DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
An Approach for Estimating Changes in Children's IQ from Lead
Dust Generated during Renovation, Repair, and Painting in
Residences and Child-Occupied Facilities
DRAFT FOR CASAC REVIEW ON JULY 9-10, 2007
Prepared By:
Office of Pollution Prevention and Toxics
U.S. Environmental Protection Agency
Washington, D.C.
June 8, 2007
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
DISCLAIMERS
Mention of product or trade name does not constitute endorsement or recommendation for use by
the U.S. Environmental Protection Agency.
This information is distributed solely for the purpose of pre-dissemination peer review under
applicable information quality guidelines. It has not been formally disseminated by the U.S.
EPA. As such, it does not represent and should not be construed to represent any agency
determination or policy.
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
TABLE OF CONTENTS
1. Scope of the Document
2. Neurocognitive Effects of Pb in Children: Pb and IQ
2.1 Blood Pb as a Biomarker
2.2 Neurocognitive Effects of Pb in Children
2.2.1. Epidemiologic Studies on Neurocognitive Effects of Pb in Children ....
2.3 Influence of Timing of Exposure
2.4 Summary of Data: Pb and IQ in Children
3. Development of Background and Activity-related Inputs
3.1 Exposure Periods and Phases for the RRP Activities
3.1.1 Indoor Air Exposures
3.1.2 Indoor Dust Exposures
3.1.3 Outdoor Soil Exposures
3.2 Overview of Approaches for Estimating Air, Indoor Dust, and Outdoor Soil
Concentrations
3.2.1 Air
3.2.2 Indoor Dust
3.2.3 Outdoor Soil
3.3 Background Pb Levels by Media Type
3.3.1 Air
3.3.1.1 Data Sources
3.3.1.2 Ambient Air
3.3.1.3 Indoor Air
3.3.2 Indoor Dust
3.3.2.1 Data Sources
3.3.2.2 Development of Values
3.3.3 Outdoor Soil
3.3.3.1 Data Sources
3.3.3.2 Development of Values
3.4 Activity-Related Inputs by Media Type
3.4.1 Data Sources
3.4.1.1 EFSS
3.4.1.2 NAHBLSWP
3.4.2 Information Regarding the OPPT Dust Study
3.4.3 Single Activity Example
3.4.3.1 Indoor Air
3.4.3.2 Indoor Dust
3.4.3.3 Outdoor Soil
3.4.4 Multiple Activities Example
3.4.4.1 Air
3.4.4.2 Indoor Dust
3.4.4.3 Outdoor Soil
13
13
13
i
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
3.5 Uncertainties and Limitations 47
4. Estimates of Media Concentrations 49
4.1 Air 51
4.1.1 Approach 53
4.1.1.1 Pre-renovation 53
4.1.1.2 Renovation 53
Calculate Whole Building Activity Concentrations during the Dust Generating Phase
54
Calculate Whole Building Concentrations across Activities during the Dust Generating
Phase 55
Calculate Concentrations during the Settling Phase 55
Calculate Concentrations for Background Phase 57
4.1.1.3 Post-renovation 57
4.1.2 Example for Single RRP Activity 58
4.1.3 Example for Multiple RRP Activities 58
4.2 Indoor Dust 58
4.2.1 Approach 61
4.2.1.1 Pre-renovation 61
Convert Wipe Loadings to Vacuum Loadings 61
Convert Loadings to Concentrations 62
4.2.1.2 Renovation 62
Convert Wipe Loadings to Vacuum Loadings 63
Calculate Whole Building Activity Loadings 64
Calculate Whole Building Total Loadings 65
Convert Loadings to Concentrations 66
4.2.1.3 Post-renovation 66
Calculate Change in Loadings Based on Routine Cleaning 67
Convert Loadings to Concentrations 68
4.2.2 Example for Single RRP Activity 69
4.2.3 Example for Multiple RRP Activities 71
4.3 Outdoor Soil 73
4.3.1 Pre-renovation 75
4.3.2 Renovation 75
ii
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Calculate Concentrations in Different Regions of Yard 75
Calculate Whole Yard Average Concentration 76
4.3.3 Post-renovation 76
4.4 Drinking Water 76
4.5 Diet 77
4.6 Sensitivity and Monte Carlo Analyses 77
4.6.1 Overview 77
4.6.2 Sensitivity Analysis 78
4.6.2.1 Approach 78
4.6.2.2 Results - Sensitivity Analysis Excluding Cleaning Efficiency 80
4.6.2.3 Results - Cleaning Efficiency Sensitivity Analysis 80
4.6.3 Monte Carlo Analysis 81
4.6.3.1 Approach 81
4.6.3.2 Monte Carlo Results 82
4.7 Uncertainties and Limitations 84
5. Estimates of Blood Pb Levels 87
5.1 Selected Model(s) 88
5.1.1 Ilil BK Model 88
5.1.2 Leggett Model 89
5.2 Derivation of Blood Pb Estimation Inputs 91
5.2.1 Time Series of Media Concentrations: Exposure to Renovation Activity at Different
Ages 91
5.2.2 Parameter Inputs Related to Pb Intake and Absorption (Uptake) 93
5.2.2.1 Ilil BK Model 93
5.2.2.2. Leggett Model 97
5.3 Estimation of Blood Pb Levels for Two RRP Examples 98
5.3.1 IEUBK Model Implementation 98
5.3.2 LeggettModel 98
5.4 Examples 98
5.4.1 Estimated Blood Pb Levels Associated With Background Exposures 98
5.4.2 Blood Pb Level Estimation for the Single and Multiple RRP Acitivity Examples ..100
5.4.3 Ways to Compare Blood Pb Changes in Association with Control Options 101
5.5 Uncertainties and Limitations 106
6. Estimates of IQ Change 108
6.1 Selected Model (Lanphear et al. 2005) 108
6.2.1 Estimated IQ Changes Associated With Background Exposures 110
6.2.2 Estimated IQ Changes for the Single and Multiple RRP Activity Examples Ill
6.2.3 Ways to Compare IQ Changes in Association with Control Options 114
6.3 Uncertainties and Limitations 120
7. Discussion of Uncertainties and Their Impact on Results 121
7.1 Input Data and Assumptions 121
Hi
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
7.2 Modeling Approaches 122
7.3 Conclusions 123
8 References 124
iv
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
LIST OF EXHIBITS
Exhibit Page
Exhibit 3-1. Crosswalk of Activity Names Used in the Examples and Job Descriptions from the
OPPT Dust Study 20
Exhibit 3-2. Exposure Periods and Associated Phases for Indoor Dust, Indoor Air, and
Outdoor Soil 22
Exhibit 3-3. Indoor Air Exposure Periods and Associated Phases 23
Exhibit 3-4. Indoor Dust Exposure Periods and Associated Phases 25
Exhibit 3-5. Outdoor Soil Exposure Periods and Associated Phases 27
Exhibit 3-6. Flow Chart of Steps Used in this Approach to Calculate Indoor Air
Concentrations 30
Exhibit 3-7. Flow Chart of Steps Used in this Approach to Calculate Indoor Dust
Concentrations 31
Exhibit 3-8. Flow Chart of Steps Used in this Approach to Calculate Outdoor Soil
Concentrations 32
Exhibit 3-9. Background Pb Loadings in Floor Dust Used in this Approach By Building
Vintage 34
Exhibit 3-10. Background Soil Pb Concentrations Used in this Approach by Building
Vintage 35
Exhibit 3-11. Control Options Used this Approach and the Phases Used in the OPPT Dust
Study 39
Exhibit 3-12. Sampling Locations in the OPPT Dust Study and Areas of Building or Yard They
Represent in This Approach 40
Exhibit 3-13. Exposure Periods and Phases of This Approach and OPPT Dust Study Sample
Types Used for Each 40
Exhibit 3-14. Crosswalk Between the Loadings Used in this Approach and the Loadings
Provided in the OPPT Dust Study for Outdoor Soil 45
Exhibit 4-1. Dimension Types, Names, and IDs Used in this Approach 50
v
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Exhibit 4-2. Approach for Estimating Indoor Air Concentrations 52
Exhibit 4-3. Indoor Air Concentrations for Single Indoor Activity Example 58
Exhibit 4-4. Indoor Air Concentrations for Multiple Indoor Activities Example 58
Exhibit 4-5. Approach for Estimating Indoor Dust Concentrations 60
Exhibit 4-6. Whole-House Indoor Dust Concentrations for the Single RRP Activity Example for
the Different Renovation Phases 70
Exhibit 4-7. Indoor Dust Concentrations Time Series for the Single Activity Example, <1940
Vintage 71
Exhibit 4-8. Whole-House Indoor Dust Concentrations for the Multiple Activities Example for
the Different Renovation Phases 72
Exhibit 4-9. Indoor Dust Concentrations Time Series for the Single Activity Example and
<1940 Vintage 73
Exhibit 4-10. Approach for Estimating Outdoor Soil Concentrations 74
Exhibit 4-11. Summary of Non-Water Dietary Pb Intake Estimates 77
Exhibit 4-12. Inputs Included in Sensitivity Analysis 79
Exhibit 4-13. Input Variables Used in the Monte Carlo Analysis 82
Exhibit 4-14. Monte Carlo Percentiles for Week 0 Concentration for the < 1940 Single Activity
Example 83
Exhibit 4-15. Monte Carlo Percentiles for Week 0 Concentration for the < 1940 Multiple
Activities Example 84
Exhibit 5-1. Flow Chart Illustrating the Approach to Calculate Blood Pb Levels 87
Exhibit 5-2. Definition of Modeled Hypothetical Children and the Time in Their Life When
They Experience the RRP Activity 92
Exhibit 5-3. IEUBK Parameter Input Values Used in this Approach 94
Exhibit 5-4. Comparison of Background Lifetime Average Blood Pb Levels from the Leggett
Model, IEUBK Model, and the NHANES Survey in |j,g/dL 99
Exhibit 5-5. Background Lifetime Average Blood Pb Levels for All Indoor Dust Concentration
Percentiles in |j,g/dL 99
VI
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Exhibit 5-6. Lifetime Average Blood Pb Levels for Child 2, Single Activity Example in |j,g/dL..100
Exhibit 5-7. Lifetime Average Blood Pb Levels for Child 2, Multiple Activity Example in
Hg/dL 101
Exhibit 5-8. Incremental Changes in Lifetime Average Blood Pb Levels due to the Renovation
Activity for Child 2, Single Activity Example in |j,g/dL 102
Exhibit 5-9. Incremental Changes in Blood Pb Levels due to the Renovation Activity for Child
2, Multiple Activities Example in |j,g/dL 103
Exhibit 5-10. Time Series of Incremental Blood Pb Changes Due to the RRP Activity for the <
1940 Vintage, Single Activity Example 104
Exhibit 5-11. Time Series of Incremental Blood Pb Changes Due to the RRP Activity for the
<1940 Vintage, Multiple Activities Example 105
Exhibit 5-12. Changes in Blood Pb for each Control Option Compared to the Base Control
Option for the Median Indoor Dust, Single Activity Example in |j,g/dL 105
Exhibit 5-13. Changes in Blood Pb for each Control Option Compared to the Base Control
Option for the Median Indoor Dust, Multiple Activities Example in |j,g/dL 106
Exhibit 6-1. Flow Chart Illustrating the Estimation of IQ Changes from Blood Pb Levels 108
Exhibit 6-2. IQ Reductions Due to Background Exposure Concentrations Calculated from Blood
Pb Levels from the IEUBK Model 110
Exhibit 6-3. IQ Changes Due to Background Exposure Concentrations Calculated from Blood
Pb Levels from the IEUBK Model 110
Exhibit 6-4. IQ Changes for Child 2 Using Blood Pb Values from the IEUBK Model for the
Single Activity Example Ill
Exhibit 6-5. IQ Changes for Child 2 Using Blood Pb Values from the Leggett Model for the
Single Activity Example 112
Exhibit 6-6. IQ Changes for Child 2 Using Blood Pb Values from the IEUBK Model for the
Multiple Activities Example 113
Exhibit 6-7. IQ Changes for Child 2 Using Blood Pb Values from the Leggett Model for the
Multiple Activities Example 114
Exhibit 6-8. Incremental Changes in IQ Due to the Renovation Activity for Child 2 Using Blood
Pb Levels from the IEUBK Model for the Single Activity Example 115
VJJ
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Exhibit 6-9. Incremental Changes in IQ Due to the Renovation Activity for Child 2 Using Blood
Pb Levels from the Leggett Model for the Single Activity Example 116
Exhibit 6-10. Incremental Changes in IQ Due to the Renovation Activity for Child 2 Using
Blood Pb Levels from the IEUBK Model for the Multiple Activities
Example 117
Exhibit 6-11. Incremental Changes in IQ Due to the Renovation Activity for Child 2 Using
Blood Pb Levels from the Leggett Model for the Multiple Activities
Example 118
Exhibit 6-12. Comparsion of IQ Changes for Child 2 Across Control Options for the
Median Case Using Blood Pb Levels from the IEUBK Model for the Single
Activity Example 119
Exhibit 6-13. Comparsion of IQ Changes for Child 2 Across Control Options for the
Median Case Using Blood Pb Levels from the Leggett Model for the Single
Activity Example 119
Exhibit 6-14. Comparsion of IQ Changes for Child 2 Across Control Options for the Median
Case Using Blood Pb Levels from the IEUBK Model for the Multiple Activities
Example 119
Exhibit 6-15. Comparison of IQ Changes for Child 2 Across Control Options for the Median
Case Using Blood Pb Levels from the Leggett Model for the Multiple Activities
Example 119
VJJJ
-------
DRAFT FOR CASAC REVIEW ON JULY 9-10, 2007
1.0 SCOPE OF THE DOCUMENT
The U.S. Environmental Protection Agency (EPA) has proposed new requirements to reduce
exposure to lead hazards created by renovation, repair, and painting activities that disturb lead-
based paint in housing and child-occupied facilities. The Federal Register Notice for the Lead
Renovation, Repair, and Painting (LRRP) proposed rule for residences is available at
http://edocket.access.gpo.gov/2006/06-71.htm and the Supplemental Proposed Rule on Child-
Occupied Facilities is available at http://www.epa.gov/lead/pubs/renovation.htm. This action
supports the Federal government's goal of eliminating childhood lead poisoning by 2010. The
proposed rule establishes requirements for training renovators and dust sampling technicians;
certifying renovators, dust sampling technicians, and renovation firms; accrediting providers of
renovation and dust sampling technician training; and for renovation work practices. These
requirements apply to "target housing," where any child under age 6 resides. "Target housing" is
defined in section 401 of the Toxic Substances Control Act (TSCA) as any housing constructed
before 1978, except housing for the elderly or persons with disabilities (unless any child under
age 6 resides or is expected to reside in such housing) or any 0-bedroom dwelling. Initially the
rule would apply to all renovations for compensation performed in target housing where a child
with an increased blood lead level resides, and rental and owner-occupied target housing built
before 1960, unless the person performing the renovation obtains a statement signed by the
owner-occupant that the renovation will occur in the owner's residence and that no child under
age 6 resides there. EPA has proposed to phase in the applicability of this proposal to all rental
target housing and owner-occupied target housing built in the years 1960 through 1977 where a
child under age 6 resides.
The Supplemental Proposed Rule on Child-Occupied Facilities is intended to add child-occupied
facilities to the universe of buildings covered by the initial proposal. A child-occupied facility
will be defined as a building, or a portion of a building, constructed prior to 1978, visited
regularly by the same child, under six years of age, on at least two different days within any
week (Sunday through Saturday period), provided that each day's visit lasts at least three hours
and the combined weekly visits last at least six hours, and the combined annual visits last at least
sixty hours. Child-occupied facilities may be located in public or commercial buildings or in
target housing. In public or commercial buildings, the child-occupied facility will only
encompass those only those areas that are routinely used by children under age six, such as
restrooms and cafeterias, and the adjacent exterior walls. Areas that children under age six only
pass through, such as hallways, stairways, and garages, will not be included.
The EPA is currently developing the final LRRP rule. This LRRP rulemaking is based on the
TSCA Title IV statutory requirement that EPA revise the lead abatement regulations to apply to
renovation and remodeling activities that create lead-based paint hazards. The statutory term
"lead-based paint hazards" includes, but is not limited to any condition that causes exposure to
lead from lead-contaminated dust and lead-contaminated soil. EPA has promulgated quantitative
lead-based paint hazard standards. Thus, a primary consideration in developing the regulations
is the extent to which the lead-based paint hazards resulting from renovation and remodeling
activities are eliminated. This is different from other TSCA rulemakings in which a "no
unreasonable risk" determination must be made. However, as with other rulemakings, which are
determined to be "significant regulatory action[s]" under Executive Order 12866 - Regulatory
9
-------
DRAFT FOR CASAC REVIEW ON JULY 9-10, 2007
Planning and Review, EPA is required to conduct an economic analysis of the costs and benefits
associated with the rulemakings.
To support the benefits assessment in the economic analysis, EPA's Office of Pollution
Prevention and Toxics (OPPT) has developed the draft document entitled "An Approach for
Estimating Changes in Children's IQ from Lead Dust Generated during Renovation, Repair and
Painting in Residences and Child-Occupied Facilities". The quantified benefits analysis will be
based primarily on changes in neurocognitive function in children (as measured by IQ) due to
lead exposure from specific renovation, repair and painting (RRP) activities. OPPT is using data
from a variety of sources, including the American Housing Survey (US Census Bureau, 1997,
2003) and the Property Owners and Managers Survey (US Census Bureau, 1995), to determine
the specific types and frequencies of RRP activities that occur in residences and child-occupied
facilities. There are obviously many types and ranges of RRP activities that can occur in any
given residence or child-occupied facility. Thus, one residence may have a kitchen remodeled,
another may have one room repainted, and another may have multiple activities such as window
replacements, painting, and a kitchen and bathroom remodeled. OPPT is currently examining
the types and number of RRP activities or combination of activities that may occur in U.S.
residences and child-occupied facilities, although the exact number has not yet been determined.
The number is anticipated to be in the hundreds.
In support of the economic analysis, it is necessary to have a scientifically sound approach for
estimating changes in children's IQ from lead exposure due to a variety of RRP activities in
residences and child-occupied facilities. The purpose of this document is to outline the general
approach that will be used. This method can then be used to "build" all the houses and child-
occupied facilities required for the economic analysis.
This document describes an approach for estimating lead exposures and resulting changes in IQ
for children under age 6 that could result from various RRP activities. The approach is designed
to characterize lead exposures in residences and child-occupied facilities, with and without the
various control options that are under consideration for the final rule. The general steps involved
in estimating changes in neurocognitive function in children from lead exposure due to RRP
activities are shown in the following flow chart.
Estimate dust lead generated from specific renovation activity(s)
I
Convert dust lead loadings to dust lead concentrations
I
Estimate blood lead
I
Estimate IQ change
10
-------
DRAFT FOR CASAC REVIEW ON JULY 9-10, 2007
Two examples are provided to illustrate the approach presented in this document. The first is for
a residence with a single RRP activity, window replacement. The second example is for a
residence with multiple RRP activities. Data from the U.S. Census Bureau were examined to
identify a likely combination of RRP activities for this example. The example for multiple RRP
activities describes a residence with eight RRP activities, including a bathroom renovation, a
kitchen renovation, 10 door or window replacements, interior painting, HVAC work, electrical
wiring work, plumbing work, and installation of a security system.
In this document, estimates of dust lead loading are made for the two examples, the house with
the single RRP activity (window replacement), and the house with eight RRP activities.
Estimates are made for each example with and without the requirements of the LRRP proposed
rule. The dust lead loadings are then converted to dust lead concentrations. For each of the two
examples, a distribution of blood lead levels is estimated for children under age 6. Finally, the
distribution of IQ change due to the resultant lead exposure is characterized for the two
examples. The examples used in this document are only two examples of a variety of activity
combinations. The two examples used in this document illustrate the approach that will be used
in the economic analysis. An extensive analysis of the health effects associated with lead
exposure is provided in EPA's recently released final Air Quality Criteria Document (AQCD)
for lead (US EPA, 2006). The AQCD is used as the source for the information on lead and
neurocognitive function in children that is presented in Section 2 of this document.
There are multiple aspects to the step of estimating lead in any specific renovation activity. Two
biokinetic models were used to estimate blood lead levels in children. These include the IEUBK
model (US EPA, 2005a), and the Leggett model (Leggett, 1993). The following schematic
expands the previous flow chart to show these steps.
11
-------
DRAFT FOR CASAC REVIEW ON JULY 9-10, 2007
The examples provided in this document are for residences. The input data are selected to
represent typical housing in the U.S. expected to have lead paint. The variability in these data is
characterized to the extent possible and analyzed using sensitivity and Monte Carlo techniques.
The age of the housing (i.e., vintage) is an important consideration when collecting these input
data since the concentration of lead in paint will vary by the age of the housing and housing
component. When data stratified on vintage are available, separate input values can be used to
characterize exposure media concentrations for different vintages. The vintages considered in
this document include houses built prior to 1940, houses built between 1940 and 1959, and
houses built between 1960 and 1979. Although no specific examples of child-occupied facilities
are provided in this document, the general approaches used to estimate exposure media
concentrations will be similar to the approaches used for residences, with some slight
modifications.
12
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
2. NEUROCOGNITIVE EFFECTS OF PB IN CHILDREN: PB AND IQ
EPA's Final Air Quality Criteria Document (AQCD) for Pb (US EPA, 2006a) provides an
extensive review of the most recent data available on Pb. This document updates the 1986
AQCD for Pb (US EPA, 1986a), associated Addendum (US EPA, 1986b) and the 1990
Supplement (US EPA, 1990). In an effort not to duplicate efforts, pertinent sections in this
document are taken directly from the Final 2006 AQCD for lead (Pb), hereafter referred to as the
AQCD.
2.1 Blood Pb as a Biomarker
Section 4 of the AQCD presents a comprehensive discussion of the use of blood Pb as both a
biomarker of Pb body burden and a biomarker of exposure (see US EPA 2006a, Sections 4.3.1.4,
4.3.1.5, and 4.3.1.6). In addition, a summary of these sections is presented in Section 8.3.2 of the
AQCD (US EPA, 2006a). A summary of the pertinent points made in these sections follows.
Blood Pb concentration is often used in epidemiologic and toxicologic studies as an index of
exposure and body burden. The prevalence of the use of blood Pb as an exposure metric is
mainly due to the feasibility of incorporating its measurement into human studies relative to
other potential dose indicators (e.g., Pb in kidney, plasma, urine, or bone) and due to the variety
of health effects associated with Pb exposure that have been reported in the literature. The Pb is
exchanged between blood and bone and blood and the soft tissues. The exchanges between the
blood and bone vary with duration and intensity of exposure, age, and other variables.
Resorption of bone in pregnant or nursing women or in postmenopausal women (through
osteoporosis) results in a mobilization of Pb from bone to circulation. Therefore, the blood Pb
concentration measured in an individual will be determined by the recent exposure history of the
individual as well as by the long-term exposure history that gives rise to accumulated bone Pb
stores.
In comparison to adults, bone mineral is turning over much more quickly in children as a result
of growth. Therefore, changes in blood Pb concentration in children are thought to more closely
parallel changes in total body burden. Several recent studies have shown that blood Pb levels
reflect Pb exposures (Abadin and Wheeler, 1997; Lanphear et al., 1998; Succop et al., 1998).
These studies have shown that Pb levels in air, dust, and soil are reflected by blood Pb levels in
humans.
2.2 Neurocognitive Effects of Pb in Children
Emphasis in this document is placed on discussion of the neurocognitive effects of Pb exposure
in children. As described in detail in Section 6.2 of the AQCD, neurocognitive effects in
children are of particular concern due to the increasingly lower levels at which they have been
reported and the potential for lifelong impact (US EPA, 2006a). The negative influence of Pb on
neurobehavior has been reported with remarkable consistency across numerous studies of
various study designs, populations studied, and developmental assessment protocols, even
following adjustment for numerous confounding factors. Collectively, the prospective cohort
and cross-sectional studies have reported that Pb exposure affects intellectual attainment of
preschool and school age children at blood Pb levels less than 10 micrograms per deciliter
13
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
([j,g/dL) (most clearly in the 5 to 10 |ig/dL range, but possibly lower). Several studies have
reported quantitative relationships between measures of intelligence quotient (IQ) and current
blood Pb levels for children aged 2 to 11 years old. Children are particularly at risk due to
sources of exposure, mode of entry, rate of absorption and retention, and partitioning of Pb in
soft and hard tissues. As described in Section 6.2.11 of the AQCD, the neurocognitive effects
reported in children appear to persist into adolescence and young adulthood in the absence of
marked reductions in environmental exposure to Pb (US EPA, 2006a). Excessive accumulation
of Pb in childhood has latent and/or persistent adverse health effects on both the peripheral and
central nervous systems of human adults assessed 19 to 29 years later (see US EPA, 2006a;
Section 8.4.2.8). Also, studies in humans and animals suggest that chelation treatment can
transiently decrease blood Pb levels—but it does not appear to reverse or ameliorate Pb-induced
cognitive deficits or behavioral problems (see US EPA, 2006a, Sections 5.3.5, 6.2.11, and 8.5.2).
Given the strength of evidence for association with blood Pb levels below 10 ug/dL, the strength
of the dose-response information at these exposure levels, and the persistence of cognitive effects
in the absence of reduced exposure levels; neurocognitive impact, specifically decrement in IQ in
young children, is the focus of the approach outlined in this document.
2.2.1. Epidemiologic Studies on Neurocognitive Effects of Pb in Children
As described in Section 6.2.3 of the AQCD, the most widely used measure of cognitive function
in epidemiologic studies is the IQ score (US EPA, 2006a). An IQ score is a global measure
reflecting the integration of numerous behavioral processes. The effects of Pb on human
neurocognitive ability have been assessed in epidemiologic studies largely by use of age-
appropriate, standardized IQ tests. A broad net approach using global assessments of cognition,
such as IQ, has proven to be the most consistently sensitive across studies of various design and
sample characteristics. Such measures combine subscales that are representative of a broad
number of underlying cognitive functions; thus, they are likely to pick up exposure-related
deficits across cohorts that differ in their functional expressions of toxicity. Global measures of
IQ have also been used so widely because of their outstanding psychometric properties.
Although the definition of "intelligence" is quite abstract, IQ remains a useful outcome measure
as it is correlated with important measures of life success, such as academic achievement,
earnings, and social status.
Epidemiologic studies linking Pb exposure to health effects are presented in detail in Chapter 6
of the AQCD, and Chapter 5 presents the toxicologic data (US EPA, 2006a). Synthesis of the
most salient health-related findings and conclusions are presented in Chapter 8 of the AQCD (US
EPA, 2006a). Based on the AQCD, it is evident that neurotoxic effects in children are of great
public health concern (US EPA, 2006a). There are numerous epidemiologic studies, as well as
extensive experimental animal evidence substantiating the plausibility of these epidemiologic
findings, that demonstrate inverse relationships between children's blood Pb concentrations and
IQ.
The studies identified in the AQCD that are most relevant to the approach outlined in this
document are Canfield et al. (2003) and Lanphear et al. (2005) (US EPA, 2006a). These large,
well-conducted studies provide both qualitative and quantitative evidence of neurocognitive
deficits, measured by IQ, in children at blood Pb levels less than 10 (j,g/dL. Two other studies,
14
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Kordas et al. 2006 and Tellez-Rojo et al. 2006, also indicate detrimental effects on specific
cognitive abilities (but not specifically IQ) in children with blood Pb levels below 10 (j,g/dL.
Descriptions of the Canfield et al. (2003) and Lanphear et al. (2005) studies, taken from Sections
6.2.3.1.9 and 6.2.3.1.11 of the AQCD, are presented below (US EPA, 2006a).
The Rochester prospective study, initiated in 1994, examined the relationship between blood Pb
levels and IQ at 3 and 5 years of age in 172, predominantly African-American, lower
socioeconomic status (SES) children (Canfield et al., 2003). Participants were enrolled when
children were 5 to 7 months of age in what was originally a study of Pb dust control methods
(Lanphear et al. 1999). Blood Pb concentrations were assessed at 6-month intervals until 2 years
and annually thereafter. No data were available on prenatal exposure. The measure of IQ was
the abbreviated Stanford-Binet Intelligence Scale-4th Edition (SBIS-4). Potential confounders
assessed included gender, birth weight, iron status, Home Observation for Measurement of the
Environment (HOME) Inventory scores, maternal IQ, SES, and tobacco use during pregnancy.
Blood Pb concentrations in the Rochester cohort (Canfield et al., 2003) were quite low for an
urban population as this study was conducted after public health measures to reduce blood Pb
levels in children were already having a dramatic impact in the United States (U.S.) population.
Blood Pb levels peaked at 2 years of age (mean 9.7 ^ig/dL). The mean lifetime average blood Pb
concentration was 7.7 (J,g/dL at the age of 3 years and 7.4 (J,g/dL at the age of 5 years. At 5 years
of age, 56% of the children had a peak blood Pb concentration below 10 (j,g/dL. Following
adjustment for covariates, there were significant inverse associations with full scale IQ at both 3
and 5 years of age for all blood Pb variables, including lifetime average up to age of behavioral
assessment.
The effect of Pb on IQ was estimated in all children using lifetime average, peak, concurrent, and
average in infancy (6 to 24 months of age) blood Pb levels. The effects of Pb on IQ for the
subgroup of children whose peak Pb concentration never exceeded 10 |ig/dL also was estimated.
Covariate-adjusted changes in IQ for each 1 (j,g/dL increase in blood Pb concentration for all
children and children with peak blood Pb concentrations below 10 |ig/dL were estimated. In all
cases, the effect estimates were larger in the subsample of children with peak blood Pb
concentrations below 10 (j,g/dL. For example, the overall estimate including all children
indicated that an increase in the lifetime average blood Pb concentration of 1 (J,g/dL was
associated with a decrease of 0.46 points [95% Confidence Interval (CI): 0.15, 0.76] in IQ. In
comparison, a 1 |ig/dL increase in lifetime average Pb concentration was associated with a
decline of 1.37 points (95% CI: 0.17, 2.56) in children with peak blood Pb concentrations below
10 |ig/dL. In an accompanying editorial of the Canfield et al. (2003) study, Rogan and Ware
(2003) noted that the steepness in the concentration-response relationship below 10 (J,g/dL might
have been influenced by 10 children with blood Pb concentrations at or below 5 |ig/dL and IQs
above 115. However, they added that it was unlikely that the associations reported by Canfield
et al. were solely due to these values. Regression diagnostics performed by Canfield et al.
identified only one potential outlier (a child who had a low IQ and low Pb concentration);
however, this value was retained in all analyses as it did not pass the discordancy test. In the
Rochester study, the relationship between children's IQ score and their blood Pb level was found
to be nonlinear. A semiparametric analysis indicated a decline of IQ of 7.4 points for a lifetime
average blood Pb concentration of up to 10 (J,g/dL, while for levels between 10 to 30 (j,g/dL a
more gradual decrease of approximately 2.5 points IQ was estimated. The authors concluded
15
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
that the most important aspect of their findings was that effects below 10 (j,g/dL that have been
observed in previous cross-sectional studies (e.g., Chiodo et al., 2004; Fulton et al., 1987; and
Lanphear et al., 2000) were confirmed in this rigorous prospective longitudinal investigation.
Lanphear et al. (2005) reported on a pooled analysis of seven prospective studies that were
initiated prior to 1995. The analysis involved 1,333 children with complete data on confounding
factors that were essential in the multivariable analyses. The participating sites included Boston,
Massachusetts; Cincinnati, Ohio; Cleveland, Ohio; Rochester, New York; Mexico City; Port
Pirie, Australia; and Kosovo, Yugoslavia. A prospective cohort study conducted in Sydney,
Australia was not included because the authors were unable to contact the investigators (Cooney
et al. 1989 and 1991). The sample size of 175 for children aged 7 years in the Sydney cohort and
the wide confidence intervals of the effect estimates, as implied by the lack of significant
associations, indicate that the non-availability of this study is unlikely to influence the results of
the pooled analysis by Lanphear et al. (2005).
The primary outcome measure was full scale IQ measured at school age (mean age at IQ testing
was 6.9 years). All children were assessed with an age-appropriate version of the Wechsler
scales. Four measures of Pb exposure were examined: concurrent blood Pb (blood Pb level
closest in time to the IQ test), maximum blood Pb level (peak blood Pb measured at any time
prior to the IQ test), average lifetime blood Pb (mean blood Pb from 6 months to the concurrent
blood Pb test), and early childhood blood Pb (defined as the mean blood Pb from 6 to 24
months). A pooled analysis of the relationship between cord blood Pb levels and IQ also was
conducted in the subsample for which cord blood Pb tests were available.
Multivariate regression models were developed adjusting the effect of blood Pb for site as well
as assessing ten common covariates likely to be confounders of the relationship between Pb and
cognitive development, including HOME Inventory scores, birth weight, maternal education and
IQ, and prenatal substance abuse. A thorough statistical analytic strategy was employed to
determine the linearity or nonlinearity of the relationship between blood Pb levels and full-scale
IQ. Regression diagnostics also were performed to ascertain whether Pb coefficients were
affected by collinearity or influential observations. The fit of all four measures of postnatal
blood Pb levels was compared using the magnitude of the model square of the correlation
coefficient (R2). The blood Pb measure with the largest R2 (adjusted for the same covariates)
was nominated a priori as the preferred blood Pb index relating Pb exposure to IQ in subsequent
inspections of the relationships. The primary analysis was done using a fixed-effects model,
although a mixed model treating sites as random effects was also examined.
The median lifetime average blood Pb concentration was 12.4 |ig/dL (5th to 95th percentile, 4.1
to 34.8) with about 18% of the children having peak blood Pb levels below 10 (j,g/dL. The 5th to
95th percentile concurrent blood Pb levels ranged from 2.4 to 30 (J,g/dL. The mean IQ of all
children was 93.2 [Standard Deviation (SD) 19.2] but this varied greatly between studies. All
four measures of postnatal exposure were highly correlated. However, the concurrent blood Pb
level exhibited the strongest relationship with IQ, as assessed by R2. Nevertheless, the results of
the regression analyses for all blood Pb measures were very similar. Multivariable analysis
resulted in a six-term model including log of concurrent blood Pb, study site, maternal IQ,
HOME Inventory scores, birth weight, and maternal education.
16
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Various models, including the linear model, cubic spline function, the log-linear model, and the
piece-wise linear model, were investigated in this analysis. The shape of the dose-response
relationship was determined to be non-linear; the log-linear model was found to provide the
strongest relationship for the data. Using the log-linear models, the authors estimated a
decrement of 1.9 points (95% CI: 1.2, 2.6) in full scale IQ for a doubling of concurrent blood
Pb. However, the IQ point decrements associated with an increase in blood Pb from below 1 to
10 |ig/dL compared to 10 to 20 |ig/dL were 6.2 points (95% CI: 3.8, 8.6) versus 1.9 points (95%
CI: 1.2, 2.6). The individual effect estimates for the seven studies used in the pooled analysis
also generally indicate steeper slopes in studies with lower blood Pb levels compared to those
with higher blood Pb.
Ernhart (2006) expressed the concern that one study site was driving the results and that the
HOME Inventory score was not always measured with the IQ test. Other limitations were also
mentioned, such as the use of capillary finger stick for the early blood Pb tests rather than venous
blood Pb samples. Lanphear et al. (2006) noted that though they agree that using an early
measure of the HOME Inventory in the Rochester cohort was a potential limitation, excluding
this cohort, from the pooled analysis changed the coefficient by less than 3%. Sensitivity
analyses reported in Lanphear et al. (2005) indicated that no single study was responsible for the
estimated relationship of Pb and deficits in IQ; thus, diminishing concerns about unique
attributes or potential limitations for any specific sites.
In summary, the log-linear model in Lanphear et al. (2005) estimated a decline of 6.2 points in
full scale IQ for an increase in concurrent blood Pb levels from less than 1 to 10 |ig/dL. This
effect estimate was comparable to the 7.4 point decrement in IQ for an increase in lifetime mean
blood Pb levels up to 10 |ig/dL observed in the Rochester study (Canfield et al., 2003), as well as
other studies presented in the AQCD (US EPA, 2006a).
2.3 Influence of Timing of Exposure
Epidemiological studies investigating blood Pb and IQ effects have considered various blood Pb
metrics, including but not limited to blood Pb levels concurrent to the time of the IQ
measurement taken, average over the "lifetime" of the child at measurement, peak or maximum
levels at a specific age range, and early childhood concentrations (usually the mean
concentration for 6 to 24 months of age). All of these blood Pb metrics have been correlated
with IQ measurements.
Available studies do not provide a definitive answer to the question of whether Pb-associated
neurodevelopmental deficits are the result of exposure during a circumscribed critical period or
of cumulative exposure. Although support can be cited for the conclusion that it is exposure
within the first few postnatal years that is most important in determining long-term outcomes
(Bellinger et al., 1992), other studies suggest that concurrent blood-Pb level is as predictive, or
perhaps more predictive, of long-term outcomes than are early blood-Pb levels (Canfield et al.,
2003; Dietrich et al,. 1993 a, b; Tong et al., 1996; Wasserman et al., 2000). Because of the
complex kinetics of Pb, an accumulative toxicant, it is extremely difficult to draw strong
conclusions from these observational studies about windows of heightened vulnerability in
children. The high degree of intra-individual "tracking" of blood Pb levels over time, especially
among children in environments providing substantial, chronic exposure opportunities (e.g.,
17
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
residence near a smelter or in older urban dwellings in poor repair), poses formidable obstacles
to identifying the time interval during which exposure to Pb caused the health effects measured
in a study. It could be that damage occurred during a circumscribed period when the critical
substrate was undergoing rapid development, but that the high correlation between serial blood
Pb levels impeded identification of the special significance of exposure at that time.
Under such circumstances, an index of cumulative blood Pb level or concurrent blood Pb level,
which might be a good marker of overall body burden under conditions of relatively steady-state
exposure, could be expected to bear the strongest association with the effect. Under these
circumstances, however, it would not necessarily be correct to conclude that it was the later
exposures, incurred around the time that the effect was detected, that were responsible for
producing the effect. While some observations in children as old as adolescence indicate that
exposure biomarkers measured concurrently are the strongest predictors of late outcomes, the
interpretation of these observations with regard to critical windows of vulnerability remains
uncertain. Additional research will be needed to distinguish effects that reflect the influence of
later Pb exposures from effects that reflect the persistence of effects resulting from exposure
during some prior critical window. Resolving this issue solely on the basis of data from
observational studies will be difficult due to the high intercorrelation among blood Pb measures
taken at different ages.
2.4 Summary of Data: Pb and IQ in Children
As stated in the AQCD, the effects of Pb on neurobehavior have been reported with remarkable
consistency across numerous studies of various designs, populations studied, and developmental
assessment protocols (US EPA, 2006a). The negative impact of Pb on IQ and other
neurobehavioral outcomes persists in most recent studies following adjustment for numerous
confounding factors including social class, quality of caregiving, and parental intelligence.
Moreover, these effects appear to persist into adolescence and young adulthood in the absence of
marked reductions in environmental exposure to Pb. In addition, although there are no direct
animal tests parallel to human IQ tests, "in animals a wide variety of tests that assess attention,
learning, and memory suggest that Pb exposure results in a global deficit in functioning, just as it
is indicated by decrements in IQ scores in children" (US EPA, 2006a).
Neurotoxic effects in children are among those best substantiated as occurring at blood-Pb
concentrations as low as 5 to 10 (j,g/dL (or possibly lower). Consistently, several recent
epidemiologic studies have observed significant Pb-induced IQ decrements in children with peak
blood Pb levels below 10 (j,g/dL (e.g., Canfield et al., 2003 and Lanphear et al., 2005) and, in
some cases possibly below 5 (j,g/dL (Bellinger and Needleman, 2003 and Tellez-Rojo et al.,
2006). An international pooled analysis of seven prospective studies with a total of 1,333
children estimated a decline of 6.2 IQ points for an increase in concurrent blood Pb levels from
less than 1 to 10 |ig/dL.
Due to the strength of evidence for association with blood Pb levels below 10 ug/dL, and the
strength of the dose-response information at these exposure levels, neurocognitive impact,
specifically decrement in IQ in young children, is the focus of this approach.
18
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
3. DEVELOPMENT OF BACKGROUND AND ACTIVITY-RELATED
INPUTS
This chapter presents exposure data that can be used to describe various RRP activities included
in the proposed rule. It provides the exposure data used to calculate background exposure values
for various media and also presents the method used to estimate lead content in these media. It
also describes the source of data for RRP activities and how these data can be used to "build"
houses undergoing any desired single or multiple RRP activity. These methods can then be used
to "build" all the houses required for the benefits analysis. The chapter provides the context
necessary to understand how background and activity-related inputs are developed (Sections 3.1
and 3.2), describes the approach for characterizing background conditions (Section 3.3), and
describes the approach for characterizing the activity-related inputs (Section 3.4). Throughout
this chapter and subsequent chapters, two "real world" examples are provided to illustrate how
this approach would be used to develop these background and activity-related inputs for a single
activity and for multiple activities.
The data presented in Exhibit 3-1 for the two examples are drawn from the Draft Final Report on
Characterization of Dust Lead Levels after Renovation, Repair, and Painting Activities (Battelle,
2007), hereafter referred to as the OPPT Dust Study. The approach described in this document
uses data from the OPPT Dust Study to estimate the comparative lead exposure impact of each
type of RRP activity. For each activity, the relevant exposure media are identified - indoor dust
and indoor air for inside activities and outdoor soil for outside activities. For some of these
activities, it is possible that there are contributions to Pb concentrations from both indoor and
outdoor sources. In particular, it is known that a significant proportion of indoor lead dust is due
to outdoor dust and soil. The approach is limited in this regard because the OPPT Dust Study
data only include Pb loadings for either a limited selection of air, dust, and soil for each job. This
is recognized as a limitation in using these data.
19
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Exhibit 3-1. Crosswalk of Activity Names Used in the Examples
and Job Descriptions from the OPPT Dust Study
THIS APPROACH
OPPT DUST STUDY ACTIVITY
Activity Name
Job Description
Experiment Number
Single Activity Example- Window Replacement
Window replacement
(Window Replacement)
Window replacement (x1)
Window replacement #1 (Experiment #s
41-44)
Window replacement #2 (Experiment #s
9-12)
Multiple Activities Example- Kitchen Renovation, Bathroom Renovation, 10 Door or Window
Replacements, Interior Painting, HVAC, Wiring, Plumbing, & Security System
Renovating kitchen
(Kitchen Renovation)
(Bathroom Renovation)
Kitchen gut (x 2)
Kitchen gut #1 (Experiment #s 49-51, 76)
Kitchen gut #2 (Experiment #s 67-70)
Window replacements
(Window Replacements)
(Door Replacements)
Window replacement
(x10)
Window replacement #1 (Experiment #s
41-44)
Window replacement #2 (Experiment #s
9-12)
Interior flat component LBP
removal, scraping
(Interior Painting)
Dry scrape (x1)
Dry scrape #1 (Experiment #s 5-8)
Dry scrape #2 (Experiment #s 26-29)
Cut-outs
(HVAC)
(Wiring)
(Plumbing)
(Security System)
Cut-outs (x4)
Cut-outs #1 (Experiment #s 22-25)
Cut-outs #2 (Experiment #s 45, 46, 71,
72)
In this approach, exposures to the following media are considered:
• Air (ambient and indoor)
• Indoor dust
• Outdoor soil
• Diet
• Drinking water
Exposures to drinking water and diet are assumed to be unaffected by RRP activities and thus are
characterized using reasonable, national-scale default values (as discussed in Sections 4.4 and
4.5, respectively). The characterization of exposures for air, indoor dust, and outdoor soil (which
are considered potentially impacted by RRP activities) involves first developing an
understanding of typical background conditions and estimating the magnitude and timing of Pb
dust generated by RRP activities.
20
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
The approach is designed to characterize Pb exposures in two types of buildings: residences
with children under six years of age and COFs. For the purposes of this approach, a COF is
defined as a building, or a portion of a building, constructed prior to 1978, visited regularly by
the same child, under age 6, on at least two different days within any week, provided that each
day's visit lasts at least 3 hours and the combined weekly visit lasts at least 6 hours, and the
combined annual visits last at least 60 hours. Examples of COFs are daycare centers, preschools,
and kindergarten classrooms. As for residences, the variability in COF input data is
characterized to the extent possible and analyzed using sensitivity and Monte Carlo analyses. No
data sufficient to characterize the age (vintage) of daycare centers or schools in the U.S. have
been identified and thus vintage is not considered in the approach for COFs.
The overall approaches used to estimate exposure media concentrations for each building type
are identical; however, to the extent possible, separate input values are applied for each type.
For residences, input data are selected from the OPPT Dust Study to represent typical housing in
the U.S. expected to have leaded paint. The variability in these data is characterized using
variability estimates from the OPPT Dust Study and analyzed using sensitivity and Monte Carlo
analyses. The age of housing (i.e., vintage) is an important consideration when collecting these
input data because lead concentrations vary with the age of housing and housing component.
When data stratified on vintage are available, separate input values should be used to
characterize exposure media concentrations for different vintages. The vintages considered in
this approach include houses built prior to 1940, houses built between 1940 and 1959, and
houses built between 1960 and 1979.
For the LRRP rule, several control options are being considered for indoor and outdoor RRP
activities. They were examined in developing this approach.
The following four combinations of these control methods are included in this approach for
indoor activities:
• No plastic, baseline cleaning (Control Option 0)
• No plastic, rule cleaning (Control Option 1)
• Rule plastic, baseline cleaning (Control Option 2)
• Rule plastic, rule cleaning (Control Option 3, proposed rule requirements)
The following two control combinations are included in this approach for outdoor activities:
• No plastic (Control Option A)
• With plastic (Control Option B, proposed rule requirements)
The control options are described in more detail in Section 3.4.2.
21
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
3.1 Exposure Periods and Phases for the RRP Activities
The first step in developing an approach for characterizing air, indoor dust, and outdoor soil
concentrations is to develop an understanding of how Pb concentrations in these media are
expected to change over time. Based on this understanding, the exposure duration can be
divided into discrete components that can be characterized separately and then combined to
define the concentrations over the course of the exposure. This section describes how these
temporal components were developed for this approach.
The exposure duration is assumed to be six years, which covers the entire range of children's
ages addressed by the RRP rule. This exposure duration is divided into three exposure periods:
Pre-renovation, Renovation, and Post-renovation. The Pre-renovation exposure period
represents the period of exposure before initiation of the RRP activity or activities and thus
consists of background contributions only. The Renovation exposure period represents the
period of exposure beginning with the initiation of the RRP activity or activities and concluding
with the completion of the renovation and any contractor cleaning. The Post-renovation
exposure period represents the period of exposure following the renovation and any contractor
cleaning and ending when the child reaches six years of age. Each exposure period can be
subdivided into phases, which can differ across indoor air, indoor dust, and outdoor soil (see
Exhibit 3-2). Within some exposure periods, multiple phases may occur within a medium
(indoor dust, indoor air, and outdoor soil), and some coincide with similar phases in the other
media. The phases for each of the media are described in greater detail in the following sections.
Exhibit 3-2. Exposure Periods and Associated Phases
for Indoor Dust, Indoor Air, and Outdoor Soil
Exposure Period
Phase
Indoor Dust
Indoor Air
Outdoor Soil
Pre-renovation
Pre-renovation
(Background)
Pre-renovation
(Background)
Pre-renovation
(Background)
Renovation
(Dust Generating)
Renovation
(Dust Generating)
Renovation
(Renovation)
Renovation
Renovation
Renovation
(Settling)
(After Baseline Cleaning)
Renovation
(Background)
Post-Renovation
Post-Renovation
(Routine Cleaning)
Post-Renovation
Post-Renovation
Post-Renovation
(Background)
(Background)
(Post-Renovation)
3.1.1 Indoor Air Exposures
In the approach for indoor air exposures, the Pre-renovation and Post-renovation exposure
periods both consist of only Background phases. The Renovation exposure period consists of
three phases: Dust Generating; Settling; and Background. In this approach, exposure
concentrations are estimated for each phase separately.
22
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
During the Pre-renovation (Background) phase, exposures are represented by the estimated
constant background indoor air concentration. During the Renovation (Dust Generating) phase,
exposure concentrations are represented by the estimated Pb concentrations during the portion of
the renovation where RRP activities are creating leaded dust (e.g., during demolition). During
the Renovation (Settling) phase, exposure concentrations are represented by the estimated Pb
concentrations during the period of time following completion of the Dust Generating activities
during which the dust in air settles onto the floor. During the Renovation (Background) phase,
exposure concentrations are represented by the estimated Pb concentrations during the period of
time from the end of the Settling phase until the end of the Renovation period. During the Post-
renovation (Background) phase, exposure concentrations are represented by background
concentrations. Exhibit 3-3 provides a detailed graph of the indoor air exposure periods and
phases used in the approach.
Exhibit 3-3. Indoor Air Exposure Periods and Associated Phases
PRE-RENOVATION
RENOVATION POST-RENOVATION
1 II
II 1
<— Dust generating
Settling
phase
Background
Background Background
\
Time
23
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
3.1.2 Indoor Dust Exposures
In the approach for indoor dust exposures, the Pre-renovation exposure period consists of only a
Background phase. The Renovation exposure period is divided into two phases: Dust
Generating and After Baseline Cleaning. The Post-renovation exposure period is divided into
two phases: Routine Cleaning and Background phases. Just as for indoor air, exposure
concentrations are estimated for each phase separately in this approach.
During the Pre-renovation (Background) phase, exposures are represented by the estimated
constant background indoor dust concentrations. During the Renovation (Dust Generating)
phase, exposure concentrations are represented by the estimated Pb concentrations during the
portion of the renovation when RRP activities are creating leaded dust (e.g., during demolition).
Exposure concentrations for the Renovation (After Baseline Cleaning) phase are characterized
using Pb loading data collected after the baseline cleaning (i.e., the one-time cleaning conducted
by the contractor immediately following completion of the RRP activity) and before any routine
cleaning is performed. The Post-renovation (Routine Cleaning) phase begins at the end of the
Renovation exposure period and continues until the Post-renovation (Routine Cleaning) dust
concentrations have decreased due to routine cleaning performed by the building occupant to
background dust concentrations. The initial Pb concentration for the Post-renovation (Routine
Cleaning) phase is determined based on the Control Option being evaluated, with the Renovation
(After Baseline Cleaning) concentration used if only baseline cleaning is performed and Post-
verification cleaning concentration used if both baseline and verification cleanings are performed
(see Section 3.4.2 for more explanation of the different cleaning Options). During the final
phase, Post-renovation (Background), exposure concentrations are represented by the estimated
background concentration. Exhibit 3-4 provides a graph of the indoor dust exposure periods and
phases used in this approach.
24
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Exhibit 3-4. Indoor Dust Exposure Periods and Associated Phases
without verification cleaning with verification cleaning
25
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
3.1.3 Outdoor Soil Exposures
In the approach for outdoor soil exposures, none of the exposure periods consists of more than
one phase. For the Pre-renovation (Background) phase, exposures are represented by the
estimated constant background outdoor soil concentrations. For the Renovation period, exposure
concentrations are represented by the estimated Pb concentrations during the activity. Note that
for Control Options where plastic sheeting is used, exposure from outdoor soil can decrease
slightly during this period. This occurs because the Pb loadings associated with the RRP
activities typically result in lower concentrations than background and the plastic sheeting
prevents exposures to the underlying soil during this period. Although the Pb concentration in
outdoor soil does not actually decrease, the Pb exposure that occurs during contact with outdoor
soil can be lower during this period.
For the Post-renovation period, exposure concentrations are represented by the estimated Pb
concentrations following the completion of the activity and any removal of Pb associated with
the relevant Control Option (e.g., removal of plastic with Pb-containing debris). Note that for
Control Options that do not include plastic sheeting (i.e., Base Control Option and Control
Option 1), Renovation period concentrations will be the same as Post-renovation concentrations.
Unlike for indoor dust and air, there are no assumed loss processes that result in a reduction of
the Post-renovation soil concentrations (i.e., no cleaning) and thus soil concentrations are
assumed to remain at Post-renovation phase levels for the remainder of the exposure period.
Exhibit 3-5 provides a graph of the outdoor soil exposure periods and phases used in this
approach.
26
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Exhibit 3-5. Outdoor Soil Exposure Periods and Associated Phases
PRE-RENOVATION
RENOVATION
POST-RENOVATION
o-
o
si
Q.
O
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
3.2 Overview of Approaches for Estimating Air, Indoor Dust, and Outdoor Soil
Concentrations
This section presents summaries of the approaches for estimating air, indoor dust, and outdoor
soil concentrations. These summaries provide an understanding for how the background and
activity-related inputs developed as described in Sections 3.3 and 3.4 are used to estimate
exposure media concentrations.
3.2.1 Air
Concentrations in both ambient (i.e., outdoor) and indoor air are required for characterizing Pb
exposures. In this approach, it is assumed that RRP activities do not contribute to ambient air Pb
concentrations and thus ambient air concentrations are characterized using only background
contributions. Conversely, indoor air concentrations are estimated by considering the
contributions of both background and RRP activity-related sources. Exhibit 3-6 provides an
overview of how indoor air concentrations are characterized in this approach. Note that the steps
in black are described in Sections 3.4 and 3.5; the steps in grey are described in Chapter 4. This
flowchart provides the steps involved in characterizing indoor air concentrations for a multiple
RRP activity scenario. For single activity scenarios, the step labeled "Average concentrations
across all activities" would not be required. For multiple activity scenarios, the steps in the left
column prior to "Average concentrations across all activities" would be repeated for all activities
and these individual results would be combined in this step to characterize the overall impacts
from all of the involved activities.
3.2.2 Indoor Dust
Indoor dust concentrations are estimated in this approach by considering the contributions of
both background and RRP activity-related sources including from exterior activities. Exhibit 3-7
provides an overview of how indoor dust concentrations are characterized. Note that the steps in
black are described in Sections 3.4 and 3.5; the steps in grey are described in Chapter 4. This
flowchart provides the steps involved in characterizing indoor dust concentrations for a multiple
RRP activity scenario. For single activity scenarios, the step labeled "Sum loadings across all
activities and add background" would consist of summing the Pb loadings for background and
the single activity. For multiple activity scenarios, the steps in the left column prior to "Sum
loadings across all activities and add background" would be repeated for all activities and these
individual results would be combined in this step to characterize the overall impacts from all of
the involved activities.
3.2.3 Outdoor Soil
In the approach presented in this document, outdoor soil concentrations are estimated by
considering the contributions of both background and RRP activity-related sources. Exhibit 3-8
provides an overview of how outdoor soil concentrations are characterized. Note that the steps
in black are described in Sections 3.4 and 3.5; the steps in grey are described in Chapter 4. This
flowchart provides the steps involved in characterizing outdoor soil concentrations for a multiple
RRP activity scenario. For single activity scenarios, the step labeled "Sum concentrations across
all activities and add background" would consist of summing the Pb concentrations for
28
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
background and the single activity. For multiple activity scenarios, the steps in the left column
prior to "Sum concentrations across all activities and add background" are repeated for all
activities and these individual results are combined in this step to characterize the overall impacts
from all of the involved activities.
29
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Exhibit 3-6. Flow Chart of Steps Used in this Approach to Calculate Indoor Air Concentrations
30
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Exhibit 3-7. Flow Chart of Steps Used in this Approach to Calculate Indoor Dust Concentrations
31
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Exhibit 3-8. Flow Chart of Steps Used in this Approach to Calculate Outdoor Soil
Concentrations
OPPT Dust Study
(Battelle 2007)
Activity 1 ...n
Activity Loadings
By phase of renovation, location
in yard, and control option
National Survey of
Lead and Allergens
in Housing
(HUD 2002)
Calculate nationally representative background
outdoor soil concentration by building vintage
Load-to-Concentration conversion
using soil mixing depth and density
Activity 1.
Activity Concentrations
By phase of renovation, location
in yard, and control option
Calculate area-weighted
concentrations for yard
Activity 1..
Yard-Wide
Activity Concentrations
By phase of renovation
and control option
Yard-Wide
Background
Concentrations
By building vintage
_ Sum concentrations across all activities _
and add background
All activities
Yard-Wide
Total Concentrations
By phase of renovation, building
vintage, and control option
Yard-Wide
Total Concentrations
Time Series for before, during, and
after renovation by building vintage,
and control option
32
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
3.3 Background Pb Levels by Media Type
A child's exposure to Pb over the first six years of life consists of exposure to both Pb released
as a result of RRP activities and background Pb concentrations. Therefore, background levels of
Pb in indoor air, indoor dust, and outdoor soil represent an important component of total Pb
exposure for children. It is necessary to characterize these background levels for accurate
estimation of blood Pb levels and to allow for the determination of the portion of the blood Pb
levels attributable to RRP activities under different Control options.
3.3.1 Air
3.3.1.1 Data Sources
The background Pb concentration data for ambient and indoor air used in this approach comes
from a review of the 2005 annual average total suspended particulate (TSP) monitoring data for
Pb contained in EPA's Air Quality Systems (AQS) database (USEPA 2006b).
3.3.1.2 Ambient Air
Background ambient air Pb concentrations used in this approach come from 2005 annual average
total suspended particulate (TSP) monitoring data for Pb contained in EPA's Air Quality
Systems (AQS) database (USEPA 2006b). The range of concentrations in this database is quite
large, with a 5th percentile concentration of 0.002 microgram per cubic meter (|ig/m3) and a 95th
percentile concentration of 0.37 |ig/m3. Based on these data, the median concentration (0.025
|ig/m3) was selected as the background exposure concentration. This value is likely biased high
because Pb monitors are often located in areas with nearby Pb emission sources.
3.3.1.3 Indoor Air
No representative background indoor air concentration data were located for this approach. As a
result, it was assumed that these concentrations would equal the background ambient air Pb
concentration (0.025 |ig/m3) presented in Section 3.3.1.2. It is unclear whether this assumption
would tend to positively or negatively bias blood Pb level results since no comparative indoor air
concentration data were found. The assumption is a recognized limitation of this approach.
33
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
3.3.2 Indoor Dust
3.3.2.1 Data Sources
The National Survey of Lead and Allergens in Housing (NSLAH) (HUD 2002) provided the Pb
loading data that are used to estimate background indoor dust Pb concentrations used in this
approach. The NSLAH data set was selected from a number of potential studies, which are
described in the Risk Analysis to Support Standards for Lead in Paint, Dust, and Soil (USEPA
1998), including the HUD National Survey of Lead-Based Paint in Housing (NSLBPH), HUD
Grantees Evaluation of HUD Lead-Based Paint Hazard Control Grant Program ("HUD
Grantees"), Lead-Based Paint Abatement and Repair & Maintenance (R&M) Study in Baltimore,
and the Rochester Lead-in-Dust Study. The HUD (2002) data set was selected based on a study
design that provides data that are representative of all housing groups throughout the U.S. and
focused on homes with children (HUD 2002). It is also the largest and most recent survey
completed that used wipe samples in accordance with ASTM E1728-95 (USEPA 1998).
3.3.2.2 Development of Values
The HUD (2002) Pb dust loading data used in this approach are limited to weighted values for
those houses containing Pb based paint (LBP), and they were broken down into three building
vintages (< 1940, 1940 to 1959, and 1960 to 1979). The values are weighted to reflect the
sampling stages with the goal of summing to the U.S. housing stock that does not exclude
children. Values below the limit of detection are handled in the same manner as described in
HUD (2002), with values that were below 0.375 microgram per square feet (|ig/ft2) set equal to
0.375 jug/ft2, which is equal to one quarter of the detection limit. Three values for background
indoor dust levels were developed. The Mid value is used in deterministic simulations
(described in Chapter 4) to characterize a "best estimate." The Low and High values were used
to develop input distributions that are used in the probabilistic simulations. The low, mid, and
high background indoor Pb dust loading values used in this approach, which are shown in
Exhibit 3-9, are the 5th, 50th, and 95th percentile values of the data set.
Exhibit 3-9. Background Pb Loadings in Floor Dust Used in This Approach bv Building Vintage
Building Vintage
Pb Loadings (ng/ft2)
Low
Mid
High
<1940
0.1
0.6
5.7
1940 to 1959
0.1
0.3
4.1
1960 to 1979
0.1
0.2
1.7
3.3.3 Outdoor Soil
3.3.3.1 Data Sources
The HUD (2002) report served as the source of the background outdoor soil Pb concentration
data used in this approach. A number of alternative data sets were considered, including the
HUD National Survey of Lead-Based Paint in Housing (NSLBPH) (HUD 2002), HUD Grantees
34
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Evaluation of HUD Lead-Based Paint Hazard Control Grant Program ("HUD Grantees"), Lead-
Based Paint Abatement and Repair & Maintenance (R&M) Study in Baltimore, Rochester Lead-
in-Dust, and a number of other studies discussed in Chapter 3 of U.S. EPA's 2000 supplemental
report (USEPA 2000). The HUD (2000) data set was chosen because it is the most recent large
survey of soil Pb concentrations that is designed to be representative of all housing groups
throughout the U.S. and focus on homes with children (HUD 2002).
3.3.3.2 Development of Values
The HUD (2002) soil concentration data used in this approach are limited to values for houses
containing LBP, with their sampling weights, and they are broken down into three building
vintages (< 1940, 1940 to 1959, and 1960 to 1979). Values below zero were handled according
to the manner described in HUD (2002), with values that were below 0 microgram per gram
(|ig/g) set equal to 5 jug/g. Three values for background outdoor soil levels were developed. The
Mid value is used in deterministic simulations (described in Chapter 4) to characterize the "best
estimate." The Low and High values were used to develop distributions for this input that are
used in the probabilistic simulations. The low, mid, and high background outdoor Pb soil
concentration loading values used in this approach, which are shown in Exhibit 3-10, are the 5th,
50th, and 95th percentile values of the data set.
Exhibit 3-10. Background Soil Pb Concentrations Used
in This Approach by Building Vintage
Building Vintage
Soil Pb Concentrations (ng/g)
Low
Mid
High
<1940
40.6
426.8
3,225.8
1940 to 1959
31.2
119.6
1,680.4
1960 to 1979
7.5
90.5
1,332.1
3.4 Activity-Related Inputs by Media Type
In order to create a full picture of Pb exposure for a child, it is necessary to examine Pb levels
generated during RRP activities along with the background Pb levels discussed in Section 3.3. A
RRP event can consist of a single activity (e.g., replacing a window) or several activities (e.g.,
renovating a kitchen and replacing several windows). The goal of this section is to describe the
methods used in this approach to build the activity-related inputs. The methods applied to
develop these inputs for the single activity and multiple activities examples are provided in
Sections 3.4.3 and 3.4.4, respectively. These sections describe the methods using two "real
world" examples. Unlike Chapter 4, separate sections do not describe the approach and
examples because the methods used to develop these inputs are much simpler than those
described in Chapter 4 and they are more easily presented through the use of examples.
The two examples are indoor events that do not involve Pb in soil tracked in from the outdoors.
The approach will be augmented to account for Pb in tracked-in soil when the approach is
applied for an outdoor RRP event.
35
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
In this approach, it is assumed that indoor air, indoor dust, and outdoor soil are the three media
types that contain increased levels of Pb as the result of RRP activities. It is a recognized
limitation of this analysis that the Pb concentration of outdoor air is assumed to be unaffected by
RRP activities. Although the tendency of this limitation would be to underestimate total
exposures, the increased ventilation in the outdoor environment as compared to indoor
ventilation would likely reduce the effect.
3.4.1 Data Sources
The OPPT Dust Study (Battelle 2007) provides the data used in this approach to determine the
activity-related Pb levels. Two other data sources were also considered for the analysis: U.S.
EPA's Lead Exposure Associated with Renovation and Remodeling Activities: Environmental
Field Sampling Study (hereafter referred to as "EFSS") (USEPA 1997) and the National
Association of Home Bui 1 ders's Lead-Safe Work Practices Survey Project Report (hereafter
referred to "NAHB LSWP") (NAHB 2006). The OPPT Dust Study was selected over these
studies because it provides the combinations of activities and control options that were necessary
to analyze the RRP rule. Additional detail on the reasons the other studies were not chosen is
provided below.
3.4.1.1 EFSS
The EFSS study was not used for this approach primarily because its use requires a significant
increase in the use of assumptions that have minimal supporting data when compared to use of
the OPPT Dust Study. Use of the EFSS necessitates the "building" of activities by estimating
the number of tasks (e.g., sawing, drilling, etc.) required to complete each activity. Data
supporting the development of these assumed equivalencies are quite limited. Such assumptions
also mask the ways in which activities performed sequentially (e.g., drilling in a piece of wood
followed by sawing that wood) might produce different levels of dust compared to performing
those tasks separately. For the OPPT Dust Study, the activities of interest were performed in
their entirety.
In addition, the EFSS did not provide measurements of the effectiveness of the different Control
Options that were being evaluated, which necessitated the use of highly uncertain assumptions
regarding control efficiencies. The EFSS also did not provide loading estimates after cleanings
were completed, which would have led to reliance on uncertain assumptions regarding cleaning
efficiency. The data provided in the OPPT Dust Study were specific to the Control Options of
interest for this approach and provided post-work and post-cleaning measurements of Pb dust
that allowed analysis of the efficacy of contractor cleaning (Battelle 2007).
3.4.1.2 NAHB LSWP
The objective of the NAHB LSWP study was to measure the amount of Pb dust generated during
"typical" RRP activities and assess whether routine RRP activities increased Pb dust levels in the
work area and property. This study was not used to supplement the OPPT Dust Study due to
four key limitations.
Properties included in the NAHB LSWP study contained significantly higher initial (Pre-
renovation) Pb levels (by 1 to 2 orders of magnitude) than the background indoor dust levels (see
36
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Section 3.3.2). In the NAHB LSWP study, unlike the OPPT Dust Study, the properties were not
cleaned prior to the start of the activity, which resulted in pre-work dust levels being higher than
post-work levels for 89 percent of the RRP activities evaluated. This makes it difficult to
estimate what the exposures would have been if the properties did not have significant dust
levels at the beginning of the study.
For both the NAHB LSWP and OPPT Dust Study, post-work dust levels were collected after
RRP activities were completed. In the OPPT Dust Study, these samples were collected both
after the work was completed (before any cleanup) and following the cleaning. In the NAHB
LSWP study, post-work samples were only collected after the final cleanup was performed, so
there is no measurement of potential dust loadings associated with particular activity types and
no information on cleaning efficiencies. Such measurements and information are necessary in
order to be able to apply the approach described in this document.
The NAHB LSWP did not use the same range of work practice Control Options as the OPPT
Dust Study, which limited its utility in evaluating the exposure scenarios of interest. Three types
of work practice Control Options were evaluated in the NAHB LSWP study: routine, modified
LSWP, and EPA/HUD LSWP. The routine Control Option could be assumed to be similar to the
Base Control Option (baseline cleaning, no plastic sheeting) in this approach and the EPA/HUD
LSWP Controls could be assumed to be similar to the Control Option 3 (rule cleaning, plastic
sheeting), but the modified LSWP Controls are not clearly similar to any of the Control Options
assessed in this approach. The NAHB LSWP study also examined the EPA/HUD LSWP
practices infrequently, with only 5 of the 60 events1 evaluated using the EPA/HUD LSWP
practices.
For NAHB LSWP, a total of eight interior activities were evaluated, but these RRP activities
differed from the interior activities that were evaluated in the OPPT Dust Study and this
approach (e.g., no door planing was evaluated). Other interior jobs were similar (e.g., window
replacement, door replacement, kitchen work), but they were not necessarily performed in a
similar manner. Finally, there were no exterior jobs evaluated in NAHB LSWP, so no exterior
soil or air samples were available.
3.4.2 Information Regarding the OPPT Dust Study
The OPPT Dust Study (Battelle 2007) provides the activity-related Pb concentrations and
loadings, and thus it is critical to understand how its structure relates to the structure of this
approach. This section presents three principal relationships that help clarify how this approach
uses data from the OPPT Dust Study.
The first relationship is between the control options discussed in this approach and the phases
used in the OPPT Dust Study. The relationship is summarized for indoor and outdoor data in
Exhibit 3-11. It is important to note that the OPPT Dust Study did not have separate experiments
for the two outdoor Control Options (with and without plastic sheeting) in this approach. The
OPPT Dust Study did, however, provide loading data for above and below the plastic sheeting,
1 An event is defined as an RRP activity conducted with one of the work practice Control Options in place.
37
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
and these data were used to estimate separate results for the two outdoor Control Options as
described in Section 3.4.3.3.
In order to describe this relationship, it is necessary to understand the control options that are
being considered for the RRP rule (and therefore examined in this approach) for indoor and
outdoor activities.
For indoor activities, there are three methods for controlling lead released during RRP activities:
baseline cleaning; rule cleaning; and plastic sheeting. Each of these methods is described briefly
below.
• Baseline Cleaning - A series of steps used to remove dust that is representative of current
standard cleaning practices used by RRP contractors (i.e., ones that are used now in the
absence of a RRP rule). These steps include sweeping and vacuuming with a non-HEPA
vacuum. (Battelle 2007)
• Rule Cleaning - The cleaning required by under the proposed RRP rule. It is comprised
of HEP A vacuuming and wet-mopping with two-bucket method. (Battelle 2007)
• Plastic (interior) - Plastic coverings were set up in the workspace by the RRP contractor
according to the proposed RRP rule. This included closing and sealing all doors into the
workspace; covering all doors within the workspace that had to be used while the job was
being performed with plastic sheeting in a manner that allowed workers to pass through,
while confining dust and debris to the workspace; covering the entire floor surface with
taped-down plastic sheeting in the workspace; and taping plastic to the outside of the
window frame, if a window was being replaced as part of a job, so that no debris could
fall to the ground outside the window. The floor of the primary decontamination area
(just outside the workspace; either a hallway or tool room) was also securely covered
with plastic. (Battelle 2007)
The following four combinations of these control methods are included in this approach for
indoor activities:
• No plastic, baseline cleaning
• No plastic, rule cleaning
• Plastic, baseline cleaning
• Plastic, rule cleaning
For outdoor activities, the only control measure being considered in this approach is the use of
plastic sheeting, which is briefly described below.
38
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
• Plastic (exterior): The exact size and location of the Rule Containment Plastic took into
account any physical constraints such as property lines, fences, nearby houses, and the
placement of the vertical containment structure. The plastic was set up by the RRP
contractor in accordance with the proposed RRP rule. This included laying securely
taped- or weighed-down plastic on top of the containment plastic extending out from the
edge of the building a reasonable distance to collect falling debris. All doors and
windows within 20 feet of and below the work area were closed. (Battelle 2007)
The following two control combinations are included in this approach for outdoor activities:
• No plastic sheeting
• With plastic sheeting
Exhibit 3-11. Control Options Used in this Approach and the Phases
Used in the OPPT Dust Study
This Approach
OPPT Dust Study
Control Option
(CO) Name
Description
Control
Option
(CO) ID
Exposure Phase ID
Description
Indoor
Base Control
Option
No plastic sheeting,
Baseline cleaning
Q
Phase IV
No plastic and baseline
cleaning
Control Option 1
No plastic sheeting,
Rule cleaning
1
Phase III
No plastic and rule
cleaning
Control Option 2
Plastic sheeting,
Baseline cleaning
2
Phase II
Plastic coverings and
baseline cleaning
Control Option 3
Plastic sheeting,
Rule cleaning
3
Phase I
Plastic coverings and rule
cleaning
Outdoor
Control Option A
No plastic sheeting
A
N/A
See footnote a
Control Option B
Plastic sheeting
B
N/A
See footnote a
a Note that the OPPT Dust Study did not have two separate experiments for the two outdoor Control Options (A and B)
per se. It did provide loading data above and below the plastic sheeting, which were used to estimate separate results
for Control Options A and B as described in Section 3.4.3.3.
The second relationship is between the sampling location or area in the yard in the OPPT Dust
Study (Battelle 2007) and the area of the building or yard that those sample locations are used to
represent in this approach. For indoor sampling, samples that were taken in the Work Room in
the OPPT Dust Study were used to represent the Workspace in this approach, while the Tool
Room and Observation Room samples, respectively, were used to represent the Adjacent Room
and the Rest of Building in this approach. For outdoor sampling, OPPT Dust Study samples that
were taken on plastic or under plastic were used to represent the Dripline in this approach
(different combinations of the On Plastic and Under Plastic samples were used to represent
conditions under different Control Options as described in Section 3.4.3.3). A summary of this
relationship is provided in Exhibit 3-12.
39
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Exhibit 3-12. Sampling Locations in the OPPT Dust Study and Areas of Building
or Yard They Represent in This Approach
Location in Building Used in This Approach
Sampling Rooms in OPPT Dust Study
Workspace
Work Room
Adjacent Room
Tool Room
Rest of Building
Observation Room
Areas of Yard in This Approach
Sampling Areas in Yard in OPPT Dust Study
Dripline
On Plastic and Under Plastica
Nearby
Near Plastic
a Note that both the On Plastic and Under Plastic designations in the OPPT Dust Study indicate that the sample was taken
within the area referred to in this approach as the Dripline.
The third relationship is between the sample types reported in the OPPT Dust Study and the
exposure periods and phases in this approach for which they are used. The sample types
describe the times at which the samples were taken over the course of a particular RRP activity.
A summary of this relationship is provided in Exhibit 3-13. It is important to note that both the
Post Cleaning and the Post Verification sample types are sometimes used in this approach for the
indoor dust Post-renovation (Routine Cleaning) phase, depending on which Control Option is
being considered.
Exhibit 3-13. Exposure Periods and Phases of This Approach
and OPPT Dust Study Sample Types Used for Each
This Approach
OPPT Dust Study
Exposure Phase (PH) Name
Phase(PH)
ID
Sample Types3
Indoor Dust
Pre-renovation (Background)
1
N/A
Renovation (Dust Generating)
2
Post Work
Renovation (After Baseline
Cleaning)
3
Post Cleaning
Post-renovation (Routine Cleaning)
4
Post Cleaning or Post Verification b
Post-renovation (Background)
5
N/A
Indoor Air
Pre-renovation (Background)
i
N/A
Renovation (Dust Generating)
ii
Post Work
Renovation (Settling)
iii
N/A
Renovation (Background)
iv
N/A
Post-renovation (Background)
V
N/A
Outdoor Soil
Pre-renovation (Background)
A
N/A
Renovation
B
Post Work
Post-renovation
C
Post Work
a The sample types describe the time at which the sample was taken over the course of a particular RRP activity.
b It is important to note that both the Post Cleaning and the Post Verification sample types are used in this
approach for the indoor dust Post-renovation (Routine Cleaning) phase depending on which Control Option is
being considered, as is described in Section 3.3.2).
Please note that the remainder of this approach will primarily use the language identified in
Exhibit 3-11, Exhibit 3-12, and Exhibit 3-13 as the nomenclature of this approach. In the cases
40
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
where the nomenclature of the OPPT Dust Study is used, a DS marker will appear in parentheses
following the term (e.g., the indoor dust Post Work (DS) samples were used in this approach to
determine the indoor dust loadings for the Renovation (Dust Generating) exposure period and
phase). It may be useful to return to the three exhibits to trace the nomenclature of this approach
to the appropriate OPPT Dust Study terminology.
3.4.3 Single Activity Example
The development of Pb air concentrations, indoor dust loadings, and outdoor soil loadings
resulting from the replacement of a single window (based on OPPT Dust Study data) is presented
in this section. This information is provided as an example of the methodology used in this
approach to calculate concentrations and loadings for a single RRP activity.
3.4.3.1 Indoor Air
The OPPT Dust Study provides window replacement, Post Work (DS) Pb air concentration data
for the Workspace, Adjacent Room, and Rest of Building under the conditions of the four
Control Options. The first step to estimate activity-related indoor air concentrations is to group
samples for each unique combination of location and Control Option. The geometric mean,
minimum, and maximum concentrations from each group are used as the mid, min, and max
concentrations, respectively, in this approach. These indoor air concentrations are displayed in
Exhibit A-4 in Appendix A. The mid concentration serves as the input for the deterministic
simulations to generate "best estimate" results (as described in Chapter 4), while the min and
max provide the range for the probabilistic modeling. The geometric mean is selected as the mid
value because it corrects for positively skewed data by assuming a log normal distribution of the
data. In reviewing the concentrations, it is important to note that all data points that are below
the limit of detection are set to one half the limit of detection.
Although the OPPT Dust Study does provide Post Cleaning (DS) and Post Verification (DS)
samples, neither of these is used for indoor air in this approach. For a single activity such as a
window replacement, the entire renovation activity is assumed to happen within a one week
Renovation (Dust Generating) phase, during which the Post Work (DS) concentrations were
assumed to prevail. During the following week, which corresponds to the Renovation (Settling)
phase in this approach, the concentrations are assumed to remain at the Post Work (DS) level for
the length of the sampling period (i.e., the time over which the sample was collected), which
varied across the samples but was typically two hours. For the remainder of the Renovation
(Settling) phase, the concentrations are decreased towards background using decay coefficients
from Choe et al. (2000). Once background is reached, the concentration is assumed to remain at
that level for the duration of the Renovation (Settling) and Renovation (Background) phases. A
generalized example of how these concentrations change over the course of the exposure periods
and phases is provided in Exhibit 3-3.
One shortcoming of the data in the OPPT Dust Study is that the duration of the sampling period
was not consistent across samples. This is important, particularly during the Post Work (DS)
period, because concentrations can change relatively quickly, which means that a shorter
sampling period may not be representative of the results if the sampling period had been
extended. The selected approach deals with this by using the actual sampling period and
41
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
applying a decay function from the end of this period until the concentration reached
background. An alternative approach would have been to convert each sample into an eight-hour
average, which would normalize all of the samples to the same sampling time. This approach
was not chosen because it would have assumed that the air concentration was zero for all times
within the eight hours when the sampling was not performed, which would have likely biased the
air concentrations low.
The post-cleanup and, if applicable, post-verification measurements of Pb dust loadings are
expected to be lower than the Pb dust loadings measured post-work, because the post work
measurement occurs prior to any clean-up and/or verification. This trend should apply to all dust
samples - those collected in the Workspace, the area adjacent to the Workspace, and the Rest of
Building. The dust study data for the single activity example of window replacements, however,
support this assumption only in some cases.
The post-work measurements of Pb air concentrations are expected to be lower in areas adjacent
to the Workspace and the Rest of Building when the plastic sheeting is used, because the plastic
sheeting is employed to contain the air-borne dust and prevent it from moving outside of the
Workspace. Thus, it is expected that the area adjacent to the Workspace and the Rest of
Building will have higher Pb air concentrations in the Base Control Option and Control Option 1,
which do not involve plastic sheeting. Similarly, the Workspace might be expected to have
higher Pb air concentrations in Control Options 2 and 3, because the plastic sheeting used in
those control options should limit the movement of dust to areas outside the Workspace.
Because post cleaning and post verification samples are not considered here, the type of cleaning
- baseline and rule - is not anticipated to affect the concentrations of interest (i.e., post work
concentrations in this case). The dust study data for the single activity example of window
replacements, however, support this assumption only in some cases. The Workspace air Pb
concentrations for Control Options 2 and 3 are indeed higher than the Workspace air Pb
concentrations measured for the Base Control Option and Control Option 1. Unexpectedly,
Control Option 2 also reports the highest air Pb concentration for the area adjacent to the
Workspace, at 8.42 |ig/m3, while the air Pb concentration for areas adjacent to the Workspace
are lower for the Base Control Option and Control Option 1 (3.68 (j,g/m3 and 4.62 (J,g/m3,
respectively). Under Control Option 3, the air Pb concentration for the area adjacent to the
Workspace is the lowest, at 2.23 (j,g/m3. These relationships illustrate the dependence of any
conclusions for any RRP activity or combination on the dust study results used.
3.4.3.2 Indoor Dust
The OPPT Dust Study provides window replacement Post Work (DS), Post Cleaning (DS), and
Post Verification (DS) floor and window sill Pb dust loadings for the Workspace, Adjacent
Room, and Rest of Building under the conditions of the four Control Options. In some cases,
bulk debris samples (i.e., samples containing excess debris that could not be collected with a dust
wipe but were analyzed separately for their Pb content) were collected and analyzed, and were
included with floor and sill wipe results. Because Pb contained in bulk debris is accessible for
exposure, this approach incorporates bulk debris results into the analysis as described below.
The floor and window sill dust samples are grouped by each unique combination of sample type
(i.e., Post Work (DS), Post Cleaning (DS), or Post Verification (DS)), location, Control Option,
42
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
and bulk debris type (i.e., with or without bulk debris). The geometric means are calculated and
the minimum and maximum detected loadings are identified for both floor and window sill dust
groups. For floor and sill dust groups that differed only by bulk debris type (i.e., data groups
with the same sample type, location, and Control Option that differ only in that one group
included bulk debris while the other did not), a geometric mean is taken across the geometric
means for the groups. In other words, each group has a geometric mean, so a geometric mean is
taken of these geometric means (i.e., geometric means of geometric means (GMM) were
calculated). The geometric means or GMMs (where applicable) for each group are selected as
the mid values for indoor dust loading. These mid values, which are presented in Exhibit A-6 in
Appendix A, are used in the deterministic simulations to generate "best estimate" results (as
described in Chapter 4). The geometric means and GMMs are used as mid values because
geometric means correct for positively skewed data by assuming a log normal distribution. In
reviewing the loadings, it is important to note that all data points that are below the limit of
detection are set to one half the limit of detection for GM and GMM calculation.
The maximum detected loadings for each floor dust group were modified to include 10 percent
of the maximum window sill loading value for the corresponding group. The addition of the
window sill loading data to the floor dust loading results is intended to account for the potential
contribution of window sill dust to a child's overall Pb exposure. In developing the methodology
for this approach, it was noted that window sills are an important source of Pb exposure that may
not be fully captured by the floor dust wipe (i.e., it is unclear that a window sill with high dust
levels would necessarily contribute significantly to the surrounding floor dust levels). As a
result, a moderate proportion of the window sill data was incorporated to approximate the
maximum floor dust loading. These modified maximum loadings and the minimum loadings
served as the high and low indoor dust loading values, respectively, and they were used in the
development of the indoor dust loading input distributions for the probabilistic modeling (as
described in Chapter 4). These values are presented in Exhibit A-6 in Appendix A.
The loadings assumed to be representative of each exposure period and phase are summarized in
Exhibit 3-13. After a one-week Renovation (Dust Generating) phase in which the loading is
assumed to be characterized by Post Work (DS) samples, it is assumed that the RRP contractor
would stop actively generating dust (e.g., through demolition activities). The loading levels are
assumed to drop after the Renovation (Dust Generating) phase to a level that is equivalent to
either the Post Cleaning (DS) or Post Verification (DS) sample loading. The choice of Post
Cleaning (DS) or Post Verification (DS) is based on what type of cleaning is performed. When
verification cleaning is performed (i.e., Control Options 1 or 3), it is assumed that the Post
Verification (DS) loading is representative because that is the value in the OPPT Dust Study
after intensive rule cleaning has been performed. Conversely, when only baseline cleaning is
performed (i.e., Base Control Option or Control Option 2), the Post Cleaning (DS) value is
assumed to be representative. These Post Verification (DS) or Post Cleaning (DS) values are
assumed to be the starting point for the Post Renovation (Routine Cleaning) phase, over which
the loading values are decayed until they reach background. A generalized example of how
concentrations, which are calculated from these loadings using a formula described in Section
4.2, change over the course of the exposure periods and phases is provided in Exhibit 3-4. It is
important to note, however, that in the single activity example, it is assumed that the Renovation
(After Baseline Cleaning) phase has a length of zero weeks (because the end of the dust-
generating activities coincides with the end of the renovation).
43
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
In general, one would expect the dust loadings to decrease from the post work measurements to
the post cleaning measurements and from the post cleaning measurements to the post verification
measurements. This general trend, however, is not always present in the OPPT data. For
example, in window replacements for the Base Control Option, dust loadings in areas adjacent to
the Workspace increased between the post work (4.1 jag/ft2) and post cleaning measurements
(12.8 (J,g/ft2). Similarly, the loading values for the Rest of Building increased between the post
work and post cleaning measurements, from 1 j_ig/ft2 to 1.5 |ig/ft2, respectively. This trend may
be due to the fact that cleaning only occurred in the Workspace, and the cleaning may have
generated air-borne dust which spread to other regions of the house or that somehow different
areas were measured across phases. For Control Options 1 and 3, plastic was placed over the
door in an effort to avoid such spreading from the Workspace to other portions of the house. In
some instances, however, (e.g., for the rest of the building for Control Option 1), the loading
again increased between the post work and post cleaning phases.
In addition, instances exist in the OPPT Dust Study where the post verification loadings are
higher than the post cleaning loadings. For example, in Control Option 3 the dust loadings for
the rest of the building decreased from the post work to post cleaning samples, but increased
from 1.4 jag/ft2 (post cleaning) to 1.6 j_ig/ft2 (post verification). In general, these differences are
small and may be attributable to measuring error. Although in general the Workspace loadings
decreases after the post work and post cleaning phases, such a trend is not a general rule in the
adjacent and rest-of-building areas.
3.4.3.3 Outdoor Soil
No example is provided in this approach for exposures from outdoor RRP activities. As a result,
this section presents a generic discussion of the methodology that will be used to estimate
activity-related outdoor soil concentrations.
The OPPT Dust Study provides outdoor dust wipe sample loadings for only one exposure period,
and it does not provide data for separate experiments to permit evaluating the relative
effectiveness of Control Options A and B. As a result, this approach estimates the loadings for
the Renovation and Post-renovation exposure periods using different combinations of the Post
Work (DS) samples measured on Top of Rule Plastic (DS), Under Rule Plastic (DS), and
Outside Rule Plastic (DS), as is shown in Exhibit 3-14. For simplicity, this discussion is broken
up by Control Option.
For Control Option A (i.e., no plastic), it is assumed that all dust that is deposited on a particular
area of soil remains in that area and undergoes no loss processes (e.g., erosion). In the Dripline,
it is assumed for the Renovation and Post-renovation exposure periods that the total available Pb
dust available for exposure is equal to the Post Work dust loading on Top of Rule Plastic (DS)
plus the dust loading Under Rule Plastic (DS). The reason for this assumption is that under
Control Option A, no plastic would be used in the Dripline for a RRP activity. In the Nearby
area, for which the OPPT Dust Study only provides data on top of the containment plastic, it is
assumed for the Renovation and Post-renovation exposure periods that the total available Pb dust
available for exposure is equal to the Post Work (DS) dust load on Top of Containment Plastic
(DS). During an actual RRP project, this containment plastic would not be used (it is not a
requirement under either Control Option A or B), so all the dust falling in the Nearby area would
44
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
be available for exposure. A generalized example of how concentrations, which are calculated
from these loadings using a formula described in Section 4.1, change over the course of the
exposure periods and phases is provided in Exhibit 3-5.
For Control Option B (i.e., plastic), it is assumed that plastic is used to prevent the loading of
some dust onto soil. In the Dripline, it is assumed that for the Renovation exposure period, the
total available Pb dust available for exposure is equal to the Post Work (DS) dust loading on Top
of Rule Plastic (DS). The reason for this assumption is that under Control Option B, the Dripline
plastic would prevent lead contamination of the dust under the plastic, and prevent exposure to
dust that is sitting on top of the plastic. For the Post Renovation (DS) exposure period, the
loading is assumed to equal the Post Work (DS) dust loading Under Rule Plastic (DS). The
rationale behind this assumption is that when the plastic is removed after the job is complete, the
dust on top of the plastic is also removed, leaving only the dust under the plastic for exposure. In
the Nearby area, for which the OPPT Dust Study only provides data on top of the containment
plastic, it is assumed for the Renovation and Post-renovation (DS) exposure periods that the total
available Pb dust available for exposure calculations is equal to the Post Work (DS) dust load on
Top of Containment Plastic (DS). During an actual RRP project, this containment plastic would
not be used (it is not a requirement under either Control Option A or B), so all the dust falling in
the Nearby area would be available for exposure calculations.
Exhibit 3-14. Crosswalk between the Loadings Used in this Approach and the Loadings
Provided in the OPPT Dust Study for Outdoor Soil
Loadings Used in this Approach
Loadings Provided in the OPPT Dust Study
No Plastic
(Control Option A)
Dripline loading - Renovation
Top of rule plastic plus Under rule plastic - Post Work
Dripline loading - Post-renovation
Top of rule plastic plus Under rule plastic - Post Work
Nearby loading - Renovation and Post-
renovation
Top of containment plastic - Post Work
Plastic (Control Option B)
Dripline loading - Renovation
Top of rule plastic - Post Work
Dripline loading - Post-renovation
Under rule plastic - Post Work
Nearby loading - Renovation and Post-
renovation
Top of containment plastic - Post Work
For some outdoor measurements, wipe samples were not the only type of sample taken in the
OPPT Dust Study. In some cases, bulk debris samples (i.e., samples containing excess debris
that could not be collected with a dust wipe but were analyzed separately for their Pb content)
were collected, analyzed, and included with some wipe sample results. Because Pb contained in
bulk debris is accessible for exposure, this approach incorporates bulk debris results into the
analysis as described below.
After setting all the data points that are below the limit of detection equal to one half the limit of
detection, the dust samples are grouped by each unique combination of sampling area in yard,
phase, and bulk debris type (i.e., with or without bulk debris). The geometric means are
calculated and the minimum and maximum detected loadings are identified. For groups that
differ only by bulk debris type (i.e., data groups with the same sample type, location, and Control
Option that differ only in that one group included bulk debris while the other did not), a
45
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
geometric mean is taken across the geometric means for the groups (i.e., geometric means of
geometric means (GMM) are calculated). The geometric means or GMMs (where applicable)
for each group are selected as the mid values for indoor dust loading. These mid values are used
in the deterministic simulations to generate "best estimate" results (as described in Chapter 4).
The maximum and minimum loadings serve as the high and low outdoor soil loading values,
respectively, and are used to develop the outdoor soil loading input distributions used in the
probabilistic modeling (as described in Chapter 4).
It is important to note that the OPPT Dust Study reports Pre Work (DS) and Post Work (DS) soil
concentrations for each of the exterior RRP activities. These measurements are not used for two
primary reasons. First, RRP activities were repeated at the various sites in the study, so only the
first activity at a given property would give valid soil concentrations for that activity. The use of
data from only these first activities would result in an extremely small number of observations,
thus decreasing predictive power. Second, the results of using this approach must be applicable
to a range of different locations in the United States. Using these soil concentrations would bias
results towards the small number of locations where the OPPT Dust Study experiments were
performed. As a result, outdoor soil concentrations associated with particular RRP activities are
estimated using estimated background concentrations in combination with the OPPT Dust
Study's estimates of exterior dust loadings.
3.4.4 Multiple Activities Example
For the purpose of illustrating the methodology of this approach for estimating childhood
exposures to multi-activity RRP events, an example scenario consisting of kitchen renovation,
bathroom renovation, 10 door or window replacements, interior painting, HVAC (heating,
ventilation, and air conditioning), wiring, plumbing, and installation of a security system is
examined here. This section presents the approach for developing the Pb levels from the OPPT
Dust Study for a multiple RRP activity scenario. Because the OPPT Dust Study does not include
experiments for all of these RRP events, the example scenario is assumed to be most similar to
two kitchen renovations, 10 window replacements, one interior flat component LBP removal
scraping event, and four cut-outs.
3.4.4.1 Air
The approach for estimating activity-related air concentration for the multiple activities example
is exactly the same as the activity-related air concentration estimation approach shown in Section
3.4.3.1. The process is simply repeated for each of the individual activities.
As in the single activities case, it is expected that the area adjacent to the Workspace and the
Rest of Building will have higher Pb air concentrations in the Base Control Option and Control
Option 1, which do not involve plastic sheeting, while the Workspace is expected to have higher
Pb air concentrations in Control Options 2 and 3, because the plastic sheeting is intended to limit
the movement of dust to areas outside the Workspace. This trend, however, may not generally
be followed for all the renovation activities comprising any multiple activities cases since
calculations will reflect any anomalies in the dust study data and sums are taken to create
aggregates for estimation. In this particular multiple activities example, the exceptions occur
46
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
most frequently for the window replacements case (discussed above), but other deviations are
seen for the other activities as well.
3.4.4.2 Indoor Dust
The approach for estimating activity-related indoor dust loadings for a multiple activities case is
exactly the same as the activity-related indoor dust loading estimation approach shown in
Section 3.4.3.2. The process is simply repeated for each of the individual activities.
As in the single activities case, it is expected that post work values should be the highest, with
post cleaning and post verification levels each subsequently decreasing. Again, however,
deviations occur. Because the multiple activity weighs the window replacements activity ten
times (that is, it assumes that ten window replacements are part of the multiple activity), the
deviations noted for this case become exacerbated for the multiple activity example. In addition,
other individual deviations also occur in the other activities.
3.4.4.3 Outdoor Soil
There are no outdoor activities in the multiple activities example, but the approach for estimating
activity-related outdoor soil loadings for the multiple activities example is exactly the same as
the outdoor soil loading estimation approach shown in Section 3.4.3.3. The process will simply
be repeated for each individual activity.
3.5 Uncertainties and Limitations
The primary source of uncertainty in the background and activity-related data is limited sample
size for the activity-related data. The need to parse the activity data into small groupings based
on Control Options, sampling location, and exposure periods and phases, significantly limits the
predictive power of the OPPT Dust Study data.
Another significant uncertainty is the lack of activity-related data that were specific to the control
options and exposure periods and phases of this approach (e.g., separate experiments were not
conducted for outdoor soil's Control Option A and B and Post Work (DS) samples were the only
usable air concentration values for indoor air). This data limitation necessitates assumptions
regarding the appropriate methods for estimating loadings.
The fact that loadings for the various control options are not always taken from the same job
sites in the OPPT Dust Study is another important source of uncertainty. For example, one
control option for window replacement may have been tested in one building while another
control option for that activity might have been tested in another. This study design limits the
comparability of the control options because differences in loading values might be explained by
differing building characteristics rather than by control options.
An additional uncertainty is that the Post Work (DS) air concentrations from the OPPT Dust
Study were often reported as non-detects. This occurred when either the measured value was
deemed below the measurement threshold (not enough mass collected in the machine) or when
the sampling time was too short. In this approach, all samples are used, except for a single
sample with a sampling time of 5 minutes that was deemed too short to be representative and
47
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
was excluded from the analysis (the next shortest sampling time was 80 minutes). Thus, there is
uncertainty surrounding the smallest air concentrations based on experimental error.
A limitation of the approach is that blood lead models (i.e., IEUBK and Leggett) do not
explicitly account for window sill loadings. As a result, this approach factors these loading
samples into the dust loadings by adding 10 percent of the window sill loading to the maximum
floor loading. It is recognized that this procedure adds uncertainty to the floor samples, but this
approach was taken to ensure that potential childhood exposures to lead on window sills are
factored into the analysis.
A source of uncertainty in this approach is the median air concentration (0.025 |ig/m3) used as
the background Pb air concentration, which comes from site monitoring data in EPA's Air
Quality Systems database (USEPA 2006c). It is likely biased high because Pb monitors are often
located in areas with nearby Pb emission sources. Modeling conducted for this approach,
however, indicates that background ambient air concentrations are a small contributor to overall
blood Pb levels, so the effects of this potential bias are expected to be minimal.
Another limitation of this approach is that the indoor air concentrations are assumed to equal the
outdoor air concentrations. It is unclear whether this assumption would tend to positively or
negatively bias blood Pb level results since no comparative indoor air concentration data were
found.
In this approach, it is assumed that indoor air, indoor dust, and outdoor soil are the three media
that contain increased levels of Pb as the result of RRP activities. It is a recognized limitation of
this analysis that the Pb concentration of outdoor air may also rise as the result of a renovation.
Although the tendency of this limitation would be to underestimate total exposures, the increased
ventilation in the outdoor environment as compared to indoor ventilation would likelymitigate
the effect.
A further limitation of this approach is that in calculating the "mid" values for the dust loadings,
geometric means are used to combine the available data into a single central tendency estimate.
The use of geometric means implicitly assumes that the distribution of the loadings is lognormal.
In general, the measurements tend to be positively skewed, indicating that lognormal is a better
representation of the data than normal; this assumption, however, represents an uncertainty and
may tend to produce either lower or higher mid estimates than if an arithmetic mean is used.
The fact that some measurements from the OPPT Dust Study are not incorporated in this
approach is also a limitation. For example, measurements from the halls are not used since these
areas have poorly specified and variable containment properties compared to enclosed rooms. In
addition, some interior renovation activities had exterior samples which were taken to determine
how much indoor dust was tracked into the outdoor environment. Little information about the
yard location of the samples is available, however, so these samples are not incorporated in this
approach. It is important to note, however, that this approach accomodates specified outdoor soil
concentration.
48
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
4. ESTIMATES OF MEDIA CONCENTRATIONS
This chapter describes the approach for estimating Pb concentrations in the relevant exposure
media, including air, indoor dust, outdoor soil, drinking water, and diet. It describes how the
approaches for air, indoor dust, and outdoor soil build on the background and activity-related
inputs developed as described in Chapter 3 and how these inputs are used to characterize air,
indoor dust, and outdoor soil concentrations. It also describes how concentrations in drinking
water and diet were characterized. The approaches for estimating indoor dust and outdoor soil
concentrations include both deterministic and probabilistic components. The approaches for air,
drinking water, and diet include only a deterministic component. Water and diet concentrations
are assumed to be unaffected by renovation activities. In addition, the activity-related air
concentrations have a lower effect on blood Pb levels than the activity-related dust and soil
concentrations, partially because the dust and soil levels remain elevated longer and partially
because the blood Pb models are more sensitive to dust and soil inputs using the range of dust,
soil, and air concentrations measured during and after renovation activities.
Because of this increased sensitivity, the dust and soil concentrations include a probabilistic
component to determine the range of responses to different ranges of dust and soil values. The
deterministic components for air, indoor dust, and outdoor soil (described in Sections 4.1 through
4.3, respectively) estimate concentrations for a given scenario based on a single set of "best
estimate" inputs and assumptions. The approaches for drinking water and diet rely upon
national-level data to characterize concentrations. The probabilistic components for indoor dust
and outdoor soil (described in Section 4.6) involve both sensitivity and Monte Carlo analyses.
The sensitivity analysis characterizes the sensitivity of the estimated media concentrations to
changes in each selected input, considering the expected variability of that input as well as
potential influential characteristics like cleaning efficiency and the use of a loading concentration
conversion, and its results provide insight into which inputs are most important. The Monte
Carlo analysis characterizes the range of estimated media concentrations given the distributions
of potential values for selected inputs. Its results help to better characterize the potential range of
estimated media concentrations given the uncertainty and variability in the input values. To
illustrate both the deterministic and probabilistic components of the approach, two example
applications, one for a single RRP activity example and another for a multiple RRP activities
example, are provided for air and indoor dust in Sections 4.1 and 4.2.2 The chapter concludes
with Section 4.7, which provides a discussion of the uncertainties and limitations of the
approach.
Exhibit 4-1 provides a key for the different dimensions (e.g., control options, exposure phases)
that are used in the formulas throughout this chapter.
2 Note that no examples are provided for outdoor soil, drinking water, and diet.
49
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Exhibit 4-1. Dimension Types, Names, and IDs used in this Approach
Dimension Type
Dimension Name
Dimension ID
Location (LOC)
Indoor
Workspace
1
Adjacent
2
Rest of Building
3
Location (LOC)
Outdoor
Dripline
A
Nearby
B
Rest of Yard
C
Exposure Phase(PH)
Indoor Dust
Pre-Renovation (Background)
1
Renovation (Dust Generating)
2
Renovation (After Baseline Cleaning)
3
Post-Renovation (Routine Cleaning)
4
Post-Renovation (Background)
5
Phase(PH)
Indoor Air
Pre-Renovation (Background)
Renovation (Dust Generating)
i
Renovation (Settling)
i
Renovation (Background)
V
Post-Renovation (Background)
V
Phase(PH)
Outdoor Soil
Pre-Renovation (Background)
A
Renovation
B
Post-Renovation
C
Control Option (CO)
Indoor Dust and Air
Base Control Option
0
Control Option 1
1
Control Option 2
2
Control Option 3
3
Control Option (CO)
Outdoor Soil
Control Option A
A
Control Option B
B
Activity (ACT)
Window replacement(s)
1
Interior flat component LBP removal, scraping
2
Renovating kitchen
3
Cut-Outs
4
Sampling Method
(SM)
Wipe
W
Blue Nozzle
BN
Vintage (VIN)
< 1940
1
1940 to 1959
2
1960 to 1979
3
50
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
4.1 Air
Two types of air concentration data are required for this approach: ambient air concentrations
(i.e., outdoor air) and indoor air concentrations. Ambient air concentrations are developed as
described in Section 3.3.1.2. It is assumed that RRP activities do not contribute to ambient air
concentration and thus the ambient concentrations used in this approach are based solely on
background estimates. Indoor air concentrations are estimated as summarized in Exhibit 4-2.
The elements of this process described in Chapter 3 are shown in grey and the elements
described in this section are shown in black (note that this exhibit is identical to Exhibit 3-6, with
the exception of which elements are in grey and black). Section 4.1.1 describes how the general
approach illustrated in Exhibit 4-2 is applied to estimate indoor air concentrations during each
phase of the different exposure periods. Estimated indoor air concentrations for the single RRP
activity and multiple RRP activities examples are provided in Sections 4.1.2 and 4.1.3.
51
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Exhibit 4-2. Approach for Estimating Indoor Air Concentrations
52
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
4.1.1 Approach
Exhibit 3-2 presents an overview of the different exposure periods, and phases comprising each
exposure period, defined in this approach for characterizing indoor air concentrations. The
specifics of the approach used to characterize indoor air concentrations for each of these
exposure periods are provided in the following sections.
4.1.1.1 Pre-renovation
The Pre-renovation exposure period represents the period of exposure before initiation of the
RRP activity or activities and thus consists of background contributions only. The approach for
estimating background indoor air concentrations is described in Section 3.3.1.3.
4.1.1.2 Renovation
The Renovation exposure period represents the period of exposure beginning with the initiation
of the RRP activity or activities and concluding with the completion of the renovation and any
contractor cleaning. In this approach, there are three phases of the Renovation period for indoor
air concentrations: the Dust Generating phase, the Settling phase, and the Background phase.
The Dust Generating phase represents the portion of the renovation where RRP activities are
creating leaded dust (e.g., during demolition). The Settling phase represents the period of time
following completion of the Dust Generating activities during which the dust in air settles onto
the floor. The Background phase represents the period of time from the end of the Settling phase
until the end of the Renovation period. In this approach, there are four steps involved in
estimating air concentrations for this exposure period: (1) calculate whole building RRP
activity-related concentrations during the Dust Generating phase; (2) combine activity
concentrations to calculate whole building concentrations for the Dust Generating phase; (3)
estimate the Settling of dust in air during the Settling phase; and (4) calculate concentrations for
background phase. These steps are described below.
53
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Calculate Whole Building Activity Concentrations during the Dust
Generating Phase
In this approach, the indoor air concentrations are developed (as described in
Chapter 3) for three different locations in the building: Workspace, Adjacent
Room, and Rest of the Building. These room-specific concentrations are combined
into a single whole house concentration for each activity using the following
equation:
ACONCpH=ii,CO,ACT -
(PA W * ACONCloc=i ,ph=u,co,act)
(PAA * ACONCwe =2,PH=H,CO,ACt) ~
(PAR * ACONCloc=3,ph=u,co,act)
(Eq. 4-1)
where:
ACONCpH=ii,CO,ACT
CO
PAW
Whole building average indoor air Pb
concentrations during phase ii (i.e., Dust Generating
phase), with Control Option CO, and Activity ACT,
in |ig/m3
Base, 1, 2, or 3
% Area of Building — Workspace
ACONCloc=i,ph=u,co,act = Indoor air Pb concentration in the Workspace (i.e.,
LOC=l) during phase ii (i.e., Dust Generating
phase), with Control Option CO, and Activity ACT,
in |ig/m3
PAA =
ACONCwc =2,PH=ii, CO,ACT -
PAR
ACONC wc =3,PH=ii, CO,ACT -
% Area of Building — Adjacent Room
Indoor air Pb concentration in the Adjacent Room
(i.e., LOC=2) during phase ii (i.e., Dust Generating
phase), with Control Option CO, and Activity ACT,
in |ig/m3
% Area of Building — Rest of Building
(calculated as 1 - PAW - PAA)
Indoor air Pb concentration in the Rest of Building
(i.e., LOC=3) during phase ii (i.e., Dust Generating
phase), with Control Option CO, and Activity ACT,
in |ig/m3
54
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Calculate Whole Building Concentrations across Activities
during the Dust Generating Phase
After calculating whole building activity air concentrations, these
concentrations are averaged together to estimate the whole building air
concentrations using the following equation:
Y.ACONC
ACONCpH=hco = ^
PH=ii,CO,ACT
n
(Eq. 4-2)
where:
ACONCpH=u, co ~
ACONCpH=ii, co, act ~
Whole building average indoor air Pb
concentrations during phase ii (i.e., Dust Generating
phase), with Control Option CO, in |ig/m3
Whole building average indoor air Pb
concentrations during phase ii (i.e., Dust Generating
phase), with Control Option CO, and Activity ACT,
in |ig/m3
n = number of activities
Calculate Concentrations during the Settling Phase
In this step, the indoor air concentrations during the Settling phase (which is
always one week long) are estimated. Estimating these concentrations requires
first estimating concentrations for each part of the phase and then calculating a
time-weighted average of these concentrations to estimate the concentrations for
the phase. Indoor air concentrations for the different parts of the phase are
calculated as follows:
Whole
Building
Cones.
(during dust
generating phase)
Decay of Air
Cones. Due to
Settling
Whole
Building
Cones.
(during dust
generating
and settling phases)
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Part 1: As described in Chapter 3, indoor air concentrations for the Dust Generating phase were
estimated using air sampling that began during the activity and continued until one hour after
completion of the activity. Because this phase is defined as beginning at the completion of the
Dust Generating phase activity, the last hour of this sampling is part of the Settling phase. As a
result, the first part of the Settling phase is represented by concentrations for the Dust Generating
phase (i.e., ACONCpH=u, co) and is assumed to last one hour.
Part 2: The second part of the Settling phase is the time during which the dust suspended during
the Dust Generating phase settles to the floor. The concentrations in this part are calculated
using the following equation:
(DecayConst*h)
ACONCph =iii@h,CO = ACONCph =ii,CO *
e (Eq. 4-3)
where:
ACONCpH=m@h,co = Whole building average indoor air Pb
concentrations at hour h of phase iii (i.e., Settling
phase), with Control Option CO, in |ig/m3
h = Time since the initiation of the Part 2 of the Settling
phase, in hours
ACONCph=u,co = Whole building average indoor air Pb
concentrations during phase ii (i.e., Dust Generating
phase), with Control Option CO, in |ig/m3
DecayConst = Decay constant for Pb dust in air, based on most
conservative value presented in Choe et al. (2000),
in hours"1
The calculation is repeated, incrementing h by one hour each time, until
ACONCph =in@h,co reaches the background concentration (generally around 15 hours).
All of the values for ACONCpH=m@h,co are then averaged together to estimate the Part 2
concentration (i.e., ACONCPH=m@part2,co)-
Part 3: The final part of the Settling phase represents the time beginning when
ACONCph =in@h,co reaches background and continuing until the Settling phase reaches one week
in duration. This part consists entirely of background contributions only. The approach for
estimating background indoor air concentrations is described in Section 3.3.1.3.
56
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
The Settling phase concentration is estimated from the concentrations for Parts 1 through 3 using
the following equation:
ACDNC ™ = ACONCPH=m9P^co * 1 hr + ACONCPH=mf^Part2 CO * Y + ACONCPH=iii@PartJCO * (167hr - Y)
vji v ^PH=iii,t(J
168 hr
(Eq. 4-4)
where:
ACONCpH=m,co = Whole building average indoor air Pb
concentrations during phase Hi (i.e., Settling phase),
with Control Option CO, in |ig/m3
ACONCpH=in@Parti,co = Whole building average indoor air Pb
concentrations during Part 1 of phase Hi (i.e.,
Settling phase), with Control Option CO, in |ig/m3
ACONCpH=iu@Part2,co = Whole building average indoor air Pb
concentrations during Part 2 of phase Hi (i.e.,
Settling phase), with Control Option CO, in |ig/m3
Y = Duration of Part 2 of phase Hi (i.e., Settling phase),
in hours
ACONCpH=in@Part3,co = Whole buildinig average indoor air Pb
concentrations during Part 3 of phase Hi (i.e.,
Settling phase), with Control Option CO, in |ig/m3
Calculate Concentrations for Background Phase
The Background phase of the Renovation period represents the time beginning at the conclusion
of the Settling phase and continuing until the end of the Renovation period. This phase consists
entirely of background contributions only. The approach for estimating background indoor air
concentrations is described in Section 3.3.1.3.
4.1.1.3 Post-renovation
The Post-renovation period represents the time beginning at the conclusion of the Renovation
period and continuing until the end of the exposure duration. This period consists entirely of
background contributions only. The approach for estimating background indoor air
concentrations is described in Section 3.3.1.3.
57
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
4.1.2 Example for Single RRP Activity
The approach presented in Section 4.1.1 was applied to the single RRP activity scenario by
selecting experiments from the OPPT Dust Study for calculation. In the example, it uses in the
input values presented in Exhibit A-4 in Appendix A. The indoor air concentrations for the
Background (where Pre-renovation, Renovation (Background), and Post-renovation all use the
same background value), Renovation (Dust Generating), and Renovation (Settling) for each of
the three housing vintages for the single activity example are shown in Exhibit 4-3.
Exhibit 4-3. Indoor Air Concentrations for Single RRP Activity Example
Exposure Phase
Concentration (ng/m3)
Base Control
Option
(no plastic;
baseline
cleaning)
Control
Option 1
(no plastic;
rule
cleaning)
Control
Option 2
(plastic;
baseline
cleaning)
Control
Option 3
(plastic;
rule
cleaning)
Background
0.025
0.025
0.025
0.025
Post-renovation (Dust
Generating)
3.06
3.57
6.30
5.14
Post-renovation
(Settling)
0.24
0.29
0.52
0.41
4.1.3 Example for Multiple RRP Activities
The approach presented in Section 4.1.1 was applied to the multiple RRP activities scenario in a
manner similar to a single RRP activity. For the multiple activities example, it uses the input
values presented in Exhibit A-5 in Appendix A. The indoor air concentrations for the
Background (where Pre-renovation, Renovation (Background), and Post-renovation all use the
same background value), Renovation (Dust Generating), and Renovation (Settling) for each of
the three housing vintages for the multiple activity example are shown in Exhibit 4-4.
Exhibit 4-4. Indoor Air Concentrations for Multiple RRP Activities Example
Exposure Phase
Concentration (|jg/m3)
Base Control
Option
(no plastic;
baseline
cleaning)
Control
Option 1
(no plastic;
rule
cleaning)
Control
Option 2
(plastic;
baseline
cleaning)
Control
Option 3
(plastic;
rule
cleaning)
Background
0.025
0.025
0.025
0.025
Post-renovation
(Dust Generating)
3.45
3.72
5.19
4.41
Post-renovation
(Settling)
0.26
0.30
0.42
0.34
4.2 Indoor Dust
58
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Indoor dust concentrations are estimated in this approach as summarized in Exhibit 4-5. The
elements of this process described in Chapter 3 are shown in grey and the elements described in
this section are shown in black (note that this exhibit is identical to Exhibit 3-7, with the
exception of which elements are in grey and black). Section 4.1.3 describes how the general
approach illustrated in Exhibit 4-5 is applied to estimate indoor dust concentrations during each
phase of the different exposure periods. Sections 4.2.2 and 4.2.3, respectively, provide estimated
indoor dust concentrations for the single and multiple activity examples.
59
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Exhibit 4-5. Approach for Estimating Indoor Dust Concentrations
60
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
4.2.1 Approach
Exhibit 3-4 presents an overview of the different exposure periods, and phases comprising each
exposure period, defined in this approach for characterizing indoor dust concentrations. The
specifics of the approaches used to characterize indoor dust concentrations for each of these
exposure periods are provided in the following sections.
4.2.1.1 Pre-renovation
The Pre-renovation exposure period represents the period of exposure before initiation of the
RRP activity or activities and thus consists of background contributions only. There are two
steps involved in estimating indoor dust concentrations for this period: (1) convert wipe sample
indoor dust loadings to vacuum sample indoor dust loadings; and (2) convert vacuum sample
indoor dust loadings to indoor dust concentrations. These steps are described below:
Convert Wipe Loadings to Vacuum Loadings
As described in Section 3.3.2, the background indoor dust loading values used for
this approach are based on indoor dust wipe samples. Because the regression
equations developed to convert indoor dust loadings to indoor dust concentrations
require that loadings be provided in terms of Blue Nozzle (BN) vacuum samples,
these wipe samples must be converted into BN samples. This conversion is
performed using the following equation for uncarpeted floors from "Conversion
Equations for Use in Section 403 Rulemaking" (USEPA 1997):
DLOADbn,bg, vin = 0.185* (DLOADw,bg, vin)°'93 1 (Eq. 4-5)
where:
DLOADbn.bg, vin = Background (BG) indoor dust Pb BN sampling
loading for building vintage VIN, |ig/m2
DLOAD^bg.vin = Background (BG) indoor dust Pb wipe sampling
loading for building vintage VIN, |ig/m2
No equation is provided for converting loadings from carpet, so it is assumed this equation could
be applied to both carpets and hard surfaces. This is recognized as a limitation of the approach.
Backgrnd
Loadings
(wipe sample)
Wipe-to-Blue Nozzle
(BN) conversion
Backgrnd
Loadings
(BN sample)
61
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Convert Loadings to Concentrations
Backgrnd
Loadings
(BN sample)
For buildings of known vintage, the BN sample loading values are converted to
indoor dust concentrations using the appropriate housing vintage-specific
regression equation described in Appendix C, which is summarized below:
Loading-to-concentration
regression
Backgrnd
Cone.
DCONCbg.vin ~ ^
(InterceptVjX + SlopeVjN (In(DLOADBN )))
(Eq. 4-6)
where:
DCONCbg.vin = Background (BG) indoor dust concentration for
building vintage VIN, in |ig/g
DLOADBn,bg = Background (BG) indoor dust Pb BN sampling
loading, in jug/ft2
Although this regression introduces uncertainty, conversion remains a necessary step, since the
blood Pb models are configured to accept concentrations and not loadings. Under an assumption
that the HUD-derived relationships can be generalized, this regression represents the best known
relationship available.
The Renovation exposure period represents the period of exposure beginning with the initiation
of the RRP activity or activities and concluding with the completion of the renovation and any
contractor cleaning. There are two phases of the Renovation period for indoor dust
concentrations: the Dust Generating phase and the After Baseline Cleaning phase. The Dust
Generating phase represents the portion of the renovation where RRP activities are creating
leaded dust (e.g., during demolition). The After Baseline Cleaning phase represents the
remainder of the Renovation period and takes into account some basic cleaning after completion
of the Dust Generating activities. The calculation process to estimate indoor dust concentrations
for these phases is identical as explained in Chapter 3, but different loadings values are used to
represent the different phases. The approach for estimating indoor dust concentrations for the
Renovation period consists of four steps: (1) convert wipe sample indoor dust loadings to
vacuum sample indoor dust loadings; (2) calculate whole building RRP activity-related vacuum
indoor dust loadings; (3) calculate whole building total vacuum indoor dust loadings (including
background); and (4) convert whole building total vacuum indoor dust loadings to whole
building indoor dust concentrations. These steps are described below.
4.2.1.2 Renovation
62
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Convert Wipe Loadings to Vacuum Loadings
In this step, the activity-related and background indoor dust loading values, developed as
described in Sections 3.4.3.2 (for single activity scenarios), 3.4.4.2 (for multiple
activity scenarios), and 3.3.2 (for background), are converted from indoor dust wipe
samples to indoor dust BN vacuum samples. This step is required because the
regression equations developed to convert from indoor dust loadings to indoor dust
concentrations require that loadings be provided in terms of BN vacuum samples.
This conversion is performed for activity-related loadings using the following
equation for uncarpeted floors based on "Conversion Equations for Use in Section
403 Rulemaking" (USEPA 1997):
Activity
Loadings
(wipe sample)
Wipe-to-Blue Nozzle
(BN) conversion
Activity
Loadings
(BN sample)
DLOADbn,loc,ph,co,act ~ 0.\%5*{DLOADw,loc,ph,co,act)
0.931
(Eq. 4-7)
where:
DLOADbn,loc,ph,co,act
= Indoor dust Pb BN sampling loading in
location LOC during phase PH, with
Control Option CO, and Activity ACT, in
|ig/m2
DLOAD
W,LOC,PH, CO,ACT
= Indoor dust Pb wipe sampling loading in
location LOC during phase PH, with
Control Option CO, and Activity ACT, in
|ig/m2
No equation is provided for converting loadings from carpet, so it is assumed this equation could
be applied to both carpets and hard surfaces. This is recognized as a limitation of the approach.
Likewise, this conversion is performed for background loadings using Equation 4-1. Again, no
equation is provided for converting loadings from carpet, so it is assumed this equation can be
applied to both carpets and hard surfaces. This is recognized as a limitation of the approach.
63
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Calculate Whole Building Activity Loadings
The indoor dust loadings are developed (as described in Chapter 3) for three
different locations in the building: Workspace, Adjacent Room, and Rest of the
building. These location-specific loadings are combined into a single whole
house loading for each activity using the following equation:
DLOADbn,ph,co,act = (PAW * DLOADbn,loc=i,ph,co,act) +
(PAA * DLOADbn,loc=2,ph,co,act) +
(PAR * DLOADbnloc=3 ,ph,co,act)
(Eq. 4-8)
where:
DLOADBn,ph,co,act = Whole building average indoor dust Pb BN
sampling loadings during phase PH, with Control
Option CO, and Activity ACT, in |ig/m2
PAW = % Area of Building — Workspace
DLOAD
BN.LOC=l .PH.CO.ACT
Indoor dust Pb BN sampling loading in the
Workspace (i.e., LOC= 1) during phase PH, with
Control Option CO, and Activity ACT, in |ig/m2
PAA = % Area of Building — Adjacent Room
DLOAD
BN.LOC=2,PH, CO ACT
PAR =
DLOAD
BN,LOC=3,PH, CO ACT
Indoor dust Pb BN sampling loading in the
Adjacent Room (i.e., LOC=2) during phase PH,
with Control Option CO, and Activity ACT, in
|ig/m2
% Area of Building — Rest of Building (calculated
as 1 -PAW-PA A)
Indoor dust Pb BN sampling loading in the Rest of
Building (i.e., LOC=3) during phase PH, with
Control Option CO, and Activity ACT, in |ig/m2
64
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Calculate Whole Building Total Loadings
After calculating whole building average loadings for each
activity, the next step is to sum the loadings across all RRP
activities and then add the background loadings. This total loading
represents the contributions from all activities as well as the
contributions from background. Note that for scenarios with only
one RRP activity, this step consists of simply summing the whole
building loadings for that activity and the background loadings.
The equation used to calculate this whole building total loading is
provided below:
Activity 1
Activity n
DLOADBn,ph,co, vin ~ £DLOAD,,.
PH,CO,ACT
\ACT =1
A
+ DLOAD
BN,BO,VlN
(Eq. 4-9)
where:
DLOADBn,ph,co,vin = Total whole building average indoor dust Pb BN
sampling loadings during phase PH, with Control
Option CO, and building vintage VIN, in |ig/m2
DLOAD,
BN,PH,CO,ACT
Average whole building indoor dust Pb BN
sampling loadings during phase PH, with Control
Option CO, and Activity ACT, in |ig/m2
DLOAD,
'BN.BG, vin
= Background (BG) indoor dust Pb BN sampling
loadingfor building vintage VIN, |ig/m2
65
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Convert Loadings to Concentrations
The final step required to calculate indoor dust concentrations for both phases of
the Renovation period is to convert the whole building total indoor dust loadings
to indoor dust concentrations. For buildings of known vintage, the BN sample
whole building total loading values are converted to indoor dust concentrations
using the appropriate housing vintage-specific regression equation described in
Appendix C, which is summarized below:
DC ONCph, co, vin
where:
(Intercept^ +SlopeYjN* (In(DLOADBN PH co yjN)))
= £ (Eq. 4-10)
DCONCph.co.vin = Total whole building average indoor dust
concentration during phase PH, with Control
Option CO, and building vintage VIN, in |ig/g
DLOADbn,ph.co.vin = Total whole building average indoor dust Pb BN
sampling loadings during phase PH, with Control
Option CO, and building vintage VIN, in |ig/m2
4.2.1.3 Post-renovation
The Post-renovation exposure period represents the period of exposure beginning with the
completion of the renovation and any contractor cleaning and ending with the end of the time
period for which exposures are being characterized. There are two phases of the Post-renovation
period for indoor dust concentrations: the Routine Cleaning phase and the Background phase.
The Routine Cleaning phase represents the period of time between the completion of the
renovation and the time at which indoor dust concentrations return to background levels, and this
phase accounts for the decrease in indoor dust concentrations that occurs due to routine cleaning
of the building. The Background phase represents only contributions from background sources
of indoor dust. The approach for estimating concentrations during the Post-renovation
(Background) phase is identical to the approach used for the Pre-renovation period (presented in
Section 4.2.1.1) and thus is not repeated here. The approach for estimating indoor dust
concentrations for the Routine Cleaning phase consists of two steps: (1) calculating the change
in indoor dust loadings based on routine cleaning; and (2) converting the resulting indoor dust
loadings to indoor dust concentrations. These steps are described below.
66
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Calculate Change in Loadings Based on Routine Cleaning
Indoor dust Pb loadings in the Routine Cleaning phase are a function of
the loadings at the completion of the renovation and the frequency and
efficiency of routine cleaning. Unlike previous phases which used a single
loading to represent the entire phase, a time series of loadings is developed
for this phase to account for the decrease in dust loadings that occurs over
time with repeated routine cleanings. The indoor dust loading for cleaning
iteration Xis calculated using the following equation:
DLOADbn,ph=4@x,co, vin
where:
= DLOAD
1BN,PH=4@X-l,CO, VIN
: RoutineCleanEfficiencyx (Eq. 4-11)
DLOAD
1BN,PH=4@X; CO, VIN
DLOAD
BN,PH=4@X-1, CO, VIN
RoutineCleanEfficiencyx =
Total whole building average indoor dust BN
sample loading following cleaning X during phase 4
(i.e., routine cleaning phase), with Control Option
CO, and building of vintage VIN, in |ig/g
Total whole building average indoor dust BN
sample loading following cleaning X-l for phase 4
(i.e., routine cleaning phase), with Control Option
CO, and building of vintage VIN, in |ig/g
Post-activity cleanup efficiency for the cleaning
iteration X; calculated as a weighted average based
on the percentage of the building that is carpeted
(the remainder is assumed to be covered by hard
surfaces), the cleaning efficiency for carpet (which
is a function of how many times the carpet has been
cleaned), and the cleaning efficiency for hard
surfaces (which is a function of the Pb loading
remaining after the previous cleaning).
67
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
The indoor dust loading for time Tis calculated using the following equations:
DLOADbn,ph =4@T,CO,VIN ~ DLOADbn,ph =4@X, CO, VIN
X = T * CleanFrequency
where:
(Eq. 4-12)
(Eq. 4-13)
DLOAD
BN,PH=4@T, CO, VIN
Total whole building average indoor dust BN
sample loading for week T during phase 4 (i.e.,
routine cleaning phase), with Control Option CO,
and building of vintage VIN, in |ig/g
Cleaning iteration, unitless
Time, in weeks
Total whole building average indoor dust BN
sample loading following cleaning X during phase 4
(i.e., routine cleaning phase), with Control Option
CO, and building of vintage VIN, in |ig/g
CleanFrequency = Frequency of routine cleaning, in cleanings/week
Note that DLOADbn,ph=4@x=o,co,vin is assumed to be equal to DLOADBn,ph=4@t=o,co,vin, which is
the initial loading for the Routine Cleaning phase and is developed as described in Chapter 3.
DLOAD
X
T
'BN,PH=4@X, CO, VIN
Convert Loadings to Concentrations
The final step required to calculate indoor dust concentrations for the routine
cleaning phase is to convert the time series of whole building total indoor dust
loadings to a time series of indoor dust concentrations. For buildings of known
vintage, the Blue Nozzle (BN) sample whole building total loading values are
converted to indoor dust concentrations using the vintage-specific regression
equation described in Appendix C, which is summarized below:
DCONCph =4@T,CO, VIN
where:
(InterceptyjN+SlopevlN \n(DLOADBN PH=4@T co yjN)))
= £ (Eq. 4-14)
DCONC PH=4@T,CO,
VIN
Total whole building average indoor dust
concentration during week T of phase 4 (i.e., routine
cleaning phase), with Control Option CO, and
building vintage VIN, in |ig/g
68
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
DLOADBn,ph=4@t,co,vin = Total whole building average indoor dust Pb BN
sampling loadings during week T of phase 4 (i.e.,
routine cleaning phase), with Control Option CO,
and building vintage VIN, in |ig/m2
For buildings of unknown vintage, the following non-vintage-specific regression equation
(described in Appendix C) is used to convert Blue Nozzle (BN) sample loadings to
concentrations.
(InterceptyjN=unknown + SlopeyjN=unknown In(DLOADBN PH=4(^T co )))
DCONC ph =4@T,CO ~ & (Eq. 4-15)
where:
DCONCph=4@t,co = Total whole building average indoor dust
concentration during week T of phase 4 (i.e., routine
cleaning phase), with Control Option CO, and
building of unknown vintage VIN=unknown, in
l-ig/g
DLOADBn,ph=4@t,co = Total whole building average indoor dust Pb BN
sampling loadings during week T of phase 4 (i.e.,
routine cleaning phase), with Control Option CO,
and building of unknown vintage VIN=unknown, in
|ig/m2
4.2.2 Example for Single RRP Activity
The approach presented above is applied to the single RRP activity scenario using the inputs
provided in Exhibit A-6 in Appendix A. The indoor dust concentrations for the Background
(where Pre-renovation and Post-renovation both use the same background value), Post-
renovation (Dust Generating), Post-renovation (After Baseline Cleaning), and Post-renovation
(Routine Cleaning, Week 0) for each of the three housing vintages for the single activity example
are shown in Exhibit 4-6. This particular type of table can be used to examine patterns in time or
across controls. In general, the concentrations seen in either example or in future calculations
using this approach will reflect whatever relationships occur in the underlying OPPT Dust Study.
In fact, at times in the underlying OPPT Dust Study, the concentration associated with Post-
renovation (After Baseline Cleaning) is actually higher than the Post-renovation (Dust
Generating) concentration, and the Post-renovation (Routine Cleaning, Week 0) is higher than
the Post-renovation (After Baseline Cleaning) concentration.
69
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Exhibit 4-6. Whole-House Indoor Dust Concentrations for the Single RRP
Activity Example for the Different Renovation Phases
Concentration (pg Pb/g dust)
Exposure Phase
Baseline
Control
Option
Control
Control
Control
Option 1
Option 2
Option 3
Vintage <1940
Pre-renovation (Background)
201
201
201
201
Renovation (Dust Generating)
1,941
2,923
3,800
2,406
Renovation (After Baseline Cleaning)
674
621
639
482
Post-renovation (Routine Cleaning, Week 0)
674
686
639
524
Vintage 1940 to 1959
Pre-renovation (Background)
86
86
86
86
Renovation (Dust Generating)
1,029
1,534
1,983
1,269
Renovation (After Baseline Cleaning)
362
333
343
257
Post-renovation (Routine Cleaning, Week 0)
362
368
343
280
Vintage 1960 to 1979
Pre-renovation (Background)
64
64
64
64
Renovation (Dust Generating)
551
760
933
652
Renovation (After Baseline Cleaning)
237
222
227
179
Post-renovation (Routine Cleaning, Week 0)
237
241
227
193
After Post-renovation (Routine Cleaning, Week 0), the routine cleaning decreases the
concentrations over time. The values for the indoor dust concentrations at Post-renovation Week
0 (at the beginning of the routine cleaning phase) and Post-renovation Week 10 (in the middle of
the routine cleaning phase) are given in tables by vintage in Appendix E. In addition, the Weeks
to Background model metric gives the number of routine-cleaning weeks required to clean away
the renovation "spike" and return to background indoor dust levels. The number of weeks tends
to increase for later vintages, because the background value itself is lower in these later vintages
and more cleaning weeks are required.
An example of a full time-series containing the Pre-renovation, Renovation, and Post-renovation
phases for the < 1940 vintage for all Control Options is shown in Exhibit 4-7. Note that the
actual time-series extends 312 weeks (6 years); it has been truncated in this figure for
presentation purposes. The concentrations are plotted on a log scale and thus exponential
decreases in concentration during the Post-renovation (Routine Cleaning) phase appear as
straight lines.
70
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Exhibit 4-7. Indoor Dust Concentrations Time Series for the Single Activity Example,
< 1940 Vintage
10000
D>
o>
A
tfl
3
a
n
o.
1000
100
-15
— Baseline Control Option
- Control Option 1
¦—Control Option 2
Control Option 3
15 25
Weeks
35
45
55
4.2.3 Example for Multiple RRP Activities
The approach presented above was applied to the multiple RRP activities scenario using the
inputs provided in Exhibit A-7 in Appendix A. The multiple indoor activities example includes a
combination of four separate renovation activities, where the whole house loadings are summed
across activities and then converted to concentrations. The indoor dust concentrations for the
Background (where Pre-renovation and Post-renovation both use the same Background value),
Post-renovation (Dust Generating), Post-renovation (After Baseline Cleaning), and Post-
renovation (Routine Cleaning, Week 0) for each of the three housing vintages for the multiple
activity example are shown in Exhibit 4-8. This particular type of table can be used to examine
patterns in time or across controls.
71
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Exhibit 4-8. Whole-House Indoor Dust Concentrations for the Multiple Activities Example
for the Different Renovation Phases
Concentration (pg Pb/g dust)
Exposure Phase
Baseline
Control
Option
Control
Control
Control
Option 1
Option 2
Option 3
Vintage <1940
Pre-renovation (Background)
201
201
201
201
Renovation (Dust Generating)
8,588
9,803
11,713
9,136
Renovation (After Baseline Cleaning)
2,075
1,950
1,932
1,483
Post-renovation (Routine Cleaning, Week 0)
2,075
3,185
1,932
1,549
Vintage 1940 to 1959
Pre-renovation (Background)
86
86
86
86
Renovation (Dust Generating)
4,395
5,000
5,949
4,668
Renovation (After Baseline Cleaning)
1,098
1,033
1,024
790
Post-renovation (Routine Cleaning, Week 0)
1,098
1,669
1,024
824
Vintage 1960 to 1979
Pre-renovation (Background)
64
64
64
64
Renovation (Dust Generating)
1,766
1,958
2,250
1,853
Renovation (After Baseline Cleaning)
581
553
549
446
Post-renovation (Routine Cleaning, Week 0)
581
813
549
461
The values for the indoor dust concentrations at Post-renovation Week 0 and Week 10 and the
Weeks to Background metric are given in tables by vintage in Appendix E. In general, the
concentrations and Weeks to Background are all higher in the multiple activities example
compared to the single activity example. The Week 0 concentrations tend to increase for the
multiple activities example by a factor of two to five.
An example of the full time series of dust concentrations for the < 1940 vintage for all Control
Options is shown in Exhibit 4-9. Such plots permit diagnostics and comparisons across Control
Options in light of the phases mapped in Exhibit 3-4. For example, in addition to the largest
spike in concentration due to the Renovation (Dust Generating) phase, there is also the eleven
week Post-renovation (post-baseline cleaning) phase. Control Option 1 suffers another spike in
concentration at the end of this phase; this spike may occur because the post-verification OPPT
Dust Study values lead to whole house concentrations higher than the post-cleaning values.
72
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Exhibit 4-9. Indoor Dust Concentrations Time Series for the Multiple Activity Example
and < 1940 Vintage
Weeks
4.3 Outdoor Soil
The approach for estimating outdoor soil concentrations is summarized in Exhibit 4-10. The
elements of this process described in Chapter 3 are shown in grey and the elements described in
this section are shown in black (note that this exhibit is identical to Exhibit 3-8, with the
exception of which elements are in grey and black). This section describes how the general
approach illustrated in Exhibit 4-10 is applied to estimate indoor dust concentrations during each
phase of the different exposure periods.
73
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
74
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
4.3.1 Pre-renovation
The Pre-renovation exposure period represents the period of exposure before initiation of the
RRP activity or activities and thus consists of background contributions only. The approach for
estimating background outdoor soil concentrations is described in Section 3.3.3.
4.3.2 Renovation
Outdoor soil concentrations during the Renovation exposure period are estimated in this
approach based on the Pb loading associated with the activity-Control Option combination, the
size of the yard, the size of the areas impacted by the activity, and the background concentration.
There are two steps involved in calculating outdoor soil concentrations during the Renovation
exposure period: (1) calculating soil concentrations in the different regions of the yard; and (2)
calculating whole yard average concentrations. These steps are described below.
Calculate Concentrations in Different Regions of Yard
Concentrations are estimated for three sections of the yard; dripline (roughly adjacent to house or
building); nearby area (adjacent to dripline); and remainder of yard. The concentration in the
remainder of the yard is assumed to be equivalent to the background soil concentration, which is
developed as described in Section 3.3.3.2. The concentrations in the dripline and nearby areas
are calculated based on the estimated loadings from the activity or activities in the two areas and
assumed soil characteristics. These concentrations are estimated using the following equation:
ft2
SLOADjnr Arrrn*0W\-J
LOC,ACT,CO 2
SCONCloc,act,co = ——— — cm_ (Eq. 4-16)
where:
SoilDepth * SoilDensity
SCONCLoc,act,co = Outdoor soil concentration in location LOC (where
LOC = dripline [A] or nearby area [B]) during
Activity ACT for Control Option CO, in |ig/g
SLOADLoc,act,co = Pb loading in location LOC (where LOC = dripline
[A] or nearby area [B]) during Activity ACT for
Control Option CO, in jug/ft2
SoilDepth = Soil mixing depth, in cm
SoilDensity = Density of soil, in g/cm3
75
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Calculate Whole Yard Average Concentration
The whole yard soil concentration is calculated as an area-weighted average of the
concentrations in the dripline, nearby area, and remainder of yard using the following equation:
SCONCact.co = PercArea,Loc=A * SCONCloc=aact,co + PercAreawc=B * SCONCloc=b,act,co +
PercAreaLoc=Rest * SCONCbg
(Eq. 4-17)
where:
PercAreaLoc=A = Percent of yard comprised of the dripline area (i.e.,
LOC = A)
SCONCloc=a,act,co = Outdoor soil concentration in dripline area (i.e.,
LOC = A) during Activity ACT for Control Option
CO, in |ig/g
PercAreaLoc=B = Percent of yard comprised of the nearby area (i.e.,
LOC = B)
SCONCloc=b,act,co = Outdoor soil concentration in nearby area (i.e., LOC
= B) during Activity ACT for Control Option CO, in
l-ig/g
PercAreaLOc=Rest = Percent of yard not comprised by the dripline and
nearby areas (calculated as 1- PercAreaLoc=A -
PercAreaLoc=B)
SCONCbg = Background outdoor soil Pb concentration, in |ig/g
(see Section 3.3.3.2)
4.3.3 Post-renovation
For the No Plastic Control Option, it is assumed that there is no cleanup or degradation of Pb in
outdoor soil; therefore, the Post-renovation exposure period concentrations are identical to the
Renovation period concentrations. For the With Plastic Control Option, outdoor soil
concentrations for all Control Options during the Post-renovation exposure period are estimated
using the same approach used for the Renovation period, with the Post-renovation exposure
period Pb loadings used instead of the Renovation period loadings.
4.4 Drinking Water
All of the exposed populations (i.e., children under six years of age) are assumed to consume
drinking water with the same "typical" Pb exposure concentration. While there is a rather large
76
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
amount of data in the literature, in many cases, the measurements represent "first-draw" samples,
non-random ("priority") samples, or samples from communities where Pb levels were known to
be elevated. After reviewing the literature, the average drinking water concentration exposure
was estimated to be 4.61 |ig/L, based on data from two recent studies of residential water
concentrations in U.S. and Canadian homes and apartments (Moir et al. 1996, Clayton et al.
1999). The range of values seen in these studies (0.84 to 16 |ig/L) was considered to be
representative of randomly sampled residential water in houses constructed since Pb pipe and
solder were banned for residential use. The selected value is close to the "default" value (4.0
(j,g/L) recommended for use with the IEUBK blood Pb model when evaluating the blood Pb
impacts of soil contamination (USEPA 1994). Much higher values have been encountered in
homes with Pb piping and/or very corrosive water. For purposes of this approach, it is assumed
that drinking water exposures are not impacted by RRP activities, and all of the exposed children
are assumed to receive the same age-specific exposures.
4.5 Diet
It is expected that young children will be exposed to Pb in the foods they consume. In this
approach, all exposed children are assumed to receive the age-specific estimates of dietary Pb
intake developed by EPA's Office of Solid Waste and Emergency Response (USEPA 2006c).
EPA developed these estimates by analyzing food consumption data from the NHANES III
survey conducted by the National Center for Health Statistics, and food residue data from the
U.S. Food and Drug Administration's (FDA) Total Dietary Study. The daily intake values
published by EPA and summarized in Exhibit 4-11 are considerably lower than those developed
using the same methodology in the 1980s and 1990s. Pb concentrations in food have decreased
dramatically since the prohibition of Pb solder in food containers in 1982. For purposes of this
approach, it is assumed that dietary exposures are not impacted by RRP activities.
Exhibit 4-11 Summary of Non-Water Dietary Pb Intake Estimates a
Age
(months)
Updated Dietary Pb
Intake Estimate (|jg/day)
Oto 11
3.16
12 to 23
2.60
24 to 35
2.87
36 to 47
2.74
48 to 59
2.61
60 to 71
2.74
72 to 84
2.99
a Data derived from U SEP A 2006b.
4.6 Sensitivity and Monte Carlo Analyses
4.6.1 Overview
Given the variability in characteristics of U.S. buildings covered by the proposed RRP rule and
the uncertainty and variability in Pb dust loading data for the selected activities, it is important to
77
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
characterize the impacts of this variability and uncertainty on the estimated dust concentrations.
The approach selected involves two stages: (1) a sensitivity analysis designed to estimate the
influence of the different inputs on the estimated dust concentrations; and (2) a Monte Carlo
analysis that estimates the reasonable range of dust concentrations given the uncertainty and
variability of these parameters. Both of these analyses were performed for both the single
activity and multiple activities examples for all three building vintages (< 1940, 1940 to 1959,
and 1960 to 1979). Complete tables of sensitivity and Monte Carlo analysis results for these
examples can be found in Appendix E. Recall that indoor air and outdoor soil are not included in
the sensitivity analysis. Air concentrations are taken directly from the OPPT Dust Study
(Battelle 2007) and do not require additional calculations. Variability from that study was used
in the Monte Carlo analyses. The outdoor soil concentrations used for the single indoor activity
and multiple indoor activities examples are constant background values and thus are not varied in
the uncertainty analyses. This approach, however, could be applied to outdoor soil
concentrations for scenarios involving outdoor activities or indoor activities with outdoor loading
estimates.
4.6.2 Sensitivity Analysis
4.6.2.1 Approach
A sensitivity analysis was performed to evaluate the sensitivity of the dust concentrations over
time to changes in inputs within a reasonable range. The selected approach considers the relative
impact of each input on estimated indoor dust concentrations, independent of the range of
possible values. Two measures of sensitivity, elasticity and sensitivity score, were estimated for
each input. Elasticity indicates "structural" sensitivity, while sensitivity score indicates "actual"
sensitivity after accounting for the estimated variability in an input property. The elasticity is
calculated as the percent change in a model output resulting from a one percent change in an
input value. The elasticity provides information useful for understanding how the model
operates and is used to compare with expected results, given knowledge of the model and the
processes being simulated. The sensitivity score, which is calculated as the product of the
elasticity and the coefficient of variation (CV), is useful in the context of assessing the influence
of input properties, or how the variability of the input property affects the variability of the
results.
For the sensitivity analysis, the inputs that are expected to most strongly influence the Post-
renovation Week 0 and Week 10 concentrations and the Weeks to Background metric were
selected. These inputs are listed in Exhibit 4-12.
78
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Exhibit 4-12 Inputs Included in Sensitivity Analysis
Input Variable
Varies by
% Area of Building - Adjacent Room
Activity
% Area of Building - Workspace
Activity
% Area of Building - Carpet
None
Cleaning Frequency
None
Load-concentration Intercept
Vintage
Load-concentration Slope
Vintage
Background Loading - Indoor
Vintage
Adjacent Loading - Post Cleanup
Control Option and Activity
Adjacent Loading - Post Verification
Control Option and Activity
Rest of Building Loading - Post Cleanup
Control Option and Activity
Rest of Building Loading - Post
Verification
Control Option and Activity
Workspace Loading - Post Cleanup
Control Option and Activity
Workspace Loading - Post Verification
Control Option and Activity
In each case, the input value was increased by 10 percent and the changes in the Post-renovation
Week 0 and Week 10 concentrations and the Weeks to Background metric were assessed.
The sensitivity of estimated indoor dust concentrations to one additional input, routine cleaning
efficiency, was assessed separately for feasibility reasons. Given the variable nature of the
routine cleaning efficiency estimates, this input parameter could not be analyzed in the same way
as the constant inputs. To analyze the sensitivity of the modeling results to this input, three
additional sets of cleaning efficiencies for carpet and hard surface (representing minimum,
maximum, and arithmetic mean efficiencies for a different data set than was used in the primary
methodology) were developed (see Appendix B). Indoor dust concentrations for the single
activity and the multiple activities examples included in the sensitivity analysis were then re-
calculated using these efficiencies. The results using the three sets of cleaning efficiencies
derived using the alternative methodology were then compared to the results using the primary
methodology for the single activity and multiple activities examples and the percent change
calculated.
The complete results of both parts of the sensitivity analysis with the examples are presented in
Appendix E and summarized below.
79
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
4.6.2.2 Results - Sensitivity Analysis Excluding Cleaning Efficiency
The indoor dust concentrations are most sensitive to the Load-concentration Slope and Intercept.
This suggests that this conversion equation is driving the concentration variability in this portion
of parameter space. The Weeks to Background metric is relatively insensitive to this equation
because the equation affects the loadings dependent on activities and the Background Loadings
in a similar way.
The indoor dust concentrations at Week 0 and Week 10 are relatively insensitive to the other
variables as measured by the elasticity. Control Options 1 and 3 have no sensitivity to the Post
Cleaning loadings and Baseline Control and Control Option 4 have no sensitivity to the Post
Verification loadings. This result derives from the fact that the Week 0 concentrations are either
Post Cleaning or Post Verification, depending on the Control Option.
The Weeks to Background metric is most sensitive to the % Building Carpet, the Background
Loading, and the Cleaning Frequency, because these are the primary drivers of the cleaning
efficiency and the number of iterations required to reach background.
The Week 0 concentration, the Week 10 concentration, and the Weeks to Background tend to
have higher elasticities in the earlier vintage, compared to the later vintages.
Elasticities are comparable across the single and multiple activities examples. The exception is
the load-concentration slope, which has a higher elasticity in the multiple activities case.
The sensitivity analysis was performed by varying parameters about their mean value.
Elasticities are not constant in all parts of parameter space because the model is non-linear.
Thus, other variables may prove more important when modeled concentrations are far from the
mean values. These variables will be discussed in Section 4.6.3.
4.6.2.3 Results - Cleaning Efficiency Sensitivity Analysis
The alternative cleaning efficiency results for hard floor tend to be lower than the primary
cleaning efficiency results for the range of loadings explored in both the single and multiple
activity examples. The cleaning efficiency remains at the lower bound of the alternative
cleaning efficiency equation (1 percent) for all whole-building loadings in both the multiple and
single activity examples in the 1960 to 1979 vintage.
Using the alternative cleaning efficiency equations tended to create higher initial carpet cleaning
efficiencies and lower initial hard wood efficiencies. Comparing the primary cleaning efficiency
results to the results derived from using the alternative cleaning efficiency's Mean equations in
the Workspace compartment, carpet efficiency went from 53 percent to 67 percent and hard floor
efficiency went from 13 percent to 1 percent in the first cleaning week. Because the buildings
are assumed to be 72 percent carpet, the overall cleaning efficiency went up in the initial weeks,
creating correspondingly lower concentrations.
80
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
A lower cutoff of 1 percent was imposed, however, on both equations in the alternative cleaning
efficiency methodology (hard floor and carpet). Because of the particular value of the Pb
loading, this meant the alternative methology's Mean equation always gave the lower cutoff
(1%) for hard wood, whereas the primary methodology always saw a value of 13% (the Mid
value for the particular loading range). Conversely, by the fourth iteration, carpet cleaning
efficiencies in the Workspace were very small in both methodologies. Thus, the overall cleaning
efficiencies were smaller in the alternative cleaning efficiency methodology in these later weeks.
The Week 10 concentrations and the Weeks to Background metric were correspondingly higher
for the alternate cleaning efficiency methodology. In particular, it took twice as long to return to
background for the single activity example in the 1960 to 1979 vintage.
The scenario tested used the vintage with the lowest background concentration (the 1960 to 1979
vintage). Other scenarios which have higher background concentrations and higher Weeks to
Background would be less sensitive to the cleaning efficiency.
4.6.3 Monte Carlo Analysis
4.6.3.1 Approach
The underlying variability in the input parameters leads to variability in the time series of indoor
dust Pb concentrations. To capture this variability in the blood Pb and IQ estimates, Monte
Carlo simulations were performed for both the single activity and multiple activities examples
for all vintages and all Control Options. Because the full time series (rather than just Post-
renovation Week 0 and Week 10) were needed as inputs to the blood Pb modeling, additional
variables were added to the Monte Carlo analysis. In addition, the Load-concentration equation
slope was not included because adequate upper and lower bounds could not be determined with
any confidence. The Load-concentration equation intercept, however, is included. The variables
included in the analysis are summarized in Exhibit 4-13.
An approximate distribution was obtained for each input parameter by choosing either a normal
or lognormal distribution, selecting the Mid value as the mean (normal) or geometric mean
(lognormal), and using the standard deviation (SD) provided in the data or calculated based on
the low and high input parameter values (see Appendix D for details). The only exception is the
Cleaning Frequency, which was modeled using a uniform distribution. In a given Monte Carlo
iteration, a value was randomly selected (sampled) from these distributions for each input
parameter. In general, these parameters were treated as independent variables. The exception is
the % Area of Building - Adjacent, which must be less than 100 percent minus the % Area of
Building - Workspace. If the sampled value did not satisfy this criterion, it was resampled until
the criterion was met.
Based on the sampled input parameters, the resulting indoor dust concentrations time series were
calculated. This process was repeated 20,000 times to adequately resolve the tails of the input
parameter distributions.3 Aggregating the 20,000 model output values for each week gives
3 A more complete description of the Monte Carlo tool is available in Appendix D.
81
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
information about the approximate distribution of the indoor dust concentration for that week.
The maximum, 99th percentile, 95th percentile, 75th percentile, mean, median, 25th percentile, 5th
percentile, and minimum values are reported for each metric in each Control Option (see
Appendix E).
Exhibit 4-13 Tnpul Variables Used in the Monk* Carlo Analysis
Input Variable
Varies by
% Area of Building - Adjacent Room
Activity
% Area of Building - Workspace
Activity
% Area of Building - Carpet
None
Cleaning Frequency
None
Load-concentration Intercept
Vintage
Background Loading - Indoor
Vintage
Adjacent Loading - Post Work
Control Option and Activity
Adjacent Loading - Post Cleanup
Control Option and Activity
Adjacent Loading - Post Verification
Control Option and Activity
Rest of Building Loading - Post Work
Control Option and Activity
Rest of Building Loading - Post Cleanup
Control Option and Activity
Rest of Building Loading - Post
Verification
Control Option and Activity
Workspace Loading - Post Work
Control Option and Activity
Workspace Loading - Post Cleanup
Control Option and Activity
Workspace Loading - Post Verification
Control Option and Activity
4.6.3.2 Monte Carlo Results
The percentiles, mean, maximum, and minimum values for the single activity and multiple
activities examples for all vintages are shown in Appendix E for the Post-renovation Week 0 and
Week 10 concentrations and the Weeks to Background metric. The overall range in the model
metrics is large. For example, the Single Activity < 1940 vintage under the Baseline Control
Option, the Week 0 concentrations have a maximum value of 12,577 |J,g/g, a minimum value of
33 ng/g, and a mean of 906 |~ig/g.
Exhibit 4-14 shows the 95th, 75th, 50th, 25th, and 5th percentiles for the Week 0 concentration for
the < 1940 housing vintage in the single activity example across the different Control Options.
The curves indicate that while the 5th percentile cases are nearly uniform across Control Options,
the 95th percentile see considerable spread, with the Base Control Option and Control Option 1
having higher 95th percentile concentrations than Control Option 2 and Control Option 3. This
shape indicates that the distributions for the less strict Control Options (Base and Control Option
1) are wider than the Control Option 2 and Control Option 3 cases.
82
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Exhibit 4-14. Monte Carlo Percentiles for Week 0 Concentration
for the < 1940 Single Activity Example
2,500
2,000 -
~ Base Control Option
¦ Control Option 1
~ Control Option 2
~ Control Option 3
5 1,500
O
o
o 1,000 -
500 -
Percentile
rth
75th, 50th,
25th, and 5th percentiles for the Week 0 concentration for
Exhibit 4-15 shows the 95
the < 1940 housing vintage in the multiple activities example across the different Control
Options.
In this case, Control Option 1 is always substantially higher than the other Control Options for
all percentiles. This phenomenon is caused by a high measurement of the tool room loading in
the Three-Cutouts renovation activity in the OPPT Dust Study. Because the low, mid, and high
estimates were all high for this case, all percentiles are much higher than the other Control
Options.
83
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Exhibit 4-15. Monte Carlo Percentiles for Week 0 Concentration
for the < 1940 Multiple Activities Example
14,000
¦ Base Control Option
12,000 ¦ Control Option 1
~ Control Option 2
~ Control Option 3
10,000
O
O
O
c
d)
Q1
8,000
6,000 -
4,000 -
2,000
4.7 Uncertainties and Limitations
Whole house loadings are created by taking area averages across different rooms. These area
averages are based on assumptions about the percent of the house occupied by each room for
each renovation activity. These percentages are uncertain, however, since a window
replacement, for example, could occur in either a living room (large "footprint") or a bathroom
(small "footprint"). These percentages have been varied by overall activity (kitchen renovation
versus window replacement versus three-cutouts, etc), but they have not been varied across
different instances of the same activity (all window replacements use the same percentages).
Thus, these assumptions introduce uncertainty into the overall whole house exposure levels.
Indoor dust loadings are summed across the different activities to represent total loadings from
multiple renovation projects. This technique is applied to whole house loadings and assumes
that the average loadings are additive since they occur in the same total surface area. This
process, however, smears out any gradients that occur in different locations of the house due to
the different activities. It also does not take into account the different "footprints" of different
activities in different portions of the house.
Indoor air concentrations are averaged across the different activities to represent total air
concentrations from multiple renovation projects. Unlike the dust concentrations, the air
84
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
concentrations are averaged (rather than summed) to account for the fact that concentrations will
never stay elevated for a full week (the smallest time increment in the exposure model) due to
settling. Thus, averaging instead of summing acknowledges that different activities may lead to
elevated air concentrations at different times during the week. This averaging represents
uncertainty, however, since the actual duration of elevated air from each activity in each part of
the house is not resolved.
Whole house concentrations are calculated for each exposure phase for each Control Option.
These Control Options can only be compared cautiously, however, owing to the underlying
methodology of the OPPT Dust Study. In the Dust Study, different Phases (or Control Options,
according to the vocabulary of this approach) were generally completed in different houses. In
general, the different Control Options for window replacements occurred in at least two different
houses, while kitchen replacements occurred in completely mutually exclusive houses. If each
house has different Pb content in the paint and other media, there is no reason to expect that a
stricter Control Option (e.g., Control Option 3) in one house will give lower dust concentrations
than a less strict Control Option (e.g., Base Control Option) in another house. Thus, the use of
Dust Study data limits the degree of comparison possible between Control Options. The overall
exposure level (and the subsequent blood Pb level and IQ reduction) for each Control Option
relative to background, however, can be assessed.
Children are assumed to occupy the entire house or the entire yard equally during the renovation
project. No effort is made to ensure that children remain outside the work area during the
renovation. Thus, this approach gives a conservative estimate of exposure. This assumption is
particularly suspect in the multiple activities example, when the project likely encompasses the
entire house and the child likely would be removed to another house during the renovation. The
current approach theoretically allows this alternative to be explored by setting the exposure to
background levels during the Renovation period, but this alternative has not been explored in the
current examples. For outdoor RRP activities especially, and for COFs, it is likely that a child
could occupy only part of the space but, also, outdoor soil may contribute differentially to indoor
exposures. This uncertainty will be handled by including an additional step prior to proceding to
the calculations of chapter 5.
The conversion of loadings to concentrations introduces significant uncertainty into the modeled
dust concentrations. The regression itself is based on limited data and the regression is
recognized as a limitation of the study. In addition, the multiple activity example Dust
Generating phase loadings are higher than the maximum loadings used in the regression,
reflecting that the regression model is being extrapolated for this case, which introduces
additional uncertainty. Finally, this conversion is being used for both carpeted and hard-floor
surfaces, when the wipe samples only apply to hard floors. Although this regression introduces
uncertainty, conversion remains a necessary step, since the blood Pb models are configured to
accept concentrations and not loadings. Under an assumption that the HUD-derived
relationships can be generalized, this regression represents the best known relationship available.
An alternate approach to obtaining concentrations could be to use the loadings in concert with
assumptions about the amounts (mass) of dust in the renovated spaces, the amount taken up on
contact by the child, and the amounts transferred to the child's mouth. Such an approach is
85
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
adopted, for instance, in the draft All-Ages Lead Model (US EPA, 2005b), which is not being
employed for blood Pb modeling in this analysis.
The routine cleaning efficiency data are another significant limitation of this approach. There
are limited studies that are appropriate for this type of approach and even those that are
appropriate have significant limitations. These data play an important role in determining how
quickly the post-activity concentrations return to background for indoor activities. Thus, the
long-term average exposure levels could be either over or under estimated as a result of the
cleaning efficiency data. Because the renovation activity level concentrations are by far the
largest "spike" in exposure, however, they are the most significant contributor to blood Pb levels
and IQ loss, and long-term slight elevations relative to background are less important.
The cleaning efficiency is applied to the whole house loadings, rather than to individual rooms.
By averaging out the room-to-room gradient, cleaning efficiency on hard floor (which is based
on the loading level) may be either over or under estimated.
In order to calculate the range of dust concentrations for each week of the child's life, Monte
Carlo runs are performed using assumptions about the underlying distribution of each input
parameter. For each parameter, a low, mid, and high value is estimated, and a coefficient of
variation is estimated from these values. In general, the low, mid, and high values came from
only a limited number of data values, suggesting the distribution likely is not adequately
captured. In addition, the calculation of coefficient of variation implements assumptions about
the shape of the distribution (normal or log-normal) and the values represented by low and high
(assumed to be two standard deviations from the mean). In both steps, significant uncertainty is
introduced into the input parameters, suggesting that the estimates for the "tails" of the Monte
Carlo dust predictions are highly uncertain. The median Monte Carlo estimates, based on the
mid estimates, are more robust.
The Monte Carlo tool (see Appendix D) is run for 20,000 iterations. This number is likely
sufficient to resolve the middle portions of the distributions. No tests are made, however, to
determine if 20,000 iterations produces convergence in the tails of the distributions. Thus, there
are additional uncertainties in the upper and lower percentiles for the Monte Carlo dust
predictions.
86
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
5. ESTIMATES OF BLOOD PB LEVELS
This chapter describes the approaches and methods that can be used to evaluate children's blood
Pb levels from RRP activities from the time of birth to age six. The general approach is shown
in the flowchart in Exhibit 5-1. Pb concentrations in the activity-influenced exposure media
(indoor air, indoor dust, and outdoor soil) and background exposure media (ambient air, drinking
water, and diet), as well as Pb exposure and intake assumptions, serve as inputs for biokinetic
blood Pb models (discussed in Section 5.2). It is recognized that some portion of Pb in the
background exposure media may derive from RRP-activity-related sources, for example, if the
exposed child consumes homegrown vegetables grown in Pb-contaminated garden soil.
However, this proportion is assumed to be small for most of the exposure scenarios considered in
this analysis.
Exhibit 5-1. Flow Chart Illustrating the Approach to Calculate Blood Pb Levels
In each exposure pathway, the relationship between exposure concentration and Pb uptake
(absorbed dose) is defined by a range of factors related to physiological processes and to the
chemical and physical properties of the exposure media. These factors include respiratory
volume, soil and dust ingestion rates, and gastrointestinal (GI) absorption fractions for diet,
water, and soil/dust, which determine how much Pb is absorbed from each medium. Values for
these factors differ with age and across exposure media, as discussed in Section 5.2.
In this approach, blood Pb levels for six different hypothetical children are modeled from birth
until six years of age. Exposure profiles are defined for each child so as to simulate the
occurrence of the renovation project at the beginning of a different year of their life (birth, first
birthday, second birthday, etc.). Prior to the renovation, these children experience background
level exposures from all media, and after the renovation the dust and air concentrations are
decreased due to routine cleaning and settling. Activity-related Pb exposure concentrations vary
weekly depending on the activities and year of life being modeled.4 Ambient air, drinking water,
4 For the examples presented in this document, outdoor soil concentrations remain constant throughout the
simulations. This is a result of the types of activities being evaluated and the data sources (specifically, the OPPT
Dust Study), which did not include contributions to outdoor soil concentrations from indoor activities. The
87
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
and dietary Pb exposures are assigned the same age-specific values in all of the exposure
scenarios, as described in Section 5.2.
5.1 Selected Model(s)
Two different biokinetic models are applied to predict blood Pb levels in this approach: the
Integrated Exposure Uptake Biokinetic (IEUBK) Model for Children (hereafter referred to as the
IEUBK model) and the International Commission for Radiation Protection (ICRP) model
(hereafter referred to as the Leggett model). Both models are well documented, widely used, and
have been subject to a range of testing and calibration exercises (see Section 4.4 of USEPA
2006a).
A third model, which is empirical, uses a regression-based approach for predicting blood lead
levels on the basis of environmental concentrations and other variables (Lanphear et al., 1998).
There are, however, obstacles to using it for the estimation of blood lead from exposure during
RRP activities. The biggest obstacle is the lack of a dynamic component or term. Because it
predicts blood leads only at a single point in time, its results cannot easily be compared to the
time-varying results of the biokinetic models. In addition, there is nothing in the development of
the empirical model that predicts how it would respond to short-term variations in exposure, such
as those anticipated with renovation activities. Thus, it has not been applied in the RRP analysis.
5.1.1 IEUBK Model
The IEUBK Lead Model Version 1.0 Build 263 (USEPA, 2005a) was used to model blood Pb
levels for all exposure scenarios. The IEUBK model consists of three main modules: the
exposure module, the uptake module, and the biokinetic module. The exposure module accepts
exposure concentration inputs for each of six exposure media: air, food (excluding water),
water, soil, indoor dust, and other. Exposure, intake, and uptake factors are combined with
exposure concentrations in the various media to estimate Pb uptake through the inhalation and
ingestion pathways. The IEUBK model provides default values for many input parameters that
can be adjusted by the user as appropriate for specific applications. These parameters include
medium-specific GI absorption fractions, and age-specific respiratory volumes, water intake,
dietary intakes of specific food classes, and soil and dust ingestion rates. The selection of model
input parameter values for this approach is discussed in Section 5.2.
The exposure module also includes default age-specific estimates of time spent outdoors, as well
as age-specific estimates of outdoor and indoor air Pb concentrations, inhalation rates, and
respiratory tract absorption fractions, all of which are used to estimate age-specific Pb inhalation
uptakes. The inhalation pathway absorption fraction combines both the deposition pattern of
inhaled Pb in the respiratory tract and absorption of deposited Pb, either directly from the
respiratory tract or from the GI tract after mucociliary clearance. Ingestion uptake of Pb is
calculated using absorption fractions that are specific to the ingested media (food, water, soil, or
approach presented here, however, can easily accommodate characterization of assumed or measured contributions
to outdoor soil, either from indoor or outdoor activities.
88
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
dust). Total GI Pb uptake is modeled as being composed of a saturable and an unsaturable
component.
In the biokinetic component of the IEUBK model, the movement of absorbed Pb (from ingestion
and inhalation) through various body "compartments" is simulated. The model is termed
"biokinetic," rather than "pharmacokinetic," because transfers between compartments are
controlled by first-order transfer coefficients (equivalent to first-order rate constants), rather than
being perfusion-controlled. A more complete description of the derivation and structure of the
IEUBK model can be found in USEPA (1994) and White et al. (1998).
The original purpose of the IEUBK model was to predict changes in long-term (annual average)
blood Pb concentrations based on constant or slowly varying exposures (USEPA 1994). The
approximate temporal resolution of the biokinetic component of the IEUBK is about one month
(USEPA, 2006a). The IEUBK is thus not ideal for use in estimating the impacts of short-term
variations in Pb exposures (such as the one-week peak exposures anticipated with renovation
activities). In the approach outlined in this document, Pb exposures and blood Pb variability for
each 7-day exposure period are modeled using the IEUBK's batch capability. Lifetime average
blood Pb concentrations (derived from the batch results) are used to estimate IQ loss, as
explained in Chapter 6.
5.1.2 Leggett Model
Leggett (1993) developed a sophisticated biokinetic model for Pb exposures, intended primarily
for use in radiological dosimetry, which has since been adopted as the standard model for that
purpose by the International Commission for Radiation Protection (ICRP). The model is capable
of predicting changes in blood Pb levels for exposed individuals over their entire lifespan (birth
to 90 years old). The compartmental structure of the Leggett model is more complex than that of
the IEUBK model, and was patterned after similar models developed by the ICRP to model the
age-specific biokinetics of calcium-like radionuclides (Leggett, 1993).
Differences between the structures of the Leggett model and the IEUBK model include:
89
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
• The published version of the Leggett model lacks the multipathway exposure module of
the IEUBK. The Leggett model accepts total respiratory and ingestion intakes
(administered doses) as inputs and calculates Pb uptake using age-specific absorption
factors.
• The central exchange compartment in the Leggett model is "diffusible plasma," rather
than the plasma-extracellular fluid (ECF) compartment used in the IEUBK model.
• The trabecular and cortical bone compartments in the Leggett model are divided into two
subcompartments each, one exchangeable, and one "non-exchangeable." Pb in the "non
exchangeable" compartments of both types of bone can be remobilized, but only
relatively slowly as a result of bone remodeling.
• Urinary excretion is modeled in the Leggett model as part of an integrated kidney
compartment, rather than separately as in the IEUBK model.
• In the Leggett model, the liver is modeled as two compartments with rapid and
moderately rapid Pb exchange, respectively. Other soft tissues are modeled as having
three compartments with differing exchange rates. Pb in brain tissue is explicitly
modeled.
Like the IEUBK model, the Leggett model is biokinetic, with exchange between compartments
modeled using first-order "transfer rates" (equivalent to rate constants). The values for the
transfer rates were estimated using a range of data from adult human radioactive tracer studies,
autopsy data from adults and children, and data from animal studies related to the absorption,
deposition, and excretion of Pb and chemically similar elements (Leggett, 1993).
Importantly, the Leggett model differs from the IEUBK model in that data from short-term
studies (on the time-scale of hours to days) are used to estimate parameter values for the most
rapid of the uptake and exchange processes, and thus the time resolution of the Leggett model is
much finer than that of the IEUBK model. Without violating any of the assumptions used to
derive the model the user may specify step lengths much shorter than those used in the IEUBK,
depending on the degree of time resolution required in the blood Pb predictions. The Leggett
model is capable of modeling the impacts of very short-term (even acute, one-time) Pb
exposures.
Leggett developed estimates of biokinetic parameters for six age categories: newborn (0 to 100
days), 1 year, 5 years, 10 years, 15 years, and 25 years and older, with age-specific transfer
parameters for children estimated by interpolation between the nearest values. Transfer factors
for children were adjusted to take into account the more rapid bone turnover (calcium/Pb
addition and resorption) in children compared with adults.
Predictions from the Leggett model have been compared with the deterministic predictions of
blood Pb levels generated by the IEUBK model, using the IEUBK default inputs (Pounds and
Leggett, 1998). In that comparison, the Leggett model predictions were substantially higher than
those of the IEUBK model. As described in Section 5.4, the same pattern appears during the
analysis of blood Pb impacts of the single and multiple RRP activities examples.
90
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
5.2 Derivation of Blood Pb Estimation Inputs
In order to obtain estimates of blood Pb, both the time series of the media inputs (air, indoor
dust, outdoor soil, diet, and drinking water) and the "exposure factor" values that govern intake
and absorption processes must be specified.
5.2.1 Time Series of Media Concentrations: Exposure to Renovation Activity at
Different Ages
Indoor dust, air (ambient and indoor), outdoor soil, and water concentrations and dietary Pb
intakes are the necessary exposure concentration inputs for the blood Pb models. Indoor dust
and air concentrations were estimated for each exposure period and phase of the renovation
activities, as discussed in Section 3.1. For the indoor single activity and multiple activities
examples presented below, the outdoor soil Pb concentration was always assumed to remain at
background levels throughout the life of each theoretical child. In addition, the drinking water
and dietary concentrations were set at age-specific background levels in all the scenarios (see
discussion below).
In order to determine the range of blood Pb levels associated with the range of exposure model
parameters, the Monte Carlo time series estimates of indoor dust concentrations were used as
inputs to the blood Pb models. For each week of the exposure model, the 5th percentile, median,
mean, and 95th percentile dust Pb concentrations are calculated from the 20,000 Monte Carlo
iterations. Each of the percentile values becomes a separate weekly time series for indoor dust
concentrations, and is run separately using the same air, outdoor soil, water, and dietary Pb
inputs.
Indoor air and indoor dust Pb concentrations are estimated in weekly increments. Each scenario
involves estimating blood Pb concentration profiles for a child associated with the specified
series of weekly exposure concentration estimates corresponding to the Renovation (Dust
Generating), Renovation (After Baseline Cleaning), Post-renovation (Routine Cleaning), and
Post-renovation (Background) phases (see Section 3.1 for phase descriptions). Depending on the
activities and scenarios being evaluated, dust exposures could remain elevated above background
for many weeks.
It is expected that these complex exposure scenarios will have different impacts on estimated
blood Pb levels, depending on the age of the child when renovation occurs. In order to explore
how the blood Pb levels change when the RRP activity occurs in a different year of a child's life,
six different hypothetical children are modeled for each exposure period, as shown in Exhibit
5-2. Child 1 experiences the renovation beginning at birth. The remaining weeks after the
renovation follow the weekly Post-renovation concentrations from the indoor dust and indoor air
exposure modeling. Child 2 experiences Pre-renovation (Background) levels during the first
year of life and the renovation occurs at the beginning of the second year of life, with the
remaining weeks after renovation following the Post-renovation concentrations, and so on for
Child 3, Child 4, Child 5, and Child 6.
91
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Exhibit 5-2. Definition of Modeled Hypothetical Children and
Child 1
Child 2
Child 3
Child 4
Child 5
Child 6
Age 0 to 1
(Year 1)
Renovation
Occurs
Pre-
renovation
Pre-
renovation
Pre-
renovation
Pre-
renovation
Pre-
renovation
Time
*
Age 1 to 2
(Year 2)
Post-
renovation
Renovation
Occurs
Pre-
renovation
Pre-
renovation
Pre-
renovation
Pre-
renovation
Age 2 to 3
(Year 3)
Post-
renovation
Post-
renovation
Renovation
Occurs
Pre-
renovation
Pre-
renovation
Pre-
renovation
Age 3 to 4
(Year 4)
Post-
renovation
Post-
renovation
Post-
renovation
Renovation
Occurs
Pre-
renovation
Pre-
renovation
Age 4 to 5
(Year 5)
Post-
renovation
Post-
renovation
Post-
renovation
Post-
renovation
Renovation
Occurs
Pre-
renovation
Age 5 to 6
(Year 6)
Post-
renovation
Post-
renovation
Post-
renovation
Post-
renovation
Post-
renovation
Renovation
Occurs
As noted above, the exposure concentrations and Pb intake from background pathways (i.e.,
those assumed not to be impacted by renovation-drinking water and non-water diet) were also
parameter inputs to the blood Pb model. Drinking water concentrations were assigned a single,
constant value for all ages (see Section 4.4). In addition to drinking water, it is expected that
young children will be exposed to Pb in the foods they consume. The development of inputs for
dietary exposures is presented in Section 4.5.
92
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
5.2.2 Parameter Inputs Related to Pb Intake and Absorption (Uptake)
As discussed previously, there are a number of model inputs that govern how the exposure
concentrations are converted to absorbed Pb dose (uptake) estimates. These variables represent
the physiological and behavioral characteristics of the exposed population and the chemical and
physical properties of the exposure media that govern exposure and absorption by inhalation and
ingestion.
5.2.2.1 IEUBK Model
Because substantial data have become available since the IEUBK automated default values were
last updated, more recent information available in the literature was used to modify some of the
IEUBK inputs. These modifications are also being used in other Agency analyses (e.g., AQCD
for lead). A number of the values, listed in Exhibit 5-3, differ from the suggested default values
in the most current version of the IEUBK (USEPA 2005a). Children's daily ventilation rate
estimates were based on the EPA Exposure Factors Handbook (US EPA, 2002a). The inhalation
absorption fraction was set to 0.42, which falls within the range of 0.25 to 0.45 estimated in the
past for areas not influenced by point sources (US EPA, 1989), and is the IEUBK default.
Estimated children's direct water ingestion values were interpolated from values in EPA's
Children-Specific Exposure Factors Handbook (US EPA, 2002b); the GI absorption fraction of
Pb from water (and diet) was retained at the IEUBK default value of 50 percent, and is consistent
with OAQPS previous analyses of Pb uptake (US EPA, 1989). Age-specific dietary intake
values for Pb were revised to reflect the latest analyses of FDA and NHANES III data on food
consumption pattern and Pb residue levels (US EPA, 2006d).
Age-specific soil and indoor dust ingestion rates were retained at the IEUBK default values.
Similarly, the weighting factor for soil and indoor dust ingestion was also left at 45 percent soil,
despite limited data supporting this specific value (US EPA, 1989; US EPA, 1994). The IEUBK
generic default value for GI absorption of Pb from soil and indoor dust (0.30, or 30 percent) was
used. This value is generally consistent with more recently reported values, although individual
estimates vary widely. Finally, for maternal blood Pb, the IEUBK default value of 2.5 ug/dL
was used.
93
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Exhibit 5-3. IEUBK Parameter Input Values Used in this Approach
Parameter
IEUBK
Parameter
Name
Leggett
Variable
ID
Parameter Value
Basis/Derivation
IEUBK Default Age Ranges (Years)
Oto 1
1 to 2
2 to 3
3 to 4
4 to 5
5 to 6
Inhalation
Daily
ventilation rate
(m3/day)
Ventilation rate
RSVOL
4
5.1
6
6.8
7.8
8.8
ICRP (2002), with
interpolation for
intermediate ages.
Absolute
inhalation
absorption
fraction
(unitless)
Lung
absorption
AFLNG
0.42
USEPA (1989),
Appendix A
Indoor air Pb
concentration
Indoor air Pb
concentration
(percentage of
outdoor)
BRETH
100%
IEUBK default,
USEPA (1994)
Time spent
outdoors
Time spent
outdoors
(hours/day)
1
2
3
4
4
4
Drinking Water Ingestion
Water
consumption
(L/day)
Water
consumption
(L/day)
H20CNS
0.34
0.31
0.31
0.33
0.36
0.39
Based on value for
infants, 1 to 3 yr
olds, 1 to10 yr olds
(with trend lines
used to interpolate
intermediate age
ranges) (USEPA
2002b).
Water Pb
concentration
(pg/L)
Pb
concentration
in drinking
water (pg/L)
H20CONC
4.61
GM of values
reported in studies
of United States
and Canadian
populations
(residential water)
(Moir et al. 1996,
Clayton et al. 1999,
as cited in USEPA
(2006b), Section
3.3 Table 3-10).
Absolute
absorption
(unitless)
Total percent
accessible
H20A
50%
Assumed similar to
dietary absorption
(see "Total percent
accessible" under
Ingestion-Diet
below).
94
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Parameter
IEUBK
Parameter
Name
Leggett
Variable
ID
Parameter Value
Basis/Derivation
IEUBK Default Age Ranges (Years)
Oto 1
1 to 2
2 to 3
3 to 4
4 to 5
5 to 6
Diet
Dietary Pb
intake
(pg/day)
Daily Pb intake
(pg/day)
EAT
3.16
2.6
2.87
2.74
2.61
2.74
Estimates based on
(a) Pb food residue
data from U.S. Food
and Drug
Administration Total
Diet Study (USFDA
2001), and (b) food
consumption data
from NHANES III
(CDC 1997).
Absolute
absorption
(unitless)
Total percent
accessible
Fl
50%
Alexander et al.
(1974) and Ziegler
et al. (1978), as
cited in USEPA
2006 (Section
4.2.1). These two
dietary balance
studies suggest that
40-50% of ingested
Pb is absorbed by
children (2 weeks to
8 years of age).
Soil/Indoor Dust Ingestion
Soil/dust
weighting
factor
(unitless)
Soil/dust
ingestion
weighting
factor (percent
soil)
SLDST
45%
This is the percent
of total ingestion
that is soil. Value
reflects best
judgment and
consideration
(results published
by van Wijnen et al.
(1990), as cited in
(USEPA 1989). The
van Wijnen et al.
study looked at
tracer studies of
ingestion rates for
rainy days and non
-rainy days. It was
assumed that rainy
days were
associated with all
soil ingestion and
non-rainy days were
associated with a
combination of soil
and dust with the
delta representing
soil.
95
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Parameter
IEUBK
Parameter
Name
Leggett
Variable
ID
Parameter Value
Basis/Derivation
IEUBK Default Age Ranges (Years)
Oto 1
1 to 2
2 to 3
3 to 4
4 to 5
5 to 6
Total dust +
soil ingestion
(mg/day)
Amount of
soil/dust
ingested daily
(mg)
SLDSTIN
G
85
135
135
135
100
90
USEPA 1989, which
was based on
multiple studies
focusing on
children.
Absolute Gl
absorption
(soil and dust)
(unitless)
Total percent
accessible
SLAF,
DSTAF
0.30 for both soil and dust
USEPA (1989)
reflects evidence
that Pb in dust and
soil is as accessible
as dietary Pb and
that dust/soil
ingestion may occur
away from
mealtimes (resulting
in enhanced
absorption relative
to exposure during
meal events).
Other
Maternal
blood Pb
(Hg/dL)
Maternal blood
Pb
concentration
at childbirth,
pg/dL
2.5
IEUBK default,
USEPA (1994)
96
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
5.2.2.2. Leggett Model
Because the Leggett model accepts inhalation and ingestion uptakes rather than media
concentrations, air, dust, water, and soil concentrations are converted to uptakes using the same
parameters as outlined above for the IEUBK model. Specifically, Pb uptake by the inhalation
pathway is calculated as follows:
BRTCRN = RSVOL * BRETH * AFLNG (Eq. 5-1)
where:
BRTCRN = Pb uptake by inhalation ([j,g/day)
RSVOL = age-specific respiratory volume (m3/day)
BRETH = modeled exposure concentration ([j,g/m3)
AFLNG = inhalation absorption fraction (unitless)
The Pb uptake through ingestion was calculated as:
EATCRN = FOODUP + H20UP + SOILUP + DUSTUP (Eq. 5-2)
FOODUP = EAT * FI (Eq. 5-3)
H20UP = H20CNS * H20C0NC * H20A (Eq. 5-4)
SOILUP = SLDSTING * SLDST * SOILPB * SLAF (Eq. 5-5)
DUSTUP = SLDSTING * (1-SLDST) * DUSTPB * DSTAF (Eq. 5-6)
where:
EATCRN = total uptake through ingestion ([j,g/day)
FOODUP = uptake from food ([j,g/day)
EAT = daily Pb intake from non-water diet (^ig/day)
FI = GI absorption fraction for diet (unitless)
H20UP = uptake from drinking ([j,g/day)
H20CNS = direct residential drinking water consumption
(1/day)
H20C0NC = average Pb concentration in drinking water ([j,g/day)
H20A = GI absorption fraction for water (unitless)
SOILUP = daily Pb uptake from soil ([j,g/day)
SLDSTING = daily soil + dust ingestion (g/day)
SLDST = fraction of soil + dust ingestion that is soil
SOILPB = average soil Pb concentration (jag/g)
SLAF = absolute GI absorption fraction for soil Pb (unitless)
DUSTU = daily Pb uptake from indoor dust (|ig/day)
DUSTPB = average indoor dust Pb concentration (|ig/g)
DSTAF = absolute GI absorption fraction for dust Pb
(unitless)
97
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
The time-varying inhalation and ingestion uptakes (BRTCRN and EATCRN) were then put
directly into the Leggett model. Note that the Leggett variable IDs are included in Exhibit 5-3
with their corresponding IEUBK input parameters.
5.3 Estimation of Blood Pb Levels for Two RRP Examples
5.3.1 IEUBK Model Implementation
To implement the IEUBK model for the indoor residential single activity and multiple activities
examples, the weekly air and dust concentrations, the yearly dietary uptakes, and the constant
soil and drinking water concentrations are input into the IEUBK model. As noted above,
multiple percentile estimates of weekly dust concentration generated by Monte Carlo modeling
are run for each scenario for 312 weeks (ages 0 to 6). Because the IEUBK is a steady state
model, it was run in batch mode, with separate blood Pb calculations for each week of input data,
where the input concentrations are assumed to capture the midpoint of that week, and for each of
the six hypothetical children (Exhibit 5-2). The outputs are weekly blood Pb concentration
estimates for each of these children. The weekly estimates then are averaged to produce two Pb
metrics, quarterly averages and lifetime averages. Quarterly averages are created by averaging
over each quarter of the child's life, ages 0 to 6. The results are also averaged over the entire
lifetime of each child to provide lifetime averages as input for the IQ estimation model.
5.3.2 Leggett Model
To implement the Leggett model, a module has been added to the original FORTRAN code
(Pounds 2000) to allow the model to be run in batch mode. The intake and uptake equations are
implemented separately for each example using Excel® spreadsheet calculations and macros.
The ingestion and inhalation uptakes are input into the model assuming 100 percent absorption,
since the absorption fraction was already implemented in the uptake calculations. The model is
then run for each of the six theoretical children, with inhalation and ingestion uptakes varying
weekly and using a time step of 0.01 days. The resulting blood Pb levels are output daily, and
these values are in turn averaged over each quarter and over the entire lifetime of each child.
5.4 Examples
Below are estimated blood Pb levels associated with background exposures, and for the single
and multiple activity RRP examples. For the single and multiple activity RRP examples, the
data are presented for one- to two-year-old children (Child 2). Complete sets of tables for all
theoretical children can be found in Appendix F. Each section describes the steps of analysis
that could be undertaken to estimate blood Pb levels associated with any specific renovation
activity or activities.
5.4.1 Estimated Blood Pb Levels Associated With Background Exposures
The blood Pb models were first run using constant background values for all media
concentrations, without any RRP activities, for a child from birth to age six. The estimated
background concentrations for dust and soil vary by building vintage. Thus, blood Pb levels
associated with background exposures also vary with building age. Estimated median lifetime
98
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
average blood Pb levels for the Leggett and IEUBK models and the 50th percentile blood lead
levels reported in NHANES for children 1-5 years old by vintage are summarized in Exhibit
5-4. This comparison allows the use of NHANES data as a benchmark to evaluate the
performance of the blood Pb models. However, this comparison is valid only to the extent that
the exposure patterns of the NHANES study subjects (1-5 years old) match those of the modeled
children (0-6 years old). In addition, it should be noted that the houses included in the NHANES
survey do not all contain lead-based paint. As expected based on prior comparisons, the Leggett
model blood Pb estimates are two to three times higher than the IEUBK values. The IEUBK
values are closer to the NHANES values which can suggest that the Leggett model may be over-
predicting the blood Pb values. The NHANES values more likely reflect steady state, as do the
IEUBK values, whereas the Leggett values can accommodate current kinetics for RRP activities
underway or recently completed.
Exhibit 5-4. Comparison of Background Lifetime Average Blood Pb Levels
from the Leggett Model, IEUBK Model, and the NHANES Survey in |j,g/dL
Vintage
Leggett Model
Median
Estimate
IEUBK Model
Median
Estimate
NHANES 50th
percentile
<1940
13.9
4.5
2.6
1940 to 1959
6.3
2.2
2.1 to 2.8
1960 to 1979
5.4
1.9
1.6 to 1.8
Exhibit 5-5 shows the background blood Pb levels estimated for one- to two-year-old children
(Child 2) living in different age (vintage) houses. In addition to the medians, the exhibit shows
the blood Pb estimates associated with the mean and the 5th and 95th percentile indoor dust
exposure concentrations. In this case, the variation in background blood Pb levels across
percentiles is due to the assumed distribution of the background indoor dust concentrations. The
median values that are compared to the NHANES data are shown in bold. In general, the
background values vary by a factor of two between the 95th percentile and 5th percentile cases.
In addition, the IEUBK results are again consistently two to three times lower than the Leggett
results.
Exhibit 5-5. Background Lifetime Average Blood Pb Levels
for All Indoor Dust Concentration Percentiles in |j,g/dL
Percentile
Leggett
IEUBK
<1940
1940 to
1959
1960 to
1979
<1940
1940 to
1959
1960 to
1979
95th Percentile
25.3
11.5
9.7
7.9
3.8
3.3
Median
13.9
6.3
5.3
4.5
2.2
1.9
Mean
15.5
7.0
6.0
5.0
2.4
2.1
5th Percentile
10.8
5.0
4.3
3.6
1.8
1.5
99
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
5.4.2 Blood Pb Level Estimation for the Single and Multiple RRP Acitivity
Examples
The time-varying RRP activity dust and air concentrations are then used as model inputs, along
with the background soil, water, and dietary values, to estimate the blood Pb levels for each of
the six children in the single activity example or the multiple activity example. These runs are
performed separately for all four Control Options, for each building vintage, and for each of the
dust concentration Monte Carlo metrics (5th percentile, median, mean, and 95th percentile).
The results can be presented as lifetime average Pb levels for each theoretical child. Exhibit 5-6
shows the lifetime average blood Pb levels for the child who experiences the single RRP activity
example starting at their first birthday (Child 2). The blood Pb levels are summarized for all four
Monte Carlo percentiles, for all housing vintages, and for all Control Options, with the median
values in bold to represent the central tendency estimate. As expected, the blood Pb levels tend
to decrease with newer buildings. In general, the 95th percentiles estimates are on the order of
three times the 5th percentile values.
Exhibit 5-6. Lifetime Average Blood Pb Levels for Child 2,
Single Activity Example in |j,g/dL
Leggett
IEUBK
Control Option and Percentile
<1940
1940 to
1959
1960 to
1979
<1940
1940 to
1959
1960 to
1979
Base Control Option, 95th Percentile
28.16
14.77
12.94
8.74
4.71
4.19
Control Option 1, 95th Percentile
28.30
14.97
13.08
8.76
4.74
4.22
Control Option 2, 95th Percentile
28.16
14.80
12.87
8.67
4.66
4.15
Control Option 3, 95th Percentile
27.16
13.68
11.98
8.38
4.40
3.91
Base Control Option, Median
15.38
7.27
6.05
4.94
2.49
2.09
Control Option 1, Median
15.51
7.36
6.10
4.97
2.51
2.11
Control Option 2, Median
15.41
7.32
6.07
4.93
2.49
2.10
Control Option 3, Median
14.92
7.01
5.89
4.81
2.41
2.04
Base Control Option, Mean
17.39
8.30
7.06
5.50
2.81
2.42
Control Option 1, Mean
17.55
8.41
7.13
5.53
2.83
2.43
Control Option 2, Mean
17.45
8.36
7.07
5.49
2.81
2.41
Control Option 3, Mean
16.79
7.93
6.77
5.34
2.69
2.33
Base Control Option, 5th Percentile
11.23
5.25
4.50
3.75
1.84
1.59
Control Option 1, 5th Percentile
11.28
5.27
4.52
3.76
1.85
1.59
Control Option 2, 5th Percentile
11.28
5.29
4.54
3.76
1.85
1.60
Control Option 3, 5th Percentile
11.15
5.21
4.49
3.72
1.83
1.58
Exhibit 5-7 shows the lifetime average blood Pb levels for the child who experiences the
multiple RRP activity example starting at their first birthday (Child 2). As in the single RRP
100
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
activity example, these runs were performed separately for all four Control Options, for each
building vintage, and for each of the indoor dust concentration percentiles from the Monte Carlo
simulations. The absolute blood Pb levels associated with the multiple activities scenario are all
higher than corresponding values in the single activities example. This is to be expected, since
the exposure concentrations and durations typically will be higher when multiple renovation
activities are undertaken. Because both models use the same background soil, diet, and water
concentrations/uptakes, the differences in the air and dust concentrations only partly determine
the overall blood Pb levels.
Exhibit 5-7. Lifetime Average Blood Pb Levels for Child 2,
Multiple Activities Example in |j,g/dL
Leggett
IEUBK
Control Option and Percentile
<1940
1940 to
1959
1960 to
1979
<1940
1940 to
1959
1960 to
1979
Base Control Option, 95th Percentile
34.50
20.45
16.52
10.42
5.91
4.94
Control Option 1, 95th Percentile
37.38
23.16
18.47
11.67
6.82
5.54
Control Option 2, 95th Percentile
33.81
19.82
16.11
10.19
5.73
4.83
Control Option 3, 95th Percentile
31.77
17.94
14.81
9.59
5.27
4.50
Base Control Option, Median
18.58
9.36
7.03
5.63
2.94
2.34
Control Option 1, Median
19.30
9.80
7.30
5.85
3.08
2.42
Control Option 2, Median
18.43
9.27
6.97
5.57
2.90
2.32
Control Option 3, Median
17.31
8.47
6.59
5.31
2.73
2.22
Base Control Option, Mean
21.43
11.17
8.56
6.38
3.41
2.76
Control Option 1, Mean
22.78
12.16
9.16
6.83
3.70
2.95
Control Option 2, Mean
21.09
10.93
8.41
6.28
3.33
2.72
Control Option 3, Mean
19.78
9.94
7.85
5.97
3.12
2.58
Base Control Option, 5th Percentile
12.41
5.86
4.73
4.01
2.00
1.65
Control Option 1, 5th Percentile
12.54
5.93
4.78
4.05
2.03
1.67
Control Option 2, 5th Percentile
12.39
5.86
4.74
4.00
2.00
1.65
Control Option 3, 5th Percentile
11.96
5.63
4.65
3.91
1.94
1.63
5.4.3 Ways to Compare Blood Pb Changes in Association with Control Options
This section presents three ways of looking at the association of the various control options and
changes in blood Pb including tabulating quantiles of the distribution of lifetime average blood
Pb level changes relative to background, graphing changes in blood Pb levels across ages for the
theoretical children, and tabulating changes in blood Pb for each numbered control option
compared to the Base Control option. Increments in blood Pb levels associated with the
renovation activities are calculated by subtracting the background values from the lifetime
average blood Pb values. In the single activity example, background values are subtracted from
101
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
those in Exhibit 5-6, and shown in Exhibit 5-8. For the multiple activity example, background
values5 are subtracted from those in Exhibit 5-7, and are shown in Exhibit 5-9.
Exhibit 5-8. Incremental Changes in Lifetime Average Blood Pb Levels due to
the Renovation Activity for Child 2, Single Activity Example in |j,g/dL
Leggett
IEUBK
Control Option and Percentile
<1940
1940 to
1959
1960 to
1979
<1940
1940 to
1959
1960 to
1979
Base Control Option, 95th Percentile
2.82
3.29
3.21
0.89
0.87
0.90
Control Option 1, 95th Percentile
2.96
3.48
3.34
0.91
0.90
0.93
Control Option 2, 95th Percentile
2.82
3.32
3.13
0.82
0.82
0.85
Control Option 3, 95th Percentile
1.82
2.20
2.24
0.53
0.56
0.62
Base Control Option, Median
1.51
0.98
0.71
0.40
0.30
0.22
Control Option 1, Median
1.65
1.07
0.75
0.42
0.32
0.23
Control Option 2, Median
1.55
1.03
0.73
0.39
0.30
0.22
Control Option 3, Median
1.05
0.73
0.54
0.27
0.21
0.17
Base Control Option, Mean
1.85
1.30
1.08
0.47
0.38
0.33
Control Option 1, Mean
2.02
1.41
1.15
0.50
0.40
0.35
Control Option 2, Mean
1.91
1.35
1.09
0.45
0.38
0.33
Control Option 3, Mean
1.26
0.92
0.79
0.31
0.26
0.24
Base Control Option, 5th Percentile
0.46
0.28
0.17
0.13
0.09
0.05
Control Option 1, 5th Percentile
0.51
0.31
0.19
0.15
0.10
0.06
Control Option 2, 5th Percentile
0.51
0.33
0.22
0.15
0.10
0.07
Control Option 3, 5th Percentile
0.38
0.25
0.17
0.11
0.08
0.05
5 Note that the background values in this multiple activities are marginally different than those presented in Exhibit
5-5. These small differences arise from small differences in the Monte Carlo sampling of the Background Loading
between the single activity and multiple activities examples.
102
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Exhibit 5-9. Incremental Changes in Blood Pb Levels due to the Renovation Activity
for Child 2, Multiple Activities Example in |j,g/dL
Leggett
IEUBK
Control Option and Percentile
<1940
1940 to
1959
1960 to
1979
<1940
1940 to
1959
1960 to
1979
Base Control Option, 95th Percentile
9.33
9.03
6.76
2.63
2.09
1.64
Control Option 1, 95th Percentile
12.22
11.73
8.70
3.87
3.00
2.24
Control Option 2, 95th Percentile
8.64
8.39
6.34
2.39
1.91
1.53
Control Option 3, 95th Percentile
6.61
6.51
5.04
1.79
1.45
1.20
Base Control Option, Median
4.75
3.08
1.68
1.09
0.75
0.46
Control Option 1, Median
5.47
3.52
1.95
1.31
0.89
0.55
Control Option 2, Median
4.61
2.99
1.62
1.03
0.72
0.44
Control Option 3, Median
3.49
2.19
1.24
0.77
0.54
0.35
Base Control Option, Mean
5.90
4.17
2.56
1.35
0.98
0.67
Control Option 1, Mean
7.25
5.16
3.16
1.80
1.27
0.86
Control Option 2, Mean
5.56
3.92
2.42
1.25
0.90
0.63
Control Option 3, Mean
4.25
2.94
1.86
0.93
0.69
0.49
Base Control Option, 5th Percentile
1.63
0.89
0.40
0.40
0.25
0.12
Control Option 1, 5th Percentile
1.76
0.96
0.45
0.43
0.28
0.13
Control Option 2, 5th Percentile
1.61
0.89
0.41
0.39
0.25
0.12
Control Option 3, 5th Percentile
1.18
0.66
0.32
0.29
0.19
0.10
The time-series of the incremental changes in blood Pb levels associated with the single and
multiple RRP activity examples for the Base Control Option for each of the six theoretical
children are shown in Exhibit 5-10 and 5-11, respectively. Age-specific blood Pb levels
associated with the background media concentrations have been subtracted from these curves.
These time-series are represented using the quarterly-averaged blood Pb values, and averages
over these curves for the entire six years are used to derive the the lifetime averages presented in
Exhibit 5-10 and 5-11. Figures like Exhibit 5-10 and 5-11 permit comparisons of the pattern
across ages with other similar figures for different control options when plotted on the same
axes.
For each hypothetical child, there is a spike during the year of renovation followed by an
elevated "tail" of continued blood Pb elevelation during the Renovation (Post Cleaning) phase.
The increments in blood Pb levels estimated with the Leggett model (solid lines) are consistently
higher than the IEUBK results (dashed lines) for all children. In addition, the Leggett results
tend to sustain higher blood Pb levels relative to the background for a longer portion of the
child's life. Such differences arise from differences in the biokinetic modules between the
Leggett and IEUBK models. This could be either because the Pb excretion rates in the IEUBK
are greater than in Leggett or because the partitioning of the Pb body burden in the IEUBK
model is weighted more towards the non-blood compartments than in the Leggett model. In both
models, children who are two years old when renovation occurs usually have the highest spike in
103
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
blood Pb for a given increase in media concentrations. This result follows from changes in age-
specific biokinetics and from the competing effects of generally decreasing dietary intakes with
age and soil/dust intakes which peak in years 2, 3, and 4.
Exhibit 5-10. Time Series of Incremental Blood Pb Changes Due to
the RRP Activity for the < 1940 Vintage, Single Activity Example
7
6
_ 5
¦o
3
T3
ro *
a) 4
u
o
o
CQ
.E 3
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Exhibit 5-11. Time Series of Incremental Blood Pb Changes Due to
the RRP Activity for the < 1940 Vintage, Multiple Activities Example
Age, Years
The difference in blood Pb between each of the Control Options and the Base Control Option for
blood Pb estimates is shown in Exhibit 5-12 for the single RRP activity example, and in Exhibit
5-13 for the multiple RRP activity example. These were calculated using median indoor dust
concentrations. Negative values in the table indicate that the Control Option represents a
reduction in blood Pb levels compared to the Base Control Option and vice versa. Tables like
Exhibits 5-12 and 5-13 translate figures like Exhibits 5-10 and 5-11 into summary measures that
can be compared for analysis of proposed rule control options.
Exhibit 5-12. Changes in Blood Pb for each Control Option Compared to the Base Control
Control Option and
Percentile
Leggett
IEUBK
<1940
1940 to
1959
1960 to
1979
<1940
1940 to
1959
1960 to
1979
Control Option 1,
Median
0.14
0.09
0.05
0.03
0.02
0.01
Control Option 2,
Median
0.04
0.05
0.02
-0.01
0.00
0.00
Control Option 3,
Median
-0.46
-0.26
-0.16
-0.13
-0.08
-0.05
105
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Exhibit 5-13. Changes in Blood Pb for each Control Option Compared to the Base Control
Option for the Median Indoor Dust, Multiple Activities Example in |j,g/dL
Control Option
and Percentile
Leggett
IEUBK
<1940
1940 to
1959
1960 to
1979
<1940
1940 to 1959
1960 to
1979
Control Option 1,
Median
0.72
0.44
0.27
0.22
0.14
0.09
Control Option 2,
Median
-0.14
-0.09
-0.06
-0.06
-0.04
-0.02
Control Option 3,
Median
-1.26
-0.89
-0.44
-0.32
-0.21
-0.12
5.5 Uncertainties and Limitations
It is likely that uncertainties in exposure concentration inputs contribute to the overall
uncertainty in any blood Pb estimates. The difficulties involved in deriving estimates of indoor
air and dust Pb concentrations for different renovation activities on the basis of limited data are
discussed in Section 3 and 4 and Appendix A. Whatever uncertainties are involved in the
estimation of exposure concentrations will be propagated through the exposure, intake, and
uptake assessment, and through the biokinetic blood Pb modeling.
Exposure factor values used to estimate inputs to the biokinetic models (respiratory volume,
water intake, GI absorption fractions, etc.) are anticipated to contribute a somewhat lesser degree
of uncertainty to blood Pb estimates. Because exposure to indoor dust represents such a large
proportion of total Pb uptake associated with indoor renovation activities, exposure factors in this
pathway, such as dust ingestion rates and GI absorption, are likely to be the important
contributors. Even with outdoor activities, a portion of outdoor soil is assumed to be available to
a child indoors.
It is well-known that the IEUBK and Leggett biokinetic models tend to predict different blood
Pb levels when given the same exposure inputs (Pounds and Leggett 1998, USEPA 2006d). The
renovation examples reported here show that predictions by the Leggett model are about three
times those predicted by the IEUBK. As discussed above, the relative performance of the two
models suggests some possible explanations: (1) more rapid excretion of Pb (to urine and feces)
from the IEUBK, or (2) differences in partitioning of Pb between blood and other body
"compartments." The IEUBK seems to maintain a higher proportion of Pb in bone and other
slowly-exchanging compartments, while Leggett keeps more Pb in the blood (erythrocyte)
compartment for longer durations.
The difference in the predictions from the two biokinetic models illustrates the approximate
contribution of "model uncertainty" to the overall uncertainty in blood Pb predictions. As
discussed in Section 5.1.2, the supporting database for the Leggett model is somewhat more
extensive than that for IEUBK, and the Leggett model is structurally more sophisticated than the
IEUBK. In addition, applying the IEUBK to estimate the impacts of short-term fluctuations in
Pb exposure (weekly in this analysis) is stretching the IEUBK to the limits of its temporal
resolution. On the other hand, as shown in Exhibit 5-6, the IEUBK blood Pb predictions
106
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
associated with what are considered to be typical Background exposures in the single activity
case can result in central tendency blood Pb estimates that are much closer to the results obtained
from large population surveys (NHANES) than the Leggett model. Regrettably, there are little
data available concerning the patterns of Pb exposures of the NHANES participants.
107
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
6. ESTIMATES OF IQ CHANGE
Once the blood Pb levels are estimated from the media concentrations, these values are
converted to IQ changes using regression equations, as shown in Exhibit 6-1. As discussed in
Chapter 2, two such available sets of regression equations include those from Lanphear et al.
(2005) and from Canfield et al. (2003). For the current approach, the Lanphear et al. (2005)
models were selected because the population in that study was much larger, including subjects
from several countries, from populations with different patterns of Pb exposure, and from a wide
range of socioeconomic strata. The larger number of subjects in the Lanphear et al. (2005) study
afforded a higher degree of precision in identifying and characterizing blood Pb-IQ relationships,
and allowed the use of more sophisticated statistical models to evaluate the data.
Exhibit 6-1. Flow Chart Illustrating the Estimation of IQ Changes from Blood Pb Levels
6.1 Selected Model (Lanphear et al. 2005)
As discussed in Chapter 2, Lanphear et al. (2005) derived regression relationships between
several blood Pb metrics and IQ test results based on linear, cubic spline, log-linear, and
piecewise linear equations. Lanphear et al (2005) found that the log-linear model provided the
strongest relationships for the data; this relationship was selected as the primary regression
equation for the current approach. In addition, a piecewise-linear equation provided by Lanphear
et al (2005) is used for purposes of comparison in the current approach to address the limitations
and uncertainties associated with the log-linear model. For the blood Pb metric, the current
approach selects the lifetime average blood Pb since it more fully takes into account the
renovation activity exposures of children in all six age groups. The alternate option, using the
concurrent average, would require averaging near the ages when IQ was measured for the test
subjects used in the Lanphear et al (2005) regression. In this case, this would involve an average
during the fifth or sixth year of life and would give relatively smaller blood Pb levels for the
108
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
children who experience the renovation earlier in their lives and have nearly reached background
by the sixth year, particularly with the IEUBK model.
The Lanphear et al. (2005) study included a relatively high proportion of children with low blood
Pb levels. For example, approximately 18 percent of the children had blood Pb levels below 10
ug/dL and about eight percent had blood Pb levels that never exceeded 7.5 ug/dL. The statistical
model with the strongest relationship between blood Pb and IQ levels in the Lanphear et al.
(2005) data set is log-linear. This model has the property that the blood Pb-IQ slope increases
rapidly at low blood Pb levels, and goes to infinity at zero blood Pb. Thus, it cannot be used to
accurately predict IQ loss at very low blood Pb levels.
To address the limitations and uncertainties associated with the log-linear model, two strategies
have been used. The first strategy uses a "cutpoint" of 1 ug/dL and does not estimate IQ change
below that. The rationale for selecting this particular cutpoint is to apply the model within the
range of the Lanphear et al. (2005) data; almost all of the observed blood Pb values in Lanphear
et al (2005) were above 1 ug/dL. This methodology is intended to give a conservative estimate of
IQ changes due to the relatively small changes in blood Pb in newer homes and in the 5th
percentile dust lead concentration cases. The equation for the log-linear model used in the
analysis is:
IQ change = 0 for PbB < 1 |ig/dL
} (Eq. 6-1)
IQ change = - 3.04 * In (blood Pb/1 |ig/dL) for PbB > 1 |ig/dL
where:
PbB = Lifetime average of the blood Pb level
The second strategy adopts a piecewise linear model developed by Lanphear et al. (2005) with
lifetime blood Pb data. This model has the advantage that while it captures the difference in
slope between the lower and higher blood Pb ranges, the slope does not increase to infinity as
blood Pb decreases. For the piecewise-linear model, the slope changes at 10 ug/dL, so the
equation (Eq 6-2) is:
IQ change = 0 for PbB < 1 ug/dL
IQ change = -0.80 * (PbB - 1 ug/dL) for 1 < PbB < 10 ug/dL } (Eq. 6-2)
IQ change = -0.13 * (PbB - 10 ug/dL) -7.2 for PbB >10 ug/dL
where:
PbB = Lifetime average of the blood Pb level
Since this model has a shallower slope, estimates based on like inputs will be uniformly smaller
than those from the log-linear fit. The coefficients for this model, however, were developed by
109
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Lanphear et al (2005) for concurrent levels and may differ from ones that might pertain to a
preconceived linear fit to lifetime Lanphear et al (2005) data.
6.2 Examples
The lifetime average blood Pb for both the Leggett and IEUBK models were used to calculate
estimated IQ changes for all vintages, all Control Options, and all indoor dust Monte Carlo
percentiles (5th, median, mean, and 95th) for the single and multiple RRP activity examples using
both the log-linear and piecewise linear models. These calculations are presented below for
Child 2, and calculations for all children are presented in Appendix G.
6.2.1 Estimated IQ Changes Associated With Background Exposures
Exhibit 6-2 and 6-3 show the IQ changes associated with the background Pb exposure
concentrations and their blood Pb estimates from the IEUBK and Leggett models, respectively.
In this table, the sign convention is such that negative values represent reductions in IQ. In
general, the Leggett model results follow similar trends to the IEUBK results, although with
different magnitudes. As discussed above, the piecewise linear model results in smaller IQ
changes than the log-linear model in every vintage.
Exhibit 6-2. IQ Changes Due to Background Pb Exposure Concentrations
Log-Linear
Piecewise Linear
Percentile
<1940
1940 to
1959
1960 to
1979
<1940
1940 to
1959
1960 to
1979
95th Percentile
-6.3
-4.1
-3.6
-5.5
-2.3
-1.8
Median
-4.6
-2.4
-1.9
-2.8
-1.0
-0.7
Mean
-4.9
-2.7
-2.2
-3.2
-1.1
-0.9
5th Percentile
-3.9
-1.7
-1.3
-2.1
-0.6
-0.4
Exhibit 6-3. IQ Changes Due to Background Pb Exposure Concentrations
Log-Linear
Piecewise Linear
Percentile
<1940
1940 to
1959
1960 to
1979
<1940
1940 to
1959
1960 to
1979
95th Percentile
-9.8
-7.4
-6.9
-9.2
-7.4
-7.0
Median
-8.0
-5.6
-5.1
-7.7
-4.2
-3.5
Mean
-8.3
-5.9
-5.4
-7.9
-4.8
-4.0
5th Percentile
-7.2
-4.9
-4.5
-7.3
-3.2
-2.7
110
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
6.2.2 Estimated IQ Changes for the Single and Multiple RRP Activity Examples
Exhibits 6-4 and 6-5 show the IQ changes for Child 2 using blood Pb levels from the IEUBK and
Leggett models, respectively, in the single activity example. The estimates derived using median
dust concentrations are highlighted in bold. The reductions in IQ tend to be largest for the older
homes, as expected. As in the background case, the predictions using the piecewise linear model
tend to be lower than those using the log-linear model, although exceptions do occur for the
higher indoor dust percentiles.
Exhibit 6-4. IQ Changes for Child 2 Using Blood Pb Values
from the IEUBK Model for the Single Activity Example
Log-Linear
Piecewise Linear
Control Option and Percentile
<1940
1940 to
1959
1960 to
1979
<1940
1940 to
1959
1960 to
1979
Base Control Option, 95th Percentile
-6.6
-4.7
-4.4
-6.2
-3.0
-2.6
Control Option 1, 95th Percentile
-6.6
-4.7
-4.4
-6.2
-3.0
-2.6
Control Option 2, 95th Percentile
-6.6
-4.7
-4.3
-6.1
-2.9
-2.5
Control Option 3, 95th Percentile
-6.5
-4.5
-4.1
-5.9
-2.7
-2.3
Base Control Option, Median
-4.9
-2.8
-2.2
-3.2
-1.2
-0.9
Control Option 1, Median
-4.9
-2.8
-2.3
-3.2
-1.2
-0.9
Control Option 2, Median
-4.9
-2.8
-2.2
-3.1
-1.2
-0.9
Control Option 3, Median
-4.8
-2.7
-2.2
-3.1
-1.1
-0.8
Base Control Option, Mean
-5.2
-3.1
-2.7
-3.6
-1.4
-1.1
Control Option 1, Mean
-5.2
-3.2
-2.7
-3.6
-1.5
-1.1
Control Option 2, Mean
-5.2
-3.1
-2.7
-3.6
-1.4
-1.1
Control Option 3, Mean
-5.1
-3.0
-2.6
-3.5
-1.4
-1.1
Base Control Option, 5th Percentile
-4.0
-1.9
-1.4
-2.2
-0.7
-0.5
Control Option 1, 5th Percentile
-4.0
-1.9
-1.4
-2.2
-0.7
-0.5
Control Option 2, 5th Percentile
-4.0
-1.9
-1.4
-2.2
-0.7
-0.5
Control Option 3, 5th Percentile
-4.0
-1.8
-1.4
-2.2
-0.7
-0.5
111
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Exhibit 6-5. IQ Changes for Child 2 Using Blood Pb Values
from the Leggett Model for the Single Activity Example
Log-Linear
Piecewise Linear
Control Option and Percentile
<1940
1940 to
1959
1960 to
1979
<1940
1940 to
1959
1960 to
1979
Base Control Option, 95th Percentile
-10.15
-8.19
-7.78
-9.56
-7.82
-7.58
Control Option 1, 95th Percentile
-10.16
-8.23
-7.82
-9.58
-7.85
-7.60
Control Option 2, 95th Percentile
-10.15
-8.19
-7.77
-9.56
-7.82
-7.57
Control Option 3, 95th Percentile
-10.04
-7.95
-7.55
-9.43
-7.68
-7.46
Base Control Option, Median
-8.31
-6.03
-5.47
-7.90
-5.02
-4.04
Control Option 1, Median
-8.33
-6.07
-5.50
-7.92
-5.09
-4.08
Control Option 2, Median
-8.31
-6.05
-5.48
-7.90
-5.05
-4.06
Control Option 3, Median
-8.22
-5.92
-5.39
-7.84
-4.81
-3.91
Base Control Option, Mean
-8.68
-6.43
-5.94
-8.16
-5.84
-4.85
Control Option 1, Mean
-8.71
-6.47
-5.97
-8.18
-5.93
-4.90
Control Option 2, Mean
-8.69
-6.45
-5.95
-8.17
-5.89
-4.86
Control Option 3, Mean
-8.57
-6.29
-5.81
-8.08
-5.54
-4.62
Base Control Option, 5th Percentile
-7.35
-5.04
-4.57
-7.36
-3.40
-2.80
Control Option 1, 5th Percentile
-7.37
-5.06
-4.58
-7.37
-3.42
-2.81
Control Option 2, 5th Percentile
-7.37
-5.07
-4.60
-7.37
-3.43
-2.83
Control Option 3, 5th Percentile
-7.33
-5.02
-4.57
-7.35
-3.37
-2.79
Exhibits 6-6 and 6-7 show the IQ changes for Child 2 using blood Pb levels from the IEUBK and
Leggett models, respectively, in the multiple activity example. Again, the predictions using the
piecewise linear model tend to be lower than those using the log-linear model, although
exceptions do occur for the higher indoor dust percentiles.
112
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Exhibit 6-6. IQ Changes for Child 2 Using Blood Pb Values
from the IEUBK Model for the Multiple Activities Example
Log-Linear
Piecewise-Linear
Control Option and Percentile
<1940
1940 to
1959
1960 to
1979
<1940
1940 to
1959
1960 to
1979
Base Control Option, 95th Percentile
-7.1
-5.4
-4.9
-7.3
-3.9
-3.2
Control Option 1, 95th Percentile
-7.5
-5.8
-5.2
-7.4
-4.7
-3.6
Control Option 2, 95th Percentile
-7.1
-5.3
-4.8
-7.2
-3.8
-3.1
Control Option 3, 95th Percentile
-6.9
-5.1
-4.6
-6.9
-3.4
-2.8
Base Control Option, Median
-5.3
-3.3
-2.6
-3.7
-1.6
-1.1
Control Option 1, Median
-5.4
-3.4
-2.7
-3.9
-1.7
-1.1
Control Option 2, Median
-5.2
-3.2
-2.6
-3.7
-1.5
-1.1
Control Option 3, Median
-5.1
-3.1
-2.4
-3.4
-1.4
-1.0
Base Control Option, Mean
-5.6
-3.7
-3.1
-4.3
-1.9
-1.4
Control Option 1, Mean
-5.8
-4.0
-3.3
-4.7
-2.2
-1.6
Control Option 2, Mean
-5.6
-3.7
-3.0
-4.2
-1.9
-1.4
Control Option 3, Mean
-5.4
-3.5
-2.9
-4.0
-1.7
-1.3
Base Control Option, 5th Percentile
-4.2
-2.1
-1.5
-2.4
-0.8
-0.5
Control Option 1, 5th Percentile
-4.3
-2.1
-1.6
-2.4
-0.8
-0.5
Control Option 2, 5th Percentile
-4.2
-2.1
-1.5
-2.4
-0.8
-0.5
Control Option 3, 5th Percentile
-4.1
-2.0
-1.5
-2.3
-0.8
-0.5
113
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Exhibit 6-7. IQ Changes for Child 2 Using Blood Pb Values
from the Leggett Model for the Multiple Activities Example
Control Option and Percentile
Log Linear
Piecewise Linear
<1940
1940 to
1959
1960 to
1979
<1940
1940 to
1959
1960 to
1979
Base Control Option, 95th Percentile
-10.76
-9.18
-8.53
-10.38
-8.56
-8.05
Control Option 1, 95th Percentile
-11.01
-9.55
-8.86
-10.76
-8.91
-8.30
Control Option 2, 95th Percentile
-10.70
-9.08
-8.45
-10.29
-8.48
-7.99
Control Option 3, 95th Percentile
-10.51
-8.78
-8.19
-10.03
-8.23
-7.82
Base Control Option, Median
-8.88
-6.80
-5.93
-8.32
-6.69
-4.82
Control Option 1, Median
-9.00
-6.94
-6.04
-8.41
-7.04
-5.04
Control Option 2, Median
-8.86
-6.77
-5.90
-8.30
-6.61
-4.78
Control Option 3, Median
-8.67
-6.49
-5.73
-8.15
-5.97
-4.47
Base Control Option, Mean
-9.32
-7.34
-6.53
-8.69
-7.35
-6.05
Control Option 1, Mean
-9.50
-7.59
-6.73
-8.86
-7.48
-6.53
Control Option 2, Mean
-9.27
-7.27
-6.47
-8.64
-7.32
-5.93
Control Option 3, Mean
-9.07
-6.98
-6.26
-8.47
-7.15
-5.48
Base Control Option, 5th Percentile
-7.66
-5.37
-4.72
-7.51
-3.89
-2.98
Control Option 1, 5th Percentile
-7.69
-5.41
-4.75
-7.53
-3.94
-3.02
Control Option 2, 5th Percentile
-7.65
-5.37
-4.73
-7.51
-3.89
-2.99
Control Option 3, 5th Percentile
-7.54
-5.25
-4.67
-7.46
-3.70
-2.92
6.2.3 Ways to Compare IQ Changes in Association with Control Options
This section presents two ways of looking at the association of the various control options and
changes in IQ including tabulating quantiles of the distribution of IQ changes relative to
background, and tabulating changes in IQ for each numbered control option compared to the
Base Control option. Increments in IQ levels associated with the renovation activities are
calculated by subtracting the IQ levels associated with the background values from the IQ levels
associated with the lifetime average blood Pb values of the activity-specific example.
Exhibits 6-8 and 6-9 show the incremental IQ reductions relative to background exposures
associated with the single activity example using the IEUBK and Leggett models, respectively.
Background values using the IEUBK model shown in Exhibit 6-2 are subtracted from those in
Exhibit 6-4, and shown in Exhibit 6-8. Background values using the Leggett model shown in
Exhibit 6-3 are subtracted from those in Exhibit 6-5, and shown in Exhibit 6-9. Exhibits 6-10
and 6-11 show the incremental IQ reductions relative to background exposures associated with
the multiple activity example using the IEUBK and Leggett models, respectively. Background
values using the IEUBK model shown in Exhibit 6-2 are subtracted from those in Exhibit 6-6,
and shown in Exhibit 6-10. Background values using the Leggett model shown in Exhibit 6-3
114
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
are subtracted from those in Exhibit 6-7, and shown in Exhibit 6-11. For these examples, unlike
blood Pb levels, these incremental changes do not decrease with newer houses. Although the
blood Pb levels follow this trend, the IQ regression equations give steeper slopes for lower blood
Pb levels. Thus, the same change relative to background in blood Pb for the pre-1940 vintage
(higher blood Pb levels) will give a lower associated IQ loss than the same change in blood Pb
for 1960-1979 vintage homes (lower blood Pb levels). These two competing effects (incremental
blood Pb increasing with increasing vintage offset by IQ slope decreases with increasing vintage)
combine such that the middle vintage sometimes has the highest predicted incremental IQ
change.
Exhibit 6-8. Incremental Changes in IQ for Child 2 Using Blood Pb Levels from the IEUBK
Log-Linear
Piecewise Linear
Control Option and Percentile
<1940
1940 to
1959
1960 to
1979
<1940
1940 to
1959
1960 to
1979
Base Control Option, 95th Percentile
-0.3
-0.6
-0.7
-0.7
-0.7
-0.7
Control Option 1, 95th Percentile
-0.3
-0.6
-0.8
-0.7
-0.7
-0.7
Control Option 2, 95th Percentile
-0.3
-0.6
-0.7
-0.7
-0.7
-0.7
Control Option 3, 95th Percentile
-0.2
-0.4
-0.5
-0.4
-0.4
-0.5
Base Control Option, Median
-0.3
-0.4
-0.3
-0.3
-0.2
-0.2
Control Option 1, Median
-0.3
-0.4
-0.4
-0.3
-0.3
-0.2
Control Option 2, Median
-0.3
-0.4
-0.3
-0.3
-0.2
-0.2
Control Option 3, Median
-0.2
-0.3
-0.3
-0.2
-0.2
-0.1
Base Control Option, Mean
-0.3
-0.4
-0.4
-0.4
-0.3
-0.3
Control Option 1, Mean
-0.3
-0.5
-0.5
-0.4
-0.3
-0.3
Control Option 2, Mean
-0.3
-0.4
-0.4
-0.4
-0.3
-0.3
Control Option 3, Mean
-0.2
-0.3
-0.3
-0.2
-0.2
-0.2
Base Control Option, 5th Percentile
-0.1
-0.1
-0.1
-0.1
-0.1
0.0
Control Option 1, 5th Percentile
-0.1
-0.2
-0.1
-0.1
-0.1
0.0
Control Option 2, 5th Percentile
-0.1
-0.2
-0.1
-0.1
-0.1
-0.1
Control Option 3, 5th Percentile
-0.1
-0.1
-0.1
-0.1
-0.1
0.0
115
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Exhibit 6-9. Incremental Changes in IQ for Child 2 Using Blood Pb Levels from the Leggett
Control Option and Percentile
Log Linear
Piecewise Linear
<1940
1940 to
1959
1960 to
1979
<1940
1940 to
1959
1960 to
1979
Base Control Option, 95th Percentile
-0.32
-0.77
-0.87
-0.37
-0.43
-0.60
Control Option 1, 95th Percentile
-0.34
-0.81
-0.90
-0.39
-0.45
-0.61
Control Option 2, 95th Percentile
-0.32
-0.77
-0.85
-0.37
-0.43
-0.59
Control Option 3, 95th Percentile
-0.21
-0.53
-0.63
-0.24
-0.29
-0.47
Base Control Option, Median
-0.31
-0.44
-0.38
-0.34
-0.52
-0.50
Control Option 1, Median
-0.34
-0.48
-0.40
-0.21
-0.86
-0.60
Control Option 2, Median
-0.32
-0.46
-0.39
-0.20
-0.82
-0.58
Control Option 3, Median
-0.22
-0.33
-0.29
-0.14
-0.58
-0.43
Base Control Option, Mean
-0.34
-0.52
-0.50
-0.24
-1.04
-0.86
Control Option 1, Mean
-0.37
-0.56
-0.53
-0.26
-1.12
-0.92
Control Option 2, Mean
-0.35
-0.54
-0.51
-0.25
-1.08
-0.87
Control Option 3, Mean
-0.24
-0.38
-0.38
-0.16
-0.74
-0.63
Base Control Option, 5th Percentile
-0.13
-0.17
-0.12
-0.06
-0.23
-0.14
Control Option 1, 5th Percentile
-0.14
-0.18
-0.13
-0.07
-0.25
-0.15
Control Option 2, 5th Percentile
-0.14
-0.19
-0.15
-0.07
-0.26
-0.17
Control Option 3, 5th Percentile
-0.11
-0.15
-0.11
-0.05
-0.20
-0.13
116
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Exhibit 6-10. Incremental Changes in IQ for Child 2 Using Blood Pb Levels from the IEUBK
Model for the Multiple Activity Example
Control Option and Percentile
Log-Linear
Piecewise-Linear
<1940
1940 to
1959
1960 to
1979
<1940
1940 to
1959
1960 to
1979
Base Control Option, 95th Percentile
-0.9
-1.3
-1.2
-1.8
-1.7
-1.3
Control Option 1, 95th Percentile
-1.2
-1.8
-1.6
-2.0
-2.4
-1.8
Control Option 2, 95th Percentile
-0.8
-1.2
-1.2
-1.8
-1.5
-1.2
Control Option 3, 95th Percentile
-0.6
-1.0
-0.9
-1.4
-1.2
-1.0
Base Control Option, Median
-0.7
-0.9
-0.7
-0.9
-0.6
-0.4
Control Option 1, Median
-0.8
-1.0
-0.8
-1.1
-0.7
-0.4
Control Option 2, Median
-0.6
-0.9
-0.6
-0.8
-0.6
-0.4
Control Option 3, Median
-0.5
-0.7
-0.5
-0.6
-0.4
-0.3
Base Control Option, Mean
-0.7
-1.0
-0.8
-1.1
-0.8
-0.5
Control Option 1, Mean
-0.9
-1.3
-1.0
-1.4
-1.0
-0.7
Control Option 2, Mean
-0.7
-1.0
-0.8
-1.0
-0.7
-0.5
Control Option 3, Mean
-0.5
-0.8
-0.6
-0.7
-0.5
-0.4
Base Control Option, 5th Percentile
-0.3
-0.4
-0.2
-0.3
-0.2
-0.1
Control Option 1, 5th Percentile
-0.3
-0.4
-0.3
-0.3
-0.2
-0.1
Control Option 2, 5th Percentile
-0.3
-0.4
-0.2
-0.3
-0.2
-0.1
Control Option 3, 5th Percentile
-0.2
-0.3
-0.2
-0.2
-0.2
-0.1
117
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Exhibit 6-11. Incremental Changes in IQ for Child 2 Using Blood Pb Levels from the Leggett
Model for the Multiple Activities Example
Log Linear
Piecewise Linear
Control Option and Percentile
<1940
1940 to
1959
1960 to
1979
<1940
1940 to
1959
1960 to
1979
Base Control Option, 95th Percentile
-0.96
-1.77
-1.60
-1.21
-1.17
-1.04
Control Option 1, 95th Percentile
-1.20
-2.15
-1.94
-1.59
-1.53
-1.29
Control Option 2, 95th Percentile
-0.90
-1.67
-1.52
-1.12
-1.09
-0.98
Control Option 3, 95th Percentile
-0.71
-1.37
-1.26
-0.86
-0.85
-0.81
Base Control Option, Median
-0.90
-1.21
-0.83
-0.62
-2.46
-1.34
Control Option 1, Median
-1.01
-1.35
-0.94
-0.71
-2.81
-1.56
Control Option 2, Median
-0.87
-1.18
-0.81
-0.60
-2.39
-1.30
Control Option 3, Median
-0.68
-0.91
-0.63
-0.45
-1.75
-0.99
Base Control Option, Mean
-0.98
-1.42
-1.08
-0.77
-2.55
-2.05
Control Option 1, Mean
-1.16
-1.68
-1.29
-0.94
-2.68
-2.53
Control Option 2, Mean
-0.93
-1.35
-1.03
-0.72
-2.52
-1.94
Control Option 3, Mean
-0.74
-1.06
-0.82
-0.45
-1.75
-0.99
Base Control Option, 5th Percentile
-0.43
-0.50
-0.27
-0.21
-0.71
-0.32
Control Option 1, 5th Percentile
-0.46
-0.54
-0.30
-0.23
-0.77
-0.36
Control Option 2, 5th Percentile
-0.42
-0.50
-0.28
-0.21
-0.71
-0.33
Control Option 3, 5th Percentile
-0.32
-0.38
-0.22
-0.15
-0.53
-0.26
Exhibits 6-12 and 6-13 show the IQ change associated with each Control Option compared with
the Base Control Option using blood Pb levels from the IEUBK and Leggett models,
respectively, for the single activity example. Exhibits 6-14 and 6-15 show the IQ change
associated with each Control Option compared with the Base Control Option using blood Pb
levels from the IEUBK and Leggett models, respectively, for the multiple activity example. In
these tables, a positive number indicates that the Control Option gives a smaller IQ change, and
thus provides a benefit compared to the Base Control Option. These tables provide summary
measures that can be compared for analysis of proposed rule control options.
118
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Exhibit 6-12. Comparison of IQ Changes for Child 2 Across Control Options for the Median
Case Using Blood Pb T.cvcls from the TEITVK Model for the Single Activity Example
Log-Linear
Piecewise Linear
Control Option
<1940
1940 to
1959
1960 to
1979
<1940
1940 to
1959
1960 to
1979
Control Option 1, Median
-0.02
-0.03
-0.02
-0.02
-0.02
-0.01
Control Option 2, Median
0.01
0.00
-0.01
0.01
0.00
0.00
Control Option 3, Median
0.08
0.10
0.08
0.10
0.07
0.04
Exhibit 6-13. Comparison of IQ Changes for Child 2 Across Control Options for the Median
Case Using Blood Pb Levels from the Leggett Model for the Single Activity Example
Log-Linear
Piecewise Linear
Control Option
<1940
1940 to
1959
1960 to
1979
<1940
1940 to
1959
1960 to
1979
Control Option 1, Median
-0.03
-0.04
-0.02
0.01
0.01
0.01
Control Option 2, Median
-0.01
-0.02
-0.01
0.00
-0.04
-0.02
Control Option 3, Median
0.09
0.11
0.08
0.06
0.21
0.13
Exhibit 6-14. Comparison of IQ Changes for Child 2 Across Control Options for the Median
Case Using Blood Pb T.cvcls from the TEITSK Model for the Multiple Activities Example
Log-Linear
Piecewise-Linear
Control Option
<1940
1940 to
1959
1960 to
1979
<1940
1940 to
1959
1960 to
1979
Control Option 1, Median
-0.12
-0.14
-0.11
-0.18
-0.11
-0.07
Control Option 2, Median
0.03
0.04
0.02
0.05
0.03
0.02
Control Option 3, Median
0.18
0.23
0.16
0.26
0.17
0.09
Exhibit 6-15. Comparison of IQ Changes for Child 2 Across Control Options for the Median
Case Using Blood Pb Levels from the Leggett Model for the Multiple Activities Example
Log-Linear
Piecewise-Linear
Control Option
<1940
1940 to
1959
1960 to
1979
<1940
1940 to
1959
1960 to
1979
Control Option 1, Median
-0.12
-0.14
-0.12
0.01
0.01
0.01
Control Option 2, Median
0.02
0.03
0.02
0.02
0.07
0.04
Control Option 3, Median
0.21
0.30
0.20
0.16
0.71
0.35
119
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
6.3 Uncertainties and Limitations
The estimates in IQ change are based on the blood Pb estimates described in Section 5.
Substantial uncertainty is associated with the blood Pb estimates due to uncertainties fin the
exposure data and models. Thus, overall uncertainty is carried throughout the approach and
ultimately to the estimates of IQ change.
Lanphear et al. (2005) compared model shapes and Pb metrics. The log-linear model provides
the strongest relationship as shown by the magnitude of its slope coefficient. Yet some estimates
of that coefficient have 95 percent confidence limits that differ by a factor of about 2.5. In the
same paper, they show prediction limits on regression estimates that span a range of about 10 IQ
points across the entire range of observed blood Pb levels. The estimated blood Pb-IQ slope
parameters also differ depending on the blood Pb metric (peak, concurrent, or lifetime) that is
used to fit the models. In this approach, estimated changes in lifetime blood Pb levels are used to
estimate IQ changes. The use of the lifetime blood Pb-IQ model for this approach is intended to
reduce the effects of age-specific differences occurring in exposure conditions when there can be
time lapses between Pb exposure and Pb measurement.
As discussed in Section 6.1, strategies can be taken to address the limitations and uncertainties
associated with the log-linear IQ model. These include the use of a "cutpoint" and the use of the
piecewise linear model. A "cutpoint" of one ug/dL maintains predictions within the range of the
data used to derive the Lanphear et al. (2005) model. The piecewise linear comparison captures
the difference in slope between the lower and higher blood Pb ranges.
120
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
7. Discussion of Uncertainties and Their Impact on Results
This chapter summarizes the overall uncertainties and limitations of the approach for generating
estimates of children's IQ changes associated with exposures to Pb dust from RRP-related
activities inside residences. In addition, this chapter identifies those aspects of the approach that
will be modified for application to child-occupied facilities. This chapter does not repeat the
detailed descriptions of the uncertainties and limitations associated with the different
components of the analysis provided in Sections 3.5 (for background and activity-related inputs),
4.7 (for characterization of exposure media concentrations), 5.5 (for blood Pb modeling), and 6.3
(for children's IQ change characterization). Instead, it places the uncertainties and limitations of
each component in the more general context of its contributions to overall result uncertainty.
There are two primary categories of uncertainties and limitations associated with this approach:
uncertainty and limitations associated with the selected input data and assumptions, which are
discussed in Section 7.1; and model uncertainty and general limitations of the modeling
approaches, which are discussed in Section 7.2. Section 7.3 provides conclusions about the
overall quality of the approach.
7.1 Input Data and Assumptions
Any model-based approach is limited by the quality of its input data. The approach described
here is fairly data intensive and thus understanding the limitations and uncertainties associated
with the input data and assumptions is critical. In addition, any uncertainties and limitations of
the data will be propagated throughout the remainder of the downstream calculations and thus
can have implications for several different parts of the approach.
For this approach, the input data most likely to significantly impact the results primarily are
related to the characterization of exposure media concentrations. The most significant limitation
is the source of the activity-related Pb loading data, the OPPT Dust Study (Battelle 2007). As
discussed in Chapter 3, several sources of activity data were evaluated and this source was
selected based primarily on the fact that its design most closely corresponded to the needs of this
approach. However, it is limited by a fairly small sample size, particularly in light of the number
of different scenarios evaluated using this approach. One of the most significant limitations of
the small sample size is that the four Control Options evaluated were never all evaluated at the
same building for the same activity. This is significant because it limits the ability to compare
results across control strategies because many of the factors that impact the generated Pb
loadings (e.g., Pb content of paint) can vary significantly across the different buildings. This
limitation is illustrated by the fact that the estimated Pb levels, particularly for indoor air, are
often not consistent with expectations (e.g., more stringent controls often result in higher indoor
air concentrations). The sample size, in combination with the method used to estimate input
distributions, also limits the ability to extrapolate from these results to other locations. The
distributions of Pb loadings used in the Monte Carlo analysis were developed strictly from the
OPPT Dust Study samples and thus likely do not capture the full range of potential loadings.
The cleaning efficiency data used in the approach are also highly uncertain and have the
potential to affect Post-renovation period indoor dust concentrations. The available data on
routine cleaning efficiency are very limited, and the data identified for this approach were not
121
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
developed with this use in mind. In addition, these data do not cover the full range of Pb
loadings evaluated using this approach, particularly on the lower end, and thus involve a number
of assumptions and some extrapolation when used in this way. As demonstrated in the
sensitivity analysis, these data have the potential to have substantial impacts on Post-renovation
period indoor dust concentrations. However, the impacts on the estimated blood Pb levels and
IQ changes may be limited because these data appear to be more heavily influenced by
renovation period concentrations, which are unaffected by routine cleaning efficiency.
This approach uses a single concentration for each building and yard for each time, rather than
characterizing the concentration gradients within the building and yard. This assumption
effectively assumes that children spend their time equally throughout the building and yard and
does not account for the fact that children may spend more or less time in locations with higher
concentrations. Although this assumption may be fairly reasonable for capturing average
exposures for most residences, it may be problematic for COFs because children in schools and
daycare centers are likely to spend the large majority of their time in a fairly limited percentage
of the building. In addition, it does not do a very good job at capturing the high- and low-end
tails of the exposure concentration distributions. For COFs loadings will be estimated for three
different room types (workspace, adjacent, rest of COF), rather than obtaining a single COF
average, and a single workspace room will be assumed.
As illustrated in the results in Chapters 5 and 6, background indoor dust concentrations are the
most significant contributors to overall blood Pb levels and IQ changes for the examples
evaluated in this document. These data are based on fairly recent HUD data and are believed to
be representative of typical US housing with LBP. Other studies generally had somewhat higher
overall background levels and thus the selected concentrations may somewhat underestimate
background contributions.
7.2 Modeling Approaches
There are four primary model-related uncertainties and limitations associated with this approach:
the indoor dust loading-to-concentration conversion, the selection of blood Pb models, the
selection of concentration-response functions for IQ change, and the selection of a cutpoint for
the concentration-response functions. The dust loading-to-concentration conversion, which is
described in detail in Appendix C, is based on a fairly limited data set that is not specific to
renovation activities. In addition, the estimated dust loadings for the multiple activities example
are higher than the range of loadings from which the load-to-concentration relationship was
developed and there is significant uncertainty in extending this relationship beyond the data or to
another data set. This conversion has an impact on both estimated activity-related and
background indoor dust concentrations.
The selection of blood Pb models is a critical element of this approach because it provides the
link between the exposure media concentrations and the measures of IQ change. Two biokinetic
models were considered for this approach: the IEUBK and Leggett models. The Leggett model
appears to be a more appropriate model for RRP activities because it can accommodate shorter
time steps and can account for the impact of previous exposures. However, the results from the
Leggett model appear to be higher than would be expected (i.e., it generates results that in some
cases do not seem to be supported by observed levels), even for the scenario with only
122
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
background contributions. The IEUBK model appears to generate results more consistent with
expectations, but has limitations of its own. For example, it is a steady-state model designed to
characterize blood Pb levels for long-term exposures, not peak exposures such as those
associated with RRP activities. In addition, both models require indoor dust concentrations,
instead of loadings, which introduces the need for the loading-to-concentration conversion
discussed above. The blood Pb levels estimated from the two biokinetic models differ by a
factor of two to three; this may provide some insight into the range of model uncertainty in the
blood Pb models, but does not capture the full range of uncertainty associated with the blood Pb
estimates (which include the uncertainties and limitations of all upstream steps).
Log-linear and piecewise linear models have been considered for proceeding from blood Pb to
IQ change, using Lanphear et al. (2005). In implementing either model, a cutpoint of 1 ug/dL
was used in estimating IQ change. The log-linear model is interpreted to give the change in IQ
relative to the IQ at the cutpoint. A cutpoint of 1 ug/dL was chosen to ensure that no blood Pb-
IQ relationship was used beyond the range of blood Pb levels used to derive it. Data in the lower
blood Pb ranges are fairly limited and the regression model near the limit of the data is
necessarily uncertain. Implementing a higher cutoff than the 1 ug/dL would decrease the change
in IQ associated with all blood Pb estimates, and thus the IQ changes are very sensitive to the
cutpoint used.
7.3 Conclusions
Overall, this approach appears to give useful estimates. Because of the complexity of the
methods and models used to estimate blood Pb and IQ loss, the overall degree of uncertainty in
estimates of changes in IQ cannot be estimated. However, it is important to note that any
systematic modeling errors apply more or less equally to estimates of change in IQ for the Base
Control case and for the alternate Control Options.
It is important to note that this approach has limitations in terms of its applicability to some types
of COFs, particularly those that are significantly different from residences (e.g., schools, daycare
centers). This approach uses whole building indoor dust and air concentrations and whole yard
outdoor soil concentrations to characterize exposures. This essentially equates to assuming
children are just as likely to spend time in a given room in the house or location in the yard as
they are in any other room or location in the yard. This assumption is likely to be more
reasonable for residences than for schools and daycare centers, where children under six years of
age are likely to be restricted to a limited number of locations in the building. However, it will
be relatively straightforward to modify this approach to better accommodate COFs by adjusting
the media concentrations to account for where children spend their time when in a COF.
123
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
8 REFERENCES
Abadin, H.G. and Wheeler, J. S. 1997. Guidance for risk assessment of exposure to lead: a site-
specific, multimedia approach. In: Andrews, J.S.; Frumkin, H.; Johnson, B.L.; Mehlman, M.A.;
Xintaras, C.; Bucsela, J. A., eds. Hazardous Waste and Public Health: International Congress On
the Health Effects of Hazardous Waste. Princeton, NJ: Princeton Scientific Publishing Co. 477-
485.
Alexander, F.W.; Clayton, B.E.; and Delves, H.T. 1974. Mineral and trace-metal balances in
children receiving normal and synthetic diets. Q. J. Med. 43: 89-111.
Battelle. 2007. Draft Final Report on Characterization of Dust Lead Levels after Renovation,
Repair, and Painting Activities. Prepared by Battelle (Columbus, OH) for the Office of Pollution
Prevention and Toxics (Washington, D.C.). January 23.
Bellinger, D.C.; Stiles, K.M.; and Needleman, H.L. 1992. Low-level lead exposure, intelligence
and academic achievement: a long-term follow-up study. Pediatrics. 90: 855-861.
Bellinger, D. C. and Needleman, H.L. 2003. Intellectual impairment and blood lead levels
[letter]. New England Journal of Medicine. 349: 500.
Canfield, R.L.; Henderson, C.R., Jr.; Cory-Slechta, D.A.; Cox, C.; Jusko, T.A.; and Lanphear, B.
P. 2003. Intellectual impairment in children with blood lead concentrations below 10 jag per
deciliter. New England Journal of Medicine. 348:1517-1526.
Centers for Disease Control and Prevention (CDC). 1997. Third National Health and Nutrition
Examination Survey (NHANES III), 1988-94, Dietary Recall Data. Available at:
http://www.cdc.gov/nchs/products/elec_prods/subject/nhanes3.htm.
Centers for Disease Control and Prevention (CDC). 2005. National Health and Nutrition
Examination Surveys 2003-2004. Available at:
http://www.cdc.gov/nchs/about/major/nhanes/nhanes2003-2004/nhanes03_04.htm
Chiodo, L.M.; Jacobson, S.W.; and Jacobson, J.L. 2004. Neurodevelopmental effects of
postnatal lead exposure at very low levels. Neurotoxicology and Teratology. 26: 359-371.
Choe, K.T.; Trunova, M.; Grinshpuna, S.A.; Willekea, K.; Harneya, J.; Trakumasa, S.;
Mainelisa, G.; Bornscheina, R.; Clarka, S.; and Friedmanb, W. 2000. Particle Settling After
Lead-Based Paint Abatement Work and Clearance Waiting Period. American Industrial Hygiene
Association. 61: 798-807.
Clayton, C.A.; Pellizzari, E.D.; Whitmore, R.W.; Perritt, R.L.; and Quackenboss, J. J. 1999.
National human exposure assessment survey (NHEXAS): distributions and associations of lead,
arsenic, and volatile organic compounds in EPA Region 5. J. Exposure Anal. Environ.
Epidemiol. 9: 381-392.
Cooney, G.H.; Bell, A.; McBride, W.; and Carter, C. 1989. Low-level exposures to lead: the
Sydney lead study. Developmental Medicine and Child Neurology. 31: 640-649.
124
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Cooney, G.; Bell, A.; and Stavrou, C. 1991. Low level exposures to lead and neurobehavioural
development: the Sydney study at seven years. In: Farmer, J.G., ed. International Conference:
Heavy Metals in the Environment, v. 1. September. Edinburgh, United Kingdom: CEP
Consultants, Ltd. 16-19.
Dietrich, K.N.; Berger, O.G.; Succop, P.A.; Hammond, P.B.; and Bornschein, R.L. 1993a. The
developmental consequences of low to moderate prenatal and postnatal lead exposure:
intellectual attainment in the Cincinnati Lead Study Cohort following school entry.
Neurotoxicology and Teratology. 15: 37-44.
Dietrich, K.N.; Berger, O.G.; and Succop, P.A. 1993b. Lead exposure and the motor
developmental status of urban six-year-old children in the Cincinnati prospective study.
Pediatrics. 91: 301-307.
Ernhart, C.B. 2006. Effects of lead on IQ in children [letter]. Environmental Health
Perspectives. 114: A85-A86.
Fulton, M.; Raab, G.; Thomson, G.; Laxen, D.; Hunter, R.; and Hepburn, W. 1987. Influence of
blood lead on the ability and attainment of children in Edinburgh. Lancet. (8544): 1221-1226.
International Commission on Radiological Protection (ICRP). 2002. ICRP Publication 89:
Basic anatomical and physiological data for use in radiological protection. Annals of the ICRP
32 (3-4): 100.
Kordas, K.; Canfield, R.L.; Lopez, P.; Rosado, J.L.; Vargas, G.G.; Cebrian, M.E.; Rico, J.A.;
Ronquillo, D.; and Stoltzfus, R.J. 2006. Deficits in cognitive function and achievement in
Mexican first-graders with low blood lead concentrations. Environmental Research. 100: 371-
386.
Lanphear, B.P.; Burgoon, D.A.; Rust, S.W.; Eberly, S.; and Galke, W. 1998. Environmental
exposures to lead and urban children's blood lead levels. Environmental Research. 76: 120-130.
Lanphear, B.P.; Howard, C.; Eberly, S.; Auinger, P.; Kolassa, J.; Weitzman, M.; Schaffer, S. J.;
and Alexander, K. 1999. Primary prevention of childhood lead exposure: a randomized trial of
dust control. Pediatrics. 103: 772-777.
Lanphear, B.P.; Dietrich, K.; Auinger, P.; and Cox, C. 2000. Cognitive deficits associated with
blood lead concentrations < 10 |ig/dL in U.S. children and adolescents. Public Health Reports.
115: 521-529.
Lanphear, B.P.; Hornung, R.; Khoury, J.; Yolton, K.; Baghurst, P.; Bellinger, D.C.; Canfield,
R.L.; Dietrich, K.N.; Bornschein, R.; Greene, T.; Rothenberg, S.J.; Needleman, H.L.; Schnaas,
L.; Wasserman, G.; Graziano, J.; and Roberts, R. 2005. Low-level environmental lead exposure
and children's intellectual function: an international pooled analysis. Environmental Health
Perspectives. 113: 894-899.
Lanphear, B.P.; Hornung, R.; Khoury, J.; Yolton, K.; and Dietrich, K.N. 2006. Lead and IQ in
children: Lanphear et al. respond [letter]. Environmental Health Perspectives. 114: A86-A87.
125
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
Leggett,R.W. 1993. An Age-Specific Kinetic Model of Lead Metabolism in Humans. Environ
Health Perspect 101: 598-616.
Moir, C. M.; Freedman, B.; and McCurdy, R. 1996. Metal mobilization from water-distribution
systems of buildings serviced by lead-pipe mains. J. Can. Water Resour. 21: 45-52.
NAHB. 2006. Lead-Safe Work Practices Survey Project Report. Prepared by Atrium
Environmental Health and Safety Services (Reston, VA) for the National Association of Home
Builders. November.
Pounds J.G. and Leggett J. 1998. The ICRP Age-Specific Biokinetic Model for Lead:
Validations, Empirical Comparisons, and Explorations. Environ. Health Pers. 106 (S6): 1505-
1511.
Pounds, JG. 2000. An Operators Manual for the Leggett Age-Dependent Biokinetic Model for
Lead. (Version 1.1)
Rogan, W.J. and Ware, J.H. 2003. Exposure to lead in children — how low is low enough?
New England Journal ofMedicine. 348: 1515-1516.
Succop, P.; Bornschein, R.; Brown, K.; and Tseng, C.-Y. 1998. An empirical comparison of
lead exposure pathway models. Environmental Health Perspectives. Suppl. 106(6): 1577-1583.
Tellez-Rojo, M.M.; Bellinger, D.C.; Arroyo-Quiroz, C.; Lamadrid-Figueroa, H.; Mercado-
Garcia, A.; Schnaas-Arrieta, L.; Wright, R.O.; Hernandez-Avila, M.; and Hu, H. 2006.
Longitudinal associations between blood lead concentrations <10 |ig/dL and neurobehavioral
development in environmentally-exposed children in Mexico City. Pediatrics. 118: e323-e330.
Tong, S.; Baghurst, P.; McMichael, A.; Sawyer, M.; and Mudge, J. 1996. Lifetime exposure to
environmental lead and children's intelligence at 11-13 years: the Port Pirie cohort study. British
Medical Journal. 312: 1569-1575.
U.S. Census Bureau. 1995. Property Owners and Managers Survey (POMS). Available at:
http://www.census.gov/hhes/www/housing/poms/poms.html.
U.S. Census Bureau. 1997. 1997 National American Housing Survey Data. Revised 11/29/99.
Available at: http://www.huduser.org/datasets/ahs/ahsdata97.html.
U.S. Census Bureau. 2003. 2003 National American Housing Survey Data. Revised 02/14/06.
Available at: http://www.huduser.org/datasets/ahs/ahsdata03.html.
U.S. Dept. of Housing and Urban Development (HUD). 2002. National Survey of Lead and
Allergens in Housing: Final Report, Revision 6. Office of Lead Hazard Control. October 31.
Available at: http://www.hud.g0v/0ff1ces/lead/techstudies/survey.cfm#natsurvey.
126
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
U.S. Environmental Protection Agency (USEPA). 1986a. Air quality criteria for lead. Research
Triangle Park, NC: Office of Health and Environmental Assessment, Environmental Criteria and
Assessment Office. EPA/600/8-83/028aF-dF. 4v. Available at: NTIS, Springfield, VA; PB87-
142378.
U.S. Environmental Protection Agency (USEPA). 1986b. Lead effects on cardiovascular
function, early development, and stature: an addendum to U.S. EPA Air Quality Criteria for
Lead, 1986. In: Air quality criteria for lead. v. 1. Research Triangle Park, NC: Office of Health
and Environmental Assessment, Environmental Criteria and Assessment Office. A1-A67. EPA-
600/8-83/028aF. Available from: NTIS, Springfield, VA; PB87-142378.
U.S. Environmental Protection Agency (USEP A). 1989. Review of the National Ambient Air
Quality Standard for Lead: Exposure Analysis Methodology and Validation. Research Triangle
Park, NC: Office of Air Quality Planning and Standards. EPA-450/2-89-011.
U.S. Environmental Protection Agency (USEPA). 1990. Air quality criteria for lead:
supplement to the 1986 addendum. Research Triangle Park, NC: Office of Health and
Environmental Assessment, Environmental Criteria and Assessment Office. EPA/600/8-
89/049F. Available from: NTIS, Springfield, VA; PB91-138420.
U.S. Environmental Protection Agency (USEPA). 1994. Technical Support Document:
Parameters and Equations Used in the Integrated Exposure Uptake Biokinetic Model for Lead in
Children (v.099d). Office of Solid Waste. EPA 540/R-94/040.
U.S. Environmental Protection Agency (USEPA). 1997. Lead Exposure Associated with
Renovation and Remodeling Activities: Environmental Field Sampling Study, Volume I:
Technical Report. Prepared by Battelle (Columbus, OH) for the Office of Pollution Prevention
and Toxics (Washington, D.C.). EPA 747-R-96-007.
U.S. Environmental Protection Agency (USEPA). 1998. Risk Analysis to Support Standards for
Lead in Paint, Dust, and Soil. Office of Pollution Prevention and Toxics. EPA 747-R-97-006.
U.S. Environmental Protection Agency (USEPA). 2000. Risk Analysis to Support Standards for
Lead in Paint, Dust, and Soil: Supplemental Report. Office of Pollution, Prevention and Toxics.
December. (EPA 747-R-00-004). Available at:
http://www.epa.gov/oppt/lead/pubs/403risksupp.htm.
U.S. Environmental Protection Agency (USEPA). 2002a. Exposure Factors Handbook.
National Center for Environmental Assessment, Office of Research & Development.
U.S. Environmental Protection Agency (USEPA). 2002b. Children-Specific Exposure Factors
Handbook Interim Draft. National Center for Environmental Assessment, Office of Research and
Development. EPA-600-P-00-002B.
U.S. Environmental Protection Agency (USEPA). 2005a. Integrated Exposure Uptake
Biokinetic Model for Lead in Children, Windows® version (IEUBKwin vl.O build 263).
Available at: http://www.epa.gov/superfund/1 ead/products.htm.
127
-------
DRAFT FOR CAS AC REVIEW ON JULY 9-10, 2007
U.S. Environmental Protection Agency (USEPA) 2005b. All-Ages Lead Model (AALM)
Version 1.05 (External Review Draft). U.S. Environmental Protection Agency, Washington, DC,
EPA/600/C-05/013, 2005. Available at:
http://cfpub. epa.gov/ncea/cfm/recordisplay. cfm?deid=139314
U.S. Environmental Protection Agency (USEPA). 2006a. Air Quality Criteria Document for
Lead. National Center for Environmental Assessment, Research Triangle Park. Available at:
http://cfpub. epa.gov/ncea/cfm/recordisplay. cfm?deid=l 58823.
U.S. Environmental Protection Agency (USEPA). 2006b. AirData: Access to Air Pollution
Data through the Air Quality Systems (AQS) database. Available at:
http://www.epa.gov/air/data.
U.S. Environmental Protection Agency (USEPA). 2006c. Specific estimates of dietary Pb
intake developed by EPA's Office of Solid Waste and Emergency Response. Available at:
http://www.epa.gOv/superfund/lead/ieubkfaq.htm#FDA.
U.S. Environmental Protection Agency (USEPA). 2006d. Air Quality Criteria for Lead,
Volume I of II. Research Triangle Park, NC: National Center for Environmental Assessment -
RTP Office. EPA/600/R-5/144aF. October. Available at:
http://cfpub. epa.gov/ncea/cfm/recordisplay. cfm?deid=l 58823.
U.S. Food and Drug Administration (FDA). 2001. Total Diet Study. U. S. Food and Drug
Administration Center for Food Safety and Applied Nutrition Office of Plant and Dairy Foods
and Beverages (June 2001). Available at: http://www.cfsan.fda.gov/~comm/tds-toc.html.
van Wijnen, J.H.; Clausing, P.; and Brunekreef B. 1990. Estimated soil ingestion by children.
Environ Res. 51 (2): 147-62.
Wasserman, G.A.; Liu, X.; Popovac, D.; Factor-Litvak, P.; Kline, J.; Waternaux, C.; Lolacono,
N.; and Graziano, J. H. 2000. The Yugoslavia prospective lead industry study: contributions of
prenatal and postnatal lead exposure to early intelligence. Neurotoxicology and Teratology. 22:
811-818.
White, P.D.; Van Leeuwen, P.; Davis, B.D.; Maddaloni, M.; Hogan, K.A.; and Marcus, A.H.
1998. The conceptual structure of the integrated exposure uptake biokinetic model for lead in
children. Environmental Health Perspectives. 106 (S6): 1513-1530.
Ziegler, E. E.; Edwards, B. B.; Jensen, R. L.; Mahaffey, K. R.; and Fomon, S. J. 1978.
Absorption and retention of lead by infants. Pediatr. Res. 12:29-34.
128
------- |