United States
Environmental Protection
Agency
National Center for
Environmental Assessment
Research Triangle Park, NC 27711
EPA/600/R-95/139
September 1995
External Review Draft
EPA
Urban Soil
Lead Abatement
Demonstration
Project
Review
Draft
(Do Not
Cite or Quote)
EPA Integrated Report
NOTICE
This document is a preliminary draft. It has not been formally
released by EPA and should not at this stage be construed to
represent Agency policy. It is being circulated for comment on its
technical accuracy and policy implications.
-------
-------
EXTERNAL REVIEW DRAFT
EPA/600/R-95/139
URBAN SOIL LEAD ABATEMENT
DEMONSTRATION PROJECT
EPA INTEGRATED REPORT
National Center for Environmental Assessment
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
September 1, 1995
DRAFT-DO NOT QUOTE OR CITE
-------
DISCLAIMER
This document is an external draft for review purposes only and does not constitute
U.S. Environmental Protection Agency policy. Mention of trade names or commercial
products does not constitute endorsement or recommendation for use.
September 1, 1995
11
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE OF CONTENTS
Page
LIST OF TABLES vii
LIST OF FIGURES „ x
LIST OF PARTICIPANTS xv
LIST OF REVIEWERS xix
LIST OF ABBREVIATIONS, ACRONYMS, AND TERMS xxi
1. EXECUTIVE SUMMARY 1-1
1.1 BACKGROUND AND OVERVIEW 1-1
1.1.1 Comparison of Study Hypotheses 1-3
1.1.2 Study Design and Conduct 1-6
1.1.3 Intervention Procedures 1-6
1.2 SUMMARY OF INDIVIDUAL STUDY REPORTS 1-12
1.2.1 Boston Study 1-12
1.2.2 Baltimore Study 1-13
1.2.3 Cincinnati Study 1-14
1.2.4 Individual Study Conclusions 1-14
1.3 SUMMARY OF EPA INTEGRATED ASSESSMENT RESULTS
AND FINDINGS . 1-16
1.3.1 Quality of the Data 1-16
1.3.2 Effectiveness and Persistency of Intervention 1-17
1.3.3 EPA Integrated Report Results 1-18
1.4 INTEGRATED PROJECT CONCLUSIONS 1-21
2. BACKGROUND AND OVERVIEW OF PROJECT 2-1
2.1 PROJECT BACKGROUND 2-1
2.1.1 The Urban Lead Problem 2-1
2.1.2 Legislative Background 2-1
2.1.3 Site Selection 2-2
2.2 INTEGRATION OF THE THREE STUDIES 2-5
2.2.1 Study Hypotheses 2-5
2.2.2 General Study Design 2-6
2.2.3 Study Groups 2-8
2.2.4 Project Activity Schedule 2-11
2.2.5 Environmental and Biological Measurements
of Exposure 2-11
2.2.5.1 Blood Lead . .' 2-14
2.2.5.2 Hand Lead 2-15
2.2.5.3 House Dust 2-16
2.2.6 Intervention Strategies 2-17
September 1, 1995
in
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE OF CONTENTS (cont'd)
Page
2.3 EXTERNAL FACTORS THAT COULD INFLUENCE
PROJECT RESULTS AND INTERPRETATION 2-18
2.3.1 Cycles and Trends in Environmental Lead
Concentrations • • • 2-19
2.3.2 Unexplained and Unexpected Sources of Lead 2^-22
2.3.3 Movement of Lead in Soil and Dust 2-23
2.3.4 Other Factors 2-24
3. METHODS INTERCOMPARISON AND QUALITY
ASSURANCE/QUALITY CONTROL 3-1
3.1 INTERCOMPARISON OF LABORATORY METHODS FOR
SOIL AND DUST MEASUREMENTS 3-2
3.1.1 Round Robin Intercalation Exercise I 3-3
3.1.2 Quality Assurance/Quality Control Standards
and Audits 3-11
3.1.3 Round Robin Intercalibration Exercise II 3-12
3.1.4 Biweight Distribution and Final Interlaboratory
Calibration 3-14
3.1.5 Disposition of Audit Data 3-15
3.2 QUALITY ASSURANCE AND QUALITY CONTROL FOR
HAND DUST 3-19
3.3 QUALITY ASSURANCE AND QUALITY CONTROL FOR
BLOOD LEAD 3-19
3.4 DATABASE QUALITY 3-19
4. INDIVIDUAL STUDIES 4-1
4.1 INDIVIDUAL STUDY INTERVENTION STRATEGIES
AND SAMPLE PLANS 4-1
4.1.1 Boston Study 4-1
4.1.2 Baltimore Study 4-4
4.1.3 Cincinnati Study 4-5
4.2 DESCRIPTION OF THE DATA 4-8
4.2.1 Measures of Central Tendency for Property Level
Soil and Dust 4-13
4.2.2 Adjustments and Corrections to the Data 4-15
4.2.2.1 Subjects Dropped from Study 4-15
4.2.2.2 Unit Conversion 4-16
4.3 DESIGN DIFFERENCES 4-16
4.4 INDIVIDUAL STUDY CONCLUSIONS 4-18
5. RESULTS OF INTEGRATED ANALYSES 5-1
5.1 BASIC STRATEGY FOR EVALUATING ABATEMENT
EFFECTIVENESS 5-1
September 1, 1995
IV,
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE OF CONTENTS (cont'd)
Page
5.1.1 General Discussion of Conceptual Approaches 5-2
5.1.1.1 Basic Strategies for Evaluating
Abatement Effectiveness 5-2
5.1.1.2 Expected Impact of Intervention 5-8
5.1.2 Conceptual Approach to Differences in Group
Means 5-13
5.1.3 Conceptual Approach to Pre- and Postabatement
Differences in Individuals 5-15
5.1.4 Conceptual Approaches to Repeated Measures
Analyses 5-17
5.1.5 Conceptual Approach to Structural Equation
Modeling 5-20
5.1.6 Comparison of Interventions Across Studies 5-23
5.2 DIFFERENCES IN GROUP MEANS 5-24
5.2.1 Changes in Mean Soil Concentrations 5-24
5.2.2 Changes hi Exterior Dust Concentrations and
Loadings 5-29
5.2.3 Changes in Interior Dust Concentrations and
Loadings 5-32
5.2.4 Changes in Hand Dust Loadings 5-51
5.2.5 Changes in Blood Lead Concentrations 5-51
5.2.5.1 Baltimore Study Blood Lead Data 5-51
5.2.5.2 Boston Study Blood Lead Data 5-53
5.2.5.3 Cincinnati Study Blood Lead Data 5-53
5.3 PRE- AND POSTABATEMENT DIFFERENCES
IN INDIVIDUALS 5-53
5.3.1 Individual Changes in Blood Lead and Soil Lead 5-53
5.4 COMPARISON BY REPEATED MEASURES ANALYSIS .... 5-66
5.4.1 Baltimore Study 5-66
5.4.2 Boston Study 5-70
5.4.3 Cincinnati Study 5-74
5.4.4 Repeated Measures Analyses Adjusted for
Environmental Analysis and Demographics 5-79
5.4.4.1 Results from Boston Study 5-79
5.4.4.2 Results of Baltimore Study 5-81
5.4.4.3 Results of the Cincinnati Study . 5-83
5.5 COMPARISONS USING STRUCTURAL EQUATIONS
MODELS 5-87
5.5.1 General Issues in Structural Equation Modeling 5-91
5.5.2 Results of Structural Equation Model Analyses 5-92
5.5.2.1 Baltimore Study 5-92
5.5.2.2 Boston Study 5-96
5.5.2.3 Cincinnati Study 5-99
September 1, 1995
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE OF CONTENTS (cont'd)
5.6 SUMMARY OF STATISTICAL ANALYSES 5-106
5.6.1 General Observations 5-106
5.6.1.1 Combining Studies 5-106
•5.6.1.2 Measurement Error 5-106
5.6.2 Summary of Results 5-107
5.6.3 Limitations of the Statistical Methods 5-108
5.6.4 Comparison Across the Three Studies 5-109
6. INTEGRATED SUMMARY AND CONCLUSIONS 6-1
6.1 PROJECT OVERVIEW 6-1
6.2 SUMMARY OF FINDINGS 6-3
6.2.1 EPA Integrated Report Results 6-3
6.2.2 Application of Findings to Conceptual Framework
of Soil Lead Exposure Pathway 6-7
6.3 INTEGRATED PROJECT CONCLUSIONS 6-10
7. REFERENCES 7-1
APPENDIX A: GROUP MEAN PARAMETERS FOR EACH STUDY
BY SAMPLE TYPE, TREATMENT GROUP,
AND ROUND A-l
September 1, 1995
VI
DRAFT-DO NOT QUOTE OR CITE
-------
LIST OF TABLES
Number
1-1 Description of Study Groups and Types of Intervention . . .
1-2 Number of Project Participants by Round . . .
1-3 Soil Abatement Statistics for the Three Studies
2-1 Treatment Group Nomenclature with Cross-Reference to
Individual Reports
2-2 Number of Project Participants by Treatment Group
and Round
3-1 Wet Chemistry and Instrumental Methods Used for the First
Intercalibration Study
3-2 Analytical Results of the First Intercalibration Study:
Lead Concentration in the Total and Fine Fractions of
10 Soils from Each Study
3-3 Soil and Dust Audit Program Results
3-4 Preliminary and Final Biweight Distributions for Soil and
Dust Audit Program
3-5 Results of the Final Intercalibration Study . . .
3-6 Consensus Values and Correction Factors from the
Final Intercalibration Program
3-7 Quality Control Results for Centers for Disease Control
and Prevention Blind Pool Lead Analyses . .
4-1 Soil Abatement Statistics for the Three Studies
4-2 Summary of Boston Study Data
4-3 Summary of Baltimore Study Data
4-4 Summary of Cincinnati Study Data
4-5 Design Differences Between the Three Studies
1-5
1-7
1-10
2-10
2-12
3-4
3-5
3-13
3-14
3-16
3-17
3-20
4-2
4-9
4-10
4-11
4-17
September 1, 1995
vn
DRAFT-DO NOT QUOTE OR CITE
-------
LIST OF TABLES (cont'd)
Number
5-1 Statistical Significance of Baltimore Repeated Measures
Analyses for Blood Lead, Rounds 3 and 6 (Pre- and
Postabatement), After Covariate Adjustment .
5-2 Statistical Significance of Baltimore Repeated Measures
Analyses for the Logarithm of Hand Lead, Rounds 3 and 6
(Pre- and Postabatement), Covariate Adjustment
5-3 Statistical Significance of Boston Repeated Measures
Analyses for Blood Lead, Rounds 1 and 3 (Pre- and
Postabatement), After Covariate Adjustment
5-4 Statistical Significance of Boston Repeated Measures
Analyses for Natural Logarithm of Hand Lead, Rounds 1 and 3
(Pre- and Postabatement), After Covariate Adjustment
5-5 Statistical Significance of Cincinnati Repeated Measures
Analyses for Blood Lead, Rounds 1 and 4 (12 Months),
After Covariate Adjustment
5-6 Statistical Significance of Cincinnati Repeated Measures
Analysis for Hand Lead, Rounds 1 and 4 (12 Months),
After Covariate Adjustment
5-7 Repeated Measures Analyses of Blood Lead in Boston Study
for First Year After Abatement, Adjusted for Differences
in Environmental Indices and Demographics
5-8 Repeated Measures Analyses of Blood Lead in Baltimore Study
for First Year After Abatement, Adjusted for Differences
in Environmental Indices and Demographics
5-9 Repeated Measures Analyses of Blood Lead hi Cincinnati Study
for First Year After Abatement, Adjusted for Differences
in Environmental Indices
5-10 Repeated Measures Analyses of Blood Lead in Cincinnati Study
for First Year After Abatement, Adjusted for Differences
hi Environmental Indices: Mohawk Versus Pendleton
5-11 Baltimore Structural Equation Model Full Information
Maximum Likelihood Method
5-68
5-69
5-74
5-74
5-75
5-75
5-80
5-82
5-84
5-86
5-94
September 1, 1995
vin
DRAFT-DO NOT QUOTE OR CITE
-------
Number
5-12
5-13
5-14
5-15
LIST OF TABLES (cont'd)
Baltimore Structural Equation Model Full Information
Maximum Likelihood Method
Boston Structural Equation Model Blood Lead Versus
Dust Lead Loading Full Information Maximum
Likelihood Method
Boston Structural Equation Model Blood Lead Versus Dust
Lead Concentration Full Information Maximum
Likelihood Method
Cincinnati Structural Equation Model Blood Lead Versus
Sidewalk Dust Lead Concentration Iterated Two-Stage
Least Squares Method
5-95
5-98
5-99
5-103
September 1, 1995
IX
DRAFT-DO NOT QUOTE OR CITE
-------
LIST OF FIGURES
Number Page
1-1 Generalized concept of the sources and pathways of lead
exposure in humans 1-8
1-2 Typical pathways of childhood exposure to lead in dust 1-8
2-1 Project activity schedule showing round designations and
time periods for sampling and interviewing and the tune
periods for soil abatement 2-13
2-2 Generalized concept of the sources and pathways of lead
exposure in humans 2-14
2-3 Typical pathways of childhood exposure to lead in dust 2-15
2-4 Literature values for seasonal patterns for childhood
blood lead (age 25 to 36 mo) 2-19
2-5 Literature values for seasonal patterns for blood lead
in children and adults (NHANES II, age 6 mo to 74 years) 2-20
2-6 Predicted differences in blood lead and hand lead during
early childhood, based on empirical data 2-21
3-1 Comparison of uncorrected data for two wet chemistry methods
of soil analysis showing the comparability of hot and cold
nitric acid for the Cincinnati laboratory 3-6
3-2 Comparison of uncorrected data for atomic absorption spectroscopic
analysis by two laboratories (Baltimore and Cincinnati) using
the hot nitric acid method of soil analysis 3-7
3-3 Interlaboratory comparison of uncorrected data for the X-ray
fluorescence method of soil analysis showing the comparability
of the Boston and Georgia Institute of Technology laboratories ... 3-8
3-4 Interlaboratory comparison of uncorrected data for soil
analysis showing the comparability of inductively coupled
plasma emission spectroscopy and atomic absorption spectroscopy
for the Baltimore and Cincinnati laboratories 3-8
3-5 Comparison of uncorrected data for soil analysis showing the
comparability of inductively coupled plasma emission spectroscopy
and atomic absorption spectroscopy within the Baltimore
laboratory 3-9
September 1, 1995
DRAFT-DO NOT QUOTE OR CITE
-------
Number
3-6
3-7
3-8
3-9
4-1
4-2
4-3
5-1
5-2
5-3
5-4
5-5
5-6
LIST OF FIGURES (cont'd)
Interlaboratory comparison of unconnected data for soil analysis
showing the comparability of X-ray fluorescence and atomic
absorption spectroscopy for the Cincinnati and Boston
laboratories
Interlaboratory comparison of uncorrected data for soil analysis
showing the comparability of X-ray fluorescence and atomic
absorption spectroscopy for the Baltimore and Boston
laboratories
Departures from consensus dust values for each of the
three studies
Departures from consensus soil values for each of the
three studies
Pathway intervention scheme for dust exposure (Boston Soil
Abatement Study)
Pathway intervention scheme for dust exposure (Baltimore Soil
Abatement Study)
Pathway intervention scheme for dust exposure (Cincinnati Soil
Abatement Study)
Hypothetical representation of the expected decrease in blood lead,
following abatement
A simple approach that compares lead variables before and
after abatement comparable to Strategy 1
Page
A more complex approach that uses covariate adjustments with
repeated measures analysis, comparable to Strategy 2
A structural equation modeling approach comparable to
Strategy 3
Schematic representation of expected outcomes for treatment and
control groups
Schematic representation of the potential interpretations
that might be reached from the various abatement outcomes ....
3-10
3-10
3-18
3-18
4-2
4-4
4-6
5-3
5-6
5-7
5-8
5-14
5-16
September 1, 1995
xi DRAFT-DO NOT QUOTE OR CITE
-------
LIST OF FIGURES (cont'd)
Number
5-7 Hypothetical representation of common statistical parameters for a
single group and a single round
5-8 Boston soil lead concentrations by study group show the
effectiveness and persistency of soil abatement
5-9 Cincinnati soil lead concentrations
5-10 Exterior dust lead concentrations from the street samples
in the Cincinnati study
5-11 Exterior dust lead concentrations from the sidewalk samples
hi the Cincinnati study
5-12 Boston floor dust lead concentration
5-13 Boston floor dust load
5-14 Boston floor dust lead load
5-15 Boston window dust lead concentrations
5-16 Boston window dust load
5-17 Boston window dust lead load
5-18 Cincinnati floor dust lead concentrations
5-19 Cincinnati floor dust load
5-20 Cincinnati floor dust lead load
5-21 Cincinnati window dust lead concentration
5-22 Cincinnati window dust load
5-23 Cincinnati window dust lead load
5-24 Cincinnati entry dust lead concentration
5-25 Cincinnati entry dust load
5-26 Cincinnati entry dust lead load
Page
5-26
5-27
5-28
5-30
5-31
5-33
5-34
5-35
5-36
5-37
5-38
5-39
5-40
5-41
5-42
5-43
5-44
5-45
5-46
5-47
September 1, 1995
xn
DRAFT-DO NOT QUOTE OR CITE
-------
Number
LIST OF FIGURES (cont'd)
5-27 The Boston hand lead load
5-28 Baltimore hand lead load
5-29 Cincinnati hand lead load
5-30 Baltimore blood lead concentrations
5-31 Boston blood lead concentrations
5-32 Cincinnati blood lead concentrations
5-33 Double-difference plot of the change in soil lead versus the
change in blood for the Baltimore study
5-34 Double-difference plot for Boston soil and blood lead data .
5-35 Double-difference plot for Boston floor dust lead
concentrations and blood lead concentrations
5-36 Double-difference plot for Boston floor dust lead loading
and blood lead concentrations
5-37 Double-difference plot for Cincinnati entry dust lead
concentrations and blood lead concentrations
5-38 Double-difference plot for Cincinnati entry dust lead
concentrations and blood lead concentrations
5-39 Double-difference plot for Cincinnati entry dust lead loading
and blood lead concentrations
5-40 Double-difference plot for Cincinnati floor dust lead
concentrations and blood lead concentrations
5-41 Double-difference plot for Cincinnati floor dust lead loading
and blood lead concentrations
5-42 Change in preabatement geometric mean blood lead levels in
Baltimore study 1 year after abatement
5-43 Change in preabatement geometric mean hand lead levels in
Baltimore study 1 year after abatement
Page
5-48
5-49
5-50
5-52
5-54
5-55
5-57
5-58
5-59
5-60
5-61
5-62
5-63
5-64
5-65
5-67
5-67
September 1, 1995
xm
DRAFT-DO NOT QUOTE OR CITE
-------
LIST OF FIGURES (cont'd)
Number
5-44 Change in preabatement geometric mean blood lead levels in
Boston study 1 year after abatement
5-45 Change in preabatement geometric mean hand lead levels in
Boston study 1 year after abatement
5-46 Change in preabatement geometric mean floor dust lead
concentration in Boston study 1 year after abatement
5-47 Change in preabatement geometric mean floor dust lead loading
in Boston study 1 year after abatement
5-48 Change hi preabatement geometric mean blood lead levels in
Cincinnati study 1 year after abatement
5-49 Change in preabatement geometric mean hand lead levels in
Cincinnati study 1 year after abatement
5-50 Change in preabatement geometric mean floor dust lead
concentrations in Cincinnati study 1 year after abatement . . . .
5-51 Change in preabatement geometric mean floor dust lead loading
in Cincinnati study 1 year after abatement
5-52 Explanation of the terms and features of the structural equation
model diagram in Figures 5-53, 5-54, and 5-55
5-53 Structural equation model for childhood exposure in Baltimore .
5-54 Structural equation model for Boston
5-55 Structural equation model for Cincinnati
6-1 Total amounts of lead in various compartments of a child's
environment, using the assumptions for concentration
or lead loading
Page
5-70
5-71
5-71
5-72
5-76
5-77
5-77
5-78
5-89
5-89
5-90
5-90
6-9
September 1, 1995
xiv
DRAFT-DO NOT QUOTE OR CITE
-------
LIST OF PARTICIPANTS
Urban Soil Lead Abatement Demonstration Projects
Dr. Ann Aschengrau
Boston University School of Public Health
80 East Concord Street, T-355
Boston, MA 02118
Dr. David Bellinger
Children's Hospital
Gardner House Room 455
300 Longwood Avenue
Boston, MA 02115
Dr. Robert Bornschein
University of Cincinnati
Department of Environmental Health
3223 Eden Avenue #56
Cincinnati, OH 45267-0056
Ms. Dawn Boyer
Inorganic Chemistry Department
Lockheed ESC
1050 East Flamingo
Las Vegas, NV 89119
Ms. Merrill Brophy
MDE/Lead & Soil Project
Maryland Department of the Environment
2500 Broening Highway
Baltimore, MD 21224
Dr. Richard Brunker
U.S. EPA - Region III
Site Support Section MD-3HW26
841 Chestnut Street
Philadelphia, PA 19107
Mr. Barry Chambers
Maryland Department of the Environment
2500 Broening Highway
Baltimore, MD 21224
Dr. Rufus Chancy
U.S. Department of Agriculture
ARC, Building 318 BARC-East
Beltsville, MD 20705
Dr. Julian Chisolm
Kennedy Institute
707 N. Broadway
Baltimore, MD 21205
Dr. Scott Clark
University of Cincinnati
Department of Environmental Health
Mail Stop 56
3223 Eden Avenue
Cincinnati, OH 45267-0056
Ms. Linda Conway-Mundew
University of Cincinnati
Department of Environmental Health
1142 Main Street
Cincinnati, OH 45210
Dr. Robert Elias
U.S. EPA
National Center for Environmental
Assessment
Mail Drop 52
Research Triangle Park, NC 27711
Dr. Katherine Farrell
Anne Arundell County Health Department
3 Harry S. Truman Parkway
Annapolis, MD 21401
Ms. Beverly Fletcher
U.S. EPA - Region I
Environmental Services Division
60 Westview Street
Lexington, MA 02173
September 1, 1995
xv
DRAFT-DO NOT QUOTE OR CITE
-------
LIST OF PARTICIPANTS (cont'd)
Urban Soil Lead Abatement Demonstration Projects (cont'd)
Mrs. Barbara Gordon
Cincinnati Health Department
3101 Burnet Avenue, Room 309
Cincinnati, OH 45229
Ms. Jo Ann Grote
University of Cincinnati
Department of Environmental Health
1142 Main Street
Cincinnati, OH 45210
Mr. Bill Hanson
Cincinnati Health Department
UC Soil Project
1142 Main Street
Cincinnati, OH 45210
Mr. Reginald Harris
U.S. EPA - Region ffl
Site Support Section MD-3HW15
841 Chestnut Street
Philadelphia, PA 19107
Mr. Ronald Jones
Cleveland Department of Public Health
1925 East St. Claire Ave.
Cleveland, OH 44114
Dr. Boon Lim
Environmental Health Program
Maryland Department of the Environment
2500 Broening Highway
Baltimore, MD 21224
Dr. Allan Marcus
U.S. EPA
National Center for Environmental
Assessment
Mail Drop 52
Research Triangle Park, NC 27711
Dr. Tom Matte
Public Health Service - Region II
26 Federal Plaza, Room 3337
New York, NY 10278
Ms. Lisa Matthews
U.S. EPA - OS 230
401 M Street, SW
Washington, DC 20460
Mr. J. Todd Scott
The Cadmus Group
3580 Cinderbed Road
Suite 2400
Newington, VA 22122
Mr. Dave Mclntyre
U.S. EPA - Region I
Environmental Services Division
60 Westview Street
Lexington, MA 02173
Mr. William Menrath
University of Cincinnati
Department of Environmental Health
1142 Main Street
Cincinnati, OH 45210
Dr. Winkey Pan
University of Cincinnati
Department of Environmental Health
3223 Eden Avenue
Cincinnati, OH 45267-0056
Dr. Dan Paschal
Centers for Disease Control
1600 Clifton Road, NE
Mail Stop F-18
Atlanta, GA 30333
September 1, 1995
xvi
DRAFT-DO NOT QUOTE OR CITE
-------
LIST OF PARTICIPANTS (cont'd)
Urban Soil Lead Abatement Demonstration Projects (cont'd)
Ms. Sandy Roda
University of Cincinnati
Department of Environmental Health
3223 Eden Avenue
Cincinnati, OH 45267-0056
Dr. Charles Rohde
Biostatistics
Johns Hopkins University
615 N. Wolfe Street
Baltimore, MD 21205
Ms. Penny Schmitgen
University of Cincinnati
Department of Environmental Health
3223 Eden Avenue
Cincinnati, OH 45210
Dr. James Simpson
Centers for Disease Control
CEHIC/EHHE, Mail Stop F28
1600 Clifton Road
Atlanta, GA 30333
Dr. Tom Spittler
U.S. EPA - Region I
Environmental Services Division
60 Westview Street
Lexington, MA 02173
Mr. Warren Strauss
MDE/Lead & Soil Project
Maryland Department of the Environment
2500 Broening Highway
Baltimore, MD 21224
Dr. Paul A. Succop
University of Cincinnati
Department of Environmental Health
3223 Eden Avenue
Cincinnati, OH 45267-0056
Dr. Pat VanLeeuwen
U.S. EPA - Region V
Technical Support Unit 5HR-11
230 S. Dearborn Street
Chicago, IL 60404
Dr. Harold A. Vincent
U.S. EPA
Quality Assurance Division
Environmental Monitoring Systems
Laboratory - Las Vegas
P.O. Box 93478
Las Vegas, NV 89193-3478
Dr. Michael Weitzman
Chief of Pediatrics
Rochester General Hospital
1425 Portland Avenue
Rochester, NY 14621
September 1, 1995
xvn
DRAFT-DO NOT QUOTE OR CITE
-------
-------
LIST OF REVIEWERS
September 1, 1995
xix
DRAFT-DO NOT QUOTE OR CITE
-------
-------
LIST OF ABBREVIATIONS, ACRONYMS, AND TERMS
AAS
ANCOVA
BALP
BALSP
BOSP
BOS PI
BOS SPI
CDC
CIN I-SE
CINNT
CIN SEI
dL
Double blind
Dust loading
ECAO/RTP
EPA
GLIM
Atomic absorption spectroscopy
Analysis of covariance
Baltimore Study Group with paint intervention
Baltimore Study Group with soil and paint intervention
Boston Study Group with paint intervention
Boston Study Group with paint and interior dust
intervention
Boston Study Group with soil, paint, and interior dust
intervention
Centers for Disease Control and Prevention
Cincinnati Study Group with interior dust intervention,
followed by soil and exterior dust intervention (second
year)
Cincinnati Study Group with no treatment
Cincinnati Study Group with soil, exterior dust, and
interior dust intervention
Deciliter; used here as a measure of blood lead in
micrograms per deciliter
Analytical audit sample where analyst knows neither that
the sample is an audit sample nor the concentration
Mass of dust per unit area
Environmental Criteria and Assessment Office/Research
Triangle Park (now National Center for Environmental
Assessment/Research Triangle Park)
U.S. Environmental Protection Agency
Numerical Algorithms Group software package for a
general linear model
September 1, 1995
xxi
DRAFT-DO NOT QUOTE OR CITE
-------
LIST OF ABBREVIATIONS, ACRONYMS, AND TERMS (cont'd)
GLM
Hand dust
HEPA
ICP
Lead concentration
Lead loading
MGLH
NHANESH
ORD
OSWER
P-value
Pb
Project
P-XRF
QA/QC
Repeated measures analysis
Round
SARA
September 1, 1995
SAS procedure for general linear models approximately
equivalent to Systat MGLH
Sample taken by wiping the child's hand thoroughly;
a measure estimating the ingestion of lead
High-efficiency particle accumulator
Inductively coupled plasma emission spectroscopy
Mass of lead per mass of medium (soil, dust, water)
Mass of lead per unit area
Systat procedure for general linear models approximately
equivalent to SAS GLM
National Health Assessment and Nutrition Examination
Survey II
Office of Research and Development
Office of Solid Waste and Emergency Response
Statistical term for the likelihood that an observed effect
differs from zero
Lead
In this report, "project" refers collectively to the three
individual studies that compose the Urban Soil Abatement
Demonstration Project.
Field or Portable XRF used in this study for paint
measurements
Quality assurance/quality control
Statistical procedure for analyzing normally distributed
responses collected longitudinally
3JI
Period of sampling and data collection during study
Superfund Amendments and Reauthorization Act
xxii DRAFT-DO NOT QUOTE OR CITE
-------
LIST OF ABBREVIATIONS, ACRONYMS, AND TERMS (cont'd)
SAS
SES
Single blind
Study
SYSTAT
USLADP
XRF
Statistical software package
Socioeconomic status
Analytical audit sample where analyst knows sample is an
audit sample but doesn't know concentration (see Double
blind)
In this report, "study" refers to one of the three
individual soil abatement studies that compose the Urban
Soil Abatement Demonstration Project.
Statistical software package
Urban Soil Lead Abatement Demonstration Project
Laboratory scale X-ray fluorescence instrument used in
this study for soil and dust analysis (see P-XRF)
September 1, 1995
xxm
DRAFT-DO NOT QUOTE OR CITE
-------
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
1. EXECUTIVE SUMMARY
1.1 BACKGROUND AND OVERVIEW
In the past 25 years, concern for children with lead poisoning has steadily increased
with mounting evidence for the subtle but serious metabolic and developmental effects of lead
exposure levels previously thought to be safe. Childhood lead poisoning was formerly
considered a severe medical problem usually traced to swallowed chips of peeling lead-based
paint. Scientific evidence has systematically revealed deleterious effects of lead at lower
levels of exposure. Agencies such as the U.S. Environmental Protection Agency (EPA) and
the Centers for Disease Control and Prevention (CDC) have repeatedly lowered the level of
concern for children's lead burden that recommends environmental or clinical intervention
from a blood lead level of 30 jtig/dL established in 1978 by CDC to 25 jig/dL in 1985, just
prior to the start of this project, then to the present level of 10 /xg/dL, which was defined hi
October 1991 by CDC as a blood lead level that should trigger community-wide prevention
activities if observed in many children.
The relationship between soil lead and blood lead is an indirect relationship hi the sense
that children most commonly do not eat soil directly but ingest small amounts of dust
derived, in part, from this soil. In the child's environment, dust is only one of several
sources of lead that also include food, air, and drinking water. Likewise, the lead hi blood
reflects not only recent exposure from these sources but also the biokinetic processes that
distribute and redistribute lead between blood and other body tissues, especially bone tissue.
The Urban Soil Lead Abatement Demonstration Project (USLADP), known also as the
Three City Lead Study, was authorized hi 1986 under Section lll(b)(6) of the Superfund
Amendments and Reauthorization Act (SARA), which mandated that EPA conduct soil lead
abatement projects in up to three U.S. cities (SMSA's). The purpose of the project was to
determine whether abatement of lead in soil could reduce the lead in blood of inner city
children. It did not attempt to compare the relative effectiveness of alternative soil abatement
methods.
This report, then, is an integrated assessment of data from three coordinated
longitudinal studies of children in urban neighborhoods of three cities (Boston, Baltimore,
September 1, 1995
1-1
DRAFT-DO NOT QUOTE OR CITE
-------
1 Cincinnati), where intervention into soil lead exposure pathways was expected to reduce the
2 children's blood lead. Many cross-sectional studies of childhood lead exposure have
3 previously shown that differences in soil lead exposure are associated with differences hi
4 blood lead concentrations, but they did not evaluate the effectiveness of intervention steps hi
5 terms of demonstrating that reductions in external exposure to lead from soil result in
6 reductions in blood lead concentrations. Thus, a unique aspect of this project is that it
7 measures response to intervention, not to contamination. Because of the physiology of lead
8 mobilization in body tissues, there is a difference between the rate of change in a population
9 with increasing lead exposure and in one with decreasing exposure. In other words, the
10 decrease hi blood lead concentrations in response to intervention was not expected to be at
11 the same rate as an increase in blood lead concentrations in response to increasing exposure.
12 The project began in December 1986 with the appointment of an EPA steering
13 committee to develop recommendations for implementing the SARA lead-in-soil
14 demonstration project. A panel of experts was formed in early 1987 to assist EPA in
15 defining a set of criteria for selection of sites and the minimum requirements for a study at
16 each site. The panel also met in mid 1987 to discuss technical issues and study designs and
17 to evaluate technical criteria for selection of urban areas as potential soil-lead abatement
18 demonstration project sites, ultimately leading by the end of 1987 to the selection of Boston,
19 Baltimore, and Cincinnati as the participating cities.
20 The individual studies were each designed around the concept of participating families
21 within a definable neighborhood. These families and their living units were part of a study
22 group, either a treatment group or a control group. Each study group was sampled during
23 preabatement and postabatement phases of the studies carried out in each city. Prior to and
24 after abatement, blood lead levels were ascertained and the environment of the child was
25 extensively evaluated through measurements of lead in soil, dust, drinking water, and paint,
26 and through questionnaires about activity patterns, eating habits, family activities, and
27 socioeconomic status (SES). The objective of the preabatement phase was to determine the
28 baseline exposure history and status (stability of the blood lead and environmental measures)
29 prior to abatement. During the postabatement phase, samples were taken to confirm
30 effectiveness of abatement actions in reducing lead in the abated media, to measure the
31 duration of the effect of soil abatement, and to detect possible recontamination. Blood lead
September 1, 1995
1-2
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
measurements were also obtained postabatement to ascertain abatement impacts at various
postabatement intervals.
Research teams in each city included state and/or local health department personnel,
academic researchers from local universities, and/or various other institutions (including hi
Boston participation by EPA Region I Laboratory personnel). Because of the complex nature
of this exposure assessment, intermediate exposure indices, such as street dust, house dust,
and hand dust were measured hi some study groups. Protocols for these measurements were
developed by a Scientific Coordinating Committee composed of representatives from each
study, the three EPA regional offices, the CDC, EPA/Office of Solid Waste and Emergency
Response, and EPA/Office of Research and Development.
1.1.1 Comparison of Study Hypotheses
The Scientific Coordinating Committee attempted to establish uniformity among the
three studies for major aspects of the project. This required a study plan from each city that
was discussed and reviewed at several early planning workshops. Although there were
differences in form and content, each study plan contained
• a statement of the objectives of the study;
• a testable hypothesis that provided direction and focus to the study;
• protocols for collecting and analyzing the data;
• an array of treatment groups that addressed all features of the hypothesis;
• measures to be taken to ensure that all phases of the study would be conducted as
planned; and
• procedures by which the results of the study would be processed, analyzed, and
interpreted.
The objectives, protocols for sampling and analysis, quality assurance/quality control
(QA/QC) plans, and data processing procedures were nearly identical for all three studies.
Elements that differed among the three studies were the hypotheses and the array of
treatment groups. The hypotheses differed only slightly, as seen from the following
statements.
September 1, 1995
1-3
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
The central hypothesis of the USLADP is:
A reduction of lead in residential soil accessible to children will
result in a decrease in their blood lead levels.
The formal statement of the Boston hypothesis is:
A significant reduction (equal to or greater than 1,000 ug/g) of lead
in soil accessible to children will result in a mean decrease of at
least 3 fig/dL in the blood lead levels of children living in areas with
multiple possible sources of lead exposure and a high incidence of
lead poisoning.
The Baltimore hypothesis, stated in the null form, is:
A significant reduction of lead (*> 1,000 pg/g) in residential soil
accessible to children will not result in a significant decrease
(3 to 6 ng/dL) in their blood lead levels.
The Cincinnati hypothesis was separated into two parts:
(1) A reduction of lead in residential soil accessible to children will result
in a decrease in their blood lead levels.
(2) Interior dust abatement, when carried out in conjunction with exterior
dust and soil abatement, would result in a greater reduction in blood
lead than would be obtained with interior dust abatement alone, or
exterior dust and soil abatement alone.
Secondary hypotheses in the Cincinnati study are:
(3) A reduction of lead in residential soil accessible to children will result
in a decrease in their hand lead levels.
(4) Interior dust abatement, when carried out in conjunction with exterior
dust and soil abatement, would result in a greater reduction in hand
lead than would be obtained with interior dust abatement alone, or
exterior dust and soil abatement alone.
The array of treatment groups differed considerably among the three studies
(Table 1-1). In each study, the treatment groups had several features in common. The
groups were taken from demographically similar neighborhoods. All groups had some prior
evidence of elevated lead exposure, usually a greater than average number of public health
reports of lead poisoning. Three phases were employed in each study: preabatement
September 1, 1995
1-4
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 1-1. DESCRIPTION OF STUDY GROUPS AND TYPES OF
INTERVENTION
Treatment Group
Namea
Cross-Reference to
Individual Study
Report
Description of Treatment
BOS SPI
BOS PI
BOSP
BALSP
BAL P-Clb
BAL P-C2b
BAL P-C3b
CIN SEI
CIN I-SEC
CIN NT
Study Group
Control Group A
Control Group B
Study Area
Study Area Low
Control Area High
Control Area Low
Area A
AreaB
Area C
BOSTON
Soil and interior dust abatement, and
interior paint stabilization at beginning of
first year, no further treatment
Interior dust abatement and ulterior paint
stabilization at beginning of first year
Interior paint stabilization at beginning of
first year
BALTIMORE
Soil abatement and exterior paint
stabilization at beginning of first year, no
further treatment
Exterior paint stabilization at beginning of
first year, no further treatment because soil
lead not above cutoff level
Exterior paint stabilization at beginning of
first year, no further treatment
Exterior paint stabilization at beginning of
first year, no further treatment
CINCINNATI
Soil, exterior dust, and interior dust
abatement at beginning of first year, no
further treatment
Interior dust abatement at beginning of first
year, soil and exterior dust abatement at
beginning of second year, no further
treatment
No treatment, soil and interior dust
abatement at end of study
The treatment group designation indicates the location of the study (BOS = Boston, BAL = Baltimore,
CIN = Cincinnati), the type of treatment (S = soil abatement, E = exterior dust abatement, I = interior dust
abatement, P = loose paint stabilization, NT = no treatment).
Treated as one group in the Baltimore report, analyzed separately in this report.
Treated as one group in the Cincinnati report, analyzed as individual neighborhoods in this report.
September 1, 1995
1-5
DRAFT-DO NOT QUOTE OR CITE
-------
1 baseline phase for 3 to 18 mo; abatement or intervention (except for controls) phase, and
2 postabatement follow-up for 10 to 23 mo.
3
4 1.1.2 Study Design and Conduct
5 Table 1-1 describes the study groups and the forms of intervention employed in each of
6 the three cities. The Cincinnati study design used intervention on the neighborhood scale,
7 where the soil in parks, play areas and other common grounds were abated, and paved
8 surfaces in the neighborhood were cleaned of exterior dust. In Boston and Baltimore, only
9 soil on individual properties was abated. Table 1-2 shows the number of subjects
10 participating in different phases of the three studies in relation to the respective participant
11 groups for each city. The general characteristics are that soil lead concentrations are
12 typically high in Boston, where it is also common to find lead hi both exterior and interior
13 paint, as well as in drinking water. In the Boston areas studied, housing is typically single
14 and multi-family units with relatively large lot sizes. In the Baltimore neighborhoods, the
15 houses were mixed single and multifamily, and the lots were smaller than Boston lots, with
16 typical yards less than 100 m2. Nearly every house had lead-based paint. Residential units
17 in Cincinnati were mostly multifamily with little or no soil on the residential parcel of land.
18
19 1.1.3 Intervention Procedures
20 Figure 1-1 illustrates the generalized concept of human exposure to lead, showing the
21 pathways of lead from the several sources in the human environment to four compartments
22 immediately proximal to the individual. In the past decade, dramatic reductions hi exposure
23 to lead in air and food have occurred as a result of regulatory and voluntary programs to
24 reduce lead in gasoline and canned food. Figure 1-2 expands the critical dust route to show
25 the complexity of the many routes of dust exposure for the typical child. The strategies for
26 intervention used in this project were designed to interrupt the movement of lead along one
27 or more of these dust pathways.
28 There were three forms of intervention hi this project: (1) soil abatement, (2) dust
29 removal, and (3) paint stabilization. Soil abatement was by excavation and removal. Dust
30 intervention was by vacuuming, wet mopping, and, in some cases, replacement of rugs and
31 upholstered furniture. Cincinnati and Boston performed interior dust abatement, and
September 1, 1995
1-6
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 1-2. NUMBER OF PROJECT PARTICIPANTS BY ROUND3
Study
BOSTON
Middate
Children6
Famlies0
Properties'1
BALTIMORE
Middate
Children1"
Families0
Properties'1
Round 1
10/17/89 '
150
125
100
Round 1
10/25/88
408
290
260
Round 3
4/9/90
146
121
96
Round 2
4/1/89
322
226
207
Round 4
9/12/90
147
122
97
Round 3
.
2/17/90
269
181
160
Round 5
7/20/91
92
77
67
Round 4 Round 5
1/27/91 6/7/91
200 196
133 128
117 114
Round 6
9/3/91
187
126
112
CINCINNATI
Middate
Children"
Families0
Properties'5
Round 1
7/6/89
201
71
141
Round 3
11/14/89
185
67
129
Round 4
7/1/90
219
66
124
Round 6
11/17/90
198
94
124
Round 7
6/16/91
169
82
124
"Number shown is based on samples taken and does not include individuals enrolled but not sampled.
Intervention is shown by the vertical dashed lines.
bBased on number of children sampled for blood. Some children may not have been included in the statistical
analyses.
°Based on number of households sampled for dust.
dBased on number of properties (Boston, Baltimore) or soil parcels (Cincinnati) sampled.
1
2
3
4
5
Cincinnati also removed neighborhood exterior dust with mechanical sweepers and hand
tools. Dust intervention was not expected to be permanent, because dust continually moves
through the human environment. Instead, the removal of dust with elevated lead
concentrations was to expedite the impact of soil abatement on the child's environment.
September 1, 1995
1-7
DRAFT-DO NOT QUOTE OR CITE
-------
Auto
Emissions
industrlaJ
I Emissions |
Crustal
Weathering
Figure 1-1. Generalized concept of the sources and pathways of lead exposure hi
humans.
Atmospheric
Particles
/
Soil
Exterior Paint
Dust
Local
Fugitive
Dust
^
'
Exterior
Dust
Interior
Dust
«
Secondary^)
Occupational
Dust }
Figure 1-2. Typical pathways of childhood exposure to lead in dust.
September 1, 1995
1-8
DRAFT-DO NOT QUOTE OR CITE
-------
1 In the home, house dust is a mixture of street dust and soil, interior and exterior paint
2 dust, workplace dust carried home by adults, and dust generated from human activities within
3 the household. It is believed that most of the mass of the interior dust originates from soil
4 immediately exterior to the home, but this can vary greatly by the types of family activities
5 and by neighborhood characteristics. Nevertheless, in the absence of lead-based paint inside
6 the home, it would seem reasonable to assume that most of the lead hi household dust comes
7 from soil and other sources immediately outside the home.
8 Many of the Boston and Baltimore households selected for the project had chipping and
9 peeling lead-based paint, both interior and exterior. In order to reduce the impact of this
10 paint, the walls and other surfaces were scraped and smoothed, then repainted. It is
11 important to note that this approach in not a full scale paint abatement and was not designed
12 to permanently protect the child from lead-based paint. Paint stabilization was used on
13 ulterior surfaces in Boston, and on exterior surfaces in Baltimore. Paint stabilization was not
14 used in Cincinnati because the lead-based paint was believed to have been removed from
15 these homes in the early 1970s as part of a housing rehabilitation project.
16 In order to accurately measure the effectiveness and persistency achieved by soil
17 abatement and the impact of this abatement on reducing lead exposure for children, .the
18 sampling and analysis plans for soil and dust required robust quality control and quality
19 assurance objectives. Protocols were developed to define sampling schemes that characterize
20 the expected exposure to soil for children; collect, transfer, and store samples without
21 contamination; and analyze soil, dust, handwipe, and blood samples in a manner that would
22 maximize interlaboratory comparison. The original design focussed on sampling blood lead
23 during the late summer, as it was known that the seasonal blood lead cycle peaks during this
24 tune. Where this schedule could not be adhered to, an effort was made to schedule the
25 follow-up blood lead sampling at a comparable time in the cycle.
26 Information on area treated and volume of soil removed from each of the three cities
27 properties appears in Table 1-3. A total of 35 Boston properties were abated during the
28 study. In Baltimore, 63 properties in the BAL SP treatment group (see Table 1-3) were
29 abated between August and November 1990. An additional seven properties that did not
30 meet the requirements for abatement were transferred to a control group. Unpaved surfaces
31
September 1, 1995
1-9
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 1-3. SOIL ABATEMENT STATISTICS FOR THE THREE STUDIES
Number of properties3
Surface area (m2)
Volume soil removed (m3)
Surface area/property (m2)
Volume soil/property (m3)
Boston
35
7,198
1,212
200
34
Baltimore
63
4,100"
690
73
11"
Cincinnati
171
12,089
1,813
71
11
"Includes only properties abated during the study. Properties abated at the end of the study, where no further
sampling was reported, are not included in this analysis, but are included in the individual study reports.
In Cincinnati, a property is the location of the soil abatement, not the location of the child's residence.
bSurface area not provided by Baltimore report. This was calculated using Boston volume-to-surface ratio,
which is equivalent to an average removal depth of 17 cm.
1 were divided into areas on each property (usually front, back, and one side) and any area
2 with the maximum soil lead concentration above 500 /xg/g was abated entirely.
3 Within each of six neighborhoods, the Cincinnati study identified all sites with soil
4 cover as discrete study sites. The decision to abate was based on soil lead concentrations for
5 each parcel of land, and for the depth to which the lead had penetrated. Lead was measured
6 at two depths, the top 2 cm and from 13 to 15 cm. If the average concentration of the top
7 and bottom samples was greater than or equal to 500 ^g/g, the soil was removed and
8 replaced. If the average of the top samples exceeded 500 /xg/g, but the average of the
9 bottom samples was less than 500 /xg/g, the soil was also abated. Ground cover was
10 reestablished on abated soils and some unabated soils according to protocols described in the
11 Cincinnati report.
12 Exterior dust abatement was performed in the Cincinnati study only. The approach to
13 this abatement was to clean all types of hard surfaces where dust might collect, using vacuum
14 equipment that they tested and found to remove about 95% of the available dust on the area.
15 The dust surface categories were streets, alleys, sidewalks, parking lots, steps, and porches.
16 Dust measurements were made in a manner that determined the lead concentration
17 (micrograms of lead per gram of dust), the dust loading (milligrams of dust per square
18 meter), and the lead loading (micrograms of lead per square meter) for the surface measured.
19 This required that a dry vacuum sample be taken over a prescribed area, usually 0.25 to
September 1, 1995
1-10
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
0.50 m2. It is important to note that dust abatement is not expected to cause an immediate
change in the lead concentration on dust surfaces, only in the dust and lead loading.
Household dust was abated in the Boston and Cincinnati studies, but not in Baltimore.
The BOS SPI and CIN SEI groups (see Table 1-1) received interior dust abatement at the
same time as soil abatement, the BOS PI group received interior dust abatement without soil
abatement, and the three CIN I-SE neighborhoods received interior dust abatement in the
first year, followed by soil and exterior dust abatement in the second year.
In Boston, interior dust abatement was performed after loose paint stabilization. Hard
surfaces (floors, woodwork, window wells, and some furniture) were vacuumed, as were soft
surfaces such as rugs and upholstered furniture. Hard surfaces were also wiped following
vacuuming. Common entries and stairways outside the apartment were not abated.
The Cincinnati group performed interior dust abatement after exterior dust abatement.
Vacuuming was followed by wet wiping with a detergent. They vacuumed hard surfaces and
replaced one to three carpets and two items of upholstered furniture per housing unit. Their
previous studies had shown that these soft items could not be cleaned effectively with
vacuuming alone.
Most homes in the Cincinnati group had undergone extensive remodeling, believed to
have removed the lead-based paint 20 years prior to the project, but in Boston and Baltimore
lead-based paint occurred in nearly every home. Because full paint abatement was not within
the scope of this project, the alternative was to retard the rate of movement of lead from
painted surfaces to household dust to the extent possible. The interior surfaces of all Boston
homes and the exterior surfaces of all Baltimore homes received loose paint stabilization
approximately one week before soil abatement.
In Boston, loose paint stabilization consisted of removing chipping and peeling paint
and washing the surfaces. Window wells were painted with a fresh coat of primer.
Baltimore homes were wet scraped over the chipping and peeling surfaces, followed by
vacuuming. The entire surface was primed and painted with two coats of latex paint.
September 1, 1995
1-11
DRAFT-DO NOT QUOTE OR CITE
-------
1 1.2 SUMMARY OF INDIVIDUAL STUDY REPORTS
2 Following the completion of data collection and analyses, the research teams in each
3 city prepared individual study reports characterizing in detail the study design, procedures,
4 and results obtained in their respective cities. Some of the more reliant features of each
5 study and key findings reported by the individual city investigators are summarized next.
6
7 1.2.1 Boston Study
8 The Boston study retained 149 of the original 152 children enrolled, although
9 22 children moved to a new location while continuing in the study. Children with blood lead
10 concentrations below 7 ^g/dL or above 24 /zg/dL had been excluded from the study and two
11 children were dropped from some aspects of the data analysis when they developed lead
12 poisoning, probably due to exposure to lead-based paint abatement debris at a location
13 outside of their home.
14 Baseline characteristics (age, SES, soil lead, dust lead, drinking water lead, and paint
15 lead) were similar for the three study groups (BOS P, BOS PI, BOS SPI). The preabatement
16 blood lead concentration was higher for BOS P. The proportion of Hispanics was higher hi
17 BOS P than in BOS PI or BOS SPI, and the proportion of blacks was lower. There was a
18 larger proportion of male than female children in BOS P.
19 Data were analyzed by analysis of covariance (ANCOVA), which showed a significant
20 effect of intervention for both the BOS PI and BOS SPI groups. These results did not
21 change following adjustment for age, sex, SES, or any other variable except race and paint.
22 When the paint variable was controlled, the blood lead declines were diminished and the
23 results were borderline statistically significant. When the race variable was added, the blood
24 lead declines were also diminished and the results were not statistically significant.
25 Participants were chosen to be representative of the population of urban preschool
26 children who are at risk of lead exposure. The Boston Childhood Lead Poisoning Prevention
27 Program identified potential participants from neighborhoods with the highest rates of lead
28 poisoning. Because study candidates with blood lead levels below 7 /xg/dL or in excess of
29 24 /tg/dL at baseline were excluded from the study, no conclusion about the effect of abating
30 lead contaminated soil for children outside of this range can be made. Similarly, a different
31 effect might have been found for children who had a greater blood lead contribution from
September 1, 1995
1-12
DRAFT-DO NOT QUOTE OR CITE
-------
1 soil, such as in communities with smelters or other stationary sources where soil lead levels
2 are substantially higher than those seen in this study, or where differences in soil properties
3 result in differences in bioavailability.
4 Follow-up blood lead measurements were made in Boston 11 months after intervention
5 and again at 23 months.
6
7 1.2.2 Baltimore Study
8 The Baltimore study recruited 472 children, of whom 185 completed the study.
9 Of those that completed the study, none were excluded from analysis. The recruited children
10 were from two neighborhoods, originally intended to be a treatment and a control group.
11 Because soil concentrations were lower than expected, some properties in the treatment group
12 did not receive soil abatement. The Baltimore report transferred these properties to the
13 control group. In this report, the unabated properties in the treatment group are treated as a
14 separate control group.
15 Because of logistical problems, there was an extended delay between recruitment and
16 soil abatement that accounted for most of the attrition from the project. In their report, the
17 Baltimore group applied several statistical models to the two populations to evaluate the
18 potential bias from loss of participating children. These analyses showed that the two
19 populations remained virtually identical in demographic, biological and environmental
20 properties.
21 The Baltimore study provided limited information on the impact of house dust as a part
22 of the change in lead in the child's environment. The study design focused on changes in
23 biological parameters, hand dust and blood lead, over an extended period of time. There
24 were no measurements of exterior dust, no interior paint stabilization, and no interior dust
25 abatement. Except for the abated properties, there were no follow-up measurements of soil
26 lead concentrations.
27 Including the prestudy screening measurements of hand dust and blood lead in the
28 original cohort of participants, the Baltimore study made six rounds of biological
29 measurements that spanned 20 months, including postabatement measurements made at 2, 7,
30 and 10 months following abatement.
31
September 1, 1995
1-13
DRAFT-DO NOT QUOTE OR CITE
-------
1 1.2.3 Cincinnati Study
2 The Cincinnati study recruited 307 children, including 16 children born to participating
3 families during the study, and an additional 50 children who were recruited after the
4 beginning of the study. In their primary data analysis, the Cincinnati group excluded these
5 66 children who were recruited after the start of the study, plus 31 children who were living
6 in nonrehabilitated housing suspected of having lead-based paint, and four children (in two
7 families) who had become lead-poisoned from other causes. Thus, data for 206 children
v
8 were analyzed in the Cincinnati report and these 206 children were included in this integrated
9 report along with 7 of the 31 children living in nonrehabilitated housing. The remaining
10 24 were dropped because of insufficient follow-up data.
11 The Cincinnati study abated soil on 140 parcels of land scattered throughout six
12 neighborhoods. If soil were the only source of lead in the neighborhoods, exterior and
13 interior dust should have responded to the reduction in soil lead concentrations. However,
14 exterior dust lead loading decreased only slightly following both soil and dust abatement, and
15 returned to preabatement levels within one year. Corresponding changes in house dust, hand
16 lead, and blood lead that paralleled changes hi exterior dust. Interior dust returned to
17 preabatement levels about one year after abatement. Because blood lead concentrations also
18 decreased in the control area, the Cincinnati group concluded that there is no evidence for
19 the impact of soil and dust abatement on blood lead concentrations. However, this integrated
20 report concludes, through a more detailed structural equation analysis, that there is a strong
21 relationship between entry dust and interior dust in this subset of the Cincinnati study, where
22 the impact of lead-based paint was minimized.
23 Postabatement measurements in the Cincinnati were made at 2, 10, 14, and 21 months
24 following abatement in the first year, and at 3 and 10 months following abatement in the
25 second year.
26
27 1.2.4 Individual Study Conclusions
28 The Baltimore group stated their conclusions as follows:
29 • "Statistical analysis of the data from the Baltimore Lead in Soil Project provides no
30 evidence that the soil abatement has a direct impact on the blood lead level of
31 children in the study."
32
September 1, 1995
1-14
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
• "In the presence of lead-based paint in the children's homes, abatement of soil lead
alone provides no direct impact on the blood lead levels of children."
The basis for these statements consisted of an adjusted and unadjusted analysis of
selected covariates. The natural log of the blood lead of children in the treatment group
showed no significant difference from the natural log of the blood lead of children in the
control group, even when adjustments were made for age, SES, hand lead, season, dust, soil,
sex, weak mouthing behavior, or strong mouthing behavior. These analyses were made on
two sets of data. The first set consisted of all children enrolled in Rounds one and six. The
second group consisted only of children enrolled in all six rounds.
In their report following the first phase of their study, the Boston group stated their
conclusions as follows:
• "...this intervention study suggests that an average 1,856ppm reduction in soil lead
levels results in a 0.8-1.6 ft,g/dL reduction in the blood lead levels of urban children
•with multiple potential sources of exposure to lead."
Following the second phase of the study, they concluded (Aschengrau et al., 1994):
• "The combined results from both phases suggest that a soil lead reduction of
2,060 ppm is associated with a 2.2 to 2.70 fig/dL decline in blood lead levels. wl
The basis for their conclusions consisted of an analysis of variance comparing mean
blood lead changes among the three intervention groups, paired t-tests for within group
effects, and analysis of covariance with one-at-a-time adjustment for age, SES, race, sex,
paint, water, and mouthing behavior. The analysis of covariance was performed using no
transformation of blood lead data, which appeared to be normally distributed.
The Cincinnati conclusions can be paraphrased from their report as follows:
• Following interior and exterior dust and soil lead abatement, blood lead
concentrations decreased (in Area A) from 8.9 to 7.0 (21%) but increased to
8.7 jug/DL at 10 mo postabatement. Following interior dust abatement alone blood
lead concentrations decreased from 10.6 to 9.2 (13%) 4 mo postabatement and were
18% below preabatement 10 mo postabatement. With no abatement, blood lead
levels decreased by 29 and 6% during these same time periods. Other comparisons
also revealed no effects of the soil or dust abatement.
35
36
'This value for soil, 2,060 ppm, cited in their published report, was not adjusted by the Boston group with
the interlaboratory correction factor of 1.037 in Table 3-6.
September 1, 1995
1-15
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
• There was no evidence that blood lead levels were reduced by soil lead or dust
abatement in Area A (with soil, exterior dust, interior dust abatement). There was
a slight reduction (net reduction over control area) of 0.6 /tg/dL hi Area B that
might be attributed to interior dust abatement. This difference is not statistically
significant.
The basis for the Cincinnati conclusions was a comparison of geometric mean blood
lead concentrations in the three treatment groups between Rounds 1 and 4.
1.3 SUMMARY OF EPA INTEGRATED ASSESSMENT RESULTS AND
FINDINGS
The original data sets for each of the three participating cities were submitted to EPA,
along with the individual study reports alluded to above. Further analysis of the data were
conducted by EPA staff in ORD, specially in the Environmental Criteria and Assessment
Office/Research Triangle Park, NC (ECAO/RTP, now the National Center for
Environmental Assessment [NCEA-RTP]). The present intergrated report presents
information on the additional EPA statistical analyses and their results, as summarized here.
From the perspective of the child's environment, changes in the soil lead concentration
are expected to bring about changes in the house dust concentration, the hand dust, and the
blood lead concentration. In each of the three studies, the soil lead concentrations were
reduced to approximately 25 to 200 /*g/g in the study area, and for many treatment groups,
there was a reduction of group mean blood leads, although not always statistically significant.
1.3.1 Quality of the Data
In the absence of certified standards for soil and dust, it was necessary to implement a
program that would ensure that chemical analyses performed by the three participating
laboratories would be internally accurate and externally consistent with similar analyses by
other researchers. This program consisted of identifying acceptable analytical and
instrumental methods, establishing a set of soil and dust standards, and monitoring the
performance of the participating laboratories through an external audit program.
Because chemical extraction of an estimated 75,000 soil and dust samples per study
>fj
presented a costly burden on the project both in terms of tune and expense, and because of
September 1, 1995
1-16
DRAFT-DO NOT QUOTE OR CITE
-------
1 the advantage of nondestructive analysis for a project of this nature, the Scientific
2 Coordinating Panel recommended the use of laboratory scale X-ray fluoresence (XRF) for
3 soil analysis on the condition that a suitable set of common standards could be prepared for a
4 broad concentration range and that a rigorous audit program be established to ensure
5 continued analytical accuracy. Two groups, Boston and Baltimore, elected to use laboratory
6 XRF for interior dust analysis also, whereas Cincinnati opted for hot nitric acid extraction
7 with atomic absorption spectroscopy (AAS) for interior dust and XRF for exterior dust.
8 During the study, the Baltimore group recognized problems with analyzing dust by XRF
9 when the sample size was small, less than 100 mg. They reanalyzed the dust samples by
10 AAS and reported both measurements. In Boston, this problem was solved by compositing
11 the floor dust samples for XRF analysis, reporting one floor dust sample per housing unit.
12 During the project, there were two rounds of soil and dust interlaboratory calibration
13 exercises, one near the beginning and one at the completion of the soil and dust analyses.
14 These exercises, which involved the three participating laboratories and two additional
15 laboratories for each exercise, provided the basis for the evaluation of the performance of
16 each laboratory in the audit sample program, and for the conversion factors used to compare
17 soil and dust data between laboratories.
18 Each study maintained rigorous standards for database quality. These included double
19 entry, 100% visual confirmation, and standard procedures for detecting outliers. Some
20 errors were found during the preparation of this report and corrected prior to use hi this
21 report. None of these errors would have impacted the conclusions drawn by the individual
22 study.
23
24 1.3.2 Effectiveness and Persistency of Intervention
25 Soil abatement reduced soil concentrations in all three studies and there was no
26 evidence of soil recontamination in either Boston or Cincinnati. There were no follow-up
27 measures of soil in Baltimore that would detect recontamination. There was some evidence
28 for exterior dust recontamination in Cincinnati. The Cincinnati group suggests that this
29 might be caused by chipping and peeling lead-based paint from the exterior surfaces of
30 nearby buildings not included in the project.
September 1, 1995
1-17
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Interior dust abatement was persistent in both Boston and Cincinnati, even though
some recontamination occurred in Cincinnati in response to the exterior dust recontamination.
Paint stabilization appeared to have some impact on exposure, but there were no measures of
persistency.
1.3.3 EPA Integrated Report Results
This integrated assessment looks at the three individual studies collectively to
determine if a broad overview can be taken of the project results when each study is placed
in its correct perspective.
The key findings of this integrated assessment with regard to the Boston study are as
follows:
1. The median preabatement concentration of lead in soil was relatively high in
Boston, averaging about 2,400 /^g/g with few samples below 1,000 /-ig/g.
2. Abatement of the soil effectively reduced the median concentration of lead in the
soil to about 150 jug/g (an average decrease of about 2,300 /^g/g).
3. Soil was clearly a part of the exposure pathway to the child, contributing
significantly to house dust lead.
4. Other sources of lead, such as interior lead-based paint were minimized by
stabilization.
5. The reductions of lead-in both soil and house dust persisted for at least two years.
6. Blood lead levels were reduced by approximately 1.6 /xg/dL at 10 mo after soil lead
abatement.
7. Additional reductions hi blood lead of about 1.0 /xg/dL (relative to non-abated) were
observed at 22 mo postabatement for children hi houses where the soil lead was
abated and the interior house dust lead was consequently reduced and remained low.
Thus, in the Boston study, the abatement of soil resulted in a measureable, statistically
significant decline in blood lead concentrations in children, and this decline continued for at
least two years. It appears that the following conditions were present, and perhaps necessary
for this effect: (a) a notably elevated starting soil lead concentration (e.g., in excess of
1,000 to 2,000 jug/g); (b) a marked reduction of more than 1,000 jig/g in soil lead
September 1, 1995
1-18
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
consequent to soil abatement accompanied by (c) a parallel marked and persisting decrease in
house dust lead.
These conclusions are consistent with those reported by the Boston research team. This
integrated assessment found no basis for modifying their conclusions, although we choose not
to express these findings as a broadly generalizeable linear relationship between soil and
blood, such as change in micrograms of lead per deciliter of blood per change in micrograms
of lead per gram of soil, because we believe that such a linear expression of abatement
effects is highly site specific for the soil-to-blood relationship. We found evidence that the
dust-to-blood relationship is more significant and, perhaps, more linear than the soil-to-blood
relationship.
With regard to the Baltimore analyses conducted for this integrated assessment, the
participants in the abatement neighborhood that did not receive abatement were treated as a
separate control group, rather than combined with the nonabatement neighborhood (as the
Baltimore research team did). The reason for this was to establish a control group not
influenced by differences between neighborhoods. This alternative approach used in this
integrated assessment had little impact on the statistical significance of soil abatement effects
as reported by the Baltimore research team.
The key findings of this integrated assessment for Baltimore are:
1. The preabatement concentrations of lead in soil were notably lower (i.e., averaging
around 500 to 700 /-tg/g, with few over 1,000 jug/g) than in Boston.
2. The actual reduction of lead in soil by abatement was small (a change of about
400 fjig/g), compared to the Boston study (a change of about 2,300 /xg/g).
3. Measurements of blood lead were made for only ten months following abatement;
and no significant decreases in blood lead consequent to soil abatement were
observed compared to non-abatement control group children.
4. Except for exterior lead-based paint, there was no control of other sources of lead,
such as the stabilization of interior lead-based paint (as done in Boston) or
abatement of house dust (as done in Boston and Cincinnati).
5. Follow-up measurements of soil (except immediately postabatement) were not made
to establish the persistency of soil abatement, and its possible effects on house dust.
Thus, in Baltimore, where starting soil lead concentrations were much lower than in
Boston and soil abatement resulted in much smaller decreases in soil lead levels and no
September 1, 1995
1-19
DRAFT-DO NOT QUOTE OR CITE
-------
1 interior paint stabilization or dust abatement was performed, no detectable effects of soil lead
2 abatement on blood lead levels were found.
3 These conclusions are consistent with those reported by the Baltimore research group,
4 and are not inconsistent with those above for the Boston study. At soil concentratons much
5 lower than the Boston study, the Baltimore group would have likely been able to see only a
6 very modest change in blood lead concentrations (perhaps less than 0.2 /ig/dL) assuming
7 similarity between the study groups in Boston and Baltimore and the same linear relationship
8 between change in soil concentration and change in blood lead. Furthermore, the ulterior
9 paint stabilization and house dust abatement performed in Boston perhaps enhanced and
10 reinforced the impact of soil abatement on childhood blood lead, whereas in Baltimore, any
11 possible small impact of soil abatement would have likely been swamped by the large
12 reservoir of lead in the interior paint and the large unabated amounts of lead in interior house
13 dust.
14 As for the Cincinnati study, because of differences in the neighborhoods, we found that
15 combining neighborhoods into treatment groups often obscures important effects, and chose
16 to analyze each of the six Cincinnati neighborhoods as separate treatment groups. One
17 neighborhood, Back Street, had an insufficient number of participants and was dropped from
18 some analyses. The Back Street group started with nine families, but by Round 5 there was
19 only one participating family in the study. We also found that the two control
20 neighborhoods, Glencoe and Mohawk, were substantially different, and that the three
21 remaining treatment groups, Pendleton, Dandridge, and Findlay, were more comparable,
22 both demographically and in geographic proximity, to Mohawk than to Glencoe.
23 On this basis, we concluded that, in most cases, the effect of soil abatement could not
24 be clearly determined, and offer the following explanation for this conclusion:
25
26 1. Most of the soil parcels in each neighborhood were not adjacent to the living units,
27 and this soil was therefore not the primary source of lead in house dust. Evidence
28 for this statement includes the observation that street dust lead concentrations are
29 much higher than soil concentrations, indicating there is a large source of lead
30 contributing to street dust in addition to soil lead.
31
32 2. The preabatement median soil lead concentrations in the three treatment groups
33 were about 300 /ng/g in Pendleton, 700 /itg/g in Findlay, and 800 /*g/g in
34 Dandridge, and the postabatement soil concentrations were less than 100 jiig/g, so
35 that the reduction of lead in soil was small, as in Baltimore.
September 1, 1995
1-20
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
Evidence for the impact of dust abatement or dust and soil abatement consists of a
statistically significant difference between changes in blood lead between Rounds 1 and 4,
approximately one year apart. Some Cincinnati neighborhoods showed decreased blood lead
concentrations in response to dust abatement or dust and soil abatement. The two
neighborhoods that received only interior dust abatement in the first year, Dandridge and
Findlay, showed a small decrease in blood lead concentrations, compared to large increases
in the nearest control group, Mohawk. The treatment group that received soil, exterior dust
and interior dust abatement, Pendleton, showed a smaller effect than did the Dandridge and
Findlay neighborhoods. After consultation with the Cincinnati research team, we suspect
that there was recontamination of street dust in Pendleton during the study, probably caused
by demolition of nearby buildings in the neighborhood.
The consistent theme across the outcomes for all three studies is that soil abatement
must be both effective and persistent in markedly reducing soil lead concentrations
accompanied by a corresponding reduction in house dust lead in order to result in any
detectable reduction of blood lead. The location of the soil relative to the exposure
environment of the child is important. In this project, the movement of lead from soil or
street dust into the home seems to be a key factor in determining blood lead concentrations.
Although these USLADP results provide substantial evidence for the link between soil or
street dust and house dust lead, there is insufficient information by which to clearly quantify
this relationship in terms of the lowest level of soil or street dust lead reduction that will
yield a measurable decrease of lead in blood.
1.4 INTEGRATED PROJECT CONCLUSIONS
The main conclusions of this Integrated Report report are two-fold:
(1) When soil is a significant source of lead in the child's environment, the abatement
of that soil will result in a reduction in exposure that -will, under certain
conditions, cause a reduction in childhood blood lead concentrations.
(2) Although these conditions for a reduction in blood are not fully understood, it is
likely that four factors are important: (1) the past history of exposure of the child
to lead, as reflected in the preabatement blood lead; (2) the magnitude of the
reduction in soil lead concentrations; (3) the magnitude of other sources of lead
September 1, 1995
1-21
DRAFT-DO NOT QUOTE OR CITE
-------
1 exposure, relative to soil; and (4) a direct exposure pathway between soil and the
2 child.
3
4 The basis for the first conclusion is: in Boston, where the soil lead concentrations were
5 high and the contribution from lead-based paint was reduced by paint stabilization, there was
6 a measurable reduction of blood lead concentrations. This reduction continued to increase
7 for two years following abatement in Boston.
8 Conversely, in Baltimore and Cincinnati, where soil was not a significant source of lead
9 relative to other sources, there was no measurable reduction of blood lead except in cases
10 where those sources were also removed or abated. In Baltimore, these sources may have
11 been interior lead-based paint that was not stabilized, or house dust that was not abated.
12 In Cincinnati, the principle source of lead seemed to be neighborhood dust that may have
13 been contaminated with lead-based paint.
14 The basis for the second conclusion is: in those cases where all important elements of
15 the exposure pathway were available for assessment, the structural equation model analyses
16 showed that preabatement blood lead concentration was a major predictor of postabatement
17 blood lead, suggesting that the remobilization of bone lead is a major component of the
18 measured blood lead.
19 All other factors being equal, the measurable reduction in blood lead was observed only
20 at higher concentrations of soil lead. In the absence of information about other sources of
21 lead, no clear statement can be made about the possibility of smaller reductions in blood lead
22 at lower soil lead concentrations.
23 In spite of the recent successes in reducing exposure to lead by removing lead from
24 gasoline and canned food, lead exposure remains a complex issue. This integrated
25 assessment attempts to assess exposure to lead in soil and house dust. Lead in soil and
26 lead-based paint are closely linked in the child's environment. If there is exterior lead-based
27 paint, then soil lead is likely to be elevated with a consequent elevation in house dust lead.
28 If there is interior lead-based paint, then efforts to reduce the impact of soil lead on house
29 dust will be only partially effective. The maximum reduction in lead exposure will not be
30 achieved unless both paint and soil abatement are implemented.
31 There is evidence from all three studies that lead moves through the child's
32 environment. This means that lead in soil contributes to lead in street or playground dust,
September 1, 1995
1-22
DRAFT-DO NOT QUOTE OR CITE
-------
1 lead in exterior paint contributes to lead in soil, and lead in street dust contributes to lead in
2 house dust. A more detailed analysis of the data may show the relative contribution from
3 two or more sources, but the present analyses imply that this transfer takes place.
4 The analysis of the data from the three studies showed evidence that blood lead
5 responds to changes in house dust lead. There is also evidence for the continued impact of
6 other, independent sources following abatement of one source. This means that abatement of
7 soil or exterior paint does not necessarily reduce the contribution of lead from other sources
8 such as interior lead-based paint.
9 The conclusions of this report suggest that soil abatement alone will have little or no
10 effect on reducing exposure to lead unless there is a substantial amount of lead in soil and
11 unless this soil lead is the primary source of lead in house dust. At a minimum, when
12 implemented, both soil abatement and interior dust removal should both be performed to be
13 fully effective. Conversely, soil abatement should be considered in conjunction with paint
14 abatement when it is likely that soil will otherwise continue to contaminate house dust after a
15 paint abatement is completed.
16 From one perspective, decisions about soil abatement should be made on an individual
17 home basis. For an individual home, the owner or renter needs to know that the property is
18 safe for children. This report shows that, on an individual house basis, soil abatement may
19 reduce the movement of lead into the home and its incorporation into house dust. The
20 magnitude of this reduction depends on the concentration of lead in the soil, the amount of
21 soil-derived dust that moves into the home, the frequency of cleaning hi the home and the
22 cleanability of the home. The number and ages of children and the presence of
23 indoor/outdoor pets are factors known to increase this rate of dust movement, whereas
24 frequent cleaning with an effective vacuum cleaner, use of entry dust mats, and removing
25 shoes at the door serve to reduce the impact of soil lead on house dust.
26 From another perspective, soil abatement at the neighborhood level poses problems not
27 pertinent to individual homes. Playground, vacant lot, and other plots of soil may pose an
28 immediate problem if they are accessible to children and there is a direct pathway for dust
29 generated by this soil to enter the home. Likewise, sources of lead other than soil may
30 contribute more to exterior dust than soil itself. The evidence in this report suggests that the
31 key to reducing lead exposure at the neighborhood level is to abate significant sources of lead
September 1, 1995
1-23
DRAFT-DO NOT QUOTE OR CITE
-------
1 contributing to exterior dust, in addition to the soil and paint abatement that would be
2 performed on an individual property.
September 1, 1995
1-24
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
2. BACKGROUND AND OVERVIEW OF PROJECT
2.1 PROJECT BACKGROUND
2.1.1 The Urban Lead Problem
Children are exposed to lead through complex pathways from multiple sources. In the
mid 1980s, attention to sources of childhood lead exposure focused on urban environments
with high concentrations of lead in soil, where there was an apparent correlation with the
incidence of high blood lead concentrations. At that time, there were several other sources
of exposure that could potentially account for unusually high blood lead in a population of
urban children. Among these were lead in the air (primarily from automobile emissions),
lead in food (primarily from canned foods with lead soldered side seams), lead in drinking
water (primarily from lead pipes or newly soldered copper pipes), and lead hi paint. The
lead hi the soil was believed to be a mixture of lead from the atmosphere and lead from
exterior paint. Regulations were in place that would largely remove lead from gasoline by
the end of 1986, and there was a voluntary program among food processors to phase out
cans with lead soldered side seams. Renewed public interest in paint abatement emerged hi
the late 1980's concurrent with the start of this project.
Soil abatement had been performed in many nonurban residential areas with elevated
soil lead. The decision to abate soil was usually based in part on the distribution of blood
lead within the population of children. There was limited experience on the effectiveness of
this abatement and little or no opportunity for follow-up studies of the results. There were
little data from controlled evaluations because the intent of abatement was remediation, not
experimentation.
2.1.2 Legislative Background
In the mid 1980s, the scientific evidence for a correlation between soil lead and blood
lead was sufficient to warrant concern for the health of children, but not strong enough to
support a large scale program for soil lead abatement. Consequently, the Urban Soil Lead
Abatement Demonstration Project (USLADP), known also as the Three City Study, was
September 1, 1995
2-1
DRAFT-DO NOT QUOTE OR CITE
-------
1 authorized in 1986 under Section lll(b)(6) of the Superfund Amendments and
2 Reauthorization Act (SARA).
3 SARA called for EPA to conduct a "pilot program for the removal, decontamination, or
4 other actions with respect to lead-contaminated soil in one to three different metropolitan
5 areas."
6 Although not specified in the amendment, the legislative history focused on lead-based
7 paint as the source of lead in soil in urban residential areas. In response to the Superfund
8 mandate, USLADP was designed to evaluate the effectiveness of removal of lead-
9 contaminated soil in urban residential areas as a means to reduce blood lead levels of young,
i
10 preschool children residing in abated residences or neighborhoods. It did not attempt to
11 evaluate the relative effectiveness of different soil abatement technologies per se, but rather
12 focussed on determining the extent to which the blood lead levels of children less than six
13 years old ( as a key risk group for lead health effects) could be reduced by intervention to
14 decrease soil lead concentrations.
15 The EPA's Office of Solid Waste and Emergency Response (OSWER) had lead
16 responsibility for overall implementation of the project, as a Superfund-mandated activity.
17 Administrative and financial management responsibilities, it was decided, were to be
18 delegated to EPA regional offices for the geographic areas containing those cities selected for
19 inclusion in the project. EPA's Office of Research and Development was asked to provide
20 technical oversight and coordination assistance to help integrate scientific activities across the
21 cities selected. An EPA Steering Committee was set up to oversee site selection and
22 initiation of the project.
23 In 1987, EPA convened a set of experts to advise on the design of the project and to
24 develop selection criteria for study sites. Six cities submitted proposals, and Boston,
25 Baltimore, and Cincinnati were chosen by the following site selection process.
26
27 2.1.3 Site Selection
28 The three cities were selected based on an evaluation of each proposal in relationship to
29 the following site selection criteria, as recommended by the experts.
30 A. To be considered for selection, a metropolitan area must have:
31
September 1, 1995
2-2
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
1. Agreement by the appropriate EPA regional office to provide general project
oversight, and to disburse the funds.
2. An established entity, preferably the state, documented as willing to be responsible
for removing and disposing of lead contaminated soil. This included identification
of an appropriate facility within the state for disposal of the soil, facilitation of
permits, community relations and education, and any other activities necessary to
expeditiously provide for safe disposal.
3. The administrative infrastructure to carry out a large scale project. This included a
key government department with appropriate authority to coordinate the project,
and generally included active participation by the state, by community groups, and
by all the different metropolitan departments with some responsibility for the
project.
4. Access to scientific and medical expertise to ensure that sampling and analysis were
properly conducted, and access to medical care needed for any children found to
have lead toxicity.
5. Evidence that there are children with elevated blood lead levels (25 /*g/dL as
defined by the CDC in its 1985 childhood lead screening guidelines), and soil in
residential areas with lead levels of 1,500 jug/g or greater.1 It would be desirable
for lead-based paint to be established as a major contributor to the soil lead levels.
B. To be considered for selection, a metropolitan area should have:
6. A documented high incidence of children with elevated blood lead levels in the
proposed study areas. This meant that the municipality supported an active
childhood lead screening program.
7. A pattern of high density population in study areas. The number of children
available for evaluation as part of the project was important to the statistical
validity of the study.
8. Availability of other sources of funding for portions of the project not funded by
SARA. Such items might include de-leading the outside of houses, or intensive
interior vacuuming to remove residual leaded dust.
The Steering Committee reviewed proposals from six metropolitan areas: Boston,
Baltimore, Cincinnati, Minneapolis, Detroit, and East St. Louis. These were reviewed on
40
41
42
43
44
1 Note that the stipulated soil value of 1,500 /^g/g was interpreted as a significant number of soil parcels in
which at least one soil measurement exceeded this value. Reports in this document of means or median values
below 1,500 /zg/g for individual soil parcels or entire treatment groups should not be misinterpreted as failure
to meet the original selection criteria.
September 1, 1995
2-3
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
December 3 and 4, 1987, by the Steering Committee and the set of expert consultants.
Boston, Baltimore, and Cincinnati were selected based on the following key points:
1. The Boston investigators proposed to select three groups of families randomly from
several neighborhoods known to have soil lead concentrations hi the range of 2000
to 5000 /Ltg/g. One of these groups would receive only paint stabilization; a second
group would receive paint stabilization and dust abatement, and the third group
would receive soil abatement, dust abatement, and paint stabilization.
2. The Boston proposal involved collaboration among Boston City Hospital, Boston
University, and the EPA Region I Laboratory (for conduct of analysis of lead in
soil, dust, etc.). This collaborative group also had demonstrated experience with
collection, analysis, and assessment of soil and blood lead data in inner city
neighborhoods of Boston.
3. Cincinnati proposed a neighborhood level abatement study where housing units had
been previously gutted and rehabilitated approximately 20 years ago, and were
thought to be free of lead-based paint. The Cincinnati sites contained soil lead
from. 220 to 900 /ig/g, exterior surface dust (primarily from paved areas) from
2,000 to 5,000 /xg/g, and a number of children with blood lead concentrations
above 25 /zg/dL.
4. The Cincinnati proposal was prepared by the University of Cincinnati and
demonstrated a high degree of organizational infrastructure, with commitments
from the City of Cincinnati. There was an established infrastructure of
neighborhood associations that was perceived to be a plus for the project.
5. The Baltimore project proposed individual housing units with soil lead
concentrations in excess of 1,000 /zg/g. Lead-based paint had been abated in some,
but not all houses.
6. The Baltimore proposal was prepared by the State of Maryland and showed a
satisfactory level of organizational infrastructure and local scientific expertise;
problems with the proposed statistical approach were resolved by consultation with
the Steering Committee.
With the selection of Boston, Cincinnati, and Baltimore, a Scientific Coordinating
Committee was established to provide scientific and technical support for the three studies
and to coordinate the exchange of scientific information. This committee was composed of
representatives from the research teams of each of the three cities, the three EPA regional
offices (Regions I, III, and V), the Office of Solid Waste and Emergency Response, the
Environmental Criteria and Assessment Office/Research Triangle Park, NC (now the
National Center for Environmental Assessment/RTF), and the Centers for Disease Control
September 1, 1995
2-4
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
and Prevention. The task of organizing, scheduling, and conduct of meetings of the
Scientific Coordinating Committee was assigned to ECAO/RTP. Major policy decisions
remained with the Steering Committee.
The funding mechanisms were set into place individually through the respective EPA
regional offices (Regions I, III, and V). Each of these regional offices set up an independent
funding mechanism and oversight plan. The regional project officer became the liaison to
the Steering Committee and to the Scientific Coordinating Committee. Each city submitted a
work plan, which included the project description, organization, operation plan, and
reporting mechanisms, and the Quality Assurance (QA) plan. These work plans required
more than one year to complete and acquire Regional approval. In the meantime, the
projects were staffed and made operational. Community relations programs were initiated
that began the process of recruiting the study participants. Coordination between the three
cities was accomplished through a series of workshops, organized and convened by
ECAO/RTP, approximately three per year.
This integrated assessment includes a review of the hypotheses and study designs of the
individual studies (Chapter 2), a report of the methods intercomparison and quality
assurance/quality control program (Chapter 3), a summary of the individual study results and
conclusions reported by the three cities (Chapter 4), a description and explanation of the
statistical procedures performed as part of this EPA integrated assessment and the results of
these procedures (Chapter 5), and a summary of key findings and conclusions derived from
this assessment (Chapter 6).
2.2 INTEGRATION OF THE THREE STUDIES
2.2.1 Study Hypotheses
To place this project in perspective, it is helpful to look at the similarities and
differences among the three studies. They are similar in that their hypotheses and study
designs were drawn from the same general hypothesis, namely, that removing lead from soil
will reduce lead exposure.
September 1, 1995
2-5
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
The central hypothesis of the USLADP is
A reduction of lead in residential soil accessible to children will
result in a decrease in their blood lead levels.
Each study chose to develop a specific hypothesis that could be tested by data and
observations from their own study design. The formal statement of the Boston hypothesis is
A significant reduction (equal to or greater than 1,000 \ig/g) of lead
in soil accessible to children will result in a mean decrease of at
least 3 pg/dL in the blood lead levels of children living in areas with
multiple possible sources of lead exposure and a high incidence of
lead poisoning.
The Baltimore hypothesis, stated in the null form, is
A significant reduction of lead (> 1,000 pg/g) in residential soil
accessible to children will not result iifia*significant decrease (3 to
6 ng/dL) in their blood lead levels.
The Cincinnati hypothesis, separated into two'parts, is
(1) A reduction of lead in residential soil accessible to children will
result in a decrease in their blood lead levels.
(2) Interior dust abatement, when carried out in conjunction with
exterior dust and soil abatement, would result in a greater
reduction in blood lead than would be obtained with interior dust
abatement alone, or exterior dust and soil abatement alone.
Secondary hypotheses in the Cincinnati study are
(3) A reduction of lead in residential soil accessible to children will
result in a decrease in their hand lead levels.
••«?'•
(4) Interior dust abatement, when carried out in conjunction with
exterior dust and soil abatement, would result in a greater
reduction in hand lead than would be obtained with interior dust
abatement alone, or exterior dust and soil abatement alone.
2.2.2 General Study Design
The project objective was to measure the relationship between soil lead and blood lead.
This is an indirect relationship in the sense that children most commonly do not eat soil
September 1, 1995
2-6
DRAFT-DO NOT QUOTE OR CITE
-------
1 directly but usually ingest small amounts of dust derived, in part, from this soil. Likewise,
2 the lead in blood reflects not only recent exposure from all environmental sources, but the
3 remobilization of lead from bone tissue.
4 Each study was designed around the concept of participating families within a definable
5 neighborhood. There were a total of twelve neighborhoods in the project, six in Cincinnati,
6 four in Boston, and two in Baltimore. Except in Boston, these neighborhoods constituted the
7 treatment and control groups in the study. In Boston, families in the treatment group were
8 randomly assigned from volunteers from each of the four neighborhoods, as were families in
9 the control group. For each treatment group, there was a preabatement, abatement, and
10 postabatement phase. The immediate residential environment of the child was extensively
11 evaluated prior to and after abatement, through measurements of lead in soil, dust, drinking
12 water, and paint, and through interviews about activity patterns, eating habits, family
13 activities, and socioeconomic status. Parallel environmental and biological measurements, as
14 well as interviews, were taken in the control groups, but without abatement. The objective
15 of the preabatement phase was to achieve a clear understanding of the exposure history and
16 status (stability of the blood lead and environmental measures) prior to abatement. During
17 the abatement phase, attention was given to preventing any possible exposure that might
18 result from the abatement activities. During the postabatement phase, the project was
19 designed to determine the duration of the effect of soil abatement and to detect possible
20 recontamination.
21 The array of treatment groups differed considerably among the three studies. Each
22 treatment group, however, had several features in common. All groups were taken from one
23 to three demographically similar neighborhoods. All groups had some prior evidence of
24 elevated lead exposure, usually a greater than average number of public health reports of lead
25 poisoning. Each group received the same pattern of treatment: baseline phase for 3 to
26 18 months, intervention (except for controls), and follow-up for 12 to 24 months.
27 In each treatment group, even the controls, there was an attempt to minimize the impact
28 of chipping and peeling lead-based paint. In Boston, this was done by paint stabilization of
29 ulterior paint. In Baltimore, only exterior paint was stabilized. Therefore, in these two
30 studies, the effects of soil abatement should be evaluated in the context of some intervention
31 for lead-based paint. In Cincinnati, most of the living units may have been abated of lead-
September 1, 1995
2-7
DRAFT-DO NOT QUOTE OR CITE
-------
1 based paint more than 20 years before the start of the study. In the case of those that had
2 lead-based paint, the lead-based paint was measured but not treated prior to the study.
3 The Boston and Baltimore studies used a parallel intervention scheme, compared to the
4 staggered scheme used in Cincinnati. In other words, intervention in Boston (and Baltimore)
5 took place at the same time for all treatment groups, and the follow-up period was of the
6 same duration. But in Cincinnati, the soil and exterior dust intervention was delayed for
7 three neighborhoods, such that follow-up varied between 12 and 24 months. Throughout all
8 phases of each study, the timing of the blood lead measurements was planned according to a
9 seasonal cycle of blood lead levels that peaks in the late summer and according to an
10 age-related pattern that peaks at 18 to 24 months.
11 The complex nature of this project required measurement of exposure indices, such as
12 street dust, house dust, and hand dust, that are in the pathway between soil and blood. New
13 sampling and analysis protocols for these measurements, not generally available in the
14 scientific literature, were developed during the initial coordinating workshops.
15 The studies differ in several respects. The two pathways: (a) soil -* exterior dust and
16 (b) paint -> house dust differ slightly among the studies, as do the intervention strategies to
17 interrupt the flow of lead along these pathways. Collectively, these differences in study
18 design broaden the scope of the project to cover aspects of lead exposure intervention not
19 possible through the study of a single neighborhood or even a single city.
20
21 2.2.3 Study Groups
22 Variations in the nature and form of intervention were included in the study designs to
23 take advantage of the unique characteristics of the cities and their housing types. For
24 example, soil lead concentrations are typically high in Boston, where it is also common to
25 find elevated concentrations of lead in drinking water and in both exterior and ulterior paint.
26 In the areas studied, housing is typically multi-unit with some single family units with
27 relatively large soil cover in accompanying yards. In the Baltimore neighborhoods, nearly
28 every house had lead-based paint, the houses were mixed single and multifamily, and the soil
29 areas were smaller; typically less than one hundred square meters. On the other hand,
30 houses in Cincinnati were selected because they were thought to be relatively free of interior
31 lead-based paint, which might obscure the contribution of soil lead to house dust lead. As it
September 1, 1995
2-8
DRAFT-DO NOT QUOTE OR CITE
-------
1 happened, these neighborhoods were mostly multifamily housing with little or no soil on the
2 residential parcel of land. The Cincinnati study design used intervention on the
3 neighborhood scale, where the soil in parks, play areas, and other common grounds were
4 abated, and exterior dust on paved surfaces in the neighborhood removed.
5 Detailed information on study design and methods of analysis can be found in the
6 appended individual reports for each city, fable 2-1 summarizes the study design
7 characteristics for each of the three studies and their respective neighborhood groups. The
8 nonmenclature for these groups has been standardized for this report. With the exception of
9 the Cincinnati control group (CIN NT), all groups received some form of intervention during
10 the study.
11 For the purposes of consistency, certain descriptive terms that are used differently hi
12 the three individual study reports are standardized here and described in the glossary of this
13 document. One example is the use of the terms "study" and "project". In order to avoid
14 confusion, the term "study" refers to one of the three separate community studies, and the
15 terni "project" is used in reference to the three studies collectively. Similarly, the terms
16 "treatment group" and "control group" are generally preferred in this report as a "study
17 group".
18 The names that identify the individual treatment groups have been modified in this
19 report to assist the reader in remembering the type of intervention performed on each group.
20 Table 2-1 lists these names, with a brief description and the corresponding term in the report
21 of each separate study. This nomenclature identifies location of the study and the nature of
22 the intervention. For example, BOS SPI refers to the Boston group that received Soil, Paint,
23 and Interior dust intervention. A hyphen is used to indicate intervention in two different
24 rounds, as in CIN I-SE, where interior dust abatement took place about one year before soil
25 and exterior dust abatement. The reader may want to become familiar with this
26 nomenclature for the ten groups of participants in the project, as the data and results will be
27 presented using these designations without further explanation. One further note: The. BOS
28 PI, BOS P and CIN NT groups each received soil abatement at the end of the study.
29 Because no data were reported following this intervention, the designation "-S" was not used.
September 1, 1995
2-9
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 2-1. TREATMENT GROUP NOMENCLATURE WITH
CROSS-REFERENCE TO INDIVIDUAL REPORTS
Treatment Group
Name3
Cross-Reference to
Individual Study
Report
Description of Treatment
BOSTON
BOS SPI
BOS PI
BOSP
BALTIMORE
BALSP
BAL P-Clb
BAL P-C2b
BAL P-C3b
CINCINNATI
CIN SEI (P)
CIN NT (G,M)
Study Group
Control Group A
Control Group B
Study Area
Study Area Low
Control Area High
Control Area Low
Area A
CIN I-SE (B,D,F)C Area B
Area C
Soil and interior dust abatement, and interior
paint stabilization at beginning of first year, no
further treatment.
Interior dust abatement and interior paint
stabilization at beginning of first year.
Interior paint stabilization at beginning of first
year.
Soil abatement and exterior paint stabilization
at beginning of first year, no further treatment.
Exterior paint stabilization at beginning of first
year; because soil was not above cut off level,
no further treatment.
Exterior paint stabilization at beginning of first
year, no further treatment; soil above cut off
level.
Exterior paint stabilization at beginning of first
year; soil lead was not above cut off level; no
further treatment.
Soil, exterior dust, and interior dust abatement
at beginning of first year, no further treatment.
Includes only the Pendleton neighborhood.
Interior dust abatement at beginning of first
year, soil and exterior dust abatement at
beginning of second year, no further treatment.
Includes the Back St., Dandridge, and Findlay
neighborhoods.
No treatment; soil and interior dust abatement
following last sampling round. Includes the
Glencoe and Mohawk neighborhoods.
'The treatment group designation indicates the location of the study (BOS = Boston, BAL = Baltimore,
CIN = Cincinnati), the type of treatment (S = soil abatement, E = exterior dust abatement, I = interior dust
abatement, P = loose paint stabilization, NT = no treatment).
'Treated as one group in the Baltimore report, analyzed separately in this report.
Treated as one group in the Cincinnati report, analyzed as individual neighborhoods in this report.
September 1, 1995
2-10
DRAFT-DO NOT QUOTE OR CITE
-------
1 Other departures here from the terminology of the respective individual study reports
2 are conversion to a common system of units (metric where possible) and standard terms for
3 phases, stages, or rounds of the project. The term "round" refers to a distinct period of
4 time when one or more measurements were made. Other activities, such as soil abatement,
5 occurred between rounds. There is no consistent pattern for when abatement occurred (i.e.,
6 after Round 1, Round 3, etc.) for the different individual cities.
7 The numbers of participating children, families, and properties appear in Table 2-2.
8 Because of attrition and recruitment in Baltimore and Cincinnati, these numbers do not
9 accurately represent the number of participants present for the duration of the study. In this
10 report, subsets of these participants were statistically analyzed for specific purposes and to
11 meet specific statistical requirements, and these subsets may not be the same subsets used by
12 the individual study teams in their statistical analysis described in their respective individual
13 city reports.
14
15 2.2.4 Project Activity Schedule
16 The project activity schedule, shown in Figure 2-1, illustrates the major intervention
17 and measurement activities of the individual studies and the sequence and duration of these
18 activities. The frequency and timing of sampling relative to abatement and seasonal cycles
19 are important issues in the study design. These time lines are the actual occurrence of these
20 events and they differ somewhat from the planned schedule. The original design focused on
21 sampling blood lead during the late summer, as it was known that the seasonal cycle for
22 blood lead reaches a peak during this period.
23
24 2.2.5 Environmental and Biological Measurements of Exposure
25 Figure 2-2 illustrates the generalized concept of the pathways and sources of human
26 exposure to lead, showing the routes of lead from the several sources in the human
27 environment to four compartments (inhaled air, dusts, food, drinking water) proximal to the
28 individual. One of these proximal sources, dust, is the primary route of concern hi this
29 project. Figure 2-3 expands this dust route to show the complexity of the many routes of
30 dust exposure for the typical child. The intervention strategies used in this project were
31 designed to interrupt the movement of lead along one or more of these pathways.
September 1, 1995
2-11
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 2-2. NUMBER OF PROJECT PARTICIPANTS BY TREATMENT
GROUP AND ROUND3
BOSTON
Middate of round
Children6
Famlies0
Properties'1
BALTIMORE
Middate of round
Children11
Families11
Treatment Group
BOS SPI
BOS PI
BOSP
BOS SPI
BOS PI
BOSP
BOS SPI
BOS PI
BOSP
BALSP
BALP
BALSP
BALP
Rl
(PRE)
10/17/89
52
51
47
150
43
43
39
125
34
36
30
100
R3
(POST 1)
4/9/90
52
48
46
146
43
40
38
121
34
33
29
96
Rl R2
10/25/88 4/1/89
212 168
196 154
408 322
155 121
135 105
290 226
R4 R5
(POST 2) (Phase 2)
9/12/90 7/20/91
52 33
49 33
46 26
147 92
43 28
41 27
38 22
122 77
34 24
34 24
29 19
97 87
R3
2/17/90
154
115
269
103
•_7g
181
R4
1/27/91
112
88
200
76
57
133
R5
6/7/91
107
89
196
71
57
128
R6
9/3/91
104
83
187
71
55
126
Properties'1
CINCINNATI
Middate of round
Children11
Families0
Parcels'1
BALSP
BALP
CIN SEI (P)
CIN I-SE (B,D,F)
CIN NT (G,M)
CIN SEI (P)
CIN I-SE (B,D,F)
CIN NT (G,M)
CIN SEI (P)
CIN I-SE (B,D,F)
CIN NT (G,M)
141
119
260
Rl
(POD
7/6/89
54
86
61
201
31
58
40
129
55
74e
86
215
112
95
207
R3
(P03)
11/14/89
52
81
52
185
30
56
37
123
39
121e
85
245
91
69
160
R4
(P05)
7/1/90
46
92
81
219
31
56
35
122
39
121
85
245
66
51
117
R6
(P07)
11/17/90
37
87
74
198
31
74
63
168
40
119
84
243
63
51
114
R7
(P09)
6/16/91
31
77
61
169
30
60
52
142
40
121
84
245
62
50
112
Round designations (Rl, R2, etc.) are not the same as used in the Boston and Cincinnati study reports. Their round designations are
shown in parentheses. Some rounds are omitted from this table because blood lead data were not collected. Intervention, shown by the
dashed lines, occurred between Rl and R3 in Boston, R3 and R4 in Baltimore, Rl and R3 in the first year of the Cincinnati study, and R4
and R6 in the second year. Middates are the mean blood sampling dates.
Based on number of children sampled for blood.
Based on number of households sampled for dust.
Based on number of soil areas sampled.
Dandridge was added to the Cincinnati study after the soil sampling for Rl, but before the completion of all other Rl sampling. This
accounts for the sharp increase in the number of soil parcels between Rl and R3, with little change in the number of children or families.
September 1, 1995
2-12
DRAFT-DO NOT QUOTE OR CITE
-------
- Jul-Sep :Oct-Deo
; 1988
BOSTON :
Soil Sampled
Soil Abated
Dust Sampled
Dust Abated
Handwipes Collected
Blood Sampled
BALTIMORE
Soil Sampled :
Soil Abated '-
Dust Sampled :
1988
Handwipes Collected ; I — ™ — I
Blood Sampled : I "-1 1
Jan-Mar
1989
|
I
I
Apr-Jun
1989
i - 09
1
|
1
CINCINNATI i i I
Soil Sampled ; :
Soil Abated : : :
Exterior Dust Sampled '- '-_ '•
Exterior Dust Abated : : :
Interior Dust Sampled ; ; ;
Interior Dust Abated : :
R2
R1
H
Jul-Sep -Oct-Dao" Jan-Mar^ Apr-Jun
1989 : 1989 : 1990 : 1990
: : :
- j R2 " '
W I
I . I
I : I
1 i J I
| Hi
tvaH
1 R1 ! 1 ' H~i
J : i :
i : ' :
1 ~m i 1 " ra
1 J 1 .
1 '•*" : I 1
1 : : 1 r^
•H
ill
FM " " 1
; : 1
R1 1
1
I : i m i
1 : }•••"••]
1 ' 1 R? 1
1 : 1 "" 1
; ;
1 lB2l IB3I I
1 rn rn -
i — i i
I— I 1R2J |R3l -
i— i i
PH R M :
: M :
Handwipes Collected : : : : HH H H
Jul-Sep -Oct-Deo Jan-Mar:Apr-Jun: Jul-Sep :
1990 i 1990
H
i
i
n
I
I
R9
I
I
I
I
u
era
I
I
TtA
I
I
H-
H
; ;
1 R4 1 tod tag)
I - I rn rn
= H i
h«H H
1 H i
H»H H R
1991
l
I
H
H
i H i '-.
1991 : 1991
;
R
I
|BS|
R
H
j
R R]
IBS bet
n n
I
PH
-
:
:
|_az^
~
|a«|i |a^ |w| j : |H^ ; :
BloodSampled j : : : HJH : H : M: : M : : R : :
Figure 2-1. Project activity schedule showing the round designations and time periods
for sampling and interviewing, and the time periods for soil abatement.
Paint stabilization in Boston and Baltimore was performed during the soil
abatement period prior to any other intervention. Abatement in Cincinnati
that was performed after the final sampling round (as a courtesy to
participants) is not shown in this figure.
1 Exposure is the amount of a substance that comes into contact with an absorbing
2 surface over a specific period of time. In the case of lead, the absorbing surface can be the
3 gastrointestinal tract or the lungs. Exposure is measured in micrograms of lead per day.
4 Thus, an exposure of 10 /ig/day represents a total ingestion and inhalation of 10 micrograms
5 of lead from all sources; a fraction of this 10 micrograms would be absorbed into the body.
6 In this project, blood lead was used as an indicator of exposure, and reductions hi blood lead
7 concentrations were expected as a result of any combination of the interventions described
8 above. The units for blood are micrograms of lead per deciliter of blood and they are not
September 1, 1995
2-13
DRAFT-DO NOT QUOTE OR CITE
-------
Auto |
Emissions I
Crustal
Weathering
i_
Surface and
Ground Water
Paint,
Industrial
Dusts
Solder
Lead Glazes
Drinking
Water
Figure 2-2. Generalized concept of the sources and pathways of lead exposure in
humans.
1 compatible with the normal units of exposure, micrograms of lead per day. This illustrates
2 that lead hi one deciliter of blood reflects cumulative exposure for an unknown number of
3 days plus an unknown amount of lead mobilized from bone tissue. Other indicators of
4 potential exposure are hand lead and house dust. The amount of lead on the child's hands is
5 believed to be closely related to the child's blood lead and to the dust lead in the child's
6 environment.
7
8 2.2.5.1 Blood Lead
9 The amount of ingested lead that is actually absorbed in the gastrointestinal tract
10 depends in part on the bioavailability of the particular form of lead. The amount of absorbed
11 lead that reaches specific body tissues depends on the biokinetics of lead in the human body.
September 1, 1995
2-14
DRAFT-DO NOT QUOTE OR CITE
-------
Crustal |
Weathering I
Figure 2-3. Typical pathways of childhood exposure to lead in dust.
1 Blood tissue is in dynamic equilibrium with all other body tissues, including bone tissue,
2 where the lead is stored for longer periods of time. The relationship between blood lead and
3 the onset of health effects of lead, depends largely on the distribution of lead to the target
4 tissues, including the red blood cells themselves. Blood lead, then, is a convenient indicator
5 of both exposure and potential health risk to the child. This situation becomes important
6 when measuring the rate at which blood lead concentrations might decline following
7 abatement. For a child with lead stored in bone tissue following a long history of high lead
8 exposure, the decline in blood lead might be expected to be slower than for a child with low
9 previous exposure.
10
11 2.2.5.2 Hand Lead
12 Because blood lead reflects exposure to lead from all environmental sources, a second
13 exposure indicator, hand lead, was used to focus directly on the immediate pathway of dust
14 into the child. The units of measure are micrograms of lead per pair of hands, and like
15 blood lead, this measure does not reflect the rate at which lead moves into the body in units
September 1, 1995 2-15 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
of micrograms of lead per day. Instead, this hand dust is a measure of lead loading on the
hand. It is a measure of the "dirtiness" of the hand in the same sense that dust loading is a
measure of the dirtiness of the floor. Hand dust loading could possibly be converted to
micrograms of lead per day if there were a measure of the area of the hand mouthed by the
child and the frequency of hand to mouth activity during each day.
2.2.5.3 House Dust
House dust is a mixture of lead from many sources, including soil, street dust, interior
paint, and biological sources such as insects, pets, and humans. The units of measurement
are ptg Pb/g (lead concentration), jug Pb/m2 (lead loading), and mg dust/m2 (dust loading).
When expressed as micrograms of lead per gram, the measurement can be converted to an
exposure measurement by assuming a specific amount of dust ingested per day, usually about
100 mg/day for preschool children. Exposure to household dust then becomes micrograms
per day:
Pb Concentration X Ingestion = Exposure
V-gPb gdust = figPb
(2-1)
gdust
day day
In a similar manner, exposure to food, drinking water, and inhaled air can be expressed
as /tg/day, and these three sources, circa 1990, normally account for about 5, 1, and 0.1 /zg
Pb/day respectively. If the lead concentration in household dust is 200 jttg/g and dust
ingestion is 0.1 g/day, the exposure is 20 peg/day or much more than the other sources
combined. In this project, the maximum lead concentration in household dust was
107,000 Atg/g.
By a different calculation, childhood lead exposure may be expressed as a function of
dust lead loading. In this case, the ingestion parameter is in units of m2/day:
Pb Loading x Ingestion = Exposure
X
(2-2)
day day
September 1, 1995
2-16
DRAFT-DO NOT QUOTE OR CITE
-------
1 The ingestion parameter estimates the effective contact area for the child's hands (assuming
2 all dust is ingested by hand-to-mouth activity). Literature reports of childhood lead exposure
3 based on contact area are not known.
4
5 2.2.6 Intervention Strategies
6 Intervention is defined here as the interruption of the flow of lead along an exposure
7 pathway. Soil abatement is one form of intervention. If done correctly, this abatement
8 should establish an effective and persistent barrier to the movement of lead through the
9 child's exposure pathways. Other forms of intervention used in this project were exterior
10 dust abatement, interior dust abatement, and paint stabilization. Because dust is a very
11 mobile constituent of the human environment, exterior and interior dust abatement would not
12 be expected to form a permanent barrier to lead unless other sources of lead, such as soil,
13 were also abated. Likewise, the form of paint stabilization used in Boston and Baltimore,
14 where chipping and peeling paint was removed and the walls repainted, was not intended to
15 be permanent lead-based paint abatement.
16 The strategy for soil abatement was to remove all soil with concentrations above a
17 specific level (500 pig/g for Baltimore and Cincinnati, 1,000 /*g/g for Boston), and replace
18 this soil with clean soil in the range of 25 to 100 /ig/g lead concentration. This method,
19 called excavation and removal, was used in all three studies. In some cases, repair and
20 maintenance of ground cover was used where the soil concentrations did not warrant
21 excavation and removal.
22 To further interrupt the flow of lead along the exposure pathways, entire neighborhoods
23 in Cincinnati were cleaned of exterior dust using street cleaning vacuum equipment and hand
24 tools.
25 Interior house dust is believed to be a major direct lead exposure pathway for children.
26 Because household dust typically contains a mixture of lead from several sources (e.g., soil,
27 interior/exterior paint, air, etc.), abating house dust temporarily separates such sources from
28 the child's environment. Their recontamination of house dust and consequent impact on the
29 child's lead exposure can be evaluated by comprehensive measurements of the household dust
30 that include changes in lead concentration, lead loading, and dust loading. Understanding the
31 expected impact of abatement on these three parameters.is critical to interpreting the
September 1, 1995
2-17
DRAFT-DO NOT QUOTE OR CITE
-------
1 observed changes in blood lead concentrations. Following dust abatement, there should be
2 an immediate decrease in the dust loading, with no change in the lead concentration for those
3 groups that did not receive soil, exterior dust, or paint intervention. The rate at which this
4 dust loading returns to preabatement levels reflects the rate of movement of dust from other
i
5 sources into the home, the frequency of cleaning, and the "cleanability" of the home. (Many
6 inner city homes have surfaces that are cracked, pitted, or in disrepair and are difficult to
7 clean effectively.)
8 The effectiveness of both paint stabilization and soil and dust abatement can be
9 observed by changes in the lead concentrations of house dust. In the presence of lead-based
10 paint, the concentration of lead in house dust is expected to be greater than 1,500 to
11 2,000 jig/g, whereas without the influence of lead-based paint, the house dust is expected to
12 be comparable to external dust and soil (U.S. Environmental Protection Agency, 1986).
13 House dust is a mixture of dusts from many sources within and outside the home.
14 In the absence of lead-based paint inside the home, it would seem reasonable to assume that
15 most of the lead in household dust comes from soil and other sources external to the home.
16 Therefore, to enhance the impact of soil abatement, interior dust abatement was carried out
17 for some treatment groups in Boston and Cincinnati.
18 Many of the Boston and Baltimore households selected for the project had chipping and
19 peeling paint, both interior and exterior. In order to reduce the impact of lead-based paint,
20 the walls and other surfaces were scraped and smoothed, then repainted. It is important to
21 note that no attempt was made to remove all lead-based paint, nor to isolate intact paint from
22 the child. Paint stabilization was used on interior surfaces in Boston and on exterior surfaces
23 in Baltimore. Paint stabilization was not used in Cincinnati because most of the lead-based
24 paint was believed to have been removed from these homes in the early 1970s.
25
26
27 2.3 EXTERNAL FACTORS THAT COULD INFLUENCE PROJECT
28 RESULTS AND INTERPRETATION
29 The Scientific Coordinating Panel recognized that several extraneous factors might
30 influence the outcome of the project and that these factors were generally beyond the control
31 of the investigators. Among these are seasonal cycles and time trends of childhood blood
September 1, 1995
2-18
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
lead concentrations, unexplained or unexpected sources of lead in the children's homes or
neighborhoods, changes in public perception and avoidance of lead exposure hazards, and
movement of lead in soil either down the soil column or laterally with surface runoff or as
fugitive dust.
2.3.1 Cycles and Trends in Environmental Lead Concentrations
Figure 2-4 illustrates a pattern of childhood blood lead concentrations for Chicago
during the 1970s, showing a seasonal cycle and a downward trend throughout the decade.
The National Health Assessment and Nutrition Examination Survey II (NHANES II) data for
the entire country and all age groups (Figure 2-5) show a similar seasonal cycle and
downward trend during the last half of that decade. (Seasonal patterns from the
NHANES m data of 1988 through 1991 are not yet available.)
o
m
CD
50
40
30
20
10
1 I \ \
\\ \\
\ \ \ \ \ \
1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980
Year
Figure 2-4. Literature values for seasonal patterns for childhood blood lead (age 25 to
36 mo).
Source: U.S. Environmental Protection Agency (1986).
September 1, 1995
2-19
DRAFT-DO NOT QUOTE OR CITE
-------
25
20
1
2 15
10
I
WINTER 1976
(FEB.)
WINTER 1977
(FEB.)
WINTER 1978 FALL 1978 WINTER 1979
(FEB.) (OCT.) (FEB.)
I
WINTER 1980
(FEB.) _
_L
J_
_L
_L
10 15 20 25 30 35
Chronological Order (1 unit - 28 days)
40
45
50
55
Figure 2-5. Literature values for seasonal patterns for blood lead in children and adults
(NHANES II, age 6 mo to 74 years).
Source: Annest et al. (1983).
1 Investigators have known about this seasonal pattern for some time. Most
2 epidemiological studies are planned so that measurements can be taken at the peak of this
3 cycle, generally during the late summer. Studies of large numbers of children show a
4 sinusoidal pattern, even when the measurements do not include sequential measurements for
5 the same child. During the development of the study designs, it was apparent that
6 understanding of the seasonal cycles and temporal trends in blood lead would play an
7 important part in the interpretation of data collected over several years.
8 There is a question as to whether the seasonal cycle for blood lead concentrations is
9 caused by fluctuations in exposure or by physiological processes that regulate the biokinetic
10 distribution of lead within the body. Some investigators have attributed fluctuations in blood
11 lead concentrations to changing environmental lead concentrations or changing activity
12 patterns. During the late summer months, the child may eat food or dust with high lead
13 concentrations or ingest more dust during outdoor play. This project was designed to
14 measure changes in lead concentrations in soil and dust, but not changes in activity patterns.
15 The observations made on these fluctuations and the interpretation of these observations are
16 reported in Section 5.2.5.
September 1, 1995
2-20
DRAFT-DO NOT QUOTE OR CITE
-------
1 Although this project was designed to maximize the measurements of blood lead during
2 the late summer for each of the three studies, measurements were made during other times of
3 the year in order to observe changes immediately after abatement. These sequential
4 measurements show a similar cycle when all children are grouped together.
5 Two other patterns, long-term time trends and early childhood patterns dependent on
6 age, are applicable to this project. Little is known about age related patterns, but one study
7 in Cincinnati, prior to the project, showed a pattern of blood lead changes during early
8 childhood growth patterns (Figure 2-6).
9
10
^ 8
o>
o
cr
"O
en
.Q
DL
PbH
PbB
_L
u 0 1 2 3 4 56
Age (years)
Figure 2-6. Predicted differences in blood lead (PbB) and hand lead (PbH) during early
childhood, based on empirical data.
Source: Bornschein et al. (1988).
1 Long-term downward trends were documented for child blood lead concentrations
2 during the 1970s and 1980s and have been attributed to decreasing concentrations of lead in
3 food and air. Data for this project were analyzed for decreasing concentrations of lead in
4 soil or dust and the results are reported in Chapter 5. The QA/QC measures reported in
September 1, 1995
2-21
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
detail in Chapter 4 rule out the possibility of this trend being caused by a measurement
artifact such as analytical drift.
2.3.2 Unexplained and Unexpected Sources of Lead
Occasionally, measurements of environmental lead are higher than expected and
difficult to explain. Atmospheric deposition can be a reasonable explanation, because this
route can change much more abruptly than soil, dust, food or drinking water. This section
discusses the possibility that the observed fluctuation hi street dust and house dust can be
attributed to changes in air concentration alone. Because this project began after the national
phasedown of lead hi gasoline, the air concentrations of lead in these cities had decreased to
about 0.1 /ig/m3 by the start the project.2 The following is a theoretical calculation of the
amount of lead that could be transferred to soil or dust at this concentration and from this
source alone.
Atmospheric deposition during the project was assumed to be typical for air
concentrations that averaged 0.1 jig/m3 (1.0 X 10'7 /tg/cm3). At a deposition rate of
0.2 cm/s, this would accumulate 0.6 ^g/cm2-year at the soil surface. Assuming that this lead
would be retained in the upper 1 cm of soil surface (therefore 1 cm2 of soil surface equals
1 cm3 of soil), then the annual increment would be 0.6 /jg/cm3. Because 1 cm3 of soil
weighs about 2 g, the annual incremental increase in lead concentration would be
0.3 fig Pb/g soil, an insignificant annual contribution for soils that average several hundred
micrograms per gram. The calculation for annual deposition to a surface is
1 x 10-7
Pb
X 0.2 EL x 3.15 X 107
cm-
year
= 0.6
Pb
cm* year
(2-3)
For the accumulation of dust on hard surfaces, however, the same calculation indicates
a potentially greater influence of atmospheric lead. Converting to units of lead loading, the
0.6 jag/cm2-year becomes 6,000 /xg/m2-year, or 16 j*g/m2-day. Therefore, 0.1 jtig/m3 in air
concentration could account for a change of 16 /ig Pb/m2 per day in the dust lead loading to
2 The 1989 maximum quarterly average air lead concentration for the metropolitan statistical areas of Boston,
Baltimore, and Cincinnati were 0.08, 0.11, and 0.11 ^g/m3, respectively (U.S. Environmental Protection
Agency, 1991a).
September 1, 1995
2-22
DRAFT-DO NOT QUOTE OR CITE
-------
1 a surface. An accumulation of 160 )Ug/m2 over 10 days is in the range of the observed
2 changes in surface dust loading in this project.
3
4 2.3.3 Movement of Lead in Soil and Dust
5 There are several reasons why localized soil lead fluctuations might occur. Changes hi
6 soil lead concentration independent of intervention that might increase lead concentration are:
7 atmospheric deposition (relatively minor as discussed above), exterior paint chipping and
8 chalking, and human activity such as household waste dumping (motor oil, etc). Soil lead
9 concentrations might decrease if lead leaches downward into the lower soil horizon, or if
10 surface dust shifts by reentrainment. The downward leaching of lead through the soil profile
11 mass occurs at a very slow rate, approximately a few millimeters per decade (Grant et al.,
12 1990). The reentrainment of dust at the soil surface is usually in equilibrium with the local
13 environment, such that inputs would equal outputs by this pathway. This would not be the
14 case if there is flaking or peeling lead-based paint within the neighborhood or an industrial
15 source of fugitive dust in the vicinity of the neighborhood. A limited effort was made to
16 monitor and' control the impact of lead-based paint on soil concentrations. In Baltimore,
17 buildings with exterior lead-based paint were stabilized by removal of the chipping and
18 peeling paint, done in a manner to avoid contaminating the soil. In Boston, homes were
19 selected with less then 30% exterior chipping and peeling paint, by area. In Cincinnati,
20 neighborhoods with mostly rehabilitated houses were selected. There were no attempts hi
21 any of the studies to control the introduction of lead to the soil by human activity such as
22 household waste dumping.
23 Lead in household dust is a mixture of dust brought into the house from outside and
24 dust generated from within the home. Studies have shown that as much as 85% of the mass
25 of dust comes from outside the home and much of this is apparently brought hi on the feet of
26 children and pets (Roberts et al., 1991). Household dust lead concentrations are usually
27 similar to the soil concentration in the immediate vicinity of the house, unless there are
28 internal sources of lead, such as lead-based paint. Thus, changes in soil concentrations are
29 likely to be reflected by changes in household dust concentrations within a few days and
30 probably reach equilibrium within a few months, depending on the relative contribution from
September 1, 1995
2-23
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
soil and other sources, the frequency and efficiency of house cleaning, and the cleanability of
the house.
2.3.4 Other Factors
In the following chapters, this report discusses several issues that identify possible
limitations of the studies. This detailed assessment: (1) examines measurement methods used
and related QA/QC data to ascertain that adequate measures were taken to produce data of
good quality that can be compared across the three studies; (2) examines the study designs to
determine if the individual study groups are comparable within each study and if comparisons
are possible across the three studies; and (3) performs rigorous statistical analyses that
attempt to quantify differences between study groups and identify specific exposure factors
that may be responsible for the differences.
With respect to the QA/QC data, it should be noted that there are no estimates of
sampling reproducibility for any of the environmental or biological measurements. This
would have required collecting duplicate samples for a specified percentage of the samples.
In retrospect, the following observations are worth noting:
1. Duplicate soil samples would not have been informative unless the entire soil parcel
was sampled in duplicate. In this report, the reproducible number is the arithmetic
mean of all soil samples from the parcel;
2. Duplicate sampling of house dust would have identified reproducibility of lead
concentration, but probably not lead loading, which changes on a daily basis.
Duplicate sampling of house dust may also have impacted the child's environment if
a substantial amount of the targeted play areas were sampled.
Nevertheless, this report recognizes the limitations of statistical analysis due to the
absence of an estimate of sampling error.
There are several exposure-related factors other than those measured by environmental
sampling that must be taken into account during the statistical analyses. Among these are
seasonal patterns in weather (especially rainfall as it affects dust loading and mobility),
activity patterns (which affect indoor/outdoor play patterns), and possible physiological
growth cycles (which affect remobilization of lead from bone tissue). Age of the child may
also impact exposure by differences in activity patterns, body size, and parental supervision.
September 1, 1995
2-24
DRAFT-DO NOT QUOTE OR CITE
-------
1 For the most part, this report is only able to ascertain that all groups within a study were
2 impacted equally by these and other confounding factors during the study.
3
September 1, 1995
2-25
DRAFT-DO NOT QUOTE OR CITE
-------
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
3. METHODS INTERCOMPARISON AND QUALITY
ASSURANCE/QUALITY CONTROL
Specific details on measurement methodology employed in each study may be found hi
the appended individual reports. This chapter describes the initial evaluation of several
methods for soil, dust, hand wipe, and blood sampling and analysis that were considered by
the Scientific Coordinating Committee, and the basis for selection of these methods by the
participating research teams.
Soil sampling methodology was determined by agreement that a 2-cm core would be
taken according to a prescribed pattern about a randomly selected point, and that this point
would be selected based on the size and shape of the plot of soil. These procedures are
described hi the individual reports, and no further assessment was made here of the
representativeness of this sampling procedure.
Interior dust sampling methods were evaluated based on the desirability of dust load
information. This required that a dry sample be taken (as opposed to a wet wipe) hi order to
determine the mass of dust collected as a function of area (dust load). Although the sampling
devices differed, the basic protocol called for a vacuum pump that collected the dust sample
on a filter pad at a prescribed flow rate and using a prescribed pattern of moving the pump
nozzle over the sample area. No further attempt was made to calibrate the collection devices
between the individual studies.
Hand wipe samples were taken according to procedures developed by the Cincinnati
group in previous studies. Field blanks and lot blanks were determined by each group.
There were some differences hi the timing of the hand wipe sample as reported by the
individual study teams.
Blood samples were taken according to methods prescribed by CDC hi their blood lead
certification program. The analysis of blood for health indicators other than lead differed
among the three groups. Blood data other than lead concentration were not used hi this
integrated assessment.
The procedures and results of interlaboratory comparisons of analytical methodology
and the results of the QA/QC plan for the individual studies are described in the following
September 1, 1995
3-1
DRAFT-DO NOT QUOTE OR CITE
-------
1 sections. These procedures and their results were reviewed and evaluated throughout the
2 project at the scheduled workshops and during monthly teleconference calls.
3 The research team for each study prepared a sampling and analysis plan that included
4 rigorous QA/QC objectives. These plans included protocols that: defined sampling schemes
5 designed to characterize the expected exposure to soil for children; described how to collect,
6 transfer, and store samples without contamination; and described how to analyze samples
7 with the maximum degree of accuracy and precision. Throughout the project, several
8 intercalibration exercises were performed to guarantee that the analytical results for
9 measurements of soil, dust, handwipes, and blood would be accurate and that the data would
10 be comparable.
11
12
13 3.1 INTERCOMPARISON OF LABORATORY METHODS FOR SOIL
14 AND DUST MEASUREMENTS
15 The objective of the laboratory intercomparison and QA/QC program was to ensure that
16 the three studies could achieve a high standard of expertise in the analysis of soil and dust
17 samples, and that each of the three laboratories would be expected to get reasonably similar
18 results when analyzing the same soil sample. The framework for the intercomparison effort
19 was two round robin calibration exercises, one at the beginning and one near the end of the
20 project. In each calibration exercise, two additional laboratories were invited to participate
21 in order to determine some measure of comparability with other studies reported in the
22 scientific literature. All laboratories reported their results independently. In the time period
23 between these two calibration exercises, the effectiveness of the individual QA/QC programs
24 was also monitored by inserting double blind audit samples into the sample stream of each
25 study to measure the persistency of analytical precision throughout the study and to monitor
26 analytical drift.
27 The participating cities recognized the need for standardizing the sampling and
28 analytical protocols so that data from each study could be compared. This standardization
29 was accomplished for soil and dust by measuring the analytical difference between each of
30 the three labs. Common standards were prepared and a program for assuring data quality
31 was put into place. A three step program was agreed to that involved: (1) a round robin
September 1, 1995
3-2
DRAFT-DO NOT QUOTE OR CITE
-------
1 calibration study of soil samples to measure differences between laboratories and differences
2 between analytical methods and instrumentation; (2) a double blind audit system for soil and
3 dust to monitor the performance of each laboratory during the project; and (3) a second
4 round robin calibration study to determine the arithmetic correction factor that would
5 normalize dust and soil data to a common project basis. This program ensured that analyses
6 performed by each of the three participating laboratories would be internally accurate and
7 externally consistent with similar analyses by other research laboratories.
8 Intercalation exercise I was conducted prior to the beginning of each study using soil
9 and dust samples collected from representative neighborhoods in each city. Intercalibration
10 exercise II was conducted near the end of the sampling phase of the project using aliquots of
11 soil and dust samples collected at the beginning of the sampling phase, some of which were
12 used for QA/QC monitoring during the project.
13
14 3.1.1 Round Robin Intercalibration Exercise I
15 At the beginning of this project, the methods proposed by each study for soil and dust
16 analysis were reviewed by the Scientific Coordinating Panel. The preferred method, hot
17 nitric acid digestion followed by atomic absorption spectroscopy (AAS), was time consuming
18 and expensive. The number of samples was expected to exceed 75,000 per study, so more
19 rapid and less expensive methods were evaluated. Laboratory scale X-ray fluorescence
20 (XRF) spectroscopy and inductively coupled plasma (ICP) emission spectroscopy were
21 proposed, and a cold nitric acid extraction method for AAS was also considered.
22 In May 1988, prior to the beginning of each study, each of the three laboratories
23 collected ten soil samples from areas similar to those that would be included in their study.
24 One of the samples from Cincinnati was a street dust sample of very high lead concentration.
25 The other 29 samples were selected from soils with lead concentrations expected to range
26 from 250 to 8,000 fJLg/g. The samples were dried and sieved according to the study
27 protocols. Approximately 200 g of each sample were sent to the other two laboratories and
28 to an outside lab at Georgia Tech Research Institute (GTRI). Table 3-1 shows the
29 instrumentation and method of analysis used by each laboratory. In making these analyses,
30 each laboratory used its own internal standards for instrumental calibration and shared a
September 1, 1995
3-3
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 3-1. WET CHEMISTRY AND INSTRUMENTAL METHODS USED FOR
THE FIRST INTERCALIBRATION STUDY
Participating Laboratories
Method* Boston
Hot HNO3/AAS
Cold HNO3/AAS
Hot HNCyiCP
XRF X
Baltimore
X
X
Cincinnati
X
X
GTRI" USDAC
X
X
"HNO3 = Nitric acid; AAS = Atomic absorption spectroscopy; ICP = Inductively coupled plasma emission
spectroscopy; XRF = X-ray fluorescence.
bGTRI = Georgia Tech Research Institute.
"USDA = U.S. Department of Agriculture.
1 common set of five standards provided by Dr. Rufus Chancy at the U.S. Department of
2 Agriculture. The intercalibration exercise successfully established a baseline for cross study
3 comparison of soil and dust results.
4 In summary, the test conditions were that each laboratory would be provided with
5 instructions for preparing the samples (drying, sieving, and chemical extraction) but would
6 use their own internal standards and instrumental settings. They would have access to a set
7 of external standards (from U.S. Department of Agriculture) with known values from which
8 they could make corrections if necessary.
9 Each of the three study laboratories sent aliquots of 10 samples to the other two
10 participating laboratories and to two external laboratories. One of the samples from
11 Cincinnati was a street dust sample with a lead concentration in excess of 15,000 /ig/g. The
12 other 29 samples were soils. The samples were subdivided by sieving during preparation to
13 a "total" and "fine" fraction. Thus there were 30 samples, each with two size fractions
14 analyzed by each of five laboratories using either one or two analytical methods. The
15 analytical and wet chemistry methods used are shown in Table 3-1, and the results of the
16 analyses appear in Table 3-2.
17 The cold nitric acid extraction method was found to be essentially equivalent to the hot
18 nitric acid extraction method for soils with lead concentrations up to 8,000 jtg/g (Figure 3-1)
19 for the samples analyzed hi this study. The AAS method used by Cincinnati and Baltimore
September 1, 1995
3-4
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 3-2. ANALYTICAL RESULTS OF THE FIRST
INTERCALffiRATION STUDY: LEAD CONCENTRATION (jtg/g)
IN THE TOTAL AND FINE FRACTIONS OF 10 SOILS FROM EACH STUDY
Sample
Fraction0
IT
2T
3T
4T
5T
6T
7T
8T
9T
10T
11T
12T
1ST
14T
1ST
16T
17T
1ST
19T
20T
21T
22T
23T
24T
26T
27T
28T
29T
SOT
IF
2F
3F
4F
5F
6F
7F
8F
9F
10F
11F
12F
13F
14F
15F
16F
17F
18F
19F
20F
21F
22F
Boston
XRF
1,200
1,750
400
550
1,100
1,450
1,000
500
550
1,450
250
800
100
700
550
220
220
75
50
4,800
500
950
1,700
2,400
2,800
3,800
5,200
4,000
6,500
1,500
2,650
500
1,600
1,700
2,400
1,200
600
650
2,200
220
1,800
100
800
620
300
100
100
50
5,100
550
1,100
Baltimore
Hot HNO3
AAS
1,418
2,893
492
619
1,058
2,323
1,359
683
608
1,649
484
1,069
2,200
1,754
264
126
106
9
15,792
496
850
1,559
2,260
2,484
3,846
5,092
5,097
7,995
1,545
3,540
625
1,814
1,793
3,137
1,344
723
686
2,398
356
2,707
96
100
796
3,200
118
142
7,866
606
1,118
Hot HNO3
ICP
1,324
2,544
389
462
882
1,955
1,098
535
485
1,330
365
878
53
1,701
1,410
200
62
48
7
12,030
372
698
1,298
1,880
2,119
3,440
4,667
4,510
6,560
1,421
2,921
507
1,554
1,475
2,387
1,105
598
558
1,946
244
2,220
68
779
616
236
73
85
10
6,000
506
916
Cincinnati
Hot HN03
AAS
1,552
2,868
387
423
964
1,876
1,383
491
455
1,679
316
.1,850
63
2,068
747
253
59
74
2
14,593
387
837
1,567
2,284
2,754
4,337
5,454
5,586
8,467
1,560
3,335
478
1,678
1,689
2,835
1,306
595
593
1,808
267
2,683
68
926
635
237
73
91
3
8,109
480
1,069
Cold HNO3
AAS
1,215
2,211
466
415
854
1,722
990
725
417
1,228
348
1,103
45
1,713
785
295
58
61
3
8,147
378
739
1,368
2,003
2,401
3,835
4,747
4,700
7,502
1,404
3,127
508
1,595
1,971
2,009
1,184
298
601
1,116
277
2,683
64
818
642
239
66
87
2
7,432
467
944
GTRP
XRF
1,174
1,912
400
500
980
1,524
651
400
261
1,660
180
900
100
652
505
187
30
100
20
4,817
383
717
1,390
2,021
2,331
3,500
4,460
3,280
4,704
1,223
2,263
440
1234
1,290
2,134
815
490
375
1,980
180
1,680
100
693
600
236
100
100
30
4,780
505
980
USDAb
Cold HNO3
AAS
1,338
2,695
417
464
988
1,808
1,473
726
605
1,764
304
1,944
73
1,710
825
286
83
111
13
14,733
1,120
1,761
2,561
2,472
4,983
3,184
6,473
10,042
1,569
3,273
515
1,824
1,683
2,682
1,297
672
630
280
2,610
89
895
664
242
80
92
20
8,451
470
904
September 1, 1995
3-5
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 3-2 (cont'd). ANALYTICAL RESULTS OF THE FIRST
INTERCALffiRATION STUDY: LEAD CONCENTRATION (/tg/g)
IN THE TOTAL AND FINE FRACTIONS OF 10 SOILS FROM EACH STUDY
Sample
Fraction6
23F
24F
25F
26F
27F
28F
29F
30F
Boston
•
XRF
1,700
2,200
2,200
2,800
4,000
3,100
4,500
8,000
Baltimore
Hot HN03
AAS
1,679
2,331
2,372
2,899
4,833
3,087
5,896
8,555
Hot HNO3
ICP
1,424
2,014
2,000
2,402
3,969
2,616
4,717
7,443
Cincinnati
Hot HN03
AAS
1,710
2,328
1,665
2,946
4,531
3,073
5,606
8,679
Cold HNO3
AAS
1,431
2,010
2,089
2,568
4,130
2,720
4,869
7,789
GTRI"
XRF
1,320
1,940
2,005
2,249
3,739
2,445
4,240
6,015
USDAb
Cold HNO3
AAS
1,640
2,492
3,156
4,979
'6,194
6,680
9,754
*GTRI = Georgia Tech Research Institute.
bUSDA = U.S. Department of Agriculture.
T = Total fraction, F = Fine fraction.
Cincinnati Hot HNO
O
Thousands
Figure 3-1. Comparison of uncorrected data for two wet chemistry methods of soil
analysis showing the comparability of hot and cold nitric acid for the
Cincinnati laboratory. The straight line indicates a slope of 1.
September 1, 1995
3-6
DRAFT-DO NOT QUOTE OR CITE
-------
1 was also equivalent (Figure 3-2), showing a high degree of comparability between these two
2 laboratories under these test conditions.
Cincinnati AAS Hot HNO3 (ng/g)
Thousands
Figure 3-2. Comparison of unconnected data for atomic absorption spectroscopic
analysis by two laboratories (Baltimore and Cincinnati) using the hot nitric
acid method of soil analysis. The straight line indicates a slope of 1.
3 The interlaboratory comparison of XRF between the Boston and GTRI Laboratories
4 showed the method was acceptable, although not fully linear above 5,000 |wg/g. There were
5 no soil standards available above 2,000 pig/g, so the analysts had some difficulty calibrating
6 their XRF instruments above this level. The data of Figure 3-3 suggest a systematic
7 difference between the two laboratories that could be corrected with a more uniform
8 calibration. Both interlaboratory (Cincinnati and Baltimore in Figure 3-4) and intralaboratory
9 (Baltimore in Figure 3-5) comparisons of AAS versus ICP demonstrated equivalency between
10 these two instrumental methods. These comparisons showed that there is likewise a
11 systematic difference that can be statistically corrected.
September 1, 1995
3-7
DRAFT-DO NOT QUOTE OR CITE
-------
1 23456789
Boston XRF (ixg/g)
Thousands
Figure 3-3. Interlaboratory comparison of unconnected data for the X-ray fluorescence
method of soil analysis showing the comparability of the Boston and
Georgia Institute of Technology laboratories. The straight line indicates a
slope of 1.
10
Cincinnati AAS
Thousands
15
20
Figure 3-4. Interlaboratory comparison of uncorrected data for soil analysis showing
the comparability of inductively coupled plasma emission spectroscopy and
atomic absorption spectroscopy for the Baltimore and Cincinnati
laboratories. The straight line indicates a slope of 1.
September 1, 1995
3-8
DRAFT-DO NOT QUOTE OR CITE
-------
I
M
0-1
2
£
o
I
CQ
5 10 15
Baltimore AAS Hot HNO3 (ng/g)
Thousands
20
Figure 3-5. Comparison of uncorreeled data for soil analysis showing the comparability
of inductively coupled plasma emission spectroscopy and atomic absorption
spectroscopy within the Baltimore laboratory. The straight line indicates a
slope of 1.
1 Finally, the interlaboratory comparison of XRF versus AAS (Boston and Cincinnati in
2 Figure 3-6, and Boston and Baltimore in Figure 3-7) led to the conclusion that, if suitable
3 soil standards at higher concentrations could be made available, XRF would be an acceptable
4 alternative method to AAS for soil analysis.
5 The Scientific Coordinating Panel recommended the use of XRF for soil analysis on the
6 condition that a suitable set of common standards could be prepared for a broader
7 concentration range and that a rigorous audit program be established to ensure continued
8 analytical accuracy. This recommendation was based on the interlaboratory comparison
9 study, the awareness that chemical extraction of a large number of soil samples presented a
10 costly burden on the project both in terms of time and expense, and the value of
li nondestructive analysis in preserving the samples for reanalysis. The Round Robin I
12 calibration exercise also revealed the need for a broader scale calibration exercise to
13 determine the arithmetic correction factor for converting the data to a common basis.
14 For routine analyses, two groups, Boston and Baltimore, elected to use XRF for
15 interior dust analysis also, whereas Cincinnati opted for hot nitric extraction with AAS for
September 1, 1995 3.9 DRAFT-DO NOT QUOTE OR CITE
-------
5 10
Cincinnati Hot HNO3 AAS
Thousands
Figure 3-6. Interlaboratory comparison of uncorrected data for soil analysis showing
the comparability of X-ray fluorescence and atomic absorption spectroscopy
for the Cincinnati and Boston laboratories. The straight line indicates a
slope of 1.
20r
Baltimore Hot HNO3 AAS
Thousands
Figure 3-7. Interlaboratory comparison of uncorrected data for soil analysis showing
the comparability of X-ray fluorescence and atomic absorption spectroscopy
for the Baltimore and Boston laboratories. The straight line indicates a
slope of 1.
September 1, 1995
3-10
DRAFT-DO NOT QUOTE OR CITE
-------
1 interior dust and XRF for exterior dust. During the study, Baltimore recognized problems
2 with analyzing dust by XRF when the sample size was small, less than 100 mg. They
3 reanalyzed the dust samples by AAS and reported both measurements. In Boston, this
4 problem was solved by compositing the floor dust samples for XRF analysis, reporting one
5 floor dust sample per housing unit.
6
7 3.1.2 Quality Assurance/Quality Control Standards and Audits
8 After the first intercalibration exercise, a set of nine interlaboratory standards was
9 prepared to monitor the QA/QC performance of soil and dust analysis throughout the project.
10 These were prepared from three soil samples and two dust from each of the three studies,
11 collected in bulk (about 30 kg), hi a range thought to be high, medium, and low for that
12 area. Seven of the soil samples and five of the dust samples were dried, sieved, and
13 analyzed at the EPA Environmental Monitoring Systems Laboratory in Las Vegas, NV
14 (EMSL/LV). Following homogenization, approximately fifty aliquots of each of the samples
15 were analyzed by laboratory scale XRF at the EMSL/LV laboratory to estimate the
16 acceptable range for a single laboratory. Three of the nine soils were distributed to the
17 participating cities for use as interlaboratory reference standards. The remaining six were
18 used as double blind external audits.
19 Each city appointed a QA/QC officer who was not directly involved with the analysis
20 of the soil samples, but who had access to the soil sample preparation stream on a daily
21 basis. This person mailed prelabeled soil sample containers with typical sample numbers to
22 the EMSL/LV laboratory. Approximately 20 g samples from one of the six external audit
23 materials typical for each city were placed in the sample containers fully disguised as field
24 soil samples and returned to the QA/QC officer hi lots of 20 to 30. The identification
25 numbers and soil concentration values were monitored by the project QA/QC officer at
26 ECAO/RTP. Each city's QA/QC officer inserted the double blind samples into the sample
27 stream on a random basis at a frequency that would ensure about four QA/QC samples per
28 analytical day. These were occasionally placed as duplicates in the same batch to provide
29 information about replication within the batch.
30 The preliminary acceptance range for the double blind audit samples was established
31 using the original 50 XRF analyses by the Las Vegas laboratory discussed above. As the
September 1, 1995
3-11
DRAFT-DO NOT QUOTE OR CITE
-------
1 analytical results were reviewed by the study QA/QC officer, the audit sample results were
2 sent to the project QA/QC officer at ECAO/RTP. If the audit samples were outside the
3 acceptable range, the study QA/QC officer was informed and could recommend either
4 reanalysis or flagging the data for that entire batch. The initial acceptable range for the six
5 audit samples was based on analyses by a single laboratory (EMSL/LV). This range was
6 adjusted for interlaboratory variation after the Intercalibration Exercise II. Final decisions on
7 the disposition of the audit sample anomalies were deferred until the completion of the
8 second intercalibration exercise near the end of the study.
9 The results of the double-blind audit program are given hi Table 3-3 based on the final
10 biweight distributions hi Table 3-4. The preliminary biweight distributions, shown also in
11 Table 3-4, contained no measure of interlaboratory variability because the preliminary
12 analyses were performed by only the EMSL-LV laboratory. These values could only be used
13 in a preliminary assessment of the audit program to identify and flag batches of soil samples
14 that might need to be reanalyzed pending the determination of the final biweight
15 distributions.
16 The laboratories were found to be systematically low or high. This was not of major
17 concern, as these discrepancies could be resolved by a more detailed intercalibration exercise
18 and statistical correction at the end of the study. The Cincinnati group elected to make a
19 midcourse change hi instrumental parameters that reduced this difference, and they described
20 this procedure hi their report. Occasionally, the measured audit sample was sporadically
21 high or low, hi which case the laboratory investigated the problem and resolved it. Most of
22 these discrepancies occurred for dust samples where the sample size for XRF analysis was
23 below 200 mg. The Boston group found, but did not report in detail, that a calibration curve
24 for XRF analysis using standards that were also less than 200 mg would provide a suitable
25 correction to the original data. They elected, however, to composite their floor dust
26 samples.
27
28 3.1.3 Round Robin Intercalibration Exercise II
29 Near the end of the project, aliquots of the nine soil and six dust audit samples used
30 during the project were redistributed to the three study laboratories for single blind analysis.
31 The analyst was aware that the samples were audit samples, but did not know their
September 1, 1995
3-12 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 3-3. SOIL AND DUST AUDIT PROGRAM RESULTS
Study/ Audit Sample
BOSTON DUST (XRF)
BAL 03
CIN01
GIN 02
BOSTON SOIL (XRF)
BOSM
BALH
CINL
CINH
BALTIMORE DUST (XRF)
BAL 02
CIN 01
BOS 01
BALTIMORE SOIL (XRF)
BOSM
BALH
CINL
CINH
Number of
Samples
N/Ab
N/A
N/A
N/A
N/A
N/A
N/A
8
10
10
15
15
15
15
Mean
0*g/g)
1,232
2,671
331
6,786
1,044
399
14,074
218
3,280
14,444
5,046
838
286
11,290
Range
(/*g/g)
980-1,441
2,075-3,228
115-461
6,015-7,549
747-1,244
207-570
11,407-16,592
159-281
800-3,660
14,080-14,920
4,800-5,200
433-916
266-307
10,100-12,500
Percent Within
Final Biweight
Distribution2
92
100
65
100
73
61
50
100
90
100
100
60
100
53
CINCINNATI DUST (AAS)
BAL 03 34
BOS 01 35
CIN 01 38
CIN 02 26
CINCINNATI SOIL (XRF)
1,727
24,104
2,683
259
1,322-2,687
20,266-27,962
2,070-3,163
200-393
N/A
N/A
100
100
BOSM
BALH
CINL
CINH
32
49
130
31
5,580
885
263
12,304
4,759-6,107
822-1,012
244-310
9,838-13,632
100
100
100
N/A
These percentages include audit samples for which analyses were outside the biweight distribution range and
for which the action required by the QA/QC plan, such as reanalysis of the entire batch, was implemented.
bN/A = Not available.
September 1, 1995
3-13
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 3-4. PRELIMINARY AND FINAL BIWEIGHT DISTRIBUTIONS FOR SOIL
AND DUST AUDIT PROGRAM
Sample Audit
Type Sample
1
2
3
4
5
6
7
Dust
Dust
Dust
Dust
Dust
Soil
Soil
Soil
Soil
Soil
Soil
Soil
Soil
Soil
Soil
Soil
Soil
Soil
BAL01
BAL02
BAL03
CIN01
CIN02
BOSL
BOSM
BOSH
BALL
BALH
CINL
CINH
REF5
REF6
REF7
REF8
REF9
REF10
Preliminary Values (/xg/g)
Mean
78
331
1,480
2,851
252
3,131
6,090
14,483
639
923
303
13,585
Low
58
288
1,346
2,660
216
2,858
5,748
13,071
555
850
284
12,872
concentrations. These measurements were
acceptability for the
study to
3.1.4
audit samples, and for
values common to the
project.
High
99
374
1,613
3,042
288
3,405
6,431
15,895
724
997
322
14,297
the basis for
adjusting the
Final Values
(Mg/g)
Mean Low
1
2
3
6
13
1
12
1
2
3
84
309
,438
,617
233
,101
,219
,369
626
,017
315
,729
413
936
,042
,354
,913
735
establishing
4
138
1,091
1,422
93
2,283
4,742
11,980
468
847
204
11,361
258
738
758
1,950
2,943
615
1
3
3
7
14
1
14
1
1
2
4
High
163
480
,786
,812
372
,919
,696
,754
783
,187
426
,096
568
,134
,326
,759
,888
854
the final range of
soil and dust measurements
in each
Biweight Distribution and Final Interlaboratory Calibration
The nine soil and five dust samples that were used
samples were reanalyzed in a more detailed round robin
September 1, 1995
3-14
for external standards and audit
t
exercise near the end of the project.
DRAFT-DO
NOT QUOTE OR CITE
-------
1 The purpose of this exercise was to determine the correction factor for statistically converting
2 the soil and dust data from each study to a common basis and to revise the biweight
3 distribution values for the audit samples to reflect the multilaboratory variance and systematic
4 differences between laboratories. Additional analyses by A AS were performed by Baltimore
5 and Cincinnati for soil and dust, even though only dust was analyzed by AAS during the
6 study. Boston and Las Vegas analyzed the samples by ICP for the purposes of obtaining a
7 broader perspective on the application of this method. The data from this exercise are in
8 Table 3-5. They are the basis for determining the consensus values and correction factors
9 that appear in Table 3-6.
10 The data evaluation subcommittee of the Scientific Coordinating Panel was appointed to
11 determine the consensus values and methods of statistical interpretation of the intercalibration
12 results. Several methods were discussed in great detail. Tests were made for outliers using
13 the method of Barnett and Lewis (1984), and none were found. The data were of good
14 quality and were highly linear. The r2 values ranged from 0.997 to 0.999 using a consensus
15 based on the simple arithmetic means of the reported values. The subcommittee chose to
16 explore alternatives to the arithmetic mean and eventually settled on a multiplicative model
17 weighted for within-laboratory variance. The model was run with GLIM statistical software,
18 Version 3.77, Update 2, and gave consensus values and correction factors shown in
19 Table 3-6. Although great care was taken to evaluate several alternatives to simple
20 regression, the consensus values produced by the GLIM procedure differed only slightly
21 from those of a simple linear regression. The correction factors on Table 3-6 were used by
22 the three studies to convert their soil and dust data to a common project basis. A plot of the
23 dust (Figure 3-8) and soil (Figure 3-9) reported values versus the consensus means derived
24 from the GLIM analysis illustrates the reliability of this method.
25
26 3.1.5 Disposition of Audit Data
27 Based on the results of the second intercalibration exercise, a consensus value was
28 determined for each dust and soil sample, biweight distributions were determined for those
29 that had been used in the audit program. This new distribution incorporated interlaboratory
30 variation. When the correction factor is applied to the reported results, the revised number
31 should lie between the upper and lower boundaries of the biweight distribution. Table 3-3
September 1, 1995
3-15
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 3-5. RESULTS OF THE FINAL INTERCALIBRATION STUDY
1
2
3
4
5
6
7
8
Sample
DUST1
DUST2
DUSTS
DUST4
DUSTS
SOIL1
SOIL2
SOILS
SOIL4
SOILS
SOIL6
SOIL?
SOILS
SOIL9
SOIL10
SOIL11
SOIL12
SOIL13
SOIL14
SOIL15
BOSK
120
320
1,430
2,000
280
450
900
1,050
2,200
3,800
710
650
950
2,800
5,600
12,500
310
12,000
810
1,450
lists the percentage
of the discrepancies
BOSX
510
910
1,100
2,300
4,000
770
930
930
2,900
5,300
13,000
290
12,000
850
1,600
of these
XRF
BAL
121
482
1,686
3,771
267
388
808
961
2,100
3,486
640
559
896
2,514
5,200
11,000
283
10,500
793
1,400
AAS
CIN
92
329
1,307
2,924
233
441
1,033
1,080
2,555
4,227
789
675
1,036
3,126
6,493
15,963
305
14,156
929
1,705
LV
78
288
1,288
2,456
212
310
833
923
2,264
3,974
611
532
798
2,972
5,956
15,984
286
13,530
763
1,509
audit sample values that fell
were resolved by
When the audit sample
BAL
15
201
1,363
2,335
150
383
1,001
1,100
2,468
4,044
741
567
1,032
3,401
6,861
13,175
321
13,000
875
1,731
CIN
66
236
1,581
2,451
273
452
1,013
1,120
2,502
4,251
798
650
1,067
3,263
6,937
13,955
379
13,195
986
1,766
within these new
ICP
BOS
94
284
1,428
2,109
244
401
850
972
2,230
3,748
699
597
944
3,148
5,932
12,652
300
13,167
907
1,631
boundaries.
LV
72
307
1,346
2,296
191
379
912
1,006
2,286
3,843
660
626
998
3,158
6,360
12,608
294
11,440
900
1,650
Most
the corrective measures taken by the laboratories.
values fell outside the boundaries of
distribution, the batches were flagged
. The options could then be
the statistical analysis, reanalyze the samples,
evidence
equal to
or use the
the final
biweight
to exclude these data from
original data based on other
that the data are correct. The quality of soil and dust analysis in
or greater than the generally
this project
was
acceptable standards for reporting soil and dust data in
the scientific literature.
September 1, 1995
3-16 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 3-6. CONSENSUS VALUES AND CORRECTION FACTORS FROM
THE FINAL INTERCALIBRATION PROGRAM
Sample
DUST1
DUST2
DUSTS
DUST4
DUSTS
Study
BOS
BAL
CIN
XRF
Interlaboratory Consensus Values
92.8
342.7
1,319.0
2,943.4
228.3
Interlaboratory Correction
1.1527
0.7803
1.0074
AAS
for Dust 0*g/g)
54.2
221.9
1,492.2
2,378.1
232.4
Factors"
1.0416
0.9616
ICP
81.7
283.4
1,362.3
2,133.4
206.2
1.0707
Interlaboratory Consensus Values for Soil (ptg/g)
Sample
SOIL1
SOIL2
SOILS
SOIL4
SOILS
SOIL6
SOIL?
SOILS
SOIL9
SOIL10
SOIL11
SOIL12
SOIL13
SOIL14
SOIL15
460.2
960.7
1,140.5
2,493.5
4,139.3
761.0
664.1
1,062.3
2,987.8
6,175.2
13,120.7
335.3
12,498.5
941.3
1,663.2
430.5
1,002.1
1,106.2
2,474.2
4,164.1
776.9
,623.3
1,049.4
3,272.6
6,863.2
13,645.4
361.5
13,041.6
949.5
1,744.1
426.6
909.6
1,018.8
2,342.1
3,706.1
736.1
656.0
1,005.4
3,274.9
6,411.5
13,224.7
323.6
13,080.0
923.3
1,716.8
Interlaboratory Correction Factors for Soil"
Study
BOS
BAL
CIN
1.0370
1.1909
0.8698
1.0166
0.9839
1.0166
The correction factor is the value that the reported soil or dust measurement should be multiplied by in order
to adjust each value to a common basis among all three studies.
September 1, 1995
3-17
DRAFT-DO NOT QUOTE OR CITE
-------
3000
500
1000 1500 2000
Consensus XRF (iig/g)
2500
3000
n Boston o Baltimore P> Cincinnati x EMSL-LV
Figure 3-8. Departures from consensus dust values for each of the three studies.
o>
0>
CO
20
15
«
s™
CL
0)
CC
5 10
Consensus XRF
Thousands
15
D Boston o Baltimore t> Cincinnati x EMSL-LV
Figure 3-9. Departures from consensus soil values for each of the three studies.
September 1, 1995 3-18 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
3.2 QUALITY ASSURANCE AND QUALITY CONTROL FOR HAND
DUST
The collection and analysis of hand wipes is an innovative procedure developed just
prior to the beginning of the project. There were few published reports of the measurement
techniques, no certified standards, no internal standards, and little information on which to
base decisions for acceptable analytical precision. Double blind audit samples were provided
to the study QA/QC officer as an external control for hand wipe analysis. These were
prepared as simulated samples by placing a known amount of an appropriate solution of lead
nitrate onto the blank hand wipe at the EMSL/LV laboratory, wrapping and labeling
according to the field protocol and returning to the participating laboratory for insertion into
the sample scheme. There was no attempt to determine interlaboratory variance or to
calculate correction factors. The study QA/QC officer was responsible for reporting
problems to the laboratory director.
3.3 QUALITY ASSURANCE AND QUALITY CONTROL FOR BLOOD
LEAD
The QA/QC program for blood analysis was directed by Dr. Dan Paschal of the
Centers for Disease Control and Prevention (CDC) using the protocols developed for the
CDC blood lead certification program. Each laboratory received double blind bovine blood
samples from CDC Blind Pool 1 and Blind Pool 2. The data from this QA/QC program are
in Table 3-7. These data report the number of exceedances to be zero for all three studies.
An exceedance occurs when the mean of two replicates exceeds the range established by
CDC. The data also report the probability of analytical drift during the period of analysis.
There was evidence for drift in the Boston Blind Pool 2 and marginal evidence in Cincinnati
Blind Pool 1.
3.4 DATABASE QUALITY
Each study maintained rigorous standards for database quality. These included double
entry, 100% visual confirmation, and standard statistical procedures for detecting outliers.
September 1, 1995
3-19 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 3-7. QUALITY CONTROL RESULTS FOR
CENTERS FOR DISEASE CONTROL AND
PREVENTION BLIND POOL BLOOD LEAD ANALYSES
Study
Boston
Baltimore
Cincinnati
Dates
Jul 89 - Aug 91
Aug 88 - Oct 90
Aug 88 - Oct 90
n
123
66
53
Blind Pool 1
Number of
Exceedances1
0
0
0
Blind Pool 2
Drift2
0.2092
0.6382
0.0672
n
112
59
48
Number of
Exceedances1
0
0
0
Drift2
0.0389
0.4748
0.4732
'Number of samples that exceeded the range established by CDC for each batch of QC blood analyses within
a pool.
2The drift test probability is a P-value for the test of the hypothesis that the slope of the difference between
the reported values and the CDC accepted value is significantly greater than zero. A P-value less than 0.05
indicates this slope may be greater than zero and that some analytical drift may have occurred over time, but
the direction of this possible drift is not indicated by this statistic.
1 In reviewing the data for statistical analyses contained in this Integrated Report, some
2 errors were found, confirmed, and corrected prior to use in this assessment. None of these
3 errors would have impacted the conclusions drawn by the individual study reports.
4 This evaluation of the QA/QC data shows that the three studies were comparable in
5 their ability to meet the requirements of their QA/QC program. Furthermore, their
6 performance on the audit program and intercalibration exercises suggests that the data are
7 comparable among the three studies, with the appropriate correction factors shown in
8 Table 3-6. While the QC data for Boston blood lead analyses suggest the possibility of
9 analytical drift for part of the period where blood lead data were being corrected, the
10 statistical methods for evaluating abatement effectiveness used by the investigators and by
11 this assessment would compensate for any possible analytical drift.
September 1, 1995
3-20
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
4. INDIVIDUAL STUDIES
4.1 INDIVIDUAL STUDY INTERVENTION STRATEGIES AND
SAMPLE PLANS
4.1.1 Boston Study
The pathway intervention scheme for Boston is shown in Figure 4-1. The approach to
soil abatement was to remove the top 15 cm of soil, apply a synthetic fabric, and cover with
a layer of about 20 cm of clean topsoil. The new soil was covered with sod or seeded with
grass and watered through dry months. Areas not resodded were covered with a bark mulch.
Some driveways and walkways were covered with 5 cm soil and 15 cm gravel or crushed
bank (stone with dust). On four properties, the driveway and yard were capped with 7.5 cm
asphalt without soil removal, at the owner's request. A total of 93 Boston properties,
including those abated at the end of the project, were abated in this manner. The
information on area treated and volume of soil removed from these properties appears in
Table 4-1. The method of excavation was by small mechanical loader (Bobcat) and hand
labor, for the most part. Initially, six properties were abated with a large vacuum device
mounted on a truck, but this proved unsatisfactory due to the size and lack of
maneuverability. During one extreme cold spell, it was necessary to remove large blocks of
frozen soil, often greater than 15 cm thick, by loosening with a jackhammer.
Interior dust abatement was performed after loose paint stabilization. Families spent
the day off-site during interior dust abatement. Hard surfaces (floors, woodwork, window
wells, and some furniture) were vacuumed with a High-Efficiency Particle Accumulator
(HEPA) vacuum, as were soft surfaces such as rugs and upholstered furniture. Hard
surfaces were also wiped with a wet cloth (an oil treated rag was used on furniture)
following vacuuming. Common entries and stairways outside the apartment were not abated.
In Boston, loose paint stabilization consisted of removing chipping and peeling paint
with a HEPA vacuum and washing the surfaces with a trisodium phosphate and water
solution. Window wells were painted with a fresh coat of primer.
Although subsequent measurements of lead-based paint were made, no measurements
were made of the movement of lead from paint to house dust that would reflect the
September 1, 1995
4-1
DRAFT-DO NOT QUOTE OR CITE
-------
h
Atmospheric
I Particles
\ ./
s
Onil
Exterior Paint
Dust
Secondary ^
Occupational
Dust J
• Full Abatement
= Stabilization
Figure 4-1. Pathway intervention scheme for dust exposure (Boston Soil Abatement
Study). Bold-line rectangles indicate pathway components monitored by
sequential sampling.
TABLE 4-1. SOIL ABATEMENT STATISTICS FOR THE THREE STUDIES
Number of properties1
Surface area (m2)
Volume soil removed (m3)
Surface area/property (m2)
Volume soil/property (m3)
Boston
36
7,198
1,212
200
34
Baltimore
63
4,100b
690
73
11
Cincinnati
171
12,089
1,813
71
11
"Includes only properties abated during study. Properties abated at the end of the study, where no further
sampling was reported, are not included in this analysis, but are included in the individual study reports.
In Cincinnati, a property is the location of the soil abatement, not the location of the child's residence.
bSurface area not provided by Baltimore report. This was calculated using Boston volume-to-surface ratio,
which is equivalent to an average removal depth of 17 cm.
September 1, 1995
4-2
DRAFT-DO NOT QUOTE OR CITE
-------
1 effectiveness or persistency of paint stabilization. It was believed that any contamination
2 from lead-based paint would be readily apparent in the dust samples.
3 The Boston study retained 149 of the original 152 children enrolled. Twenty-two of the
4 149 children moved to a new location but were retained in the study. Children with blood
5 lead concentrations below 7 /-tg/dL or above 24 jtig/dL had been excluded from the study and
6 two of the 149 children were dropped from the data analysis when they developed lead
7 poisoning, probably due to exposure to lead-based paint outside their home.
8 Baseline characteristics (age, SES as derived from the Hollingshead Index, soil lead,
9 dust lead, drinking water lead, and paint lead) were similar for the three Boston study groups
10 (BOS P, BOS PI, BOS SPI). The preabatement blood lead concentration was higher for BOS
11 P. The proportion of Hispanics was higher in BOS P than in BOS PI or BOS SPI, and the
12 proportion of Blacks was lower. There was a larger proportion of male children in BOS P.
13 Data were analyzed by comparison of group means using analysis of covariance
14 (ANCOVA), which showed a significant effect of group assignment (intervention) for both
15 the BOS PI and BOS SPI groups. These results did not change with age, sex, socioeconomic
16 status, or any other variable except race and paint loading (P-XRF measurement). When the
17 paint loading was controlled, the blood lead declines were diminished; when the race variable
18 was added, the blood lead declines were also diminished and the results were not statistically
19 significant.
20 The Boston study has some limitations. Participants were chosen to be representative
21 of the population of urban preschool children who were already at risk of lead exposure.
22 The Boston Childhood Lead Poisoning Prevention Program was used to identify potential
23 participants from neighborhoods with the highest rates of lead poisoning. Because no study
24 subjects had blood lead levels below 7 pig/dL or in excess of 24 /ig/dL at baseline,
25 extrapolation of the effect of lead contaminated soil abatement for children above or below
26 this range is difficult.
27 Follow-up blood lead measurements were made in Boston eleven months after
28 intervention and again at 23 months.
September 1, 1995
4-3
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
4.1.2 Baltimore Study
In Baltimore, 63 properties in BAL SP were abated between August and November
1990. An additional seven properties that did not meet the requirements for abatement were
transferred to the control group (BAL P). The pathway intervention scheme is shown in
Figure 4-2. Soil surfaces were divided into parcels on each property, usually front, back,
and one side; and any parcel with soil lead concentrations above 500 jug/g was abated
entirely. Soil and ground cover were removed down to 15 cm and replaced to the original
level with soil having a lead concentration less than 50 /ig/g. These areas were sodded or
reseeded as appropriate. Bare areas were prepped and reseeded even if soil lead
concentrations did not warrant excavation. Additional abatement statistics appear in
Table 4-1.
[ Atmospheric
Particles
^
s
9nil
Exterior Paint
Dust
Interior Paint
Dust
Local
Fugitive
Dust
Exterior
Dust
Interior
Dust
Hand
Dust
Secondary
| Occupational |
Dust
Food
Child
Full Abatement
Stabilization
Figure 4-2. Pathway intervention scheme for dust exposure (Baltimore Soil Abatement
Study). Bold-line rectangles indicate pathway components monitored by
sequential sampling.
September 1, 1995
4.4
DRAFT-DO NOT QUOTE OR CITE
-------
1 The exterior painted surfaces of Baltimore homes were wet scraped over the chipping
2 and peeling surfaces, followed by HEPA vacuuming. The entire surface was primed and
3 painted with two coats of latex paint.
4 The Baltimore study recruited 472 children, of whom 185 completed the study.
5 Of those that completed the study, none were excluded from analysis. The recruited children
6 were from two neighborhoods, originally intended to be a treatment and a control group.
7 Because soil concentrations were lower than expected, some properties hi the treatment group
8 did not receive soil abatement. In their analysis, the Baltimore group transferred these
9 properties to the control group.
10 Because of logistical problems, there was an extended delay between recruitment and
11 soil abatement that accounted for most of the attrition of the participating families from the
12 study. In their report, the Baltimore group applied several statistical models to the two
13 populations to evaluate the potential bias from loss of participating children. These analyses
14 showed the two populations remained virtually identical in demographic, biological and
15 environmental characteristics.
16 The Baltimore study design focused on changes in biological parameters, hand dust and
17 blood lead, over an extended period of tune. The study provided limited information on
18 changes in the movement of lead in the child's environment in response to intervention.
19 Repeat measurements of soil were on abated properties only, to confirm abatement. There
20 were no abatement measurements of exterior dust, no interior paint stabilization, and no
21 ulterior dust abatement.
22 Including the prestudy screening measurements of hand dust and blood lead hi the
23 original cohort of participants, the Baltimore study made six rounds of biological
24 measurements that spanned twenty months.
25
26 4.1.3 Cincinnati Study
27 The pathway scheme for the Cincinnati study is shown in Figure 4-3. Within each of
28 six neighborhoods, the Cincinnati study identified all sites with soil cover as discrete study
29 sites. The decision to abate was based on soil lead concentrations for each parcel of land,
30 and for the depth to which the lead had penetrated. Lead was measured at two depths, the
September 1, 1995
4-5
DRAFT-DO NOT QUOTE OR CITE
-------
f Atmospheric
Particles
V /
Local
Fugitive
Dust
>
Soil
Exterior Paint
Dust
Secondary^)
Occupational
Dust
Full Abatement
Lead Based Paint Previously Removed
Figure 4-3. Pathway intervention scheme for dust exposure (Cincinnati Soil Abatement
Study). Bold-line rectangles indicate pathway components monitored by
sequential sampling.
1 top 2 cm and from 13 to 15 cm. If the average concentration of the top and bottom samples
2 was 500 /zg/g or greater, the soil was removed and replaced, regardless of the adequacy of
3 the top cover. If the average of the top samples exceeded 500 £tg/g, the soil was also abated.
4 Initially, there was an option to cultivate by roto-tilling, but this approach was abandoned as
5 not feasible in this study. For areas where the top concentration was greater than or equal to
6 300 jtig/g, and the average concentration of the top and bottom samples was less than
7 500 jtig/g and the cover was inadequate, the soil was resodded. Excavation was by front end
8 loader, backhoe, and hand tools down to 15 cm, and the replacement soil lead concentration
9 was less than 50 /ig/g. Further abatement statistics can be found in Table 4-1.
10 The approach to exterior dust abatement was to identify all types of exterior hard
11 surfaces in the neighborhood where dust might collect, to obtain permission to sample and
12 abate these areas, and to clean them once with vacuum equipment, suitable for the area.
13 This vacuum equipment had previously been tested and shown to remove about 95 % of the
September 1, 1995
4-6
DRAFT-DO NOT QUOTE OR CITE
-------
1 available dust on the area. The groups of surfaces selected were streets, alleys, sidewalks,
2 parking lots, steps, and porches. For data analysis in the Cincinnati report, these were
3 grouped as (1) targeted areas adjacent to'the exterior of the buildings where children lived,
4 such as steps, porches, and sidewalks; (2) streets, sidewalks, and alleys throughout the study
5 neighborhoods; and (3) parking lots and other paved areas throughout the study
6 neighborhoods.
7 The exterior dust measurements in the Cincinnati study (and the interior dust
8 measurements of all three studies) were made in a manner that determined the lead
9 concentration (/*g Pb/g dust), the dust loading (mg dust/m2), and the lead loading (pig Pb/m2)
10 for the surface measured. This required that a dry vacuum sample be taken over a
11 prescribed area, usually 0.25 to 0.5 m2. It is important to note that dust abatement is not
12 expected to cause an immediate change in the lead concentration on dust surfaces, only the
13 dust and lead loading.
14 The Cincinnati group performed interior dust abatement after exterior dust abatement,
15 moving the families off-site during this activity. Vacuuming of noncarpeted areas, which
16 was done two times, at a prescribed rate of 1 m2/min, was followed by wet wiping with a
17 detergent. They replaced one to three carpets and two items of upholstered furniture per
18 housing unit. Their previous studies had shown that these soft items could not be cleaned
19 effectively with vacuuming alone. Where carpets could not be replaced, these were vacuum
20 cleaned three times at a rate of 1 mVmin, recognizing the limitations of this method.
21 The Cincinnati study recruited 307 children, including 16 children born to participating
22 families during the study, and an additional 50 children who were recruited after the
23 beginning of the study. In their main data analysis, the Cincinnati group excluded these
24 children who were recruited after the start of the study, plus 31 children who were living in
25 nonrehabilitated housing suspected of having lead-based paint, and four children (in two
26 families) who had become lead-poisoned from other causes. Thus, data for 206 children
27 were analyzed in the Cincinnati report.
28 The Cincinnati study abated soil on 140 parcels of land scattered throughout the
29 neighborhoods. In CIN SEI, where soil abatement was performed in the first year, the
30 arithmetic mean concentration dropped from 680 /*g/g down to 134 /zg/g. In the two groups
September 1, 1995
4.7 DRAFT-DO NOT QUOTE OR CITE
-------
1 where soil abatement occurred in the second year, CIN I-SE-1 and CIN I-SE-2, the soil lead
2 concentration dropped from 262 to 125 ywg/g and 724 to 233 /tg/g, respectively.
3 If soil were the only source of lead in the neighborhoods, exterior and interior dust
4 should have responded to the reduction in soil lead concentrations. Exterior dust lead
5 loading decreased only slightly following soil and dust abatement, but returned to
6 preabatement levels within one year. The analysis of exterior dust should provide a measure
7 intermediate between external sources, such as soil, and house dust. In the case where the
8 soil was abated, then abatement of external dust should speed up the rate at which the impact
9 of this soil abatement can be observed on the interior dust of homes. But soil is not the only
10 source of exterior lead, especially if the distance between the soil and the living unit entry
11 way is more than a few hundred feet. In this case, the recontamination of exterior dust from
12 sources other than soil complicates the interpretation of the movement of soil lead into the
13 home or to exterior play areas.
14 Household dust was abated in.the Boston and Cincinnati studies, but not hi Baltimore.
15 The BOS SPI and CIN SEI groups received interior dust abatement at the same tune as soil
16 abatement, the BOS PI received interior dust abatement without soil abatement, and the CIN
17 I-SE received interior dust abatement in the first year followed by soil and exterior dust
18 abatement in the second year.
19
20
21 4.2 DESCRIPTION OF THE DATA
22 This section focuses on the actual data that formed the basis for the conclusions reached
23 by the individual study reports. These data consist of measurements of soil, exterior dust
24 (sometimes referred to as street dust), interior dust (house dust), hand dust, blood lead,
25 exterior paint, ulterior paint, and drinking water. The age of the child and the date of
26 collection were also included in some analyses. Tables 4-2, 4-3, and 4-4 summarize key
27 data for all three studies. For the most part, these data are the bases for the results and
28 conclusions presented in the individual city reports, and also for the statistical analyses in
29 Chapter 5 of this integrated assessment.
September 1, 1995
4-8
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 4-2. SUMMARY OF BOSTON STUDY DATA
Median Soil Pb Cone, (/xg/g)
BOS SPI
BOS PI
BOSP
Median Floor Dust Pb Cone. 0*g/g)
BOS SPI
BOS PI
BOSP
Median Floor Dust Load (mg/m2)
BOS SPI
BOS PI
BOSP
Median Floor Dust Pb Load (/ig/m2)
BOS SPI
BOS PI
BOSP
Median Window Dust Pb Cone, (jtg/g)
BOS SPI
BOS PI
BOSP
Median Window Dust Load (mg/m2)
BOS SPI
BOS PI
BOSP
Median Window Dust Pb Load (jtg/m2)
BOS SPI
BOS PI
BOSP
Median Hand Pb Load (/jg/pair)
BOS SPI
BOS PI
BOSP
Median Blood Pb Cone. 0*g/dL)
BOS SPI
BOS PI
BOSP
GM Blood Pb Cone. (/ig/dL)
BOS SPI
BOS PI
BOSP
Round 1
2,396
2,307
2,275
2,100
2,240
2,200
24
24
40
52
59
75
13,240
19,667
17,400
293
304
239
7,005
7,196
4,179
6.75
6.75
5.75
13
12
12
12.36
11.70
11.49
Round 2
125
-
-
1,040
1,105
-
36
19
-
40
24
-
9,967
2,400
-
104
31
-
1,392
88
-
4.0
5.5
3.5
10
8
9
9.11
8.01
9.19
Round 3
115
2,084
2,212
845
1,150
950
23
26
28
23
27
27
11,217
10,000
15,500
474
380
239
4,728
4,624
4,441
3.5
2.0
4.5
10
11
11.5
9.90
10.74
10.75
Round 4
-
-
-
760
1,030
1,300
15
17
19
16
18
21
21,125
15,650
12,667
373
570
504
5,735
5,697
5,559
-
-
-
-
-
-
-
-
-
Rounds
193
278
220
726
806
862
31
31
37
24
28
37
8,780
6,870
12,350
919
500
797
5,402
2,553
6,018
12.5
7.15
9.2
10
8
10
9.07
7.11
8.85
September 1, 1995
4-9
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 4-3. SUMMARY OF BALTIMORE STUDY DATA
1
2
3
4
5
6
7
8
Round 1 Round 2 Round 3 Round 4 Round 5 Round
Median Soil Pb Cone. 0*g/g)
BALSP 440 . - - 22
BALP ,409 - - - - -
Median Floor Dust Pb Cone (/xg/g)
BALSP 1,600 - - 1,068
BALP 1,850 - - 1,150
Median Floor Dust Load (mg/m2)
BALSP 40 - 37
BALP 37 - - 38 - -
Median Floor Dust Lead Load (fig/m2)
BALSP 73 38
BALP 72 41
Median Hand Pb Load (fig/pair)
BALSP 10.7 12.9 7.4 8.5 12.6 14.9
BALP 13.6 14.8 9.5 6.0 17.3 13.0
Median Blood Pb Cone. (/ig/dL)
BALSP 12.4 11.0 9.8 8.8 9.9 10.4
BALP 10.6 10.2 9.2 7.4 8.0 8.0
GM Blood Pb Cone. Og/dL)
BALSP 11.0 9.9 9.7 8.6 9.6 9.7
BALP 10.9 10.5 9.1 7.8 8.1 8.4
Each study produced similar information about the occurrence of lead in the
6
environment. The data sets among the studies are not perfectly comparable, however, in that
they differed hi the tuning of the collection relative to intervention (see Figure 2-1), the
spatial distribution of the sampling points relative to the expected exposure to the child, and
the manner in which the data were reduced to a central tendency.
Data were collected hi rounds. That is, during a specific period of tune, samples were
taken of soil, dust, etc. , for a specific objective, such as establishing the concentration of
lead prior to intervention. Usually a round lasted for several weeks, perhaps three to
September 1, 1995 4-10 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 4-4. SUMMARY OF CINCINNATI STUDY DATA
Round 1 Round 2
Round 3
Round 4
Round 5 Round 6 Round 7
Median Soil Pb Cone, (ftg/g)
CIN SEI
CIN I-SE
CIN NT
Median Street Dust Pb
CIN SEI
CIN I-SE
CIN NT
680
237
339
Cone, (/tg/g)
134
247
346
3,937 3,398
3,665 3,416
1,583 1,156
142
240
330
2,118
3,411
891
103
262
256
2,559
2,275
968
122 166 132
125 182 138
331 267 266
3,231
3,040
1,086
Median Street Dust Load (mg/m2)
CIN SEI
CIN I-SE
CIN NT
Median Street Dust Pb
CIN SEI
CIN I-SE
CIN NT
Median Floor Dust Pb
CIN SEI
CIN I-SE
CIN NT
454
649
624
Load (/tg/m2)
242
561
755
1,162 789
2,364 1,618
1,005 957
Cone. (/*g/g)
362
395
229
346
388
224
363
326
481
641
1127
498
325
408
209
452
420
477
968
943
587
474
431
213
310
126
654
808
371
442
158
163
162
Median Floor Dust Load (mg/m2)
CIN SEI
CIN I-SE
CIN NT
Median Floor Dust Pb
CIN SEI
CIN I-SE
CIN NT
Median Window Dust
CIN SEI
CIN I-SE
CIN NT
Median Window Dust
CIN SEI
CIN I-SE
CIN NT
Median Window Dust
CIN SEI
CIN I-SE
CIN NT
418
167
147
Load (/tg/m2)
158
69
35
Pb Cone, (jtg/g)
1,509 1
2,000 1
983
Load (mg/m2)
134
38
126
76
18
32
,287
,572
816
710 433
1,258 380
2,170 2,534
Pb Load (/ig/m2)
983
2,548
1,782 1
426
360
,111
135
117
161
54
58
32
922
1,306
548
254
269
324
242
286
172
197
392
200
130
243
34
1,920
2,017
1,399
4,524
9,860
8,573
15,385
26,364
12,849
76
108
92
502
592
302
966
615
648
397
358
227
September 1, 1995
4-11
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 4-4 (cont'd). SUMMARY OF CINCINNATI STUDY DATA
Round 1 Round 2
Round 3
Round 4
Round 5 Round 6
Round 7
Median Mat Dust Pb Cone. G*g/g)
GIN SEI
CINI-SE
CINNT
109
132
100
738
939
373
549
702
349
767
722
405
659
889
332
„
-
-
_
.
-
Median Mat Dust Load Incremental Increase
Per Day (mg/m2/day)
GIN SEI
GIN I-SE
CINNT
.
-
-
Median Mat Dust Pb Load Incremental
6.5
18.7
1.8
Increase
7.7
4.7
2.0
4.4
4.9
2.7
28.2
16.6
12.2
_
.
-
_
.
-
Per Day (/tg/nrVday)
GIN SEI
GIN I-SE
CINNT
Median Entry
GIN SEI
GIN I-SE
CINNT
Median Entry
GIN SEI
GIN I-SE
CINNT
Median Entry
CIN SEI
GIN I-SE
CINNT
Median Hand
CIN SEI
CIN I-SE
CINNT
Median Blood
CIN SEI
CIN I-SE
CINNT
GM Blood Pb
CIN SEI
CIN I-SE
CINNT
_
-
-
Dust Pb Cone, (/tg/g)
334
425
290
Dust Load (mg/m2)
386
272
348
Dust Pb Load Otg/m2)
112
95
157
Pb Load (fig/pair)
6.0
7.0
3.0
Pb Cone. (jig/dL)
9.2
10.8
9.0
Cone. (/tg/dL)
8.8
10.8
8.3
6.54
7.65
3.30
606
492
367
113
70
238
104
38
80
5.0
7.0
4.0
-
-
-
-
-
-
7.62
5.14
4.67
433
468
317
230
142
294
167
70
88
5.0
5.0
3.0
7.0
9.2
5.9
6.9
9.3
5.7
2.38
3.20
0.99
491
632
286
590
1,394
373
250
588
106
12.0
10.0
5.5
8.0
8.9
6.8
8.8
8.6
6.8
9.80
8.02
5.29
211
102
84
12,671
17,889
14,509
2,502
2,700
1,714
12.5
8.0
7.0
_
-
-
_
-
-
_
-
-
382
598
317
97
161
148
56
103
58
_
_
-
7.9
8.0
6.4
8.2
7.6
7.2
_
_
-
488
615
284
301
513
1,080
150
302
264
_
_
-
8.3
8.8
7.8
8.7
8.9
7.8
September 1, 1995
4-12
DRAFT-DO NOT QUOTE OR CITE
-------
1 four months. It may be important to know when a sample was taken during a round,
2 especially following intervention, in order to evaluate the impact on exposure. Consider the
3 pathway from soil =» street dust =* house dust =» hand lead => blood lead. One would expect,
4 if soil alone (not house dust) were abated and the exposure were mainly through house dust,
5 there would be a lag in tune between abatement and response, and the impact of intervention
6 might become greater with increasing time. Conversely, the impact of intervention might be
7 reduced with time if there were recontamination, as would be expected if house dust were
8 abated but soil or other sources were not.
9 Data linkages are important to the interpretation of the results. Specifically, it is
10 important to know how well the data link (e.g., between soil concentration measurements and
11 house dust concentration measurements) actually represent the hypothesized pathway between
12 soil and house dust. Through these data linkages, it is ultimately possible to construct a
13 simple exposure scenario for the individual child and to analyze these scenarios by structural
14 equation modeling. For example, a young child may spend most of the time indoors,
15 whereupon the exposure scenario becomes the lead that is available to the child through food,
16 drinking water, air, and dust (see Figure 2-1). Each of these proximal sources of lead is
17 influenced by one or more other sources of lead more remote from the immediate exposure
18 of the child.
19 Data are also linked by a primary identifier or index. Some data are linked to the
20 individual child, such as blood lead and hand lead. Some are specific for the living unit or
21 family, and some are specific for the property. It is important to be aware of this distinction
22 because of the duplication effect that can occur when there are several siblings in a family
23 and several families hi a dwelling. This means that a single numerical value for soil such as
24 a mean or median for the premises could be heavily weighted if there were, for example,
25 five children living on the same property.
26
27 4.2.1 Measures of Central Tendency for Property Level Soil and Dust
28 For soil and dust, there is a need to reduce multiple measurements within a round to a
29 single representative data point for each property or living unit. In order to determine the
30 appropriate central tendency for this measurement, the participating groups discussed several
31 alternatives at great length without reaching a consensus. Therefore, different measures of
September 1, 1995
4-13
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
central tendency were reported in each of the three studies. The following is an extended
discussion of each of these measures, followed by an argument for the use of the arithmetic
mean as the best measure hi these circumstances.
The procedures for selecting a representative soil sample were based on the statistical
distribution of data hi each study. The Boston study used the median, giving no weight to
extreme values. The Cincinnati study used the geometric mean, a method that is often used
when the measured values are lognormally distributed, because it gives lesser weight to
extreme values. The geometric mean is always lower than the arithmetic mean for any set of
positive values and therefore may be an underestimate of the exposure to the child.
The distribution problem was approached differently hi Baltimore, where the tri-mean
was calculated as the weighted average of the first, second, and third quartiles:
X =
2
-------
1 evaluated. In the case of outside play activities, a sample would be taken at each location
2 where the child played and this sample would be weighted according to factors such as the
3 time spent playing there and the frequency of hand-to-mouth activity during that time.
4 Because this information is not available, a simplifying assumption is that weight should be
5 given to the location of the sample rather than concentration. Location, not lead
6 concentration, is the basis of choice for the child's play environment. An exposure weighted
7 mean of the soil samples would seem to be the most direct approach. This would be an
8 arithmetic mean of soil values corrected for the degree of exposure to the child. For
9 example, a sample taken from bare soil in an area observed to be a play area would be given
10 a high weighting factor for exposure. Grass covered areas with limited accessibility would
11 be weighted on the low end of exposure. Although cumbersome, this method is feasible
12 because such information was collected at the time of sampling in each study. The drawback
13 is that the method emphasizes the direct, outdoor playtime contact between the child and the
14 exterior dust, and does not consider other routes of dust exposure, such as soil => household
15 dust.
16 An alternative solution is to consider that the child has equal exposure to the entire
17 surface of the soil. In this case, the perfect sample would be to scrape up this upper 2 cm of
18 soil, homogenize it and take a sample. Theoretically, this is equivalent to sampling in a
19 random pattern and taking the arithmetic mean of these samples. In this project, random
20 locations were taken along lines specifically selected to represent the expected high- and low-
21 concentration areas of the plot of soil. In this sense, the arithmetic mean is the best measure
22 of the central tendency of soil data for a property, and is the statistic used in this report. For
23 populations of children at the neighborhood or higher level, the median or geometric mean is
24 often the preferred measure of central tendency.
25
26 4.2.2 Adjustments and Corrections to the Data
27 4.2.2.1 Subjects Dropped from Study
28 During the analysis of their data, the Boston group discovered that two children of the
29 same family had apparently become exposed to lead-based paint abatement debris while
30 staying at a house outside their neighborhood during a time when it was being remodeled.
31 Both siblings had blood lead concentrations that had tripled in less than five months, between
September 1, 1995
4-15
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
Rounds 1 and 3, from 10 to 35 ptg/dL and 17 to 43 /zg/dL. The Boston group analyzed their
data with and without these children, eventually excluding these data from the analyses used
to test their hypothesis. This Integrated Report accepts the conclusion that the data are
outliers and also dropped them from further analysis.
4.2.2.2 Unit Conversion
All data were converted to common units, usually metric. No further corrections were
made for analytical blanks or similar analytical adjustments, other than as reported by each
individual city research team.
4.3 DESIGN DIFFERENCES
Table 4-5 describes the design differences among the three studies. While considerable
effort was made to coordinate the study designs so as to assure the highest possible degree of
comparability among study results, the investigators in the three cities faced different design
issues that precluded carrying out completely identical or equivalent studies. Thus, although
participant recruitment and certain other aspects were similar across the three cities, some
salient differences are also worth noting.
The first difference was that there were different levels of remediation or treatment
among the cities. Boston used two comparison or reference groups in addition to the soil
abatment group, whereas Baltimore used only one such group. In the Cincinnati study, there
were three levels of intervention. Also, the trigger level for soil lead removal varied
somewhat across the cities. In the Baltimore and Cincinnati, a maximum level of 500 ppm
or greater in the parcel or residential property triggered soil removal. In contrast, all Boston
yards from which soil was removed initially had soil lead much higher than 500 ppm, most
in excess of 1,000 to 2,000 ppm. Properties recruited in the Boston study were scattered
across four large neighborhoods or urban areas, although households were assigned at
random to the treatment group for soil removal and not specifically limited to any given
neighborhood. The Baltimore study was carried out in two large neighborhoods, with soil
lead removal restricted to only one of the neighborhoods (Lower Park Heights). Most
houses above the soil lead trigger level in the Lower Park Heights neigborhood in the
September 1, 1995
4-16
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 4-5. DESIGN DIFFERENCES BETWEEN THE THREE STUDIES
1
2
3
4
Design Feature
Number of treatment groups
Number of rounds with blood Pb
measurement
Interval between abatement and final
Pb measurement (months)
Soil removal trigger level 0*g/g)
Paint stabilization
Number of neighborhoods
Participant recruitment
Treatment assignment to participants
Control groups with no intervention
Age structure of participants (%)
Ethnicity (%)
Black
Hispanic
White
Other
Male/female ratio
Blood sample collection
blood
0-1
1-2
2-3
3-4
4-5
5-6
6-»
Rl
R2
R3
R4
R5
R6
R7
Boston
3
4
22
1,000
Interior
4
Volunteer
Random
No
2.7
24.0
34.0
34.7
4.7
51
15
7
27
. 47/53
1-2 mo preabate
3-4 mo after Rl
10 mo after Rl
22 mo after Rl
Baltimore had yard soil removed, but some did not, and
soil removed. The Cincinnati study was
and exterior dust removal only
Baltimore
2
6
10
500
Exterior
2
Volunteer
By Neighborhood
No
8.6
17.6
18.1
18.4
20.3
14.5
2.5
100
0
0
0
48/52
24 mo preabate
12 mo preabate
5-8 mo preabate
8-10 mo after R3
14-16 mo after R3
18-20 mo after R3
Cincinnati
3
5
20
500
None
6
Volunteer
By Neighborhood
Yes
29.9
17.2
17.6
15.8
14.0
5.4
97
0
2
1
44/56
1-2 mo preabate
3-4 mo after Rl
11 mo after Rl
16-18 mo after Rl
22-24 mo after Rl
no house in Walbrook junction had
carried out in six smaller neighborhoods, with soil
carried in the Pendleton
study, all parcels in Pendleton above the
neighborhood. In
the Cincinnati
soil lead trigger level had soil removed.
September 1, 1995
4-17
DRAFT-DO NOT QUOTE OR CITE
-------
1 Paint was stabilized inside all Boston houses and outside all Baltimore houses, but not
2 in Cincinnati where it was believed that only gut-rehab houses had been recruited into the
3 study. No Baltimore residence received interior abatement, either of dust or lead paint,
4 whereas as the majority of the residences in the Boston and Cincinnati studies received
5 interior dust abatement whether or not they were in the soil removal treatment group.
6 Demographic differences among study populations should also be noted, The age
7 distribution of children at the tune of abatement differed among the three studies. The
8 Baltimore group had more children of age at least four years, since many of the children had
9 been initially recruited up to 2 years earlier. Almost all of the children initially recruited hi
10 the Baltimore study were of African-American ancestry; by the final phase of the study, 100
11 percent of the study group was African-American. The Cincinnati study group was slightly
12 more diverse, with a small percentage of Caucasians of Appalachian origin. The Boston
13 group was the most diverse, with substantial subgroups of white and Cape Verdean children,
14 and also with a large percentage of African-American children. Percentages of male and
15 female children differed somewhat among the cities. While all of these inner city households
16 tended to be economically disadvantaged, the majority of the households in Baltimore were
17 occupied by the property owner, which was uncommon in the other two cities.
18 Lastly, as for biological measurements indexing changes in lead exposure, each study
19 involved collection of preabatement and postabatement blood samples and their analyses.
20 However, the numbers of sampling points varied across the studies. The studies had four to
21 six rounds of blood lead collection, with one to three pre-abatement rounds, a short-term
22 post-abatement round (about two or three months), and two to three rounds up to two years
23 post-abatement.
24
25
26 4.4 INDIVIDUAL STUDY CONCLUSIONS
27 In their report following the first phase of their study, the Boston group stated their
28 conclusions:
29 "...this intervention study suggests that an average 1,856ppm reduction in soil
30 lead levels results in a 0.8-1.6 ^g/dL reduction in the blood lead levels of urban
31 children with multiple potential sources of exposure to lead."
32 Following the second phase of the study, they concluded (Aschengrau et al., 1994):
September 1, 1995
4-18
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
"The combined results from both phases suggest that a soil lead reduction of
2,060ppm1 is associated with a 2.2 to 2.70 ng/dL decline in blood lead levels."
The basis for their initial conclusions consisted of an analysis of variance comparing
mean blood lead changes among the three intervention groups, paired t-tests for within group
effects, and analysis of covariance with one-at-a-time adjustment for age, SES, race, sex,
paint, water, and mouthing behavior. The analysis of covariance was performed using no
transformation of blood lead data, which appeared to be normally distributed.
The conclusions from the second phase of the study are based on additional analyses of
phase one and phase two data using two-way analysis of variance (ANOVA) with repeated
measures. Soil was abated for the two original control groups (BOS PI and BOS P) at the
beginning of phase 2. The reduction in blood lead is based on pre- and postabatement
measurements of all three groups.
The Baltimore group stated their conclusions as follows:
"Statistical analysis of the data from the Baltimore Lead in Soil Project provides
no evidence that the soil abatement has a direct impact on the blood lead level of
children in the study."
"In the presence of lead-based paint in the children's homes, abatement of soil
lead alone provides no direct impact on the blood lead levels of children."
The basis for these statements consisted of an adjusted and unadjusted analysis of
selected covariates. The natural log of the blood lead of children in the treatment group
showed no significant difference from the natural log of the blood lead of children in the
control group, even when adjustments were made for: age, SES, hand lead, season, dust,
soil, sex, weak mouthing behavior, or strong mouthing behavior. These analyses were made
on two sets of data. The first set consisted of all children enrolled in rounds one and six.
The .second group consisted only of children enrolled in all six rounds.
The Cincinnati conclusions can be paraphrased as follows based on their individual
report:
Following interior and exterior dust and soil lead abatement, blood lead
concentrations decreased (in Area A) from 8.9 to 7.0 (21%) but increased to 8.7,
33
34
35
1 This value for soil, 2,060 ppm, cited in their published report, was not adjusted by the Boston group with the
interlaboratory correction factor of 1.037 in Table 3-6.
September 1, 1995
4-19
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
10 months postabatement. Following interior dust abatement alone blood lead
concentrations decreased from 10.6 to 9.2 (13%) four months postabatement and
were 18% below preabatement 10 months postabatement. With no abatement,
blood lead levels decreased by 29 and 6% during these same time periods. Other
comparisons also revealed no effects of the soil or dust abatement.
There was no evidence that blood lead levels were reduced by soil lead or dust
abatement in Area A (with soil, exterior dust, interior dust abatement). There was
a slight reduction (net reduction over control area) of 0.6 ng/dL in Area B that
might be attributed to interior dust abatement. This difference is not statistically
significant.
The basis for the Cincinnati conclusions was a comparison of environmental and blood
lead data for the three treatment groups from Rounds 1, 3, 4, 6, and 7 and of additional
environmental data from Rounds 2 and 5.
September 1, 1995
4-20
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
5. RESULTS OF INTEGRATED ANALYSES
5.1 BASIC STRATEGY FOR EVALUATING ABATEMENT
EFFECTIVENESS
Abatement effectiveness is assessed by comparing changes in critical measurements
before and after abatement. Changes in blood lead levels, in hand lead levels, and hi
household dust lead levels are expected to occur in response to abatement but may also occur
even without environmental interventions. Blood lead concentrations in young children often
increase up to ages 2 or 3 years, which are peak ages for ingestion of soil and dust during
play, and then decrease slowly in older children (U.S. Environmental Protection Agency,
1986; Clark et al. 1988). Hand lead loadings increase steadily with age (Bornschein et al.,
1988). House dust lead levels may increase as changes in sources or exposure pathways
cause change in house dust lead levels to occur.
Each individual report reached its conclusion based partially or entirely on linear
regression using analysis of covariance. With this statistical method, when either or both the
measurement error or sampling error of the independent or predictor variable are unknown,
then the estimated regression effect (reduction of blood lead per unit reduction in soil lead)
may be reduced or attenuated., Part of the potential attenuation attributable to "simultaneous
equation bias" is addressed in this integrated report by the use of structural equation models
so that effects size estimates derived by that method are likely more accurately characterized.
This integrated assessment also addresses the question of whether there are effects of
intervention other than soil abatement that might reduce childhood lead exposure. Some of
these Intervention strategies, such as paint stabilization, interior dust abatement, and
neighborhood level exterior dust abatement, were used in this project and an evaluation of
their effectiveness is also reported below.
Finally, this report contains some information on the reliability of childhood lead
exposure measures other than blood. In this respect, data on handwipes and house dust are
interpreted as predictors of childhood lead exposure.
September 1, 1995
5-1
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
5.1.1 General Discussion of Conceptual Approaches
5.1.1.1 Basic Strategies for Evaluating Abatement Effectiveness
Childhood blood lead concentrations are, to some extent, a measure of the recent
history of lead exposure and may respond to environmental changes in lead within a time
frame of a few months. Reductions in blood lead due to reductions in exposure might be
somewhat attenuated by the remobilization of lead in bone tissue as shown in Figure 5-1.
This figure shows the complexity of biokinetic translocations of lead when the total body
burden is decreasing. If the total lead exposure of the child decreases, there seems to be no
doubt that the blood lead concentrations would decrease, but measurements of this decrease
would be complicated by the remobilization of bone tissue lead, and interpretation of these
measurements would be complicated by the uncertainty that the reduction in exposure might
not be fully attributable to reductions hi soil lead exposure.
Changes in blood lead must be interpreted in the context of four time-dependent effects
that are independent of each other as follows:
(1) the typical seasonal changes in children's blood lead concentrations, found
hi virtually every longitudinal study, that usually indicate a peak in
concentration during the late summer months;
(2) the changes that occur with age during early childhood that usually peak
between 18 and 27 months;
(3) long-term changes in national baseline levels of exposure, believed to be
mostly from reductions of lead in gasoline and in food, that are reflected
hi a downward trend for childhood blood lead levels observed since 1978;
or
(4) changes that can be attributed to interventions of this project.
Several different analytical strategies may be used to evaluate the effectiveness of lead
abatement or intervention methods: comparison of simple changes for different treatment
groups; comparison of adjusted changes among different treatment groups where the
adjustment normalizes the preabatement treatment and control groups; and comparison of
adjusted changes among different treatment groups where the adjustments both normalize the
groups to a common starting point and account for different rates of change during the study.
These strategies could be applied to any of the lead measurements used to compare abatement
September 1, 1995
5-2
DRAFT-DO NOT QUOTE OR CITE
-------
Combined Preabatement
Steady-state Blood Lead
Lead in Blood
from Bone
3
Combined Postabatement
Blood Level-A+B + C
A - Rapid Elimination
of Stored Lead
from Soft Tissue
B - Slow Elimination of
Stored Lead from Bone
^ Postabatement Steady-state
r ir ir ir i^r^S:.:?.;;? ^ ^ ™ " " ™ ™====:
~~~^ C - Buildup to •
Postabatement Steady-state
.Jrom Postabatement Exposure
Abatement Year 1
Year 2
Time
Figure 5-1. Hypothetical representation of the expected decrease in blood lead, (solid
curved line) following abatement. This rate of decrease is less than might
be expected from exposure reduction alone. This is because blood also
contains lead recently released from storage in bone and soft tissue.
1 effectiveness: blood lead concentration, hand lead loading, dust lead concentrations, dust lead
2 loading, or soil lead concentration. Each of these three analytical strategies represents a
3 different perspective on the importance of the components of the entire exposure pathway and
4 on the possible changes that may occur, either as a consequence of intervention or because of
5 other unplanned changes during the course of the study.
6 In the simplest approach, the best comparisons are the lead variables before and after
7 the abatement was carried out. In general, the lead levels would be expected to be different,
8 with or without abatement, so that it is necessary to compare the changes that occurred in the
9 soil or dust abatement groups with the change that occurred in the nonabatement groups.
10 The statistical methods that would commonly be used here are paired-sample tests, looking at
11 the difference between the lead levels or logarithms of lead levels before and after abatement.
September 1, 1995
5-3
DRAFT-DO NOT QUOTE OR CITE
-------
1 If the lead levels are measured at more than two tune points or phases, then a simple
2 repeated measures analysis of some sort would be used.
3 The second analytical strategy recognizes that the treatment groups may not be entirely
4 equivalent to each other. It would therefore be necessary to adjust the "starting line" for
5 different groups to a common baseline so that all subsequent comparisons could be made as
6 if everything else were equal, except for the experimental interventions or treatments. Some
7 of the initial adjustment factors could also be lead related variables. For example, the
8 comparison of blood lead concentrations may need to be adjusted for differences in soil lead
9 concentrations in different yards, because one would expect (everything else being equal) that
10 children who live hi houses with higher soil lead would start with higher blood lead
11 concentrations than children who started in houses with lower soil lead. Similarly, it may be
12 useful to adjust for other nonlead factors such as the child's age. Repeated measures
13 analyses with adjustments for covariates (multiple regression or multivariate general linear
14 model) are appropriate statistical methods for carrying out the second strategy.
15 The Boston study offers the fewest complications in using the second strategy, because
16 treatments were randomly assigned to houses and there is little reason to believe that there
17 may be some intrinsic confounding effect between treatment group and either blood lead or
18 environmental lead. Adjustments for environmental lead as covariates should therefore
19 clarify comparisons of the effectiveness of different treatments for individual children in the
20 Boston study. The Baltimore and Cincinnati studies are more difficult to interpret, because
21 the treatment groups were assigned by geographical area or location, not randomly selected
22 from within the same group. There were substantial differences in soil lead and dust lead
23 concentration between neighborhoods.
24 Several comparisons could be carried out using the second strategy. These include:
25 comparisons of treatment group effect on blood lead concentration, adjusted for initial hand
26 lead, dust lead, and soil lead; comparisons of treatment group effect on hand lead, adjusted
27 for initial differences in dust lead and soil lead; comparisons of treatment group effect on
28 dust lead, adjusted for initial differences in soil lead; and even comparisons of soil lead
29 before and after treatment, to determine whether soil lead in the soil lead abatement group
30 remained at reduced levels or was recontaminated.
September 1, 1995
5-4
DRAFT-DO NOT QUOTE OR CITE
-------
1 The third strategy uses structural equation modeling to combine the seemingly unrelated
2 tests of the changes in blood lead and other lead variables. The basis for testing the changes
3 simultaneously is the assumption that current blood lead and environmental lead levels reflect
4 recent lead exposure, and that changes in exposure will lead to changes in lead levels further
5 along the pathways from source to child. The appropriate statistical methodology for this
6 strategy involves testing group differences in models with simultaneous equations for
7 different environmental lead variables. Separate model equations would be needed for dust
8 lead concentration and for total dust loading.
9 Key characteristics of each of the three strategies are illustrated graphically in
10 Figures 5-2 through 5-4. Figure 5-2 shows four separate models for blood lead, hand lead,
11 dust lead, and soil lead, as they would be tested using Strategy 1. Figure 5-3 extends each
12 of these to models with covariate adjustments as the most detailed implementation of Strategy
13 2. The third strategy is illustrated in Figure 5-4. The interconnected nature of the lead
14 measurements over time is shown explicitly, reflecting the hypothesis that changes in dust
15 lead, hand lead, and blood lead are quantifiable effects of changes in lead source terms such
16 as lead in soil and lead in paint.
17 In their individual reports, all three research teams used Strategy 1 as their primary
18 statistical tool and the main basis for their conclusions. The Boston and Baltimore teams also
19 reported results of statistical analyses using Strategy 2, and the Cincinnati group used
20 structural equation modeling to report some of their results.
21 The statistical analyses conducted as part of this EPA integrated assessment were aimed
22 at addressing the following questions:
23
24 • DID THE ABATEMENT OR INTERVENTION HAVE AN EFFECT? This
25 hypothesis is tested statistically by.the interaction between the intervention group and
26 the phase or year. If the statistical significance or P value of the interaction terms is
27 larger than a conventional value such as 0.05, one would conclude that there is no
28 effect of the abatement or intervention (parallel group mean profiles not significantly
29 different).
30
31 • WAS THE EFFECT IN THE EXPECTED DIRECTION? Abatements and other
32 interventions are expected to reduce blood lead, hand lead, or dust lead levels more
33 than in nonabatement or control groups. That is, if group 1 is the control group and
34 group 2 is the intervention group, one would expect pre- versus postabatement
35 differences in the treatment (intervention) group to be larger than the pre- versus
36 postabatement difference in the control group.
September 1, 1995
5-5
DRAFT-DO NOT QUOTE OR CITE
-------
Blood Pb - Before
Abatement
Blood Pb - After
Hand Wipe Pb - Before
Hand Wipe Pb - After
Dust Pb - Before
Dust Pb - After
Soil Pb - Before
Soil Pb - After
Figure 5-2. A simple approach that compares lead variables before and after abatement
comparable to Strategy 1.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
• WAS THERE AN OVERALL DIFFERENCE BETWEEN PHASES? This
hypothesis is tested statistically by the mean within-subject difference between the
preabatement and postabatement groups averaged across all intervention groups. If the
statistical significance or P value of the phase term is larger than a conventional value
such as 0.05, one would conclude that there is no difference in overall level over time.
As noted above, lead levels are expected to change over time with or without
interventions.
• WAS THERE AN OVERALL DIFFERENCE BETWEEN GROUPS? This
hypothesis is tested statistically by the mean between-group differences averaged across
preabatement and postabatement groups. If the statistical significance or P value of the
phase term is larger than a conventional value such as 0.05, one would conclude that
there is no difference hi overall group mean levels. Group mean lead levels are
expected to differ when different interventions are associated with different
neighborhoods, as in Baltimore and Cincinnati.
• WAS THERE A CHANGE IN THE RELATIONSHIP BETWEEN THE RESPONSE
VARIABLE AND THE COVARIATES AFTER ABATEMENT? Many factors affect
blood lead, hand lead, dust lead, dust loading, and other indicators of lead exposure.
Blood lead depends on hand lead and on environmental lead exposure indices, dust lead
depends on lead in soil and paint, and so on. Blood lead may also depend on child age,
on behavioral variables such as the frequency of outdoor play, on
September 1, 1995
5-6
DRAFT-DO NOT QUOTE OR CITE
-------
Blood Pb -
Before
Abatement
I .
Hand Wipe Pb - Before
Dust Pb -
Before
Soil Pb -
Before
Hand Wipe Pb - Before
Dust Pb -
Before
Soil Pb -
Before
Blood Pb
-After
Hand Wipe Pb-After
Dust Pb
-After
Soil Pb
-After
Hand Wipe Pb-After
Dust Pb - After
Soil Pb
-After
Dust Pb -
Before
Soil Pb -
Before
Soil Pb -
Before
Dust Pb - After
Soil Pb
-After
Soil Pb-After
Figure 5-3. A more complex approach that uses covariate adjustments with repeated
measures analysis, comparable to Strategy 2.
1
2
3
4
5
6
7
8
9
10
household socioeconomic indicators such as parental education, and on demographic
factors such as race or ethnicity. These factors may modify the effectiveness of
abatement. One way to test for this is to include the covariate in the analysis as an
adjustment factor so that the baseline levels can be tested as if all children started out at
the same level. A similar argument may apply to adjustments of postabatement blood
lead. The effect of the covariate may be assumed to have changed over the course of
abatement (possibly as a consequence of abatement) if the ^three-way interaction between
the treatment group, the phase of the study, and the covariate is statistically significant.
September 1, 1995
5-7
DRAFT-DO NOT QUOTE OR CITE
-------
X
X
*
Blood Pb - Before
t
Hand Wipe Pb - Before
t
Dust Pb - Before
t
Soil Pb - Before
Abatement
1
/
f
^
Blood Pb - After
f
Hand Wipe Pb - After
f
Dust Pb - After
f
Soil Pb - After
Figure 5-4. A structural equation modeling approach comparable to Strategy 3.
1 5.1.1.2 Expected Impact of Intervention
2 Impact of Soil Abatement on Exterior and Interior Dust
3 The key to understanding the impact of soil (and exterior dust) abatement on interior
4 dust is to observe changes in the three components of the interior dust measurement: lead
5 concentration (micrograms of lead per gram of dust), lead loading (micrograms of lead per
6 square meter), and dust loading (milligrams of dust per square meter). Where there was no
7 interior dust abatement, the lead concentration in interior dust should decrease gradually over
8 time, provided mat the influence of lead-based paint has been minimized. Also, the lead
9 loading should decrease if the dust loading remains constant or the lead loading is normalized
10 to dust loading. This normalization is believed to correct for differences in housekeeping
11 efficiency. If interior dust abatement has occurred, the lead concentration should decrease
12 markedly and remain low where the influence of lead-based paint is minimal, and the lead
13 loading and dust loading should decrease and then increase in tandem.
14 The impact of lead-based paint can be minimized in three ways: (1) observe only cases
15 where there is no lead-based paint; (2) stabilize the paint so that the rate of incorporation to
16 house dust is minimized; and (3) compare measurements where the influence of lead-based
17 paint is probably high relative to soil to areas where the influence of soil is high. A crude
September 1, 1995
5-8
DRAFT-DO NOT QUOTE OR CITE
-------
1 measure of the rate of recontamination of house dust from lead-based paint can be observed
2 from the changes in window well dust lead concentrations following interior dust abatement,
3 for units with and without lead-based paint.
4 The analysis of three types of internal dust measurements, (1) entry, (2) floor, and
5 (3) window well, can provide additional information about the impact of soil abatement. The
6 entry measurement probably shows the greatest influence of exterior lead from soil and dust.
7 If the entryway to the housing unit is somewhat removed from the building entrance, such as
8 an apartment on the second or third floor, then a comparison of these two measurements
9 should demonstrate the effect of soil lead on multifamily houses. Likewise, where ulterior
10 dust abatement has taken place, the rate of recontamination of interior dust should be
11 entry > floor > window well.
12 Exterior dust was measured and abated in Cincinnati only. In this study, the results
13 suggest a recontamination rate for exterior dust of less than two weeks, and that the source
14 of this recontamination is not the soil. With a neighborhood level perturbance of this type, it
15 is not possible to measure the impact of soil abatement on house dust directly. However, if
16 abatement is considered on the broader scope, where neighborhood cleanup would include
17 soil, external dust, and any other sources of lead external to the home, then the house dust
18 measurements made immediately inside the homes can be used as a measure of this "total
19 neighborhood abatement". For those cases in the Cincinnati study where there was no
20 immediate recontamination of this entryway dust, this measurement may sometimes be used
21 as a surrogate for soil abatement. To make this determination, it is also necessary to
22 evaluate the fraction of exposure that would derive directly from soil or from playground
23 dust, which would not be included in the interpretation of house dust alone.
24
25 Impact of Soil and Dust Abatement on Hand Lead Loading
26 It was expected that hand dust would serve as an surrogate measure of changes in
27 exposure following abatement to augment information about blood lead changes. Hand dust
28 reflects the child's recent exposure (since the latest hand washing), but is a measure only of
29 lead loading, not lead concentration or dust loading, because the total amount of dust is not
30 measured. Consequently, it is not possible to determine the source of lead (soil or paint) by
31 differences in concentration, nor is it possible to correct for housekeeping effectiveness by
September 1, 1995
5-9
DRAFT-DO NOT QUOTE OR CITE
-------
1 observing changes in dust loading, as with house dust. It seems plausible that the amount of
2 dust (not mud or dirt) on the hand reaches equilibrium after a short period of time, perhaps
3 30 min to 2 h. The dustiness of the house would affect only the rate at which this
4 equilibrium is reached, not the total amount of dust at equilibrium.
5
6 Impact of Soil and Dust Abatement on Blood Lead Concentrations
7 Blood lead concentrations should respond to soil and dust abatement through the impact
8 of abatement on two routes of exposure: (1) hand-to-mouth activity, reflecting the impact of
9 interior house dust and exterior play area dust on exposure; and (2) food contamination,
10 reflecting the incorporation of house dust in food during kitchen preparation. There was no
11 measure of the incorporation of house dust into food during this project. Intuitively, the
12 impact of interior dust abatement should be the same, or at least comparable, for food and
13 hand dust. In some homes, however, lead-based paint is more common in kitchens and
14 bathrooms, and the rate of return of dust from lead-based paint following stabilization would
15 have a greater impact on food than hand dust. There is a limited amount of data, not yet
16 analyzed, where kitchen floor dust can be compared to bedrooms and other living areas, and
17 likewise for window wells. Most of these data, however, are from the Cincinnati study,
18 where there was a minimum influence of lead-based paint.
19 The Baltimore study showed no influence of soil abatement on blood lead
20 concentrations. The Baltimore study did not measure the impact of soil abatement in the
21 absence of interior lead-based paint, and it is possible that soil abatement would be swamped
22 by the presence of paint lead in the house dust. This negative result is an important finding
23 of this study and the integrated project that suggests, in the absence of interior dust
24 abatement and interior paint stabilization (or abatement), soil, exterior dust, and exterior
25 paint abatement will have little impact on childhood lead exposure.
26 The Cincinnati study showed no effect of soil abatement alone on the blood lead
27 concentrations, but showed a positive effect of interior dust abatement and a marginal effect
28 of total abatement when the interior-entry dust immediately inside the home was used as a
29 surrogate of neighborhood lead abatement. The importance of these findings is that when the
30 sources of lead that recontaminate exterior dust can be identified and abated, the impact of
31 neighborhood-level abatement will be greater than single dwelling unit abatement alone.
September 1, 1995
5-10
DRAFT-DO NOT QUOTE OR CITE
-------
1 Effect of Lead Abatement or Intervention on Blood Lead Over Time
2 One of the most important limitations in carrying out a longitudinal lead abatement or
3 intervention study over time is that reductions in blood lead are limited to some fraction of
4 the total amount of lead stored in the child's body prior to abatement. Even if lead-burdened
5 children were completely removed from lead exposure, a significant amount of lead would
6 still be present hi the child's blood due to the slow release of lead from the large amounts
7 stored in the body, mostly in the bones. Autopsy data show that as much as 60 to 70% of
8 the lead in a child's body is stored in the skeletal system, especially in the hard (or cortical)
9 part of long bones such as the femur and the tibia (Barry, 1981). In adults this percentage is
10 even larger, 90 or 95%. Lead is retained in cortical bone for many years, and even though
11 bone remodeling hi young children is very rapid, these large body burdens contained in the
12 bone constitute a significant internal source of lead exposure for several years after exposure
13 has stopped.
14 The persistence of elevated blood lead concentrations has some important public health
15 implications. No matter how effective the environmental intervention, children can be
16 expected to retain a fairly high fraction of their initial blood lead concentration for a period
17 of several years. Because the health effects of lead exposure are believed to be cumulative,
18 increasing as the total internal dose (years of exposure tunes micrograms per deciliter of
19 blood lead), there may be substantial postremediation internal exposure and consequent health
20 effects even after a successful intervention.
21 Reduction of environmental lead exposure should not be expected to produce a
22 complete reduction of elevated blood lead levels attributable to the preabatement exposure.
23 Blood lead levels are expected to be more persistent when there is long-term exposure to
24 higher preabatement environmental lead from any source or medium. Much of the lead hi
25 the blood is distributed to other tissues before being eliminated from the body. Lead is
26 avidly accumulated in the child's skeletal tissues, along with calcium needed for further
27 growth and development. However, lead is released only very slowly from skeletal tissues,
28 and this skeletal lead burden may become an internal source of blood lead even after the
29 source of the lead exposure has been removed. Therefore, the postabatement blood lead
30 level will not only reflect exposure to the new postabatement environmental lead levels, but
31 will also in part reflect retention of skeletal lead from historical preabatement exposure. The
September 1, 1995
5-11
DRAFT-DO NOT QUOTE OR CITE
-------
1 long-term stability of blood lead levels in a stationary exposure environment has been noted
2 by a number of authors (David et al., 1982; Rabinowitz, 1987).
3 Persistence of elevated blood lead after abatement has both biological and
4 environmental components. The biological component is the resorption of skeletal lead.
5 In adults, recent stable lead isotope studies (Smith et al. 1995) suggest that 30 to 65% of the
6 circulating lead in adults is due to skeletal lead, which is consistent with other estimates.
7 Although a somewhat lower percentage may be appropriate for children rather than adults, it
8 is clear that even in children a substantial fraction of blood lead has a skeletal origin.
9 The environmental component of persistence is the child's remaining exposure to other
10 nonremediated lead media, such as lead in diet, drinking water, or air. This was illustrated
11 in Figure 5-1, which shows a blood lead profile (for an individual, or possibly as a
12 population mean) before and after a hypothetical lead abatement. The steady-state blood lead
13 concentrations are shown as flat curves, although in reality there may be substantial age-
14 dependent changes during the course of abatement even when environmental lead
15 concentrations remain constant. Assuming that environmental concentrations remain constant
16 after abatement (they may not; see below), the child's blood lead would eventually reach a
17 new steady-state concentration at a much lower level. At any given time after abatement, the
18 child's blood lead is a mixture of three components, denoted "A", "B", and "C" in
19 Figure 5-1. Component A shows the relatively rapid decrease in blood lead from elimination
20 of preabatement lead deposits in blood and soft tissues. Component B shows the contribution
21 of preabatement skeletal lead to post-abatement blood lead, which is much slower because the
22 large skeletal burden hi cortical bone is eliminated on a time scale of several years. Almost
23 all of the stored lead will eventually be eliminated. However, the contribution of
24 preabatement deposits of lead now stored as an internal source of exposure may be
25 quantitatively significant compared to remaining postremediation environmental exposure
26 media.
27 The combination of persistent internal exposure and persistent baseline external
28 exposure amounts to a post-abatement blood lead contribution of about 50 or 60% of the
29 preabatement blood lead starting value at 8 to 12 months after abatement. This means that
30 any environmental abatement or intervention can achieve at most a 40 to 50% reduction in
31 child blood lead concentrations within a year after abatement (see Figure 5-1).
September 1, 1995
5-12
DRAFT-DO NOT QUOTE OR CITE
-------
1 Soil Lead Remediation Effects Modeled by Environmental Pathways for Lead
2 Soil lead remediation in residential yards is expected to have both direct and indirect
3 effects on childhood lead exposure. The direct effect of removing lead contaminated soils is
4 to deny access to the lead in the soil. However, most children do not eat large quantities of
5 soil. Some children may regularly ingest a large amount of soil (a condition known as pica
6 for soil), and some adults are known to experience geophagia, but these are untypical
7 conditions and are not appropriate for assessing soil risks for the majority of children. For
8 most children, direct exposure to lead in soil is likely to come from fine particles of loose
9 soil or exterior surface dust that adhere to the child's hands and are transferred to the child's
10 face and mouth during hand-to-mouth contact that is part of normal behavior for preschool
11 children and infants.
12 The larger part of the contribution of lead in soil is as a source of lead in household
13 dust. Soil hi the residential yard may be tracked into the house by its occupants (including
14 pets), and fine exterior dust particles may become re-entrained and carried into the house as
15 micro-scale air contaminants. Fine dust particles may adhere to the child's hands, and may
16 contaminate food during its preparation. Dust is usually a more important medium of lead
17 intake than is soil. This is an indirect soil lead exposure pathway, from soil to house dust to
18 the child's blood.
19 It is therefore necessary to model lead exposure through multiple pathways or exposure
20 media hi order to accurately characterize the complete effects of soil abatement. Time-
21 dependent modeling of changes in environmental media and exposure pathways is a parallel
22 process to time-dependent modeling of blood lead changes as noted in the preceding
23 subsection.
24
25 5.1.2 Conceptual Approach to Differences in Group Means
26 The basis for simple analyses of abatement effectiveness is comparison of changes hi
27 mean blood lead in groups of children who received different interventions. The basis for
28 interpreting such tests will be discussed before any formal statistical techniques are applied.
29 Figure 5-5 sketches the probable outcomes of a soil abatement study (in general, any
30 intervention study). All of the studies assigned a control group who received no soil lead
September 1, 1995
5-13
DRAFT-DO NOT QUOTE OR CITE
-------
m
Outcome C1
•AC
Outcome A1
Control Group
Outcome C2
Soil Abatement Group /
Outcome A2
Outcome C3
•AC
t
Outcome A3
AA
Figure 5-5. Schematic representation of expected outcomes for treatment and control
groups.
1 abatement during the first year of the study. This is shown by a flat line connecting soil lead
2 measured before (denoted B) and after (denoted A) the abatement period, because soil lead
3 concentrations are expected to show little decrease during a year or two of study. The
4 probable responses of blood lead are either no change in blood lead (denoted outcome Cl) or
5 a measurable decrease hi blood lead (denoted outcome C2). The straight lines in outcomes
6 Cl and C2 connect mean blood lead measured in the control group before (denoted Bc) and
7 after abatement (denoted Ac). Similar results could conceivably occur in the soil abatement
8 group, whose outcomes are denoted Al and A2, and whose observed mean blood lead before
9 and after abatement are denoted BA and AA respectively.
10 Figure 5-6 shows all possible combinations of outcomes for the control group and the
11 ' abatement group that could lead to different conclusions. The preabatement blood lead
12 concentrations of these groups are shown as possibly different, because in the Baltimore and
13 Cincinnati studies the soil abatement group was in a distinctly different neighborhood from
September 1, 1995
5-14
DRAFT-DO NOT QUOTE OR CITE
-------
1 the intended control group and had a different mean blood lead. Outcomes Cl and Al
2 occurring together show that blood did not change in either the soil abatement group or the
3 control group, suggesting that there was no effect of the abatement. Outcomes Cl and A2
4 occurring together show that blood decreased in the soil abatement group and did not change
5 in the control group, suggesting that there was a beneficial effect of the abatement.
6 Outcomes C2 and Al occurring together show that blood lead did not decrease in the soil
7 abatement group and did decrease in the control group, suggesting that there might be a
8 possible negative effect of the abatement compared to doing nothing that was not done for the
9 control group. Outcomes C2 and A2 occurring together show that blood decreased in both
10 the soil abatement group and in the control group, but the nature of the effect depends on the
11 magnitude of the changes between the two groups, which are denoted as Types 1,2, and 3
12 changes. In Type 1, blood decreased by the same amount in both groups, suggesting no
13 effect of abatement. In Type 2, blood decreased by a greater amount hi the abatement group
14 than in the control group, suggesting a beneficial effect of abatement. In Type 3, blood
15 decreased by a greater amount in the control group than hi the abatement group, suggesting a
16 possible negative effect of abatement. Again, these are hypothetical outcomes that illustrate
17 the possibilities in interpreting the results of a longitudinal study. It is clearly not adequate
18 to look at changes hi blood lead in a single treatment group hi the absence of an appropriate
19 reference group or control group.
20
21 5.1.3 Conceptual Approach to Pre- and Postabatement Differences in
22 Individuals
23 A potential problem arises in simple comparisons of group mean values during a
24 longitudinal study when different individuals are present at different phases of the study. For
25 example, some individuals in the preabatement phase of the study may have dropped out by
26 the tune of the postabatement phase, whereas other individuals who were not in the
27 preabatement phase may have been recruited into the postabatement phase (e.g., infant
28 siblings who reached enrollment age status during the study). Although it would be
29 reassuring to think that attrition and recruitment do not depend on the treatment group, and
30 that children lost or gained during the progress of the study are no different from those
September 1, 1995
5-15
DRAFT-DO NOT QUOTE OR CITE
-------
OUTCOME
C1.A1
Type 4
C1.A2
TypeS
C2.A1
TypeB
C2.A2
Typel
C2.A2
Type 2
C2.A2
TypeS
Blood Pb
Blood Pb
Blood Pb
Blood Pb
i Blood Pb
* B
i Blood Pb
AC
INTERPRETATION
No Effect of Abatement
Positive Effect of Abatement:
Reduces Blood Lead
Negative Effect of Abatement:
Reduces Blood Lead Less
Than Would Otherwise
Have Occured
No Effect of Abatement
Relative to Control
Positive Effect of Abatement:
Reduces Blood Lead More
Than Would Otherwise
Have Occured
Negative Effect of Abatement
Reduces Blood Lead Less
Than Would Otherwise
Have Occured
Figure 5-6. Schematic representation of the potential interpretations that might be
reached from the various abatement outcomes.
1
2
3
4
5
6
7
8
9
10
11
enrolled throughout the study, this cannot be guaranteed. One of the simplest solutions is to
limit the analyses to children who were present during all phases of the study.
When the analyses are restricted to subjects with both pre- and postabatement data, then
abatement effectiveness may be assessed by simply taking differences of blood lead
concentrations or differences of their logarithms. Unfortunately, blood lead differences
ignore the intrinsic persistence of blood lead concentrations over time. The only part of the
preabatement blood lead concentration that can be reduced by intervention is the
nonpersistent part,
removable blood lead = fraction of preabatement blood lead
September 1, 1995
5-16
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
where the fraction for one year postabatement may be about 50%. The difference between
preabatement and postabatement blood lead cannot be larger than the amount of removable
blood lead. In other words,
preabatement — postabatement blood lead < fraction of preabatement blood lead.
This suggests that a better index for abatement effectiveness might be a partial difference:
postabatement — (1 — fraction) preabatement blood lead > 0.
Unfortunately, the value of this fraction is not known well enough to define a priori the
partial difference for use as an index of lead effectiveness, because the1 value of the retained
fraction of lead depends on the time since abatement and the child's age, and probably on
other factors as well.
5.1.4 Conceptual Approaches to Repeated Measures Analyses
The simple comparison of typical values of blood lead concentrations among treatment
groups at different phases of these longitudinal studies has certain limitations that may not be
obvious to the reader. These limitations are the same whether blood leads are characterized
by the group mean, geometric mean, median or other percentile values. The first is that
some of the children in any treatment group are probably not exactly the same children at
one phase of the study as at a subsequent phase. Some children will almost certainly be lost
to follow-up by moving or by refusal to participate (normal processes of attrition in
longitudinal studies), whereas other children may be added by recruitment (such as at
Round 3 in the Baltimore study) or as additional members of households where other
children are already enrolled in the study. Since children who are lost to follow-up or who
are added to the study may differ in some systematic ways from children who were retained
throughout the study, it may be prudent to analyze data from these children who were not
present separately from those who were present at all relevant phases. On the other hand, if
study results are restricted only to children who were present at certain specific pre- or
postabatement phases of the study, then repeated measurements on the same child at
September 1, 1995
5-17
DRAFT-DO NOT QUOTE OR CITE
-------
1 different phases of the study are not statistically independent of each other. Although data
2 from one treatment group at a given phase are independent of data from a different group,
3 data on the same group at a different phase are not independent of data from an earlier
4 phase.
5 Data from the same individual at different phases of a study can be analyzed as
6 "repeated measurements" techniques. "Repeated measurements analyses" is a statistical term
7 usually applied to a certain kind of mixed model multivariate analysis of covariance in which
8 it is assumed that there are several distinct kinds of predictors for the response variable (such
9 as blood lead):
10 (i) Repeated observation phases (for example, pre- and postabatement rounds);
11 (H) Within-individual non-random differences or fixed effects attributable to specific
12 covariates (for example, hand lead or dust lead loading at each round);
13 (Hi) Within-individual random differences not attributable to specific covariates or
14 treatment groups (random error at each round);
15 ftv) Between-individual non-random differences (fixed effects) attributable to specific
16 treatment groups or between-group covariates (for example, the treatment group
17 could be a control group or soil abatement group or neighborhood, and the
18 average soil lead concentration or percentage of non-gut-rehab houses within a
19 neighborhood could be a numeric covariate);
20 (v) Between-individual random differences attributable to other factors (for example,
21 being in different households or families, when there are some households with
22 multiple children enrolled in the study);
23 (vi) Between-individual random differences not attributable to specific covariate or
24 other factors (a random intercept term).
25
26 Let us provide an explicit mathematical model to illustrate these points. This model
27 will be a linear model of the sort that could be fitted using SAS PROC MIXED or similar
28 statistical programs. We will first define the subscripts corresponding to each case:
29 g = group index, such as neighborhood or treatment group (treatment groups are often
30 denoted RGP for remediation group in the models we used);
31 h = household or other "nested" unit within each treatment group (often denoted FMID
32 in the models we used);
33 I = individual index or identifier (denoted KDID in the models we used);
34 j = round or phase of the study.
35 The generic form of the model is defined as follows:
36 Yy = Ggj + Hh(g) + Ii(gh) + Xy Bgj
37 + ey.
September 1, 1995
5-18
DRAFT-DO NOT QUOTE OR CITE
-------
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
In the above sequence of effects, the response variable for child I at round j is denoted Yy,
and the other terms are identified as follows:
(i) Repeated observation phases, denoted j;
(ii) Within-individual non-random differences or fixed effects attributable to specific
covariates (for fixed effect of predictor X in child I at round j, denoted Xy Bgj);
(Hi) Within-individual random differences not attributable to specific covariates or
treatment groups (denoted ey for child I at round j);
(iv) Between-individual non-random differences (fixed effects) attributable to specific
treatment groups or between-group covariates (denoted Ggj for treatment group g
at round j in this example);
(v) Between-individual random differences attributable to other factors (denoted Hh(g)
for household h in group g in this example);
(vi) Between-individual random differences not attributable to specific covariate or
other factors (denoted Ii(gh) for child I in group g, household h, in this example).
Hypotheses about treatment group effects could be formulated in terms of contrasts,
which are pre-specified linear combinations of group effect estimates, for example:
Difference in group g between rounds j = 1 and j = 2
= Ggi - Gg2;
Difference between groups g=l and g=2 at round j
= Gy - G2j;
Effect of treatment g=2 relative to treatment g=3 between rounds 1 and 4
= G21 — GU ~ (G3i ~ G34)
also = G21 - G31 - (G24 - G34);
Effect of treatment g=2 relative to average of treatments g=l and g=3 between
rounds 1 and 4
= G21 - GM - 0.5 (Gu - G14) - 0.5 (G31 - G34);
Difference in effect of covariate adjustment at round 4 between groups 1 and 2
= B14 — B^ per unit of X.
Several approaches are evaluated for analyzing the longitudinal data from the three
cities using "repeated measures" models. Several convenient computer implementations of
the method are available. We tried three versions and found that in many cases, the ability
to identify differences among interventions was greatly improved by including covariates in
the analyses. For example, child blood lead is known to change with age. When age is
included as a covariate, some of the variation in blood lead differences before and after
abatement can be attributed to the age of the child when the abatement was carried out. This
September 1, 1995
5-19
DRAFT-DO NOT QUOTE OR CITE
-------
1 increases the ability to estimate the relationship between blood lead and other variables, such
2 as soil lead. Similarly, the effect of abatement may depend on changes in proximate
3 exposure variables such as house dust lead. The effects of changes in house dust lead may
4 be different at different ages, however, so that other covariates that may be useful in the
5 analyses include interactions between age, house dust lead, and treatment group.
6 The use of baseline preabatement environmental or demographic measurements as
7 covariates allows one to proceed as if all groups had the same starting values. The use of
8 differences in environmental measurements before and after abatement allows one to proceed
9 as if individuals responded similarly to similar changes in lead exposure, which is a
10 fundamental assumption in a remediation and intervention program. It might even be useful
11 to evaluate treatment effects adjusted only for the final postabatement values of the covariates
12 if one assumed that blood lead differences reflected only the final post-abatement lead
13 exposures. In general, differences hi environmental indices before and after abatement were
14 found to be more predictive of blood lead changes than the absolute baseline or final values.
15 Repeated measures analyses can be carried out using standard statistical programs for
16 analyses of general linear models. PROC GLM in the SAS statistical package (SAS, 1990)
17 and the MGLH procedure in the SYSTAT statistical package (SYSTAT, 1990) were used for
18 most of the analyses. Analyses of repeated measures models with time-vary ing covariates
19 cannot be conveniently carried out using these programs, so some analyses were therefore
20 done using the P2V and P5V programs in the BMDP (BMDP, 1993) statistical package.
21 Repeated measures models with more than two phases or tune points may require specific
22 assumptions about tune correlation structure in some programs, which can be done using
23 generalized estimating equation (GEE) approaches such as that used in some of the Baltimore
24 analyses, but no such assumptions are needed when comparing outcomes at only two time
25 points, pre- and postabatement.
26
27 5.1.5 Conceptual Approach to Structural Equation Modeling
28 Even though statistical models could be based on the partial differences of blood lead
29 levels between pre- and postabatement phases, the environmental exposure variables are
30 themselves more or less correlated with earlier measurements of the exposure variables.
31 This violates one of the most important assumptions about linear regression models, and
September 1, 1995
5-20
DRAFT-DO NOT QUOTE OR CITE
-------
1 generally about linear models such as the analysis of variance and the analysis of covariance.
2 That assumption is that the predictor variables or regressors are known without statistical
3 error. Although the statistical error is usually called "measurement error" (Fuller, 1987), the
4 errors include many other kinds of variability. In environmental epidemiology, the most
5 common measurement errors in exposure include behavior or activity pattern variability,
6 repeat sampling variability, sampling location variability, as well as analytical error. That is,
7 the observed value of the predictor, such as floor dust lead loading, may not perfectly reflect
8 the activity of the child and the child's actual exposure to dust lead over time.
9 One way to deal with this is to predict the precursor exposure variables in an
10 environmental model. For example, suppose that blood lead is predicted by hand lead, soil
11 and dust lead, and by a preceding value of the blood lead. Hand lead may then be predicted
12 by current dust and soil lead levels, and dust lead by current soil lead, so that in addition to
13 the direct effect of soil lead on blood lead, there are indirect effects from soil to dust to hand
14 to blood, and from soil to hand to blood. This approach allows estimation of the
15 measurement error variance in the precursor lead exposure variables in terms of residual
16 deviations between the observed exposure variable and its best estimate from its own
17 precursors. If the model is correct, this approach will essentially eliminate the bias
18 introduced by measurement errors. The usual bias in estimating a regression coefficient or
19 effect size of intervention will be to deflate or attenuate the estimate (i.e., to shrink the
20 estimate towards 0, which reduces both its magnitude and its statistical significance).
21 However, with multiple correlated predictors such as lead soil and dust variables for a single
22 residential premises used in these analyses, this attenuation may not occur (Klepper et al.,
23 1993).
24 Structural Equation Modeling is a computational approach that allows estimation of sets
25 of inter-related linear or nonlinear models (Buncher et al., 1991). This has been widely used
26 for cross-sectional environmental pathway modeling (Bornschein et al., 1985, 1988, 1990;
27 Marcus, 1991, 1992). Applications to longitudinal lead studies have recently been developed
28 (Marcus, 1991; Menton et al., 1994; Marcus and Elias, 1994). PROC MODEL program hi
29 the SAS ETS computer package (SAS, 1992) allows estimation of either linear or nonlinear
30 models. This procedure is believed to result in unbiased or less biased estimates of
31 regression coefficients than other estimation procedures that do not include fitting
September 1, 1995
5-21
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
simultaneous equations for blood lead to predictor variables such as lead in paint, soil, or
dust.
The most complete and technically correct evaluation of these studies requires a
simultaneous assessment of changes hi blood lead levels and changes hi environmental lead
pathways following soil lead or dust lead abatement. Underlying any analysis of tune-
dependent relationships are the following assumptions:
(1) Both preabatement and postabatement blood lead levels reflect, hi part,
contemporary environmental lead exposures that can be characterized by
measurements of lead levels in soil, dust, paint, and other media;
(2) Postabatement blood lead levels may also reflect, in part, preabatement blood
lead levels due to the contribution of preabatement body burdens of lead
(principally hi the skeleton) from earlier exposures;
(3) Postabatement dust lead levels may also reflect, in part, preabatement dust lead
levels due to mixing of incompletely abated or unidentified sources of lead in
dust for which preabatement dust lead levels are a surrogate indicator;
(4) Postabatement soil lead levels may also reflect, in part, preabatement soil lead
levels due to mixing of incompletely abated or unidentified sources of lead in soil
for which preabatement soil lead levels are a surrogate indicator;
(5) Even when lead-based paint has been stabilized, lead paint levels measured by
P-XRF may also help to predict postabatement soil and dust lead levels from
incompletely abated or unidentified sources of lead in soil and dust for which
lead-based paint levels are a surrogate indicator.
These models were fitted using indicator or "dummy" variables for different study or
treatment groups. Sometimes these indicator variables were used as "switches", for example
when postabatement soil lead concentration is modeled as a fraction of preabatement soil lead
for soil nonabatement groups, but as a new replacement value for the soil abatement groups.
At other tunes, indicator variables were used when the data suggested that the effect of
abatement was to modify the regression coefficient for the predicted variable (for example,
floor dust lead concentration) for a pathway. In that case, separate coefficients were fitted to
the product of the treatment group indicator and the predictor variable (for example, entry
dust lead concentration) as well as separate intercept terms for each treatment group. Apart
from this, the underlying assumptions in the Structural Equation Model approach are that
abatement effects can be characterized by concentrations or loadings of appropriate
September 1, 1995
5-22
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
environmental lead exposure variables, a concept that allows inferences about effects of
hypothetical abatements at other levels of lead exposure.
5.1.6 Comparison of Interventions Across Studies
There were substantial differences among the three studies that complicated a direct
comparison of intervention effectiveness. The differences included:
(1) different levels of soil lead abatement and intervention. Although all three studies
excavated soil associated with child exposure, the Baltimore and Boston studies
removed soil hi the yard surrounding the child's home, usually a single detached
dwelling unit. The Cincinnati study had most children in multi-family units, and
removed soil and exterior dust from common play areas and accessible areas in the
neighborhood. The Baltimore study did not include exterior dust abatement,
whereas the Boston and Cincinnati studies were accompanied by substantial interior
dust abatement.
(2) different "control" groups. The Baltimore control group used homes in a different
distant neighborhood than the soil abatement homes. These homes had exterior
paint stabilized in order to avoid further soil contamination, and the soil abatement
group houses also had exterior paint stabilization. There was also a de facto
control group in the soil abatement neighborhood, because houses with soil lead
below 500 ppm were not abated. The Boston control group consisted of houses in
the same neighborhoods as the houses that received soil and dust abatement. The
Cincinnati control group houses received no treatment of any sort, and were
located hi neighborhoods that were some distance away from the abated
neighborhoods.
Other conditions will facilitate comparison of the studies:
(1) all three studies have blood lead measurements that were made in late summer or
early autumn (July to October) during the peak blood lead'season, at least 8 months
after abatement but not more than 15 months afterward;
(2) all studies have baseline or preabatement blood lead levels taken not more than
18 months before the summer-fall postabatement blood lead level in the same child,
so that individual pre- and postabatement differences may be compared;
(3) all studies have hand lead data that were taken at or about the same time as the
blood lead data, and may be used as proximate indicators of actual environmental
soil and dust lead exposure or contact;
(4) all studies have preabatement residential dust lead levels linked to each child, and
preabatement soil or entry-area dust lead levels as indicators of environmental
exposure for each child;
September 1, 1995
5-23
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
(5) all studies have used the same or nearly identical protocols for blood lead and hand
lead sampling and analyses;
(6) soil sampling and analysis protocols are very similar across studies; and
(7) dust lead sampling and analysis were done by somewhat different methods, but
were calibrated to produce comparable dust lead and soil lead concentrations across
all studies.
The application of many hypothesis tests to the same set or subset of data may greatly
distort the overall significance level of the entire decision-making process. This problem of
multiple comparisons can be controlled by testing only hypotheses that are specified in
advance. Because tests of the across-study hypotheses depend on the results of preceding
tests on the pooling of certain groups within studies, the exact number of tunes that each data
set is used in a test cannot be stated, but is not more than six tests. An extremely
conservative approach is to assign experiment-wise significance at level alpha (for example,
alpha = 0.05) only to those tests whose individual test-wise significance is at level alpha /
(number of tests). That is, to assert that all of the results of six tests involving the same data
set are significant at level 0.05, each test should be carried out at level 0.05/6 = 0.0083.
Some authors argue mat this adjustment, which is called the Bonferroni correction, is
exceptionally conservative and that no adjustments are needed for multiple comparisons
(Rothman, 1990). P levels are provided for each test to assist the reader who wishes to form
his or her own judgements of the meaning of the results of the analyses. The decision level
alpha of any statistical test is a subjectively chosen number. For most users of these tests,
the conventional choice of alpha = 0.05 with the conservative decision to use an experiment-
wise Bonferroni adjustment based on five tests per group per variable would suggest a
test-wise level of 0.01 in order to decisively reject the hypothesis of no change, difference,
or effect.
5.2 DIFFERENCES IN GROUP MEANS
5.2.1 Changes in Mean Soil Concentrations
Differences hi group means are presented hi the following set of figures. The subsets
of participants in these figures are not necessarily the same as in comparable presentations in
September 1, 1995
5-24
DRAFT-DO NOT QUOTE OR CITE
-------
1 the individual reports. Therefore, the number of participants may also differ. In the Boston
2 study analyses, we used the same subset of children as in the Boston report, excluding the
3 same two children who had become lead-poisoned. For the Baltimore data, we chose to
4 assign the small group of participants from the treatment group whose properties were not
5 abated to a separate control group, rather than merge them with the main control group. We
6 also report data for all children for a specific round, rather than all children in round one or
7 children in all six rounds, as Baltimore reported. We treat the Cincinnati neighborhoods as
8 individual treatment groups and include all children recruited, except for the four children
9 were undergoing treatment for lead poisoning.
10 The presentation of these group mean data uses a similar format for all of the figures in
11 this series. Each treatment group is represented in each round by a box and whisker plot.
12 Each box has a mark approximately midway that shows the median value for the group and
13 these medians are connected by a line between boxes. The upper and lower ends of the box
14 mark the 3rd and 1st quartiles (75th and 25th percentiles) respectively. The tick marks on
15 the upper and lower whiskers show the location of the 84th and 16th percentiles,
16 respectively. (These two statistics are useful in estimating geometric distributions.) The
17 diamond on the line or in the box shows the location of the arithmetic mean. These
18 statistical parameters are shown in Figure 5-7, expanded for clarity. The data for these plots
19 are given in Appendix A, Table A-l.
20 In order to form an effective, permanent barrier between the source of lead and the
21 human environment, soil abatement must reduce the concentration of lead in the soil in a
22 manner that is persistent for a period of years. In each of the three studies, measurements
23 were made prior to abatement and immediately after abatement (within three months).
24 Followup measurements were made periodically until the end of the study in Cincinnati and
25 Boston. The results of these soil analyses are graphically illustrated in Figures 5-8 and 5-9.
26 These data show, for all three studies, a substantial reduction in the amount of lead in abated
27 soil areas. In Boston and Cincinnati, where follow-up soil measurements were taken, this
28 reduction persisted for the duration of the study. In Baltimore, the postabatement
29 measurements were made only in the locations where soil had been excavated and removed.
September 1, 1995
5-25
DRAFT-DO NOT QUOTE OR CITE
-------
84th Percentile
75th Percentile
Arithmetic Mean
Median
25th Percentile
16th Percentile
Figure 5-7. Hypothetical representation of common statistical parameters for a single
group and a single round.
1 Each study was able to achieve the targeted concentration for abated soil. The median
2 soil concentrations following abatement are not substantially higher than the specifications for
3 clean soil. The amount of soil lead reduction actually achieved directly influences the
4 expected changes in dust lead and blood lead. In Section 5.3, an attempt will be made to
5 evaluate the treatment/response relationship for each step of the pathway of lead in the
6 human environment.
7 To determine the effectiveness and persistency of soil abatement, the mean for each
8 parcel of land was taken for each round where soil measurements were made. The median
9 of these parcel means for the Boston and Cincinnati studies show that abated soil
10 concentrations (BOS SPI and CIN SEI) dropped significantly after abatement (Figures 5-8
11 and 5-9) whereas unabated soil (BOS PI, BOS P, and CIN NT) appear to decrease only
12 slightly, if at all. The Cincinnati groups CIN I-SE(B) and CIN I-SE(D), and CIN I-SE(F),
13 which received soil and exterior dust abatement later (during the second year), showed a
September 1, 1995
5-26
DRAFT-DO NOT QUOTE OR CITE
-------
10,000
1,000
I 1M
10
BOS SPI
10,0001
Oct89
1,000
Apr 90 Jul91
Jan 90 ~L ~"~
100
Soil & Dust
Abate
10
Oct89
B
BOS PI
Apr 90
Jul91
Soil
Abate
RD1
RD 2 RD 3
Sample Round
RDS
RD1
RD2 RD3 RDS
Sample Round
10,000
1,000
.a
a
100
10
BOSP
Oct89
Apr 90
Jul91
Dust
Abate
Soil
Abate
RD 1 RD 2 RD 3
Sample Round
RDS
Figure 5-8. Boston soil lead concentrations (on a log scale) by study group show the
effectiveness and persistency of soil abatement. Note the decrease in soil
lead concentrations (RD 2) immediately post soil abatement and persisting
through RD 2, RD 3, and RD 5 for BOS SPI Group (Panel A); no soil lead
sampling in RD 2 for other two groups (BOS PI and BOS I); RD 3 values
for those two groups similar to their RD 1 soil lead concentrations; and the
later marked decrease in their RD 5 soil lead values following soil
abatement after RD 3.
September 1, 1995
5-27
DRAFT-DO NOT QUOTE OR CITE
-------
100,000
10,000
1,000
CINSEI(P)
JulM
I
I
So«»
Dux
HD1 RD2 RD3 RD4 RDS RDf RD7
Sampfe Round
100,000
KMXXI
1,000
i
100
10
B
CMI-SE(D)
100,000
]!
£
1,000
RD1 RD2 RDS RD4 RDS RD8 RD7
Swipto Round
n CWKT(a)
muss
i
Y" X r1! Juso
RD1 RD2RD3 RD4 RD5RD6 RD7
Scmpl* Round
100,000
10,000
1,000
10
I
I
I
CIM1-3E(F)
I
100,000
10,000
1,000
RD1 RDZ RDS RD4 RD8 RD» RD7
Simple Round
_ CMNT(M)
JJ89
T •?,%.»•
RD1 RD2 RDS RD4 RD« RDS RD7
Sunpl* Round
Figure 5-9. Cincinnati soil lead concentrations (log scale). Data are shown by
neighborhood and reflect abatement in the first or second year of the study.
There were no soil samples taken in the Dandridge neighborhood (Panel B)
during round 1.
September 1, 1995
5_28 DRAFT-DO NOT QUOTE OR CITE
-------
1 postabatement decrease in the range expected. Follow-up measurements of exteriior dust
2 after this second year abatement were limited to targeted entry areas.
3 There appears to be a general downward trend of soil lead concentrations., Although
4 not statistically significant for any individual group, the fact that all treatment groups where
5 the soil remained unabated show this phenomenon lends some credence to this observation.
6 Analysis of QA/QC audit samples shows this trend cannot be attributed to analytical drift
7 (see Section 3.1). Soil lead concentrations vary widely over relatively small distances.
8 Because it was not feasible to return to the exact spot for sequential soil samples, two
9 sequential samples may vary widely.
10
11 5.2.2 Changes in Exterior Dust Concentrations and Loadings
12 In Cincinnati, exterior street and sidewalk dust concentrations remained relatively
13 constant throughout the study (Figures 5-10 and 5-11). This indicates that even though the
14 relative contribution of lead from other sources may have changed over time, exterior dust
15 abatement did not seem to be impacted by the contribution from these sources.
16 If the major source of the lead in exterior dust is soil and the soil parcels are abated
17 prior to or at the same time as external dust abatement, then the lead concentration of dust on
18 the streets and sidewalks should slowly decrease to a level comparable to the new soil
19 concentration. This does not appear to be the case. Furthermore, the exterior dust lead
20 concentrations in Cincinnati are much higher than the soil concentrations, suggesting a source
21 or sources with higher lead concentrations than soil that mix with leaded dust from soil to
22 form exterior dust. A possible conclusion is that sources of lead in exterior dust other than
23 soil impacted each neighborhood differently. This is reasonable because the neighborhoods
24 are geographically separated. Interpretation of the spatial distribution of the Cincinnati data
25 is not possible without more information on the location of the dust samples.
26 For Boston and Baltimore, the question arises that there may also be external sources of
27 lead other than soil that contribute to household dust and to the exposure of children during
28 outside activities. Because there were no measurements of exterior dust in these studies,
29 little evidence is available to accept or reject this hypothesis. However, hi the context of
30 exposure pathways, the parcels of soil in Boston and Baltimore were on the individual
September 1, 1995
5-29
DRAFT-DO NOT QUOTE OR CITE
-------
100^)00
10,000
1,000
CIHSEI(P)
*"• JUI90
NovW U
Soil & Dint
Abitt
100,000
1,000
10
B
RD1 RD2 RD3 RD4 HDS
Sample Round
100,000
10,000
1,000
StpM
R01 HD2 RD3 RD4 RD5
Simple Round
100,000
10,000
100
D
C!NNT(O)
RD1 RD2 RD3 RD4 RD5
Sample Round
CINM>E(F)
100,000
10,000
IflOO
RD1 RD2 RD3 RD4 BOB
Sample Round
CIHNT(M)
Sop 00
RD1 RD2 RD3 RD4 RDS
Sara pte Round
Figure 5-10. Exterior dust lead concentrations (log scale) from the street samples in the
Cincinnati study. Data are by neighborhood. Exterior dust samples were
not reported for rounds 6 and 7.
September 1, 1995
5-30 DRAFT-DO NOT QUOTE OR CITE
-------
100,000
I 10,000
1,000
10
CINSE](P)
NovM T
100,000
| 10,000
1,000
B
RD1 H0t RDS RD4 RDS
SampM Round
CINI-SEfD) 100,000
1,000
SottftDuM
Abate
H
CIHkSE(F)
SollliDuat
Abah
RD1 RD2 RDS RD4 RDS
Svnpls Round
1001000
10KJOO
1,000
CINNT(Q) 100,000
10,000
1,000
RD1 RD2 RD3 RD4 RDS
Samp!* Round
CINNT(H)
10
JUI90
T
Sop 90
RD1 RD2 RDS RD4 RDS
Sampfl* Round
RD1 RD2 RDS RD4 RDS
Santpl* Round
Figure 5-11. Exterior dust lead concentrations (log scale) from the sidewalk samples in
the Cincinnati study. Data are by neighborhood. Exterior dust samples
were not reported for rounds 6 and 7.
September 1, 1995
5-31 DRAFT-DO NOT QUOTE OR CITE
-------
1 properties, whereas in Cincinnati, most soil parcels were in areas separated spatially from the
2 living units, such as parks and vacant lots.
3
4 5.2.3 Changes in Interior Dust Concentrations and Loadings
5 Interior dust is measured in both concentration and surface loading. Concentration is
6 measured in micrograms of lead per gram of dust, whereas loading is measured in milligrams
7 of lead per square meter. When dust abatement is performed, the amount of dust changes,
8 but the concentration of lead in the dust does not. Therefore, there should be no change in
9 dust lead concentration unless the source of the dust changes. Where soil abatement has
10 been performed in connection with dust abatement, the dust lead concentration should also
11 decrease abruptly if the soil is the major component of the dust. If there is a mixture of dust
12 sources and only one has been abated, the lead concentration would change less abruptly,
13 according to the contribution from each source.
14 The data for the Boston study interior dust are shown in Figures 5-12 through 5-17.
15 In both BOS SPI and BOS PI, there was a general decrease in the floor dust lead loading
16 following interior dust abatement, as shown in Figure 5-14, and further decreases were
17 observed at 7 to 12 months after abatement. In the window wells, however, the lead loading
18 decreased immediately after dust abatement (Figure 5-17) persisted for a few months, then
19 returned to original levels by 12 months after abatement. The high concentrations of lead in
20 individual measurements of window well dust (5,000 to 22,000 /tig/g) indicate lead-based
21 paint was present (Figure 5-15).
22 The Cincinnati study (Figures 5-18 through 5-20) found an immediate reduction in floor
23 dust lead loading that persisted for at least 5 months, followed by an increase by 12 months
24 to 70% of the preabatement level in CIN SEI, where soil abatement had taken place, and to
25 nearly twice the preabatement interior dust level in CIN I-SE-1 and CIN I-SE-2, where soil
26 had not yet been abated. Similar patterns were observed in the window wells
27 (Figures 5-21 through 5-23) and entry ways (Figures 5-24 through 5-26). The window well
28 concentrations were lower in Cincinnati (1,000 to 2,300 jiig/g) than in Boston, suggesting a
29 minimum influence of lead-based paint.
September 1, 1995
5-32
DRAFT-DO NOT QUOTE OR CITE
-------
100,000
10,000
Oct89
5
JQ
BU
g 1,000
100
BOS SPI
100,000
10,000
_L
Jan 90
Apr 90
— Sep90
~
JU191
1,000
Soil & Dust
Abate
100
B
BOS PI
Octt89
Jan 90
I
Sep90
Ap_r_90 _^ JU|91
Dust
Abate
Soil
Abate
RD1
RD2 RD3 RD4
Sample Round
100,000
RD5
10,000
a
t>
a 1,000
100
RD1 RD2 RD3 RD4 RD5
Sample Round
BOSP
OcK89
Apr 90
Sep90
Jul91
I
Soil
Abate
RD1 RD2 RD3 RD4
Sample Round
RD5
Figure 5-12. Boston floor dust lead concentration. While dust abatement alone may
temporarily reduce the total dust lead loading (see Figure 5-14), it may not
change the concentration of lead in any remaining dust.
September 1, 1995
5-33
DRAFT-DO NOT QUOTE OR CITE
-------
1,000
100
10
BOS SPI
Oc<89
I
i
Soli & Dust
Abate
, „ Apr90 -r
Jan 90 !T_
Sep90 pt
RD1 RD2 RD3 RD4 RD5
Sample Round
1,000n
1,000
100
10
g BOS PI
Oct89 Ja"9° Apr90
Tig s^90 pq
— rn
u 5 M
Dust Soil
Abate Abate
0 S
RD1 RD2 RD3 RD4 RD 5
Sample Round
BOSP
Apr 90
iG^
g 100
"S5
•o
1
o m
Oct89
I
=
I
T
—
—
in
JulS
Sep90
I
H^
—
i
I
Soil
Abate
RD1 RD2 RD3 RD4 RD 5
Sample Round
Figure 5-13. Boston floor dust load (log scale). The absence of a decrease following
interior dust abatement in the BOS SPI and BOS PI groups suggest that
house dust loadings may be replenished back to preabatement levels in a
time period shorter than the interval between Round 1 and Round 2.
September 1, 1995
5-34
DRAFT-DO NOT QUOTE OR CITE
-------
BOS PI
1Q,UUU.U
1,000.0
^^,
*
au
| 100.0
i 10.0
I
1.0
0.1
A BOSSPI 10,000.0
1,000.0
Oct89 nn
— Apr 90
-r- Jan 90 ~~ Jul91
[I ~ T T 100.0
Hun sep9o pi
U . i°-°
1.0
Soil & Dust
Abate
1 8 — i 1 1 1 n-i
uw rt
B
Oct89
^ Jan 90 Sep 90
-j- AR^90 JuI91
r*i *
y y u
Dust Soil
Abate Abate
a i
RD1 RD2 RD3 RD4 RD5
Sample Round
10,000.0
RD1 RD2 RD3 RD4 RD5
Sample Round
BOSP
1,000.0
100.0
10.0
1.0
m
Oct89
. I Argo
-^- Scp 90
i
y 5 y
Soil
Abate
i
RD1 RD2 RD3 RD4 RD5
Sample Round
Figure 5-14. Boston floor dust lead load (log scale). Even though the dust load in
Figure 5-13 indicates a quick recovery, the lead load did not recover
immediately, indicating that the source of the lead was cut off, at least
temporarily.
September 1, 1995
5.35 DRAFT-DO NOT QUOTE OR CITE
-------
1,000,000
100,000
10,000
1,000
100
10
BOS SPI
Oct89
Jan 90
Apr 90
Sap 90
Jul91
1,000,000
100,000
10,000
1,000
100
Soil & Dust
Abate
10
B
BOS PI
Oct89 Sop 90
Apr 90
X
Jan 90
IE
Jul 91
I
Soil
Abate
B
RD1 RD2 RD3 RD4 RD5
Sample Round
1,000,000
o
U
100,000
10,000
i 1,000
3
Q
I
ej 100
10
RD1
BOSP
RD2 RD3 RD4 RDS
Sample Round.
Oct89
Apr 90
I
Sep90
I
Jul 91
! T
Soil
Abi
bate
RD1 RD2 RDS RD4 RDS
Sample Round
Figure 5-15. Boston window dust lead concentrations (log scale). Paint stabilization and
soil abatement appear to have been effective and persistent for several
hundred days, similar to floor dust. The recovery observed between April
and July 1990 was not observed for the floor dust load data.
September 1, 1995
5-36 DRAFT-DO NOT QUOTE OR CITE
-------
0,000] BOS SPI
1,000
100
10
1
A
10,000i
Jul91
Oct89
I
Sep
Apr 90 T
Jan 90
^
I
—
I
90
I
I
1,000
100
10
Soil & Dust
Abate
B
1
BOS PI
B
Sep 90
APJ^0 P=; Jul91
Oct89 T _ pr
M -" — =
~ Jan90 U X
y t i ^
2
Dust Soil
Abate Abate
1 8 — , 1 1 % ,
RD1 RD2 RDS RD4 RDS
Sample Round
10,0001
10
RD1 RD2 RDS RD4 RDS
Sample Round
BOSP
Oct89
^P90 Jul91
Apr90 T -[-
T X
I
L
Soil
Abate
FID1 RD2 RDS RD4 RDS
Sample Round
Figure 5-16. Boston window dust load (log scale). These data show the effectiveness of
window dust abatement, which appears to recover after about 150 days,
similar to floor dust loads observed in Figure 5-13.
September 1, 1995
5.37 DRAFT-DO NOT QUOTE OR CITE
-------
1,000,000.0
100,000.0
^JL 10,000.0
a
-i 1,000.0
J
1 100.0
0
I 1ao
1.0
0.1
BOSSPI 1,000,000.0i
A
100,000.0
Jul91
Oct89 Sep90 p
[• Jan 90 5 Pi ~" 10,000.0
- y -1- 1,000.0
re
100.0
10.O
1.0
Soil & Dust
Abate
, • , nt
- BOS PI
D
Oct89 Sep90
~ Apr 90
n s S Jul91
U Jan90 y I
n
i
Dust soil
Abate Abate
3 B
RD1 RD2 RD3 RD4 RDS
Sample Round
1,000,000.0
100,000.0
,«.
"^
a
•n
1
10,000.0
1,000.0
Oct 89 Sep 90
T Apr90 ^ X
n
h
u -
n i
T
T ^ 1
t; 100.0
I
10.O
1.0
RD1 RD2 RDS RD4 RDS
Sample Round
BOSP
Jul 91
s
Al
oil
•bate
RD1 RD2 RD3 RD4 RDS
Sample Round
Figure 5-17. Boston window dust lead load (log scale). As with floor dust lead loads,
the window data indicate that both paint and soil sources of lead were
interrupted, at least temporarily. The data appear to be consistent with
Figure 5-14.
September 1, 1995
5-38 DRAFT-DO NOT QUOTE OR CITE
-------
100,000
10400
1,000
£ 100
i
I ,.
CMSEI(P)
Duat
RD1 RD2 HDS RD4 RD8
Simple Round
100,000
1CMXM
1,000
g 100
a
i
G 10
B
CINI-SEflJ)
Dual
Atate
100,000
10,000
1,000
g 100
1
i .
RD1 HD2 RD3 RD4 RD5
Sample Round
CINNT(GI)
D
RD1 RD2 RD3 RD4 RD5
Sample Round
100,000
1,000
100
«NI-SE(F)
Abat*
100,000
10,000
1,000
100
HD1 RD2 RD3 RD4 RD<
Simpls Round
C1NKT(M)
RD1 RD2 RD3 RD4 RDI
Sample Round
Figure 5-18. Cincinnati floor dust lead concentrations (log scale). The small changes in
lead concentrations across all sampling points suggest that the sources of
lead and their relative contributions to housedust lead did not change as a
result of the abatement activities.
September 1, 1995
5-39 DRAFT-DO NOT QUOTE OR CITE
-------
100,000
10,000
1>00°
100
10
CWSEI(P)
a sep««
-T-
SolIlDlut
Abate
RD1 RD2 RD3 RD4 HOS
Smmpto Round
100,000
1,000
100
10
CJNI-SE(D)
100
R01 RD2 RDS RD4 RD£
Simple Round
•) ONNT(Q)
_ NOVS8
RD1 R02 RD3 RD< RDS
S*mpl> Round
10JXM
100
10
CINt-SE(F)
uia
StpM
100,000
10,000
1,000
10
RD1 RD2 t\03 RD4 RDS
Sanipla Round
CINNT(M)
RD1 RD2 RD3 RD4 RDS
Simple Round
Figure 5-19. Cincinnati floor dust load (log scale). These data confirm the effectiveness
of the household dust abatement and show that this reduction was
persistent for as much as 60 days.
September 1, 1995
5-40 DRAFT-DO NOT QUOTE OR CITE
-------
100.000
lOtOOO
1
*
1.0M
100
CIN8EI(P)
JJBD
X
S«p89
Soil 4 Dim
HIM RD2 RDS RD4 ROD
Sunptt Round
100,000
10,000
» 1-oao
I
S 100
&
10
1
B
Cl
J
jiita
fj NWW
I
L
H
it
^r
t-SEO
0
Si
SWM H
»)
£
100,000
10,000
n 1,000
100
B U i
y
Dud
Ab*t*
s
10
1— 1
c
*i«,
I
i
n
y
Diut
Mats
B
C*«EO,
Jdeo
T Sw9
rgg 1 1 pL
3 1 -
2
. i
100,000
10,000
1,000
100
10
RD1 RD2 HDJ RD4 RO6
8ampl*Roynd
_ CINNT(Q)
100,000
10,000
1,000
RD1 RD2 RD3 RD4 RDB
Sunpfe Round
CINNT(M)
RD1 RD2 RD3 RD4 RD5
Stnpli Round
RD1 RD2 RD3 RD4 RD8
Sample Round
Figure 5-20. Cincinnati floor dust lead load (log scale). The data suggest that the
sources of lead were interrupted by the abatement activities, but that at
least one source recovered after November 1989.
September 1, 1995
5-41 DRAFT-DO NOT QUOTE OR CITE
-------
100,000
1,000
100
10
CIN SEI (P)
S.D8S
1 SepSO
Soil
-------
1,000,000
100,000
rf-
10,000
1
!1>MO
1 100
10
1
A
JJIB
T S._p8»
- fl
1
IJOlllcDu*!
Abatt
8
CMSE1(P)
Jut!
I
-
Hovao 1
5 T
tl
y
0
T
I
RIM RD2 RD3 RD4
Sampfe Round
1,000,000
•tt»jat»
10,000
B
CINI-SE(»)
I
Dust
Alxrt.
1^)00,000
100,000
I 10.000
1JVX
100
10
RD1 RD2 RD3 RD4 RD5
Sampla Round
CINNT(G)
T
I
I
RD1 RD2 RD3 RD4 RD5
Round
1,000,000
100^00
10,000
1,000
100
10
CWhSE(F)
mm
I
Duat
Abats
RD1 RD2 RD3 RD4
Sarnpte Round
100,000
10/XX)
ll
-------
1MOWOO
1,000,000
<-
£100,000
1 10JMO
1
| 1.000
1 100
10-
«•
. ONSBIP)
*0» *"«
•*!? * £
pi
SNovSQ
T .3^80
B Q
^
SoHtDutt
Atato
s
RD1 HO 2 DOS HD 4 HDS
S«mpt» Round
10,000,000
1*00*00
C 100*00
1*00
100
10
B
s«pn
RD1 RD2 RDS R04 ROS
Susphi Round
18^00,000
1^00*00
•I 100*00
J 10*00
1 1*«
| 100
10
1
Q ONHr
-------
1,000,000
100,000
1 10,000
• »
[ woo
I
o
£
J 100
I 10
CINSEI(P)
pSO
JUI89
I
Sop 89 Jul«0
Nov 89 T
JunSI
V90_
Soil 4 Dim
M»t*
RD1RD2 RD3 RD4RD5 RD6RD7
Sample Round
100,000
•* 10^00
1,000
100
10
B
CINI-SE(D)
SepS9 Jul90
JulM
Sepj»
Dual
Abate
ioo,ooa
10,00»
RD1 RD2 RD3 RD4 RDS RD6 HD7
Simpli Round
_ CINNT(G)
SvpSO
Nov 90
RD1 RD2 RDS Rl)4 RDS RDS RD7
Samplo Round
LOOO.OOOn
100,000-
10,000
1,000
100
CMI-SE(F)
NovM S»p90
JuliS
Dust
AbttB
1,000^00
100,000
10,000
1,000
10
RD1 RD2 RD3 RD4RD5 ROS RD7
Sample Round
CINNT(M)
-ScpJM
JuISS
— Sen 19
EE Nov 89 !
Jun»1
NovwT
RD1 RD2 RDS RD4 RDS RD« RD7
Sample Round
Figure 5-24. Cincinnati entry dust lead concentration (log scale). The entry way subset
of the floor dust shows a pattern different from the complete floor dust
data of Figure 5-18. Note the three additional rounds, September 1990,
November 1990, and June 1991.
September 1, 1995
5-45
DRAFT-DO NOT QUOTE OR CITE
-------
1,000,000'
100,000
«
J
10,000
1,000
u
!
10
CINSEI(P)
s*p*o
JulM
NOVM •
NOV
I —
NovM
Soil 4 DuU
Abst*
RD1RDZRDI RD4RD5RD8RD7
Simple Round
1,000,000
100^00
10,000
1,000
B
CINhSE(D)
S«p90
JulM
Julia
Novta
j JunSI
NovM A
8*pSS
RD1 RD2 RDSRD4 RD8 RDB RD7
Samp!* Round
1,000,000
100,000
rf-
•£ 10,000
I
I
1 100
RD1RD2 RD3 RD4 RD5RD8 RD7
Simple Round
100,000
10,000
1,000
CINK3E(F)
Sep«0
Jul90
I
Jul«9
NovBO
j Sep88 T .
Du*t
Ab«l»
JunSI
Nov 90 S
I
1,000,000|
100,000
10,000
1,000
100
10
RD1 RD2RD3 RD4 RD$RD
-------
10,OMW»0
1.000,000
If
m 100,000
]'
| 10JMO
!
J 1,000
I 100
10
1
CWSEI(P)
A
Jul £8
f
-
I
T-T-R- "
-
n:
~r \ i
I _]_ NovCO
D T 0 "
-C U
SOII&DU*
a
tfntojaoa
k loo^oo
1,000
100
10
10,000,000
1,000,000
f
c 100,000
J 10^)00
!
I 1,000
D
i 100
10
Q
XI M
FID1RD2 RDIRD4 RDIRDtRDT
S»mpl« Round
lO^XMMIOO-
1,000,000-
Jlri 8i
I Sap SO JunSI
Dint
RD1 RD2 RD3RD4 F1D5RDOHD7
Svnpm Round
ON NT (Q)
SeaSO
S NovW
RD1RD2 RD3 RD4 RDS RD8 RD7
Sampto Round
10,000
1,000
CtHHE{F)
s*pw
thwBo
1,000,000
lOOtOOO-
10,000-
1,000
RD1R02RD3 RD4 RD5RD* RD7
Smmpto Round
cwnrpi)
JunM
JulOO
Septa
RD1 RD2 RD3 RD4RD5RD6 RD7
3»mpl» Round
Figure 5-26. Cincinnati entry dust lead load (log scale).
September 1, 1995
5-47
DRAFT-DO NOT QUOTE OR CITE
-------
40
35
30
25
20
15
10
5
0
A
BOS SPI Jul
Apr 90
Oct89
I Jan 90
I . (
•-
IE
-
I
Soil & Dust
Abate
••
I J
91 40
35
1 30
25
20
15
10
5
BOS PI
° Jul 91
Apr 90
Jan 90 1
Oct89 T n
a 1
IK
3 ? y i
Dust Soil
Abate Abate
• a
RD1
RD2 RD3
Sample Round
40
c
35
30
RD5
RD1
RD2 RD3 RD5
Sample Round
BOSP
Apr 90
-p Jul 91
•= 25
f 20
a.
1 15
10
5
0
Oct89
I
• —
—
RD1
Jan 90
IT
rr:
I
—
I
Soil
Abate
i
RD2 RD3 RD5
Sample Round
Figure 5-27. Boston hand lead load. The Boston hand lead load increased in all three
groups, hi contrast to the blood lead concentrations shown in Figure 5-31.
September 1, 1995
5-48
DRAFT-DO NOT QUOTE OR CITE
-------
50
45
40
35
£" 3Q
5 MV
A 25
JQ
O.
•g 20
15
10
5
n
BALSP 50
A
Jun91
Oi
:t
T
I
A|
88
?«•
—
b:
89
Jan 91
Feb90 T
Pi li
^*.
Soil
-^
I
•^
b;
ai
IlAbatei:
45
40
35
30
25
91
20
15
10
5
n
BALP-1
b
O
ot
T
rd
Apr 89
88
^»
F«
IE
ib
I
—
b:
Jun91
90
Jan 91
-^
Sep91
I
Ft T LT
1 I ,_. 1
RD1 RD2 RD3 RD4 RD5 RD 6
Sample Round
RD1 RD2 RD3 RD4 RD5 RD6
Sample Round
50
45
40
35
f 30
J 25
A
O.
i 20
15
10
5
n
c
BAL P-2
Apr 89
-r Sep91
Oct88
p:
in
<••»
rr
Jun91
Feb 90 D
-T Jan 91 "* 4,
I 1 a F
— b:
T U
RD1 RD2 RD3 RD4 RD5 RD6
Sample Round
Figure 5-28. Baltimore hand lead load. There were no sequential measurements of
Baltimore house dust to compare with the hand lead load.
September 1, 1995
5-49
DRAFT-DO NOT QUOTE OR CITE
-------
80
48
40
CIN8EI(P)
£
1 »o
I »
1 w
S
15
10
a
Q
Jul 90
S»p89
Jul 89
ST NoviS
I
—
-
S*pM
•"••H H I
Dutt LJ LJ
RD1 RD2 RDS RD4
Snnol» Round
SO _ CINI-SE{D) 50
45
40
38
30
25
20
15
10
a
b
45
40
1. 1.. Jul »0
JulS* 35
1 SO
S«p89
Kov89
n n S
~l
25
S«p90
• T
1 16
- 10
Aba* 1— 1 AM>
C
~
~
I
RDS
C
IN
\-SKff)
S«pM
Jul SO
Jul 89
I
-r NovM
n i
^®$ S
in
RD1 R02 RDS RO4 RDS
Sampto Round
CO
45
40
38
30
25
20
15
10
5
Q
cwmr(G) «o
45
40
35
30
25
20
15
*T T 10
TTTR a :
RD1
I
I
M5
RD2 RDS RD4 RDS
S*mpl« Round
p
Juita
^
RD1 RD2 RDS RD4 RDS
Samp)* Round
-
RD1
CINIir(M)
s_,M
JulW
- I r
S
u
y NOVM
-"- H
I
[
RD2 RDS RD4 HD5
Sample Round
Figure 5-29. Cincinnati hand lead load. The pattern of hand lead load change, both
increases and decreases, appears to follow the pattern of floor dust lead
load in Figure 5-20.
September 1, 1995
5-50
DRAFT-DO NOT QUOTE OR CITE
-------
1 5.2.4 Changes in Hand Dust Loadings
2 Because hand-to-mouth activity is one route by which lead may be ingested, the amount
3 of lead on the child's hand is an indicator of exposure. Only lead loading information is
4 available because it was necessary to take the sample with wet wipes and there is no measure
5 of the amount of dust removed. The units of measurement are micrograms per pair of hands
6 rather than micrograms per square meter.
1 In Boston, there was a general increase in hand lead throughout the study
2 (Figure 5-27). Although there is no explanation for this increase, there appears to be less of
3 an effect for the groups that received soil, and dust intervention, and this reduction is greatest
4 for the group that received soil, dust, and paint intervention.
5 Baltimore hand lead values did not follow a discernable pattern (Figure 5-28) and there
6 appear to be no systematic differences among the groups.
7 In Cincinnati, the hand dust lead load (Figure 5-29) appears to follow the pattern of
8 change observed in the floor dust lead load (Figure 5-20). This is an important link in the
9 exposure pathway that measures actual external contact with the child's dust environment.
10 Hand lead loadings were expected to respond more quickly to environmental changes than
11 blood lead concentrations. The hand lead data were informative and showed a number of
12 similar patterns across the three studies. The discussion below of the relationship of hand
13 lead to blood lead will shed further light on this critical pathway.
14
15 5.2.5 Changes in Blood Lead Concentrations
16 5.2.5.1 Baltimore Study Blood Lead Data
17 The blood lead concentrations for the three Baltimore groups are shown in Figure 5-30.
18 The data are for all children participating in the round. They show that the groups were
19 similar prior to soil abatement between Rounds 3 and 4. Following abatement, the groups
20 responded according to treatment, but the difference was not significant 10 months after
21 abatement. The lack of postabatement measurements of soil and house dust limits the ability
22 to interpret these data by more than a simple analysis of variance.
September 1, 1995
5-51
DRAFT-DO NOT QUOTE OR CITE
-------
3U
25
20
15
10
5
0
A
Oi
5t
I
«•»
38
•
i
rl
I
31
Feb
$9 -
II
SAL:
SP
30
25
20
90 Jun 91
Jan 91 Sep91 1g
TIT
Soil
-
r
r
—
r
10
5
Abate
i
— 0
B BALP-1
Oct88
T Apr 89 Sep 91
rJL, -r Feb 90 Jun 91 _
1 T Jan91 T 1
_ ^.
y y Q g g
-
r
RD1 RD2 RD3 RD4 RD5 RD6
Sample Round
RD1 RD2 RD3 RD4 RD5 RD6
Sample Round
30
25
20
15
£ 10
BAL P-2
Oct88
i
Apr 89
T Feb
I T
| J_,
Feb 90 sep 91
Jun 91
Jan 91 -r -r
I
—i 1 1 1 —i 1
RD1 RD2 RD3 RD4 RD5 RD6
Sample Round
Figure 5-30. Baltimore blood lead concentrations. There appears to be little difference
between study groups.
September 1, 1995
5-52 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
5.2.5.2 Boston Study Blood Lead Data
The blood lead concentrations for the Boston study are shown in Figure 5-31, where
they graphically illustrate the conclusions of the Boston report, that intervention probably
accounted for a decrease of 0.8 to 1.5 /xg/dL in the blood lead. The observation that all
three Boston study groups experienced an increase hi blood lead concentrations between
Round 3 (April 1990) and Round 4 (September 1990) is consistent with similar observations
in the hand dust lead load and, to a lesser degree, the window dust lead load. The apparent
absence of a comparable increase in floor dust lead load runs counter to the expected pattern
of the floor dust lead load being the primary route for dust exposure hi children.
5.2.5.3 Cincinnati Study Blood Lead Data
The wealth of information from the more detailed measurements of household dust hi
the Cincinnati study presents a proportionally greater challenge to the modeling of dust
exposure pathways. The blood lead concentrations shown in Figure 5-32 correspond roughly
to the changes observed in the hand dust lead loads of Figure 5-29. And there are several
pohits where the blood lead concentrations are consistent with the observed changes hi the
various forms of house dust. The floor and window dust lead loads are especially indicative
of the exposure route, and the mat dust lead load seems to account for the increase hi blood
lead concentrations after November 1990. The group that received soil abatement hi the first
year, CIN SEI, continued to show increasing blood lead concentrations through the following
year, and the CIN I-SE(B) and CIN I-SE(D), and CIN I-SE(F) groups continued to decrease
following soil and exterior dust abatement in the second year.
5.3 PRE- AND POSTABATEMENT DIFFERENCES IN INDIVIDUALS
5.3.1 Individual Changes in Blood Lead and Soil Lead
Section 5.2 provides a visual presentation of longitudinal changes hi population means
for specific parameters over the course of the study. This section presents information on an
individual child basis through the use of a series of double difference plots where the
difference between pre- and postabatement blood lead concentrations are plotted against the
September 1, 1995
5-53
DRAFT-DO NOT QUOTE OR CITE
-------
25-
»
15
1»
&
«v
^ BOSSPI *°
Apr 90 20
Oct88 JUI91
I
I
Jan 90
I
=
-^
-r
^ +
15
.*
I
10
5
Soit&Dust
Abate
1 -,,- n
B
Oct89
I
Jan 90
:= T
I I
*
Dust
Abate
$
BOS PI
Apr 90
I
. Jul91
I
bd
Soil
Abate
i
RD1 RD2 RD3 RD5
Sample Round
25
20
15
10
RD1
BOSP
RD2 R03 RD5
Sample Round
Oct89
Apr 90
. n~ T J«l»l
Jan 90 I _|_
I
I
I
Soil
Abate
RD1 RD2 RD3 RD5
Sample Round
figure 5-31. Boston blood lead concentrations.
September 1, 1995
5.54 DRAFT-DO NOT QUOTE OR CITE
-------
20
11
10
CW8E1(P)
Juita
-r JulOO Juntl
I -r NOVM -
"^ Novftt rLl
T
RD1 RDS RD4 RDt RD7
SampM Round
SO
25
i
> 20
15
B 10
I
•
A
_ Om$E(D> *
D
25
JunQI
JlHW
T JulM)
HovW
rh T T
1 w^-*°
"•
i U u
I
-
:^
20
15
10
T 6
£a& ££.
B . i , 1 — A
c
All
JultS
I
_
r
Nov8» r
5 -
0
_c
" Jun
1 NovM
• I f
M J
ac
Sail
SS
i 1 H 1 1
RD1
RD3 RD4 RDS RD7
Sample Hound
RD1 RD3 RIM RD8 RD7
Samptt Round
HDU RDS RD4 HD8 RDT
Sampl* Round
so
as
28
15
10
5
A
CMKT((1) SO
25
20
15
juiaa
T „„_ JUI8° Jun 91
n "0*-aB -T- NOT 90 10
R 1 1 T J
§ 1 1 i
c
E
JulCS Julsa
I
=
I
i y'"?
s:
I
I D b
WNT(H>
Junt
i
90
U
RD1 RDS RD4 RDt RDT
Sample Round
Figure 5-32. Cincinnati blood lead concentrations. Compare to hand lead load patterns
in Figure 5-29.
September 1, 1995
5.55 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
difference in pre- and postabatement soil lead concentrations, dust lead concentrations or dust
lead loadings.
Most children in each neighborhood experienced some change in blood lead, either an
increase or decrease, during the course of the study. Some of this change is due to changes
brought about by intervention. Another part may be due to seasonal effects, age
(see Figure 2-6), or changes in exposure not related to intervention.
A child exposed to decreasing soil lead concentrations is expected to experience a
decrease in blood lead concentration. In Figure 5-33, this child would be represented in the
lower left quadrant (III). Conversely, a child exposed to increasing lead concentrations
should experience an increase in blood lead concentrations. This child would be represented
in the upper right quadrant (I). If there were no other factors involved, all children should
be hi the upper right or lower left quadrants, or centered around the origin if there were little
or no change. If the relationship between blood lead and soil lead were strictly linear, and if
blood lead concentrations increased by the same mechanism as they decrease, all points
would lie on a straight line passing through the origin.
In these studies, there does not appear to be a linear response for any of the double
difference plots, and there are many cases where data lie in one of the excluded quadrants II
and IV, indicating blood lead increased when environmental lead decreased, or vice versa
(Figures 5-33 to 5-41).
This type of plot is especially helpful to the reader hi understanding the variability of
the measurements and the possible significance of patterns or clusters. They are designed to
show the interaction of only two variables at a tune, not the multiple interactions of several
variables. In Section 5.4, statistical techniques such as repeated measures analysis and
structural equation modeling are used to extract information from the systematic variability
using more appropriate methods for comparison than observed on these double difference
plots but in the context of several variables interacting at the same tune.
There are a few observations worth noting in the double difference plots. In Boston
and Baltimore, the more intense interventions (BAL SP and BOS SPI) placed a greater
number of points in quadrant III. Even though soil seemed to have a greater impact than
floor dust (Figures 5-34 through 5-36), later analyses hi Sections 5.4 and 5.5 suggest
otherwise. Entry way dust lead concentrations and loadings in Cincinnati do not seem to
September 1, 1995
5-56
DRAFT-DO NOT QUOTE OR CITE
-------
15
CO
D5
6
II
* BAL BP *
x BAL P-2
A
; *
^ •&• *
* * * *
* * * ** * * *
V # x**
£ *fr* * *
*** ** * *^
•sV
•&•
III *
1
xx
X
X
IV
-15
-1250
-750 -250
Change in Soil Lead
250
Figure 5-33. Double-difference plot of the change in soil lead versus the change in blood
for the Baltimore study. Except for a few measurements in BAL P-2,
postabatement soil measurements were taken in BAL SP only.
September 1, 1995
5-57
DRAFT-DO NOT QUOTE OR CITE
-------
10
D)
i"
Q>
\^
8
m
0
o
O
-10
-15
1
II
—
ft
-A- A
ft
ft
*ff &
ft
ft
ft
_
III
1
1 1 I
x xo
•CO
•A 0
O O
ft >O Ox O®5
X ft QCO
ft ftft (3&1 iV i^rx>^3 x
ft ftffr O ft O x XO® >s
ft ft ft •& X iKx OX X
ft x
ft ft ftxQ
ftft* ft
ft O
ft O >
x
I 1 1
1
1
O
-
x
xx o
OOx
5D x x O
©XX
£x x Ox
XD O
O O x
D
x
* BOS SP\
o x BOS PI _
o BOSP
IV
i i
-5000-4000-3000-2000-1000 0 1000 2000 3000
Change in Soil Lead (|ig/g)
Figure 5-34. Double-difference plot for Boston soil and blood lead data.
September 1, 1995
5-58
DRAFT-DO NOT QUOTE OR CITE
-------
o>
TJ
8
CD
6
1U
5
-5
10
^ f^
I O
II
-
_
O
x
* BOS SPI
x BOS PI
o BOSP
i 1
O Ox
O
o
O O x
O * Ogx
x QsO
O TS5rX©i
•6 xCX <3fxx •&
•iV x x OSS
x x C
* O -to -A
•sV . -tt XS
•&•
x
O
•&
I I
I
—
&•
000
OO ^r
DO
O
•ft
-
IV
-25000 -15000 -5000 0 5000
Change in Floor Dust Lead Concentration (|ig/g)
Figure 5-35. Double-difference plot for Boston floor dust lead concentrations and blood
lead concentrations.
September 1, 1995
5-59
DRAFT-DO NOT QUOTE OR CITE
-------
10
I
i"
•g
.2
m
0
-5
>
I
6
-10 -
-15
II
-
-
__
III
1 1 1 1 1
* BOS SPI
x BOS PI
o BOSP
o
o
1 1
XQ
O
it
00 x
o x ® •& a
X("l H7"\wsin~A
\^J '^/•Rfii^'
W 7* 7S >ZO T^ZPVt&f
x •& it iQX © -tg
xx o x O x •&© 4
it •&• X i^feC
x O xc
•k *
"^T "^f "A" X!
* A
x
C
*
1 I I
1
1
-
O
) O O x O
O
X X
x
it •&• —
—
IV
1
-600 -500 -400 -300 -200 -100 0 100 200
Change in Floor Dust Lead Loading (ng/m2)
Figure 5-36. Double-difference plot for Boston floor dust lead loading and blood lead
concentrations.
September 1, 1995
5-60
DRAFT-DO NOT QUOTE OR CITE
-------
20
"§ 10
CD
c
I
6
.-10 -
-20
-4000 -2000 . 0 2000 4000 6000
Change in Entry Dust Lead Concentration (^g/g)
Figure 5-37. Double-difference plot for Cincinnati entry dust lead concentrations and
blood lead concentrations.
September 1, 1995
5-61
DRAFT-DO NOT QUOTE OR CITE
-------
I
E
^O
QQ
.C
0
D)
I
O
20
10
0
-10
-20
II
_
III
1
*CINSEI
x CIN l-SE-1
+ CINI-SE-2
o CIN NT
o
i
o
* ^
X
x 1 <$
X O Q
1
* -tr
+ *
O
X
frO O +
-------
I
20
10 -
*M*
§
CD
0
D>
I
6
0
-10
-20
II
X
-
•&
X
X
_
III
o
x
* CIN SEI
x CIN l-SE-1
+ CIN l-SE-2
o CIN NT
i
o
•fr ,5
C
*c
x^
-fi
1
° cfi
1
i
-j
1
1
—
0 + o
_ x
4. X 4.
1 "
*y _A_ -I- 1 f^$
I x x o
w & +
4.
IV
1
-15000 -5000 0 5000
Change in Entry Dust Lead Loading
15000
Figure 5-39. Double-difference plot for Cincinnati entry dust lead loading and blood
lead concentrations.
September 1, 1995
5-63
DRAFT-DO NOT QUOTE OR CITE
-------
20
I> 10
0
jo
CD
c
-10
03
O
-20
II
* CIN SEI
x x CIN l-SE-1 *
+ CIN l-SE-2
o CIN NT °
•ft-
x +.
•ft- * x,
•&
•6- + xr>~^
x ' x 1 $O
x x+ i 0 o x
x flSsr*
§ H- o J <
x £• -
++
III
I I
1
+
TV
+ *r
X
I_O x
Qfr , i!r
O + X
r+ x
+ & x x
« 9** *
ip+^+ x x
-98x+i o
+ +
^
+
IV
-1500 -1000 -500 0 500 1000
Change in Floor Dust Lead Concentration
Figure 5-40. Double-difference plot for Cincinnati floor dust lead concentrations and
blood lead concentrations.
September 1, 1995
5-64 DRAFT-DO NOT QUOTE OR CITE
-------
20
I
10
^^
8
CD
(D
D>
CO
6
0
-10
-20
II
•&
•fe
+
- +
III
* GIN SEI
x GIN l-SE-1
+ CINI-SE-2
o GIN NT
*
t +
•i
-------
1 have a significant impact on blood lead concentrations, although later analyses show this
2 variable is important hi estimating the influence of exterior dust or soil on house dust. Floor
3 dust lead concentrations and lead loadings in Cincinnati (Figures 5-40 and 5-41) show a large
4 number of points hi quadrant IV. The increase in exterior dust that eventually impacted
5 interior dusts may not have yet caused an increase in blood lead concentrations.
6
7
8 5.4 COMPARISON BY REPEATED MEASURES ANALYSIS
9 5.4.1 Baltimore Study
10 The Baltimore study results for blood lead are shown in Figure 5-42, and for hand lead
11 in Figure 5-43. For each of the three groups, the central points show the geometric mean
12 and the ends of the bars around the points show the uncertainty of the geometric mean as
13 measured by the geometric standard error, where the upper bar is the geometric mean
14 multiplied by one geometric standard error and the lower bar is the geometric mean divided
15 by one geometric standard error. The geometric standard error is a factor equal to
16 exponent (SEL), where SEL is the standard error of the mean logarithm of hand lead or
17 blood lead. The ends of the bars also define a 68% confidence interval for the geometric
18 mean of natural log. The intervals are based on an assumed normal distribution for the
19 natural logarithm of the geometric mean, and so are not quite symmetric around the
20 geometric mean. Each measurement made before abatement must be paired with a
21 measurement made after abatement hi order to calculate the effect of the abatement, so that
22 the statistical uncertainty of the intervention differences cannot be calculated from the
23 separate standard errors shown in these figures. Preabatement is Round 3, and
24 postabatement is one year later, Round 6.
25 The geometric mean blood lead profiles for the BAL P-l and BAL P-2 control groups
26 hi Figure 5-42 are almost parallel and horizontal, similar to the example in Figure 5-6.
27 There is a slight decrease hi the BAL SP blood lead levels between Rounds 3 and 6,
28 resembling Figure 5-6. This suggests that there was a slight decrease in blood lead levels hi
29 Baltimore soil abatement children relative to either control group. However, hypothesis tests
30 hi Table 5-1 showed no significant differences hi blood lead rates of change related to soil
31 abatement.
September 1, 1995
5-66
DRAFT-DO NOT QUOTE OR CITE
-------
12
11
'10
Round 3
Round6
Figure 5-42. Change in preabatement geometric mean blood lead levels in Baltimore
study \ year after abatement.
13
11
Round 3
Rounds
Figure 5-43. Change in preabatement geometric mean hand lead levels in Baltimore
study 1 year after abatement.
September 1, 1995
5_67 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 5-1. STATISTICAL SIGNIFICANCE OF BALTIMORE REPEATED
MEASURES ANALYSES FOR BLOOD LEAD, ROUNDS 3 AND 6
(PRE- AND POSTABATEMENT), AFTER COVARIATE ADJUSTMENT
Significance of Effect
Covariate
None
Log of Soil Lead
Log of Dust Lead Loading
Log of AA Dust Con.
Log of XRF Dust Cone.
Log of Interior Paint + 1
Log of Exterior Paint 4- 1
Age in years (categorical)
N
149
149
145
149
149
143
136
149
Time * Group
0 4247
0.3780
0.2012
0.6212
0.3144
0.1783
0.8418
0.4043
Time * Covariate
0.0361*
0.0125*
0.6304
0.6560
0.1564
0.3878
0.5761
Time * Group * Covariate
0.3217
0.2106
0.7895
0.4741
0.0988+
0.7342
0.8224
'In this chapter, the convention for indication significance by ranges of p-values is:
+p -0.05 to 0.10
*p = 0.01 to 0.05
"p - 0.005 to 0.01
"*"p » 0.001 to 0.005
""p < 0.001
1 The geometric mean hand lead profiles for the BAL P-l and BAL SP groups in
2 Figure 5-43 are almost identical and increase during the study, whereas the profile for
3 BAL P-2 is nearly horizontal. The interpretation of Figure 5-43 is that hand lead levels in
4 the soil abatement group rose at a faster rate than in the control groups. However, when
5 adjusted for initial floor dust lead concentration before abatement, the rate of increase of
6 hand lead levels was significantly less than in the Baltimore P-l control group. When
7 adjusted for initial floor dust lead concentration before abatement, the rate of increase of
8 hand lead levels in the low-soil adjacent control group BAL P-2 was significantly greater
9 than in the Baltimore P-l control group. Without adjusting for the preabatement dust lead
10 concentration, then as shown in Figure 5-43, the rate of increase of hand lead levels in the
11 low-soil adjacent control group BAL P-2 appears to be significantly less than in the
12 Baltimore P-l control group.
13 The statistical significance of the covariate-adjusted repeated measures analyses is
14 shown in Table 5-1. Comparisons of changes in blood lead concentrations showed no effect
15 of treatment group, with or without adjustments, with all P values > 0.178. Similar lack of
16 significant treatment group effect was shown when the covariates were tested one at a time,
September 1, 1995
5-68
DRAFT-DO NOT QUOTE OR CITE
-------
1 except for a marginal effect of interior lead paint (P = 0.10). Because interior lead paint
2 was not abated, there may have been some mitigation of any beneficial soil lead abatement
3 effect that might have occurred by the nonremediated ulterior lead-based paint. There was,
4 however, a significant positive statistical relationship of blood lead reduction to the
5 preabatement soil lead concentration and dust lead loading. Remediating households with
6 higher soil lead had more benefit than remediating those with lower soil lead, but the higher
7 nonremediated dust lead and interior lead paint loadings offset any beneficial effects of soil
8 lead remediation that might have occurred.
9 Hand lead loadings show many statistically significant relationships to the study group
10 hi Table 5-2. There are also significant interactions of study group with covariates, but these
11 effects show little relation to soil abatement. Detailed examination of these relationships (not
12 shown here) finds that the increase in hand lead is different between the two control groups,
13 BAL P-l and BAL P-2, and that there is little difference between the control group and the
14 soil abatement group, BAL SP, in Area 1. It is of some interest that there is usually a larger
15 difference in the average change hi lead between the two neighborhood control groups than
16 between the control group and soil abatement group hi the same neighborhood hi Baltimore.
17
18
TABLE 5-2. STATISTICAL SIGNIFICANCE OF BALTIMORE REPEATED
MEASURES ANALYSES FOR THE LOGARITHM OF HAND LEAD,
ROUNDS 3 AND 6 (PRE- AND POST ABATEMENT), COVARIATE ADJUSTMENT
Significance of Effect
Covariate
None
Log of Soil Lead
Log of Dust Lead Loading
Log of AAS Dust Cone.
Log of XRF Dust Cone.
Age in years (categorical)
N
288
288
274
288
288
288
Time * Group
0.0015**
0.0448*
0.0366*
0.1869
0.6598
0.0032**
Time * Covariate
0.1324
0.7750
0.4023
0.7519
0.4465
Time * Group * Covariate
0.0186*
0.0011"
0.0071**
0.0419*
0.5888
September 1, 1995
5-69
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
Age plays a significant role in hand lead loading, but not in blood lead differences
among study groups. The child's age appears to be a useful variance-reducing covariate that
can explain some of the differences among children, but is not useful as a significant
modifier of the soil abatement effect hi the Baltimore study.
5.4.2 Boston Study
The Boston study results for blood lead are shown in Figure 5-44, for hand lead hi
Figure 5-45, for floor dust lead concentration in Figure 5-46, and for floor dust lead
loadingin Figure 5-47. For each of the three groups, the central points show the geometric
mean and the ends of the bars around the points show the uncertainty of the geometric mean,
calculated for one geometric standard error, as in Section 5.4.1.
131 rp
12
11
10
Round 1
Round 4
Figure 5-44. Change in preabatement geometric mean blood lead levels in Boston study
1 year after abatement.
September 1, 1995
5-70
DRAFT-DO NOT QUOTE OR CITE
-------
21
20
19
18
«
17
S
'l6
i
! 15
14
13
12
11
Round!
Round 4
Figure 5-45. Change in preabatement geometric mean hand lead levels in Boston study
1 year after abatement.
3500]
3000
n.
"gzsoo
2000
1600
1000
500
Round 1
Round 4
Figure 5-46. Change in preabatement geometric mean floor dust lead concentration in
Boston study 1 year after abatement.
September 1, 1995
5-71 DRAFT-DO NOT QUOTE OR CITE
-------
120
100
80
en
"§
.3
60
40
20
Round 1
Round 4
Figure 5-47. Change in preabatement geometric mean floor dust lead loading in Boston
study 1 year after abatement.
1 The geometric mean blood lead profiles for the BOS PI and BOS P control groups after
2 one year, shown in Figure 5-44, are almost identical and intersecting, as in Figure 5-6 for
3 example. There is a much greater decrease in the BOS SPI blood lead levels between
4 Rounds 1 and 4, even more greatly resembling Figure 5-6. This suggests that there was a
5 slightly greater decrease hi blood lead levels hi the Boston interior dust abatement children
6 in BOS PI than in the negative control group BOS P. Boston children in the soil abatement
7 group BOS SPI showed a much greater decrease hi blood lead relative to either control
8 groups BOS PI and BOS P, demonstrating a beneficial soil abatement effect that was
9 statistically significant.
10 The geometric mean hand lead profiles for the BOS SPI and BOS PI groups in
11 Figure 5-45 are almost parallel and increase during the study, whereas the profile for BOS P
12 increased much more rapidly. The interpretation of Figure 5-45 is that hand lead levels in
13 the control group BOS P rose at a faster rate than in the soil or dust abatement groups.
September 1, 1995
5-72
DRAFT-DO NOT QUOTE OR CITE
-------
1 However, none of these differences were statistically significant even when adjusted for
2 initial lead exposures.
3 The geometric mean floor dust lead concentration profiles for the BOS SPI and BOS PI
4 groups in Figure 5-46 are almost parallel and decreased rapidly during the study, whereas the
5 concentration profile for BOS P decreased more slowly. The interpretation of Figure 5-46 is
6 that the soil and dust abatements both had a beneficial effect in reducing floor dust lead
7 levels more than hi the control group BOS P. However, none of these differences were
8 statistically significant even when adjusted for initial soil and paint lead exposures.
9 The geometric mean floor dust lead loading profiles for the BOS SPI and BOS PI
10 groups hi Figure 5-47 are almost parallel and decreased rapidly during the study, whereas the
11 loading profile for BOS P decreased more rapidly. The interpretation of Figure 5-47 is that
12 the soil and dust abatements had little effect hi reducing floor dust lead loadings. However,
13 none of these differences were statistically significant even when adjusted for initial soil and
14 paint lead exposures.
15 The Boston study showed clear and statistically significant differences hi the decrease of
16 blood lead between Rounds 1 and 3, as shown hi Table 5-3. When the relationship was
17 adjusted for initial soil lead, dust lead, or paint lead, the differences among treatment groups
18 became nonsignificant. This suggests that the quantitative characterization of abatement by
19 change hi soil lead or dust lead is sufficiently strong hi the Boston study that remediation
20 group effect is largely subsumed by the changes hi environmental lead concentrations. The
21 environmental changes hi the Boston study are twofold: large and persistent reductions hi
22 soil lead and dust lead in the soil abatement group, and small changes hi the other two
23 groups. The corresponding effects are moderately large reductions in blood lead the first
24 year after abatement hi the soil abatement group. Blood lead continues to decrease hi the
25 second postabatement year hi those households where recontamination did not occur, as
26 expected from the biokinetics of lead storage in bone.
27 Unlike the Baltimore study, hand lead loadings hi Boston showed little relation to soil
28 or dust abatement, as seen in Table 5-4. Reasons for this difference are not obvious.
29 The Boston study also found that child age was an important and highly significant
30 covariate for changes in blood lead. As hi the Baltimore study, there was no strong evidence
31 that age modified the effect of soil abatement versus other treatments.
September 1, 1995
5-73
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 5-3. STATISTICAL SIGNIFICANCE OF BOSTON REPEATED
MEASURES ANALYSES FOR BLOOD LEAD, ROUNDS 1 AND 3
(PRE- AND POSTABATEMENT), AFTER COVARIATE ADJUSTMENT
Significance of Effect
Covariate
None
Log of Soil Lead
Log of Dust Lead Loading
Log of Dust Lead Cone.
Log of 1 + Chipped Paint
Log of 1 + Interior XRF
Age in years (categorical)
Sex
N
147
147,
147
133
132
141
147
147
Time * Group
0.0074"
0.4589
0.4046
0.9932
0.0774+
0.7993
0.0004"*
0.0107*
Time * Covariate
0'.8844
0.2138
0.3890
0.4375
0.7961
0.2800
0.6425
Time * Group * Covariate
0.5644
0.4516
0.8453
0.4937
0.8645
0.1695
0.6497
TABLE 5-4. STATISTICAL SIGNIFICANCE OF BOSTON REPEATED
MEASURES ANALYSES FOR NATURAL LOGARITHM OF HAND LEAD,
ROUNDS 1 AND 3 (PRE- AND POSTABATEMENT),
AFTER COVARIATE ADJUSTMENT
Significance of Effect
Covariate
None
Log of Soil Lead
Log of Dust Lead Loading
Log of Dust Lead Cone.
Log of 1 + Chipped Paint
Log of 1 + Interior XRF
Age in years (categorical)
Sex
N
150
150
150
136
134
150
150
Time * Group
0.0781+
0.3102
0.7893
0.6985
0.3190
0.8924
0.0840+
Time * Covariate
0.5085
0.6812
0.6148
0.3909
0.4400
0.6808
Time * Group * Covariate
0.3873
0.6643
0.7412
0.7912
0.4007
0.9521
1 5.4.3 Cincinnati Study
2 The results on significant neighborhood treatment group effects for the Cincinnati study
3 are shown hi Tables 5-5 and 5-6. There was a significant difference in blood lead changes
4 among the Cincinnati neighborhoods, which also became nonsignificant when adjusted for
5 differences hi dust lead concentrations or loadings in the residence unit interior entry or
6 floor. This suggests that preabatement environmental dust lead characterizes changes in the
7 child's blood lead at least as well as does the remediation group for the neighborhood. Even
September 1, 1995
5-74
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 5-5. STATISTICAL SIGNIFICANCE OF CINCINNATI REPEATED
MEASURES ANALYSES FOR BLOOD LEAD, ROUNDS 1 AND 4
(12 MONTHS), AFTER COVARIATE ADJUSTMENT
Significance of Effect
Covariate
None
Log Dust Cone. Floor
Log Dust Cone. Entry
Log Lead Load Floor
Log Lead Load Entry
Log XRF Interior Trim
Log XRF Interior Wall
Log XRF Exterior Trim
Log XRF Exterior Wall
Age (years)
N
156
146
139
146
143
153
154
154
132
156
Time * Group
0.0477*
0.9990
0.8364
0.9883
0.3106
0.5036
0.0280*
0.0161*
0.1237
0.0521+
Time * Covariate
0.9753
0.3386
0.2217
0.5823
0.6762
0.9342
0.2827
0.7934
0.0001****
Time * Group * Covariate
0.9912
0.9050
0.8812
0.7317
0.1190'
0.4964
0.0026*
0.4410
0.0438*
TABLE 5-6. STATISTICAL SIGNIFICANCE OF CINCINNATI REPEATED
MEASURES ANALYSIS FOR HAND LEAD, ROUNDS 1 AND 4
(12 MONTHS), AFTER COVARIATE ADJUSTMENT
Signficance Effect
Covariate
None
Log Dust Cone. Floor
Log Dust Entry Floor
Log Lead Load Floor
Log Lead Load Entry
Age (years)
N
—
Ill
106
111
110
120
Time * Group
—
0.8142
. 0.4226
0.9513
0.9172
0.2119
Time * Covariate
—
0.6746
0.7115
0.9860
0.3734
0.0406*
Time * Group * Covariate
—
0.7780
0.3937
0.9530
0.9077
0.9179
1 though the Cincinnati study was largely restricted to gut-rehab housing, interior lead-based
2 paint on walls, and exterior lead-based paint on trim were significantly related to blood lead
3 changes hi different neighborhoods. Finally, there were significant age-related effects on
4 blood lead changes during the study that were also related to the neighborhood or equivalent
5 treatment group.
September 1, 1995
5-75
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
Hand lead loadings showed no significant relationship to remediation group,
neighborhood, environmental covariates, but did show an age effect, as shown hi Table 5-6.
The Cincinnati study results for blood lead are shown hi Figure 5-48, for hand lead hi
Figure 5-49, for floor dust lead concentration hi Figure 5-50, and for floor dust lead loading
hi Figure 5-51. For each of the four groups, the central points show the geometric mean and
the ends of the bars around the points show the uncertainty of the geometric mean, calculated
for one geometric standard error as described hi Section 5.4.1.
14
13
12
£=• 44
I"
10
Round 1
Round 4
figure 5-48. Change in preabatement geometric mean blood lead levels in Cincinnati
study 1 year after abatement.
1 The geometric mean blood lead profiles for the CIN I-SE-1, CIN I-SE-2, and CIN NT
2 control groups hi Figure 5-48 are almost parallel and nonintersecting, as in Figure 5-6 for
3 example. There is an increase hi the CIN SEI blood lead levels between Rounds 1 and 5,
4 somewhat resembling Figure 5-6. This suggests that there was a moderate increase hi blood
5 lead levels hi the Cincinnati soil abatement children hi CIN SEI than hi the positive or
6 negative control groups. The unexpected direction of the soil abatement effect was
September 1, 1995
5-76
DRAFT-DO NOT QUOTE OR CITE
-------
18
16
"8 10
Is
Round 1
Round 4
Figure 5-49. Change in preabatement geometric mean hand lead levels in
Cincinnati study 1 year after abatement.
500
400
1
o>
S 300
200
100
CINNT
Round 1
Round 4
Figure 5-50. Change hi preabatement geometric mean floor dust lead
concentrations in Cincinnati study 1 year after abatement.
September 1, 1995
5-77 DRAFT-DO NOT QUOTE OR CITE
-------
800
700
I
:600
600
400
300
200
100
Round 1
Round 4
Figure 5-51. Change in preabatement geometric mean floor dust lead loading in
Cincinnati study 1 year after abatement.
1 statistically significant relative to the no-treatment group CIN NT, but not relative to the
2 interior dust abatement groups.
3 The geometric mean hand lead profiles for the CIN SEI and CIN I-SE-2 groups in
4 Figure 5-49 are almost parallel and increase during the study, whereas the profiles for
5 CIN I-SE-1 and CIN NT have a flatter slope. The interpretation of Figure 5-49 is that hand
6 lead levels in the control groups CIN NT and CIN I-SE-1 rose at a slower rate than in the
7 soil or dust abatement groups CIN SEI and CIN I-SE-2. The only differences that were
8 statistically significant were between CIN I-SE-1 and CIN I-SE-2, and only when adjusted
9 for initial floor dust and entrance dust lead concentrations.
10 The geometric mean floor dust lead concentration profiles for the CIN SEI, CIN I-SE-1
11 and CIN I-SE-2 groups in Figure 5-50 are almost parallel and increased during the study,
12 whereas the concentration profile for CIN NT decreased. The interpretation of Figure 5-50
13 is that the soil and dust abatements apparently had no effect in reducing floor dust lead
14 concentrations more than hi the control group CIN NT. However, none of these differences
15 were statistically significant except for CIN SEI versus GIN NT.
September 1, 1995
5-78
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
The geometric mean floor dust lead loading profiles for the CIN I-SE-1 and CIN I-SE-2
groups in Figure 5-51 increased rapidly with different slopes during the study, whereas the
lead loading profile for CIN SEI and CIN NT decreased. The interpretation of Figure 5-51
is that the soil and dust abatements had little effect in reducing floor dust lead loadings in
CIN I-SE-1 and CIN I-SE-2. The differences between CIN I-SE-1 and CIN I-SE-2 were
statistically significant after adjusting for entrance dust lead loading. The differences between
CIN SEI and CIN NT were statistically significant even when adjusted for initial entrance
dust lead loadings, with the rate of decrease proportionally larger in CIN NT. The pattern of
changes in dust lead loadings is not easy to interpret without invoking additional sources of
lead in these neighborhoods where there were no exterior soil or dust interventions.
5.4.4 Repeated Measures Analyses Adjusted for Environmental Analysis
and Demographics
5.4.4.1 Results from Boston Study
The results of repeated measures analyses for a variety of models are shown in
Table 5-7. Eleven models to be tested have been specified in advance, so that any model for
which there is a soil abatement effect with P value less than about 0.05 711= 0.0045 can
be regarded as showing a significant effect 8 to 10 months after soil and interior dust
abatement, with a group-wise significance level less than 0.05. The eleven models can be
described as follows:
• Soil abatement group versus Other two groups combined;
• Soil abatement group versus Other two groups combined, adjusted for change in
floor dust lead concentration from pre- to postabatement;
• Comparison of all three groups, not adjusted for covariates;
• Comparison of all three groups, adjusted for covariates one at a tune:
- Change in soil lead concentration from pre- to postabatement
- Change in floor dust lead concentration
- Change in floor dust load
- Change in floor dust lead loading
- Age at beginning of study
- Ethnicity/race category
- SES
- Sex.
September 1, 1995
5-79
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 5-7. REPEATED MEASURES ANALYSES OF BLOOD LEAD IN
BOSTON STUDY FOR FIRST YEAR AFTER ABATEMENT, ADJUSTED FOR
DIFFERENCES IN ENVIRONMENTAL INDICES AND DEMOGRAPHICS
Statistical Significance
Comparison Groups
Covariates
Comparison
df1
Soil Abatement vs.
others
Soil Abatement vs.
others
Soil, Dust, Control
Soil, Dust, Control
Soil,
Soil,
Soil,
Soil,
Soil,
Soil,
Soil,
Dust,
Dust,
Dust,
Dust,
Dust,
Dust,
Dust,
Control
Control
Control
Control
Control
Control
Control
None
Dust Pb
Cone, Floor
None
Soil Pb
Dust Pb
Cone.
Dust Load
Dust Pb
Load
Age
Ethnicity
Category
SES
Sex
1,148
Soil!
1,101 Soil!*
2, 147 SoiH
2, 143 Soil!
2,101
2,144
2,144
SoiH
Soill
Soil*
2, 144 Soil!
2,
SoiH
2,140
82
Dustl
SoiH
2,144
*Time
P
0.0394
0.0077
0.1035
0.2209
0.0084
0.0641
0.1070
0.7599
0.0020
0.0720
0.4248
Covariate * Time
df
—
1,101
1,143
1,101
1,144
1,144
1,144
3,82
Black T t
1,140
1,144
P
—
0.2537
0.5121
0.3389
0.8027
0.9894
0.4159
0.0074
0.6222
0.9487
Comp * Cov
df
—
1,101
2,143
2,101
2,144
2,144
2,144
6,82
2,140-
2,144
* Time
P
—
0.24330
0
0
0
0
0
0
0
0
.8778
.5134
.6634
.9651
.9996
.0006
.2355
.7257
'df = degrees of freedom, expressed here as two numbers: a, b. The value a is the number of degrees of
freedom of the effect being tested; the value b is the number of degrees of freedom of the residual error term
used as the basis for the hypothesis tests.
2
3
4
5
6
7
8-
9
10
11
12
Soil abatement showed a test-wise significant reduction in blood lead for four of the
models:
• Soil abatement group versus other two groups combined, P = 0.0394
• Soil abatement group versus other two groups combined, adjusted for change in floor
dust lead concentration, P = 0.0077
• Comparison of all three groups, adjusted for covariates one at a time:
- Change hi floor dust lead concentration, P = 0.0084;
- Ethnicity/race category, P = 0.0020.
Soil abatement also showed some marginally significant effects wiith other covariate
adjustments:
September 1, 1995
5-80
DRAFT-DO NOT QUOTE OR CITE
-------
1 • Comparison of all three groups, riot adjusted for covariates, P = 0.1035
2 • Comparison of all three groups, adjusted for covariates one at a time:
3 - Change in floor dust loading from pre- to postabatment, P = 0.0641
4 - Change hi floor dust lead loading, P = 0.1070
5 - SES, P = 0.0720.
6
7 There was only one significant interaction term between treatment group and covariate
8 hi the nine models that had covariate adjustments, but the interaction between abatement
9 group and ethnicity was the most significant effect among all of the treatment group and
10 covariate effects that were tested. The interaction between ethnicity/race category and
11 treatment effect had P = 0.0006. Although the soil abatement group had a significantly
12 greater reduction in blood lead than the other groups in the tests described above, the dust
13 abatement group also had a smaller but statistically significant reduction in blood lead
14 compared with the control group when race/ethnicity was taken into account. It is clear that
15 sociodemographic factors may affect the response of child blood lead to soil remediation.
16 In the Boston study, it is possible that race or ethnicity was a surrogate for type or quality of
17 housing or some other characteristic of the household that affects the response of the children
18 in a household to soil or dust abatements.
19 The soil abatement group effect ranged from about 1.3 to 1.9 /ng/dL, whereas the dust
20 abatement group effect ranged from about 0.3 to 0.6 pig/dL. Covariate effects were not
21 statistically significant modifiers of treatment group effect, except for ethnicity/race.
22 However, including the covariates and interactions in the models greatly reduced the
23 uncertainty about the treatment effect size. It appears that the treatment effect for soil
24 abatement may be partially subsumed by changes in environmental variables, particularly by
25 changes hi the floor dust lead concentration. Floor dust loading may also play a role, but it
26 is not clear from these analyses whether the role of dust loading is as a modifier of floor dust
27 lead concentration or as a sociodemograpliic surrogate variable.
28
29 5.4.4.2 Results of Baltimore Study
30 The results of the repeated measures analyses of a variety of models for the Baltimore
31 study are shown hi Table 5-8. In contrast to the Boston study, the treatment group effect
32 was never statistically significant, but the covariate effects of age and of changes in dust lead
33 loading were statistically significant. There was a broad range of ages in the Baltimore
September 1, 1995
5-81
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 5-8. REPEATED MEASURES ANALYSES OF BLOOD LEAD IN
BALTIMORE STUDY FOR FIRST YEAR AFTER ABATEMENT, ADJUSTED FOR
DIFFERENCES IN ENVIRONMENTAL INDICES AND DEMOGRAPHICS
Statistical Significance
Comparison Groups
Soil Abatement
Ctrls 1 and 2
Soil Abatement,
Ctrl 1,2
Soil, Abatement,
Ctrls 1, 2
Soil, Abatement,
Ctrls 1, 2
Soil, Abatement,
Ctrls 1, 2
Soil, Abatement,
Ctrls 1, 2
Soil Pb > 500
Soil, Abatement,
Ctrls 1, 2,
Soil Pb «£ 500
Soil, Abatement
Ctrls 1, 2
Soil, Abatement
Ctrls 1, 2
Covariates
None
Dust Pb Cone
AAS
Dust Pb Cone
XRF
Dust Load
Dust Pb Load
None
None
Age, year
category
Age, year
Dust Pb Load
Comparison
df
2,176
2,105
2,102
2,111
2,105
2,61
2,112
2,169
2,98
* Time
P
0.3357
0.6546
0.2008
0.6306
0.9530
0.3633
0.6287
0.6450
0.9610
Covariate * Time Comp * Cov * Time
df p df p
—
1,105 0.4995 2,105 0.4680
i,102 0.9409 2,102 0.1670
1,111 0.9928 2,111 0.4703
1,105 0.0727 2,105 0.0910
7,169 0.0021
7,98 0.0139
1,98 0.00206 2,98 0.0714
1 study, so age was treated as a categorical variable with seven categories: age 0 years (0 to
2 11 months), age 1 year (12 to 23 months), and so on. Blood lead increased greatly by
3 Round 6 for children less than twelve months of age at Round 3, increased slightly for
4 children who were 12 to 35 months of age at Round 3, and decreased modestly for children
5 whose age at Round 3 was greater than 35 months. Children whose households had greater
6 reductions in dust lead loading had significantly smaller increases in blood lead than children
7 whose households showed no such reduction. However, on average, blood lead increased hi
8 all three groups, with insignificantly greater increases hi the soil abatement group than in
9 Control Groups 1 or 2.
September 1, 1995 5-82 DRAFT-DO NOT QUOTE OR CITE
-------
1 5.4.4.3 Results of the Cincinnati Study
2 The results of the Cincinnati study are shown in Tables 5-9 and 5-10. Comparison of
3 soil abatement or dust abatement groups with combined control groups was much less
4 informative than comparison with separate control neighborhoods. Based on discussions with
5 the Cincinnati investigators, it appears that, in spite of the small number of subjects, the
6 Mohawk neighborhood is a more appropriate control group for the soil abatement
7 neighborhood of Pendleton than was the much more remote neighborhood of Glencoe.
8 Mohawk and Pendleton had more similar housing than Glencoe, and were located in the Ohio
9 River Valley rather than on the surrounding hills.
10 Table 5-9 shows that there are substantial differences in changes hi blood lead among
11 the six Cincinnati neighborhoods during the first postabatment year. Differences in
12 treatment group are test-wise statistically significant when adjusted for changes hi dust load
13 at the ulterior entry (P = 0.018) or for changes in lead loading at the entry (P = 0.037).
14 When adjusted for age as well, the differences among neighborhoods were more pronounced
15 when adjusted for changes hi floor dust lead concentration (P = 0.029), floor dust lead
16 loading (P = 0.034), entry dust lead loading (P = 0.002), and entry dust load (P < 0.001),
17 and nearly significant when adjusted for changes in floor dust load (P = 0.055). This is
18 even more impressive because of the small sample size for Mohawk (N = 6 including floor
19 dust measurements, N = 8 for entry dust) and for Pendleton (N = 32 to 35).
20 In general, the three neighborhoods that received only dust abatement during the first
21 year (Back Street, Dandridge, and Findlay) were not significantly different and showed the
22 largest decreases hi blood lead. Glencoe children also showed a large decrease in blood
23 lead, which differed significantly from children hi Mohawk who showed a large increase in
24 blood lead. The children in the Pendleton neighborhood where soil abatement was carried
25 out showed a very small increase in blood lead, significantly larger than the distant
26 neighborhood of Glencoe, but smaller than the children in the nearby Mohawk neighborhood.
27 Table 5-10 shows results of testing a variety of models in which the soil abatement
28 neighborhood of Pendleton is compared with the proximate control neighborhood of
29 Mohawk. The differences in goodness of fit among the models in Table 5-10 is small, with
30 residual standard deviations ranging from 2.95 to 3.16 jug/dL. The overall treatment group
31 effect is not statistically significant in any of these models, but the interaction of treatment
September 1, 1995
5-83
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 5-9. REPEATED MEASURES ANALYSES OF BLOOD LEAD
IN CINCINNATI STUDY FOR FIRST YEAR AFTER ABATEMENT,
ADJUSTED FOR DIFFERENCES IN ENVIRONMENTAL INDICES
Statistical Significance
Comparison * Time
Comparison Groups
Pendleton vs. Other
Pendleton vs.
Controls
Pendleton vs.
Glencoe, Mohawk
Controls: Glencoe
vs. Mohawk
Dust Abate: Back,
Find, Dand
Nbhds
Nbhds
Nbhds
Nbhds
Nbhds
Nbhds
Nbhds
Nbhds
Nbhds
Nbhds
Covariates
None
None
None
None
None
None
Age
Age
Dust Pb Cone.
Entry
Dust Pb Cone.
Entry
Dust Pb Cone
Floor
Dust Load
Entry
Dust Load
Floor
Dust Pb Load,
Entry
Dust Pb Load,
Floor
Age
Dust Pb Cone,
Floor
df
1,154
Pend t
1,85
Pend t
2,84
Pend t
Glen 1
1,42
Glen 1
2,66
All 1
5,150
5,144
5,114
5,120
5,125
5,120
5,125
5,127
5,125
5,119
P
0.022
Other 1
0.077
Controls 1
0.018
Moha 1 1
0.016
Moha 1 1
0.549
0.048
0.001
0.000
0.011
0.247
0.018
0.376
0.037
0.407
0.029
Covariate * Time Comp * Cov * Time
df p df p
— — — —
—
—
—
— — — —
1,144 0.000 5,144 0.015
1,114 0.000 5,114 0.009
1,114 0.148 5,114 0.066
1,120 0.835 5,120 0.060
'l,125 0.571 5,125 0.958
1,120 0.394 5,120 0.814
1,125 0.920 5,125 0.511
1,127 0.302 5,127 0.719
1,125 0.916 5,125 0.966
1,119 0.000 5,119 0.102
1,119 0.872 5,119 0.819
September 1, 1995
5-84 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 5-9 (cont'd). REPEATED MEASURES ANALYSES OF BLOOD LEAD
IN CINCINNATI STUDY FOR FIRST YEAR AFTER ABATEMENT,
ADJUSTED FOR DIFFERENCES IN ENVIRONMENTAL INDICES
Statistical Significance
Comparison Groups
Nbhds
Nbhds
Nbhds
Nbhds
Comparison * Time
Covariates df p
Age 5,114 0.000
Dust Load,
Entry
.Age 5,119 0.055
Dust Load,
Floor
Age 5,121 0.002
Dust Pb
Load, Entry
Age 5,119 0.034
Dust Pb
Load, Floor
Covariate
df
1,114
1,114
1,119
1,119
1,121
1,121
1,119
1,119
* Time
P
0.000
0.646
0.000
0.931
0.000
0.781
0.000
0.976
Comp
df
5,114
5,114
5,119
5,119
5,121
5,121
5,119
5,119
* Cov * Time
P
0.022
0.949
0.217
0.815
0.051
0.951
0.144
0.772
1 group and covariate is slightly significant after adjustment for changes in dust lead loading at
2 the entry (P = 0.037) or dust loading on the floor (P = 0.0432). Dust lead loading on the
3 floor is not significant by itself, but becomes marginally significant (P = 0.097) when dust
4 loading is included in the model (P = 0.0207). Age category is highly significant, with
5 children whose age at the beginning of the study hi Round 1 was less than 12 months, and
6 modest decreases hi blood lead for children of age 2 years or older.
7 Although the evidence for a soil abatement effect is suggestive, it is hardly conclusive
8 hi the Cincinnati study. Some children hi both the Mohawk and Pendleton neighborhoods
9 had large increases hi blood lead during the first post-abatement year, possibly associated
10 with increases in dust lead loading and dust loading. This suggests that additional sources of
11 dust exposure may have been occurring that were not under control by the study. Although
12 some recontamination from other non-abated urban sources was expected, the magnitude of
13 these effects was larger than expected. This may be one of the major challenges hi doing
14 urban soil lead remediation.
September 1, 1995
5-85
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 5-10. REPEATED MEASURES ANALYSES OF BLOOD LEAD
IN CINCINNATI STUDY FOR FIRST YEAR AFTER ABATEMENT,
ADJUSTED FOR DDJFERENCES IN ENVIRONMENTAL INDICES:
MOHAWK VERSUS PENDLETON
Statistical Significance
Residual S.D.
/ig/dl
2.97
3.09
2.98
3.03
September 1, 1995
Covariates
AgeGRP
Dust Pb Cone,
Entry
Dust Load,
Entry
Dust Pb Load,
Entry
Dust Pb Cone,
Floor
Dust Load,
Floor
Dust Pb Load
Floor
Age GRP
Dust Pb Cone,
Entry
Dust Load,
Entry
Dust Pb Load,
Entry
Age GRP
Dust Pb Cone,
Floor
Dust Load,
Floor
Dust Pb Load,
Floor
AgeGRP
Dust Pb Cone,
Entry
Dust, Pb
Load, Entry
Comparison * Time Covariate
df p df
1,24 0.6613 5,24
1,24
1,24
1,24
1,24
1,24
1,24
1,27 0.4957 5,27
1,27
1,27
1,27
1,25 0.3454 5,25
1 75
— — — — — ™ JLj^«^
I 25
1,25
1,29 0.5527 5,29
1,29
1,29
* Time
P
0.001180
0.4590
0.4888
0.3908
0.8165
0.0463
0.0902
0.000068
0.3155
0.4074
0.1340
0.000164
0.1676
0.3579
0.2046
0.000031
0.5271
0.0445
5_86 DRAFT-DO NOT
Comp * Cov * Time
df p
—
— —
— —
— —
— —
— —
— —
—
1,27 0.1667
1,27 0.4041
1,27 0.1256
—
1,25 0.1615
1,25 0.2241
1,25 0.1592
—
1,29 0.2384
1,29 0.0370
QUOTE OR CITE
-------
TABLE 5-10 (cont'd). REPEATED MEASURES ANALYSES OF BLOOD LEAD
IN CINCINNATI STUDY FOR FIRST YEAR AFTER ABATEMENT,
ADJUSTED FOR DIFFERENCES IN ENVIRONMENTAL INDICES:
MOHAWK VERSUS PENDLETON
1
2
3
4
Residual S.D.
/ag/dl Covariates
3.06 Age GRP
Dust Pb Cone,
Floor
Dust Load,
Floor
3.13 Age GRP
Dust Load,
Entry
3.08 Age GRP
Dust Pb Load,
Entry
3.03 Age GRP
Dust Load,
Floor
3.16 Age GRP
Dust Pb Load,
Floor
3.04 Age GRP
Dust Load,
Floor
2.95 Age GRP
Dust Load,
Floor
Dust Pb Load,
Floor
Statistical Significance
Comparison * Time Covariate * Time Comp * Cov * Time
df p df p df p
1,27 0.3838 5,27 0.000191 —
1,27 0.0551 . 1,27 0.0607
1,27 0.0868 1,27 0.0426
1,31 0.4635 5,31 0.000051 —
1,31 0.0780 1,31 0.0776
1,32 0.2644 5,32 0.000030
1,32 0.0369 1,32 0.0345
1,29 0.2643 5,29 0.000265 —
1,29 0.0432 1,29 0.2582
1,29 0.8991 5,29 0.000397 —
1,29 0.4564 1,29 0.2460
1,30 0.5001 5,29 0.000336 —
1,29 0.0868
1,29 0.5830 5,29 0.000128 —
1,29 0.0207
1,29 0.0970
5.5 COMPARISONS USING STRUCTURAL EQUATIONS MODELS
The effectiveness of environmental lead intervention may be assessed in any of several
ways, depending on the purposes
of the analyses. One of the most important goals in the
analysis of environmental lead data from the USLADP is the identification of the effects of
September 1, 1995
5-87
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
5
6
7
8
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
different lead interventions on environmental pathways from lead sources through different
media (especially household dust) to which the child may be exposed. A generic structural
equation model is shown in Figure 5-52, and is analogous to individual segments of
Figure 5-4. This is an environment-only model and assumes that the soil and dust lead
interventions have no effects apart from those that can be identified by differences in lead
concentrations in soil and dust, dust lead loadings, and total lead loadings, and long-term
reductions in treatment group blood lead concentrations. Although these relationships are
expressed by a series of interconnected algebraic equations, they may be more easily
understood from the environmental pathway diagrams shown in Figure 5-52. The
assumptions of the model are as follows:
1. Preabatement dust loadings depend on sociodemographic variables that affect
household dustiness, such as the age of the house, and on environmental dust
sources such as chipping and peeling interior paint;
2. Pre-abatement soil lead concentrations are independent or exogenous variables that
may depend on exterior lead-based paint and on historic deposition of airborne
lead particles from stationary sources (e.g., lead smelters or nonferrous metal
processing operations) and from mobile sources (combustion of leaded gasoline);
3. Dust lead concentrations both pre- and postabatement are related to current soil
lead concentrations at the time of measurement and to other sources such as
deteriorating interior lead-based paint;
4. Dust lead loadings are the product of dust loading per unit area and the
concentration of lead in house dust, an exact mathematical relationship denoted
"X" in the figures;
5. Blood lead concentrations are related to lead in soil and to lead loading or
concentration in house dust at or shortly before blood leads are measured, to prior
or historic lead exposures that have accumulated a (primarily skeletal) body
burden of lead that contributes to current blood lead concentrations, and on the
child's age as well as many other individual behavioral or demographic factors;
6. Soil lead concentrations change very slowly over time, in the absence of
interventions;
7. Blood lead concentrations from stored body burdens decrease relatively slowly
over time, and hi children such as those hi the Boston USLADP who have had
several years of exposure to high concentrations of environmental lead with
consequently large skeletal lead pools, stored body burdens may account for
1-year postabatement blood lead concentrations that may be as high as 66% of the
September 1, 1995
5-88
DRAFT-DO NOT QUOTE OR CITE
-------
INDEPENDENT VARIABLE
Not predicted by other
components of the
system
solid arrow indicates
significant effect
STATE OR RESPONSE VARIABLE
If input changes, output changes
dashed arrow indicates
effect not significant
STATE OR RESPONSE VARIABLE
If input changes, output changes
Figure 5-52. Explanation of the terms and features of the structural equation model
diagram in Figures 5-53, 5-54, and 5-55.
Proabatement
DusliPb
Concentration
Preabatement
Dust Lead Load
Figure 5-53. Structural equation model for childhood exposure in Baltimore.
September 1, 1995 5-89 DRAFT-DO NOT QUOTE OR CITE
-------
No Soil Sol
Abatement Abatement
tPnabatement I Preabatement
Floor Dust | Floor Dust
Load
Lead Load
Preabatement
Blood Lead
X
Peak
Postabatwnent
Blood Lead
IPostabatement I Postabatam«nt
Dust Load Du*t Lead Load
A A
Y
Figure 5-54. Structural equation model for Boston.
Ho Sol So»
\
f \
KwtitMtora
rWdhbortx
toSLwd
t
J>l
x>d
>
r
PcxtitatMnent)
N4lghbo*oodL
8ld«nIkDustp
PbCone. }
\
\
>
Po*t»bl
Inlerioi
DuitPt
//
f
temwit
Cone.
\
i
\
k
4
PoadiMtwMnt
Interior Floor
Du«lPbLo«d
/,
Postabctmwnt
Intwfor Entry
DuatPbLoKl
['o»Ub«Iemont
Interior Enby
Du*tLo*d I
A A A
Figure 5-55. Structural equation model for Cincinnati.
September 1, 1995 5_90 DRAFT-DO NOT QUOTE OR CITE
-------
1 preabatement concentrations, which may severely limit the potential effectiveness
2 of any environmental lead intervention in treating currently lead-burdened
3 children, and suggests that lead intervention may be far more effective in
4 preventing lead poisoning in children who have never been exposed to elevated
5 environmental lead.
6
7 5.5.1 General Issues in Structural Equation Modeling
8 The purpose of structural equation modeling is to elucidate pathways for environmental
9 lead exposure from source to child. From this perspective, the development and testing of
10 pathway models for urban lead is an exploratory model-building activity that does not readily
11 lend itself to hypothesis testing. It is well known that "specification searches" such as step-
12 wise regression have complicated inferential properties (Learner, 1978), and the true P level
13 for an estimated regression coefficient may be quite different from the nominal P value.
14 An up-and-down search procedure was employed that started with a plausible pathway
15 diagram, and dropped nonsignificant blocks of parameters if all estimates of the same or
16 analogous parameters in different groups were zero or nonsignificant. New parameters were
17 added for each new pathway in the model, based on prior beliefs and on sample correlation
18 coefficients.
19 Structural equation models are useful hi evaluating hypothetical causal pathways among
20 multiple variables. This is particularly useful in assessing intervention studies in which
21 changes in one part of a system can have both direct and indirect effects on other
22 components of the system. The general framework for all of the models is shown in
23 Figure 5-52. Independent variables (covariates, predictors) are those measured components
24 of a system that are not predicted from other components. The independent variables are
25 functionally independent of each other, but may be correlated with each other. It is not
26 necessary to model an explicit causal pathway among the independent variables. Independent
27 variables are shown by elliptical figures.
28 In Figure 5-52, dependent variables are shown as rectangular figures. The dependent
29 variables of the system are assumed to have some predictive relationship to the independent
30 variables and to each other. Although it is not necessary to dwell on the concept that there is
31 a "causal" implication for any proposed predictive relationship, it should be noted that hi a
32 longitudinal lead study, most of the lead in yard soil at the earlier measurement will still be
September 1, 1995
5-91
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
there at a later measurement unless the yard soil is removed; some of the lead in house dust
will be left for later collection; and some of the lead hi the child's body (even hi blood and
soft tissues) will be circulating hi blood at a later measurement. Thus, estimates of lead
concentrations in earlier samples are expected to be predictive of measurements from later
samples, which are estimated of the same quantity, hi part. The models do not depend on
causal interpretations, however, but do assume a temporal direction in which the dependent
variables depend on values of other variables measured at the same tune, or measured
previously, but not on values measured hi the future. The direction of statistical dependence
is shown by a line with an arrow. The line is solid if the relationship is statistically
significant hi the study, otherwise the line is dotted.
5.5.2 Results of Structural Equation Model Analyses
5.5.2.1 Baltimore Study
The structural equation model (denoted SEM) developed for the Baltimore study is
shown hi Figure 5-53. The model has three dependent variables with estimated parameters:
(1) Pre-abatement floor dust lead concentration measured by AAS, denoted DCFAR1.
(2) Pre-abatement blood lead concentration measured at Round 3, denoted BCR3.
(3) Post-abatement blood lead concentration measured at Round 6, denoted BCR6.
The preabatement floor dust lead loading, denoted LLFAR1, is calculated from the
preabatement floor dust lead concentration DCFAR1 and from the preabatement total floor
dust loading denoted DLFR1, which does not involve unknown parameters:
LLFAR1 = DLFR1 * DCFAR1 / 1,000
where the factor of 1,000 converts dust loading hi mg/cm2 and dust lead concentration hi
/tg/g into dust lead loading hi pg Pb/cm2.
The model also has a number of independent variables:
• SCR1 = soil lead concentration, preabatement
• SCR4 = soil lead concentration, postabatement (soil abatement group only;
otherwise, SCR4 = SCR1 if no soil abatement)
September 1, 1995
5-92
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
• EP = exterior paint lead P-XRF
• IP = interior paint lead P-XRF
• DLR1 = total dust loading, preabatement
• AGE2 = (age in months at Round 3-36 months), squared.
A linear model was fitted in logarithmic form hi order to stabilize variances. The
parameters hi the model are denoted D_, B_, and A_, with affixes:
log(DCFARl) = log(DCO + DS * SCR1 + DE * EP + DI * IP)
log(BCR3) = log(BO + BC * DCFAR1 + BL * LLFAR1 + BE * EP + BI * IP +
BS * SCR1 + BA2 * AGE2)
log(BCR6) = log(AO + AB * BCR3 + AC * DCFAR1 + AL * LLFAR1 + AE * EP
+ AI * IP + AS * SCR4 + AA2 * AGE2).
The following equation defined dust lead loading, but had no parameters to estimate:
log(LLFARl) = LC * log(DCFARl) + LD * log(DLRl) - log(l,000),
where LC = 1 and LD = 1. The estimated parameters for two such models are shown hi
Table 5-11 and 5-12. All other parameters that were determined to be nonsignificant were
set to 0 hi the analysis reported here.
In Table 5-11, interior and exterior lead paint, and lead hi soil make marginally
significant contributions to floor dust lead concentrations hi these Baltimore residences.
However, preabatement blood lead shows little relationship to dust lead loading or exterior
lead paint in this model. On the other hand, postabatement blood lead is highly correlated
with dust lead loading, but only weakly associated with lead paint once the influence of
starting blood lead (parameter AB) is taken into account. Interior lead and dust lead loading
are somewhat confounded, because including dust lead loading tends to reduce the ulterior
paint lead contribution to pre- and postabatement blood lead to nonsignificant levels.
The primary contribution of interior paint for these children appears to be as an indirect
source of house dust. In Table 5-12, the contributions of soil lead, ulterior and exterior paint
to house dust lead concentration are all statistically significant. The contribution of interior
paint to blood lead pre- and postabatement is statistically significant, but interior paint does
September 1, 1995
5-93
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 5-11. BALTIMORE STRUCTURAL EQUATION MODEL
FULL INFORMATION MAXIMUM LIKELIHOOD METHOD
Dependent
Variable
Dust Lead
Cone., AAS-
PRE
Blood Lead -
PRE
Blood Lead -
POST
R2for
Log
Model
0.0828
-0.0043
0.5459
Predictor or
Independent Variable
If Control Group 1:
Intercept
If Control Group 2:
Intercept
If Soil Abate:
Intercept
Soil Lead Cone. -
PRE
Interior Paint Lead
XRF
Exterior Paint Lead
XRF (All groups)
Intercept
Age-Squared
Dust Lead Loading -
PRE
Exterior Paint Lead
XRF
Intercept
Age-Squared
Dust Lead Loading -
PRE
Interior Paint Lead
XRF
Exterior Paint Lead
XRF
Blood Lead - PRE
Coefficient ± S.E.
1328 ± 1519
504 ± 573
-131 ± 395
1.728 ± 1.257
203 ± 132
86.0 ± 56.6
9.76 ± 1.05
-0.00066 ± 0.00070
0.79 ± 2.91
0.118 ± 0.112
3.91 + 0.33
-0.00095 ± 0.00011
14.61 ± 2.18
0.036 ± 0.056
0.012 ± 0.022
0.5629 ± 0.0274
Units
«*
A*g/g
A*g/g
A*g/g per
Mg/g
^g/g per
mg/cm2
/ng/g per
mg/cm2
Atg/dL
^ig/dL per
month2
fjLg/dL per
1,000 /ig/m2
fj,g/dL per
mg/cm2
/ig/dL per
month2
fjLg/dL per
l,000)«g/m2
/jg/dL per
mg/cm2
jwg/dL per
mg/cm2
/Ag/dL per
/ig/dL
September 1, 1995
5-94
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 5-12. BALTIMORE STRUCTURAL EQUATION MODEL
FULL INFORMATION MAXIMUM LIKELIHOOD METHOD
Dependent
Variable
Dust Lead
Cone., AAS-
PP.F
JT JVCr
Blood Lead -
PRE
Blood Lead -
POST
R2for
Log
Model
0.0699
0.0211
0.5414
-•
Predictor or
Independent Variable
If Control Group 1:
Intercept
If Control Group 2:
Intercept
If Soil Abate:
Intercept
Soil Lead Cone. -
PRE
Interior Paint Lead
XRF
Exterior Paint Lead
XRF (All groups)
Intercept
Age-Squared
Dust Lead Loading -
PRE
Interior Paint Lead
Exterior Paint Lead
XRF
Intercept
Age-Squared
Interior Paint Lead
XRF
Exterior Paint Lead
XRF
Blood Lead - PRE
Coefficient ± S.E.
1111 ± 1204
326 + 394
-182 ± 288
1.656 ± 0.790
241 + 113
87.4 ± 38.5
8.96 ± 0.83
-0.00094 + 0.00071
0.135 ± 1.866
0.697 ± 0.287
0.108 + 0.081
3.41 ± 0.70
-0.00061 + 0.00021
0.648 ± 0.197
0.025 ± 0.049
0.5533 ± 0.0694
Units
Mg/g
/*g/g
Mg/g
/*g/g per
Mg/g
/ig/g per
mg/cm2
Mg/g per
mg/cm2
Mg/dL
/zg/dL per
month2
/zg/dL per
1,000/ig/m2
/zg/dLper •
1,000 jtig/m2
/zg/dL per
mg/cm2
Mg/dL
jug/dL per
month2
jtig/dL per
mg/cm2
jttg/dL per
mg/cm2
/zg/dL per
^g/dL
September 1, 1995
5-95
DRAFT-DO NOT QUOTE OR CITE
-------
1 not make a significant contribution to blood lead when dust lead loading is included as a
2 predictor of postabatement blood lead.
3 The models presented here do not include postabatement dust lead data because there
4 were a substantial number of missing values, 25/80 hi the soil abatement group, 4/21 in the
5 Area 1 nonabatement group, and 40/76 hi the Area 2 control group. Additional analyses
6 using non-missing postabatement dust lead data may be useful.
7
1 5.5.2.2 Boston Study
2 The structural equation model (denoted SEM) developed for the Boston study is shown
3 in Figure 5-54. The preabatement blood lead model had no statistically significant
4 parameters other than the intercept, so that all preabatement lead variables are taken as
5 independent variables. The model has four dependent variables with estimated parameters:
6 • Postabatement floor dust lead concentration at Round 4, denoted DCFR4
7 • Postabatement soil lead concentration at Round 3, denoted SCR3
8 • Postabatement floor dust loading at Round 4, denoted DLFR4
9 • Postabatement blood lead concentration measured at Round 3, denoted BCR3.
10 The pre- and postabatement floor dust lead loadings, denoted LLFR1 and LLFR4
11 respectively, are calculated from the preabatement floor dust lead concentrations DCFR1 and
12 DCFR4, and from the pre- and postabatement total floor dust loadings denoted DLFR1 and
13 DLFR4, which do not involve unknown parameters:
14 LLFR1 = DLFR1 * DCFR1 / 1,000
15 LLFR4 = DLFR4 * DCFR4 / 1,000
16 where the factor of 1000 converts dust loading in mg/cm2 and dust lead concentration hi
17 into dust lead loading hi /*g Pb/cm2.
18 The model also has a number of independent variables:
19 • SCR1 = soil lead concentration, preabatement
20 • DCFR1 = floor dust lead concentration, preabatement
21 • BCR1 = blood lead concentration, preabatement
22 • IP = ulterior paint lead XRF
23 • CPTO = total area of chipped and peeling paint
24 • DLR1 = total dust loading, preabatement
September 1, 1995
5-96
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
• AGE2 = (age in months at Round 3-36 months), squared
• PRAY = property age (0 if post-1940, 1 if pre-1940).
A linear model was fitted in logarithmic form in order to stabilize variances. The
parameters in the model are denoted D_, B_, and A_, with affixes:
log(SCR3) = log(SCNO * SCR1) if no soil abatement
log(SCR3) = log (SS) if soil abatement
log(DCFR4) = log(DCO + DS * SCR1 + DI * IP + DCP * CPTO + DCC *
DCFR1)
log(DLFR4) = log(DLO + DLD * DLFR1 + DLP * CPTO)
log(BCR3) = log(AO + AB * BCR1 + AC * DCFR4 + AL * LLFR4 + AI * IP +
AS * SCR2 + AA2 * AGE2).
The following equations defined dust lead loading, but had no parameters to estimate:
log(LLFRl) = LC * log(DCFRl) + LD * log(DLRl) - log(l,000),
log(LLFR4) = LC * log(DCFR4) + LD * log(DLR4) - log(l,000),
where LC = 1 and LD = 1. The estimated parameters for two such models are shown in
Table 5-13 and 5-14. All other parameters that were determined to be nonsignificant were
set to 0 in the analysis reported here. The interior paint variables were not significant and
were omitted from Figure 5-54.
In Tables 5-13 and 5-14, lead in soil makes a significant contributions to postabatement
floor dust lead concentrations in these Boston residences. However, preabatement dust lead
shows little relationship to interior lead paint or paint condition in this model. Dust loading
is significantly correlated with dust lead loading the preceding year (parameter DLD)
nonsignificant levels.
On the other hand, postabatement blood lead in Table 5-13 is highly correlated with
dust lead loading, but only weakly associated with lead paint once the influence of starting
blood lead (parameter AB) is taken into account. As shown in Table 5-14, the correlation of
postabatement blood lead with dust lead concentration is weaker than the association with
dust lead loading. Soil lead was not a significant direct predictor of blood lead.
The primary contribution of soil lead for these children appears to be as an indirect
source of house dust. In Tables 5-13 and 5-14, the contribution of soil lead to house dust
September 1, 1995
5-97
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 5-13. BOSTON STRUCTURAL EQUATION MODEL BLOOD LEAD
VERSUS DUST LEAD LOADING FULL INFORMATION
MAXIMUM LIKELIHOOD METHOD
Dependent
Variable
Soil Lead -
POST
Dust Lead
Cone. -
POST
Dust Load -
POST
Blood Lead -
POST
R2for
Log
Model
0.8453
0.0845
0.1357
0.3908
Predictor or
Independent Variable
If Soil Abate:
Intercept
If No Soil Abate:
Soil Pb - PRE
Intercept
Dust Pb Cone. - PRE
Soil Pb Cone. -
POST
Intercept
Dust Load - PRE
Intercept
Age-Squared (Peak at
36 Months)
Dust Lead Loading -
POST
Blood Lead - PRE
Coefficient ± S.E.
129 ± 15
0.832 + 0.104
892 ± 149
0.0111 ± 0.0208
0.1697 ± 0.0775
10.43 ± 2.94
0.2736 ± 0.0834
2.38 ± 0.48
0.00021 ± 0.00100
7.99 + 4.01
0.5961 ± 0.0409
Units
,g/g
jig/g per
Mg/g
Mg/g per
/tg/g per
mg/m2
mg/m2 per
mg/m2
Mg/dL
/ng/dL per
month2
/*g/dL per
1,000 fjig/m2
/Ltg/dL per
Aig/dL
1 lead concentration are statistically significant, as is the contribution of dust lead loading to
2 blood lead. In the Boston study, soil abatement produced a persistent reduction in soil lead,
3 which was associated with a persistent reduction in dust lead that accounted for a persistent
4 reduction hi blood lead during the first year after abatement. Recent analyses (Aschengrau
5 et al., 1994) show that additional decreases in blood lead occurred in the second year as
6 well, provided no dust recontamination occurred.
September 1, 1995
5-98
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 5-14. BOSTON STRUCTURAL EQUATION MODEL BLOOD LEAD
VERSUS DUST LEAD CONCENTRATION FULL INFORMATION
MAXIMUM LIKELIHOOD METHOD
Dependent
Variable
Soil Lead -
POST
Dust Lead
Cone. -
POST
Dust Load -
POST
Blood Lead -
POST
R2for
Log
Model
0.8485
0.0611
0.1374
0.4155
Predictor or
Independent Variable
If Soil Abate: Intercept
If No Soil Abate: Soil
Pb-PRE
Intercept
Dust Pb Cone. - PRE
Soil Pb Cone. - POST
Intercept
Dust Load - PRE
Intercept
Age-Squared (Peak at
36 Months)
Dust Lead Cone. -
POST
Blood Lead - PRE
Coefficient ± S.E.
132 ± 19
0.867 ± 0.125
940 ± 158
0.0105 ± 0.0220
0.1821 ± 0.0768
10.42 ± 2.79
0.2705 ± 0.0849
3.39 ± 0.48
-0.00193 ±
0.00111
0.225 ± 0.194
0.5834 ± 0.0440
Units
!"g/g
ptg/g per
^g/g
/*g/g
jtg/g per
/ig/g per
mg/m2
mg/m2 per
mg/m2
jig/dL
jig/dL per
month2
jig/dL per
1,000 jtg/g
pig/dL per
1 5.5.2.3 Cincinnati Study
2 The structural equation model developed for the Cincinnati study is shown in
3 Figure 5-55. Because the study collected a larger number of interior and exterior
4 environmental indices than did the Baltimore or Boston studies, it was possible to develop a
5 more detailed environmental pathway model than in the other studies. The Cincinnati model
6 has twelve dependent variables with estimated parameters:
7 • Preabatement neighborhood sidewalk lead concentration at Round 1, denoted
8 DCWR1
9
September 1, 1995
5-99
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
• Preabatement interior entry dust lead concentration at Round 1, denoted DCER1
• Preabatement floor dust lead concentration at Round 1, denoted DCFR1
• Preabatement floor dust loading at Round 1, denoted DLFR1
• Preabatement blood lead concentration measured at Round 1, denoted BCR1
• Postabatement neighborhood soil lead concentration at Round 4, denoted SCR4
• Postabatement neighborhood sidewalk lead concentration at Round 4, denoted
DCWR4
• Postabatement interior entry dust lead concentration at Round 4, denoted DCER4
• Postabatement floor dust lead concentration at Round 4, denoted DCFR4
• Postabatement interior entry dust loading at Round 4, denoted DLER4
• Postabatement floor dust loading at Round 4, denoted DLFR4
• Postabatement blood lead concentration measured at Round 4, denoted BCR4.
The pre- and postabatement floor dust lead loadings, denoted LLFR1 and LLFR4
respectively, and the interior entry dust lead loadings, denoted LLER1 and LLER4
respectively, are calculated from the preabatement floor and interior entry dust lead
concentrations DCFR1, DCFR4, DCER1, and DCER4, and from the pre- and postabatement
total floor dust loadings denoted DLFR1, DLFR4, DLER1, and DLER4, which do not
involve unknown parameters:
LLFR1 = DLFR1 * DCFR1 / 1,000
LLFR4 = DLFR4 * DCFR4 / 1,000
LLER1 = DLER1 * DCER1 / 1,000
LLER4 = DLER4 * DCER4 / 1,000
where the factor of 1,000 converts dust loading hi mg/cm2 and dust lead concentration in
into dust lead loading in jig Pb/cm2.
The model also has a number of independent variables:
• SCR1 = neighborhood soil lead concentration, preabatement
• DLER1 = ulterior entry dust loading, preabatement
• XMET = exterior trim paint lead, mean XRF
September 1, 1995
5-100 DRAFT-DO NOT QUOTE OR CITE
-------
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
• XMEW = exterior wall paint lead, mean XRF
• XMIT = interior trim paint lead, mean XRF
• XMIW = interior wall paint lead, mean XRF
• AGE2 = (age in months at Round 3-36 months), squared
• SIB = number of preschool children in household
• PRAY = property age (0 if post-1940, 1 if pre-1940).
A linear model was fitted in logarithmic form in order to stabilize variances. The
parameters hi the model are denoted S_, F_, E_, L_, D_, B_, and A_, with affixes:
log(SCR4) = log(SCNO * SCR1) if no soil abatement
log(SCR4) = log (SS) if soil abatement
log(DCWRl) = log(DWl + DWS1 * SCR1)
log(DCWR4) = log(DW4. + DWS4 * SCR4)
log(DCERl) = log(DEl + DEW1 * DCWR1 + DET1 * XMET + DEW1 * XMEW
+ DEY1 * PRAY )
log(DCER4) = log(DEC4 + DEW4 * DCWR4) if CONTROL group;
log(DCER4) = log(DED4 + DEW4 * DCWR4) if DUST ABATE group;
log(DCER4) = log(DES4 + DEW4 * DCWR4) if SOIL ABATE group;
log(DCFRl) = log(DFl + DFE1 * DCER1 + DFIW1 * XMIW + DEY1 * PRAY)
log(DCFR4) = log(DFC4 + DFEC4 * DCER4) if CONTROL group;
log(DCFR4) = log(DFD4 + DFED4 * DCER4) if DUST ABATE group;
log(DCFR4) = log(DFS4 + DFES4 * DCER4) if SOIL ABATE group;
log(DLER4) = log(DLEC4 + DLE4 * DLER1) if CONTROL group;
log(DLER4) = log(DLED4 + DLE4 * DLER1) if DUST ABATE group;
log(DLER4) = log(DLES4 + DLE4 * DLER1) if SOIL ABATE group;
log(DLFRl) = log(DLFl + DLFE1 * DLER1 + DLFY1 * PRAY)
log(DLFR4) = log(DLFC4 + DLF4 * DLFR1) if CONTROL group;
log(DLFR4) = log(DLFD4 + DLF4 * DLFR1) if DUST ABATE group;
log(DLFR4) = log(DLFS4 + DLF4 * DLFR1) if SOIL ABATE group;
log(BCRl) = log(BO + AK * SIB + ACW1 * DCWR1 + AL1 * LLFR1 + AA2 *
AGE2).
log(BCR4) = log(AO + AB * BCR1 + ACW4 * DCWR4 + AL4 * LLFR4 + AA2 *
AGE2).
The following equations defined dust lead loading, but had no parameters to estimate:
September 1, 1995 5-101 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
log(LLFRl) = LC * log(DCFRl) + LD * log(DLFRl) - log(l,000),
log(LLFR4) = LC * log(DCFR4) + LD * log(DLFR4) - log(l,000),
log(LLERl) = LC * log(DCERl) + LD * log(DLERl) - log(l,000),
log(LLER4) = LC * log(DCER4) + LD * log(DLER4) - log(l,000),
where LC = 1 and LD = 1. The estimated parameters for one such model is shown in
Table 5-15. All other parameters that were determined to be nonsignificant were set to 0 in
the analysis reported here. The interior paint trim variable was not significant and was
omitted from Figure 5-55.
In Table 5-15, lead in soil makes a significant contribution to pre- and postabatement
sidewalk dust lead concentrations in these Cincinnati neighborhoods. Both pre- and
postabatement interior entry dust lead shows a statistically significant relationship to
neighborhood sidewalk dust lead concentrations. Exterior lead paint on walls and trim
contributes significantly to preabatement ulterior entry dust lead concentrations hi this model,
even though these are "gut rehab" housing units. The dust lead pathway can be traced
further by statistically significant relationships between preabatement entry dust lead and
floor dust lead concentrations, and by a marginal statistically significant relationship between
postabatement entry dust and floor dust lead concentration hi the dust abatement
neighborhoods. Dust loading is not significantly correlated with dust loading at the ulterior
entry or floor a year earlier, but preabatement floor dust loading is significantly correlated
with interior entry dust loading in the same residence. This suggests a consistent but
complex pattern of movement of particles from the soil and other sources to the sidewalk and
surface areas outside these urban residential properties, then into the individual dwelling units
within the property.
Preabatement blood lead shows a significant relationship to dust lead loading at the
interior entry, but not to dust lead loading on the unit floor or sidewalk lead concentration.
On the other hand, postabatement blood lead hi Table 5-15 is more highly correlated with
sidewalk dust lead concentration than with interior entry or floor dust lead concentration or
loading, once the influence of starting blood lead (parameter AB) is taken into account. Soil
lead was not a significant direct predictor of blood lead, but its effect as an indirect source
can be traced along the soil-to-sidewalk-to-entry-to-floor dust pathway.
September 1, 1995
5-102 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 5-15. CINCINNATI STRUCTURAL EQUATION MODEL BLOOD LEAD
VERSUS SIDEWALK DUST LEAD CONCENTRATION ITERATED
TWO-STAGE LEAST SQUARES METHOD
Dependent
Variable
Soil Lead -
POST
Dust Lead
Cone. - PRE,
Sidewalk
Dust Lead
Cone. -
POST,
Sidewalk
Dust Lead
Cone. - PRE,
Int. Entry
Dust Lead
Cone. -
POST, Int.
Entry
R2for
Log
Model
0.8999
0.3989
0.1032
0.2902
0.0893
=====
Predictor or
Independent Variable
If Soil Abate: Intercept
If No Soil Abate: Soil
Lead - PRE
Intercept
Soil Lead - PRE
Intercept
Soil Lead - POST
Intercept
Property Age (0 = new,
1 = old)
Dust Lead Cone. - PRE,
Sidewalk
Exterior Trim Paint
LeadXRF
Exterior Wall Paint Lead
XRF
If Control: Intercept
If Dust Abate: Intercept
If Soil Abate: Intercept
Dust Lead Cone. -
POST, Sidewalk (All
Groups)
Coefficient ± S.E.
129 ± 5
0.898 ± 0.025
202 ± 301
5.84 ± 0.89
1587 ± 683
6.00 ± 2.75
90 ± 48
111 ± 168
0.033 ± 0.013
48.3 ± 27.4
78.4 ± 47.4
275 ± 113
-190 ± 434
263 ± 190
0.139 + 0.079
Units
/*g/g
/ig/g per
/ig/g
/ig/g
/ig/g per
/ig/g
Mg/g
/ig/g per
/ig/g
Mg/g
,g/g
/ig/g per
Mg/g
/ig/g per
mg/cm2
/ig/g per
mg/cm2
Mg/g
/ig/g
/*g/g
/ig/g per
/*g/g
September 1, 1995
5_103 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 5-15 (cont'd). CINCINNATI STRUCTURAL EQUATION MODEL
BLOOD LEAD VERSUS SIDEWALK DUST LEAD CONCENTRATION
ITERATED TWO-STAGE LEAST SQUARES METHOD
Dependent
Variable
Dust Loading
- POST, Int.
Entry
Dust Lead
Cone. - PRE,
Floor
Dust Lead
Cone. -
POST, Floor
Dust Loading
- PRE, Floor
R2for
Log
Model
0.1929
0.2022
0.3250
0.5675
Predictor or
Independent Variable
If Control: Intercept
If Dust Abate: Intercept
If Soil Abate: Intercept
Dust Loading - PRE,
Int. Entry
Intercept
Property Age
Dust Lead Cone. - PRE,
Int. Entry
Interior Wall Paint Lead
XRF
If Control: Intercept
Dust Lead Cone. -
POST, Int. Entry
If Dust Abate: Intercept
Dust Lead Cone. -
POST, Int. Entry
If Soil Abate: Intercept
Dust Lead Cone. -
POST, Int. Entry
Intercept
Dust Loading - PRE,
Int. Entry
Property Age
Coefficient ± S.E.
474 ± 243
3753 ± 1251
-2394 ± 1346
0.0027 ± 0.0077
22 ± 61
-96.6 ± 133.9
0.976 ± 0.239
48.1 ± 42.0
191 ± 45
0 (constr.)
340 ± 127
0.315 ± 0.175
141 ± 253
0.124 ± 0.522
129 ± 25
0.125 ± 0.035
-104 ± 101
Units
mg/m2
mg/m2
mg/m2
mg/m2 per
mg/m2
Mg/g
Mg/g
Mg/g per
Mg/g
mg/g per
mg/cm2
Mg/g
Mg/g per
Mg/g
Mg/g
Mg/g per
Mg/g
Mg/g
Mg/g per
Mg/g
mg/m2
mg/m2 per
mg/m2
mg/m2
September 1, 1995
5-104 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 5-15 (cont'd). CINCINNATI STRUCTURAL EQUATION MODEL
BLOOD LEAD VERSUS SIDEWALK DUST LEAD CONCENTRATION
ITERATED TWO-STAGE LEAST SQUARES METHOD
Dependent
Variable
Dust Loading
- POST,
Floor
Blood Lead -
PRE
Blood Lead -
POST
R2for
Log
Model
0.0789
0.2879
0.3430
Predictor or
Independent Variable
If Control: Intercept
If Dust Abate: Intercept
If Soil Abate: Intercept
Dust Loading - PRE,
Floor (All Groups)
Intercept
Age for Peak Blood
Lead
Age-Squared
Number of Preschool
Children
Dust Lead Loading -
PRE, Floor
Dust Lead Cone. - PRE,
Sidewalk
Intercept
Age-Squared
Dust Lead Cone. -
POST, Sidewalk
Blood Lead - PRE
Coefficient ± S.E.
202 ± 76
278 ± 71
73 ± 117
0.0477 ± 0.0328
10.09 ± 1.45
47.2 ± 19.1
-0.0026 ± 0.0028
0.33 ± 0.67
0.191 ± 0.124
0.078 ± 0.252
0.86 ± 6.04
0.0019 ± 0.0042
0.454 ± 0.488
0.5501 + 0.4468
Units
mg/m2
mg/m2
mg/m2
mg/m2 per
mg/m2
Aig/dL
months
/ng/dL per
month
/ig/dLper
child
pig/dL per
1,000 /ig/m2
jig/dL per
l,000jug/g
Mg/dL
ptg/dLper
month
/tg/dL per
1,000 Mg/g
jwg/dL per
Mg/dL
1
2
3
4
The primary contribution of soil lead, for these children appears to be as an indirect
source of lead in house dust. In the Cincinnati study, soil abatement did not produce a
persistent reduction in dust lead or blood lead during the first year after abatement.
September 1, 1995
5-105
DRAFT-DO NOT QUOTE OR CITE
-------
1 5.6 SUMMARY OF STATISTICAL ANALYSES
2 5.6.1 General Observations
3 This integrated assessment of the USLADP includes a reevaluation of the results of the
4 analyses carried out by the original investigators and of the conclusions reached by the
5 investigators based on then: analyses. While we have largely confirmed the numerical results
6 of the analyses, other interpretations of the results are also consistent with these numerical
7 findings and, hi some cases, may be more plausible than the conclusions published by the
8 investigators. We have also extended the results of the original investigations by carrying
9 out additional analyses, using a consistent set of powerful analytical techniques not available
10 when the original reports were published.
11
12 5.6.1.1 Combining Studies
13 There were substantial differences in the design of the three studies that precluded
14 completely identical analyses of the data. It was technically possible to create a combined
15 data set, given that all three studies included data on blood lead and hand lead before and
16 after abatement, as well as carefully coordinated measures of family demographic
17 characteristics, soil and dust lead at the child's residence. However, there were substantial
18 differences hi study design, such as the. characterization of the "control" groups, pre-
19 abatement paint stabilization, age distribution at the tune of abatement, ethnic and racial
20 characteristics of the populations, and pre-abatement soil lead exposure. Mathematically
21 similar measures of effect hi each study would therefore have very different interpretations,
22 and would not be clearly generalizable to other study designs, much less to soil lead
23 abatement in other communities. However, some parameters are the same, such as the
24 persistence parameter for blood lead used hi structural equation models.
25
26 5.6.1.2 Measurement Error
27 Statistical characteristics of these studies must be interpreted in the light of so-called
28 "measurement error". QA/QC procedures were instituted to minimize analytical errors in
29 the measurement of blood lead, soil lead, and dust lead concentrations. However, a larger
30 part of the possible difficulty hi reproducing lead measurements is likely to be found in the
31 necessity of sampling highly variable phenomena. Blood lead concentrations are known to
September 1, 1995
5-106
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
change over time as a function of changes in behavior (e.g., ingestion of soil or hand
washing), diet (intake of calcium, iron, lactate, vitamins, fiber, etc.), and metabolism
(thyroid function, etc.). Soil lead concentrations may change only slowly over time, but
there are obviously serious difficulties in sampling precisely the same location at different
tunes. This raises serious questions about the appropriate method for monitoring changes hi
soil lead over tune, or characterizing soil lead for potential child exposure in a yard or some
portion of a parcel of land. Finally, there are even more serious questions about defining a
dust lead concentration or dust lead loading for child exposure. Dust lead exposure depends
on the sections of bare or carpeted floor sampled, on the selection of rooms and sampling
areas, and on variable factors such as season, frequency of opening of doors and windows,
house cleaning, and other variable factors. In spite of these difficulties, there are statistically
strong correlations among lead in soil and dust, on child's hands and in child's blood that are
found hi almost all recent studies.
5.6.2 Summary of Results
The data presented hi this section lead to the following conclusions:
(1) Soil abatement hi each study effectively reduced the concentration of lead
in the soil in the areas where soil abatement was performed.
(2) In the Boston and Cincinnati studies, the effectiveness of soil abatement
was persistent through the end of the study. There were no foliowup
measurements of soil in Baltimore to demonstrate persistency.
(3) Exterior dust abatement, performed only hi Cincinnati, was not persistent,
indicating a source of lead other than soil at the neighborhood level.
(4) Interior dust with soil abatement, as performed in Cincinnati and Boston,
appeared to respond to subsequent changes in exterior dust and soil lead in
Cincinnati. Entry way measurements of lead concentration and lead load
may be a good indicator of the movement of environmental lead into the
living unit.
(5) Hand lead measurements often reflected general trends in blood lead
measurements and may be a reasonable estimate of recent exposure.
Hand lead, as measured in these studies, can be a useful complement to
blood lead measurements.
September 1, 1995
5-107
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
(6) Paint stabilization as performed on all homes with lead-based paint in
Boston (interior) and Baltimore (exterior), was intended to reduce the
potential confounding effects from contamination of soil and dust, but in
retrospect, paint stabilization itself may be a form of intervention in this
study.
5.6.3 Limitations of the Statistical Methods
The statistical methods used here were reasonable and appropriate, and could be used
by other investigators with access to standard statistical software packages. However, the
methods have certain limitations that should be understood. The repeated measures analyses
assume only that the response variables are correlated with each other, with no implication of
temporal causality. The goodness of fit of the models was significantly improved by use of
covariate analyses. Some repeated measures analyses require that the covariates have no
time dependence. In most applications in this chapter, only two tune points (before and after
abatement) were used and the pre-post difference in environmental covariates was used.
A problem arises if the response variable must be transformed, say by a logarithmic
transformation for blood lead or for hand lead, in order to reduce skewness and to stabilize
variances across treatment groups. The implied model for the original untransformed
variable is then multiplicative in treatment effects and random variation. This is probably
acceptable for the analysis of variance, but is likely to produce a physically or biologically
meaningless specification for the covariate model when the covariates are indicators of
distinct and additive sources of lead, such as soil lead and interior lead-based paint. The
logarithmic model does not reproduce the additive nature of the separate exposure pathways.
Extension of repeated measures analyses to covariates such as environmental lead levels
that change with time can be done using a single technique, structural equation modeling.
These methods provide more powerful interpretive tools. The availability of environmental
data to characterize time-varying lead exposures in the Boston and Cincinnati studies suggests
that more powerful statistical methods, such as structural equation models, could be more
appropriate.
September 1, 1995
5-108
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
5.6.4 Comparison Across the Three Studies
The effectiveness of soil lead abatement in reducing blood lead varied greatly among
the three cities. The variability in abatement effects is probably due to substantial differences
hi lead sources and pathways among the neighborhoods hi these studies. These differences
for each study are discussed below.
The Baltimore study had two neighborhoods, Upper Park Heights and Walbrook
Junction. The area to which abatement was assigned (Park Heights) had enrolled families
whose residences did not have soil lead levels that were high enough to justify abatement.
The soil lead levels in the nonabatement premises in Park Heights that were measured hi the
preabatement phase were not significantly smaller than those of the control premises hi
Walbrook Junction. Therefore, the nonabatement houses hi Park Heights were used as an
additional control group. Unlike the other two studies, the soil abatement in Baltimore was
not accompanied by interior dust abatement. There was essentially no significant effect of
soil abatement hi the abated houses, compared to the control group. Statistical covariate
adjustment in both repeated measures analyses showed that the differences in blood lead
levels both before and after abatement were significantly dose-related to ulterior lead-based
paint and (nonabated) ulterior dust. It is likely that interior paint contributed to child lead
exposure, either directly by ingestion of paint chips, or indirectly by the hand-to-mouth
exposure pathway, as follows:
ulterior paint => ulterior dust => hands => blood.
Cross-sectional and longitudinal structural equation analyses could be used to explore this
hypothesis. However, because there were no repeated measurements of household dust lead,
it will be very difficult to assess changes in exposure over tune except by use of hand lead
data. Concerning the Baltimore study, we conclude that:
It is likely that soil lead abatement had little effect on the primary factors
responsible for elevated child blood lead levels in these two neighborhoods,
•which appear to be interior lead-based paint and interior dust lead.
The Boston study was conducted with blood and hand leads measured at one
preabatement round and at about 8 months after abatement. Soil and dust lead measurements
were available for pre- and postabatement at about the same tune. These data allowed a very
September 1, 1995
5-109
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
complete analysis of blood lead responses to changes in dust and soil lead over time.
Relative to the no treatment group, the results showed clearly that there was a persistent
reduction in blood lead levels (1.5 /zg/dL) in the soil lead abatement children, and that, on
average, the postabatement blood leads were lowest in premises that had the lowest
postabatement soil lead and dust lead loadings. Interior and exterior lead paint were not
significant predictors of blood lead for Boston children. Concerning the Boston study, we
conclude:
When soil and dust lead levels show a persistent decrease as. a result of
effective abatement, blood lead levels also show a persistent decline.
Because the Cincinnati study had collected blood lead and environmental samples hi six
Cincinnati neighborhoods, analyses comparable to those reported for the Baltimore and
Boston studies can be made. After some analyses using models similar to those for
Baltimore and Boston, it became evident that the neighborhoods within each of their
treatment group were not comparable in every way. Although there was a strong dependence
of blood lead on environmental lead, particularly on hand lead and on current floor or entry
dust lead there was no clear pattern of change or response of interior dust lead levels after
abatement.
We are inclined to accept the conclusion of the Cincinnati investigators that blood and
dust lead levels were affected differently at different tunes and places by other events not
under their control. However, the dose-dependence exhibited in the models suggests that
reducing ulterior dust lead levels did reduce blood lead levels, at least for a while. The
problem is that the abatements did not always persistently reduce dust lead levels. We
therefore conclude that:
There were additional sources of environmental lead exposure that had
different effects on the neighborhoods during the course of the Cincinnati study
and were not related to the abatement methods used in the study. It will be
necessary to use other analysis methods, such as structural equations
modeling, in order to assign changes in Cincinnati child blood lead levels to
changes in lead exposure.
September 1, 1995
5-110
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
6. INTEGRATED SUMMARY AND CONCLUSIONS
6.1 PROJECT OVERVIEW
This project focuses on the exposure environment of the individual child, looking at
three indicators of exposure: blood lead, hand lead, and house dust lead. From the
perspective of the child's environment, changes in the soil concentration are expected to
bring about changes in the house dust concentration, the hand dust loading, and the blood
lead concentration.
In the past 25 years, concern for children with lead poisoning has steadily increased
with mounting evidence for the subtle but serious metabolic and developmental effects of lead
i
exposure levels previously thought to be safe. Childhood lead poisoning was formerly
considered a severe medical problem usually traced to swallowed chips of peeling lead-based
paint. Scientific evidence has systematically revealed deleterious effects of lead at lower
levels of exposure. Agencies such as the U.S. Environmental Protection Agency and the
Centers for Disease Control and Prevention (CDC) have repeatedly lowered the level of
concern for children's lead burden that recommends environmental or clinical intervention
from a blood lead level of 30 /xg/dL established in 1978 by CDC to 25 /*g/dL in 1985, just
prior to the start of the project, then to the present level of 10 jug/dL, which was defined hi
October 1991 by CDC as a blood lead level that should trigger community-wide prevention
activities if observed in many children.
The purpose of Urban Soil Lead Abatement Demonstration Project (USLADP) was to
determine to what extent intervention in the form of soil abatement in residential
neighborhoods would be effective as a means to reduce childhood lead exposure. Each of
the three studies in the project is a longitudinal study of the impact of an altered environment
on the lead exposure of children. The studies focused on evaluation of the exposure
environment of the children living mainly in inner city neighborhoods. Measurements of
lead in key external environmental media (e.g., soil, exterior and interior dust, and paint)
were obtained prior to soil abatement, along with more direct indices of personal exposure in
terms of hand wipes and blood lead levels. Abatement of soil lead generally involved
removal of contaminated soil and replacement with "clean" soil. Postabatement lead levels
September 1, 1995
6-1
DRAFT-DO NOT QUOTE OR CITE
-------
1 in the above media and children's blood lead were remeasured at varying intervals to
2 determine the effect of soil abatement, alone or in combination with paint stabilization or
3 dust abatement, on blood lead concentrations. There are few other longitudinal studies of
4 this type, and none of this scope or duration. Because the three studies were conducted
5 using mutually agreed upon protocols, with few exceptions, a common ground exists for
6 understanding an array of information available from the three individual studies that
7 broadens the base of information beyond the limits of a single study or location.
8 Although the three studies'were conducted independently, an effort was made to
9 coordinate the critical scientific aspects of each study in order to provide comparable data at
10 their completion. This effort included seventeen workshops where the study designs,
11 sampling procedures, analytical protocols, and QA/QC requirements of each study were
12 discussed with a goal toward reaching a common agreement. In most cases, a consensus was
13 reached on the resolution of specific issues, but the individual studies were not bound to
14 conform to that consensus or to adhere to it throughout the study. This procedure produced
15 similar studies with some differences in study design and experimental procedures.
16 The individual results for each of the three cities were originally presented at an EPA-
17 sponsored symposium in August 1992. These presentations included the data analysis and
18 conclusions for each of the three individual city studies. Following this open discussion with
19 the scientific community, the three research teams submitted their respective reports to the
20 designated EPA regional offices (Boston, Region I; Baltimore, Region III; and Cincinnati,
21 Region V). These reports and their associated data sets were then provided to EPA's Office
22 of Research and Development (ORD) and Office of Solid Waste and Emergency Response
23 (OSWER) for further analysis and preparation of this Integrated Report.
24 The EPA review of the study designs, chemical analytical procedures and data quality
25 measures has found no major flaws that would cast doubt on the findings of the individual
26 reports. The data sets submitted to EPA were systemically scrutinized for errors and
27 inconsistencies, and were reviewed and revised by the principal investigators for each of the
28 three cities prior to the completion of the analyses reported here. The few data corrections
29 found to be necessary were minor and would not have altered the conclusions of the
30 individual city reports.
September 1, 1995
6-2
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
This draft integrated report has reached its present form after an extensive review
process. First, the reports of the individual studies were peer reviewed by non-EPA experts,
revised, and presented to EPA in their final form, along with the data sets that were used as
the basis for the individual reports. These data sets were then reanalyzed by EPA using
rigorous statistical techniques to extract information not easily accessible from any individual
study. An earlier draft of integrated report was next written based on those initial analyses.
Following internal review and revision, the integrated report was released in draft form for
public comment and external review at an expert workshop. Further statistical analyses
(based hi part on peer review comment recommendations) have since been carried out, and
this draft of the integrated report incorporates changes reflecting the new analyses and earlier
comments from the external experts. Another round of review and revision of the draft
report is now being carried out prior to its final release.
Electronic copies of the underlying three cities data sets will be made available to
members of the scientific community for continued review and analysis along with the
release of the final version of this report. This continuing reanalysis means that new
perspectives on the USLADP data may emerge. Although it is unlikely that major findings
have been overlooked during these extensive review phases, it is not at all unreasonable that
still further information will be retrieved and reported by the extended investigations to be
made possible by this open policy for data release.
6.2 SUMMARY OF FINDINGS
6.2.1 EPA Integrated Report Results
This integrated assessment looks at the three individual studies collectively to determine
if a broad overview can be taken of the project results when each study is placed in its
correct perspective.
The key findings of this integrated assessment with regard to the Boston study are as
follows:
1. The median preabatement concentration of lead in soil was relatively high in
Boston, averaging about 2,400 /zg/g with few samples below 1,000 /xg/g.
September 1, 1995
6-3
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
2. Abatement of the soil effectively reduced the median concentration of lead in the
soil to about 150 jug/g (an average decrease of about 2,300 /jg/g).
3. Soil was clearly a part of the exposure pathway to the child, contributing
significantly to house dust lead.
4. Other sources of lead, such as interior lead-based paint were minimized by
stabilization.
5. The reductions of lead in both soil and house dust persisted for at least two years.
6. Blood lead levels were reduced by approximately 1.6 ptg/dL at 10 mo after soil lead
, abatement.
7. Additional reductions in blood lead of about 1.0 /ig/dL (relative to non-abated), were
observed at 22 mo postabatement for children in houses where the soil lead was
abated and the interior house dust lead was consequently reduced and remained low.
Thus, in the Boston study, the abatement of soil resulted hi a, measureable, statistically
significant decline in blood lead concentrations in children, and this decline continued for at
least two years. It appears that the following conditions were present, and perhaps necessary
for this effect: (a) a notably elevated starting soil lead concentration (e.g., in excess of
1,000 to 2,000 jiig/g); (b) a marked reduction of more than 1,000 /ig/g in soil lead
consequent to soil abatement accompanied by (c) a parallel marked and persisting decrease in
house dust lead.
These conclusions are consistent with those reported by the Boston research team. This
integrated assessment found no basis for modifying their conclusions, although we choose not
to express these findings as a broadly generalizeable linear relationship between soil and
blood, such as change in micrograms of lead per deciliter of blood per change in micrograms
of lead per gram of soil, because we believe that such a linear expression of abatement
effects is highly site specific for the soil-to-blood relationship. We found evidence that the
dust-to-blood relationship is more significant and, perhaps, more linear than the soil-to-blood
relationship.
With regard to the Baltimore analyses conducted for this integrated assessment, the
participants in the abatement neighborhood that did not receive abatement were treated as a
separate control group, rather than combined with the nonabatement neighborhood (as the
Baltimore research team did). The reason for this was to establish a control group not
September 1, 1995
6-4
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
influenced by differences between neighborhoods. This alternative approach used in this
integrated assessment had little impact on the statistical significance of soil abatement effects
as reported by the Baltimore research team.
The key findings of this integrated assessment for Baltimore are:
1. The preabatement concentrations of lead in soil were notably lower (i.e., averaging
around 500 to 700 /*g/g, with few over 1,000 /ng/g) than hi Boston.
2. The actual reduction of lead in soil by abatement was small (a change of about
400 /tig/g), compared to the Boston study (a change of about 2,300
3. Measurements of blood lead were made for only ten months following abatement;
and no significant decreases in blood lead consequent to soil abatement were
observed compared to non-abatement control group children.
4. Except for exterior lead-based paint, there was no control of other sources of lead,
such as the stabilization of ulterior lead-based paint (as done in Boston) or
abatement of house dust (as done hi Boston and Cincinnati).
5. Follow-up measurements of soil (except immediately postabatement) were not made
to establish the persistency of soil abatement, and its possible effects on house dust.
Thus, in Baltimore, where starting soil lead concentrations were much lower than hi
Boston and soil abatement resulted in much smaller decreases in soil lead levels and no
ulterior paint stabilization or dust abatement was performed, no detectable effects of soil lead
abatement on blood lead levels were found.
These conclusions are consistent with those reported by the Baltimore research group,
and are not inconsistent with those above for the Boston study. At soil concentratons much
lower than the Boston study, the Baltimore group would have likely been able to see only a
very modest change in blood lead concentrations (perhaps less than 0.2 jug/dL) assuming
similarity between the study groups in Boston and Baltimore and the same linear relationship
between change in soil concentration and change in blood lead. Furthermore, the interior
paint stabilization and house dust abatement performed hi Boston perhaps enhanced and
reinforced the impact of soil abatement on childhood blood lead, whereas in Baltimore, any
possible small impact of soil abatement would have likely been swamped by the large
reservoir of lead hi the interior paint and the large unabated amounts of lead hi interior house
dust.
September 1, 1995
6-5
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
As for the Cincinnati study, because of differences in the neighborhoods, we found that
combining neighborhoods into treatment groups often obscures important effects, and chose
to analyze each of the six Cincinnati neighborhoods as separate treatment groups. One
neighborhood, Back Street, had an insufficient number of participants and was dropped from
some analyses. The Back Street group started with nine families, but by Round 5 there was
only one participating family in the study. We also found that the two control
neighborhoods, Glencoe and Mohawk, were substantially different, and that the three
remaining treatment groups, Pendleton, Dandridge, and Findlay, were more comparable,
both demographically and hi geographic proximity, to Mohawk than to Glencoe.
On this basis, we concluded that, hi most cases, the effect of soil abatement could not
be clearly determined, and offer the following explanation for this conclusion:
1. Most of the soil parcels in each neighborhood were not adjacent to the living units,
and this soil was therefore not the primary source of lead hi house dust. Evidence
for this statement includes the observation that street dust lead concentrations are
much higher than soil concentrations, indicating there is a large source of lead
contributing to street dust hi addition to soil lead.
2. The preabatement median soil lead concentrations hi the three treatment groups
were about 300 jitg/g in Pendleton, 700 /ig/g hi Findlay, and 800 jttg/g hi
Dandridge, and the postabatement soil concentrations were less than 100 j^g/g, so
that the reduction of lead hi soil was small, as in Baltimore.
Evidence for the impact of dust abatement or dust and soil abatement consists of a
statistically significant difference between changes hi blood lead between Rounds 1 and 4,
approximately one year apart. Some Cincinnati neighborhoods showed decreased blood lead
concentrations in response to dust abatement or dust and soil abatement. The two
neighborhoods that received only ulterior dust abatement in the first year, Dandridge and
Findlay, showed a small decrease hi blood lead concentrations, compared to large increases
hi the nearest control group, Mohawk. The treatment group that received soil, exterior dust
and interior dust abatement, Pendleton, showed a smaller effect than did the Dandridge and
Findlay neighborhoods. After consultation with the Cincinnati research team, we suspect
that there was recontamination of street dust hi Pendleton during the study, probably caused
by demolition of nearby buildings hi the neighborhood.
The consistent theme across the outcomes for all three studies is that soil abatement
must be both effective and persistent in markedly reducing soil lead concentrations
September 1, 1995
6-6
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
accompanied by a corresponding reduction in house dust lead in order to result in any
detectable reduction of blood lead. The location of the soil relative to the exposure
environment of the child is important. In this project, the movement of lead from soil or
street dust into the home seems to be a key factor in determining blood lead concentrations.
Although these USLADP results provide substantial evidence for the link between soil or
street dust and house dust lead, there is insufficient information by which to clearly quantify
this relationship in terms of the lowest level of soil or street dust lead reduction that will
yield a measurable decrease of lead hi blood.
6.2.2 Application of Findings to Conceptual Framework of Soil Lead
Exposure Pathway
This integrated assessment attempts to answer the following question: If residential soil
is abated will blood lead concentrations decline? To confirm or reject this soil lead/blood
lead hypothesis, this report builds a framework of logical arguments described below. Each
step of the pathway from soil to blood must be scrutinized closely and related data examined
hi detail. This means that if dust lead derived from soil is not ingested, either directly or
after passing through other sources, then blood lead concentrations cannot respond to changes
in soil lead concentrations.
1. There is a substantial amount of lead in soil.
Lead was measured in soil hi the range of less than 50 jttg/g to more than
18,000 ptg/g. If a parcel of 100 m2 had an average of 500 /tg Pb/g soil, then the
upper 2 cm of soil on this parcel (about 4,000,000 g) would contain 2 billion /tg or
two kilograms of lead. Before abatement, there was an estimated 25,000
kilograms of soil lead on the participating properties of this project.
A 2-cm soil core was deemed better than a 15-cm core commonly used in previous
studies. When there is a decreasing gradient between the top and bottom of the
15-cm core, the effect is to dilute the concentration, giving a distorted picture of
what is available at the surface. In this project, some measurements were made of
the soil concentration in the bottom 2-cm of the 15-cm core in order to determine
the depth of excavation. The Boston study reported there was not a large gradient
between the top and bottom of the 15-cm core, as had been expected.
Finally, there is little information on the types of surfaces that a child plays on.
If these surfaces are mostly soil, as opposed to asphalt or concrete, then the soil
measurement may be a good estimate of exposure. However, exterior dust is
probably a better estimate of exposure from hard play surfaces (item 5 below).
September 1, 1995
6-7
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
Exterior dust represents lead from several sources, including soil, and may also be
a better estimate of the lead transferred to household dust.
2. Lead in soil can move to other compartments of the child's environment, such as
exterior dust.
Limited evidence for this statement was shown in the Cincinnati study. In the
Cincinnati study, the relationship between soil and exterior dust was found to be
very weak, giving rise to the next statement.
3. There are sources of lead other than soil that contribute to exterior dust.
Because the changes hi lead hi soil do not account for all of the changes hi exterior
dust, it is reasonable to conclude from the Cincinnati study that there are other
sources for lead in exterior dust. In Cincinnati, the soil parcels were not on the
individual properties of the participating families, as was the case in Boston and
Baltimore. There are no measurements of exterior dust hi the Boston or Baltimore
studies.
4.
Lead in exterior dust can also move into other components of the child's
environment, such as ulterior dust.
In the Cincinnati study, when exterior dust lead concentrations changed, interior
dust lead concentrations also changed. This was especially obvious when the
exterior dust sample closest to the residence was compared to the interior floor
dust sample taken just inside the entryway door.
A living unit with 130 m2 of floor space (1,400 ft2) and 1,000 ^g Pb/m2
(a relatively high value from tables hi Section 3.3) would have 130,000 ^
or less than 1 % of the lead available from soil in paragraph 1 above (see
Figure 6-1). Additional lead would be in rugs and upholstered furniture.
of lead,
5. There are sources of lead other than exterior dust that contribute to ulterior dust.
Taken individually, none of the studies decisively demonstrated this effect. The
most obvious source of lead inside the home is lead-based paint, which was
common hi the Boston and Baltimore studies, but less important in the Cincinnati
study. Because neither Boston nor Baltimore measured exterior dust,
measurements of ulterior dust in these studies, cannot easily be broken down into
contributions from lead-based paint and from exterior dust. However, structural
equation analyses on the Boston study showed a strong influence of both interior
and exterior lead-based paint on interior dust.
6. Lead in soil can move directly onto the child's hand.
September 1, 1995
6-8
DRAFT-DO NOT QUOTE OR CITE
-------
2 kg Pb in Soil
(500 ppm)
0.13g Pb
in dust
(1,000ng/nv?
1.6 kg Pb in paint
(2 mg/cm2)
) 0.5 kg Pb in paint
(0.9 mg/cm2)
Figure 6-1. Total amounts of lead in various compartments of a child's environment,
using the assumptions for concentration (soil, top 2 cm) or lead loading
(dust and paint) in parentheses. Although house dust is only a small
fraction of the total lead in the child's environment, it is the most accessible
component. The concentrations and loadings are illustrative, not typical.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
Conceptually, the transfer of lead from soil to the child's hand is difficult to
measure. A child playing outside usually gets soil on his/her hands, but it is not
certain whether this soil is adequately represented by a composite of 2 cm soil
cores.
7. Lead in exterior dust can move directly onto the child's hand.
There is no portion of these studies that directly measures this effect. Baltimore
reported that the lead loading on hands increased during the summer months, by
inference due to the increased playtime outside. During the interviews with the
family, questions were asked in all three studies about the activity patterns of the
children, including the amount of time spent outside, but none of the studies
attempted to assess the play activities immediately before the hand wipe sample
was taken.
8. Lead in interior dust can move directly onto .the child's hand.
September 1, 1995
6-9
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
In most cases, when interior dust changed, hand dust changed. Because hand dust
lead is only a measure of the amount of lead on the hand, not the concentration nor
the amount of dust, it is difficult to make a quantitative estimate of this pathway.
It is not likely that the amount of dust on the hand is strictly a function of the
amount of dust on the playing surface, as there is probably an equilibrium effect
where some dust falls off after time. There is no aspect of these studies that could
measure this interesting problem.
9. Lead hi interior dust can also move into other components of the child's
environment, such as food.
This pathway was not investigated by any of the three studies. Measurements of
lead in food before and after kitchen preparation would be required. Conceptually,
this lead and other routes such as the direct mouthing activities on toys, furniture,
and window sills is included in the measurement of interior dust when the
assumption is made that a child ingests about 100 mg dust/day by all routes and
through all activity patterns.
10. There are sources of lead other than dust that contribute to the child's lead
exposure.
In this project, lead was measured in drinking water once or twice during each
study. Low ambient levels (ca. 0.1 /*g/m3) of lead hi air (typical of U.S.
metropolitan areas hi 1990) were assumed, as were national averages of lead hi
food. Ethnic food preferences and individual use of cosmetics or other lead
containing products were not investigated.
6.3 INTEGRATED PROJECT CONCLUSIONS
The main conclusions of this Integrated Report report are two-fold:
(1) When soil is a significant source of lead in the child's environment, the
abatement of that soil will result in a reduction in exposure that will, under
certain conditions, cause a reduction in childhood blood lead concentrations.
(2) Although these conditions for a reduction in blood are not fully understood, it is
likely that four factors are important: (1) the past history of exposure of the child
to lead, as reflected in the preabatement blood lead; (2) the magnitude of the
reduction in soil lead concentrations; (3) the magnitude of other sources of lead
exposure, relative to soil; and (4) a direct exposure pathway between soil and the
child.
The basis for the first conclusion is: hi Boston, where the soil lead concentrations were
high and the contribution from lead-based paint was reduced by paint stabilization, there was
September 1, 1995
6-10
DRAFT-DO NOT QUOTE OR CITE
-------
1 a measurable reduction of blood lead concentrations. This reduction continued to increase
2 for two years following abatement in Boston.
3 Conversely, in Baltimore and Cincinnati, where soil was not a significant source of lead
4 relative to other sources, there was no measurable reduction of blood lead except hi cases
5 where those sources were also removed or abated. In Baltimore, these sources may have
6 been interior lead-based paint that was not stabilized, or house dust that was not abated.
7 In Cincinnati, the principle source of lead seemed to be neighborhood dust that may have
8 been contaminated with lead-based paint.
9 The basis for the second conclusion is: hi those cases where all important elements of
10 the exposure pathway were available for assessment, the structural equation model analyses
11 showed that preabatement blood lead concentration was a major predictor of postabatement
12 blood lead, suggesting that the remobilization of bone lead is a major component of the
13 measured blood lead.
14 All other factors being equal, the measurable reduction in blood lead was observed only
15 at higher concentrations of soil lead. In the absence of information about other sources of
16 lead, no clear statement can be made about the possibility of smaller reductions hi blood lead
17 at lower soil lead concentrations.
18 In spite of the recent successes in reducing exposure to lead by removing lead from
19 gasoline and canned food, lead exposure remains a complex issue. This integrated
20 assessment attempts to assess exposure to lead hi soil and house dust. Lead hi soil and
21 lead-based paint are closely linked in the child's environment. If there is exterior lead-based
22 paint, then soil lead is likely to be elevated with a consequent elevation hi house dust lead.
23 If there is interior lead-based paint, then efforts to reduce the impact of soil lead on house
24 dust will be only partially effective. The maximum reduction in lead exposure will not be
25 achieved unless both paint and soil abatement are implemented.
26 There is evidence from all three studies that lead moves through the child's
27 environment. This means that lead hi soil contributes to lead hi street or playground dust,
28 lead in exterior paint contributes to lead hi soil, and lead hi street dust contributes to lead hi
29 house dust. A more detailed analysis of the data may show the relative contribution from
30 two or more sources, but the present analyses imply that this transfer takes place.
September 1, 1995
6-11
DRAFT-DO NOT QUOTE OR CITE
-------
1 The analysis of the data from the three studies showed evidence that blood lead
2 responds to changes in house dust lead. There is also evidence for the continued impact of
3 other, independent sources following abatement of one source. This means that abatement of
4 soil or exterior paint does not necessarily reduce the contribution of lead from other sources
5 such as interior lead-based paint.
6 The conclusions of this report suggest that soil abatement alone will have little or no
7 effect on reducing exposure to lead unless there is a substantial amount of lead hi soil and
8 unless this soil lead is the primary source of lead in house dust. At a minimum, when
9 implemented, both soil abatement and interior dust removal should both be performed to be
10 fully effective. Conversely, soil abatement should be considered in conjunction, with paint
11 abatement when it is likely that soil will otherwise continue to contaminate house dust after a
12 paint abatement is completed.
13 From one perspective, decisions about soil abatement should be made on an individual
14 home basis. For an individual home, the owner or renter needs to know that the property is
15 safe for children. This report shows that, on an individual house basis, soil abatement may
16 reduce the movement of lead into the home and its incorporation into house dust. The
17 magnitude of this reduction depends on the concentration of lead in the soil, the amount of
18 soil-derived dust that moves into the home, the frequency of cleaning in the home and the
19 cleanability of the home. The number and ages of children and the presence of
20 indoor/outdoor pets are factors known to increase this rate of dust movement, whereas
21 frequent cleaning with an effective vacuum cleaner, use of entry dust mats, and removing
22 shoes at the door serve to reduce the impact of soil lead on house dust.
23 From another perspective, soil abatement at the neighborhood level poses problems not
24 pertinent to individual homes. Playground, vacant lot, and other plots of soil may pose an
25 immediate problem if they are accessible to children and there is a direct pathway for dust
26 generated by this soil to enter the home. Likewise, sources of lead other than soil may
27 contribute more to exterior dust than soil itself. The evidence in this report suggests that the
28 key to reducing lead exposure at the neighborhood level is to abate significant sources of lead
29 contributing to exterior dust, in addition to the soil and paint abatement that would be
30 performed on an individual property.
September 1, 1995
6-12
DRAFT-DO NOT QUOTE OR CITE
-------
7. REFERENCES
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
Aschengrau, A.; Beiser, A.; Bellinger, D.; Copenhafer, D.; Weitzman, M.; (1994) The Impact of Soil Lead
Abatement on Urban Children's Blood Lead Levels: Phase II Results from the Boston Lead-in-Soil
Demonstration Project. Environ. Res. 67:125-148.
Annest, J.L.; Pirkle, J.L.; Makuc, D.; Neese, J.W.; Bayse, D.D.; Kovar, M.G. (1983) Chronological trend in
blood lead levels between 1976 and 1980. N. Engl. J. Med. 308:1373-1377.
Barnett and Lewis (1984) Outliers in Statistical Data. John Wiley and Sons, NY.
Barry, P.S.I.; Mossman, D.B. (1970) Lead concentrations in human tissues. Br. J. Ind Med. 27:339-351.
Bornschein R. L., Clark C. S., Pan U. W., Succop P. A. et al. (1990). Midvale Community Lead Study.
Department of Environ. Health, University Cincinnati Medical Center. July 1990.
Bornschein R. L., Clark C. S., Grote J., Peace B., Roda S., Succop P.A. (1988). Soil lead—Blood lead
relationship in a former lead mining town. In: Environmental Geochemistry and Health, Monograph
Series 4, Lead in Soil: Issues and Guidelines. (Eds) B. E. Davies and B. G. Wixson. Science Review
Limited, Northwood, England, pp. 149-160.
Bornschein R. L., Succop P., Dietrich R. N., Clark C. S., Que Hee S., Hammond P. B. (1985). The influence
of social and environmental factors on dust lead, hand lead, and blood lead levels in young children.
Environ. Res. 38: 108-118.
Buncher C. R, Succop P. A., Dietrich K. N. (1991). Structural equation modeling in environmental risk
assessment. Environ. Health Persp. 90: 209-213.
Clark, S.; Bornschein, R.; Succop, P.; Peace, B.; Ryan, J.; Kochanowski, A.; (1988) The Cincinnati Soil-lead
abatement Demonstration Project. In: Environmental Geochemistry and Health, Monograph Series 4,
Lead in Soil: Issues and Guidelines. (Eds) B. E. Davies and B. G. Wixson. Science Review Limited,
Northwood, England, pp. 287-300.
David, O.J.; Wintrob, H.L.; Arcoleo, C.G. (1982) Blood lead stability. Arch. Environ. Health 37: 147-150.
Fuller, W. A. (1987) Measurement error models. New York: John Wiley and sons.
Grant, L.D.; Elias, R.W.; Goyer, R.; Nicholson, W.; Olem, H. (1990) Indirect health effects associated with
acidic deposition. National Acid Precipitation Assessment Program. SOS/T Report 23. U.S.
Government Printing Office, Washington, DC.
Klepper, S.; Kamlet, M.S.; Frank, R.G. (1993) Regressor diagnostics for the errors-in-variables model -
an application to the health effects of pollution. J. Environ. Econ. Management 24:190-211.
Learner, E.E. (1978) Specification searches: Ad hoc inference with non-experimental data. New York: John
Wiley and Sons.
Marcus A. H. (1991c). Relationship between soil lead, dust lead, and blood lead over time: A reanalysis of the
Boston lead data. Report from Battelle Columbus Division, Arlington Office, to USEPA Office of Toxic
Substances. Contract No. 68-02-4246.
September 1, 1995
7-1
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Marcus A.H. (1992). Use of site-specific data in models for lead risk assessment and risk management.
In: An Update of Exposure and Effects of Lead, B. Beck (Ed), Fund. Appl. Toxicol. 18: 10-16.
Marcus, A. H., Elias, R. W. (1994). Estimating the contribution of lead-based paint to soil lead, dust lead, and
childhood blood lead, Lead in Paint, Soil, and Dust: Health Risks, Exposure Studies, Control Measures,
Measurement Methods, and Quality Assurance, ASTM STP 1226, Michael E. Beard and S.D. Allen Iske,
Eds., American Society for Testing and Materials, Philadelphia, 1994.
Menton R.G., Burgoon D.A., Marcus A.H. (1994). Pathways of lead contamination for the Brigham and
Women's Hospital Longitudinal Lead Study, Lead in Paint, Soil and Dust: Health Risks, Exposure
Studies, Control Measures, Measurement Methods, and Quality Assurance, ASTM STP 1226, Michael E.
Beard and S.D. Allen Iske, Eds., American Society for Testing and Materials, Philadelphia, 1994.
Rabinowitz M. B. (1987). Stable isotope mass spectrometry in childhood lead poisoning. Biological Trace
Element Research. 12: 223-229.
Roberts, J.W.; Camaan, D.E.; Spittler, T.M. (1991) Reducing lead exposure from remodeling and soil track-in
in older homes. Air and Waste Management Association Paper 91-134.2, 84th Annual Meeting and
Exhibition, Vancouver, British Columbia, June 16-21, 1991.
Rothman, K.J. (1990) No adjustments are needed for multiple comparisons. Epidemiology 1:43-46.
SAS Institute, Inc. (1993) SAS/ETS® User's Guide, Version 6, Second Edition, Gary, NC: SAS Institute, Inc.
Succop P. A., O'Flaherty, Bomschein R. L., et al. (1987). A kinetic model for estimating changes in the
concentration of lead in the blood of young children. In: International Conference: Heavy Metals in the
Environment, (Eds) Lindberg S. E., Hutchinson T. C. New Orleans, September 1987. (EP Consultants
Ltd., Edinburgh, pp. 289-291).
U.S. Environmental Protection Agency (1986) Air quality criteria for lead. Research Triangle Park, NC: Office
of Health and Environmental Assessment, Environmental Criteria and Assessment Office; EPA report no
EPA-600/8-83/028aF-dF. 4v. Available from: NTIS, Springfield, VA; PB87-142378.
Wilkinson, L. (1990) SYSTAT: The system for statistics. Version 5.03 (1991). Evanston, IL: SYSTAT, Inc.
Wilkinson, L. (1992) SYSTAT for Windows: The system for statistics. Version 5.02 (1993). Evanston, IL:
SYSTAT, Inc.
September 1, 1995
7-2
DRAFT-DO NOT QUOTE OR CITE
-------
APPENDIX A:
GROUP MEAN PARAMETERS FOR EACH STUDY BY
SAMPLE TYPE, TREATMENT GROUP, AND ROUND
The data in Table A-l were derived using the PROC UNIVARIATE feature of SAS 6.10
(SAS, 1994). The treatment groups are as described in Chapter 5, using data identical to
that plotted in Figures 5-8 through 5-32. Data for blood lead concentration and hand lead
are calculated with one value for each child; for floor and window dust, one value for each
living unit; and for soil, one value for each property or soil parcel. The group assignments
and numbers of individuals are different from the individual study reports and different also
from the summaries of these reports hi Chapter 4. In particular, the data are different from
Tables 4-2 through 4-4.
September 1, 1995
A-l
DRAFT-DO NOT QUOTE OR CITE
-------
M
^<
o
II
%*
$
co
en
||
rH
0
Pw
VO
T— 4
-
1
o
«
a
5 n,
H o
rH
CO
00
1
1
i
en
00
T-H
ON
vo
en
T-H
S
o
m
S
8
00
en
vo
a
oo
c3
rH
1
en
en
§
6
s
in
oo
T-4
vo
T-H
oo
oo
cs
CS
in
oo
I
g
00
oo
vo
T-4
VO
T-4
•n
•n
en
0
en
OH
r/1
O
§
|
OO
rH
rH
rH
O
en
ro
S
en
o
i
o
T-H
T— 4
O
in
t-
m
g
o
oo
§
o
o
cs
o
8
en
Si
§
T-H
CO
0
0
U
£
W
(§
o
o
E
rH
rH
rH
5
1
O
S
1
o
oo
in
s
CS
-5°
s5
8
OO
1
vo
in
oo
in
O
en
in
oo
en
en
§
I
O
*
O
§
o
oo
T— 1
en
en
cS
oo
VO
in
cs
oo
T— 4
VO
T-H
vo
o
in
in
oo
in
CS
T-H
in
o
o
oo
cs
cs
T— 4
o
en
T-H
CO
O
PQ
o
VO
in
1
o
T— 4
00
g
T-H
T-H
in
f-
r-
1
cs
en
es
r— 1
S
1
rH
S
en
T-H
T— 4
o
in
in
en
en
en
ON
00
T-H
1
rH
8
1
rH
g
0
T-H
VO
T-H
en
T-H
m
cs
%
00
T— 4
in
|
R
m
^
1
1
o
o
in
cs
T-H
1
en
en
T-4
OH
CO
O
m
8
rH
g
o
8.
1
VO
en1
en
1
g
en
cs
g
1
T-4
o
ON
ON
O
en
t-
ON
cs
T-H
S
in
T— (
CS
cs
VO
oo
i
o
in
m
cs
in
8
T-H
in
T— 1
O
T-H
OO
vo
00
VO
in
en
cs
T-H
ON
O
ON
§
CO
O
m
•s
,3
B3
Q
g
n
E
cs
in
00
oo
VO
r-
ON
vq
oo
in
in
•n
oo
00
T-H
in
C3
O
en
es
1l
•S
S
T-H
en
i/i
ON
in
8
CS
oo
f-
a
en
en
en
T-H
vo
T-H
T-H
oo
en
en
en
vq
in
cs
00
oq
vo
f-
pj
es
o
cs
in
T-H
in
oo
ON
OO
VO
en
PA-
%s
September 1, 1995
A-2 DRAFT-DO NOT QUOTE OR CITE
-------
vo ;
°° !
OO
CO
a
CO
oo
00
g
oo
CO
s?
ON
CO
o
CO
s
cs
i-H
CS
ON
a
CO
CO
oo :
ON
co
ON
8
a
I
S5
VO
ON
VO
CO
VO
8
%
o
CO
vd
oo
ON
ON
12
ON
00
CO
8
00
t-
CO
CO
00
CO
cs
in
co
S
ON
co
co
O
%
R
8
CO
g
VO
vo
8
VO
«s
ON
06
ON
c-
VO
ON
vd
VO
ON
co
in
•n
ON
cs
CO
55
ON
VO
•n
co
co
t-
VO
bU
•-s
a
CO
s
ON
oo
vo
VO
oo
CS
ON
00
ON
CO
a
o
s
$
00
•n
VO
CS
VO
f-^
cs
S
oo
cs
cs
OO
o
t-^
CO
CO
CO
VO
ON
a
%
oo
vo
S
g
ON
CO
co
ON
8
vd
8
vd
a
00
r-
g
VO
o
ON
82
VO
VO
So
VO
o\
a
ON
o
g
ON
g
a
OO
CO
v
ON
CO
C--
oo
cs
v
OO
•*
vd
CO
VO
00
ON
ON
"
cs
OO
v
v
co
CO
ON
cs
CS
CS
in
co
&
cs
cs
CO
cs
in
cs
o
5
co
O
PQ
CM
co
I
CU
1
co
•a
S
O
O
o
E
1
September 1, 1995
A-3 DRAFT-DO NOT QUOTE OR CITE
-------
§
3
00
cs
CO
rH
00
S?
cs
cs
S
cs
a
s
$
S
I
cs
i
§
CO
S
3
CO
CO
ON
ON
cs
«S
i
CO
a
I
00
cs
^s
]
i
I
oo
VO
cs
cs
ro
CO
cs
ON
CO
cs
I
1
t--
i
8
ON
s
ON
CS
cs
cs
•n
ON
es
1
cs
CM
in
cs
cs
cs
5?
CS
cS
VO
cs
VO
ON
ON
vo
OO
CO
VO
3
ON
00
co
VO
ON
in
V-l
cs
CS
•n
C/3
§
M
oo
8
CQ
oo
§
PL,
%
pa
PH
•s
CO
o
1
1
Q
&
g
September 1, 1995
A-4 DRAFT-DO NOT QUOTE OR CITE
-------
O
1«
•S 5
%**
1
CO
a
g ^
"S oj
£
a
u
PH
VO
*
•a
Ǥ
Treatment
Group
Sample Type
00
VO
VO
3
a
^t
VO
„
VO
O
CM
T-H
T-H
oo
8
PQ
|
Q
1
^.
S
S
CM
T-H
CM
VO
i*
VO
Os
r-H
00
in
,,
ON
^
T-H
a
00
§
to
s
1
PQ
T-H
00
CO
T-H
CO
T-H
VO
oo
oo
vn
CO
VO
co
CM
£
T-H
ON
ON
T-H
CO
T-H
0;
T-H
s
«
oo
CO
to
oo
CO
CO
ON
•n
CO
p-
CO
VO
CO
s
CO
R;
VO
S
VO
8
r-H
CO"
•<*
^J.
VO
ON
«
o
CO
rM
oo
o
1 — I
o
t>
s
«
s
CO
%
00
oo
r-
O\
f-
T-H
CO
IN
„
CM
CO
T-H
PH
00
O
PQ
in
0
ON
CO
a
CM
T-H
^
o
»— '
•n
"*
co
CO
2
T-H
r-
T-H
00
CO
CO
VO
ON
m
in
T-H
CM
CM
VO
T-H
r~
CO
-
T-H
rM
vo
1
ON
VO
00
OO
T— (
O
VO
00
CO
VO
ON
VO
T-H
ON
T-H
•n
*.
i^-
e~
T-H
8
t~
o
o
CO
8
T-H
„.
ON
in
-
00
i
w
CM
in
•*
a
8
f-
O
•n
CM
S
0
CM
OO
35
CM
I
J
VO
o
oo
«
8
8
t-~
8
CO
oo
00
CO
•n
co
CM
>H-
§5
8
8
CM
O
O
VO
O
T-H
CO
CO
in
p-
ON
CO
O
CM
O
o
-------
CO
5
§
_o
I*
g
S
CO
1*
1-
eu
VO
2
1
Treatment
Group
CO
Os
CO
1-H
^
8
8
CO
i-H
8
o
CO
Tf
in
-
CO
CO
=Q
u
JO
3
I
«
CO
o
CO
S
8
0
8
VO
m
^
„
g
£
1-H
VO
T- i 1
8
S
0
i-H
8
^
CO
Tt
in
CO
oo
00
o
T-H
i-H
8
8
0
i-H
8
•n
•n
CM
T— (
T-H
jjj
CO
Z
U
0
X>
PH
|| Cincinnati
g>
vo
S
0
cS
CM
•S?
T-H
OS
^
rM
CO
i-H
Os
S
CO
i-H
OS
0
oo
00
«
8
'
oo
in
1-H
OS
CM
•rf
•*
Os
VO
CM
8
in
CM
CO
0
oo
oo
VO
CM
1-H
o
VO
S!
i-H
CO
.
ro
OS
VO
CM
S
-
1-H
T— 4
CO
C3
S
t-H
OS
Os
CM
VO
-
B
z
u
t-H
1-H
00
CM
•n
CM
1-H
00
1-H
CM
cs
1-H
i-H
T-H
8
CM
0
1-H
VO
i-H
^
"""
CM
CM
i-H
OO
oo
CO
S -
S&^
I.
September 1, 1995
A-6 DRAFT-DO NOT QUOTE OR CITE
-------
tt
Arithmetic
Mean
g
s
CO
a
JR
S^
i— t
S
VO
T— 1
Z
1
o
^_»
a g
S cS
H
u
«'
fll ^H
§
N
1
s
CO
•n
CO
1— (
VO
p
Q
£•)
f,
&4
*-H
£
b
|
oo
50
1— 1
SI
i-H
„
'Si
•Sh
•o
i
X3
S?
VO
a
a
§
S
ro
00
oo
CM
CO
CM
oo
§
oo
CO
CO
CO
s
o\
g
VO
VO
i-H
§
W
i_l.
Z
O
i
8
i
s
CO
CM
f-
00
J\
o
CO
CO
ro
00
CO
s
f-
oo
O\
ro
VO
3
,D
VO
VO
O
OO
%
oo
»
CO
1
5?
s
oo
2?
>n
1
•H
7\
VO
VO
1— 1
(X)
n
in
CO
cs
cs
VO
CO
s
co
CO
ON
co
CO
CM
VO
VO
g
i
£
8
m
September 1, 1995
A-7 DRAFT-DO NOT QUOTE OR CITE
-------
s
«
o
•s
*c
!
»
1|
a
i
VO
»— i
-
•o
|
4-*
s ^
80
£
PI
H
ju
1
o
o
o
0
o
o
*
m
S
o
U
a!
4->
Q
8
E
"o
.s
u
1
t
1
1
1
t
'
VO
"th
S
i
'
•
'
•
,
1
-
8
1
VO
CO
o
CO
s
r-
o
i—*
v— 1
©
w
§
0
i
VO
1—4
.8
8
CO
CO
s
0
e.
in
VO
CO
s
CO
CO
VO
s
CM
CM
CO
s
CO
oo
CO
1—4
fl
CO
s?
1
CO
o\
co
CO
CO
CO
-
0
o
0
o
o
o
*— 1
•n
1
'
1
1
t
,
1
VO
1
1
1
1
I
,
'
-
CO
CO
vo
o\
CM
1—4
CO
co
CO
CM
CO
CM
-
s
w
2
§
0
§
oo
s
in
in
co
CO
S
CM
1
CO
vo
r?
•n
t— i
CO
a\
*
CO
m
o
•n
oo
r-
1
CO
co
CM
a
•*
o
0
o
o
0
o
CM
in
1
I
1
I
1
1
1
VO
1
i
i
i
i
i
i
-
1
a
B
VO
VO
s
§
CO
CM
-
S
W
2
g
u
1
r-
i-H
CM
CO
CM
1
CO
CO
co
9
CO
CM
CM
a
1— 1
in
CO
VO
o\
oo
co
00
CO
CM
in
VO
CM
a
CO
jrj
i-H
i-H
1
oo
CO
t-
1—4
co
in
in
CM
CM
•<*•
o
O
o
0
0
0
a
„
1
I
1
1
1
,
•
VO
t
•
i
\
i
i
i
-
£
CM
§
S
CO
VO
CO
1—4
s
i-H
CM
CO
1-H
e
'~~'
z
z
u
CM
OS
CM
CO
s
s
o
in
1—4
'm
CM
T— I
OO
CM
CM
CM
8
CO.
fc
CM
CO
CO
i-H
s
1—4
o\
CM
CO
c^
s
s
o\
s
8
T-H
i-H
-
0
o
0
0
o
0
V
•o
1
1
1
1
•
1
1
VO
g
£*:
|e
§S
CO
September 1, 1995
A-8 DRAFT-DO NOT QUOTE OR CITE
-------
£
B
U
B
8
o'
a
o\
00
ON
!Q
00
a
00
CO
CO
CM
8
VO
CN
S
ON
CM
(N
00
cs
00
VO
a
£
(N
VO
VO
§
j
H
0
p,
bO
"So
September I, 1995
A-9 DRAFT-DO NOT QUOTE OR CITE
-------
•c
•s
vo
o\
o\
CS
VO
00
CO
VO
ON
vS
o\
CO
VO
a
s
VO
ID
in
o\
ON
VO
P«
cs
a
00
cs
o\
VO
in
(N
VO
-------
5
I
Arit
s
9
ex
VO
VO
00
ON
O\
0\
ON
o\
CN
CN
\o
00
00
3
Ox
CM
VO
VO
O
CE
H
a
VO
CN
00
VO
VO
cs
CM
VO
VO
CM
VO
CM
•0
VO
11
2 O
§
&
CO
•a
E3
'
b
September 1, 1995
A-l 1 DRAFT-DO NOT QUOTE OR CITE
-------
CO
8
O
§
c/
CM
VO
00
o\
o\
8
s
en
CM
vo
•ri
cs
s
CN
CM
00
VO
CN
o\
CN
CN
CN
CN
in
vo
in
S
VO
CN
CM
CM
CM
S
-------
3
«p
c0 ^
^Q
o>
H
vo
N
s
ts
i—
S
en
N
cs
CM
CO
00
ON
CM
b
September 1, 1995
A-13 DRAFT-DO NOT QUOTE OR CITE
-------
tt
5*
'SS
18
-------
ti
Arit
o
VO
VO
00
£
en
oo
t-
S
OS
VO
en
in
I
10
en
en
S
VO
eN
00
00
oo
VO
en
oo
OS
5S
8
£
-------
M
><
U
J
Jl
PH
&
s
?§
s~
t— t
a
i
VO
T-H
2
1
H
S o
a o
H
1
4J
o
ro
§
8
1
3N
CO
So
CO
CN
CN
•a
£
3
Q
>
.9
"o
•a
o
X)
i
1
CO
n
oo
CN
T-H
3
CO
B
i
1
VO
3
XI
XI
§
CN
ro
oo
30
T-H
ON
-
ro
n
X>
00
o\
CO
§
§
CO
T-H
CO
CN
">
I
1
-
1
1
,
1
VO
1
1
'
1
1
1
1
t-
1
CO
CO
1
§
oo
o
8
CN
ON
CO
T-H
g
H
2
z
U
1
X)
I
g
>n
T— 1
CN
r-H
(N
T-H
3
cs
»
1
1
s
oo
a
CO
VO
5
ON
CN
CO
ON
»
VO
o
1
o
ro
CN
OO
T-H
00
•*
'
a
o
oo
Js
1
ro
T-H
ON
•n
1
1
1
1
1
,
I
VO
1
1
1
!
I
,
•
-
N
O
X}
3O
T-H
00
T— i
Tt
s
CN
ON
T-H
I
£•*
2
2
O
VO
1
1
oo
1
ro
,
CO
ON
CN
O
O
o
ON
OO
VO
XI
n
X
oo
f-
co
•
a
VO
CO
8
N
X3
VO
•n
N
1
1
a
-
i
VO
VO
T— i
x>
ro
IN
CN
a
a
m
'
1
1
,
•
,
1
VO-
'
t
•
,
1
.
1
t-
00
CO
i
a
CO
a
o
CN
CO
T-H
ON
CN
-
g
pq
CO
'f£
O
<
1
h.
"
Q
r*~*
»9
o
T-H
N
1
i
1
ro
T-H
CO
CN
HI
J
o
r>
CN
ON
OO
o
XI
CO
ro
VD
s
s
CO
n
!
£
oo
T-H
3N
s
00
T-H
T-H
*
'
£
n
r>
CO
•n
a
T-H
CM
*
CN
,
•n
1
t
n
CN
oo
CO
§
T-H
a
VO
ON
10
o
§
XI
i
oo
T-H
T-H
r-
a
CO
8
T-H
s
VO
X}
T-H
£
T-H
r-H
T-H
§
W
h-4
2
U
Sp;
O
September 1, 1995
A-16 DRAFT-DO NOT QUOTE OR CITE
-------
I*
ffl
><
02 _
si
ivig
H-l p$
* C^
o §
^3
t» „
e£ aT
gs
H O
r-r-1 25
£3 P5
C^ C3
• OH
S ^
S S
(_i
«D
t?
o
vo
VO
c-
1
en
^H
VO
T-H
§
CM
CM
T-H
£
w
z
u
3
s
1
vo
CM
vo
§
CO
CM
CM
CM
oo
s
s
1
oo
CM
CM
CM
CO
3\.
CO
CM
T-H
1
£5
CO
T-H
o
co
CM
ON
CM
^
M
s
-
vo
T-H
CM
T-H
CO
CM
m
oo
1
T-H
T-H
i
oo
vo
CM
§
s
VO
n
in
s
CO
t-~
t--
§
T-H
oo
co
in
T-H
CM
oo
t^
i
5
T-H
T-H
CO
c-
T— 1
CM
T-H
VO
oo
o
CO
T-H
s
r
z
z
u
VO
VD
a
§
8
CO
ON
oo
T-H
T-H
oo
T-H
f~
CM
CM
o\
g
95
CO
CO
CM
T-H
ON
CM
CO
T-H
in
VO
oo
CM
T-H
o
T-H
ON
CO
^
i
s
00
t-~
CM
VO
oo
CM
oo
T— t
ON
CM
in
§
In
VO
vo
1
SI
T-H
oo
co
T-H
in
CO
vo
in
CO
CO
VD
ON
oo
vo
CM
T-H
CM
CM
CO
T-H
co
t--
T-H
T— t
S;
in
ON
T-H
CM
CO
CO
CM
CO
CO
ON
T-H
g
r
z
z
0
September 1, 1995
A-17 DRAFT-DO NOT QUOTE OR CITE
-------
o
•p
PM
S
i§
»— *
a
1
vo
55
T3
O
«^
S g,
ed sw
£
|
CO
VO
s
C—
co
8
oo
CO
fc
1
ON
OO
CO
ON
CS
1
Q
£
b
•«*•
CO
>n
y— i
i— i
§
vH
£
f-
CO
i
00
S!
CO
CO
i— c
CO
CO
r-
t—
oo
(S
CO
t— t
T*
f-
o
CO
1—4
CO
i
cs
•n
CO
S
oo
VO
VO
rH
cs
oo
r-H
CO
3
*
tn
•si-
CO
§
i
r-
ON
,
CS
cs
cs
VO
1
1
co
00
1—4
t— 4
o
CO
s
o
r-
c-
co
CO
00
8
00
i-H
r-H
ON
o\
^"^
CO
|Zj
U
ON
8
i-H
S
ON
9
VO
CO
ON
CS
ON
cs
00
cs
1—4
f-
co
cs
.
>n
oo
CO
1
n
cs
t-
CO
I
t— (
1
CO
CS
in
T— t
00
o
1—4
CS
cs
VO
ON
CO
CO
ON
CO
ON
CO
OO
CS
VO
September 1, 1995
A-18 DRAFT-DO NOT QUOTE OR CITE
-------
g
I
o
Arithmeti
Mean
£
en
IS
S ^
i— C
a
1
VO
z
1
o
rt
Treatment
Group
0>
>I
H
4n
1
*
o
1
en
T— 1
VO
es
oo
T-H
T)
3
1
1
•43
es
'fj
.3
u
1
in
m
•n
S)
m
o
^
OO
T— I
*— t
g
W
O
13
If
ON
*— «
8
1
vo
S
oo
a
•*
h
§
S
t^
oo
VO
oo
a
m
oo
3\
i
S
en
1
§
ON
§
2
x>
N
VO
SI
S3
DO
ON
*
VO
in
en
en
en
vo
ON
es
en
i— 1
2
OO
T— 1
i
i
8
ON
t-~
VO
«
S
55
g
H
U
1
en
N
I
,
i-H
5
J5
i§
i
i
m
en
l-H
es
*
ON
i— i
en
es
S
^
I
S
g
«
g
S
O
en
en
en
es
n
v-H
OO
§
C
-------
><
s
{*
L*B Vjf
U 3
a§
C^ Q
C) y
f-r-t
j
VO
rH
2
1
o
§ CL,
ttf u«
H
0
«
1
o
1
1
CN
VO
§
a
in
T3
ca
,3
£
Q
|
•a
b
TH
-------
£
C/3
8
§
•c
c£
§8
CM
CM
00
CO
a
oo
cs
cs
CM
CS
S3
in
ON
CO
s;
CO
cs
VO
VO
00
ON
S3
O\
a
VO
00
in
00
i
o
CM
CM
in
CO
VO
ON
CO
CO
Os
cs
ON
co
S -
£f r-r-l
O n.
C& K^ ]
3 S
VO
CM
vo
cs
I
Oi
(S
M
1
September 1, 1995
A-21 DRAFT-DO NOT QUOTE OR CITE
-------
o
— i
a
£
PM
VO
-
1
£
§ Oi
£
|
CO
1
§
CO
m
C-J
VO
0)
s
VO
m
CO
VO
8
-
g
w
Z
u
§
u
Q
ts
£
•«-H
C3
a
1
vS
VO
§
§
VO
ON
VO
VO
vo
in
a
cs
-**
,3
1
VO
1—4
VO
1-H
VO
T—4
0
1—4
in
r—4
CO
CO
oo
1—4
CO
T— 1
1
1
r-4
r—4
VO
CO
ON
1—4
Tfr
1
1
1
1
00
R
r—4
t-
VO
s
in
'
'
•
'
1
1
1
VO
'
1
•
I
1
•
'
-
1—4
f-
co
CM
in
00
o
CO
in
«
oo
VO
CO
oo
CM
1—4
1—4
CO
CO
1—4
1
I
oo
a
1—4
CO
VO
CO
CO
T-H
CM
s
r-4
in
co
oo
i
s
CS
CO
f-
-
in
CO
CO
i-H
g
8
oo
ON
T—4
00
ON
O
oo
r—4
-
£
OO
oo
8
VO
o
VO
in
O\
1—4
oo
in
o
CM
m
i
'
1
1
•
1
1
VO
\
\
\
\
'
1
•
r-
oo
CO
cs
oo
CO
CO
CM
1—4
CM
oo
1—4
T- 4
•n
in
r— i
T—t
§
N
U
VO
s
ON
00
CO
co
CO
r—4
CO
5
-
rH
VO
CM
CO
m
CO
ON
CM
CM
CO
1— 1
1—4
s
CO
00
00
ON
CO
f-
co
CO
§3
CM
CM
ON
1— 1
in
vo
i—i
£
-
CO
CM
C-
VO
CO
r-
in
CO
vo
rH
VO
OO
oo
1—4
•n
i
4
'
1
'
1
'
VO
'
1
\
\
'
'
'
t~~
1
^H
00
CO
t-H
§
r—4
1—4
VO
ON
CO
ON
r-
co
CO
-
1
1
o
o
SI
1— 1
in
00
c—
•n
m
m
m
CO
CM
R
m
VO
OO
£
3
in
CM
r-4
in
CO
c-
m
CO
CO
oo
a
1— 1
1—4
1—4
CO
i-H
5?
T-4
1—4
in
T—4
VO
in
m
CO
-
1
oo
r— 1
1—4
1— 1
OO
in
m
VO
CM
VO
in
m
CO
in
O
I
September 1, 1995
A-22 DRAFT-DO NOT QUOTE OR CITE
-------
1
J
H
o
Arithmet
Mean
OH
OO
S
ll
T— t
a
OH
VO
T-H
z
1
o
Pi
^
S &1
M
£0
i>
S
"&1
M
GO
'
I
•
•
,
i
VO
y
6
«
-t
0
ID
u
00
'is
.1
0
1
.
i
i
t
i
r-
on
&o
_a.
?\
0
8
o
oo
T— 4
O
oo
so
cS
T— 1
67
W
00
O
C
n
U
£
Q
^,
^
^
•o
'&
OO
*
XJ
a
vo
S
s
1
CO
§
bO
"Sh
^=5.
•o
1
o
X
£
vo
o\
0
X}
n
E
T— I
o\
o
CO
t>
CO
•o
CO
1
o
s
ro
1—4
30
O
uo
CO
CO
s
1
1
'
'
1
,
1
VO
'
t
1
'
1
,
1
r-
co
CO
>
1
VO
f-
co
82
CO
t— I
oo
co
1— I
£T
z
0
§
5
s
i
CO
§
If)
VO
o
T— 1
5
CS
•o
X)
VO
oo
o\
t-
X
s
*— 1
CO
oo
cs
1— 1
CO
CO
co
September 1, 1995
A-23 DRAFT-DO NOT QUOTE OR CITE
-------
w
fe«
•q
a
a
a
s
vo
n
in
S
en
in
in
vo
VO
VO
ir
in
00
n
ts
vo
VO
5?
vo
in
vo
ts
CM
cs
in
•n
en
vo
IS
Cd ft
3 e
•a
\O
>n
10
M
O
a
CO
•a
September 1, 1995
A-24 DRAFT-DO NOT QUOTE OR CITE
-------
51
if!
fiu ^^
^ Q
a
cs
CM
CM
00
VO
CO
CM
CM
CM
•
CM
8
CM
CM
CN
CM
CM
CO
00
00
00
VO
VO
CM
CM
#
Group
U
(J
o
u
"Si
co
September 1, 1995
A-25 DRAFT-DO NOT QUOTE OR CITE
-------
------- |