scale isotopic studies
provide strong evidence that air lead has a rapid and measurable effect on
ingested lead, even in adult populations at low levels of air lead (Facchetti
et al., 1985). It is necessary, therefore, to attempt to quantify how
atmospheric emissions can influence human exposure, not only through direct
inhalation of lead-containing particles, but through ingestion of lead
deposited onto soil, dusts, vegetation, and other environmental surfaces.
Until recently, between 85-90% of airborne lead in the U.S. originated
from gasoline combustion, with most of the remainder from stationary industrial
processes such as primary and secondary lead smelting, battery plants, and
combustion of oil, coal, waste oil and municipal waste. Because of the
accelerated reduction of the lead content in gasoline and the continuing
phaseout of older lead-burning vehicles, concern regarding lead emissions
into the ambient air is shifting almost exclusively to relatively confined
areas surrounding significant stationary sources of lead and this paper's
focus will be on estimating exposures and risks in those locales.
A. Airborne Lead
1. Physical, Chemical and Spatial Characteristics
Ambient air lead levels result from current industrial emissions,
lead-contaminated road and wind-blown dust, automotive exhaust, and
solid waste combustion. The industrial contribution includes point, fugitive
-------
11-4
process, and materials handling emissions from lead mining, smelting, and
recovery operations as well as various fabrication and manufacturing processes.
Current contributions from roads and other fugitive dust sources vary with
location and the influence of historically contaminated soil.
In general, about 50% of automotive lead emissions deposits within a
few hundred meters of major roadways (Daines et al., 1970; Huntzicker et
al., 1975; Inyalls and Garbe, 1982), while the remaining particles are small
enough to remain airborne and can travel hundreds or thousands of kilometers.
This long-distance transport likely accounts for the surface contamination
of polar glaciers, oceans, and other remote locations around the ylobe
(Settle and Patterson, 1982).
In contrast to automobile exhaust, atmospheric lead emissions from
industrial plants that process lead and its products have contributed little
to the overall pollution load across large, regional areas although fallout
from these sources can be severe on a local scale. The trajectories of
atmospheric emissions from industrial sources vary with wind speed and
direction. High concentrations around lead smelters and other major
emitters can be dominated by fugitive emissions, predominately made up of
large [>7 micrometers (um)] lead particles resulting from materials
handling, furnace upsets, and furnace charging and tappiny operations
(Landriyan et al., 1975; Jennett et al., 1977; Battye et al., 198b).
Stack and vent emissions can also produce siynificant and frequent
short distance impacts, especially in complex terrain or around older
facilities (Iaccarino, 1988). Beyond the immediate area ( a 2 km) of lead
stationary sources, process emissions from stacks, predominately as lead
sulfates and oxides with a size range between 1 and 10 ym, become the major
source of lead in soils and air (Dorn et al., 1976; Roels et al., 1980;
Davidson and Osborn, 1984).
-------
11-5
2. Ambient Concentrations
As indicated in the CD (Section 7.2.1) and EPA (1989), air lead
levels in urban areas and near point sources have been markedly reduced
since 1977 by the use of unleaded gasoline in new cars equipped with
catalytic converters, the gradual phaseout of older lead-burning cars,
the lead-in-gasoline phasedown program, and steady reductions in emissions
from stationary sources in compliance with state implementation plans
toward attainment of the lead and particulate matter NAAQS and partly
because of decreased industrial production. Total atmospheric lead emissions
dropped 94% between 1978 and 1987; gasoline lead use dropped by 90% over tnat
period and industrial source emissions dropped by over one-half (EPA, 1989).
Recent (1980-1986) air quality data for both point source-oriented sites
(predominantly "SLAMS" sites), roadside sites ("NAMS", micro-scale) and other
sites (middle-scale, neighborhood), are summarized in Table 2-2 (Battye,
1988). By 1986, the only quarterly average concentrations over 1.5 pg/m3
were recorded at monitors near stationary sources. Twelve counties in
the U.S. reported quarterly averages above 1.5 pg/m3 in the 1986-1987
period; 13 more were on the margin of exceeding the NAAQS but either had
inadequate numbers of samples or were slightly below the standard level.
Trends in maximum quarterly averages between 1980 and 1986 are
illustrated in Figure 2-2. Even more dramatic declines are evident in data
confined to roadside monitoring (see Battye, 1988 and EPA, 1989).
B. Lead in Soil
The natural occurrence of lead in the earth's crust averages
5-50 pg lead/g soil in various soils* (CD, p. 7-28), although soils near
*0ne pg/g is equivalent to one part per million or 1 ppm.
-------
II-6
TABLE 2-2. FREQUENCY DISTS ISUT [ON OF MAXIMUM QUARTERLY AIR lEAI) CONCENTRATIONS
( jg/"»3) a
Percentage of sues in concent-ation Nunber 3f
r^nnoc f rnnranf r AMnne in nn/m^ 5i£g.
Site type/time'f-rame
<7.-?
0.5-1.0
1.0-1.5
i.s-2.a
>2.J
Mean
years
1980 through 1986
All monitors0
77 .S
15.4
3.2
1.5
2.3
0.39
3,279
NAMSC
76.9
17.2
4.7
1.1
0.2
0.41
471
SLAMS
73.4
15.7
2.8
1.7
3.4
0.38
1,138
Mlcroscale roadside3
24.1
j 7.9
25.9
10.3
1.7
J. 74
3d
Middle scale
45.7
47.1
7.1
0.0
0.0
0.4b
7o
Neighborhood scale
6d.7
24.3
7.0
0.0
U.O
0.37
115
198b only
AU monitors
93.0
2.5
1.2
1.6
1.8
0.17
314
NAMS
97.3
1.3
1.3
0.0
0.0
J.14
79
SLAMS
92.9
2.2
0.9
1.3
Z.I
0.21
226
Microscale roadside
100.0
0.0
0.0
0.0
0.0
J.18
9
Middle scale
100.0
0.0
0.0
0.0
0.0
0.12
9
Neighborhood scale
94.4
0.0
0.0
5.6
0.0
0.14
18
1987 only
All monitors
93.0
2.8
1.7
0.4
2.0
0.20
458
NAMS
97.4
0.0
1.3
0.0
1.3
0.20
78
SLAMS
93.1
2.3
1.7
0.0
2.9
0.20
175
*0ata are from the SAROAO and AIRS systens and '•epreseic the number of site-years for
which the maximum quarterly concentration fall witfnn the designated ranges. To oe
included, a valid site-year must have at least three quarters of rlata with at least 5
observations per quarter.
"Includes sites previously classified into categories (e.g., "urban") that do not
¦neet any of the current definitions.
CNAMS refers to the national Ambient Monitoring Station necwork, while SLAMS refers
to the State and Local Ambient Monitoring Station network. Each of these two networks
includes Both roadside and point source inonitors; noweve-, ?n the case of lead, the NANS
network tends to focus on roadside sites, while the SLAMS network incorporates ;nore
point sources and other special purpose monitors.
^Mlcroscale sites were within 5-ls meters from a major roadway. Middle scale sites
had- further setbacks and define concentrations up to several city blocks. Neighborhood
scale sites define concentrations in inore extended areas of umforn land use within
cities (0.5-4 ki»2). After 1986, data collection at these sites stopped.
Source: Battye (1988).
0.9 -
0.8 -
0.7 -
0.6 -
0.5 "
0.4 -
0.3 "
0.2 -
0.1 -
79
~ ALL
0 NAMS
A SLAMS
81
v 83
Year
i
85
87
FIGURE 2-2. TRENDS IN MAXIMUM QUARTERLY AIR LEAD CONCENTRATIONS
FOR VARIOUS MONITOR SITES, BAITYE (1988).
-------
11—7
naturally occurring outcrops can reach several hundred ppm. Urban soils
are contaminated by lead from a combination of automotive and point source
deposition as well as from lead in paint on indoor and outdoor surfaces.
Elevated soil lead concentrations (as high as 2000 ppm) have been found
within 10 feet of wood frame houses painted with lead-based paint (Ter Haar
and Aronow, 1974). Accumulations of lead in soils (and dusts) around brick
or stone structures have also been found, and can be partially attributed
to washoff lead collected on roofs, ledges, and exterior walls (Wheeler and
Rolfe, 1979), and possibly to an aerodynamic effect of the structures on
deposition of lead from nearby sources. Recent data from Baltimore (Mielke
et al., 1983) and in and around Chicago and "downstate" Illinois (LaBelle
et al., 1987) indicate that lead accumulates in soil away from painted
structures and known industrial lead sources, in a pattern at least partially
related to traffic activity and distance from nearby roads.
The upper layer* of roadside soils may contain deposited atmospheric
lead in concentrations of 10 to 500 ppm in excess of natural levels within
10 meters of the roadbed, beyond which concentrations decline abruptly,
depending on traffic density and volume and vehicle speed (Page and Ganje,
1970; Motto et al., 1970; Quarles et al, 1974; Wheeler and Rolfe, 1979;
Pierson and Brachaczek, 1976; Mielke et al., 1983; LaBelle et al., 1987).
Soil lead concentrations in a sample of city parks have also been reported
to be high, ranging from 20U to 3300 ppm (Chow et al., 1975; Zimdahl and
Hassett, 1977).
~Because of the vertical gradient of lead in soil (i.e., lead concentration
decreases with soil depth), the techniques used in sampling soils
influence the lead concentrations found. In this report, soil lead data
refer to the upper 1-5 centimeter layer, which is the most relevant for
potential intake by children.
-------
II-8
Topsoil lead concentrations around various lead point sources generally
decrease exponentially within a 3-10 km radius from 100-60,000 ppm down to
background levels (Yankel et al., 1977; Schmitt et al., 1979; Roels et al.,
1978; Diemel et al., 1981; Fugas, 1977). However, elevated soil lead
levels have even been found at distances exceeding 20-25 km from some
smelters (Wixson, 1978). Limited data indicate that lead in soil near
primary and secondary lead smelters occurs as lead oxides, sulfide, sulfate,
and elemental lead (Olson and Skogerboe, 1975; Corrin and Natusch, 1977).
Much of the lead in the atmosphere is deposited on terrestrial surfaces
where it is retained in organic complexes or adsorbed to hydrous iron
oxides near the soil surface (CO, p. 7-31). The ability of soil to immobilize
lead largely depends on soil pH and organic content (i.e., fulvic and humic
substances). Many U.S. soils' appear to have a large capacity to bind lead
with only a small fraction dissolved in soil moisture where it is available
for plant uptake. The environmental impact of lead on the biota of terrestrial
and aquatic ecosystems is discussed in Chapter 6 of the CD.
Improvements in air quality do not necessarily involve reductions in
soil lead at the same rate. The decontamination of the environment is a
very complex and slow process, particularly for soil lead which has been
shown to remain at elevated levels several years after lead manufacturing
plants have ceased operating (Prpic-Majic et al., 1984). While lead in
soil is relatively immobile, it still may enter the food chain, be ingested
directly, be mobilized by mechanical forces such as gardening, or may be
redispersed into the air and resuspended into water by wind scattering of
soil particles and dust.. Consequently, populations living in areas with
historical accumulations of lead in soil may continue to be exposed to lead
for indefinite periods despite effective controls of atmospheric emissions.
-------
11—9
C. Lead in Dust
Dust is a normal component of the home (i.e., "house dust") as well as
the outdoor environment where it can be found on sidewalks, playgrounds,
driveways, and other hard surfaces. Anthropogenic materials deposited on
these outdoor surfaces will be referred to as "street dusts". In addition,
the very top layer of soils (including leaf litter) to which people, particularly
children, come in direct contact, are considered in the criteria document
to be "soil dusts".
Both house dust and street/soil dust contain lead from atmospheric
deposition, "natural" soil, and paint chips. As with roadside soil, siynificant
correlations between lead concentrations in street dust and the proximity
and density of traffic have been found (Rolfe et al., 1977; Lau and Wony,
1982). Concentrations of lead in dusts on playgrounds, streets, and soil
surfaces also increase with higher air lead concentrations and with proximity
to stationary lead sources (Roels et al., 1980; Brunekreef et al., 1981;
Yankel et al., 1977). As summarized by Nriagu (1978), street dusts from
different U.S. cities contained between 300 and 18,000 ppm lead. A survey
of street dusts in 77 midwestern cities showed an average lead content of
1,636 ppm in residential neighborhoods and 2,413 ppm in commercial and
industrial areas (NAS, 1980). Limited chemical analyses of roadside soils
and dusts from U.S. cities found lead predominately as sulfate, along with
minor amounts of oxide and halide salts (Olson and Skogerboe, 1975; Corrin
and Natusch, 1977). Pavement dust, street dust, gutter debris, and household
dust samples contain mostly large lead particles ranging between 40 and
1000 ym (Pierson and Brachaczek, 1976; Sturges and Harrison, 198b; Duggan
and Inskip, 1985). Higher lead concentrations are usually found in the
smaller sized fractions. The concentration in a sample of street dust
-------
11—10
sieved at 100 um, for example, is usually at least two to four times greater
than that in the same sample sieved at 1,000 urn (Duggan and Inskip, 1985)
and 76% of the composite weight of house dust was measured in particles
smaller than 149 pm (Que Hee et al., 1985). This is significant given
that the bioavailability of soil lead tends to increase with decreasing
particle size (Barltrop and Meek, 1979). In addition, preliminary work
indicates that smaller particles (<10U-250 um) readily adhere to fingers
and are thus likely to be ingested (Duggan and Inskip 198b; Que Hee et al.,
1985).
Unlike lead that is incorporated into soil, lead in surface dusts,
both indoors and outdoors, is mobile and can be expected to respond over
time to a much greater degree to changes in atmospheric lead emissions.
Airborne lead deposited on streets, sidewalks and driveways is subject
to distribution by wind and water. Windblown particles associated with
dust are apt to be -redeposited within the urban environment by the complex
wind currents caused by buildings and street canyons. Following the
installation or supplementation of emission controls at lead point sources,
and significant improvements in air lead quality, significant drops in
concentrations of lead in playground dust, surface dust on soil, and in
house dust have been observed within 1 to 2 years (Morse et al., 1979;
Prpic-Majic et al., 1984). Another study found a weak correlation between
dust lead samples collected two months apart in the same houses in Wales
(Uavies et al., 1985). The authors noted that this is consistent with the
suggestion that external influences, e.g., soil, are important for dust
lead and differences in weather or season could influence the amount of
soil brought into the house and thereby cause short-term changes in dust
composition. This illustrates again the potential responsiveness of house
dust lead levels to changes in air quality.
-------
11-11
There is some evidence to suggest that this same relationship applies
in non-point source areas and that lead in surface dusts should decline
as a result of the decline in gasoline lead emissions. A survey of
rainwashed areas in the U.K. and New Zealand found that the normal
acidity of rain (pH between 4 and 5) was insufficient to completely dissolve
and transport lead particles, and that residue near streets poses a health
hazard to children who are prone to ingest street dusts (Day et al., 1979).
However, Laxen and Harrison (1977) found that a light rainfall (2 to 3
mm) is sufficient to remove 90% of the lead from the road surface, mainly
to surrounding soil and to waterways. Furthermore, based on deposition
fluxes around a major roadway over a year and concentrations of lead in the
road drainage water, Harrison et al. (1985) estimated that removal of lead
in runoff waters exceeds that deposited on the roadsides. The rate of
removal of lead from street dusts is also dependent on the frequency and
efficiency of street cleaning operations. It appears reasonable to expect
that with declining input, lead concentrations in street and household *
dusts will also drop due to precipitation, wind, street and house cleaning.
Surveys of a diverse set of homes indicate a wide range of house dust lead
levels between 10 and 35,000 ppm (EPA, 1977; Harrison, 1979; Angle and
Mclntire, 1979; Thornton et al., 1985; Clark et al., 1985) and as high as
100,000 ppm within 2 km of a smelter (Landrigan et al., 1975). Lead levels
in house dust can vary considerably depending on house cleaning practices,
as demonstrated by Charney et al., (1983), as well as on the presence and
condition of lead-based painted surfaces, the presence of cigarette smoke,
the amount of dust and soil blown into or carried into the house on clothing
and shoes (especially on those occupationally exposed to lead) and pets,
indoor sources of lead other tharr paint (e.g., soldering), the permeability
-------
11—12
of the home to outdoor air (which can vary with season), and outdoor con-
centrations of air lead. Lead concentrations in house dust are also dependent
on other constituents in the dust and it is therefore important to consider
the total amount of leaded dust as well as dust lead concentration.
To assess the impact of atmospheric lead on children's total exposure,
it is necessary to estimate the contributions of different air lead levels
to outdoor and indoor soil/dust lead levels as well as the amount of dirt a
child may ingest, both inadvertantly and deliberately. These issues are
addressed in the modeling approaches presented in Sections IV and V.
D. Lead in the Diet
The ingestion of food is a major component of most individuals'
total lead uptake, although the relative contribution is a function of their
age, and the size and type of diet. The occurrence of lead in the diet may be
a result pf a) natural sources of lead; b) deposition of airborne lead particles
onto crops, forage, feed, soils, and water; and c) the harvesting, processing,
transportation, packaging, preparation, and storage of food during w"hich lead
can be introduced at every stage either by atmospheric deposition or through
metallic contamination, particularly from plumbing or solder in cans.
Recent studies on lead content of various foods, both before and after
processing, packaging, and preparation, and on food consumption patterns
in the U.S., provide information on dietary lead intakes for different
populations (Beloian and McDowell, 1981; Wolnik et al., 1983; National Food
Processors Associations, 1982; Pennington, 1983; U.S. FDA, 1983, 1984).
Using these data for 1982-83, a "Multiple Source Food Model" was developed
in the CD that apportions lead in "typical" child and adult diets to the
following sources: natural soil lead, direct atmospheric lead added to
food crops before harvest and during processing, indirect atmospheric lead
-------
11-13
that has been incorporated in soils, lead in solder from food cans and
drinking water, and lead whose origin cannot be determined at present
(CD, Section 7.3.1.2). The model was validated by using the same parameters
along with more recent food data from 1984 and 198b to successfully predict
continuing declines in dietary lead intake. For example, dietary lead intake
for a 2-year old child has dropped from about 45-50 pg/day in 1978 to an
estimated 13.1 pg/day in 1985; comparable declines (from 45-50 pg/day to
15-20 pg/day) have been seen in different adult populations since 1982
(Flegel et al., 1988.) Even greater reductions had been achieved in the early
to mid-1970's (CD, p. 7-49). These trends are attributable to the drastic
reduction in gasoline lead emissions and the voluntary phaseout of lead-soldered
cans by U.S. manufacturers since the 1970's. This downward shift is expected
to continue up until the early 1990's as lead from both these sources drops
further. The methods by which estimates of dietary lead intake from the
Multiple Source Food Model are derived and applied to modeling total lead
exposure are discussed in Appendix A and Section V.
E. Lead in Water
Lead is a natural, usually very minor, constituent of surface and ground
waters. Atmospheric lead can enter aquatic systems through direct fallout or
in surface runoff as suspended particles or adsorbed to soil particles. Under
most conditions (pH, temperature, alkalinity), lead forms insoluble salts and
precipitates to sediments, which probably accounts for the low lead content of
U.S. surface water supplies; the average concentration ranging between 3 and 4
micrograms Pb per liter water (pg/L) (NAS, 1980).
In contrast, lead levels in household, school, and office building
drinking water can be much higher due to plumbing corrosion and subsequent
leaching of lead, ranging between 10 and 30 pg/1 on average. The combination
of corrosive (i.e., soft or acidic) water and lead pipes or lead soldered
-------
11-14
joints in distribution systems or houses creates localized zones of high lead
concentrations up to 380 ug/1 (Worth et al., 1981). The combination of new solder
and corrosive water can result in concentrations commonly over 1000 mg/L. In
general, levels are highest in samples of hot and/or stagnant "first draw" water.
Drinking water is a major source of lead exposure for many infants while they
are dependent on baby formulas during their first year, as well as for young
children. The 1986 Safe Drinking Water Amendments banned the use of lead solder or
flux and lead-bearing pipes and fittings, to be implemented and enforced by the
States by June 1988. In August 1988, EPA's Office of Drinking Water proposed revis'
the existing lead standard of 50 yg/L to 5 yg/L, measured at the entry point to the
distribution system or the treatment plant, with a "no-action level" average of 10
Mg/L» measured at the home tap.
Estimated contributions that natural, atmospheric, and solder lead make to
total lead exposure via drinking water and total diet are addressed in the modeling
approaches presented in Sections IV and V.
F. Lead in Paint
Ingestion of lead-containing paint is considered to be the most frequent
cause of severe lead intoxication among children (Chisolm, 1984; CDC, 1985).
Significant correlations have been found between the quantity and condition of lead
painted surfaces in homes and blood and fecal lead levels in inner city children
(Urban, 1976; CD, pp. 11-156 to 11-160). Significant differences in blood lead
levels have been found in relation to housing condition among children as young as
nine months, with highest levels in children living in deteriorating pre-World
War II housing, intermediate levels in well-maintained and rehabilitated older
housing, and lowest levels in children in public housing and newer units free of
lead paint (Clark et al., 1985). Other investigations indicate that in addition
to peeling lead-based paints, "intact" lead-based paint contributes to elevated
blood lead levels in children via chalking or weathering (Gilbert et a>.,
-------
11-15
1979; Galke et al., 1975) or can be exposed to children who chew or gnaw upon
the surfaces. Further risks are associated with conventional paint removal
techniques (i.e., sanding, burning heat guns) which can generate concentrations
of fine lead dust up to 100-100,000 yg/iP (NIBS, 1988; Inskip and Atterbury,
1983).
Under the 1971 Lead-Based Paint Poisoniny Prevention Act, which has
since been amended, the Department of Housing and Urban Development has .
regulatory and research responsibilities for eliminating the hazards of
lead-based paint poisoning in federally funded housing. In 1977, the
Consumer Product Safety Commission banned household paint (including toy
and furniture paint) containing more than 0.06% lead as hazardous. This
was an important step but it does not affect hazards posed by lead-based
paints already present on old housing surfaces. Prior to 1940, some interior
paints contained more than 50% (500,000 ppin) lead. Use of lead pigment
paints declined slowly between 1940 and the late 1960's after which it
declined at a rapid rate (NAS, 1972). Pope (1986a) estimates that in 1980,
the interior and/or exterior surfaces of between 21.5 and 47.3 million
households in the U.S., mostly built before 1960, contained greater than
0.7 mg/cm^ lead in painted surfaces, a level considered hazardous to young
children by the Centers for Disease Control (CDC). Between 6.2 and 13.6
million children under the age of 7 are estimated to have resided in lead-based
painted housing in 1980 (Pope, 1986a). Even if coated with low-level leaded
paint, the underlying layers of lead-based paint in older homes represent a
large reservoir of lead exposure in children, particularly if there is
peeling paint, broken or cracked plaster, or holes in the walls. The
number of lead-based painted homes with these deteriorating conditions
(i.e., "unsound") is estimated to have been between 800,000 to 2.9 million
-------
11-16
in 1980 with approximately 235,000 to 842,000 children under age 7 living
in such homes (Pope, 1986a).
Lead-based paint continues to be the major source of high-dose lead
exposure and symptomatic lead poisoning for children in the U.S. (Chisolm,
1971; CDC, 1985), and it appears that exposure to lead-based paint will
continue to be a problem for decades to come (Schneider and Lavenhar, 1986).
Between 1973 and 1980, only 10% of pre-1940 housing units and.5% of
housing units built .in the 1940's had been removed from our nation's housing
stock by demolition, disaster, or by other means such as conversion to
commercial space (Bureau of the Census, 1983). Less than half of these
removed units were located in central city areas where lead poisoning is
most prevalent. In addition, poor urban families often have no choice but
to live in poorly maintained older housing, given that the vast majority of
new housing units created from 1973 to 1980 is located outside of central
city areas and that acute shortages of modern and lead-free, low income
rental units exist in many cities (Farfel,.1985). Many rural families
are also afflicted with a lead paint risk since about 41% of rural housing
is "substandard" (ATSDR, p. V1-15).
High exposures to children living in older housing with flaking or
intact lead-based paint will not be significantly influenced by any changes
in atmospheric lead emissions. Children with pica for paint chips or
others living in deteriorating lead-based painted homes who are excessively
exposed to paint lead-contaminated dust through normal hand to mouth activity
cannot be expected to be protected by any lead NAAQS no matter how strinyent.
For this reason and because exposure to paint lead is so highly variable,
no attempt will be made here to quantitatively estimate exposures of children
who are excessively exposed to lead-based paint under various air lead
-------
11-17
levels, as is done for other exposure media. Some of the available data,
although sparse, are discussed at the end of Appendix A to demonstrate how
some estimates could be generated. Despite the small amount of data, and the
great variability in paint lead exposures, it is clear that any exposures
and blood lead levels predicted for children under various air lead levels
using the approaches in Section III will be significantly higher for children
with high paint lead exposure. Comprehensive plans for safe and effective
removal or containment of lead painted surfaces from old housing need to be
developed and implemented, otherwise pediatric plumbism will persist until
much of the present housing stock is completely rehabilitated or demolished
(Chisolin, 1986). Preventing excessive exposure to existing sources of lead
in and around housing must be addressed by an appropriate combination of
legislation, housing code inspection and enforcement, financial incentives,
parental education, and other programs.
-------
III. ESTIMATING LEAD EXPOSURES AND BLOOD LEAD LEVELS
A. Alternative Modeling Approaches
To assess the health risks associated with alternative air lead
standards, it is necessary to estimate the blood lead (PbB) levels that
would be distributed in the population(s) of concern under various air lead
concentrations. [The amount of lead measured in whole blood is an index of
the rapidly diffusible fraction of the total body burden of absorbed lead
and is generally used as the dosage, or index of exposure, in investigating
the various human health effects associated with lead.] Lead exposure occurs
through multiple media but unfortunately, longitudinal assessments of
simultaneous multi-media exposures have not yet been made. Potentially
useful longitudinal analyses are now being carried out as part of the
Cincinnati Lead Program Project (Clark et al., 1987). Three approaches
are examined here that can be used to estimate or predict the impact of inhaled
and ingested lead aerosols and compounds on the body burden of lead as indexed
by blood lead. These approaches are presented in detail in Sections IV and
V for consideration for using one, two, or all three in assessing the
protection provided by alternative lead NAAQS.
The first approach, discussed in Section IV, is to use measured rates
of absorption, or "uptake" of lead through different pathways (e.g., inhalation
and ingestion) from experimental studies together with available mathematical
("biokinetic") models from lead balance studies to project either total
body burden or the amount of lead in any of the presumed "physiological"
kinetic compartments (e.g., blood, soft tissue, bone) at any time.
The second approach, referred to as an "aggregate" model, is to
directly apply a mathematical relationship between air lead and blood lead
derived from community epidemiological studies that reflects both direct
-------
II1-2
inhalation exposure and indirect exposures via secondary deposition pro-
cesses. The third approach, referred to as a "disaggregate" model, is a
hybrid of the first two in which separate empirical relationships between
blood lead and lead intake from lead in air, food, water, dust, and soil,
available from experimental exposure and epidemiological studies of different
populations, are applied to lead concentration estimates in different
media. The nature of the relationships detected between lead concentrations
in blood and various environmental media, and the merits and limitations in
using empirical relationships to represent lead exposure through multiple
pathways in these latter two approaches, will be discussed in Section V.
8. Calculating Blood Lead Distributions Around Predicted Means
Because most of the data input to these models are generally average
estimates, all three models estimate PbB levels in terms of population mean
levels. The distribution of PbB levels in U.S. children is broad because '
there is a distribution of exposures those children face, a distribution of
behavior patterns that affect the uptake of that exposure, and a distribution
of biological absorption and excretion rates. For purposes of setting a
lead NAAQS, it is necessary to determine the blood lead distribution
across a defined population group corresponding to a given mean PbB for
that group so that, for example, percentages of individuals exceeding PbB
levels of concern can be ascertained. Determining such distributions is an
attempt to account for the significant behavioral and biological variability
within populations. It is the group of individuals with the potentially
greatest adverse response to any given lead exposure that are of greatest
concern in establishing a primary lead NAAQS.
Consistent with measurements of other metals in tissues of human
populations, the distribution of PbB levels for any relatively homogeneous
-------
111—3
population closely follows a lognormal distribution (CD, Section 11.3.4).
A lognormal distribution is completely specified by its geometric mean and
geometric standard deviation. It is possible, therefore, to calculate
percentiles of a blood lead distribution (e.g., median or 99 percentile)
around a mean PbB level by using the geometric standard deviation (GSD) of
that population's blood lead distribution. Several epidemiological studies
have indicated that the log values of measured individual PbB levels in a
uniformly exposed population are normally distributed with a variation,
including analytical variation, ranging between 1.3 and 1.4, when expressed
as a GSD (Tepper and Levin, 1975; Azar et al., 1975; Billick et al., 1979).
The NHANES II study provides the best available data on nationwide blood
lead levels in terms of quality control and sample size. In regression
analyses of these data, Schwartz (1985a) estimated a GSD of 1.428 for young
children aged 6 months to 5 years, after removing the variance in PbB
levels attributable to air lead exposure by adjusting every individual's
PbB level to what it would be at zero air lead, while allowing for variations
in background non-air lead exposure.
After excluding from the analysis children with PbB levels above
4U yg/dl, who would not be expected to be substantially affected by changes
in the lead NAAQS (for example, because of higher than average pica activity
or excessively high exposures to lead in paint), Schwartz (1985b) reapplied
the same technique and calculated a GSD of 1.419. Thus, a GSD without
attribution of any source of lead exposure except gasoline lead and industrial
air lead emissions may be taken as a rounded-off value of approximately
1.42 for the NHANES II population of children (CD, p. 11-31).
Given that the focus of this exposure analysis is on populations living
near lead point sources where blood lead variability may be different compared
-------
II1-4
to the total mix of U.S. children, a separate assessment of blood lead
GSD's for such populations is presented here. Published statistics of PbB
samples near various lead point sources have been assembled and analyzed
in Cohen (1986) and Marcus (1988a). The GSDs calculated from these point-
source/blood lead surveys are listed in Table 3-1, alony with the NHANES II
GSO for children. To different degrees, the primary origin of non-biological
variation in lead exposure for these point source populations was the smelter,
especially in the Roels and Yankel studies conducted when smelter emissions
were considerably higher. It is possible that children with exposure
to a single, intense source of lead may have relatively low variability
(essentially only due to intrinsic biologic variability) with a high
mean blood lead. The most recent study indicates a larger variance at
relatively lower exposure levels near the Montana smelter surveyed in 1983,
suggesting that multiple origins with separate variability may be present
rather than a single predominant source of variability in lead exposure
(Marcus, 1988a).
The selection of a GSD value to model populations living near various
lead point sources in the future depends on assumptions regarding variance in
exposure in the future. Perhaps the most relevant data are from the 1983
E. Helena, Montana study since they reflect the most contemporary conditions.
It can be expected that even for this population, however, total lead exposure
will have dropped substantially by 1990, the starting point of our exposure
analyses. This is attributable to recent and continuing downward trends in
lead levels in canned foods, the nationwide ban on lead solder in new construction
and plumbing repairs, the lead-in-gasoline phasedown, and perhaps increasing
public awareness regarding the dangers of lead exposure and effective avoidance
measures. All of these changes should result in not only lower mean baseline
-------
111—5
TABLE 3-1. COMPARISON OF ESTIMATED GSDs ACROSS SEVERAL STUDIES
Population/Reference
Mean PbB
(ug/dl)
Estimated
GSD1
NHANES II - total mix of
children 1-5 years old
(Schwartz, 1986b)
11-year olds living near Belgian
primary lead smelter (Roels
et al., 1980)
1-9 year olds living near Idaho
primary lead smelter (Yankel et al.,
1977)
1-5 year olds, living near Montana
primary lead smelter (CDC, 1983)
1-5 year olds living in 3 Missouri
smelter towns (Baker et al., 1977)
16.0
21.7
56.5
1-5 year olds, closest to 3 non-ferrous 15.6
smelters in U.S. (Hartwell et al., (median)
1983)
9.4
16.2
1.42
1.29
1.32
1.39
1.53
1.57
1-GSD calculations described in Cohen (1986) and in Marcus (1988a).
-------
111-6
PbB levels, but perhaps more importantly, in fewer high level exposure
situations and therefore less variance in PbB levels than existed in 1983.
Quantifying such an effect to estimate a future GSD value is difficult. It
is likely, however, that future reductions in baseline "non-point source"
lead exposures will result in a reduction in exposure variance so that the
blood lead GSD will be more reflective of intrinsic biologic variance, such
as that found in the Roels and Yankel studies. Althouyh these latter
studies were conducted during times of much higher exposures, the exposures
were dominated by high smelter emissions and soil/dust contamination,
resulting in relatively low PbB variability. For purposes of modeling
future exposures among children living near different lead point sources, a
range of GSD values can be considered: a lower bound of approximately 1.30
derived from the Yankel et al. and Roels et al. studies and an upper bound
GSD value of 1.53 derived from the 1983 CDC study of E. Helena, Montana
children. The midpoint of this range is 1.42, which coincidentally is
identical to the NHANES II GSD estimate. Until additional data are available
a range of 1.30-1.53 will therefore be assumed for children living near
lead point sources as a reasonable ranye of GSD values, and the midpoint of
1.42 will be assumed as a reasonable best estimate. In the case study exposu
analyses, the larger and more conservative GSD of l.b3 will be tested in
sensitivity analysis.
Since blood lead estimates will also be made for middle-aged men, GSD
values for adult blood lead distributions are needed. There is little
information on adult blood lead GSDs from point source studies. The
most complete data base is clearly from NHANES II, which has been
analyzed in the CD to yield a range of GSD values, depending on race and
place of residence, between 1.34 and 1.39 for adult women and between
-------
111—7
1.37 and 1.40 for adult men (CD, Table 11-9). Given the uncertainties in
extrapolating these ranges to adult populations living near point sources
under future exposure scenarios, a midpoint value of 1.37 will be chosen as
a reasonable estimate for both adult men and women.
Analytical variation which exists in any measurement has an impact
on the bias and precision of statistical estimates such as the GSDs derived
from the studies discussed here. For this reason, it is important to
recognize the magnitude of analytical variation in blood lead measurements
due to measurement variation (i.e., between measurements run at the same
time) and variation created by analyzing blood samples at different times
(CD, p. 11-29). Using GSDs adjusted in such away, although providing a
more accurate characterization of a given blood lead distribution, may not
be appropriate for this assessment because the health effects studies that
will be used to define dose-response relationships did not correct their
PbB measurements for analytical variance. These studies, like NHANES II
and the point source epidemiological surveys, generally employed the best
available measurement techniques and quality control. Using "corrected"
GSDs to predict the distribution of children's PbB levels around a mean
population PbB, and then matching those predicted PbB levels with "uncorrected"
PbB levels%derived from health studies in order to assess the risks associated
with that population mean PbB, would result in a somewhat biased assessment.
Therefore, the range of GSDs presented above that are not corrected for
analytical variance will be used in further calculations.
Table 3-2 presents blood lead distributions associated with sample
population mean PbB levels calculated using the GSD values derived above.
-------
III-8
These distributions were calculated using the formula given to describe
a lognormal distribution (Yankel et al., 1977): PbB = GM(GSD)Z
where:
GM = geometric mean PbB
GSD = geometric standard deviation
z = the number of standard deviations
PbB = value of PbB at z standard deviations
TABLE 3-2. ILLUSTRATIVE BLOOD LEAD DISTRIBUTIONS ASSUMING
DIFFERENT MEAN BLOOD LEADS AND GEOMETRIC STANDARD DEVIATIONS
Geometric Percentile
GSD
Mean
PbB
90th
95th
99th
99.5th
1.37
4
6.0
6.7
8.3
9.0
5
7.5
8.4
1U.4
11.3
6
9.U
1U.1
12.5
13.5
7
10.5
11.7
14.6
15.8
J. 42
4
6.3
7.1
9.0
. 9.9
5
7.8
8.9
11.3
12.3
6
9.4
10.7
13.6
14.8
7
11.0
12.5
15.8
17.3
-------
IV. INTEGRATED LEAD UPTAKE/BIOKINETIC MODEL
The "uptake/biokinetic" modeling approach attempts to account for the
following: a) the amount of lead in the body at any one time is the product
of dynamic interactions of partially offsetting processes of absorption,
distribution, storage, mobilization, and excretion; b) these processes vary
with the route and rate of exposure, a person's age, nutritional and health
status, and baseline exposure; c) uptake from all sources by all absorption
routes can be separately modeled, thus providing an estimate of the relative
importance of atmospheric lead exposure, either directly or indirectly, to
total body burden; and d) past and future trends in environmental lead
levels due to different control efforts and regulations are important
determinants of projected exposures.
There have been several studies measuring the intake, uptake, and
metabolism of lead in adult male volunteers, from which balance schemes have
been constructed (Kehoe, 1961; Chamberlain et al., 1978, Rabinowitz et al.,
1976, 1977). By making estimates about lead intakes and various metabolic
factors, balance schemes can be constructed for different groups with
different exposures to lead to assess the importance of specific exposure
factors under variable conditions (e.g., alternative lead NAAQS, phaseout
of lead in soldered cans and in gasoline). Pre-school age children (<_ 6
years old), pregnant women (as exposure surrogates for the fetus), and
adult men (40-59 years of aye) are specified in the CD as particularly
sensitive to lead. Of these groups, children between 2 and 3 years old
exhibit, in general, the highest blood lead levels (Mahaffey et al.,
1982; Billick et al., 1979), most likely due to their greater hand-to-mouth
activity as well as to various metabolic processes (Harley and Kneip,
1985). To predict exposure impacts of alternative air lead standards,
model balance schemes have been devised for children aged 0-6 years and
are discussed below.
-------
IV-2
A lack of adequate biokinetic data for pregnant females prevents such
modeling for the fetus. In addition, accurate biokinetic modeling of the
sensitive population of adult males would require at least 40 years of
cumulative exposure inputs. Such uptake modeling would be difficult given
the great variability in exposure histories among U.S. adults and the paucity
of data. Exposure modeling for purposes of risk estimation for adults,
discussed in Section V.A., will rely on combining air: blood lead "disaggregated"
relationships with estimated baseline "non-air" blood lead levels.
A discussion of lead's absorption, excretion, retention, and distribution
within the body under different exposure and physiological conditions is
necessary background to the uptake/biokinetic model, and is provided in
Chapter 10 of the CD. As-will be further discussed, application of this
or any lead exposure model must recognize the limitations of the available
data and the significant variability among populations in their behavioral,
exposure," and physiological characteristics. For present purposes, the
children to be considered in the model do not include the whole U.S. population;
they comprise groups with contrasting exposures to lead, some of them extreme,
in order to illustrate the variations in lead exposure and absorption
around lead point sources under different scenarios. Section IV.A presents
methodologies to estimate average daily lead uptake-from all exposure
sources under alternative air lead levels. Section IV.B presents methodologies
to relate these estimates of lead uptake to average, steady-state blood
lead levels. Although this approach predicts hypothetical outcomes and
the absolute numbers should not be used uncritically, the model does strive
to use the available data, with all its limitations, to the fullest extent
possible to provide a useful tool in eventually distinguishing the health
impacts of alternative lead NAAQS.
-------
IV-3
A. Estimates of Lead Uptake
The method employed to estimate the degree to which each environmental
source of lead contributes to a child's total daily lead uptake is based
on the distribution of ambient air lead levels that would be allowable or
expected under each standard considered. Probable exposure conditions with
respect to other exposure media such as food and dust, and average biological
absorption rates for each exposure route are also specified. The method
consists of a four-step process:
1) definition of ambient concentrations of lead for major exposure
sources (i.e., air, food, water, soil, dust);
2) determination of daily lead intake according to the relationship:
I, = Ci • [Pb]i
where I\ is the daily lead intake from source, i, C-j is the ingestion or
inhalation (i.e., consumption) per day of each lead source i and [Pb]-j is
the concentration of lead in each source i;
3) calculation of the amount of lead absorbed from each exposure
source i:
ui * li * Ai
where U-j is lead uptake for each exposure source i, I -j is the daily lead
intake from each source i, and A-j is the percent absorption of lead, via
the appropriate exposure route for the particular source; and
4) calculation of the total lead uptake from all sources, U^;
% • Hi ¦ A,)
The way daily lead uptake would be calculated is illustrated in Table 4-1
for 2-year old children living near (within 2-5 km) a lead point source.
Details of the model parameters and the line-by-line estimates, along with
-------
TABLE 4-1. ESTIMATED 1990 AVERAGE LEAD INTAKE AND UPTAKE IN 2-YEAR OLD CHILDREN
UNDER AVERAGE AIR LEAD LEVELS1
0.25 pg/m3 0.5 Ug/m3 0.75 Mg/ro3 1.0 Ma/fli"* My/m3 1.5 M9/ffl3
1.
Outdoor air lead (pg/m3)
0.25
0.5
0.75
1.0
1.25
1.5
2.
Indoor air lead (yg/m3)
.08
0.15
0.23
0.3
0.38
0.45
3.
Time spent outdoors (hours/day)
2-4
2-4
2-4
2-4
2-4
2-4
4.
Time weighted average (pg/m3)
.09-.11
.18-.21
.27-.32
.36-,42
.45-.54
.54-.63
5.
Volume of air respired (m3/day)
4-5
4-5
4-5
4-5
4-5
4-5
6.
Lead Intake from air (pg/day)
0.4-0.6
0.7-1.1
1.1-1.6
1.4-2.1
1.8-2.7
2.2-3.2
7.
% deposition/absorption in lungs
42
42
42
42
42
42
8.
Total lead uptake from lungs
0.2-0.3
0.3-0.5
0.5-0.7
0.6-0.9
0.8-1.1
0.9-1.3
(Mg/day)
9.
Dietary Lead Consumption (pg/day)
a) from solder or other metals
5.6
5.6
5.6
5.6
5.6
5.6
b) atmospheric lead
1.2
1.2
1.2
1.2
1.2
1.2
c) natural lead, Indirect atmos-
3.6
3.6
3.6
3.6
3.6
3.6
pheric, undetermined sources
10.
% absorption in gut
42-53
42-53
42-53
42-53
42-53
42-53
11.
Dietary lead uptake (pg/day)
4.4-5.5
4.4-5.5
4.4-5.5
4.4-5.5
4.4-5.5
4.4-5.5
12.
Street dust/soil lead (pg/g)
181
308
436
563
691
818
13.
Indoor dust lead (pg/g)
271
482
693
904
1115
1326
14.
Time weighted average (pg/g)
241-256
424-453
607-650
790-847
974-1044
1157-1241
15.
Amount of dirt ingested (g/day)
.08-.135 .
.08-.135
.08-.135
.08-.135
.08-.135
.08-.135
16.
Lead intake from dirt (pg/day)
19.3-34.6
33.9-61.2
48.6-87.8
63.2-114.3
77.9-140.9
92.5-167.6
17.
% dirt lead absorption in gut
25
25
25
25
25
25
18.
Lead uptake from dirt (pg/day)
4.8-8.6
8.5-15.3
12.1-21.9
15.8-28.6
19.5-35.2
23.1-41.9
19.
Total lead uptake from lung and
9.4-14.4
13.2-21.3
17.0-28.1
20.8-35.0
24.7-41.8
28.4-48.7
gut (pg/day)
^Assumptions and calculations are described In Appendix A by row number. Refers to children living near one or more lead
point sources and unaffected by lead paint. Children living 1n typical urban or rural environments remote from lead point
sources can be modeled using the same uptake calculat-ions and substituting parameter values where differences are noted
in Appendix A.
-------
IV-5
limitations of the available data base, are discussed in Appendix A. For
comparative purposes, children living in areas generally unaffected by
stationary lead emissions can also be modeled using most of the same parameter
values; exceptions (indoor/outdoor particled penetration rates, respiratory
deposition rates) are noted in Appendix A. Dust and soil lead levels
associated with given air lead concentrations would differ substantially
between point and non-point source areas; this too is briefly discussed
in Appendix A. The following points are important to consider with regard
to the uptake calculations in the table: 1) the calculations do not represent
population-wide exposure scenarios under alternative standards, only exposure for
the specified air lead concentrations. Ambient lead levels around a point
source meeting a given lead NAAQS vary depending on distance, topography,
local meteorology related to pollutant transport and diffusion, plant operating
parameters, emissions controls, particle sizes and deposition rates, etc.
Air quality dispersion modeling that accounts for these factors will be used
in the actual case study analyses to assign air lead concentrations under
alternative standards to different "receptor" points where people live around
the lead point sources; 2) the estimates in the table apply to 2-year old
children to illustrate the period of maximum exposure. Blood lead levels
estimated by the biokinetic model (discussed below) account for accumulating
exposure throughout childhood, not just a single year. It is necessary,
therefore, to make age-specific exposure estimates that reflect changes in
behavior and metabolism as a child ages. Age-specific exposure estimates are
also discussed in detail in Appendix A; and 3) the yeneral types of exposure
environments considered here do not include older homes with lead-based paint
hazards. Such conditions are discussed at the end of Appendix A to illustrate
the severity of lead-paint hazards compared to other exposures.
-------
IV—6
B. Uptake and Blood Lead Concentration
To estimate blood lead (PbB) levels under different exposure scenarios,
several different kinds of studies can be used to derive a relationship
between absorbed lead (or lead uptake) and blood lead. Available studies
include population surveys in which the blood lead concentration of
individuals or groups is correlated with measured lead concentrations in
air, food, water, soil, or dust; experiments in which volunteers are exposed
to controlled air lead concentrations and their PbB levels measured, and;
lead balance studies of individuals with measured lead intakes. The most
relevant experimental studies including, for comparative purposes, those on
adults, are discussed here along with descriptions of the analyses used
to derive uptake/PbB relationships. Relationship between blood lead and
environmental lead concentrations from population surveys are discussed as
part of the aggregate and disaggregate models in Section V.
1. Dietary Lead Ingestion Studies in Infants
Stu3ies that have measured dietary lead intake concurrently with
PbB levels among infants and toddlers are compared in the criteria document
(CO, Table 11-49). Although precise estimates of dietary lead consumption
can not be made, these studies (U.K. Central Directorate, 1982; Lacey
et al., 1985; Sherlock et al., 1982; Ryu et al., 1983) have several advantayes:
1) careful control and analysis of dietary intake; 2) minimal lead exposure
to sources other than the diet as the study population (infants) is relatively
immobile; and 3) intake/blood lead relationships can be used to estimate
the impact of total lead exposure (i.e., from all sources) on blood lead
because tracer studies (Chamberlain et al., 1978; Rabinowitz et al., 1976)
show no differences in the distribution of lead to tissues whether taken up
from lung or gut.
-------
IV-7
Both the U.K. Central Directorate and the Sherlock et al. studies
involved infants with relatively high PbB levels and high intakes. It is
well documented that the relationship between PbB and lead uptake from any
source is approximately linear at low intake levels but "flattens out" at
high intake levels (CO, Sections 11.2 and 11.3). Because the PbB levels
(< 20 yg/dl) and lead intakes of the Ryu infants are more relevant, the
criteria document concludes that the slope from this study (0.16 pg/dl per
My/day lead intake) is the best available estimate (CO, p. 11-129).
The infants in the Ryu et al. (1983) study absorbed substantial
quantities of lead from formula or whole milk in lead-solderd cans.
Even lead at much lower levels from formula in glass bottles achieved
an apparent blood lead equilibrium postnatally. Reanalyses of the Ryu
et al. data by Marcus (1989) suggested a larger slope, about 0.24 py/d1
per yg/day lead intake in formula and milk.
Using non-linear and piecewise linear models to examine PbB-dietary
lead slopes at higher intake levels, Marcus (1989) reanalyzed the Ryu et
al. data, as well as data on Scottish infants (Lacey et al., 1985) and
school children (Laxen et al., 1987) exposed across a wide range of water
lead levels. Because the diets among the infants in the Lacey et al. study
were predominantly liquid, the derived water lead/blood lead relationship
is considered a surrogate to predict the contribution of total diet to
blood lead levels. A piecewise linear model fitted the data about as well
as a cube-root or square-root model, with a PbB/water Pb slope from the
Lacey et al. study on infants of 0.254 yy/d1 per yy/L for water lead levels
below 16 pg/L, and a much lower slope of 0.042b above that inflection point;
after partial ling out the effects of house dust lead, the Laxen et al.
school-children had a PbB/water Pb slope of 0.161 pg/d1 per pg/L for water
-------
IV—8
lead levels below 15 |jg/L, and a 0.0318 slope above that level (Marcus,
1989).
If the slopes from Lacey et al. and Laxen et al. are converted to uptake
slopes by assuming 1 L/d water consumption, then the three absorption coefficients
(0.24 from Ryu et al., 0.254 from Lacey et al., 0.161 from Laxen et al.) are
reasonably consistent for low levels of dietary lead intake. Furthermore,
given the differences among the three study populations and their exposure
situations, the above results are strikingly consistent in terms of the
shape of the dose-response relationship, the location of the inflection
point, and the magnitude of the differences above and below that point.
The higher slope found for the Lacey et al. infants compared to the Laxen
et al. school children is expected given the reduction in lead gut absorption
rates with age. The uptake of lead from diet depends not only on the age
of the child, but also possibly on the chemical and physical form of the lead,
and on other components of diet that affect' lead absorption. Much lower
coefficients may be appropriate for adults and for older children who only
ingest lead with meals; even higher coefficients may be appropriate for
lead absorbed by very young children who consume lead between meals and who
have nutritional deficiencies that facilitate lead absorption.
The lower slope at high levels of water lead or dietary lead uptake
probably represents other factors that reduce absorption, e.g., much of the
"excessive" dietary lead intake may occur during meals and is thus much less
bioavailable. There is reasonable consistency among these high-uptake slopes:
0.0426 from Lacey et al. infants; 0.0318 from Laxen et al. school children;
and 0.032 from the Sherlock and Cools studies for adults (see below).
Uncertainty about the change-over between high-absorption and low-
absorption cases probably reflects incomplte information about the food and
water consumption patterns and other dietary factors. For present-day
-------
IV-9
U.S. dietary lead exposure, the steeper slopes are likely to be more
accurate.
The results from the Ryu et al. and Lacey et al. studies will be used to
represent the infant dose-response relationship: the midpoint of the slopes
(0.16 and 0.254, i.e., 0.207) estimated for low exposure levels (i.e., < 16
pg/L), and the Lacey et al. slope of 0.026 for higher exposure levels. Assuming
the same lead concentration in total diet where the slope flattens out, the
water lead "threshold" level from the Scottish studies of approximately 16 pg/L
(or 16 (jg/Kg) can be converted to a corresponding dietary intake level by
multiplying by the average dietary intake for an infant. The Cn cites a dietary
intake rate of 1.502 Kg/day for a 2-year old child (Table 7-15) and a rate of
1.0 Kg/day will be assumed for the infant. The resulting level of 16 (jg/day
will be used as the inflection pornt in applying the slopes from Ryu et al. and
Lacey et al. (0.207 and 0.026). Because lead ^take/PbB relationships are of
interest, these slopes will be combined with the midpoint of gastrointestinal
absorption rates (42-53%) cited for infants in the CO, to illustrate the
relationship between blood lead and daily lead uptake for infants at the
end of this section. A baseline (i.e., non-air) PbB of 4 ug/dl will be
assumed, based on recent studies of PbB levels and lead intake in Boston
and Cincinnati infants (Rabinowitz et al., 1986; Succop et al., 1987).
^ • Dietary _I_ntake _Studi_esj^_Adul_ts_
The relationship between blood lead and food and/or water lead levels
has been examined for adults in both experimental studies in nhich controlled
dietary supplements were administered to volunteers, and duplicate diet
studies. Again, several studies indicate non-linear blood lead responses at
high exposures but a linear relationship at relatively low intake levels
(< 100 pg/day, or < 30 (jg/dl PbB).
-------
IV-10
Two studies identified in Table 11-49 and 11-50 of the CD appear most
useful in estimating a dietary lead/blood lead relationship chiefly because
of their relatively large sample sizes and control over confounding factors:
Sherlock et al. (1982), a duplicate diet study of 31 mothers (and their children)
from Ayr, Scotland; and Cools et al. (1976) in which 11 subjects and 10 controls
were studied for blood lead response to oral dosages of lead acetate.
The latter study was an extension of the Stuik (1974) experiment, results
of. which will no.t be used here because responses were not followed long enough
for blood lead to equilibrate. The weighted mean of the slopes reported in the
CO for the Sherlock and Cools studies is 0.032 ug/dl increase in blood lead per
Mg/day intake, which is a factor of 5-6 times lower than the comparable slope
estimated for infants at low exposure levels. A gastrointestinal absorption
factor of 0.15, cited in the CD for adults, will be applied to this intake
slope to derive a blood lead/uptake relationship. A baseline PbB level of 4
Mg/dl will be assumed for adults based on projections described in Appendix C.
The same pattern of non-linear absorption seen in children will be assumed for
adults and a slope of .009 ug/dl per yg/day will be used for levels above
approximately 40 ^g/day, which corresponds to an inflection point of 15 ug/kg
lead in diet, multiplied by an approximate dietary intake of 2.89 kg/day for
middle-aged men (CD, Table 7-15).
3. Application of Chamberlain and Heard Analysis of Adult Exposure Data
An attempt to relate blood lead and absorbed lead was made by Chamberlain
and Heard (1981) using epidemiological and clinical data on adult men
(Williams et al., 1969; Kehoe, 1961; Nordman, 1975; Zurlo and Griffini,
1973; Fugas and Saric, 1979). Total dietary and airborne lead uptake was
estimated using calculations similar to those presented in the integrated
lead uptake model in Section IV.A. The following assumptions given in
Chamberlain et al., (1978) were used: a) a fraction, 0.55, of the uptake
-------
IV—11
becomes attached to red blood cells; b) the biological half-life of lead in
blood is 18 days; c) a factor, 1.3, is to be allowed for long-term resorption
and re-entry into blood of some of the lead which is stored in bone; and
d) the mass of blood is 5400 grams. With these assumptions, Chamberlain
and Heard calculated the following relationship:
& PbB = 0.55 x 18 x 1.3 = 0.34 pg/dl per ug/day
a Uptake 54 x 0.693
This relationship is comparable to that derived from the Cools and
Sherlock studies after gastrointestinal absorption is accounted for. Given
uncertainties about some of the studies analyzed by Chamberlain and Heard
regarding exposure levels and quality control, and because of the high
exposure levels, this slope is presented here only for comparative.purposes
and will not be used further.
4. Lead Balance Studies: Compartmental Models
To demonstrate a causal relationship between lead in the body and a
biological change, it would be ideal to know the amount of lead present at
the site and time of the effect. For instance, if lead is suspected to
induce neuropsychological changes, a measure of the lead level present in
nerve cells when the change occurred would be desirable. Living tissues
can rarely be sampled, and data obtained at necropsy cannot reveal the
variations in exposure throughout the individual's life. For analytical
simplicity, tissues can be grouped together on the basis of lead distribution
characteristics and body burdens of lead represented as a limited number of
distinct, homogenous, and well-mixed pools or physiological compartments
with similar kinetic properties. Mathematical biokinetic models have been
fitted to. data obtained in long-term balance studies and to lead isotope
tracer experiments that estimate the rates of input to, transfer between,
-------
IV-12
and excretion from the different compartments (Rabinowitz et al., 1976,
1977; Batschelet et al., 1979; Bernard, 1977; Mallon 1983; Harley and
Kneip, 1985; Marcus, 1985a,b,c). Differences in the predictive lead models
have been discussed by Bernard (1977), Batschelet et al. (1979) and Hammond
et al. (1981). Models are shown schematically in the CD (Figures 1U-3 and
10-4).
The choice of pools or level of aggregation in a model depends on the
available experimental data, the appropriate time scale, and the pre-
dictive uses of the analysis, and is a compromise between accuracy and
simplicity. For example, blood can be broken down into plasma and erythrocytes
and then further into plasma protein-bound lead, diffusible lead, extracellular
fluid lead, and erythrocyte proteins. Soft tissues can .be broken down into
brain (hippocampus, medulla, etc.), kidneys, liver, hair, and so on. The
hard tissue pools can be separated into compact (cortical) and cancellous
(trabecular) bones, and teeth; within each of these, separate components
may be needed to model distinct diffusion time scales (Marcus, 198ba,c).
Because the skeletal system in young children is rapidly developing and is
both large and kinetically active, it is especially important to model bone
lead in children. Until such refinements are further tested, a relatively
simple 3-5 pool model provides an adequate framework for predicting lead
distributions in children.
These models are helpful in predicting total body burden or equilibrium
levels of lead over time in any of the presumed kinetic compartments under
different exposure conditions. The basic assumption of the models is that
the mass of lead in each of the compartments changes according to a system
of coupled first-order linear differential equations with constant fractional
transfer rates. Such models predict that when the lead intake changes from
-------
IV-13
one constant level to another, there is a directly proportional change in
the mass of lead in each compartment and the attainment of a new equilibrium.
The rates and magnitudes of these changes are dependent upon the rates of
lead flux in the tissues and can theoretically be calculated from a compartmental
model of the appropriate parameters. Support for a first order kinetic model
for lead metabolism is demonstrated by calculations, using first order models
of soft tissue and bone concentrations of lead, and other elements (calcium
strontium, radium), that fit human measurements as well as by using more
complex models (Johnson and Myers, 1981; Mallon, 1983).
A four-compartment biokinetic model of lead metabolism has been developed
from data obtained in controlled single dose and chronic lead exposures of
infant (9 months) and juvenile (22 months) baboons (Mallon, 1983; Kneip et
al., 1983). Dynamic blood measurements and steady state blood and organ
lead measurements were closely fitted to predict concentrations of lead
in blood, liver, and kidney, and bone (the four compartments in which 95%
of total body lead is contained) (Heard and Chamberlain, 1984). Human
metabolism and growth patterns were applied in a computer simulation of the
model that was then successfully validated using human autopsy data.
The model parameters were revised using measured metabolic data for
each organ (e.g., bone turnover rates) for chijdren and were used to
simulate organ lead burdens and concentrations in children with constant
lead exposures from birth (Harley and Kneip, 1985). Although complete
model validation is not possible, the revised model is consistent with experi-
mental data on blood lead accumulation following dietary lead uptake among
infants (Ziegler et al., 1978), and skeletal lead accumulation following
controlled exposures in adults (Heard and Chamberlain, 1984). Given
-------
IV-14
the available data, the model appears to provide the best estimates of PbB
levels in children with continuous lead uptake over time.
Table 4-2 presents PbB levels predicted by the model for male children
under different exposures, or lead uptake levels. The values in Table 4-2
include an adjustment for the propagation of maternal lead during pregnancy
that persists throughout childhood. This adjustment was done using another
kinetic model fit to blood lead data collected longtudinally in young children
from birth to 27 months (Succop et al., 1987) and assuming a baseline maternal
PbB level of 4 pg/dl (see Section C.6). The following equation adapted from
Succup et al. was used:
dPbBt = PbB0 (e" ^ *)
where: dPbBt = added blood lead increment propagated from maternal PbB at
time t;
PbB0 = baseline maternal PbB (4 yg/dl);
and the rate constant a = 0.072.
The rate constant a = 0.072 mo-1 corresponds to a mean residence time
of 1/ a =14 months, or a half-life of 1n(2)/ a = 10 months. This almost
certainly represents resorption of that fraction of prenatal lead burden
sequestered in the neonatal skeletal tissues. The apparent mean residence
time of lead in blood in infants is much shorter (see Section V.A.5).
No differences were found by tfie biokinetic model between the sexes
except at older ages, and predicted PbB levels were highest among 2-3 year
olds, consistent with results of NHANES II, the New York City screening
program (Billick et al., 1979), and ongoing longitudinal study (McMichael et
al., 1986). These results will be applied to estimates of the integrated lead
uptake model to estimate children's PbB levels under alternative air lead
levels. It is important to note that this equilibrium blood lead model is
linear in total lead uptake at each age. This is a necessary consequence of
-------
IV-15
TABLE 4-2.
PREDICTED EQUILIBRATED BLOOD LEAD LEVELS (ug/dl) OVER TIME AMONG CHILDREN
WITH CONSTANT LEAD UPTAKES3
Lead Uptake (|jg/day)
Age
10
20
30
40
50
60
70
80
¦ M
1
TTo
'579
379
TT.9
TT.8
T7.8
2o\ 8
21.8
2
4.9
9.0
13.0
17.1
21.1
24.2
28.3
32.3
3
4.6
8.2
11 .9
15.5
19.2
22.0
25.6
29.3
4
4.5
8.2
11.8
15.4
19.0
21 .8
25.4
29.0
5
4.4
7.9
11.4
14.9
18.4
21.0
24.5
28.0
6
4.4
7.8
11.3
14.7
18.2
20.7
24.2
27.6
7
4.2
7.4
10.7
13.9
17.2
19.5
22.8
26.0
8
3.4
5.9
8.3
10.8
13.3
14.9
17.4
19.9
9
2.3
4.6
6.9
9.3
11.6
13.9
16.2
18.5
10
2.6
5.2
7.8
10.4
13.0
15.6
18.2
20.8
aFrom Harley and Kneip (1985) after accounting for propagation of maternal
blood lead during pregnancy (see text).
-------
IV-16
the linear pharmacokinetic model assumed by Harley and Kneip (CD, Appendix
11-A). The uptake/biokinetic model is thus logically consistent with
aggregate and disaggregate linear total-uptake models discussed in the
following sections.
Figure 4-1 compares the relationships between lead uptake and blood lead
derived from the various studies on infants and adults. Despite the diverse
nature of the populations, study designs, and methodologies, there is a fair
degree of consistency in the relationships. Each study found that a linear
function provided as good a fit, if not better, than other non-linear forms at
the relatively low exposure levels investigated. Some experimental and epi-
demiological evidence suggests however, that the relationship between lead
concentrations in tissues and cumulative lead intake is only approximately
linear at low levels of intake, and that successive increments in intake or
exposure result in progressively smaller contributions to blood lead concen-
trations (Azar et al., 197b; Moore, 1977; Gross, 1981; Sherlock et al., 1982).
The curves drawn in Figure 4-1 for infants and adults do in fact include smaller
slopes for lead uptake values above 20-40 ^ly/aay. This curvilinear
relationship may be due to increased renal clearance with higher blood lead
(Gross, 1981), distributional non-linearities due to differences in lead
binding sites in different tissues (Hammond et al., 1981; Marcus, 1985b;
Manton, 1985), and/or to a sizeable pool of mobile lead in bone maintained
more or less independently of uptake (Rabinowitz et al., 1977; Chamberlain,
1983). It appears however, that none of the mechanisms introduce significant
non-linearities at blood lead levels below 30 gg/dl (Marcus, 1984, 1985a,c;
Chamberlain, 1983) and that a linear mathematical model is valid for relatively
low to moderate lead exposures (CD, p. 10-31, Appdx. 11A-2). As discussed
above, at levels above 30-40 gg/dl, blood lead may be an inadequate index
-------
BLOOD LEAD (ug/dl)
30
RYU/LACEY
SHERLOCK/COOLS
25
HARLEY Sc KNEIP
20
0
20
40
60
80
100
LEAD UPTAKE (ug/day)
FIGURE 4-1. Summary of Relationships Between Daily Lead Uptake and Blood Lead for Different Age
Groups, Derived from a) Ryu et al. (1983) and Lacey et al. (1985) Studies on Infants;
b) Sherlock et al. (1982) and Cools et al. (1976) Studies on Adults; and c) Harley and
Kneip (1985) Biokinetic Model with Estimates for 2-3 year old children illustrated.
-------
IV-18
for tissue lead burdens in many children (Piomelli et al., 1984) and linear
models are likely to lose their predictive power. For this reason, tne
relationships depicted in Fiyure 4-1 are truncated at 3U gy/dl. To estimate
PbB levels above 30 pg/dl, which is now above the PbB level of health-related
concern for children, use of non-linear models discussed in the criteria
document would be required (CD, Appendix 11.B).
The compartmental biokinetic model of lead metabolism developed by Mai Ion
(1983) and Kneip et al. (1983), and revised for children by Harley and Kneip
(1985), relied on a broad array of experimental and observational measurements
of mammalian metabolism and growth patterns, and has been successfully
validated using available human experimental and autopsy data. As is the
case for any mathematical model, there are inherent limitations and associated
uncertainties. Because this model is based on the most comprehensive data
available and has been developed specifically to predict organ lead concen-
trations over time in young children with continuous lead uptake, it appears
that the outputs of the Harley and Kneip (198b) biokinetic model are the
most appropriate to predict PbB levels.in children using the integrated
lead uptake estimates presented in Section IV.A and Appendix A. PbB levels
calculated from the conjunction of the integrated lead uptake and biokinetic
models will be presented in the staff paper described in the Introduction
(along with estimates using other modeling approaches that are discussed
later) in order to estimate potential health impacts under alternative air
lead levels. Methodologies for calculating children's and adult blood lead
levels data under alternative standards using other data and other modeling
approaches will be discussed in Section V.
-------
V. STATISTICAL RELATIONSHIPS BETWEEN BLOOD LEAD AND AIRBORNE LEAD:
USE IN DISAGGREGATE AND AGGREGATE MODELS
This section addresses two other methods of determining the contribution
of air lead to total exposure and the potential health impacts associated
with alternative air lead levels. The previous section examined the re-
lationships between blood lead in children and estimated levels of lead
uptake from the air, housedust, outdoor soil and dust, and diet. The two
approaches presented here develop a direct relationship between air lead
and blood lead from various experimental and community or epidemioloyical
studies: 1) a disaggregate model in which total exposure to air lead is
assessed by separately analyzing the relationship between blood lead and
inhaled air lead, versus the associated changes in blood lead as a result
of exposure to lead that has deposited onto soil, dust, food, and water;
and 2) an aggregate model in which a single variable, air lead, serves
as an index for lead exposure through other media affected by integrated
atmospheric lead deposition.
Experimental studies include those in which adult volunteers have
been exposed to controlled levels of laboratory generated lead aerosols.
Because intake of air lead through the diet and dust was probably minimal
for the adult subjects in these experiments, these studies underestimate
the dose-response relationships that would be seen among children with
changes in air lead (Angle et al., 1985). Epidemiological (i.e., community)
studies provide correlations between air lead and blood lead in different
populations of children and adults under varying conditions of lead, from
current or previous atmospheric fallout and other sources, and yield dose-
response relationships at more relevant exposure levels.
-------
V-2
Because of the simultaneous presence of lead in multiple media
with different time profiles, the assessment and use of these studies
is not always direct. As in any attempt to model lead exposure, possible
confounding variables that influence PbB levels but which are not related
to air lead measurements (e.g., lead in paint, canned food and plumbing,
socioeconomic status, parental care, housing and play conditions, calcium
intake) cannot always be disentangled. Control for confounders can be
achieved by comparing populations that differ only in their exposure to
airborne-derived lead, although identifying such groups has been difficult.
Another possibility is to perform a multivariate statistical analysis in
which adjustments are made for some or all of the confounders before calculating
the blood lead/air lead relationship. This requires information on the
value of each confounder for each individual, which is also difficult to
estimate. In the case of lead, several confounders tend to work in the
same direction as air lead, e.g., elevated exposure and unfavorable social
conditions often are both concentrated in central cities. Consequently,
when statistical adjustment is incomplete, the relationship between air
lead and blood lead will likely be inflated. Adjustment for a
confounding variable may result in all of the shared variance between
confounder and exposure variable being attributed to the confounder; this can
inevitably lead to underestimation of the "true" impact of the exposure
variable (Rutter, 1983). In characterizing good studies of air lead/blood
lead relationships, several reviewers (e.g., Hammond et al., 1981; Brunekreef,
1984), as well as the CD, discuss key factors that should be considered. These
include a well-defined study population, a good measure of individual
exposure, measurement of blood lead with adequate quality control, a statistical
analysis model that is biologically plausible and consistent with the data,
-------
V-3
and control or measurement of important covariates (CO, p. 11-63). The
studies identified in the CO, and highlighted here, address these factors
sufficiently to establish meaningful relationships.
A. Disaggregate Model
1. Air Lead/Blood Lead Relationships in Adults
Because the relationship between blood lead and environmental exposure
is nonlinear across the range of potential exposure, but approximately linear
at lower levels (CO, p. 11-65), the focus here is on studies of adults with-
out excessive occupational or personal exposure. Longitudinal studies in
which changes in blood lead were measured in adult volunteers exposed to
controlled levels of laboratory generated aerosols, in some cases with
isotopic lead tracer, are summarized in Table 11-40 of the CD. Data from
the most relevant studies have been reanalyzed in the CD to yield blood
lead air lead "slopes" (3), where 3 measures the change in blood lead
expected for a unit change in air lead.
As noted in the CD, the blood lead inhalation slope estimates vary
appreciably from one subject to another in the experimental studies, and
from one study to another. The weighted slope and standard error estimates
from the Griffin study (1.75 +^0.35) were combined with those calculated
similarly for the Rabinowitz study (2.14 j^U.47), and the Kehoe study
(1.25 _+ 0.35), yielding a pooled weighted slope estimate of 1.64 +_ 0.22
(CD, pp. 11-99 to 11-102). Excluding the subjects in the Kehoe study exposed
to very high air lead levels (up to 36 pg/m^), results in an average slope of
approximately 1.9 pg/dl per pg/nv*.
Several deficiencies in the individual studies are discussed in the
CD such as uncontrolled or unmeasured air lead exposures outside the
-------
V-4
chambers or difficulties in determining non-inhalation blood lead baseline
levels (CD, p. 11-101). Furthermore, these experimental inhalation studies
may underestimate the overall impact of airborne lead (via deposition and
incorporation over time into soils, dusts, water, food chain). This is
much less of a problem, however, in predicting changes in adult lead
exposure in the future, compared to children. For example, although slightly
elevated adult blood lead levels have been observed in areas of high
lead contamination (Barltrop et al., 1975; Gallacher et al., 1984; Rabinowitz
et al., 1985), adults do not ingest significant amounts of dirt and (unless
under fasting conditions), normally do not absorb ingested lead at the
same rate as children.
The cross-sectional relationship between blood lead and air lead in
adults has been examined in several population studies. Azar et al. (1975)
studied five groups of men (cab drivers and office workers) who carried
personal air lead monitors and vdiose non-air lead exposures (water, smoking)
were also measured over a 24-hour period. Several alternative geometric
mean regressions have been calculated, including linear and non-linear
models assuming different non-air contributions to blood lead, and estimations
of the effect of endogenous lead stored in the skeleton using age as a
surrogate measure of cumulative exposure. None of the fitted models are
significantly different statistically (CO, p. 11-81) with a pooled slope
estimate of 1.32 +0.38 pg/dl per pg/m^ (CD, p. 11-105). Although the
other population studies typically used less accurate measures of individual
exposures (e.g., Tepper and Levin, 1975; Nordman, 1975; Johnson et al.,
1975), the range of slope estimates (1.0-2.0) is comparable to the Azar et
al. and experimental inhalation results (Snee, 1981; see CD, p. 11-98).
-------
V-5
Although the studies differed in co-variables measured, populations,
exposure conditions, etc., the most reliable and relevant studies consistently
yield inhalation slope values typically in the range of 1.3-2.0 pg/dl per ng/m3,
with a weighted average slope of 1.4. The above slope estimates are derived
from studies in which exposure to air lead was assessed over days, weeks,
or months and are based on the assumption that an equilibrium level of
blood lead is achieved within a few months after exposure begins. This is
only approximately true, since some of the absorbed lead that accumulates
in the skeleton deposits in the spongy trabecular bones (e.g., rib, vertebrae)
where it may be resorbed into the blood stream. An additional factor
therefore, should be allowed for possible re-entry of lead into the blood
transferred from bone or other long-term storage (Chamberlain, 1983).
Based on isotopic ratio and radiolabel tracer studies on adults, Chamberlain
et al. (1978) estimated this factor to be 1.3. Similar experiments later
indicated that about 20% (i.e., a factor of 1.2) of an adult male's blood
lead is from bone lead resorption (Chamberlain, 1985). Re-entry into blood
of stored lead is likely to be lower in children because of the rapid
growth, high rate of turnover and smaller reservoir of lead in their skeletal
systems. While no adjustment for bone resorption will be made for children's
slopes, it appears reasonable to apply a factor of 1.3 to the blood lead/air
lead inhalation slope of 1.4 derived above for adults, yielding a slope
estimate of about 1.8 pg/dl per yg/m^.
2. Estimating Future Adult Blood Lead Averages
Two adult populations are identified as especially sensitive to lead:
middle-aged males, 40-59 years of age; and pregnant women, as exposure
surrogates for fetuses (see CO, Sections 13.7.2 and 13.7.3). Older women
may also be at increased risk due to mobilization of bone lead after menopause
due to the processes of osteoporosis (Silbergeld and Schwartz, 1987).
-------
V-6
Approximately 95% of adult body lead burden is sequestered in bone and
this accumulation can maintain elevated blood lead levels years after high
exposures have ended (CO, p. 10-23). In most cases, fetal exposure to lead
can be expected to be dominated by maternal bone lead stores from past
exposures which, in the U.S., were much higher than current levels.
Accurately predicting future changes in fetal blood lead levels would
require estimates of maternal lead burden and information on how bone lead
stores would be transferred across the placenta. Although it is likely
that there is extensive mobilization of lead, like calcium, during periods
of physiological stress such as pregnancy, there are no biokinetic data to
quantify this dynamic process. In the absence of such data, no attempt
will be made to estimate fetal lead exposures associated with maternal PbB
levels under alternative standards. Given the sensitivity of the fetus,
however, potential risks associated with prenatal exposures will be of
major emphasis in the overall lead NAAQS assessment.
Given a background or "baseline" blood lead estimate representing non-air
lead adult exposures, and an average disaggregated blood lead: air lead
relationship, the following simple equation can be used to predict blood
lead means for adult male populations in our case studies:
*x = y0 + 1.8A
where "x = mean blood lead (pg/dl)
y0 = non-air background blood lead (pg/dl), and
A = average air lead (pg/m^)
Estimates of non-air "background" blood lead level contributions for different
adult populations that will be used in the 1990-96 case studies are derived
in Appendix C. Other issues involved in estimating fetal PbB levels are also
discussed there.
-------
V-7
3. Air Lead/Blood Lead Relationships in Children
Three population studies of children liviny near lead point sources
in which covariates were controlled (e.g., aye, sex, dust exposure) were
extensively analyzed in the CD for the most useful and relevant estimates of
inhalation slopes: Angle and Mclntire, ly79 (p = 1.92 +_ 0.60); Roels et al.,
1980(8 = 2.46 +_0.S8); and Yankel et al., 1977/Walter et al., 1980 (s = 1.53
+_ 0.064). The median slope of the three studies is 1.97 (CD, p. 11-189); its
application for use in the disaggregate model is discussed below.
4. Estimating Children's Blood Lead Levels Using Disaggregate Model
The relationship between blood lead and direct inhalation of airborne
lead provides information useful for changes in air lead on a time scale of
only several months (CD, p. 11-189). Over time, suspended lead is deposited
and incorporated into soil, dust, and water, and enters the food chain.
Since prior and current atmospheric fallout directly modify the daily
burden of invested lead, laryer chanyes in blood lead would be predicted if
the associated chanyes in the surface deposition of lead were accounted for,
•
rather than simply inhaled air lead (Angle et al., 1984). To account for
the simultaneous presence of lead in multiple environmental media, the CD
has analyzed the separate mathematical relationships between blood lead and
dietary, soil, and dust lead (described in detail in Chapter 11.4). Using
representative values for lead concentrations in these media, these relation-
ships were applied in a further analysis presented in the CD (CD, Table
13-6) as a way to estimate proportional inputs to total blood lead levels
in U.S. children. A similar disaggregate model was developed by Angle and
Mclntire (1979) and Angle et al. (1984) in forming an inteyrated lead ex-
posure function from measurements of lead in air, soil, and house dust and
relating that to PbB levels of children liviny in various areas of Omaha.
-------
V-8
The disaggregate model developed in the CD, intended to represent
the then "current (i.e., 1983-84) exposure picture," is adapted here in
Table 5-1 for different air lead levels and background levels of lead in
food, water, dust, and soil, expected in 1991). The estimates are mean
PbB levels for 2-year old children; comparable analyses for other ayes of
children can be done using the aye-specific estimates of dietary lead intake
presented in Appendix A. Calculations and assumptions used in deriviny the
estimates are summarized in footnotes to the table and are similar to those
described in Table 13-6 of the CD. The following changes were made for the
present analysis:
1) Estimated levels of dietary lead intake for children in 1990 were
substituted for the estimates for 1983-84 used in the CD. The 1990 pro-
jections, described in Appendix A, are based on the Multiple Source Food
Model developed >n the CD and uses more recent data on the downward
trends in food and water;
2) Soil and housedust lead concentrations associated with different
air lead levels are estimated based on the lony-term relationships derived
from regression analyses of data collected from concurrent measurements of
air, soil, and/or dust lead concentrations at about 45 lead point source
locations (see Table A-4). These same relationships are beiny used to
estimate surface soil and indoor dust lead levels in the uptake biokinetic
model. In contrast, soil and dust lead concentrations.under different air
lead levels were estimated in Table 13-6 of the CD by interpolation of data
from the Angle et al. (1984) study. This study is but one of many data
sets included in the aforementioned regression analysis described in
Appendix 8.
-------
V-9
TABLE 5-1. DISAGGREGATE MODEL OF CONTRIBUTIONS FROM VARIOUS MEDIA
TO 1990 MEAN BLOOD LEAD LEVELS (yg/dl) OF U.S. CHILDREN
(2 YEARS OF AGE): BACKGROUND LEVELS AND INCREMENTAL
CONTRIBUTIONS FROM AIR1
Air Lead (uq/m3)
Source
0.25
0.5
0.75
1.0
1.25
1.5
Background - nonair
food and water^
Dust3
Subtotal
1.5
0.1
TJ
1.5
0.1
T7Z
1.5
0.1
T7S
1.5
0.1
TT
1.5
0.1
tz
1.5
0.1
TZ
Background-Air
Food and Water^
0.2
0.2
0.2
0.2
0.2
0.2
Ingested Dust (with
Pb deposited from air)5
1.0
1.7
2.5
3.2
4.0
4.7
Inhaled air®
0.5
1.0
1.5
2.0
2.5
3.0
Total
3.3
4.5
5.8
7.2
8.3
9.5
^-Adapted from Table 13-6 of CD.
^Estimated dietary intake of non-air Pb in 1990 (from Cohen, 1988a,b; Appendix A)
9.2 yg/day x 0.16 pg/dl per pg/day (from Ryu et al., 1983 and Laxen et al.,
1987 as re-analyzed by Marcus, 1989).
3From CD, Chapter 7, 1/10 dust not atmospheric. Using Angle et al. (1984)
"low area" for soil and house dust levels and median regression coefficients
from Stark et al. (1982): (1/10) x (97 pg/g x 0.0022) + (324 gg/g x 0.0018).
^As in 2 above, but using 1.2 pg/day for dietary intake of atmospheric lead.
Derived for component of background Pb in food from past deposition from air
onto soil and into other media leading into human food chain.
^Regression equations of Stark et al. (1982) used along with levels of
^soil dust and house dust predicted by regression analyses of collected point
"source environmental data, described in Appendix B. For example, the mean soil
and house dust lead concentrations associated with 1.0 ug/nr in air are
approximately 630 ppm and 1035 ppm; respectively.
The effect on blood lead would be:
(630 x 0.0022) + (1035 x 0.0018) = 3.2 ug/dl per pg/m3.
6Using the median inhalation slope of 1.97 yg/dl per yg/m3 from the CD, p. 11-189.
-------
V-10
3) The soil and dust/blood lead regression coefficients used to compute
soil and housedust contributions to blood lead are from the CD analysis
of the Stark et al. (1982) study and are described in the CD as the most
reasonable median estimates (pp. 11-151, 11-191). The coefficients used in
Table 13-6 of the CD are from the Angle et al. (1984) study and represent
upper bound values.
It must be emphasized that the blood lead estimates presented in Table
5-1 represent population mean PdB levels and are illustrative only. To
assess health risks associated with such air lead exposures, it would
be necessary to calculate PbB distributions around these mean levels, as
discussed earlier in Section III.B.
The uncertainty in predicting mean blood lead using the disaggregate
model can be estimated in principle since standard errors associated with
the air Pb:PbB slopes are available (CD, p. 11-105). The differences among
the slope estimates from different studies can be ascribed to many factors
including differences in study populations, lead bioavailabilty, and to
statistical artifacts, e.g., the attenuation of slopes attributable to environ-
mental measurement uncertainties (Fuller, 1987). Differences in estimated
standard errors of the regression coefficients reflect sample size, range
of values of environmental variables in the sample, and control of other
sources of variability. A composite slope could be estimated as a weighted
average with weights inversely proportional to the variances for each of
the slopes. Total variance would appropriately include both within-study
and between-study variation and therefore all the variances and covariances
of the parameter estimates would have to be calculated in a series of
iterations. Although such analysis could be conducted, it was decided that
the information about uncertainty of predicted mean blood lead added by
this complicated procedure would not be significant, especially in view
-------
V-L L
of the large uncertainties about the form of the dose-response model,
the judgmental selection of key studies, and the contributions from non-air
sources of lead.
Since completion of the 1986 CD, several findings have become available
that supplement previous conclusions. Unpublished analysis by Johnson and
Wijnberg (1988) of the 1983 study of young children living near the East
Helena smelter (CDC, 1983) estimate slopes for both blood lead-soil lead
and blood lead-dust lead of approximately 1.4 pg/dl per 1000 ppm. This
analysis adopted modeling used by EPA in the CD to fit other similar studies
and adjusted for numerous covariates including children's age, air lead,
dust and/or soil lead depending on analysis, poor quality lead painted
housing, smoking, and secondary occupational exposure.
As indicated in many of the studies, the dust lead slope is similar to
the soil lead slope, suggesting that soil may contribute both directly and
indirectly to blood. Such hypotheses regarding cause-effect responses of
blood lead to earlier exposures have been explored in recent longitudinal
studies on blood lead and developmental indicators in Cincinnati (Bornschein
et al., 198b, 1986) and Boston (Rabinowitz et al., 198b; Bellinger et al.,
1987) children. The Cincinnati data set was analyzed by structural equation
methods (Bentler, 1980) for relationships between exterior surface scrapings,
dust lead, and blood lead. Unfortunately, the results are expressed as a
linear equation in logarithms of the environmental variables, and the
standardized regression coefficients are not directly interpretable as
overall slopes for soil or dust lead given the published information (Born-
schein et al1986).
The analyses of the Boston data used a somewhat different technique—
random effects models for longitudinal data (Ware, 1985). Unfortunately,
-------
V-12
soil lead data were not sampled longitudinally. There were repeat
measurements for blood, dust, air and water lead. The 18-month blood lead
regression model reported by Rabinowitz et al. (1985) used the logarithm
of the yard soil lead measurement as a predictor, so that the reported slope
is not directly comparable to those cited in the CD. However, the slopes at
several concentrations are similar to those cited in the CD: 8 pg/dl per
1000 ppm soil lead when soil lead = 100 ppm, 1.6 gg/dl per 1000 ppm soil lead
when soil lead = 500 ppm, and 0.8 gg/dl per 1000 ppm when soil lead = 1000
ppm. Subsequent analyses of this population using composite blood lead
averages between 6 and 24 months (and a mean soil lead level of 700 ppm)
yielded a slope of 0.9 pg/d1 per 1000 ppm soil lead, varying from 0.6 to
1.6 according to the reported amount of mouthing by the children (Rabinowitz
and Bellinger, 1988). Because the Boston study group did not generally live
in crowded conditions, in homes with deteriorating leaded paint, or have
nutritional deficiencies, their results may not be directly applicable to
children at higher risk.
5. PL3usi_t>iJj_ty_of_Di_^a^re^ated_:yj0^
Clq Etfl® Jill® L
The disaggregated air, soil, and dust lead "slopes" used here are
plausibly consistent with the uptake/biokinetic model parameters. To show
this, we use the well-known relationship (e.g., Rabinowitz et al., 1976) between
blood lead increments, dPbB, and lead intake increments, dPbl;
dPbB = dPbl * (mean residence time in blood pool)/ (volume of
distribution in blood pool)
For a two-year old child, the Harley-Kneip model predicts a mean residence
time in the blood pool of about 8 days, i.e., a blood lead half life of 8
* 0.693 - 5.6 days, consistent with Duggan's (1983) estimate of 4-6
-------
V-13
days. The blood lead pool is much larger than the blood lead volume,
including accessible parts of the soft tissues, bone and bone marrow, and
extracellular fluid. Rabinowitz et al. (1976) estimates that the volume
of this pool in adult volunteers was 7.5 - 10.8 Kg (or approximately
70-100 dl), almost twice the adult blood volume. Thus, a factor of 2 is
used to calculate the total volume of available blood lead distribution in
the equation below. Since the blood volume in infants (about 12% of body
weight) is proportionately greater than in adults (about 8% of body weight),
we will assume the blood volume in 1-5 year old children is about 10% of
body weight, adjusted for whole blood density of 1.06 kg/1. Thus, the
volume of distribution in a 10 kg child (about 2 years old) is taken as
about:
2'* 0.10 * 10 k.y/(1.06 kg/1) - 2 1 - 20 dl.
Assuming that the total amount of dirt consumed, C, is about
0.1 g/day, of which a fraction a = 0.25, approximately, ts absorbed (see
Appendix A.17), the leaded dirt is partitioned as some fraction p of soil
and 1 - p of interior house dust. Thus:
dPbl = c * a * (p * PbS + (1 - p) * PbD)
Assuming p = 0.5 i.e. equal soil and dust access,
dPbB = (0.1 g/d) * (0.25) * (0.5 PbS + 0.5 * PbD) * 8d/20 dl
= 0.005 PbS + 0.005 PbD
Thus the soil lead vs. blood lead and the dust lead vs. blood lead slopes
are both predicted to be about 0.005 pg/dl per ppm.
A variety of factors may be responsible for the somewhat smaller
empirical slope estimates, typically about 0.002 (see above). Even
though the individual household PbS and PbD levels are available in
some studies, these are not necessarily the exact exposure covariates,
but only surrogate measures of dirt lead exposure. It is well
-------
V—14
known that exposure measurement "error" tends to flatten or attenuate
regression slopes in dose-response studies (e.g., Pickles, 1982; Fuller,
1987; Brunekreef et al., 1987). Further, specification of the constants
(e.g., a < 0.25 or C < 100 mg/d) could also reduce the predicted slopes.
The agreement is nonetheless impressive.
The air lead slope is similarly calculated. From Table 4.1, the point
source uptake at 1.0 is 0.6 - 0.9 Mg/d, thus
PbB = (0.6 - 0.9) * PbA * 8d/20 dl
= (0.24 - 0.36) PbA
The estimated direct inhalation slope of 0.24 - 0.36 is much smaller
than has been cited for empirical disaggregate studies. These discrepancies
have been previously noted in Marcus (1988b) in an analysis of East Helena
data. The differences between the uptake/biokinetic model apparent
PbD slope and the disaggregate regression slope (0.00341 vs. 0.00151),
apparent PbS slopes (0.00199 vs. 0.00102), and apparent PbA slopes (0.63
vs. 1.08) were in the same direction. It is possible that the empirical
multiple regression studies have misal1ocated the variance in log(PbB),
attributing too little effect to soil and dust ingestion. We cannot determine
whether these are artifacts of the multiple regression model or of the
uptake/biokinetic simulation model. Nevertheless, the slope values used in
the disaggregate model are in general consistent with the uptake/biokinetic
model.
B. Aggregate Model
1. Air Lead/Blood Lead Relationship
In the disaggregate modeling approach, "inhalation" slopes are derived
from epidemiological studies by making adjustments for measured indirect
-------
V-15
air exposure variables (e.g., house dust lead). These inhalation slopes are
then combined with separate slopes for other exposure media (e.g., dust,
food, water) to arrive at an integrated lead exposure function. With the
exception of Angle and Mclntire (1979), the community studies from which
inhalation slopes have been derived have not simultaneously measured lead
in more than two or three media. Consequently, the integrated lead exposure
function is necessarily based on data from multiple studies involving
different populations, exposure conditions, measurement techniques, etc.
An alternative method of calculating the effect of changes in air lead on
children's blood lead is to analyze individual community studies in which
reliable "unadjusted" blood lead/air lead relationships can be derived such
that the impact of both direct (inhaled) and indirect (via dust, soil,
etc.) contributions of air lead are combined, or aggregated, in one variable
(i.e., air lead). This approach has the advantage over disaggregate modeling
(using the adjusted "inhalation" slope) in that the total impact of atmospheric
lead on children's exposure is better captured since since air lead levels,
in general, are causally related to other important exposure variables such
as hand and household dust lead (Brunekreef, 1984).
The population studies with identifiable air monitoring methods and
reliable blood lead data are summarized in Table 11-36 of the CD. None of
the studies included adult populations. Since adults do not typically
ingest much deposited lead anyway, the aggregate modeling approach will be
limited to young children. In Table 5-2, the calculated 3 values for each
study are presented according to the blood lead levels, ages, and type of
location of the investigated children. Different statistical analyses of
these studies have resulted in a range of possible values for the blood
-------
Table 5-2. BLOOD LEAD/AIR LEAD SLOPES1 IN CHILDREN FOR DIFFERENT AGES AND EXPOSURES
Type of Location (Study)
Blood Lead (ug/dl)/
Age (years) Urban | Near Point Source(s)
<25/1-7
8.5a»b (Brunekreef et al., 1983)
3.6-4.0a (Zielhuis et al., 1979; Brunekreef
et al., 1981)
<25/1-18
0.6 (Angle and 0.66a (Brunekreef, 1984
1.92 Mclntlre; 1979; 2.10a from Angle and
4.40 CD) Mclntire, 1979)
-
<25/10-15
2.46 (CD from Roels et al. 1976)
5.3a (Roels et al. 1980)
« 5.9-9.8a»b (Brunekreef, 1984 from Roels
et al., 1976, 1978, 1980)
>25/0-9
2.9 (Billick et al., 1979, 1980;
Nathanson and Nudleman, 1980)
1.07-1.52 (Yankel et al., 1977; Walter et al.,
198U; Snee, 1982b; CD)
2.4-3.3a (Brunekreef, 1984 from Yankel et al.,
1977)
>25/1-18
•
2.6-3.7a (Landrigan et al. 1975; Morse et al.,
1977)
4.6a (Brunekreef, 1984 from Roels et al., 1976,
1978, 1980)
^Slopes are 1n units of pg lead/dl blood per pg lead/m^ air.
a"Aggregate" slopes derived from group comparisons or multiple regression analyses that were unadjusted for soil or
dust lead, thereby attempting to account for influence of deposited atmospheric lead through these exposure routes.
Remaining slopes represent relationships between blood lead and air lead due solely to direct inhalation which have
been calculated by adjusting for the influence of soil and/or dust lead.
bUse of low-volume particulate samplers likely underestimated air lead exposures, and overestimated the
value, especially near point source where large particles predominate.
-------
V— 17
lead/air lead relationship (e). Unadjusted (i.e., aggregate) relationships
are presented in addition to adjusted (i.e., disaggregate) relationships
derived from regression analyses, the latter which refer to the blood
lead/air lead relationships due to direct inhalation exposure.
Relatively wide ranges of 6 values are observed for industrial and
urban areas at low and high PbB levels and for all ages of children. The
apparent trend towards smaller slopes with increasing PbB levels is consistent
with findings within some studies (Angle and Mclntire, 1979; Roels et al.,
1980) and other observations of curvilinear blood lead/exposure relationships
(Azar et al., 1975; Moore et al., 1977; Gross, 1981; Sherlock et al., 1982;
Hammond et al., 1981).
As discussed, the CD determined that the disaggregate inhalation
slopes derived from the studies by Angle and Mclntire, Roels et al., and
Yankel et al. are the most reliable given their overall quality. Aggregate
analyses of these data, either by Brunekeef (1984) or the original authors
(Roels et al., 1980) illustrate the significant decline in air:blood lead
slope when adjustments are made for factors that covary with air lead (e.g.,
soil and house dust lead), and the dependence of these slopes on exposure
levels. For example, Brunekreef (1984) segregated the Roels et al. data
and compared groups with large differences in exposure (i.e., <1 km from
smelter vs. urban/rural; 5.9) groups with small differences at high
exposure levels (i.e., <1 km vs. 2.5 km from smelter; 4.6) and groups
with small differences in exposures at low levels (i.e., 2.5 km from
smelter/urban vs. urban/rural; 9.8). As in the Brunekreef et al. (1983)
study, the use of low-volume samplers may have underestimated air lead
exposures, and overestimated the B value, especially for those children
near the smelter where large particles predominate. The slope may be
-------
V-18
underestimated for young children, however, since this study only sampled
children older than 10 years of age.
It is important to note several additional studies and analyses that
may provide equally relevant and useful information for purposes of estimating
an aggregate blood lead/air lead slope. These studies (Brunekreef et al.,
1983; Zielhuis et al. 1979; Brunekreef et al., 1981) used well-defined
study populations (children younger than 10 yrs. old; PbB levels below 25
Hg/dl), employed available quality control procedures for blood lead analysis,
and measured or controlled for important covariates (see CD, Table 11-36).
Several specific comments should be made regarding these studies:
1. The series of studies reported by Zielhuis et al. (1979) and
Brunekreef et al. (1981) included many environmental measurements (lead in
ambient and indoor air, lead in dustfall indoors and outdoors, soil,
streetdust, floordust, tapwater, and dustiness of homes) and regression
analyses to determine the impact of different variables on PbB levels
(i.e., the above environmental indices as well as distance from the smelter,
parental education, age of the child, mouthing activity, and cleanliness of the
child). Venous blood lead samples were analyzed by standard techniques and
although information on interlaboratory comparisons is not given in the
original study, quality control participation is reported by one of the
investigators in a subsequent review (Brunekreef, 1984). Air lead was
measured at two sites 0.2 and 0.4 km from the smelter in 1976 and in 1977.
The levels in Table 5-2 were taken from Brunekreef (1984) and represent
those measured at 0.4 km from the smelter. Air lead was measured at 6
sites continuously for 2 months in 1978. Brunekreef (1984) estimates
3 = 4.0 for 1976 by assuming a difference of 2.0 ^ig/m^ in average
air lead exposure levels between the 2-3 year old children with the highest
-------
V-19
and lowest PbB levels. After adjustment for parental education, the blood
lead difference of about 8 jjy/dl decreased to 7.2 ug/dl, resulting in
a slightly lower 3 estimate (3.6). For 1977, a difference of 1.0 yg/nr*
was assumed for air lead exposure levels between smelter and control children
aged 2-3 years who had PbB levels of 18.2 and 14.6 yg/dl, respectively, again
resulting in a 3 of 3.6. In 1978, only children living between 0.4 and 1.0
km of the smelter were sampled and air levels did not correlate with blood
lead within this population. Thus, no direct estimate of 3 can be derived,
although soil lead and indoor and outdoor dust lead, and therefore, accumulated
lead deposition, accounted for a significant fraction of the variance in
PbB levels (Brunekreef et al., 1981).
2. The more recent study by Brunekreef et al. (1983) on Dutch city and
suburban children measured venous blood (which was analyzed as p-art of
the European Community laboratory quality control program) and many
environmental and social variables and potential confounders (lead in
drinking water, soil, street and playground dust, hand dust, indoor dust,
mouthing behavior, dietary intakes, parental education and occupation, age
of home, etc). The very high 3 value (8.5) derived by Brunekreef (1984)
even after adjustment in a multiple regression analysis for six of the
confounders, may be related to an underestimation of ambient air lead
levels due to the fact that low volume British Smoke air monitors were
employed, in contrast to the hi-vol samplers used in most other studies.
The extent of inflation in the estimated slope as a result of any underestimated
air lead level is uncertain. Brunekreef (1984) notes however that even if
this bias is accounted for, the difference in urban and suburban air lead
levels was probably not 1arger than 0.2 yg/m^. The contrast in lead
deposition between the areas was significant (i.e., 643 vs. 220 mg/m2/day), •
-------
V-20
indicating that ongoing lead pollution accounted for a good deal of the
differences in PbB levels, although variations in historical emissions can
not be entirely ruled out. The relatively low PbB levels in this study
(« 8-13 yg/dl) could be another partial explanation of the high 0 estimates,
given that other studies used to derive 3 included children with PbB levels
on average closer to, and above, 30 yg/dl. At this point, the blood
lead/lead intake relationship is estimated to level off; below 30 ug/dl, •
this relationship appears to be approximately linear (CO, p. 11-104).
The remaining slopes listed in Table 5-2 are based on data from children
whose PbB levels, as some in Yankel et al. (1977), exceeded 25 yg/dl. As
discussed in the CD, the relationship between lead uptake and PbB levels
above about 30 ug/dl appears to be non-linear (CD, p. 10-31). Because Pb8
levels above 25-30 ug/dl exceed those where health-related concerns are
triggered (see CO, Chapter 13), these slopes are less relevant to the
present review.
In summary, the analyses using the aggregate approach assume the
same source for most lead in air, soil, and housedust, and that adjustment of
PbB levels for soil or dust lead to yield an inhalation slope, under-
estimates the "true" impact of atmospheric lead on PbB levels. A range of
3 values can be estimated from a) additional, and apparently relevant, studies
not used to derive inhalation slopes in the CO (i.e., Zielhuis et al.,
1979; Brunekreef et al., 1981, 1983) and b) aggregate analyses that include
both direct (inhalation) and indirect (via soil, dust, etc.) air lead
contributions in the key "inhalation slope" studies cited in the criteria
document. Although far from conclusive, these studies and analyses suggest
a range of possible blood lead/air lead aggregate slopes in the range of 2
-------
V-21
to 10 for young, moderately exposed children with the most reliable slopes
falling between 3 and 5 {CO, p. 11-104).
2. Predicting Children's Blood Lead Levels Using Aggregate Model
To predict PbB levels associated with alternative air lead levels using
these aggregate slopes, it is necessary to estimate the contribution to
total lead exposure from sources not affected by a proposed air quality
standard, for example, lead-soldered food cans, lead in plumbing, and
lead-based paint. It should be noted that by using measured lead concen-
trations in dust from a wide range of locations and conditions, the integrated
uptake model discussed in the previous section implicitly includes average
contributions to total exposure from lead-based paint, but excludes from
the analysis high level exposures associated with deteriorated lead-based
painted housing.
Several difficulties arise in attempting to explicitly estimate
exposures from non-air sources of lead:
1) With the exception of non-air lead in diet, few studies provide
detailed information on the relative contribution of various sources to
children's PbB levels in the U.S. Estimates must be made by inference
from earlier survey data; and
2) Because non-air contributions to PbB levels probably vary widely
in space and time among children, a single estimate for the average case
may result in a lead NAAQS that is not protective for all children. [This
can be expected, for example, for children who regularly ingest lead-based
paint.] Conversely, compliance with the air standard will provide extra
protection in areas where lead from non-air sources is below the average.
-------
V-22
In setting the 1978 lead NAAQS, the Agency estimated that non-air
sources of lead contributed 12 py/d1 on average to children's PbB levels.
Since then, significant reductions in PbB levels have occurred, attributable
not only to declines in atmospheric lead emissions but to the gradual
conversion by manufacturers to non-lead soldered cans, in some cases reduction
in the number of old, lead-painted homes, cleaner working conditions, and
by parents and public health agencies to minimize children's lead
exposure. Further declines in lead exposure can be expected as these
trends continue and lead exposure from drinking water drops, due to
increased awareness as a result of widespread, mandated public notifications,
the 1986 Safe Drinking Water Act banning the use of lead solder and
pipes in new construction and plumbing repairs, and to reductions in water
corrosivity by public water suppliers as compliance with the revised national
water regulations for lead begins. The most recent, and well-conducted,
nationwide survey of children's PbB levels was the Second National Health and
Nutrition Evaluation Survey (NHANES II) of 1976 to 1980. Appendix C describes
analyses that a) starts with the average PbB level for children from 1978,
the midpoint of the NHANES II survey, and b) estimates a 199U PbB average by
adjusting the 1978 value to account for the important changes, summarized
below, in lead exposure that have recently occurred and that can be expected
to continue:
1. Use of lead in gasoline has declined by about 90 percent since
1978; this trend will continue up through 1990-92 as compliance with the
lead in gasoline standard is completed and as the fleet of lead-burning
cars shrinks. As observed during the NHANES II survey period (see CD,
Section 11.3.6), the continued dramatic decline in gasoline lead emissions
is predicted to parallel a major shift in PbB levels.
-------
V-23
2. A significant reduction in dietary lead intake has occurred since
the late 197U's, and this trend will continue as atmospheric lead emissions
and deposition and the use of lead-soldered cans, continue to decline.
The derived range of mean PbB levels for children expected in 1990
(4.2 - 5.2 pg/dl) reflects changes that have occurred since 1978, and that
are expected to continue, in gasoline lead emissions and deposition and canned
food technology. Any other changes that are more difficult to quantify are
also probably of lesser importance for the bulk of the population. For
example, changes for children living in lead-painted housing has not been
significant on a broad scale. This range of 1990 "baseline" average estimates
represents the mean PbB levels that would be expected in 1990 in U.S. children
not exposed to atmospheric lead from lead point sources directly or indirectly,
based on the data and assumptions presented above.
Another method to estimate children's mean "non-air" PbB is to use
available data on typical background levels of lead in food, water, dust,
and soil ingested by U.S. children and the relationship between lead
taken up through these media and children's PbB levels. Table 13-6 of
the criteria document uses such data in calculating a mean PbB of 4.42
VJg/dl to be expected at an air lead level of zero. Although this value was
calculated based on 1983-84 FDA data on lead in food, it is interesting that
it is within the range described above. This range can be used in the
aggregate model approach to estimate ranges of mean blood lead levels under
alternative air lead exposures in the following equation:
mean PbB = (4.2 - 5.2 gg/dl) + (3 to 5 gg/dl per tJg/nr*) Air Pb
It must be emphasized that these estimates for non-air contributions to
average PbB levels represent average values. Many children may be at risk
for significantly higher lead exposures that cannot be prevented by atmospheric
-------
V-24
emission controls. In particular, these include children who live in
deteriorated or recently resurfaced lead-painted housing who deliberately or
inadvertently ingest paint dust through normal mouthing activities, children
exposed to high drinking water lead levels from eroding lead pipes or
solders in distribution systems, children of parents who work in lead-related
industries and are exposed to lead dust that is subsequently carried home
on clothing, or children living near lead smelters and other point sources
where historical accumulations of lead are excessive. It is important that
other regulatory agencies and public health programs, including other EPA
components, responsible for minimizing children's lead exposures from non-
air sources, or from historical accumulations of atmospheric lead deposition,
maintain or increase, where necessary, their efforts.
-------
VI. VALIDATION OF INTEGRATED LEAD UPTAKE/BIOKINETIC EXPOSURE MODEL
While all three modeling approaches identified in Section III are
useful in assessing the relative protection afforded by different
lead NAAQS, the aggregate and disaggregate blood lead-air lead models are
intended to be "equilibriun" models. However, we know that there is a great
deal of variation in past and present exposures to lead from various
sources and pathways. 0-f the three, the uptake/biokinetic model can best
estimate the changes over time of blood lead to changes in environmental
lead in rapidly developing young children.
A general description of the model is provided in Section IV and
Appendix A. Average daily uptake for young children is calculated under
conditions specified in terms of ambient air lead levels, soil and dust
lead levels that correspond to both historical and current atmospheric lead
emissions, and dietary lead levels from both water and food. Additional
exposures to paint lead can be added but are not in the present exercise
given the high degree of variability, and the inadequate data base (see
section II.F). Blood lead (PbB) levels in populations of young children, who
are the most exposed and highly susceptible to lead, are estimated over time
based on total daily lead uptake using a biokinetic compartmental model.
Parameters such as indoor air lead exposure, time spent indoors
vs. outdoors, absorption rates through the lung or gastrointestinal tract,
and amount of dirt that children typically ingest through hand-to-mouth
activity, are estimated from available data in the literature, as summarized
in Appendix A.
In this chapter, results of several validation exercises where predicted
and observed blood leads were compared are presented. The most detailed
analysis was performed with data gathered around a smelter in East Helena,
-------
VI -2
Montana. Two types of validation efforts were undertaken with this data
set: 1) in the first effort, the best data regarding such model parameters
as observed air, soil, and dust lead exposure estimates were used; 2) in the
second validation effort, predicted levels of air, soil and dust levels were
used to estimate PbB levels. The latter work was undertaken to determine
how well the model behaved when actual measurements of necessary input data
are not available. This was necessary so that the reliability of the model
could be assessed for policy analysis purposes vhen less than full information
is available to use with the model. Also, presented in this section are
validation exercises using observed blood lead estimates in locations other
than East Helena. While less effort was undertaken in these exercises,
they nonetheless provide additional information on the reliability of the
model and provide insight regarding the sensitivity of the model results
to uncertainties in the input data. Given the many uncertainties in the
input data and the biological variability that cannot be incorporated, the
results of the validation exercises presented below reveal that the lead
uptake/biokinetic model performs quite well in predicting mean blood lead
concentrations in children living near point sources of lead.
A. Validation Using 1983 East Helena Data
The data set considered first is based on a 1983 study in which the
Montana Department of Health and Environmental Sciences (MDHES) and the
Centers for Disease Control (CDC), in cooperation with EPA, measured
blood lead levels in approximately 400 children ages 1-5 living around
the ASARCO lead smelter in East Helena. The lead content of soil and
dusts around and in individual homes was measured. Airborne lead was
measured before and during the survey at 8 sites. Three study areas were
designated according to their distances from the smelter: Area 1, within
-------
VI-3
1 mile of the smelter; Area 2, 1-2.25 miles from the smelter; Area 3,
more than 5 miles from the smelter. The validation exercises included
children from Areas 1 and 2 only.
In order to estimate 1983 PbB levels in the East Helena children so they
could be matched against the measured PbB levels, exposure profiles were
generated extending as far back as 1978, depending on the children's age in
1983 and their residential location. A five-year description of monthly
ambient air lead concentrations throughout the study area was generated using
the Industrial Source Complex-Long-Term (ISC-LT) dispersion model based on
air lead concentrations, source sampling data of smelter emissions, and local
meteorological data, and accounting for dry atmospheric deposition. Area and
year-specific estimates of background lead contributions (i.e., mobile sources,
re-entrained soil, local minor point sources) were also included to account
for total ambient exposure. Only one site, about 1/4 mile from the smelter,
provided sufficiently complete air quality data over the period 1978-1983
that could be considered representative of population exposure in the vicinity
of the monitor although air lead data vere available from monitors in ten
other locations around East Helena. A total of 158 monthly average lead readings
were available from these monitors between 1980 and 1983. These averages were
compared with predicted air lead concentrations from the ISC-LT dispersion
model. As would be expected with a dispersion model using limited meteorological
data, the model predicted no more than 37% of the month-to-month variability
in monitor readings. However, simple linear regression predicting monitored
readings as a function of a constant and ISC predictions, indicated that
reasonably accurate estimates of air quality were made with the ISC model: the
constant was within the range of the estimated background concentrations and
was statistically significant, and; the coefficient for the ISC prediction
variable was not significant.1 y different from 1.0.
-------
VI-4
Estimates of soil and dust lead concentrations were also obtained using
two methodologies. In the first, case, home-specific observed values for
soil and dust lead were used. In the second case, soil and dust lead estimates
were made using models developed by McLamb (1988) and Marcus (1988c)
that describe soil and dust lead as a function of ambient lead. (For a
description of these models, see Appendix R). The latter approach was
deemed necessary to determine how well the model predicts when full
information on model inputs is not available. (This is indeed the usual
situation when the model is used to assess the impacts of alternative
regulatory programs on mean blood leads of a population of children.)
Age-specific exposure parameters used as input to the uptake model
are described in Appendix A. Estimates of dietary lead intake for different
age groups between 1978 and 1983, listed in Table A-2, were calculated
through application of the Multiple Source Food Model developed in Chapter 7
of the CD. The latter model uses year-specific FDA data on food lead content
(i.e., market basket surveys) and age-specific data on dietary patterns
throughout childhood and has since been validated using recent 1984-85
data (Flegel et al., 1988). The contribution of drinking water to dietary
lead exposure was estimated for an average tap water lead concentration
of 12 pg/1, which was the level reported for the State of Montana in
1983.
The criteria for validation need to be carefully defined. There is a great
deal of individual variability in PbB levels, even with available estimates
of lead concentrations in exposure pathways such as air, dust, and soil.
The variability is attributable to individual biological variation, to
unknown factors mediating exposure such as frequency of hand-mouth contact,
and to other unattributed sources of lead. Even with good environmental data,
the fraction of variance in the logarithm of blood lead that is accounted
-------
VI -5
for by an optimal total uptake regression model (multiple R?) rarely exceeds
30%. For the East Helena data, = 0.28 in the best fitting models (Johnson
and Wijnberg, 1988). Thus, it does not seem fruitful to use the individual
predicted PbB levels as a criterion for model validation. We used the more
realistic goal of matching the geometric mean PbB for children in a large
neighborhood that was moderately homogeneous with respect to exposure.
A more global validation involves comparing the cumulative distribution
function of observed PbB with that of the individual PbR estimates from the
uptake/biokinetic model. These comparisons from the different validation
runs are summarized in Table 6-1.
The simulations were all run at two levels, one with all parameters
set to the lower bounds for uptake parameters in Appendix A, and the second
with all parameters set to the upper bound levels in Appendix A. The
midpoint of the all-lower and all-upper bound PbB estimates, i.e., the
simple mean was found to be the best predictor.
In model run A, only observed values for air, soil and dust input
data were used. This data set was limited to the 28 observations near the
smelter where an ambient monitor was within a few blocks of the homes.
Model run B includes observations on 299 children. This expanded
data set was obtained by using dispersion modeling techniques to estimate
block specific ambient exposure estimates.1 Comparisons between runs A and B
illustrate that accurate PbB reductions result from using either observed
1 A similar validation exercise was conducted previously where average
PbB levels predicted by the uptake/biokinetic model for different census
tracts were compared to measured PbB levels in 1-5 year olds living near
twd secondary lead smelters and in a reference area, in Dallas. Results
were useful in indicating that greater refinement in the spatial scale
would be necessary in accurately modeling high exposure situations.
For the East Helena analysis, and in subsequent case-study modeling,
the highest degree of spatial refinement possible (i.e., city blocks or
block groups) is used.
-------
TABLE 6-1. COMPARISON OF INTEGRATED LEAD UPTAKE/B10K1NET1C MODEL PREDICTIONS
TO 1983 MEASOREMENTS IN EAST HELENA
Model
Run
Population3
Pb Air Quality
Dust/Soil Lead
Predicted vs. Measured
Average Blood Leadb and GSDC
28 children living in
area 1 around single
air monitor with
valid data
89 Area 1 and 210
Area 2 Children
89 Area 1 and 210
Area 2 Children
Measured at monitor
approx. 1/4 mile from
smelter
Contribution from
smelter emission
estimated by dis-
persion model (local
meteorology data)
local background Pb
levels (e.g., auto
emissions, fugitive
dusts) estimated
from 1982 source
apportionment study
Same as for Run B
except that dis-
persion modeling
used ineteoroloyical
data from a nearby
ai rport
Measured in and
outside individual
homes
Measured in and
outside individual
homes
16.3 vs. 16.5 pg/dl*
1.42 vs. 1.49
9.3 vs. 9.3 Mg/dl*
1.61 vs. I.b9
<
W
ON
Estimated from
generalized air:
soil/dust Pb
relationships from
regression analyses
of available data in
literature
9.b vs. 9.3 pg/d1 *
1.67 vs. I.b9
Note: Dietary Pb exposure estimated for all model runs based on year-specific analyses of 1980's FDA food lead
concentration data and food consumption data from Pennington (1983, 1986), and U.S.D.A. Nationwide Food
Consumption Survey, 1977-78.
aAreas as defined in the text
bGeometric means
cG'eometric Standard Deviation
*A11 differences statistically insignificant; t-tests of differences of means assuming lognormal blood lead
distribution and non-equal variances.
-------
V1-7
or estimated ambient air data. Run C, like run B, has 299 observations.
However, in the latter model run, observed measures of household-specific
soil and dust lead are replaced with estimates obtained from equations
that explain soil and dust lead levels as functions of ambient lead (see
Appendix B). Also, while model run C uses dispersion model estimates
of ambient concentrations, as does model run B, the source of meteorological
data is different: Run B uses local meteorological data, and run C uses
airport data. This difference allows runs B and C results to be compared
to assess how sensitive blood lead estimates are to less location-specific
meteorological data as well as to estimated soil and dust Pb concentrations.
Figures 6-1 and 6-2 compare the blood lead distributions predicted by
Runs B and C, respectively, with the actual 1983 measurements of East Helena
children. It is important to emphasize that the model, because of the available
input data, is designed to predict mean population responses and not individual
PbB levels. Individuals with greater than average responses in the upper
tail of the distribution comprise the population of concern, however. As
discussed in Section III.B., the response of the most affected individuals
will be estimated in case-study analyses by calculating the lognormal distribution
around a given mean PbB level using empirically-derived estimates of PbB
variance. The degree of protection under a given exposure scenario will be
characterized by the fraction of children with PbB below some level of concern.
The results of the East Helena validation are shown for individual children
in Figures 6-1 and 6-2, without the application of a GSD. The fraction, or
percentage, protected is on the vertical axis and the potential level of
concern is on the horizontal axis. As can be seen by these figures, and in
scatter plots of individual predicted vs. observed PbB levels (not shown),
there is significant individual variation at higher levels due to extra
variability not included in the model. These figures reinforce the presumption
-------
VI-8
100
90
80
70
MOOEL ESTIMATE
OBSERVED
60
50
40
30
20
10
0
0
10
20
X
SO
40
60
BLOOD LEAD CONCENTRATION. H g/dl
FIGURE 6-1. COMPARISON OF DISTRIBUTION OF MEASURED BLOOD LEAD LEVELS
IN CHILDREN, 1-5 YEARS OF AGE, LIVING WITHIN 2.25 MILES OF E. HELENA
LEAD SMELTER VS. LEVELS PREDICTED BY UPTAKE/BIOKINETIC MODEL. MEASURED
SOIL AND DUST LEAD LEVELS WERE INCLUDED IN ESTIMATING UPTAKE LEVELS.
(RUN B)
-------
VI-9~
100
90
80
70
MOOEL ESTIMATE
OBSERVED
60
SO
40
30
20
10
0
0
10
20
30
40
50
60
BLOOD LEAD CONCENTRATION, M-g/dl
FIGURE 6-2. COMPARISON OF DISTRIBUTION OF MEASURED BLOOD LEAD LEVELS
IN CHILDREN, 1-5 YEARS OF AGE, LIVING WITHIN 2.25 MILES OF E. HELENA
LEAD SMELTER VS. LEVELS PREDICTED BY UPTAKE/RIOKINETIC MODEL. ESTIMATED
SOIL AND DUST LEAD LEVELS FROM GENERALIZED RELATIONSHIPS WERE INCLUDED
IN ESTIMATING UPTAKE LEVELS. (RUN C)
-------
VI-10
that the "importation" of empirically-derived PbB variance is necessary to
fully capture variability. These figures also show that the model, as
designed, gives a useful estimate of the mean response. Using the approach
discussed previously of a) first estimating population mean PbB using the
uptake/biokinetic model and b) then calculating the PbB distribution using
an empirically-derived GSD (in this case 1.53; see Table 3-2), the modeled
PbB distribution would be nearly identical to the observed distribution since
the predicted mean is so close to the observed. Such comparisons are not
illustrated here simply because the differences would be virtually indistinguishabl
In conclusion, generally satisfactory results were obtained for
predicting geometric mean PbB levels. Model run A can be considered a
"pure" validation run that uses the best input data available to predict
mean blood leads which can be compared to mean observed blood leads.
Model runs B and C were necessary validations since exact population
locations, local meteorology, and soil and dust lead data will rarely be
available for point source areas where the model will be used for policy
analysis. Therefore, Run C can be considered a "generalized methodology"
for "export" or application to other study areas.
Although the differences between observed and predicted mean PbB levels
are not statistically significant, post hoc calibration of the model to remove
any differences was considered (Marcus and Holtzman, 1988). Both linear
and nonlinear models were examined and it was found that calibrated
models (including age-dependent ones) improved predictive ability of the
model only marginally. One alternative is to simply subtract the difference
between observed and predicted PbB levels in the generalized methodology
Run C ( -0.2 yg/dl) or use a multiplicative factor (the latter probably
more appropriate given the lognormality of the PbB distribution). Such a
-------
VI-11
calibration based on the results of a single, albeit large and detailed,
validation may be inappropriate, especially considering the uncertainties
inherent in such modeling exercises (e.g., a "positive" calibration may
be needed in a different location). Riven these uncertainties, no gost
hoc calibration of case study exposure estimates is currently being
proposed although such an adjustment could easily be accommodated if
decided upon.
Several sensitivity runs were made in order to "fine-tune" various
parameters for vJiich data are incomplete such as age-specific dirt
consumption rates and gut absorption rates for ingested lead. For example,
in estimating the rate by hhich ingested lead is absorbed, the criteria
document identifies many factors that have to be considered, among them
person's age, physiological status, medium for the lead (e.g., food vs.
paint or "dirt"), coincident ingestion of other elements in food (e.g.,
calcium) which compete with lead for intestinal absorption sites,.and the
non-linear relationship between Jead intake and blood lead (due for example
to a saturable gut absorption pathway). Differences may also be attributable
to physical or chemical differences in bioavailability of deposition from
smelter emissions, or to the higher rate of intake in these areas compared
to typical situations.
In fact, there are reasons to believe that such differences may exist
in East Helena. The lead-bearing particles that are deposited near the
smelter have a higher proportion of larger particles from fugitive emissions
(e.g., erosion from tailings piles, loading operations) whereas the more
distant deposits are predominantly smaller particles from stack emissions.
Another factor is that parents of children who lived closer to the smelter
(e.g., Area 1) may have been more sensitized to lead exposure and thus
tended to control access to leaded soil and dust more than parents in
-------
VI-12
Area 2. There may also have been differences due to nutritional factors.) To
test this, we ran the model with 20% gut absorption of dirt lead in Area 1
and 30% absorption in Area 2.
Two additional runs were made in which the same absorption factor,
either 25% or 30%, was used for both areas. Area-specific results of the
20% (Area 1) / 30% (Area 2) model were not significantly different from the
observed geometric means. Likewise, using a 25% absorption factor for both
areas yielded no significant difference between predicted and observed
geometric means for the overall study area. Results using the 30% rate
cited in the CD based on indirect chemical and animal experiments were
not as concordant. Because of the programming complexities that would be
introduced if dual absorption rates were used in all other applications
of the model, a 25% absorption rate is chosen for general use since it
yielded adequate overall results.
As indicated in Section III, the model should be regarded as valid for
estimating mean responses and it does appear that with the correct range of
soil and dust lead inputs, the model adequately predicts the geometric mean
of the observed distribution. Estimation of the higher percentiles of the
PbB distribution requires application, to the geometric mean, of a 6S0 from,
for example, NHANES II or exposure surveys around different point sources.
B. Application of Uptake/Bioki netic Model to Other Data Sets
There are other published data to test the uptake/biokinetic model
at other U.S. locations and times. One data set was collected in Omaha,
Nebraska, from 1971 to 1977 (Angle and Mclntire, 1979; Angle et al.,
1984). Another data set was collected throughout Silver Valley, Idaho, in
1974-1975 and is described in Yankel et al. (1977), and Walter et al. (1980)
with additional information given in the CD.
-------
VI-13
• * • Omaha_Data
The Omaha study measured blood leads in 242 children ages 1-5 years
and 832 children ages 6-18 years. The children were at 3 locations, denoted
C (commercial area near a small battery plant), M (mixed residential-
commercial), and S (suburban). Air lead, soil lead, and house dust lead
measured in each area are incorporated into the model to estimate average
daily lead uptake levels. These data are less satisfactory for testing
the model than the East Helena data because the soil and dust lead measurements
are not specific to each household where blood leads were taken. Furthermore,
the study was not designed for controlled geographic comparisons given
the significant demographic difference betv*een areas in SES, housing, and
racial composition. Age-specific dietary lead intakes were estimated
based on 1970's data from the FDA and Jelinek (1982) as re-analyzed in
the CD. In Table 6-2, the observed geometric mean blood lead levels for
1-5 .year olds are compared to predicted mean PbB levels for 2-year olds
in the suburban and mixed areas (blood lead data for 1-5 year olds were
not reported for the commercial area). The model appears adequate for
the suburban site, whi1e average blood lead estimated from reported soil,
dust, and air lead concentrations is significantly underpredicted for the
"mixed" site.
There are several factors that may explain the contrast in PbP
levels between the areas despite the apparent similarities in environmental
lead levels, for example:
1) Air lead levels in the mixed area were apparently much higher than
the suburban site in years prior to 1972 when soil, dust, and blood lead
measurements were taken (e.g., 1.44 pg/m^ in mixed vs. 0.73 pg/m^ in
suburban in 1970). Thus, the higher blood leads in the mixed area may
-------
TABLE 6-2. CHILDREN'S BLOOD LEAD LEVELS MEASURED IN OMAHA, 1971-1977 VS. INTEGRATED
UPTAKE/BIOKINETIC MODEL PREDICTIONS
Average Air Lead (yg/m3)
Average Soil Lead (ppm)
Averaye House Dust Lead (ppm)
Observed Mean PbB (pg/dl)
1-5 yr. olds
Predicted Mean* PbB (pg/dl)
2 yr. olds
Mixed" Site
U.26
213
653
25.6
20.2
Suburban Site
0.37
110
567
14.6
15.1
Estimated Dietary Lead Uptake (gg/day) for
both sites during early 1970 's (see text)
0-1 yr. olds 16.3 - 20.6
1-2 yr. olds 20.2 - 25.6
2-3 yr. olds 16.6 - 22.1
~Predicted mean PbB represents midpoint of lower and upper bound means estimated by model.
-------
VI-15
reflect higher historical exposure; 2) The population in the mixed area
vas predominately black with a large percentage of substandard housing
compared to the suburban area vrfiich was predominately white. Rlood lead
levels are highly dependent on demographic variables such as SES, race,
housing conditions, parental care and nutrition, all of i*hich were likely
to have had systematic interactive influences in the Omaha study areas;
3) Other unmeasured variables may have differed between the areas such as
drinking water levels (older homes in mixed area may have had lead plumbing),
paint lead, frequency of canned foods in diets; 4) Possible measurement
error related to the fact that most housedust samples were collected from
vacuum cleaner bags. There may have been differences between areas in
floor surfaces (carpet vs. wood) or frequency of vacuuming, for example.
2. S ll_ver _V a lJ_e^_Dat a
The Silver Valley study of 1.974 and 1975 covered 860 children ages
1-9 years living near the lead smelter in Kellogg, Idaho and surrounding
regions. PbB levels were quite high for children nearest the smelter.
Household dustiness, soil lead, and estimated or observed air lead concentrations
were obtained for 9 zones around the smelter and used to calculate daily
lead uptake levels. As for the Omaha comparison, dietary lead intakes
were calculated from 1970's data analyzed in the criteria document. The
geometric mean values by zone are shown in Table 6-3, along with predicted
mean PbB using the uptake/biokinetic model. Although comparisons in the
three control areas (V-VII) are limited by the fact that air lead values were
estimated there, results are generally adequate for the more distant
areas with lower exposure levels. This is the intended range of applicability
of the model. It is not surprising that blood leads are overpredicted in
areas I and II located closest to the smelter. The Harley and Kneip
compartmental model's presumption of linear kinetics in lead absorption,
-------
TABLE 6-3. CHILDREN'S BLOOD LEAD LEVELS MEASURED IN SILVER VALLEY, IDAHO 1974-1975 VS. INTEGRATED
UPTAKE/BIOKINETIC MODEL PREDICTIONS
Area I_
LL
ILL
IV
V
11
HI
Average Air Lead (pg/m3)
16.8
14.2
fi.6
3.0
0.73
0.5a
0.53
Average Soil Lead (ppm)
7,470
3,300
1,250
1,400
2,300
337
700
Average House Dust Lead (ppm)
11,700
10,300
2,400
3,300
3,400
1,800
3,900
Observed PbB (pg/dl)
2-yr. olds
72
51
36
35
35
25
35
Predicted Mean PbB^ (pg/dl)
2-yr. olds
88.5
74.1
28.4
33.7
34.6
22.3
35.8
Estimated Dietary Lead Uptake (pg/day) for
all sites during early 19701s (see text)
0-1 yr. olds 16.3 - 20.6
1-2 yr. olds 20.2 - 25.6
2-3 yr. olds 16.6 - 22.1
aAir lead concentrations in 3 control areas were estimated by ISC dispersion modeling.
^Predicted mean PbB represents midpoint of lower and upper hound means estimated hy model.
-------
VI—17
excretion, and accumulation, ufrile appropriate at relatively low to moderate
exposures, is not applicable for such extraordinarily high exposure levels
as observed in these two areas. Some adjustment for the non-linear patterns
of lead uptake would be necessary in order to reliably model PbP levels in
locales with such extremely high air, soil, and/or dust lead concentrations.
3. Other J)ata
Unpublished analyses by Phillips and Vornberg et al. (1986) for populations
living near a large industrial lead source in Herculaneum, Missouri have
similarly provided fairly good validation for the uptake/biokinetic model,
especially when measured soil and dust lead values were available.
C. Conclusions
The above discussion illustrated several validation exercises that were
undertaken to refine data inputs and to assess the reliability of the blood
lead uptake/biokinetic model. The East Helena analysis was by far'the
most intensive effort. The results of the analysis indicated that the model,
using the best available and location-specific input data, predicted overall
means essentially identical to the observed blood leads. When using less
location-specific input data (that is generally available for policy analyses),
the model predicts mean PbB levels within 2 percent of observed. Other
validation efforts, though undertaken with less effort, of data from
Omaha and Silver Valley also support the hypothesis that the uptake/biokinetic
model is reasonable for estimating blood leads when actual observations
are not available. Other independent validation efforts have also provided
general support for the efficacy of the model.
-------
VII. CONCLUSIONS AND APPLICATIONS OF LEAD EXPOSURE MODELING
Because of the pervasiveness of lead in the environment,- a large number
and variety of exposure situations require some understanding not only in
terms of present risks but for predicting hazards under future regulatory/
abatement alternatives. More than perhaps any other environmental con-
taminant, a wealth of information has accumulated not only on lead toxicity
but on lead exposure as well. This report has attempted to utilize the most
reliable of that information to develop multi-media exposure methodologies to
assess risks associated with alternative lead NAAQS and which can be applicahl
to other lead exposure scenarios. For example, blood lead levels in children
exposed to different drinking water concentrations have been estimated, in
EPA's recent review of the lead in drinking water regulation by adopting
both the disaggregate and the uptake/biokinetic modeling approaches.
Similar applications maybe appropriate for assessing the impacts of alternati
soil lead contamination/abatement scenarios (Cohen, 1988c).
Each of the three approaches described here has both advantages and
uncertainties. The aggregate model, for example, allows for a straight-
forward conversion of air lead concentration changes to the range of expected
impacts on children's blood lead levels through direct and indirect sources.
However, projecting in this approach what contribution non-air sources of
lead will make to future PbB levels requires fairly good data on current
exposures as well as trends in different exposure sources. Also, since soil
lead changes relatively slowly in response to decreases in air lead, the
change in blood lead calculated from the aggregate model may not be achieved
for a long time. The disaggregate model, with rapid change in dust lead
and slower change in soil lead, may be a more accurate predictor of near-term
effects of air lead changes on blood lead levels. The uptake/biokinetic
model, as with the other approaches, also must rely on incomplete
-------
VI1-2
data in some cases and because of its assumptions of linear absorption and
biokinetics, is limited to predicting low-moderate level exposures (i.e.,
< 25-30 pg/dl). The model's mathematical assumptions and numerical parameters,
however, combine plausible biological hypotheses, animal experimental data,
and results of observational studies. Further, it allows explicit projections
of future lead concentrations in various media and in turn can estimate
impacts of these different changes on different age groups of children.
It is this flexibility that makes the integrated uptake/biokinetic model
adaptable for a wide range of predictive exposure assessments and why it
was the focus of the validation exercises described in Section VI.
As discussed in Section I, the methodologies described in this report
will be used to conduct case study exposure analyses around various lead point
sources in the U.S. Blood lead distributions among young children and
middle-aged men living near these sources in 1996 will be estimated under
alternative lead NAAQS. Ouantitative.projections for pregnant women will
not be made. Children's PhB levels will be estimated using the uptake/biokinet
and aggregate models; adult men will be modeled using the disaggregate
approach. The range of GSDs (1.30-1.53) discussed in Section III.R will be
used for children to calculate PbB distributions around the predicted means
and results using the midpoint of 1.42 will be presented as best estimates;
a GSD of 1.37 will be used for adults. In addition to mean PbB levels,
percentages of the population above selected levels (e.g., 10, 15, 25
pg/dl) will be presented. Results of the case-study analyses will be
included in the staff paper to help develop the range of standards for the
Administrator to consider. Fetal exposures, although not quantified, will
be a major consideration. Uptake/biokinetic model results for children
will be used along with the adult men estimates in further quantitative
analyses to estimate potential monetary benefits of alternative standards.
-------
V 1 1-0
Results of the different validations indicate good concordance between,
observed and predicted average PbB levels in children living near lead point
sources. The model may also be a useful tool in estimating PbB levels in
children living with some other lead hazards, such as contaminated soils from
historical deposition near major urban roadways or closed smelters or mines.
Ongoing regulatory efforts by different components of EPA to control
concentrations of lead in air, water and soil have created a significant need
to model blood lead concentrations that delineates specific routes of lead
exposure. At present, the uptake/biokinetic model provides the best method to
achieve such specificity. Furthermore, such a model must be capable of simulating
blood lead levels under future as well as historically known exposure regimes.
(The validation efforts described in the previous section used historical
data to test and improve the predictive ability of the'model.) In fact, an
earlier but similar version of the uptake/biokinetic model has been applied
to assessments of Boston soil contamination (Beck and Tsai, 1987), exposures
around the Herculaneum, Missouri smelter (Phillips and Vornberg et al., 1986) -
and historical contamination of soils near the Bunker Hill smelter in Kellogg,
Idaho and as part of EPA's Integrated Environmental Management Project in
their efforts to develop and apply methodologies to establish public healtn
and environmental priorities.
Because the uptake/biokinetic model is linear, it cannot accurately predict
excessive exposure? (e.g., PbB levels above 3U pg/dl). Further development of
the model to incorporate non-linear absorption and biokinetic patterns is underway.
Other aspects of the model are still being assessed and validation tests will
continue as new data are available. In particular, adjustments of dirt ingestion
rate and dirt lead absorption may be desireable in calibrating the air, dust, and
soil lead slopes relative to observed regression models or to available soil
chemistry models (Cohen, 1988c). The empirical multiple regression methods used
-------
VI1-4
in the disaggregate models may also need refinement, e.g., by use of structural
equation analyses that extend the statistical methods used by the Cincinnati Lead
Program Project (Clark et al., 19B7).
-------
APPENDIX A. ESTIMATES OF LEAD UPTAKE
For each of the numbered lines in Table 4-1 in the integrated lead uptake/
biokinetic model, the assumptions and estimates used in calculating average
lead uptake for children from the various exposure pathways are discussed
in the correspond!" ngl y numbered paragraphs below. Table 4-1 presents
calculations for 2-year old children only; to estimate exposures throughout,
childhood, age-specific exposure parameters are necessary in several cases.
These are also described below.
1. Outdoor air lead: In the case-study exposure analyses of alternative,
future regulatory scenarios, ambient air lead concentrations will he assigned
to populations living in different locations, or "receptor points." Locations
will be defined as block group centroids - i.e., subdivision of census tracts.
Air lead levels will be generated around individual lead point sources through
use of the Industrial Source Complex-Long-Term (ISC-LT) dispersion model,
based on site-specific operating parameters, emissions (both stack and fugitive),
and meteorological data. Background lead concentrations from mobile sources,
re-entrained soil, and local, minor point sources will also be included. The
air lead concentrations specified in Table 4-1 do not represent exposure
scenarios for alternative lead NAAQS but starting points to illustrate how
daily lead uptake estimates are made.
2. Indoor air lead: The penetration of atmospheric lead into residential
structures depends on the size of the lead particles, meteorological conditions,
and the permeability of the windows, doors, and walls of the home. A range
of indoor/outdoor ratios has been found (0.3-0.8) for different cities and
structures (CD, Table 7-6). Near point sources viiere large airborne particles
are more prevalent and infiltration into homes is low, the ratio appears to
be closer to 0.3 (Cohen and Cohen, 1980).
-------
A-2
3. Time spent outdoors: The amount of time spent between indoor and
outdoor environments varies among young children depending on their stage
of development (i.e., infant, toddler, pre-school), season, geographical
location, and family behavior. While seasonal variations especial 1 y may be
important in explaining summertime blood lead peaks among many children,
yearly averages that smooth out such variations will be used for this modeling.
There are several reasons: 1) There is little data to quantify seasonal
differences in children's outdoor/indoor partitioning of time; 2) Regional
differences confound seasonal variations in outdoor/indoor patterns, for example,
a child in the South may spend more time outdoors in the fall and winter than in
the summer, in contrast to a child living in a northern climate, 3) Although
peak exposures are important, the model is being used to predict average
exposures over several years. Such estimates would not be expected to be
affected if seasonal differences were accounted for instead of integrating these
differences to produce yearly averages. A range.of 2-4 hours per day spent
outdoors (and therefore 20-22 hours spent indoors) is considered a reasonable
average for a 2-year old (CD, p. 7-43). The following age-specific estimated
ranges for hours spent outdoors were derived from a literature review
summarized in Pope (1985) using various studies (Hoffman et al., 1979; Rubinstein
et al., 1972; Suter, 1979; Koont.z and Robinson, 1982) and confirmed by
informal surveying of parents:
Li _0-l_ __l_-2_ _2~3_ __3-7
time spent "1-2 1-3 ~ 2-4 ~ 2-5
outdoors
(hrs/day):
4. Time weighted air lead concentrations for each level is estimated
by: [(outdoor concentration x time spent outdoors) + (indoor concentration
x time spent indoors)] 24 hours.
5. The volume of air breathed each day is dependent on age, body
size, lung capacity, altitude, and activity of the child. For instance,
-------
A-3
ventilatory volume can increase three-fold during strenuous exercise (Cotes,
1979). Phalen et al. (1985) determined average ventilation rates for males
and females from birth through age 18 from graphical fits of published
tabulated data (Altman and Dittmer, 1971, 1972). For example, a two-year
old at "low activity" is estimated to have a minute ventilation rate of
2.75 liters/minute, viiich corresponds to an average daily rate of approximately
4 m^/day while a three-year old is estimated to have a daily rate of 4.3
m^/day. Other estimates that appear in the literature are 4.7 m^/day
(ICRP, 1975) and 4 to 6 m^/day (Nutrition Foundation, 1982) for active
one-year olds, and 4.7 m^/day for a three-year old (Nutrition Foundation,
1982). These values combined with those determined by Phalen et al., and
scaling factors based on body size are used to construct the following
ranges for average age-specific daily ventilation rates.
Age (years): 0-1 1-2 2-3 3-4 4-5 5-6 6-7
Ventilation Rate
(m^/day): 2-3 3-5 4-5 4-5 5-7 5-7 6-8
6. The range of total lead intake by inhalation for each air lead
level is computed by multiplying both the estimated upper and lower bound
time-weighted average concentrations of air lead by upper and lower bound
estimates of the volume of air respired per day.
7. Respiratory deposition and absorption: Only a portion of inhaled
lead is deposited in the lungs and subsequently absorbed into the bloodstream.
The deposition efficiency of lead particles depends primarily on their size
and the physiology and rate of breathing of the individual. Available data
on lead particle size distributions, particle deposition patterns in the
lung, and respiratory absorption of lead particles were used to estimate
deposition efficiencies of airborne lead particles in young children
-------
A-4
(Cohen, 1987). A respiratory deposition/absorption rate of 25 to 45% is
calculated for young children living in non-point source areas, while a
rate of 42% is calculated for those living near point sources.
8. Total lead uptake from the air is the product of total intake
and the lung deposition/absorption factor.
9. Average dietary lead consumption: Given the wide spatial and temporal
distribution of food in this country, it appears reasonable to assume that
even people living near industrial lead sources receive roughly typical
lead levels from their diet. Any variability due to local contamination of
garden crops or kitchens near point sources can be factored in separately.
As noted in Section II.D., the Multiple Source Food Model developed in the
CD has been validated and updated using the most recent food data from 1984
and 1985 (Flegel et al., 1988). Further declines in dietary lead intake
are expected as a result of continuing reductions of lead in canned foods,
gasoline emissions, and lead in drinking water. Cohen (1988a,b), uses available
data on these downward trends and information from the Multiple Source
Food Model on food consumption patterns, lead content of various foods,
and source-specific contributions to project 1990-96 dietary lead intake
estimates for different age groups of children. The age-specific estimates
.• from Table 5 of Cohen (1988a,b) shown below as Table A-l will be incorporated
into the integrated uptake model to predict 1990-96 PbB levels among children.
Similar dietary lead intake estimates were made for 1978-83 for purposes of
validating the uptake/biokinetic model (See Section VI). These are shown in
Table A-2.
10. Gut absorption of dietary lead: Only a portion of ingested lead
is absorbed into the bloodstream from the gastrointestinal (GI) tract,
or gut, and is dependent upon .the composition of the diet and physiological
-------
A-5
TABLE A-l. AGE-SPECIFIC ESTIMATES OF TOTAL DIETARY LEAD INTAKE (pg/day) FOR 1990-1996
"Metal 1ic"
(years)
(e.g. solder)
Atmospheric
Other
Total
<1
3.4
0.8
3.3
7.5
1
4.0
1.1
3.8
8.9
2
5.6
1.2
3.6
10.4
3
5.8
1.2
3.7
10.7
4
5.9
1.1
3.8
10.8
5
6.1
1.2
4.0
11.3
6
6.3
1.3
4.3
11.9
Source: Cohen (1988a,b)
TABLE A-2. TOTAL DIETARY LEAD INTAKE (ug/day) FOR 1978-1983 FOR FIVE AGE
GROUPS OF CHILDREN3
Age (years)
1978
1979
1980
1981
1982
1983
1-2
45.8
41.2
31.4
28.8
26.0
19.3
2-3
52.9
48.0
36.9
33.8
30.6
24.1
3-4
52.7
47.8
36.9
33.7
30.6
23.0
4-5
52.7
47.8
36.9
33.8
30.7
22.0
5-6
55.6
50.3
38.7
35.5
32.2
23.2
aFrom Sledge (1986) — calculated using year-specific FDA data on food lead
content and Multiple Source Food Modeling methodology described in
Chapter 7 of CD. Assumes average drinking water lead concentration at the
tap of 12 ng/1. Used in validation exercises described in Section VI.
-------
A-6
status of the individual. Based on measurements, mainly among infants, of
lead in the diet and excreta, a gut absorption rate for ingested dietary
lead between 42 and 53% has been calculated (Alexander et al., 1973; Ziegler
et al ., 1978). Correspond!"ng rates between 7 and 15% have been estimated
from adult studies (Kehoe, 1961; Chamberlain et al., 1978; Rabinowitz
et al ., 1980). Following the approach used by Harley and Kneip (1985),
these rates were assumed for infants and adults and smoothed to yield
values for intermediate ages, as follows:
Age: 0-1 1-2 2-3 3-4 4-5 5-6 6-7
Average GI 42-53 . 42-53 30-40 30-40 30-40 30-40 18-24
Absorption Rate (%):
These rates do not reflect the wide degree of inter-subject variability
observed in the studies nor other factors that influence absorption in
children. (Such variability will be accounted for in calculating population
distributions of PbB levels around estimated average PbB levels using empirically-
derived geometric standard deviations). For example, absorption rates 2 to 4
times higher were observed in adults following fasting periods of 4 to 16
hours (Blake, 1976; Chamberlain et al., 1978, Heard and Chamberlain, 1982;
Rabinowitz et al., 1980). Such increases can be expected among children who
skip meals, a particular problem among lower income groups (Koh andCaples,
1977). Regular eating patterns do not ensure minimized dietary lead absorption,
however. Based on several animal studies, clinical investigations, and
epidemiological surveys, diets deficient in calcium, iron, phosphate, zinc,
copper, vitamin D or protein, and excesses of dietary lipid or lactose, can
be expected to increase the absorption and retention of lead (CD, Table
10-4). Data available from the 1976-1980 NHANES II indicate that as many as
22% of children aged 3-5 may have some form of iron deficiency (CD, p. 13-48).
-------
A-7
This phenomenon nay partially explain the enhanced neurotoxicity of
lead in animals with nutritional deficits (Mahaffey and Michael son, 1980).
The importance of nutritional interactions with lead absorption and possibly
toxicity is particularly significant for young children because of their large
fluxes in relative nutrient status. Although nutritional deficiencies are
more pronounced among lower income children, they exist in children of all
socio-economic strata (CD, p. 10-41).
11. Daily dietary lead uptake is obtained by multiplying rows 10 and 9.
12. and 13. Dust/soil concentrations: As discussed in Section II, the
accumulation of lead in street and household dusts and soils appears to be
directly related to the volume of traffic, and inversely related to distance
from neighborhood streets and roads, distance from lead based painted and brick
houses and buildings, and distance from lead point sources. Predicting a
relationship between different air lead levels and dust and soil lead
levels over time would require the inclusion of many complex variables such
as deposition rates, chemical and physical characteristics of the lead particles
and soils, topographic and meteorological conditions, frequency of street
washings and precipitation, background dust concentrations, and information
on transport of dust and soil into homes and buildings. Given current data,
there would be an extremely large amount of uncertainty surrounding anyone
of these variables for different locations. To predict outdoor and indoor
dust concentrations under alternative air lead concentrations, reliance is
placed on available studies that include measurements of both air levels and
dust and/or surface soil concentrations (see Table A-3).
Because of historical accumulations of relatively large lead particles
near primary and secondary lead smelters and other point sources, outdoor
soil and dust lead concentrations, and consequently potential exposures,
are significantly greater in these areas, regardless of ongoing emission
-------
A-8
TABLE A-3. SUMMARY OF ENVIRONMENTAL LEAD MEASUREMENTS FROM VARIOUS POINT-SOURCE LOCATIONS
Average
Air Lead
(uq/m3)
Average Surface
Soil/Outdoor
Dust Lead3
(uq/g)
Average Housedust
Lead
(yg/g)
Location/Reference
"Post-control"b (1973-1976)
measurements in Omaha
neighborhoods:
0.22
80
215
suburban
0.04
. 91
162
0.32
591
280
mixed (battery plant in resi-
dential neighborhood)
0.46
260
470
commerci al
(Angle and Mclntire, 1979;
Anqle, 1985)
0.30
114
-
"Post-control"b measurements (1976)
0.45
112
-
in rural area, Brussels, and near
0.8
466
-
lead smelter, Belgium (Roels et
3.67
2.560
-
al.. iy80)
0.41
690 (street dust)
1,239*
Near Arnheim secondary lead
240.(soil)
smelter, Holland (Brunkreef et
al., 1981; Diemel et al., 1981)
0.82
924
713
Near Toronto secondary lead smelters
3.01
2.416
1.550
and in city (Roberts et al., 1974)
- 2.0C
1,214 (<1.6 km)*
-
Various distances from a lead-zinc
698 (1.6 - 3.2 km)
*
smelter in British Columbia
253 (>3.2 km)*
-
(Schmitt et al., 1979)
1.6
-
1,825*
Near lead smelter in Yugoslovia
2.0
-
1.900*
(Prpic-Majic et al., 1984)
Distance range (km) from
zinc or copper smelters in:
0.13
35*
241*
(3.5 - 24.0)
0.20
243*
409*
(1.3 - 3.7) Bartlesville, OK;
0.30
829*
386*
(0.8 - 4.3)
0.31
821*
441*
(0.8 - 1.5)
0.14
75
235
(10.0 - 26.0)
0.18
115
164
( 3.5 - 21.0) Anaconda, MT;
0.09
294
210
( 2.0 - 11.0)
0.26
424
398
( 2.0 - 3.5)
0.09
58
75
( 3.4 - 68.0)
0.11
65
60
( 1.0 - 6.4) Ajo, AZ;
0.1J
77
65
( 0.5 - 2.3)
0.26
95
116
( 0.5 - 1.3)
0.36
532
263
(11.0 - 26.U)
0.56
117
201
( 5.4 - 14.5) Palmerton, PA
0.13
326
198
( 3.3 - 9.9)
0.28
331
438
( 0.3 - 2.8)
(Hartwell et al., 1983)
-------
A-9
TABLE A-3.
SUMMARY OF ENVIRONMENTAL LEAD MEASUREMENTS FROM VARIOUS POINT-SOURCE LOCATIONS
(Continued)
Average Surface
Average
Soil/Outdoor
Average Housedust
Air Lead
Oust Lead3
Lead
(yq/m3)
(UQ/q)
(UQ/q)
Location/Reference
Near El Paso primary lead smelter
1.3
427
1,479
"Post-control"b
2.7^
948e
8,623e
"Pre-control" (Landrigan et al.,
1975; Morse et al.. 1979)
0.5
337
1,800
Near Silver Valley (Kellogg), Idaho,
0.5
700
3,900
primary lead smelter; Pre-controlb
0.7
2,300
3,400
Yankel el al., 1977; Idaho Dept. He al11
3.0
1,400
3,300
and Welfare, 1977)
6.6
1,250*
2,400*
14.2
3,300*
10,300*
16.8
7,470*
11.700*
0.3
200
501
Near E. Helena Montana primary lead
0.8
158
398
smelter (CDC, 1983)
1.9
307
891
2.9
1,345
1,585
3.6
1,549
2.284
0.3
157
170
Near Herculaneum, Missouri, primary
0.8
1,822
2,040
lead smelter (Phillips and
0.3
70
850
Vornbery, 1986; Vornberg, 1987)
0.5
183
1,030
0.8
2,239
1,210
0.8
.508
975
2.2
2,558
1,610
0.8
148
630
1.1
827 .
1,600
2.8
1,458
2.080
*Oata not used to develop airtsoil:dust Pb relationships in Tables A-4 and A-5; See Text.
aSurface soil measurements represent top 0 - 2.5 cm.
b"Post-control" refers to the period of time following the application or increase in emission
controls at the lead point source(s) involved in the particular study. Conversely, "pre-
control" refers to the period of time before the application or increase in emission controls.
cAnnual mean concentration reported as an approximate value
^Geometric annual mean of air lead levels measured within 5 km of smelter
eGeometric mean of dust and soil lead levels within 6.4 km of smelter
-------
A-10
controls. This is the main reason why separate exposure estimates for individuals
living near point sources under alternative air lead levels are required.
Several data points collected from non-point source areas are excluded from
Table A-3 and from subsequent analyses. These data include those collected
near roadways in English and U.S. cities dominated, at that time, by automotive
emissions, and in some cases, significantly impacted by old lead-painted
homes (e.g., Barltrop et al., 1974; Lepow et al., 1975; Davies et al., 1987).
Only data collected near lead point sources where emissions were comparable to
current situations are used to develop outdoor soil/dust lead levels for
point sources meeting different air lead concentrations (see Appendix R).
The techniques used in the sampling and pre-analytical treatment of dusts and
soils influence the lead concentrations found. In order to make comparisons
between different studies possible, analyses are limited to those studies that
collected only the top 1-5 cm of soils (the most relevant depth for childhood
exposure), collected dusts using surface'wipes or vacuum pump, and employed
careful* analytical quality control. Although there are inevitable uncertainties
in relying on a variety of studies, an attempt has been made to use the best
available data to derive plausible relationships between air, soil, and dust
lead concentrations.
Several of the studies besides the "non-point source" studies for
vvhich data are not used to develop relationships require comment: a) The
Idaho smelter studied by Yankel et al. (1977) operated for several years
with severely limited air pollution control capacity due to a baghouse fire
in 1973. The very high soil and dust lead concentrations measured near the
smelter would not be expected today with normal 1y control 1ed emissions;
b) Similarly, the pre-control measurements reported for the El Paso smelter
are not relevant to operating facilities today. In addition, the air lead
-------
A-ll
average for the pre-control period represented a different geographical area
than represented by the soil and dust measurements; c) The high indoor dust
lead levels measured by Brunekreef et al. (1981) and Diemel et al. (1981)
near the Belgian smelter were probably due, in part, to extremely high lead
content (up to 40% by weight) in peeling paint in old houses (built around
1900; the children in these houses were excluded from analysis of blood lead
levels); d) The relatively high soil and dust lead levels in relation to
concurrent air lead concentration measured near the Bartlesville smelter by
Hartwell et al. (1983) was probably due to a drastic reduction of total
suspended particulate emissions, from 1600 tons/year to 15 tons/year, just
prior to the onset of the study; e) The air lead level reported by Schmitt
et al. (1979) for the smelter area in British Columbia was only an
approximation; f) the protocol for dust lead collection in the Yugoslovian
study (Prpic-Majic et al.) could not be verified.
Another data set excluded from analysis is the 1983 air, soil and dust
lead measurements taken around the Bunker Hill smelter complex in Kellogg, Idaho.
These data were not used because the smelter ceased operation in 1981 and current
soil and dust lead levels are still declining and thus do not represent steady
state conditions.
With these exceptions, the studies used in Table A-3 appear to have
sampled a broad spectrum of homes (e.g., both low and middle class, old and
modern, with and without leaded paint—those with high paint lead levels were
excluded) and neighborhoods that can be considered fairly representative of
current U.S. conditions in point source areas. Emphasis is placed on data
collected near operating primary and secondary lead smelters. Other lead
sources were also used, such as zinc and copper smelters, that have dispersion
patterns similar to the lead smelters, but at lower levels. The studies used
-------
A-12
for analysis had average air lead levels thought to be representative of each
defined study area, or "neighborhood." The average soil and dust lead values
are thought to characterize average exposures for children who spend time
at several households or play areas within each "neighborhood." Although
far from conclusive, the available data from point source studies suggest
the long-term equilibrium relationships between air lead (PbA) and soil/dust
lead concentrations shown in Table A-4. Appendix R contains a description
of the derivation of these equations.
Short-term (several months) departures from long run equilibrium
levels of soil and dust lead (from imposition of new alternative lead NAAOS,
for example) are also estimated in Appendix B. The short-term relationships
are shown in Table A-5. In general, a change in ambient lead will in the
short term, only induce direct changes in dust lead. In the long term, the
relationships in Table A-4 are appropriate. Unfortunately, insufficient
longitudinal data exists to explicitly estimate the temporal relationship
between ambient, soil, and dust lead.
The ranges in lines 12 and 13 are means using the long-term relationships
in Table A-4.
14. It is estimated that a young child typically sleeps about 12 hours
a day (Pope, 1986b), leaving approximately 12 waking hours in which he or she
is capable of ingesting dirt. The time weighted concentration of lead in
dust and soil that a child is exposed to is thus computed by: [(outdoor
soil/dust lead concentration x time spent outdoors) + (indoor dust lead
concentration x time spent indoors)] 12 hours.
15. Amount of dirt ingested: Hand to mouth activity (e.g., thumb
sucking and finger licking) and immature dietary habits (i.e., the retrieval
and subsequent consumption of food from dusty surfaces or soil), which are
-------
A-13
Table A-4. LONG-TERM RELATIONSHIPS RETWEEN DUST, SOIL AND AMBIENT AIR LEAD*
Soil Lead = 53 + 510 (PbA)
Dust Lead = 60 + 844 (PbA)
Table A-5. ESTIMATING SHORT-TERM* RESPONSES IN EQUILIBRIUM SOIL AND DUST
LEAD LEVELS FROM CHANGES IN AMBIENT AIR LEAD*
Change in Dust Lead = 638 (Change in PbA)
Change in Soil Lead = No Change
*Derivation described in Appendix B.
1 Duration of several months or less.
-------
A-14
normal behavorial characteristics of children up to five years of age
(Lin-Fu, 1972), make soil and dust major sources of ingested lead for
children (Charney et al., 1983). Good information on the amount of dirt a
child eats in the normal course of a day is needed for an assessment of
health risks associated with toxic materials in the environment. Rased on
a small number of direct measurements, Day et al. (1975) estimated that
under average urban conditions (and after 30 minutes of normal playground
activity), 5 to 50 mg of dirt transferred from a child's hands to a typical
"sticky sweet", and estimated that a daily intake of 2-20 sweets would result
in a dirt intake of 10-1000 mg. Lepow et al., (1974) measured a mean of
10 (jg of dirt on the hands of 22 young children which would be ingested with
each episode of hand-to-mouth activity. These authors estimated that a
child puts its hand into its mouth ten times per day which would result in
the ingestion of 100 mg (0.1 g) of dust and soil per day. An average
estimate of 100 mg of dirt ingested daily by young children has been used
in several documents (Drill et al., 1979; NAS, 1980; CH, Table 7-23) to
represent a probable value for a "typical" child.
Recently, Binder et al. (1986) and Clausing et al. (1987) estimated
children's intake of soil by measuring relatively non-absorbed elements
with high concentrations in soil as tracers. Levels of aluminum, titanium,
and either silicon or acid insoluble residue were determined in soil and
children's feces and mass balance calculations were made to estimate
soil ingestion. In both studies, soil ingestion estimates based on titanium
were highly variable and in many cases, an order of magnitude greater than
results using the other tracers. This suggests an additional source of
titanium not accounted for in the studies (e.g., paint, toothpaste, talc,
laboratory contamination) and that the titanium-based, outlying estimate
should be given less weight.
-------
A-15
For 59 children, ages 1 through 3 years, Binder et al. estimated
mean daily soil ingestion based on aluminum and silicon to be 181 and 184
mg/day respectively; geometric means for the two elements were 128 and 130
mg/day. The mean estimate based on titanium was 1834 mg/day. Of the various
factors and assumptions used in this study that could have contributed to
an inaccurate estimate of soil ingestion (e.g., assumption that the tracer
element absorption is negligible, and that there is introduction or loss
of tracer element during processing), the assumption with the most significant
potential impact is that dietary intake of the tracer elements is negligible.
Binder et al. and Sedman (1987) cite evidence that these elements do in fact
occur in the diet. The study design used by Clausing et al . provides an
indirect control for such dietary intake; 18 nursery school children and 6
hospitalized children, without soil contact, ages 2-4, were sampled. The
average of estimated soil ingestions based on al1 3 tracers was 105 mg/day
for the nursery school children and 49 mg/day for hospitalized children nfnch
was significantly different despite the small number of samples. If as the
authors assumed, all of the "soil ingestion" by hospitalized children is
background due to dietary and other non-soil sources, correction of the
nursery school average for this background would result in an estimated
average soil ingestion rate of 56 mg/day. If the same background is
subtracted from the Binder et al. estimates to adjust for dietary sources of
tracer elements, soil ingestion rates of approximately 80-135 mg/day result.
For children in the 1-4 year old age group studied by Binder and Clausing,
a range of soil ingestion between 80 and 135 mg/day will be assumed.
As noted by Binder, Clausing, Sedman, and colleagues, considerable
uncertainty surround these estimates given for example, the small number
of subjects studied, the lack of complete data on dietary intakes and
-------
A-16
gut absorption, and the uncertain representativeness of Dutch nursery school
children to U.S. children. Nevertheless, the range of the estimates is
narrow despite being derived from separate data sets and does suggest that
they reasonably reflect average soil ingestion for this age group, fiiven
that children eat a combination of soil and dust, and that the studies did
not distinguish between those two media, the estimates would be inaccurate
if concentrations of silicon and aluminum were different in house dust and
soil. Until further data are available, it will be assumed that soil ingestion
estimates from these studies reflect total "dirt" consumption rates.
To use the above estimates to determine dirt ingestion rates for other
ages of children, data on age-related changes in blood lead (Yankel et al.,
1977; Annest and Mahaffey, 1984; Billick, 1982; Quah et al., 1982) and the
prevalence of mouthing behavior (Millican et al., 1962; Barltrop, 1966) were
applied by Sedman (1987). Relative age-related changes in soil ingestion calculated
by Sedman were applied to the range derived above for the 1-4 year old age-group
to yield the following estimates:
Age (years) 0-1 1-2 2-3 3-4 4-5 5-6 6-7
Dirt Ingestion 0-85 80-135 80-135 80-135 70-100 60-90 55-85
(mg/day)
The lower bound of zero for the youngest age group was chosen because contact
with dirty surfaces is limited for the first 6 months or so until crawling
begins. In informal discussions with clinicians and other researchers,
including those who are involved in ongoing studies of soil ingestion in
children, there was general agreement that these estimated ranges are
reasonable averages given the available data and associated uncertainties.
Further research on this issue that includes specific examination of tracer
element metabolism and dietary intake, and house dust ingestion, has been
-------
A—17
stimulated by the Binder and Clausing studies. Preliminary results from this
ongoing research that can hopefully clarify the estimates presented here
are expected sometime in early 1990.
The above ranges representing average exposure estimates are derived
from studies that did not include children with pica. Dirt ingestion rates
for many children can be much higher. For example, among young children
hospitalized for asymptomatic lead poisoning and exposed to three play
environments which differed in their stimulus complexity (e.g., number and
type of toys, availability of playmate), significantly more mouthing behavior
occurred in the impoverished play setting (Madden et al., 1980). In particular,
children with pica who display patterns of repetitive hand-to-mouth
activity, or who deliberately ingest paint, plaster, paper and other non-food
items including dirt, may be exposed to considerable amounts of lead compared
to children wfco only inadvertantly ingest foreign substances. Pica occurs
to some degree in a substantial percentage of young children. Data from
NHANES II indicate that the percent of children with a history of pica is
significantly higher for those 6 months through 3 .years old (11.0%) than
for those 4 through 5 years old (3.2%) and for children living in households
with annual family incomes < $10,000 (11.9%) than for those in households
with incomes > $10,000 (6.0%) (Mahaffey and Annest, 1985). Estimates for
%
pica prevalence rates, based on more limited samples, range as high as
approximately 10 to 30% in children 1 to 6 years old to 35-50% in those 1
to 3 years old (Millican et al., 1962; Rarltrop, 1966). Children exhibiting
pica for paint are of major concern because of the high levels of lead in
some paints, with older painted surfaces containing lead in concentrations
greater than 1-10 percent (10,000-100,000 ppm). [The allovable lead content
of paint was set in 1978 by the Consumer Product Safety Commission at 0.06
percent lead (600 ppm)]. Pica for paint is believed to occur in episodes,
-------
A-18
possibly 2 to 3 times per week (NAS, 1972). It has been estimated that a
child can consume somewhat greater than 1 gram within a 24 to 36 hour
period, with cases reported of up to 20 grams within the same time period
(Sachs, 1975) and that children with pica for paint may consume 1 to 3
grams per week (NAS, 1972). Although direct data on pica children are not
yet available, a reasonable "worst case" estimate of 1 gram/day for soil
ingestion in young children has been suggested (White, 1987).
Estimates of daily lead uptake under alternative air lead levels are
presented in Table 4-1 only for children without pica. To calculate the
contribution of lead in paint to the total lead exposure of children with
pica, a worst-case estimate of 1 gram of paint chips consumed per day could
be used. Future reductions in airborne lead emissions are not likely to
significantly alter pica exposure to lead over the next few years and other
regulatory alternatives to protect these children must be considered.
16. Lead intake from dust and soil is computed by multiplying the
time weighted concentration of indoor and outdoor soil/dust concentrations
by the estimated dirt ingestion rates listed above.
17. Gut absorption of dirt: Animal experiments indicate that lead
of variable chemical forms in soil or dust is as available for absorption as
food lead (Dacre and Ter Haar, 1977) and i_n vitro studies demonstrate that the
acidity of the human stomach is adequate to extensively solubilize lead
assimilated from soil and dust (Day et al., 1979; Harrison, 1979; Duggan and
Williams, 1977). Based on these data and the fact that ingestion of such
materials occurs other than at mealtimes, allowing for potentially enhanced
absorption, the CD estimates that 30% of the lead ingested in dust and soil
is absorbed in a child (CD, p. 10-10).
As discussed in the CD, the relationship between blood lead and lead
intake is curvilinear across a broad range of blood lead values such that
-------
A—19
the relative change in blood lead becomes smaller as "baseline" blood lead or
lead intake increases (CD, p. 10-53, section 11.4.3). In the lower range
of exposures, there is no significant difference between curvilinear and linear
relationships (CD, p. 11-99, p. 11-134). A number of biological factors may
explain the curvilinear relationships such as an increasing fraction of blood
lead as plasma lead when circulating lead rises, with greater movement of
plasma lead to tissues, including bone, and increased lead excretion.
Another plausible basis is gradual saturation of lead transport proteins in
the GI tract as lead intake rises (CD, p. 10-13). In fact, improved fits
of the uptake/ biokinetic model estimates to measured blood lead data from
E. Helena resulted when GI absorption rates in children were differentiated
based on proximity to the smelter vdiich is a surrogate for lead intake
level. The best fits resulted when a GI absorption rate of 0.3, equivalent
to the average rate cited in the CD for children, was assumed for children
living beyond a one-mile radius from the smelter, and a rate of 0.2 was
assumed for children living within one mile. A close fit between observed
and predicted blood lead levels also resulted when an average rate of 25%
was used for the entire area (within 2.25 miles) of the smelter. It is of
interest to note that a GI absorption factor of 0.17 has been estimated for
paint chip ingestion in children (CD, p. 10-10). For children above 6 years
old, a slightly lower value is needed to reflect the fact that GI absorption
efficiency generally is much lower in adults. For children ages 6-7 years of
age, a GI factor of 0.15 and 0.2 will be assumed depending, as discussed
above for younger children, on their proximity to a lead point source.
An inverse relationship was found between particle size and gastro-
intestinal absorption of lead, especially in the range of 1-100 pm, such
that a 6 pm dietary lead particle was absorbed five times as efficiently
as a 197 pm particle (Barltrop and Meek, 1979). There is little information
-------
A- 20
to assess whether the conversion in the physical form of atmospheric lead-
containing particles once inside the complex yeochemical matrix of soil or
as dust would affect their bioavailability and, consequently, toxicity.
Thus, atmospheric lead particles that have deposited will be considered to
have absorption rates if ingested, independent of particle size.
18. Total lead uptake from dust and soil is obtained by multiplying
rows 16 and 17.
19. Total lead uptake is the sum of rows 8, 11, and 18.
Paint Lead Exposure Estimates
No attempt has been made to separate lead paint hazards from generalized
conditions in estimating soil and dust lead concentrations associated with
different air lead levels. The major reasons are that 1) any changes in
the lead NAAQS will have minimal effects on children exposed to dusts and
soils contaminated by the flaking, peeling* weathering or "powdering" of
lead-based paint and 2) data are not available to adequately estimate lead
concentrations resulting from lead paint contamination given their high
variability depending on many factors including housing age, extent of
deterioration, layers of paint, family behaviors, and climate. For purposes
of simply illustrating the magnitude of risks associated with lead-paint
hazards in relation to generalized exposures around point sources or in
other areas, one set of average estimates of lead levels in dusts and
soils in and around homes with lead paint, presented in the CD, can be used.
Based on studies of Hardy et al. (1971) and Ter Haar and Aronow (1974),
the CD states that soil lead and household dust lead concentrations around
and in lead-painted homes can be expected to average 2000 ppm (p. 7-62).
If this soil and dust lead level is entered into the exposure profile for
children in Table 4-1, assuming an air lead level of 0.5 pg/nr*, for example,
-------
A- 21
and a gut absorption rate of U.17 cited in the CD for paint chips (CO,
p. 10-10), ranges of total lead uptake would increase from 11.5-26.7 pg/day
to 31.2-51.1 wg/day. Even this large difference markedly underestimates
leaded paint exposure for many children since children who eat flaking
paint or gnaw lead-painted woodwork are not accounted for. Assuming pica
behavior (1 gram/day ingested rate, see #15 above) in lead-painted housing
conditions with soil and dust lead levels of 2000 ppm, the daily lead
uptake estimates cited above increase to approximately 350 My/day.
-------
APPENDIX B. ESTIMATION OF THE RELATIONSHIPS BETVEEN SOIL, nuST
AND AMBIENT LEAD
A key assumption in regulation of atmospheric lead concentrations
(PbA) is that changes in PbA will be followed by correspond!'ng changes in
soil lead concentration (PbS) and interior house dust lead concentration
(PbD) in the same vicinity. Since ingestion of soil and dust lead during
normal hand-mouth activity is believed to be the major source of lead
uptake in very young children (and worse when pica is present) it is
important to make a quantitative estimate of the magnitude of the effects
on these proximate causes of childhood lead exposure due to changes in PbA.
1. Mathei^ti_cal_J^q^del_s _far_SoiJ__and_Dust_Lea^d_vs^_Ai_r _Lead
The relationships depend on many factors, including time (t) and
physical or chemical properties of the atmospheric lead particulates
such as particle size and surface water solubility of the particulate
lead matrix. We let x denote a generic set of particle properties. The
fraction of airborne particles deposited on soil, per'day, denoted DF(x,t),
is in general a function of x and t. So is the fractional rate of removal
of surface soil lead, denoted RF(x,t), by burial, runoff, resuspension,
cleaning and other activities, and the influx rate (pg/g per day surface
PbS from non-air sources) denoted IF(x,t). Finally, in order to calculate
the Pb concentration in the atmospheric particulates, we also need the
atmospheric particulate matter concentration in particle class x at time
t, denoted PM(x,t). A formal model for combining these quantities is
described next.
A plausible mathematical model for the rate of change of PhS(x,t)
can be expressed by a differential equation
Rate of change of PbS = non-air influx + atmospheric deposition - soil removal
-------
8-2
or, symbolically,
dPbS(x,t)/dt = IF(x,t) + [PbA(x,t)/PM(x,t)] DF(x,t)
- PbS(x,t) RF(x,t) (Equation 1.1)
An equilibrium model may be obtained by setting dPbS/dt = U and
solving the resulting equation (suppressing dependence on t),
PbS(x) = Aq(x) + Ai(x) PbA(x) (Equation 1.2)
where
Ao(x) = IF(x) / RF(x) (Equation 1.3)
Ai(x) = DF(x) / [RF(x) PM(x)] (Equation 1.4)
Note that this implies that when there is more unleaded dust in the air
(larger PM(x)), then the soil lead vs. air lead slope Ai is lower.
Likewise, a non-equilibrium model can be derived from equation. 1.1. In
the case when all of the parameters are independent of time, we have a
relatively simple solution in terms of an exponential approach to the
equilibrium concentration PbS(x) in Equation 1.2,
PbS(x,t) = PbS(x,0) exp(-RF(x)t) +
(1 - exp(-RF(x)t) PbS(x) (Equation 1.5)
There may be substantial variation in the apparent soil lead-air lead
"slope" where PbS has not had sufficient time to reach near-equilibrium,
say t < 3/RF(x). Thus the soil lead removal time scale l/RF(x) is a
critical parameter.
-------
B-3
Another consequence of this model is that particle properties (x)
may be a factor in producing an apparently nonlinear relationship between
total soil lead concentration PbS
PbS = Ix PbS(x)
and total air lead concentration PbA
PbA = Ix PbA(x).
From equation 1.2, the equilibrium relation is
PbS = £x Aq(x) + Ix Ai(x) PbA(x) (Equation 1.6)
This produces an exactly linear relationship
PbS = ao + ai PbA . (Equation 1.7)
when
Ai(x) 3 DF(x) / [RF(x) PM(x)] = ai (Equation 1.8)
a0 = ExIF(x) / RF(x)
The assumption that A^(x) is a constant at each site, i.e., does not depend
on particle properties x, is plausible but untested. Empirically, the
relationship between PbS and PbA is nearly linear at low levels of PbA,
but may be somewhat nonlinear at higher PbA (McLamb, 1988; Marcus 1988b)
It is more plausible that PbS(x) is linear with the average concentration
of lead in the particles, as in Equation 1.4 so that a linear relation
between PbS and PbA is, at best, an approximation, unless all particles
are in a single x class.
Similar but more complex calculations may be used to derive a linear
PbO vs. PbA relationship.
2. Review of Data in The Literature
A number of data sets were examined that could be used to assess the
rate at which soil lead (PbS) and/or house dust lead (PbD) concentrations
change over time as a consequence of changes in air lead concentration (PbA).
-------
B-4
Most of the reported studies are of marginal value. We will first describe
the data sets we found less useful.
The Omaha studies by Angle et al. (1984) contain useful annual
data, but the spatial and temporal averaging is at a very broad scale and
much detail is lost. The lead dustfall data (PbOF, mg/cm^) was available
for most of the years in which PbA averages are yiven, but PbD and PbS
were only measured at 37 locations in 1972. The data are shown in Table
B-l. No relation between PbDF and PbA is evident.
The Roels et al. (1978, 1980) data also measured PbA for a number of
years, but PbD only for 1976, so is not useful for showing PbD changes.
The El Paso smelter data (Morse et al., 1979; Landrigan et al.,
1975; Landrigan and Baker, 1981) measured PbA, PbD, and PbS at several
locations in 1972 and 1977. In 1972 many of the most heavily lead-burdened
children lived in the Smeltertown neighborhood adjacent to the smelter.
By 1977 almost all children had been evacuated to safer neighborhoods.
In the intervening five years the mean annual air lead levels (PbA) had
declined from 10.0 to 5.b yy/m^ at 0.4 km from the smelter. The dust
samples in 1972 were collected in Smeltertown and other locations, with
exceptionally high geometric mean Pbl) value of 22,191 ppm in the ring 0
to 1.6 km from the smelter. The 1977 mean PbD was only 1,479 ppm in this
ring, and 1,461 ppm in the ring 0.8 to 1.6 km from the smelter. Likewise,
the 1972 soil lead mean was 1,791 ppm in the 0 - 1.6 km ring, and 427 ppm
in 1977.
The decline in El Paso PbS and PbD is proportionately much greater
from 1972 to 1977 than the decline in PbA. Several factors may play a
role here. The first is that the extensive publicity given to lead as a
health hazard probably caused much more extensive individual attention to
-------
B-5
TABLE B-l
ENVIRONMENTAL LEAD CONCENTRATIONS IN
OMAHA, NEBRASKA (ANGLE ET AL., 1984)
Area Year PbA PbDF PbD PbS
1970
1.66
-
-
-
1972
0.37
-
479
262
1973
0.46
26 •
-
-
1974
0.04
17
-
-
1976
0.13
5
-
-
1977
0.78
34
-
-
1970
1.44
_
1972
0.32
-
300
339
1973
0.32
11
-
-
1974
-
6
-
-
1976
0.16
1
-
-
1977
U.62
8
-
-
1970
0.73
__
1972
0.26
-
211
81
1973
0.22
7
-
-
1974
0,04
3
-
-
1976
-
-
-
-
1977
-
-
-
_
-------
B-6
household cleanliness and to control of areas in which children could be
exposed to leaded soil (Landrigan, personal communication, April 5, 1988).
Secondly, we do not know if the same locations were sampled in 1972 and
1977. Interior household dust collection would have been negligible in
Smeltertown in 1977 as only three children lived there. Finally, the air
lead levels relevant to the highest PbD in 1972 may have been much larger
than 10.0 yg/m^ in 1972. Landrigan .et al. (1975).reported a mean PbA of 92 ug/m^
at the air monitor closest to the smelter. A PbD deline from 22,191 to
1,479 ppm (15-fold) is better explained by a PbA decline from 92 to 5.b ug/m^
(17-fold). Access to unpublished data may clarify this.
The NEA source apportionment study (Cooper et al., 1981) obtained a
number of low-vol, high-vol, and dichotomous sampler TSP samples at
locations in the vicinity of the Kellogg, Idaho, smelter during 1980,
when the smelter was-still in operation. Samples were also obtained of
"aerosolizable dust." These data are not directly comparable to PbS
values. The samples of soil and road dust were sieved, aerosolized, and
samples with a dichotomous sampler. This fine surface soil fraction
probably is more relevant to the exterior dusts that adhere to a child's
fingers and are eaten (Duggan and Inskip, 1985). Since a number of
samples were obtained after the area had been covered by fine volcanic
ash from the eruption of Mt. St. Helens on May 18, 1980, "Soil samples
were collected after scraping away the overburden of ash which probably
also removed the normal aerosolizable dust layer. Special care was
exercised to ensure that the ash did not contaminate the underlying soil
when sampled. The top one or two centimeters of soil were collected by
scraping with a spatula..." (Copper et al., 1981, pp. 114-115). This
study appears to be of marginal relevance for studying time trends.
-------
B-7
3. Estimating Soil and Dust Lead Levels From Ambient Lead Levels
The assumption in most analyses of data around existing lead point
sources is that the environmental lead emissions have been nearly
constant for a sufficiently long time that lead levels in soil and dust
are nearly in dynamic equilibrium. Thus, for some parameters ao» ai, bo,
bi, b2, co, ci, we have the following linear relations for predicted
geometric means depending on the level of available information:
Predicting PbS when PbA is available:
G.M. PbS = ao + ai PbA (Equation 3.1)
Predicting PbD when both PbA and PbS are available:
G.M. PbD = by + bi PbA + b2 PbS (Equation 3.2)
Predicting PbD when only PbA is avaialble:
G.M. PbD = co + ci PbA (Equation 3.3)
Because these are predictions for geometric means and are fitted after a
log transformation, we know that ci t (bi + ai b2)- Note that ci > bi
since it includes the indirect PbA — > PbS — > PbD pathway. The values
shown in Table B-2 were obtained from two data sets. AGG refers to the
40 community averages from different lead point sources identified in Table
A-3. Table B-3 summarizes the relationships derived from each of the locations.
EH is based on a sample of households with young children obtained in
1983' in East Helena, Montana, by the Centers for Disease Control and the
Montana Department of Health and Environmental Sciences (CDC, 1983).
These equations are not intended to be used recursively. If only PbA is
available, then Equation 3.1 may be used to estimate PbS and Equation 3.3
may be used to estimate PbD directly, subsuming the intermediate
PbA —> PbS —> PbD pathway. If PbS is in fact measured (such measure-
ments being more often available than PbD), then Equation 3.2 can use the
actual information in both PbA and PbS data without an intermediate
estimate of PbS.
-------
B-8
TABLE B-2
LINEAR MODEL PARAMETER ESTIMATES FOR AIR:
SOIL AND DUST LEAD RELATIONSHIPS*
Data Set
Parameter ao
al
bo
bl
b2
CO
ci
~From (Marcus, 1988c)
G.M. PbS = ao + ai PbA
G.M. PbD = bg + bi PbA + b2 PbS
G.M. PbD = eg + ci PbA
TABLE B-3
AVERAGE DUST AND SOIL LEAD VS. AIR LEAD RELATIONSHIPS BY LOCATION1
Dust Soil
PI ace
N
cn
ci
an
ai
Ajo, AZ
4
46
195
41
204
Anaconda, MT
4
122
715
107
454
Bartlesville, OK*
4
139
974
-484
3989
Belgium
4
--
--
- 95
625
E. Helena, MT
5
263
418
6b
268
Kellogg, ID
4
1488
613
278
407
Herculaneum, MO
10
220
934
-196
994
Omaha, NE
4
128
566
59
557
Palmerton, PA
3
144
566
225
648
Toronto, ON
2
400
382
365
681
Yugoslavi a*
2
1525
188
—
--
All**
41
60
844
53
510
G.M. PbD = co + c\ PbA G.M. PbS = ag + a^ PbA
* Outliers: Dropped from analysis
**Includes Ajo, Anaconda, Belgium, E. Helena, El Paso (only one data point),
Kellogg, Missouri, Omaha, Palmerton, Toronto
AGG
53.0
510.0
31.3
638.0
0.364
60.0
844.0
EH
88.1
206.0
184.0
267.0
U.894
220.0
551.0
1-Data from studies listed in Table A-3.
-------
B-9
The data on time scales for soil and dust lead changes is inconclusive.
Our opinion is that lead in undisturbed soil matrix persists for an
extremely long time, but that soil lead concentrations in disturbed
(especially urban) environments will change, on average, over periods of
a few years to reflect changes in surface deposition. Interior dust lead
concentrations will likely change over a period of weeks to months in
response to air lead changes, depending on interior-exterior access and
interior recirculation or removal of dust. Thus the following strategy
is recommended.
The estimation of changes in PbS and PbD from a change in PbA (brought
about from a hypothetical change in the Pb NAAQS) can be estimated under
two situations—short-term and long-term. In the long-term, equations
to use in predicting equilibrium levels of PbS and PbD following a change
in PbA are illustrated by equations 3-1 and 3-3 above. Each predicts soil
or dust lead dependent only on air lead, which will generallybe the only
information available in applied policy analyses. When available, PbS
in combination with PbA should prove far more predictive of PbD than PbA
alone, since the indirect pathway from PbA to PbD through PbS is probably
a more important dust lead source (Rabinowitz et al., 1985).
The short-term changes in PbS and PbA, however, require further
consideration. In time periods measured in terms of several months, we
would expect to see little or no change in PbS from a change in PbA.
Therefore, PbS would remain constant in the short-term. In contrast,
dust lead, being a soiling phemonena, would be expected to change in the
short-term as a result of changing ambient lead concentrations. There
are, however, two components comprising the level of PbD. There is
direct deposition from the atmosphere and secondary deposition caused by
-------
B—10
such factors as reentrainment from the soil. Therefore, the coefficient,
t>i, in Equation 3.2 must be used to predict short-term changes in PbD
(from equilibrium levels) from a change in PbA.
The model is intended to be used at many sites. Of the two data sets
that were analyzed, the aggregate data set is preferable because it includes
site-to-site differences in soil and dust slopes. Slopes from any particular
location such as East Helena reflect local climate, soil type, accessibility
of lead soil and air to the interior of the house, and housecleaning
practices related to awareness of lead hazards. Thus, a model based on
the EH data set may not be as transportable to other sites and the relationships
expressed in Appendix A are based on the AGG analyses.
4. Conclusions
We have shown that the relation between soil lead, dust lead, and
air lead is expected to vary over time, even under near-equilibrium
conditions. The various slopes can be expected to depend on site-specific
properties such as the deposition rate, removal rate, total airborne
suspended particulate concentration, and non-air lead influx rate for
lead-contaminated surface soils. There are almost no data tnat can be
used to provide estimates of critical kinetic parameters such as soil or
dust lead removal or turnover rates.
We thus suggest calculating the expected changes in PbS and PbD Trom
equilibrium relationships. In the short term (a few months) we expect
little or no change in surface soil Pb, but almost complete equilibration
of the PbA (only) component of PbD and no change in the PbS component of PbD.
In the long-term (e.g., ten years after a PbA change) we can use the fully
equilibrated PbD vs. PbA slope that includes the indirect PbA —> PbS —> PbD
pathway. The data from the aggregation of communities will be used since
it will be more representative of the majority of point source locations.
-------
APPENDIX C. ESTIMATING 1990 BASELINE BLOOD LEAD AVERAGES
1. Introduction
In estimating PbB levels expected under alternative lead NAAOS, both
the disaggregate model for adults and the aggregate model for
young children require estimates of the contributions to blood lead from
sources that will be unaffected by changes in atmospheric lead related to
change in the lead NAAQS. These sources include, for example, solder in
canned foods and plumbing and historical deposition of gasoline lead.
Blood lead contributions from non-air sources of lead are explicitly
accounted for in the disaggregate model for children, and in the integrated
uptake/biokinetic model. This section describes methodology for, and results
of, estimating average blood lead contributions from non-air sources, or
"baseline" blood lead levels for children and middle-aged male populations
living in 1990. Estimates for women of child-bearing age are presented
for 1) comparative purposes; 2) they could be used to estimate changes in
fetal lead levels if biokinetic data during pregnancy becomes available; and
3) to adjust children's PbB estimates for recent reductions in maternal exposures.
1990 is the starting year for the various exposure analyses that will be
conducted as part of the lead NAAQS review.
2. Abroach .
Ideally, we would know current PbB levels in pregnant women, middle-
aged men, and young children living near point sources. Known blood lead
distributions could then be adjusted to account for the continuing downward
trends in food, water, and air lead concentrations, described in Section II,
in order to estimate baseline 1990 blood lead distributions. Because
-------
C-2
there have been no recent systematic surveys of PbB levels in the U.S.,
baseline blood lead estimates will have to be based on earlier data.
The most recent, and well-conducted, nationwide survey of PbB levels
was the Second National Health and Nutrition Evaluation Survey (NHANES II)
of 1976 to 1980. Since then, lead concentrations in different media have
dropped substantially and have apparently, and predictably, been accompanied
by significant reductions in lead exposure. The approach taken here will be:
1) start with average PbB levels from 1978, the midpoint of the NHANES II
survey; and 2) estimate 1990 blood lead averages by adjusting the 1978 values
for changes in gasoline emissions, and food and water lead concentrations,
that have recently occurred and that can be expected to continue.
The issue of fetal lead exposure will be addressed separately.
3. ^978_Bl_oo^d_LeadJ^ata
Table C-l lists mean PbB levels for selected population subgroups
measured as part of NHANES II in 1978, based on data provided to 0A0PS
by Joel Schwartz of EPA's Office of Policy, Planning and Evaluation.
These levels will serve as starting points for adjustments to account for
the exposure sources in flux, described above.
TABLE C-l. SELECTED 1978 BLOOD LEAD LEVELS FROM NHANES II
P^ul_at1_on_(A^e|
Geoiretr^c_Mean_B_l_ood Lead_Level__(_(jg;/d,l_)_
14.9
Children
(0.5 - 5 years)
Females
(15-44 years)
10.8
White Males
(40-59 years)
15.4
Black Males
(40-59 years)
17.7
-------
C-3
Black middle-aged men have not been identified in the CD as a separate
sensitive group. White middle-aged men have been the focus of the two key
population studies on lead-related blood pressure effects (British
Regional Heart Study and U.S. NHANES II) in order to avoid the effects of
confounding variables and because of less extensive data for non-whites.
Given their high risk of cardiovascular disease, it is likely that a
similar, but possibly less apparent, blood lead/blood pressure relationship
exists in black males and therefore, blood lead changes for this group will
also be estimated.
4. Adjustment for Gasoline Lead Phasedown
The strong correlation between gasoline lead consumption declines and
the downward trend in NHANES II PbB levels between 1976 and 1980 has been
described in the 1985 Regulatory Impact Analysis for the final rule on
lead in gasoline (EPA, 1985) and the CD (Section 11.3.6). Similar results
were found with U.S. childhood lead-screening data (Schwartz et al.,
1984) and in an Italian isotope experiment where fairly rapid changes in
gasoline lead produced large changes in adult and children's blood lead
(Facchetti, 1985; CD, p. 11-87).
Gasoline lead accounted for an estimated 60 percent of the lead in the
average American's blood in the second half of the 1970's, and explained
short-term seasonal increases in PbB levels from winter to summer as well
as the long-term drop. Further, the accelerated rate of decline in gasoline
lead after 1978 was paralleled by an accelerated decline in PbB levels.
Additional support for a causal association between gasoline lead and blood
lead is seen in a) the blood lead trend, which was consistent throughout the
entire distribution, and not only through truncation of the high blood lead
levels (CD, p. 11-185); b) one-month lagged gasoline lead sales being the
-------
C-4
most significant predictor of blood lead, which matches the one month half-
life of lead in blood (EPA, 1985; p. 3-24) and is consistent with results of
Rabinowitz and Needleman (1982, 1983) who found that monthly changes in
gasoline lead exposure were the probable cause of trends in umbilical cord
PbB levels in Boston births (CD, p. 11-39); and c) the yasol ine/blood lead
relationship beiny stable across demographic and yeoyraphic boundaries and
between the first and second halves of the NHANES II survey, when yasoline
lead levels were rouyhly bl)% lower.
Regression coefficients estimated from the NHANES II analysis after
controlling for age, race, sex, region of the country, season, income,
degree of urbanization, and accounting for laboratory error, changes in
dietary lead (including canned food) and lead-painted housing and other
time-trends, have been used to predict changes in blood lead due to the
gasoline lead phasedown (Annest et al., 1983; Schwartz et al., 1984).
The phasedown will have continuing effects at least up to 1990-92 and
earlier projections regarding blood lead changes can be extended using
these regression coefficients. Joel Schwartz of EPA's Office of Policy
Analysis provided the coefficients to OAQPS (personal communication to
Jeff Cohen, May 9, 1988), derived from regressing natural-log blood lead
values on gasoline values since the data best fitted a loy-normal distribution.
Similar analysis by the Centers for Disease Control and the National
Center for Health Statistics has been described oy Annest et al. (1983).
The coefficients were derived from a period when gasoline lead
consumption dropped significantly and since then, the trend has continued
to an even greater extent. As noted earlier, adults have large stores of
skeletal lead accumlated over times of higher lead exposures. Some of
this lead is mobile and can be resorbed from bones into the bloodstream
after reductions in exposure (Rabinowitz et al., 1977). The extent of
-------
such bone lead resorption and the amount of time for re-establishing new
steady-state PbB levels in response to declines in gasoline lead emissions
since NHANES II is difficult to quantify. The only available direct data
on bone lead resorption come from excessively exposed people who may not
have the same biokinetic rates as those with lower exposures. Similarly,
other studies using isotopic lead tracers may not be relevant since
isotopic exchange rates may differ from bone lead resorption rates.
Nevertheless, workers exposed for periods up to 10 years (O'Flaherty et
al., 1982; Hryhorczuk et al., 198b) and 2U years (Ahlgren et al., 1987)
in the lead industry, with PbB levels as high as 70 pg/dl (and normal
renal function) had blood lead half-lives between 20 and 70U days and
showed significant blood- lead reductions as "quickly as 3 months to 2
years. Every indication from the NHANES II blood lead/gasoline lead
analysis (e.g., short time lags, seasonal parallels), and other studies
(e.g., Facchetti et a!.,.1985) indicate that non-occupationally exposed
adult blood lead levels respond quickly (e.g., as quickly as one month)
to fluctuations in gasoline lead. Given the fact that by 1990, the current
phase of the gasoline phasedown regulation (i.e., 0.10 grams of lead/leaded
gallon—current levels are slightly lower; D. Kortum, EPA's Office of Mobile
Sources, personal communication) will have been in full effect for about 3
years, whatever chanyes predicted based on the 1976-1980 NHANES results
can reasonably be applied to estimating "baseline" 1990 blood lead levels.
Table C-2 summarizes calculations using the regression coefficients
from NHANES II along with gasoline lead to estimate changes in average
blood lead expected from changes in recent and future gasoline lead usage.
The regression coefficients were similar for the different population groups.
As observed in the NHANES II analysis, the continued dramatic decline in
-------
C-6
TABLE C-2. PREDICTED CHANGES IN AVERAGE BLOOD LEAD LEVELS
ASSOCIATED WITH GASOLINE LEAD TREND
Children
(0.5-5 years)
1978 Geometric Mean
PbB (yg/dl) 14.9
Regression Coefficient
(log PbB change/100 tons
gasoline Pb change)3 .1220
Reduction in Gasoline Pb
Consumption, 1977-78 to
1990 (tons/day)b 435
Predicted Reduction
in Gas/Lead Contribution
to 1990 Mean PbB
(ug/dl) -8.6
Women White Men Black Men
(15-44 years) (40-59 years) (40-59 year
10.8 15.4 17.7
.1558 .1216 .1216
435 435 435
-5.5 . -9.1 -1U.5
Calculated by Joel Schwartz.
Estimates listed below, of gasoline lead usage provided by John Holley
and Dave Kortum of EPA's Office of Mobile Sources, Field Operations and
Support Division; 1990-92 projections based on Turner Mason and Company,
1987 report on petroleum industry marketing:
1977-78 1988 1990 1992c
Tons/day 438 4.8 2.9 1.4
cDespite further decline in 1992 in gas-lead consumption, no additional
reduction in blood lead levels predicted by model.
-------
C-7
gasoline lead emissions is predicted to parallel a major shift in PbB
levels. This decline will be used along with adjustments expected because
of other lead exposure trends to estimate average 1990 baseline blood lead.
5. Ad^us tmeat_f or _Di_etar^_Lead_R eductions
A significant reduction in dietary lead intake has occurred since the
late 1970's, and this trend will continue as atmospheric lead emissions
and deposition, the use of lead-soldered cans, and lead levels in drinking
water all continue to decline. Source contributions to children and adult
diets are estimated in Chapter 7 of the CD, as described in Section II.n of this
paper. Age-specific estimates of dietary lead intake in children for 1978-
1985 based on that information are detailed in Sledge (1986) and projections
of future lead intakes, incorporating more recent data in Hegel (1988)
have been estimated for children in 1990 (Cohen, 1988). Estimated differences in
dietary lead intake between 1978 and 1990 are show in Table C-3. The 1982-
85 values are taken directly from Flegel et al. (1988), who showed that
earlier predictions using the modeling approach developed in the CD were
validated by actual data since made available by FDA. The 1978 estimates are
backward extrapolation's of the 1982-95 data using the calculations in Sledge
(1986). Among children, only 2-year olds are considered for simplicity
since other children would have roughly the same proportional decrease in
dietary lead over this time period. Likewise, the 1978-1985 extrapolations
for adults were calculated by assuming the same proportional declines that
occurred in 6-year olds' dietary lead intake, given that the sources of lead
intake among older children are the same as for adults (confirmed by Rob
Ellas of EPA's Environmental Criteria and Assessment Office; personal
communication to Jeff Cohen, March 9, 1988). The 1982-1985 estimates for
all 3 subgroups are consistent with updated analyses by FDA based on
recent Total Diet Study data for different age-sex categories, as presented
in the ATSDR (1988) report to Congress (Table IX-6).
-------
c-a
TABLE C-3.
TRENDS IN
INTAKE
DIETARY LEAD INTAKE (uy/day): TOTAL
(AND ATMOSPHERIC CONTRIBUTIONS)
1978a
1982<*
1983
1984
1985
199Ue
2 year old
children
52.9
(27.0)*>
30.6
21.6
21.7
13.1
10.4
(1.2)f
Adult Females
65.6
(26.3)c
37.9
27.2
27.6
14.5
11.5
(4.3)c
Adult Males
91.5
(39.0)C
55.2
36.3
42.5
19.8
15.7
(6.1)c
al978 values are extrapolations based on estimates in Sledge (1986),
assuming a mean lead in drinking water of 12 ug/L.
^Atmospheric contribution to 1978 dietary lead in children from Sledge (1986)
based on Multiple Source Food Model developed in CD, Chapter 7.
cAtmospheric contributions to 1978 and 199U adult diets derived by taking
proportional contributions to 1982-1985 diets, as provided in Flegel et
al. (1988).
d1982-85 values are from Flegel et al. (1988). Exception is 1982 value
for children which is derived in Cahen (1988a,b) assuming a mean tap water
lead level of 17 U9/L.
e1990 projections, described in Cohen (1988a,b) derived by adjusting 1982
values based on following assumptions: 1) switchover to lead-free cans
will continue such that by 1989, lead levels in canned foods will be 1U%
of 1982 values; 2) average lead level in drinking water will be 17 pg/L
in 1990.
^Atmospheric contribution to 1990 children's diet derived in Cohen (1988a)
by multivariate regression analysis using CD Multiple Source Food Model,
FDA data, and projected gasoline and ambient air lead data.
-------
C-9
The dietary intake estimates for 1982, based on the CD analysis, were
adjusted for the changes that have occurred and that are expected to continue,
due to the phaseout of lead-soldered cans, reductions of lead in drinking
water, and reductions in atmospheric lead deposition onto crops and other
vegetation, to yield 1990 dietary estimates. Again, the same proportional
change in dietary lead intake for a 6-year old, estimated between 1982-86 and
1990, can be assumed for adults since the sources of dietary lead in both
diets are the same.
Part of the reduction in dietary lead intakes is attributable to reduced
deposition of atmospheric lead particles, and although the NHANES II regression
coefficients used to adjust PbB levels for the gasoline phasedown represented
total exposure, they probably did not capture much of the latter pathway
since it takes some time (at least more than a month) for changes in gasoline
lead usage to be reflected in different foods. The gasoline lead/blood lead
coefficients would likely have been higher had a longer timeframe been analyzed.
Given the uncertainties of precisely how much dietary lead exposure
is in fact captured by the gasoline lead/blood lead coefficient, a range
of assumptions will be made. As a lower bound, we will assume that the
adjustment described previously for gasoline lead reductions, based on
1976 to 1980 data, will account for reductions in atmospheric lead
contributions to dietary lead intake. In other words, the entire reduction
in dietary intake estimated from 1978 to 1990 (52.9 yg/day - 8.8 [jg/day =
44.1 yg/day for children, for example) would be considered in addition to
the adjustment for gasoline lead reductions. Alternatively, the contribution
of atmospheric lead deposition will be "factored out" of the estimates of
dietary lead intake for 1978 and 1990 to account for reductions in gasoline
lead usage. Atmospheric contributions to children's dietary lead intake,
for example, are estimated to have been 27.0 pg/day in 1978 (Sledge,
-------
C-10
1986) and to be 1.2 yg/day in 1990 (Cohen, 1988). Factoring out atmospheric
lead deposition from the dietary lead intakes cited above yields 52.9-
27.0 = 25.9 yg/day for 1978 and 10.4-1.2 = 9.2 yg/day for 1990. The range
of dietary lead intake reductions estimated between 1976 and 1990 for
children is therefore comprised of two values--25.9-9.2 * 16.7 yg/day, and
52.9-10.4 = 42.5 yg/day. Corresponding ranges for adult females and adult
males are 32.1-55.1 yg/day and 42.9-75.8 yg/day, respectively.
To estimate the impact on blood lead levels that changes in dietary
lead have had, and will continue to have by 1990, dose-response relationships
between PbB levels and lead levels in food and/or water can be used. Studies
relating blood lead to dietary intake, measured by duplicate diets or fecal
lead determinations, are summarized in the CD and the most relevant coefficients
relating dietary lead, or "slopes" are cited in Section IV.B.l and IV.8.2 of
this report. Most available studies are either on infants or adults; Slopes,
in units of yg/dl PbB per pg dietary Pb intake/day, range from about .02 to
about 0.060 (U.K. Central Directorate, 1982; Sherlock et al., 1982) from adult
studies and for infants, from 0.16 (Ryu et al., 1983) to 0.25 (Lacey et al.,
1985) at low levels, and 0.026 at high exposure levels where the relationship
flattens out (Marcus, 1989 based on Lacey et al. data).
The Laxen et al. (1987) study on Scottish school children provides data
to relate water lead concentration to blood lead. After parti ailing out the
effects of house dust lead, the authors suggested a linear relationship between
blood lead and water lead with a slope of 0.062 gg/dl per yy/L. Reanalysis of
these data indicated that a piecewise linear relationship fit as well as non-1inei
models, with a slope of 0.161 ug/dl per ug/L for water lead concentrations
below 15 ppb and a lower slope of 0.0318 for water lead above that inflection
point (Marcus, 1989). Assuming that the kitchen water lead levels correlated
well with total dietary lead concentration, for this population, this relationship
-------
C-ll
can be used to predict the contribution of diet to blood lead. Results of the
Laxen et al. analyses can thus be used to predict changes in blood lead by
converting dietary lead intake values (in pg/day) to concentrations in terms of
py/kg, or ppb, as follows:
Dietary Lead Intake (pg/day) x slope (uy/dl)
Total Dietary Intake (kg/day) (py/Kg)
where: reduction in dietary lead intake between 197a and 199D is in the
range of 16.7 to 42.5 ug/day;
total dietary intake .for a 2-year old child is estimated at 1.502
kg/day (CD , Table 7-15); and
slope from the Lacey et al. study is estimated at 0.16 for lead
levels below 15 ppb and 0.032 for levels above 15 ppb.
Table C-4 presents results of the above calculations for children, along
with adjustments for 1978 adult PbB levels derived from multiplying the slopes
discussed in Section IV.B.2 by estimated reductions in dietary lead intakes.
These adjustments will be combined with those made previously for the gasoline
lead phasedown to yield 199U average baseline PbB levels.
TABLE C-4. ADJUSTMENTS TO 197a MEAN BLOOD LEAD LEVELS FOR DIETARY
LEAD REDUCTIONS
Range of Differences PbB Reduction Estimated
Between 1978 and 1990 By 199U Due to Dietary
Estimated Dietary Lead Lead Reduction
Intake (pg/day)a (-pg/dl )b
16.7-42.5 0.9-1.8
32.1-55.1 1.0-1.8
42.9-75.8 1.4-2.4
Children
Adult Females
Middle-aged Men
aFrom Table C-3. Ranges reflect uncertainty as to how much of the adjustment
for gasoline lead reductions, using NHANES II-derived coefficients, captures
changes in atmospheric contributions to dietary lead intake.
bSlope values used to convert dietary lead intake to average blood lead
changes are from reanalysis by Marcus, 1989 of Laxen et al.
(1987) for children (see text); and 0.032 pg/dl per pg Pb/day for adults
(Sherlock et al., 1982; Cools et al., 1976; see Section IV.B.2).
-------
C-12
6. Adjustment to Children's PbB for Reductions in Maternal Exposure
Since lead exposure has been, and will continue to be, reduced for
women since the 1970's, PbB levels in children born since then can be
expected to be lower as a result. Over the course of the NHANES II
survey, much of the reduction in yasoline lead emissions was probably
reflected in lowered maternal, as well as newborn PbB levels. Thus, it
can be assumed tha.t the adjustment to children's PbB already described
for gasoline lead changes sufficiently included any gasoline-attributable
reductions maternal lead exposure.
Reduction in maternal PbB attributable to changes in PbB dietary
lead intake, however, is not reflected in adjustments made previously.
One method to account for this is by using a kinetic model fit to blood
lead data collected longitudinally from young children from birth to 27
months (Succop et al., 1987). This model provides a rate of blood lead
change which can be applied to the amount of newborn blood lead affected
by reductions in maternal dietary lead intake estimated between 1978 and
1990. As will be shown in Section C.8 below, the ratio between newborn
PbB to maternal PbB is on average around 0.8. Thus, the followiny equation
is derived to account for the propagation of PbB reductions in women, due
to dietary changes (1.0 - 1.8 yg/dl), to 2-year olds:
[(1.0 - 1.8 ug/dl) x 0.8] * e-^t = dPbBm
where: the rate constant « 3 0.072 (Succop et al., 1987)
t = 24 months
d PbBm = PbB reduction in 2-year olds expected in 1990 due to reductions
in maternal dietary lead intake.
-------
C-13
The resulting range for dPbBm is 0.2 - 0.3 pg/dl at age 2. This
adjustment will be combined with the others for 2 year old children to
yield an expected background mean PbB for 1990.
7. Adjustment for Implementation of the Lead NAAQS
The midpoint of the NHANES II survey, 1978, was the year the current
lead NAAQS was promulgated. Since then air lead levels have dropped
dramatically throughout the country, mainly due to the gasoline phasedown,
but also because of implementation of the air standard, especially in
areas dominated by industrial point source emissions. Since the bulk of
the NHANES II sample lived in urban or rural areas, fairly remote from
major lead point sources, it will be assumed that correcting for the
gasoline lead phasedown will capture most of the reduction in air lead
exposures that has occurred since implementation of the standard.
8. Additive Adjustments to 1978 Blood Lead Level Averages
Table C-5 summarizes the derivation of mean PbB levels estimated for
four subpopulations in 1990. These averages reflect changes that have
occurred since 1978, and that are expected to continue, in gasoline lead
emissions and deposition, canned food technology, and corrosion control
for lead in drinking water. Any other changes not quantified were probably
of lesser importance for the bulk of the NHANES II population (e.y.,
reductions in industrial point source emissions). The 1990 baseline
averages can subsequently be used to model exposure changes expected
under alternative lead NAAQS between 1990 and 1996 using the disaggregate
model for adults and the aggregate model for children. The methodologies
for these models are discussed in Section V.
-------
C-14
Table C-5. DERIVATION OF 1990 "BASELINE" AVERAGE BLOOD LEAD LEVELS
Children Women White Men Black Men
1978 NHANES II 14.9 10.8 15.4 17.7
Average (yg/dl)l
Adjustment for 8.6 5.5 9.1 10.5
Gasoline Pb
Phasedown
(-yg/dl)
Adjustment for 0.9 - 1.8 1.0 - 1.8 1.4 - 2.4 1.4 - 2.4
Dietary Lead
Reductions
(-ng/dl)
Adjustment for 0.2 - 0.3 - -
Reductions in
Maternal
Exposures
(- yg/dl)
Estimated 4.2 - 5.2 3.5 - 4.3 3.9 - 4.9 4.8 - 5.8
1990 Average
(Mg/dl)!
^Geometric Mean Levels
-------
C-15
9. Estimating Fetal Exposures
Different indices of prenatal and neonatal lead exposure have been found
in longitudinal studies to correlate with delays in early mental and physical
development. Lead readily crosses the placenta and accumulates in fetal
tissues throughout development. The dynamics of lead exchange during
pregnancy are complex and poorly understood (see CO Addendum). Ideally,
serial measurements of maternal lead would have been taken in the studies
just prior to conception and throughout pregnancy and biokinetic models
would exist that can estimate actual fetal exposures. In the absence of
either, the question is whether equilibrated PbB levels in pregnant
women are adequate surrogates for critical fetal lead exposure, or is
there a better index.
Umbilical cord blood lead, for example, which has been a key effects
predictor in several studies, could be estimated by adjusting maternal PbB
predictions -and applied to available dpse-response relationships to assess
risks. Table C-6 lists available data on paired mother/child PbB concen-
trations along with ratios between average maternal and cord PbB levels.
The consistently high ratios (mean of 0.80 or 0.82, depending on the inclusion
of one study with outlying results) indicates that the mature fetus absorbs
a sizable fraction of circulating maternal lead.
Cord blood lead, however, may not accurately reflect individual
circumstances or past exposure levels. In at least some pregnancies,
there could be greater than normal transfer of lead from mother to the
fetus. Mobilization of bone lead stores during pregnancy and lactation
may be more substantial in some women, and iron and calcium deficiency, both
common in pregnancy, enhances gastro-intestinal absorption of lead (Rom, 1976).
In addition, integrated exposure levels during the course of pregnancy
-------
C-I 6
Table C-6. MATERNAL: CHILD BLOOD LEAD CONCENTRATIONS AND RATIOS
Study/Location
Number of Mean Blood Lead (ug/dl)
Subjects Maternal Cord Newborn
Ratio of
Cord; Maternal Pb
Barltrop (1968)/U.K. 29
Harris and Hoi ley
(1972)/U .S. 24
Haas (1972)/Germany 294
Zetterlund et al.
(1977)/4 areas in
Sweden
A 21
B 173
C 103
D 37
Clark (1977)/Zambia
- Control 31
- Mine 122
Hubennont et al. (1978)/
Belgium 70
Buchet et al. (1978)/
Belgium 474
Dragovic (1980)/
Yugoslavia
-Smelter 102
-Control 45
Alexander and
Delves (1981)/U.K. 165
Moore et al. (1982)/ 236
U.K.
Winneke et al. (1985)/ 114
FRG
Ernhart et al. (1986)/ 118
U.S.
McMichael et al. (1986)/
Australia
Port Porie 519
Other 172
13.9
13.2
16.9
6.1
9.2
8.4
9.4
14.7
41.2
12.0
10.4
32.6
18.6
12.2
14.0
9.3
6.5
10.8
12.3
14.9
4.4
8.0
7.3
6.8
11.8
37.0
10.1
8.3
28.3
17.3
12.0
8.2
5.9
11.0
10.U
0.78
0.93
0.88
0.72
0.87
0.87
U.72
0.80
0.90
0.84
0.80
0.87
0.93
0.86
0.88
0.91
10.8
7.7
10.1
6.2
11.2
7.5
0.94
0.81
-------
C— 17
Table C-6. MATERNAL: CHILD BLOOD LEAD CONCENTRATIONS AND RATIOS
(Continued)
Roels and 1978
Lauvterys/Bel gi um
474
Dietrich et al . (1987)/ 305
U.S.
Bel linger et al . 216
(1986)/ U.S.
Mil man et al. (1988)/
Denmark 48
Fahim et al. (1976)/
U.S.
Region I 240
Region II 177
10.1
8.0
35
13.1
14.3
8.4
6.3
6.5
20
4.3
4.6
0.83
0.79
4.6-
5.9
6.2-
7.7
0 j_57
Weighted Mean: 0.82
0.33
0.32
Weighted Mean: 0.80
-------
C-18
may not be accurately indexed by blood lead levels at delivery or by
single measurements during pregnancy. Average maternal PbB levels can
fluctuate considerably during pregnancy (CD Addendum, p. A-45), while the
equilibrium between blood and soft and skeletal tissues in the mother is
displaced, and lead is transferred to the placenta and fetal tissues;
Alexander and Delves (1981) found an almost 20% drop in maternal PbB
levels from 8 weeks until delivery.
As noted in Section V.A., there are significant uncertainties associated
with predicting PbB levels among pregnant women and certainly fetal lead
exposures, under alternative regulatory scenarios. Further, the chance of
underpredicting exposures is great. Therefore, for purposes of assessing
risks associated with alternative lead NAAQS, no quantitative estimates
will be generated of PbB levels among pregnant mothers as surrogates for
fetal exposures. Potential risks to this extremely sensitive, population
will be emphasized in qualitative terms.
-------
APPENDIX D. CASAC CLOSURE REPORT
UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON. D C. 20460
April 27, 1989
OFFICE OF
THI AOMINISTRATOR
The Honorable William Reilly
Administrator
U.S. Environmental Protection Agency
401 M Street, SW
Washington, DC 20460
Dear Mr. Reilly:
We are pleased to transmit via this letter the advice of the
Clean Air Scientific Advisory Committee (CASAC) concerning its
review of the EPA document "Review of the National Ambient Air
Quality Standards for Lead: Exposure Analysis Methodology and.
Validation" (August 1988).
This document was reviewed by the Lead Exposure Subcommittee
of CASAC on October 25, 1988. It was the unanimous consensus of the
Subcommittee that the document is scientifically adequate for use
in the standard setting process for lead as an ambient air
pollutant. The CASAC hereby endorses the report of its
Subcommittee. A detailed presentation of our views are contained
in the attached report.
We appreciate the opportunity to provide advice on this
important issue. Further advice concerning the lead national
ambient air quality standards will be contained in our closure
letter on the Lead Staff Paper.
Sincerely
Timothy La:
Chairman, Lead Exposure
Roger 0. McClellan
Chairman, Clean Air Scientific
Advisory Committee
-------
2
U. s. Environmental Protection Agency
Science Advisory Board
Clean Air Scientific Advisory Committee
LEAD EXPOSURE SUBCOMMITTEE
Dr. Timothy Larson, Environmental Engineering and Science
Program, Department of Civil Engineering,
University of Washington, Seattle, Washington
Dr. J. Julian Chisolm, Jr., Associate Professor, Johns Hopkins
School of Medicine, Francis Scott Key Medical Center,
Baltimore, Maryland
Dr. Scott Clark, Professor of Environmental Health, Department
of Environmental Health, University of Cincinnati Medical
Center, Cincinnati, Ohio
Dr. Ian von Lindem, President, Terragraphics Environmental
Engineering, Moscow, Idaho
Dr. Kathryn R. Mahaffey, Science Advisor, Office of the
pirector, NIEHS, University of Cincinnati Medical Center,
Cincinnati, Ohio
Dr. M. Granger Morgan, Head, Department of Engineering and
Public Policy, Carnegie Mellon University, Pittsburgh,
Pennsylvania
Dr. Paul Mushak, Consultant and Adjunct Professor, University
of North Carolina, Durham, North Carolina
Dr. Michael B. Rabinovitz, Investigator, Marine Biological
Laboratory, Woods Hole, Massachusetts
Dr. Robert D. Rove, Senior Vice President, RCG/Hagler, Bailly
Inc., Boulder, Colorado
Executive Secretary
Mr. A. Robert Flaak, Environmental Scientist, Science Advisory
Board (A-101F), U.S. Environmental Protection Agency,
401 M Street, SW, Washington, DC 20460
-------
3
REPORT OF THE CLEAN AIR SCIENTIFIC ADVISORY COMMITTEE
ON ITS REVIEW OF
"REVIEW OF THE NATIONAL AMBIENT AIR QUALITY STANDARDS FOR LEAD:
EXPOSURE ANALYSIS METHODOLOGY AND VALIDATION"
With the dramatic decline in the emissions of lead from mobile
sources, there has been increased interest in airborne lead near
stationary sources. Given a proposed change in lead concentrations
near stationary sources due to reduced emissions, will there be
predictable changes in blood lead levels in the surrounding
population? Exposure models are currently the only practical tool
that can address this question within the framework of the national
ambient air quality standard (NAAQS) setting process. The
scientific framework for such modeling is the basis of the document
reviewed by the Clean Air Scientific Advisory Committee's (CASAC)
Lead Exposure Subcommittee. In addition to discussing various
modeling approaches, the document also presents several validation
studies in order to compare these modeling approaches with actual
observations of blood lead levels near several point sources of
lead. It was the unanimous consensus of this Subcommittee that the
document is scientifically adequate and that the EPA staff's
proposed changes to the document discussed at the meeting
appropriately address the written comments of the Subcommittee
members. The Clean Air Scientific Advisory Committee (CASAC)
hereby endorses this report of its Lead Exposure Subcommittee.
The validation studies presented convincing evidence for a
decrease in blood lead levels with increasing distance (eut to
several kilometers) from a point source. The Subcommittee agreed
with the conclusion that any attempt to predict blood lead levels
must include all the important exposure pathways and that the
direct inhalation route of airborne lead is a relatively minor
pathway in children. The validation studies also provided
additional information on the lead levels of other important
exposure media, including soil, house dust, food and water.
Therefore, these studies provide a unique opportunity to test the
ability of various exposure/uptake models to predict mean values
of blood lead from various routes of exposure. Blood lead levels
were predicted using both a disaggregate approach as well as a
biokinetic approach. The Subcommittee recognized that several of
the inputs to the exposure models are uncertain, but felt that this
uncertainty was adequately recognized in the document. More
important, the Subcommittee concurred with the general modeling
framework and endorsed the use of the biokinetic model in children
under six years of age and the use of the disaggregate approach in
adults. The Subcommittee also strongly emphasized that these
modeling predictions were not valid for pregnant women and their
fetuses due to a lack of information on this potentially important
subpopulation. ' The Subcommittee also recommended that the
exposure model outputs include not only the predicted mean blood
lead levels as a function of downwind distance but also the
-------
4
corresponding lead levels in all exposure media including air,
soil, dust, food and water. These outputs would provide an
additional basis for evaluating model performance.
Given a predicted mean blood lead level, another important
component of the exposure model is the prediction of maximum blood
lead levels in exposed individuals. EPA's approach is to use
empirical estimates of the variance of blood lead levels in the
general population, as well as those in populations living near
lead point sources, as a predictor of peak to mean values. The
document correctly recognized that the population variance
estimates depend on many different factors (e.g., biological,
climatic, behavioral) that exposure modeling cannot fully capture
at this time. Given the uncertainties, the Subcommittee agreed
that the only reasonable assumption is to use the range of variance
estimates from the empirical data. However, because this is a
sensitive parameter, we felt the additional concern that as blood
lead levels continue to decrease in the future, the assumption of
a constant proportional variance (log-normality assumption) may be
compromised by analytical uncertainties in the measurement of blood
lead, but the Subcommittee felt that this issue was adequately
addressed in the document.
The Subcommittee felt very strongly that the results of this
modeling exercise not be taken out of context. For example,
because the available data on lead in drinking water for the
validation studies was limited, the biokinetic model in this
application was used to calculate average drinking water exposures
over time. However, the biokinetic model is sensitive to total
intake from this route and can account for variations in water lead
exposure where appropriate data are available. While the model can
be used now to evaluate relative changes in blood lead levels from
changes in water lead levels, it has not been calibrated for
absolute assessments of risk from drinking water in the same way
as done for other routes of exposure. Use of the model for other
metals was also not recommended at this time. In addition,
although the Subcommittee agreed that an appropriate application
of this approach might be for prediction of offsite lead exposures
from fugitive dust emissions, there was concern that until non-
linearities in the relationship between lead exposure and blood
lead are incorporated into the model, the model be limited to use
in areas where soil lead levels do not exceed 4000 ppm. In
addition, the model should not be used in areas where ingestion
(pica) of paint fragments is an important route of intake because
this variable was not considered in the case study validation.
Finally, the biokinetic model should not be used for predicting
adult blood lead levels at this time due to limited data regarding
historical exposures and the possible confounding factor of blood
lead coning from bone.
The Subcommittee was also asked for guidance on several
technical issues that are summarized below. As to the range of
2
-------
5
dirt ingestion rates used in the report (55-135 mg/day depending
on age), the Subcommittee agreed that this is a relatively poorly
defined parameter subject to climatic variations. Some members
felt that the value of 100 mg/day represented an upper limit for
a high risk child, whereas others felt that the use of the Binder
et al. and Clausing et al. studies was as good a choice as any
until further data are available. All members agreed that this is
an important parameter in determining total intake and that the
uncertainties have been adequately discussed in the document. In
this regard, there was agreement that the emphasis in future
research should focus on the lead levels in the surface layer of
the soil and not on the older, deeper layers. There was general
agreement that the model use a constant soil lead level in
predicting future scenarios, but that the house dust component
should track the air lead value. Finally, the approach of
interfacing the biokinetic and disaggregate models for intermediate
age groups was judged acceptable by the Subcommittee in the absence
of any other available information to the contrary.
3
-------
REFERENCES
Ahlgren, L.; Schutz, A.; Skerscing, S.; Christoffersson, J.O.; Mattson, S.
(1987) Lead in vertebral bone biopsies from active and retired lead
workers. Arch. Env. Health 42: 340-346.
Alexander, F.W.; Delves, H.T.; Clayton, B.E. (1973) The uptake and excretion
by children of lead and other contaminants. In: Barth, D.; Berlin, A.;
Engel, R.; Recht, P.; Smeets, J., eds. Environmental health aspects of
lead proceedings, international symposium: October 1972; Amsterdam,
The Netherlands. Luxembourg: Commission of the European Communities,
Centre for Information and Documentation, pp. 319-331.
Alexander, F.W.; Delves, H.T. (1981) Blood lead levels during pregnancy.
Int. Arch. Occup. Environ. Health 48: 35-39.
Altman, P.L.; Dittmer, D.S. eds. (1971) Respiration and Circulation. Fed. Am. •
Soc. for Exp. Biol., Bethesda, MD, pp. 40-41 and 78-84.
Altman, P.L.; Dittmer, eds. (1972) Biology Data Book, 2nd Ed. Federation
of American Societies for Experimental Biology, Bethesda, MD, Vol. 1,
pp. 195-201.
Angle, C.R. (1985) Personal communication to Donna Sledge, Office of Air
Quality Planning and Standards, U.S. EPA, Durham, N.C. July 27, 1985.
Angle, C.R.; Mclntire, M.S. (1979) Environmental lead and children: the
Omaha study. J. Toxicol. Environ. Health 5: 855-870.
Angle, C.R.; Mclntire, M.S. (1982) Children, the barometer of environmental
lead. Adv. Pediatr. 27: 3-31.
Angle, C.R.; Marcus, A.; Cheng, I-H, Mclntire, M.S. (1984) Omaha childhood
blood lead and environmental lead: a linear total exposure model.
Environ. Res. 35: in press.
Annest, J.L.; Pirkle, J.L.; Makug, D.; Neese, J.W.; Bayse, D.D.; Kovar, M.G.
(1983) Chronological trend in blood lead levels between 1976 and 1980.
N. Engl. J. Med. 308: 1373-1377.
Annest, J.L.; Mahaffey, K. (1984) Blood lead levels for persons ages 6 months
to 74 years, United States, 1976-80, U.S. Dept. of Health and Human Services,
Data from the National Health and Nutrition Survey, Series II, No. 233,
Hyattsville, MD.
ATSDR (1988). Agency for Toxic Substances and Disease Registry. The Nature and
Extent of Lead Poisoning in Children in the United States: A Report to Congress.
U.S. Dept. of Health and Human Services, Public Health Service, Atlanta,
July, 1988.
Azar, A.; Snee, R.D.; Habibi, K. (1975) An epidemiologic approach to
community air lead exposure using personal air samplers. In: Lead.
Environ. Qual. Saf. Suppl. 2: 254-290.
Baker, E.L., Jr.; Hayes, C.G.; Handrigan, P.J.; Handke, J.L.; Leger, R.T.;
Housworth, W.F.; Harrington, J.M. (1977) A nationwide survey of heavy
metal abosorption in children living near primary copper, lead, and
zinc.smelters. Am. J. Epidemiol. 106: 261-273.
-------
2
Barltrop, D. (1966) The prevalence of pica. Am. J. Dis. Child. 112: 116-123.
Barltrop, D.; Strehlow, C.D.; Thornton, I.; Webb, J.S. (1974) Significance
of high soil lead concentrations for childhood lead burdens. Environ.
Health Perspect. 7: 75-82.
Barltrop, 0.; Meek, F. (1979) Effect of particle size on lead absorption from
the gut. Arch. Environ. Health 34: 280-285.
Batschelet, E.; Brand, L.; Steiner, A. (1979) On the kinetics of lead in the
human body. J. Math. Biol. 8: 15-23.
Battye, B. (1988) Trends and frequency analysis of ambient lead data. Technical
memorandum to Jeff Cohen, U.S. EPA, Office of Air Quality Planning and
Standards, Durham, N.C. Alliance Technologies Corporation, September 28, 1988.
Battye, W.; Battye, R.; Browne, N.; Clowers, M.; Eichinger, J.;
Viconovic, G. (1985) Cost Assessment of Regulatory Alternatives for
Lead National Ambient Air Quality Standards: Draft Report. GCA Corporation,
GCA/Technology Division, Chapel Hill, N.C. July 1985.
Beck, B.D.; Tsai, P.Y. (1987) Risk assessment for lead contaminated soil.
Presented at Annual Meeting of Society of Toxicology, Boston.
February 25, 1987.
Bellinger, D.; Leviton, A.; Rabinowitz, M.; Needleman, H.; Waternaux, C.
(1986) Correlates of low-level lead exposure in urban children at 2 years
of age. Pediatrics 77: 826-833.
Bellinger, D.; Leviton, A.; Waternux, C.; Needleman, H.; Rabinowitz, M.
(1987) Longitudinal analyses of prenatal and postnatal lead exposure
and early cognitive development. New Engl. J. Med. 316: 1037-1043.
Beloian, A.; McDowell, M. (1981) Estimates of lead intakes among children
up to 5 years of age, 1973-1978 and 1980. Washington, DC: U.S. Food
and Drug Administration, Bureau of Foods; Division of Nutrition final
internal report. Available from: U.S. Food and Drug Administration,
Bureau of Foods, Washington, DC.
Bentler, P.M. (1980) Multivariate analysis with latent variables: causal
modeling. Ann. Rev. Psychol. 31: 419-456.
Bernard, S.R. (1977) Dosimetric data and metabolic model for lead. Health
Phys. 32: 44-46.
Billick, I.H. (1982) Prediction of pediatric blood lead levels from gasoline
consumption. Available from U.S. EPA, Central Docket Section,
Washington, D.C.; Docket No. A-81-36; Document No. IV.A.4.
Billick, I.H.; Curran, A.S.; Shier, D.R. (1979) Analysis of pediatric
blood lead levels in New York City for 1970-1976. Environ. Health
Perspect. 31: 183-190.
Billick, I.H.; Curran, A.S.; Shier, D.R. (1980) Relation of pediatric
blood lead levels to lead in gasoline. Environ. Health Perspect.
34: 213-217.
-------
3
Binder, S.; Sokal, D.; Maughan, D. (1986) Estimating soil ingestion:
the use of tracer elements in estimatiny the amount of soil ingested
by young children. Arch. Env. Health 41: 341-345.
Blake, K.C.H. (1976) Absorption of 203pb from gastrointestinal tract of man.
Environ. Res. 11: 1-4.
Bornschein, R.L.; Succop, P.; Dietrich, R.N.; Clark, C.S.; Que Hee, S.;
Harranond, P.B. (1985) The influence of social and environmental factors
on dust lead, hand lead, and blood lead levels in young children.
Environ. Res. 38: 108-118.
Bornschein, R.L.; Succop, P. Krafft, R.M.; Clark, C.S., Peace, B.;
Hammond, P.B. (1986) Exterior surface dust lead, interior house dust
lead, and childhood lead exposure in an urban environment. Conference
on Trace Metals in Environmental Health, Columbia, MO.
Briskin, J. (1988) Office of Drinking Water, U.S. EPA, Washington, D.C.
Personal communication to Jeff Cohen. July 28, iy88.
Brunekreef, B.D. (1984) The relationship between air lead and blood lead
in children: a critical review. Sci. Total Environ. 38: 79-123.
Brunekreef, 6.; Veenstra, S.J.; Biersteker, K.; Boleij, J.S.M. (ly81) The
Arnhem lead study: 1. lead uptake by 1- to 3-year-old children living
in the vicinity of a secondary lead smelter in Arnhem, The Netherlands.
Environ. Res. 25: 441-448.
Brunekreef, B.; Noy, D.; Biersteker, K.; Boleij, J. (1983) Blood lead levels
of Dutch city children and their relationship to lead in the environment.
J. Air Pollut. Control Assoc. 33: 872-876.
Brunekreef, B.; Noy, D.; Clausing P. (1987) Variability of exposure
measurements in environmental epidemiology. Am. J. Epidemiol.
125: 892-898.
Buchet, J.P.; Roels, H.; Hubermont, G.; Lauweyrs, R. (1978) Placental
transfer of lead, mercury, cadmium, and carbon monoxide in women.
Environ. Res. 15: 494-503.
Bureau of the Census (1983) Census of Housing: Components of inventory
change, United States and regions. Washington, D.C.: U.S. Dept. of
Commerce.
CDC [Centers for Disease Control] (1983) East Helena, Montana Child
Lead Study, Summer 1983. Lewis and Clark County Health Dept.;
Montana Dept. of Health and Environ. Sci., U.S. Dept. of Health
and Human Services, U.S. EPA. July 1986.
CDC [Centers for Disease Control] (198b) Preventing Lead Poisoning in Young
Children. U.S. Department of Health and Human Services. January, 1985.
Chamberlain, A.C. (1983) Effect of airborne lead on blood lead. Atmos.
Environ. 17: 677-692.
-------
4
Chamberlain, A.C. (1985) Prediction of response of blood lead to airborne and
dietary lead from voluntary experiments with lead isotopes. Proc. R. Soc.
London B 224: 149-182.
Chamberlain, A.C.; Heard, M.J. (1981) Lead tracers and lead balances.
In: Lynam, D.R.; Piantanida, L.G.; Cole, J.F., eds. Environmental
lead: proceedings of the second international symposium on environmental
lead research; December 1978; Cincinnati, OH. New York, NY: Academic
Press; pp. 175-198.
Chamberlain, A.C.; Heard, M.J.; Little, P.; Newton, D.; Wells, A.C.;
Wiffen, R.D. (1978) Investigations into lead from motor vehicles.
Harwell, United Kingdom: United Kingdom Atomic Energy Authority;
report no. AERE-R9198.
Charney, E.; Sayre, J.; Coulter, M. (1980) Increased lead absorption
in inner city children: where does the lead come from?
Pediatrics 65: 226-231.
Charney, E.; Kessler, R.; Farfel, M.; Jackson, D. (1983) Childhood lead
poisoning: a controlled trial of the effect of dust-control measures
on blood lead levels. N. Engl. J. Med. 309: 1089-1093.
Chisolm, J.J., Jr. (1984) The continuing hazard of lead exposure and
its effects in children. Neurotox. 5: 23-42.
Chisolm, J.J., Jr. (1986) Removal of lead paint from old housing: the
need for a new approach. Am. J. Public Health 76: 236-237.
.Chow, T.J.; Snyder, C.8.; Earl, J.L. (1975) Isotope ratios of lead as pollutant
source indicators. Proceedings of the symposium on isotope ratios as
pollutant source and behavior indicators; Vienna, Austria; IAEA.
Clark, A.R.L. (1977) Placental transfer of lead and its effects on the
newborn. Postgrad. Med. J. 53: 674-678.
Clark C.S.; Bornschein, R.L.; Succop, P., Que Hee, S.S.; Hammond, P.B.;
Peace, B. (1985) Condition and type of housing as an indicator of
potential environmental lead exposure and pediatric blood lead levels.
Env. Res. 38: 46-53.
Clark, C.S.; Bornschein, R.L.; Succop, P.; Hammond, P.B.; Peace, B.; Krafft, K.;
Dietrich, K. (1987) Pathways to elevated blood lead and their importance
in control strategy development. In: Lindberg, S.E.; Hutchinson, T.C.,
eds. International conference: heavy metals in the environment, vol. 1;
September; New Orleans. Edinburgh, U.K.; CEP Consultants, Ltd., Edinburgh
pp. 159-161.
Clausing, P.; Brunekreef, B.; van Wijnen, J.H. (1987) A method for estimating
soil ingestion by children. Int. Arch. Occup. Environ. Health 59: 73-82.
Cohen, J. (1986) Blood lead GSD for children living near lead point sources.
Memorandum to Joel Schwartz, Office of Policy Analysis, U.S. EPA.
Office of Air Quality Planning and Standards, Ambient Standards Rranch,
Research Triangle, N.C., April 8, 1986.
-------
5
Cohen, J. (1987) Respiratory deposition and absorption of lead particles.
Memorandum to Fred Miller and Ted Martonen, Inhalation Toxicology
Division, U.S. EPA. Office of Air Quality Planning and Standards,
Ambient Standards Branch, Research Triangle Park, N.C., October 7, 1987.
Cohen, J. (1988a) Dietary lead estimates for case study exposure analyses.
Memorandum to the Files, U.S. EPA, Office of Air Quality Planning and
Standards, Ambient Standards Branch, Research Triangle Park, N.C.
May 16, 1988.
Cohen, J. (1988b) Revisions to dietary lead estimates for case study exposure
analyses. Memorandum to the files, U.S. EPA, Office of Air Ouality Planning
and Standards, Ambient Standards Branch, RTP, September 9, 1988.
Cohen, J. (1988c) Lead exposure modeling for Superfund. Memorandum to
Dorothy Patton, Office of Research and Development, U.S. EPA. Office
of Air Quality Planning and Standards, Ambient Standards Branch, Research
Triangle Park, N.C. December 19, 1988.
Cohen, A.F.; Cohen, B.L. (1980) Protection from being indoors against inhalation
of suspended particulate matter of outdoor origin. Atmos. Environ. 14:
183-184.
Cools, A.; Salle, J.A.; Verberk, M.M; Zielhuis, R.L. (1976) Biochemical
response of male volunteers ingesting inorganic lead for 49 days. Int.
Arch. Occup. Environ. Health 38: 129-139.
Cooper, J.A.; Decesar, R.T.; Frazier, C.A.; Hovck, J.E.; Mohaw, J.F.
(1981) Determination of source contributions to air particulate lead and
cadmium levels in Kellogg, Idaho using the receptor model. NEA, Inc.,
Beaverton, OR.
Corrin, M.L.; Natusch, D.F.S. (1977) Physical and chemical characteristics
of environmental lead. In: Lead in the environment; a report and
analysis of research at Colorado State University, University of Illinois,
and University of Missouri, (W.R. Boggess, ed.); Washington, D.C.:
National Science Foundation; NSF report no. NSF/RA-770214; pp. 7-32.
Cotes, J.E. (1979) Lung function assessment and application in medicine.
Blackwell Scientific Publications, London, England, pp. 265-284.
Dacre, J.C.; Ter Haar, G.L. (1977) Lead levels in tissues from rats fed
soils containing lead. Arch. Environ. Contam. Toxicol. 6: 111-119.
Daines, R.H.; Motto, H.; Chilko, D.M. (1970) Atmospheric lead: its
relationship to traffic volume and proximity to highways. Environ.
Sci. Technol. 4: 318-322.
Davidson, C.I.; Osborn, J.F. (1984) The sizes of airborne trace metal -
containing particles. In: Nriagu, J.O.; Davidson, C.I., eds.
Toxic Metals in the Air. New York, NY: Wiley.
Davies, B.E.; Elwood, P.C.; Gallacher, J.; Ginnever, R.C. (1985) The relationship
between heavy metals in garden soils and house dusts in an old lead mining
area of North Wales, Great Britain. Env. Pollution 9: 255-266.
-------
6
Davies, P.J.A.; Thorton, I.; Watt, J.M.; Culbard, E.R.; Harvey, P.G.; Delves, H.T.;
Sherlock, J.C.; Smart, G.A.; Thomas, J.F.A.; Quinn, M.J. (1987) Relationship
between blood lead and lead intake in two year old urban children in the U.K.
In: Lindberg, S.E.; Hutchinson, T.C., eds. Int. Conf. on Heavy Metals in the
Environment, New Orleans, September, 1987 (EP Consultants, Ltd., Edinburgh,
pp. 203-205).
Day, J.P.; Fergusson, J.E.; Chee, T.M. (1979) Solubility and potential
toxicity of lead in urban street dust. Bull. Environ. Contam. Toxicol.
23: 497-502.
Day, J.P.; Hart, M.; Robinson, M.S. (1975) Lead in urban street dust.
Nature (London) 253: 343-345.
Diemel, J.A.L.; Brunekreef, B.; Boleij, J.S.M.; Biersteker, K.; Veenstra,
S.J. (1981) The Arnhem lead study: II. indoor pollution, and indoor/
outdoor relationships. Environ. Res. 25: 449-456.
Dietrich, K.M.; Kraff, K.M.; Bornschein, R.L.; Hammond, P.B.; Berger, 0.;
Succop, P.A*; Bier, M. (1987) Low-level fetal lead exposure effect on
neurobehavioral development in early infancy. Pediatrics 80: 721-730.
Dragovic, D. (1980) Lead in the blood of the women in childbed in the
umbilical cord of the nev£>orn children and in placenta. Acta. Riol.
Med. Ext. 5: 63-68.
Dorn, C.R.; Pierce, J.D.; Phillips, P.E.; Chase, G.R. (1976) Airborne Pb,
Cd, Zn and Cu concentration by particle size near a Pb smelter.
Atmos. Environ. 10: 443-446.
Drill, S.; Konz, J.; Mahar, H.; Morse, M. (1979) The environmental lead problem:
an assessment of lead in drinking water from a multi-media perspective.
Washington, D.C.: U.S. Environmental Protection Agency; EPA report no.
EPA-570/9-79-003. Available from NTIS, Springfield, VA; PB 296556.
Duggan, M.J. (1983) The uptake and excretion of lead by young children.
Arch. Env. Health 38: 246-247.
Duggan, M.J.; Williams, S. (1977) Lead-in-dust in city streets. Sci.
Total Environ. 7: 91-97.
Duggan, M. J.; Inskip, M.J. (1985) Childhood exposure to lead in surface
dust and soil: a community health problem. Public Health Rev. 13: 1-54.
EPA [U.S. Environmental Protection Agency] (1977) Air qua 1 ity criteria for
lead. Research Triangle Park, NC: U.S. Environmental Protection Agency,
Criteria and Special Studies Office; EPA report no. EPA-600/8-77-017.
Available from: NTIS, Springfield, VA; PB 280411.
EPA [U.S. Environmental Protection Agency] (1985a) Regulation of Fuels
and Additives; Gasoline Lead Content; Final Rule and Supplemental
Notice of Proposed Rulemaking. March 7, 1985, 40 CFR Part 80.
EPA [U.S. Environmental Protection Agency] (1985b) Costs and Benefits of
Reducing Lead in Gasoline. Final Report, Office of Policy Analysis,
Office of Policy, Planning and Evaluation, Washington, D.C.
-------
7
EPA [U.S. Environmental Protection Agency] (1989) National air quality
and emissions trends report, 1987 - Office of Air Quality Planning and
Standards, Research Triangle Park, N.C. EPA-450/4-89-001, March 1989.
EPA [U.S. Environmental Protection Agency] (1986) Air Quality Criteria for Lead.
Environmental Criteria and Assessment Office, Office of Research and Develop-
ment, Research Triangle Park, N.C. EPA 600/8-83-028 a-d, June 1986.
Ernhart, C.B.; Wolf, A.Q.; Kennaro, M.J.; Erhard, P.; Filipovich, H.F.;
Sokol, R.J. (198b) Intrauterine exposure to low levels of lead: the
status of the neonate. Arch. Environ. Health 41: 287-291.
Facchetti, S. (1985) Isotopic lead experiment - an update. Presented
at: Lead Environmental Health: the current issue, Durham, N.C.,
Duke University Medical Center.
Fahim, M.S.; Fahim, Z.; Hall, O.G. (1976) Effects of subtoxic lead levels on
pregnant women in the state of Missouri. Res. Comm. Chem. Path. Pharmacol.
13: 309-331.
Faoro, R. (1988) Monitoring and Reports Branch, Office of Air Quality
Planning and Standards. Personal communication to Jeff Cohen, July 28, 1988.
Farfel, M.R. (1985) Reducing lead exposure in children. Ann. Rev. Public
Health 6: 333-360.
Flegel, A.R.; Smith, D.R.; Elias, R.W. (1988) Lead contamination in food.
In: Nriagu, J.O.; Simmons., M.S., eds. Environmental Food Contamination;
Advances in Environmental Science and Technology. J. Wiley, New York.
Fugas, M. (1977) Biological significance of some metals as air pollutants.
Part I: Lead. U.S. EPA, Research Triangle Park. EPA 600/1-77-041.
Fugas, M.; Saric, M. (1979) Health study of a lead-exposed population.
In: Lynam, D.R.; Piatanida, L.G.; Cole, J.F., eds. Environmental lead:
proceedings of the second international symposium on environmental lead
research; December 1978; Cincinnati, OH, New York, NY: Academic Press.
Fuller, W.A. (1987) Measurement error models. J.A. Wiley Publishers, New York.
Galke, W.A.; Hammer, D.I.; Keil, J.E.; Lawrence, S.W. (1975) Environmental
determinants of lead burdens in children. In: Hutchinson, T.C.; Epstein, S.;
Page, Al. L.; Van Loon, J.; Davey, T.; eds. International conference on heavy
metals in the environment: symposium proceedings, vol. 3; October; Toronto,
ON, Canada. Toronto; ON, Canada: Institute for Environmental Studies;
pp. 53-74. See also: Washington, DC: U.S. Environmental Protection Agency;
EPA report no. EPA-600/J-78-022 1975. Available from: NTIS, Springfield, VA;
PB 283567.
Gallacher, J.E.J.; Elwood, P.C.; Phillips, K.M.; Davies, B.E.; .Jones, D.T.
(1984) Relation between pica and blood lead in areas of differing
lead exposure. Arch. Dis. Child. 59: 40-44.
Gilbert C.; Tuthill, R.W.; Calabrese, E.J.; Peters, H.A. (1979) A
comparison of lead hazards in the housing environment of lead poisoned
children versus nonpoisoned controls. J. Environ. Sci. Health A14: 145-168.
-------
8
Gross, S.B. (1981) Human oral and inhalation exposures to lead: summary of
Kehoe balance experiments. J. Toxicol. Environ. Health 8: 333-337.
Haas, T.; Wieck, A.G.; Schaller, K.H.; Mache, K.; Valentin, H. (1972)
The usual lead load in newborn infants and their mothers. Zbl. Rakt.
Hyg. I. Abt. Orig. B 155: 341-346.
Hammond, P.B.; O'Flaherty, E.J.; Gartside, P.S. (1981) The impact of air-
lead on blood-lead in man - a critique of the recent literature. Food
Cosmet. Toxicol. 19: 631-638.
Hardy, H.L.; Chamberlain, R.I.; Maloof, C.C.; Boylen, G.W., Jr.; Howell,
M.C. (1971) Lead as an environmental poison. Clin. Pharmacol. Ther. 12:
982-1002.
Harley, N.H.; Kneip, T.H. (1985) An integrated metabolic model for lead in
humans of all ages. Final report to the U.S. EPA, Contract No. B44899
with New York University School of Medicine, Dept. of Environmental
Medicine, January 30, 1985.
Harris, P.; Holley, M.R. (1972) Lead levels in cord blood. Pediatrics 49:
606-608.
Harrison, R.M. (1979) Toxic metals in street and household dusts. Sci.
Total Environ. 11: 89-97.
Harrison, R.M.; Johnston, W.R.; Ralph, J.C.; Wilson, S.J. (1985) The
budget of lead, copper and cadmium for a major highway. Sci. Total
Env. 46: 137-145.
Hartwell, T.D.; Handy, R.W.; Harris, B.S.; Williams, S.R.; Gehlbach, S.H.
(1983) Heavy metal exposure in populations living around zinc and copper
smelters. Arch. Environ. Health 38: 284-295.
Heard, M.J.; Chamberlain, A.C. (1982) Effect of minerals and food on
uptake of lead from the gastrointestinal tract in humans. Hum. Toxicol.
1: 411-415.
Heard, M.J.; Chamberlain, A.C. (1984) Uptake of Pb by human skelton and
comparative metabolism of Pb and alkaline earth elements. Health Physics
47: 857-862.
Hoffman, W.F.; Steinhauser, G.; Pohl, E. (1979) Dose calculations for the
respiratory tract from inhaled natural radioactive nuclides as a function
of age-I. Compartmental deposition, retention, and resulting dose.
Health Physics 37: 517-532.
Hryhorczuk, D.O.; Rabinowitz, M.B.; Hessl, S.M.; Hoffman, 0.; Hogan, M.M.;
Mallin, K.; Hollister, F.; Orris, P.; Berman, E. (1985) Elimination kinetics
of blood lead in workers with chronic lead intoxication. Am. J. Ind.
Med. 8: 33-42.
-------
9
Hubermont, G.; Buchet, J.P.; Roels, H.; Lauwerys, R. (1978) Placental
transfer of lead, mercury and cadmium in women living in a rural area;
importance of drinking water in lead exposure. Int. Arch. Occup.
Environ. Health 41: 117-124.
Huntzicker, J.J.; Friedlander, S.K.; Davidson, C.I. (1975) Material
balance for automobile-emitted lead in Los Angeles basin. Environ.
Sci. Technol. 9: 448-457.
Iaccarino, T.F. (1988) Comments - Draft Report, Review of the National
Ambient Air Quality Standards for Lead: Exposure Analysis Methodology
and Validation. NUS Corporation, Gaithersburg, MD. October 3, 1988. EPA
Docket A-83-22, II-D-26.
ICRP [International Committee on Radiological Protection] (1975) Report
of the Task Group on Reference Man: Report #23. Pergamon Press,
New York.
Idaho Department of Health and Welfare (1977) Shoshone Lead Health Project
Work Summary, January 1976. Division of Health, Boise, Idaho.
Ingalls, M.N.; Garbe, R.J. (1982) Ambient pollutant concentrations fron
mobile sources in microscale situations. From: Passenger car meeting;
June; Troy, MI. Warrendale, PA: Society of Automotive Engineers; SAE
Paper No. 820287.
Inskip, M.J.; Atterbury, I. (1983) The legacy of lead-based paint: potential
hazards to "do-it-yourself" enthusiasts and children. In: Proceedings
of an International Conference on Heavy Metals in the Environment:
Heidleberg. CEP Consultants, Edinburgh: Vol. 1, 286-289.
Jelinek, C.F. (1982) Levels of lead in the United States food supply.
J. Assoc. Off. Anal. Chem. 65: 942-946.
Jennett, J.C.; Wixson, B.J.; Bolter, E.; Lowsley, I.H.; Hemphill, D.D.;
Tranter, W.H.; Gale, N.L.; Purushotaman, K. (1977) Transport and
distribution around mines, and smelters. In: Lead in the environment:
a report and analysis of research at Colorado State University,
University of Illinois, and University of Missouri (W.R. Boggess, ed.);
Washington, D.C.: National Science Foundation; NSF report no. NSF/RA -
770214: pp. 135-178.
Johnson, D.E.; Tillery, J.B.; Prevost, R.J. (1975) Levels of platinum,
palladium, and lead in populations of Southern California.
Env. Health Perspect. 12: 27-33.
Johnson, J.R.; Myers, R.C. (1981) Alkaline earth metabolism: a model
useful in calculating organ burdens, excretion rates and committed
effective dose equivalent conversion factors. Radiat. Prot. Dosim.
1: 87-94.
-------
10
Johnson, T.; Wijnberg, L. (1988) Statistical analyses of lead exposure data
collected in East Helena, Montana: Draft report. PEI Associates, Inc.
Durham, N.C., January 1988.
Kehoe, R.A. (1961) The metabolism of lead in man in health and disease:
the normal metabolism of lead. (The Harben lectures, 1960). J.R. Inst.
Public Health Hyg. 24: 81-97.
Kneip, T.J.; Mallon, R.P.; Harley, N.H. (1983) Biokinetic modelling for
mammalian lead metabolism. Neurotox. 4: 189-192.
Koh, E.T.; Caples, V. (1977) Nutrition survey of black families in Clairborne
County of Southwest Mississippi: I. Food practices and dietary intake,
1973-1974. Dept. of Home Economics, Alcorn State Univ., Lorman, Miss.
Koontz, M.D.; Robinson, J.P. (1982) Population activity patterns-St. Louis study.
Env. Monitoring and Assessment 2: 197-212.
Labelle, S.J.; Lindahl, P.C.; Hinchman, R.R.; Ruskamp, J.; McHugh, K. (1987)
Pilot study of the relationship of regional roarl traffic to surface soil lead
levels in Illinois. Argonne National Laboratory, Energy and Environmental
Systems Division, Center for Transportation Research, Argonne, Illinois. ANL/ES-
Lacey, R.F.; Moore, M.R.; Richards, W.N. (1985) Lead in water, infant diet .
and blood: the G1asgow duplicate- diet study. Sci. Total Environ. 41:
235-257. -
Landrigan, P.J.; Gehlbach, S.H.; Rosenblum, R.F.; Shoults, J.M.; Candelaria,
R.M.; Barthel, W.F.; Liddle, J.A.; Smrek, A.L.; Staehling, N.W-.; Sanders,
J.F. (1975) Epidemic lead absorption near an ore smelter: the role of
particulate lead. N. Engl. J. Med. 292: 123-129.
Landrigan, P.J.; Baker, E.L. (1981) Exposure of children to heavy metals
from smelters: epidemiology and toxic consequences. Environ. Res. 25:
204-224.
Lau, W.M.; Wong, H.M. (1982) An ecological survey of lead contents
in roadside dusts and soils in Hong Kong. Environ. Res. 28: 39-54.
Laxen, D.P.H.; Harrison, R.M. (1977) The highway as a source of water
pollution: an appraisal with the heavy metal lead. Water Res. 11: 1-11.
Laxen, D.P.H.; Raab, G.M.; Fulton (1987) Children's blood lead and exposure
to lead in household dust and water—a basis for an environmental standard
for lead in dust. Sci. Total Env. 66: 235-244.
Lepow, M.L.; Bruckman, L.; Gillette, M.; Markowitz, S.; Robino, R.;
Kapish, J. (1975) Investigations into sources of lead in the environment
of urban children. Environ. Res. 10: 415-426.
Lepow, M.L.; Bruckman, L., Rubino, R.A., Markowitz, S., Gillette, M.;
Kapish, J. (1974) Role of airborne lead in increased body burden of lead
in Hartford children. Env. Health Perspect. 7: 99-102.
Lin-Fu, J.S. (1972) Undue absorption of lead among children - a new look
at an old problem. N. Engl. J. Med. 286: 702-710.
-------
X X
Madden, N.A., (1980). Environmental influrneces on mouthing in children
with lead intoxication. J. Ped. Psych. 5: 207.
Mahaffey, K.R.; Annest, J.L. (1985). Association of blood lead level, hand-
to-mouth activity and pica among young children. 69th Annual Meeting of the
Federation of American Societies for Experimental Biology. Anaheim, CA;
USA April 21-26, 1985. Fed. Proc. 44: 752.
Mahaffey, K.R.; Annest, J.L.; Roberts, J.; Murphy, R.S. (1982) National
estimates of blood lead levels: United States, 1976-1980: association
with selected demographic and socio-economic factors. N. Engl. J. Med.
307: 573-579.
Mahaffey, K.R.; Michaelson, I.A. (1980) The interaction between lead and
nutrition. In: Needleman, H.L., ed. Low level lead exposure: the clini-
cal implications of current research. New York, NY: Raven Press; pp.
159-200.
Mai Ion, R.P. (1983) A metabolic model of lead kinetics based upon measured
organ burdens during chronic exposure experiments with infant and
juvenile baboons. Doctoral Dissertation, New York University.
Manton, W.I. (1985) Total contribution of airborne lead to blood lead.
Br. J. Ind. Med. 42: 168-172.
Marcus, A.H. (1984) Testing alternative nonlinear kinetic models in
compartmental analysis. In: Eisenfield, J.; Delisi, C., eds. Proceedings',
second IMACS international symposium in biomedical systems modelling,
Amsterdam, The Netherlands: Elsevier Science, New York, pp. 259-268.
Marcus, A.H. (1985a) Multicompartment kinetic models for lead: I. bone
diffusion models for long-term retention. Environ. Res. 36: 441-458.
Marcus, A.H. (1985b) Multicompartment kinetic models for lead: II. linear
kinetics and variable absorption in humans without excessive lead exposures.
Environ. Res. 36: 459-472.
Marcus, A.H. (1985c) Multicompartment kinetic models for lead: part III.
lead in blood plasma and erythrocytes. Environ. Res.: 36: 473-489.
Marcus, A.H. (1988a) Geometric standard deviation of blood lead for
children in the vicinity of point sources. Battelle/Columbus Division,
Washington Operations, Washington, D.C.
Marcus, A.H. (1988b) Prediction of blood lead levels in East Helena children:
comparison of aggregate air lead, disaggregate, and uptake/biokinetic models.
Battelle/Columbus Division, Washington Operations, Washington, D.C.
Marcus, A.H. (1988c) Quantitative relationships betveen lead in air, dust, and
soil. Battelle/Columbus Division, Washington Operations, Washington, D.C.
Marcus, A.H. (1989) Relationship between childhood blood lead and lead in
water or diet. Battelle/Columbus Division, Arlington, MD.
Marcus, A.H.; Holtzman, A.P. (1988) Calibration of Run 1A. Memorandum
to Jeff Cohen, Office of Air Quality Planning and Standards, Battelle/
Columbus Division. April 5, 1988.
-------
12
McKeague, J.A.; Wolynetz, M.S. (1980) Background levels of minor elements
in some Canadian soils. Geoderma 24: 299-307.
McLamb, D. (1988) Estimates of the relationships between air, soil and dust
lead levels. Memorandum to Jeff Cohen and John Haines, Ambient Standards
Branch, Office of Air Quality Planning and Standards, U.S. EPA,
Research Triangle Park, January 12, 1988.
McMichael, A.J.; Vimpani, G.V.; Robertson, E.F.; Baghurst, P.A.; Clark, P.n.
(1986) The Port Pirie cohort study: maternal blood lead and pregnancy
outcome. J. Epi. Comm. Health 40: 18-25.
Mielke, H.W.; Anderson, J.C.; Berry, K.J.; Mielke, P.W.; Chaney, R.L,;
Leech, M. (1983) Lead concentrations in inner-city soils as a factor
in the child lead problem. Am. J. Pub. Health 73: 1366-1369.
Millican, F.K.; Layman, E.M.; Lourie, R.S. et al. (1962) The prevalence of
ingestion and mouthing of nonedihle substances by children. Clin. Proc. Child.
Hosp. 18: 207-214.
Millman, N.; Christensen, J.M.; Ibsen, K.K. (1988) Blood lead and erythrocyte
zinc protoporphyrin in mothers and newborn infants. Env. J. Pediatr.
147: 71-73
Moore, M.R. (1977) Lead in drinking water in soft water areas—health hazards.
Sci. Total Environ. 7: 109-115.
Moore, M.R.; Goldberg, A.; Pocock, S.J.; Meredith, A.; Stewart, I.M.;
Macanespie, H.; Lees, R.; Low, A. (1982) Some studies of maternal and
infant lead exposure in Glasgow. Scott. Med. J. 27: 113-122.
Morse, D.L.; Landrigan, P.J.; Rosenblum, B.F.; Hubert, J.S.; Houseworth, J.
(1979) El Paso revisited: epidemiologic follow-up of an environmental
lead problem. J. Am. Med. Assoc. 242: 739-741.
Motto, H.L.; Daines, R.H.; Chilko, D.M.; Motto, C.K. (1970) Lead in soils and
plants: its relationship to traffic volume and proximity to highways.
Environ. Sci. Techno!. 4: 231-238.
NAS [National Academy of Sciences] Committee on Biologic Effects of Atmospheric
Pollutants (1972) Lead: airborne lead in perspective. Washington, P.C.;
National Academy of Sciences.
NAS [National Academy of Sciences] Committee on Lead in the Human Environment
(1980) Lead in the human environment. Washington, D.C.: National
Academy of Sciences.
Nathanson, B.; Nudelman, H. (1980) Ambient lead concentrations in New York
City and their health implications. Bull. N.Y. Acad. Med. 56: 866-875.
National Food Processors Association; Can Manufacturers Institute, Inc.
(1982) Comprehensive supplementary report covering further research,
February 1980 through May 1983, on food borne lead in the diet of
children. Available from: U.S. Department of Health and Human Services,
Washington, D.C.; FDA docket no. 79N-0200.
-------
J. J
NIBS (1988) Lead Based Paint in Housing: Task Force Report to Roard of
Directors for National Institute of Building Sciences, Washington, D.C.
February 20, 1988.
Nordman, C.H. (1975) Environmental lead exposure in Finland: a study on
selected population groups. Helsinki, Finland: University of Helsinki.
Nriagu, J.O. (1978) Lead in soils, sediments and major rock types. In:
The biogeochemistry of lead in the environment. Part A: Ecological
cycles (J.O. Nriagu, ed.); Amsterdam, The Netherlands: Elsevier/
North-Holland Biomedical Press; pp. 15-72.
Nutrition Foundation, Inc. (1982) Assessment of the safety of lead and lead
salts in food: a report of the Nutrition Foundation's Expert Advisory
Committee. Washington, D.C. The Nutrition Foundation.
O'Flaherty, E.J.; Hammond, P.B.; Lerner, S.I. (1982) Dependence of apparent
blood lead half life on the length of previous lead exposure in humans.
Fundam. Appl. Toxicol. 2: 49-54.
Olson, K.W.; Skogerboe, R.K. (1975) Identification of soil lead compounds
from automotive sources. Environ-, Sci. Technol. 9: 227-230.
Page, A.L.; Ganje, T.J. (1970) Accumulation of lead in soils for regions of high
and low motor vehicle traffic density. Environ. Sci. Technol. 4; 140-142.
Pennington, J.A.T. (1983) Revision of the total diet study food list and
diets. J. Am. Diet. Assoc. 82: 166-173.
Phalen, R.F.;01dham, M.J.; Beavcage, C.R.; Crocker, T.T.; Mortenson, J.D.
(1985) Postnatal enlargement of human tracheobronchial airways and
implications for particle deposition. Anat. Rec. 212: 368-380.
Phillips, P.E.; Vornberg, D.L. (1986) Pediatric blood lead levels in Herculaneum,
Missouri-1984. Presented to the Society for Environmental Geochemistry
and Health at the 20th Annual Conference on Trace Substances in
Environmental Health, Columbia, M0, June 5, 1986.
Pickles, J.H. (1982) Air pollution estimation errors and what it does to
epidemiological analysis. Atm. Environ. 16: 2241-2245.
Pierson, W.R.; Brachaczek, W.W. (1976) Particulate matter associated with
vehicles on the road. Warrendale, PA: Society of Automotive Engineers;
SAE technical paper no. 760039. SAE transactions 85: 209-227.
Pope, A. (1985) Development of activity patterns for population exposure to
ozone, PEI Associates, Inc., Durham, N.C. for Tom McCurdy, Office of
Air Quality Planning and Standards, August 23, 1985.
Pope, A. (1986a) Exposure of children to lead-based paints, PEI Associates, Inc.
Durham, N.C. Prepared for Strategies and Air Standards Division, Office
of Air Quality Planning and Standards, January, 1986.
Pope, A. (1986b) Seasonal variations in the amount of time children spend indoors,
outdoors, and in motor vehicles. PEI Associates, Inc., Durham, N.C. for
Strategies and Air Standards Division, Office of Air Ouality Planning and
Standards, January 1986.
-------
14
Prpic-Majic, D.; Meczner, J.; Telisman, S.; Kersanc, A. (1984) Biological
monitoring of lead effects in a smelter community before and after
emission controls. Sci. Total Env. 32: 277-288.
Quah, R.F.; Stark, A.D.; Meigs, J.W.; DeLouise, E.R. (1982) Children's
blood lead levels in New Haven: A population-based demographic profile.
Env. Health perspect. 44: 159-164.
Quarles, H.D., III; Hanawalt, R.B.; Odum, W.E. (1974) Lead in small mammals,
plants and soil at varying distances from a highway. J. Appl. Ecol. 11:
937-949.
Que Hee, S.S.; Peace, B.; Clark, C.S.; Boyle, J.R.; Bornschein, R.L.; Hammond, P.B.
(1985) Evolution of efficient methods to sample lead sources, such as
house dust and hand dust, in the homes of children. Env. Res. 38: 77-95.
Rabinowitz, M.B.; Wetherill, G.W.; Kopple, J.D. (1976) Kinetic analysis of
lead metabolism in healthy humans. J. Clin. Invest. 58: 260-270.
Rabinowitz, M.B.; Wetherill, G.W.; Kopple, J.D. (1977) Magnitude of lead
intake from respiration by normal man. J. Lab. Clin. Med. 90: 238-248.
Rabinowitz, M.B.; Kopple, J.D.; Wetherill, G.W. (1980) Effect of food intake
and fasting on gastrointestinal lead absorption in humans. Am. J. Clin.
Nutr. 33: 1784-1788.
Rabinowitz, M.B.; Needleman, H.L. (1982) Temporal trends in the lead
concentrations of umbilical cord blood. Science (Washington, D.C.) 216:
1429-1431.
Rabinowitz, M.B.; Needleman, H.L. (1983) Petrol lead sales and umbilical
cord blood lead levels in Boston, Massachusetts [letter]. Lancet 1 (8314/5):
63.
Rabinowitz, M.; Leviton, A.; Needleman, H.; Bellinger, 0.; Waternaux, C. (1985)
Environmental correlates of infant blood lead levels. Env. Res. 38: 96-107.
Rabinowitz, M.; Leviton, A.; Needleman, H. (1986) Occurrence of elevated
protoporphyrin levels in relation to lead burden in infants. Environ.
Res. 39: 253-257.
Rabinowitz, M.B.; Bellinger, D.C. (1988) Soil lead-blood lead relationship
among Boston children. Bull Env. Contam. Toxicol. 41: 791-797.
Roberts, T.M.; Hutchinson, T.C.; Paciga, J.; Chattopadhyay, A.; Jervis, R.E.;
Van Loon, J.; Parkinson, D.K. (1974) Lead contamination around secondary
smelters: estimation of dispersal and accumulation by humans. Science
(Washington, D.C.) 186: 1120-1123.
Roels, H.A.; Buchet, J—P.; Lauwerys, R.; Hubermont, G.; Bruaux, P.; Claeys-
Thoreau, F.; Lafontaine, A.; Van Overschelde, J.; (1976) Impact of air
pollution by lead on the heme biosynthetic pathway in school-age children.
Arch. Environ. Health 31: 310-316.
-------
15
Roels, H.A.; Buchet, J-P.; Lauwerys, R.; Bruaux, P.; Claeys-Thoreau, F.;
Lafontaine, A.; Van Overschelde, J.; Verduyn, G. (1978) Lead and cadmium
absorption among children near a nonferrous metal plant: a follow-up
study of a test case. Environ. Res. 15: 290-308.
Roels, H.A.; Buchet, J-P.; Lauwerys, R.; Bruaux, P.; Claeys-Thoreau, F.;
Lafontaine, A.; Verduyn, G. (1980) Exposure to lead by the oral and the
pulmonary routes of children living in the vicinity of a primary lead
smelter. Environ. Res. 22: 81-94.
Rolfe, G.L., A. Haney, and K.A. Reinbold (1977) Environmental contamination
by lead and other heavy metals, Vol. 2: Ecosystem analysis; final
report. University of Illinois at Urbana-Champaign, Institute for
Environmental Studies, Urbana 111. NTIS, Springfield, VA: PC-AD/MF-A01.
Rom, W.N. (1976) Effects of lead on the female and reproduction: a review.
Mt. Sinai J. Med. 43: 544-52.
Rubinstein, E.A.; Comstock, G.A.; Murray, J.P. (1972) Television and social
behavior, Volume IV. National Institute of Mental Health, IJ.S. Dept. of
Health, Education, and Welfare, Rockville, MD.
Rutter, M. (1983) Low level lead exposure: sources, effects and implications.
In: Rutter, M.; Jenes, R. Russell, eds. Lead versus health: sources
and effects of low level lead exposure. New York, NY: John Wiley &
Sons; pp. 333-370.
Ryu, J.E.; Ziegler, E.E.; Nelson, S.E.; Fomon, S.J. (1983) Dietary intake
of lead and blood lead concentration in early infancy. Am. J. Dis.
Child. 137: 886-891.
Sachs, H.K. (1974) Effect of a screening program on changing patterns
in lead poisoning. Env. Health Perspect. 7: 41-45.
Sayre, J.W.; Charney, E.; Vostal, J.; Pless, I.B. (1974) House and hand dust
as a potential source of childhood lead exposure. Am. J. Dis. Child.
127: 167-170.
Schilling, R.J.; Bain, R.P. (1988) Prediction of children's blood lead levels
on the basis of household-specific soil lead levels. Am. J. Epidemiol.
128: 197-205.
Schmitt, N.; Philion, J.J.; Larsen, A.A., Harnadek, M.; Lynch, A.J. (1979)
Surface soil as a potential source of lead exposure for young children.
CMA J. 121: 1474-1478.
Schneider, D.J.; Lavenhar, M.A. (1986) Lead poisoning: more than a medical
problem. AJPH 76: 242-244.
Schwartz, J. (1985a) Office of Policy Analysis, U.S. EPA, Washington, D.C.
Modeling the percent of children with high blood lead levels.
Memorandum to Vic Hasselblad, Allan Marcus, Jeff Cohen, August 21, 1985.
-------
lfi
Schwartz, J. (1985b) Office of Policy Analysis, U.S. EPA, Washington, D.C.
Modeling the blood lead distribution in children. Memorandum to Jeff
Cohen, Office of Air Quality Planning and Standards, U.S. EPA, Research
Triangle Park, N.C., September 4, 1985.
Schwartz, J.D.; Janney, A.; Pitcher, H. (1984) The relationship between
gasoline lead and blood lead. Available from: Joel D. Schwartz,
U.S. EPA, Office of Policy Analysis, Washington, D.C.
Sedman (1987) The development of applied action levels for soil contact:
a scenario for the exposure of humans to soil in a residential setting.
State of California Department of Health Services, Toxic Substances
Control Division, April 1987.
Sherlock, J.; Smart, G.; Forbes, G.I.; Moore, M.R.; Patterson, W.J.;
Richards, W.N.; Wilson, T.S. (1982) Assessment of lead intakes and dose-
response for a population in Ayr exposed to a plumbsolvent water supply.
Hum. Toxicol. 1: 115-122.
Silbergeld, E.K.; Schwartz, J. (1987) Lead and osteoporosis: mobilization of
bone lead in postmenopausal women and possible etiologic role in bone
demineralization. In: Lindberg, S.E.; Hutchinson, T.C., eds. International
conference: heavy metals in the environment. New Orleans,
September, 1987. CEP Consultants Ltd., Edinburgh, p. 300.
Silvasi, J. (1985) Control Programs Development Division, Office of Air
Quality Planning and Standards, U.S. EPA, Research Triangle Park, N.C.
Personal communication.
Skogerboe, R.K. (1975) Analytical chemistry. In: Environmental
contamination caused by lead. Washington, D.C.: National Science
Foundation; pp. 187-203.
Sledge, D. (1986) Office of Air Quality Planning and Standard, U.S.
EPA, RTP, N.C. Calculation of dietary lead intake as input to the
lead exposure model. Memorandum to the lead file. September 12, 1986.
Snee, R.D. (1981) Evaluation of studies of the relationship between blood
lead and air lead. Int. Arch. Occup. Environ. Health 48: 219-242.
Snee, R.D. (1982) Silver Valley lead study: further analysis of the
relationship between blood lead and air lead. J. Air Pollut. Control
.Assoc. 32: 170-175.
Stark, A.D.; Quah, R.F.; Meigs, J.W.; DeLouise, E.R. (1982) The relationship
of environmental lead to blood-lead levels in children. Environ. Res.
27: 372-383.
Stuik, E.J. (1974) Biological response of male and female volunteers to
inorganic lead. Int. Arch. Arbeitsmed. 33: 83-97.
Succop, P.A.; O'Flaherty, E.J.; Bornschein, R.L.; Clark, C.S.; Krafft, K.;
Hammond, P.B.; Shukla, R. (1987) A kinetic model for estimating changes in
the concentration of lead in the blood of young children. In: Lindberg,
S.E.; Hutchinson, T.C., eds. International conference: Heavy Metals
in the Environment, New Orleans, September, 1987 (EP Consultants
Ltd., Edinburgh, pp. 289-291).
-------
Suter, L.E. (1979) Travel to School. U.S. Bureau of Census, Series P-20,
No. 392.
Tepper, L.B.; Levin, L.S. (1975) A survey of air and population lead levels
in selected American communities. In: Griffin, T.B.; Knelson, J.H.,
eds. Lead. New York, NY: Academic Press; pp. 152-195. (Coulston, F.;
Korte, F., eds. Environmental quality and safety: suppl. v. 2).
Ter Haar, G.; Aronow, R. (1974) New information on lead in dirt and dust as
related to the childhood lead problem. Environ. Health Perspect. 7: 83-89.
Thornton, I.; Culbard, E.; Moorcroft, S.; Watt, J.; Wheatley, M.; Thompson, M.
(1985) Metals in urban dusts and soils. Env. Techn. Letters 6: 137-144.
U.K. Central Directorate [United Kingdom Central Directorate on Environnental
Pollution] (1982) The Glasgow duplicate diet study (1979/1980): a joint
survey for the Department of the Environment and the Ministry of Agriculture
Fisheries and Food. London, United Kingdom: Her Majesty's Stationery
Office; pollution report no. 11.
U.S. FDA [Food and Drug Administration] (1983) Market basket survey:
preliminary results for lead analysis. Available for inspection at:
U.S. Environmental Protection Agency, Environmental Criteria and Assessment
Office, Research Triangle Park, N.C.
U.S. FDA [Food and Drug Administration] (1984) Market basket survey:
prel iminary results for lead analysis. Available for inspection at:
U.S. Environmental Protection Agency, Environmental Criteria and Assessment
Office, Research Triangle Park, N.C.
Urban, W.D. (1976) Analysis of blood lead levjsls of children surveyed in
Pittsburgh, Pennsylvania: analytical methodologies and summary results.
Washingtoo, D.C.: U.S. Dept. of Commerce, National Bureau of Standards;
report no. NBSIR76—1024. Available from NTIS, Springfield, VA; PR 255876.
Vornberg, D.L. (1987) St. Joe Lead Company, Personal Communication to
Gale Hoffnagle and Richard DeCesar in a report for Lead Industries
Association, Project 3952-P51, May 1987.
Walter, S. D.; Yankel, A. J.; von Lindern, I. H. (1980) Age-specific risk
factors for lead absorption in children. Arch. Environ. Health 35: 53-58.
Ware, J. (1985) Linear models for the analysis of longitudinal studies.
Am. Statist. 39: 95-101.
Wheeler, G. L.; Rolfe, G. L. (1979) The relationship between daily traffic
volume and the distribution of lead in roadside soil and vegetation.
Environ. Pollut. 18: 265-274.
White, P. (1987) Exposure Assessment Group, U.S. EPA, Washington, D.C.
Request for soil ingestion data by the Office of Solid Waste.
Memorandum to Peter Preuss, Office of Health and Environmental
Assessment. July 1987.
Williams, M. K.; King, E.; Walford, J. (1969) An investigation of lead
absorption in an electric accumulator factory with the use of personal
samplers. Br. J. Ind. Med. 26: 202-216.
-------
18
Winneke, G.; Beginn, V.; Ewert, T.; Havestadt, C., Kraemer, U.; Krause, C.;
Thron, H.L.; Wagner, H.M. (1985) Comparing the effects of perinatal and
later childhood lead exposure on neuropsychological outcome.
Environ. Res. 38: 155-167.
Wixson, B. G. (1978) Biogeochemical cycling of lead in the New Lead Belt of
Missouri. In: The biogeochemistry of lead in the environment, Part A:
Ecological cycles, J. 0. Nriago, ed. Elsevier/North-Holland Biomedical
Press, New York.
Wolnik, K. A.; Fricke, F. L.; Capar, S. G.; Braude, G. L.; Meyer, M. W.;
Satzger, R. 0.; Bonnin, E. (1983) Elments in major raw agricultural crops
in the United States: 1. Cadmium and lead in lettuce, peanuts, potatoes,
soybeans, sweet corn, and wheat. J. Agric. Food Chem. 31: 1240-1244.
Worth, D.; Matranga, A.; Lieberman, M.; DeVos, E.; Karelekas, P.; Ryan, C.;
Craun, C. (1981) Lead in drinking water: the contribution of household
tap water to blood lead levels. In: Lynam, 0. R.; Piantanida, L. G.;
Cole, J. F., eds. Environmental lead: proceedings of the second international
symposium on environmental lead research; December 1978; Cincinnati, OH.
New York, NY: Academt Press; pp. 199-225.
Yankel, A. J.; V) n Lindern, I. H.; Walter, S. D. (1977) The Silver Valley
lead study: the relationship of childhood lead poisoning and environmental
exposure. J. Air Pollut. Control Assoc. 27: 763-767.
Zetterlund, B.; Winberg, J.; Lundgren, G.; Johansson, G. (1977) Lead in
umbilical cord blood lead correlated with the blood lead of the mother
in areas witPI low, medium or high atmospheric pollution. Acta.
.Paediatr. Scand. 66: 169-175.
Ziegler, E. E.; Edwards, B. 8.; Jensen, R. L.; Mahaffey, R. R.; Fomon, S.J.
(1978) Absorption and retention of lead by infants. Pediatr. Res. 12:
29-34.
Zielhuis, R. L.; del Castilho, P.; Herber, R. F. M.; Wibowo, A. A. E.; Salle,
H. J. A. (1979) Concentrations of lead and other metals in blood of two and
three year-old children living near a secondary smelter. Int. Arch. Occup.
Environ. Health 42: 231-239.
Zimdahl , R. L.; Hassett, J. J. (1977) Lead in soil. In: Lead in the
Environment: A report and analysis of research at Colorado State
University, University of Illinois, and University of Missouri (W.R.
Boggess, ed.); Washington, D.C.: National Science Foundation: NSF report
no. NSF/RA-770214; pp. 93-98.
Zurlo, N.; Griffini, A. M. (1973) Lead contents in foods and beverages
consumed in Milan. In: Proceedings, international symposium on the
environment! health aspects of lead; October, 1972; Amsterdam, The
Netherlands. Luxembourg: Commission of The European Communities;
pp. 93-98.
------- |