National Survey of Lead and Allergens in Housing

Volume I:  Analysis of Lead Hazards

FINAL REPORT

Revision 7.1

October 31, 2002

Prepared for:

Office of Healthy Homes and Lead Hazard Control

U.S. Department of Housing and Urban Development

451 7th Street, N.W.

Washington, D.C.  20410

Project Officers:

Warren Friedman, Ph.D., CIH

Joey Zhou, Ph.D.

Prepared by:

Westat, Inc.

1650 Research Boulevard

Rockville, MD  20850

Robert P. Clickner, Ph.D.

David Marker, Ph.D.

Susan M. Viet, Ph.D., CIH

John Rogers

Pamela Broene

This work was conducted under HUD Contract Number C-OPC-21356.

Table of Contents

Chapter	Page

EXECUTIVE SUMMARY		ES-1

1.	INTRODUCTION 	1-1

	1.1	Background	1-1

	1.2	Survey Objectives	1-3

		1.2.1	Technical Note on the Survey Weights	1-4

	1.3	Report Organization	1-4

2.	SURVEYED HOUSING POPULATION	2-1

3.	LEAD-BASED PAINT (LBP) HAZARDS IN HOUSING	3-1

	3.1	Definition of Significant Lead-Based Paint (LBP) Hazards	3-1

	3.2	Prevalence of Significant Lead-Based Paint Hazards in Housing	3-2

	3.3	Prevalence of Lead-Related Occupations or Hobbies	3-11

4.	LEAD-BASED PAINT (LBP) IN HOUSING	4-1

	4.1	Prevalence of Lead-Based Paint 	4-1

	4.2	Prevalence of Deteriorated Lead-Based Paint 	4-8

	4.3	Paint Lead Loadings in Housing	4-10

4.4	Comparison of Prevalence of Lead-Based Paint (LBP)

			to the 1990 LBP Survey	4-14

	4.5	Amount of Lead-Based Paint in Housing	4-17

5.	DUST LEAD IN HOUSING	5-1

	5.1	Prevalence of Dust Lead in Housing	5-1

	5.2	Dust Lead Loadings in Housing	5-3

	5.3	Association between Interior Dust Lead Hazards and Interior LBP
Condition	5-19

6.	RESIDENTIAL SOIL LEAD	6-1

	6.1	Prevalence of Residential Soil Lead Over All Sampled Locations	6-1

	6.2	Association Between Soil Lead and Exterior Paint Condition	6-6

	6.3	Prevalence of Bare Soil Lead in Children’s Play Areas	6-8

	6.4	Prevalence of Bare Soil Lead in the Rest of the Yard	6-11

	6.5	Comparison of Prevalence of Soil Lead to the 1990 LBP Survey	6-13

7.	QUALITY OF THE NATIONAL SURVEY DATA	7-1

	7.1	Statistical Concepts and Terminology	7-1

	7.2	Potential for Nonresponse Bias	7-2

		7.2.1	Analysis of Completion and Response Rates	7-2

7.2.2	Comparison of  “Hard to-Recruit” Versus 

	“Easy-to-Recruit” HUs	7-7

		7.2.3	Completion Rates in the Play Area Subsample	7-8

	7.3	Quality of Field Data Collection and Analysis	7-10

				7.3.1	Field Data Collection	7-10

				7.3.2	Laboratory Quality Control Samples	7-13

				7.3.3	Laboratory Selection Quality Assurance	7-14

	7.4	Paint Testing Quality Assurance	7-14

List of Appendices

Appendix A	SELECTED ADDITIONAL TABLES AND FIGURES	A-1

APPENDIX B	COMPARISON OF PROTOCOLS FOR THE HUD 1990 SURVEY

				OF LEAD-BASED PAINT (LBP) AND THE HUD NATIONAL

				SURVEY OF LEAD AND ALLERGIES IN HOUSING	B-1

APPENDIX C	Correcting for Classification Bias Due to Measurement 

				Error	C-1

APPENDIX D	CALCULATION OF SOIL RELATED ESTIMATES	D-1

List of Tables

Table ES.1	Summary Estimates of Prevalence of Lead-Based Paint and
Lead-Based

		Paint Hazards	ES-2

Table ES.2	Type of Lead-Based Paint Hazard	ES-3

Table 2.1	Characteristics of the Survey Population, with Comparisons to
the American

		Housing Survey (AHS) and the Current Population Survey (CPS)	2-4

Table 3.1	Prevalence of Housing Units with Significant Lead-Based Paint
(LBP) Hazards,

		by Selected Characteristics	3-5

Table 3.2	Prevalence of Significant Lead-Based Paint (LBP) Hazards by
Location in

		the Building	3-8

Table 3.3	Prevalence of Significant Lead-Based Paint (LBP) Hazards in
Housing Units

		with a Child Under 6 years of Age by Type of Hazard	3-9

Table 3.4	Prevalence of Significant Lead-Based Paint (LBP) Hazards in
Housing Units by 

		Type of Hazard and HU Age	3-10

Table 3.5	Prevalence of Housing Units with Selected Lead-Related
Characteristics	3-12

Table 3.6	Prevalence of Selected Lead-Related Characteristics in Homes
with Significant

	Interior LBP Hazards	3-12

Table 4.1	Prevalence of Lead-Based Paint (LBP) by Selected Housing Unit
(HU) 

		Characteristics	4-4

Table 4.2	Prevalence of Lead-Based Paint (LBP) by Location in the
Building	4-7

Table 4.3	Prevalence of Deteriorated and Significantly Deteriorated
Lead-Based Paint 

		(LBP) by Location in the Building	4-8

Table 4.4	Distribution of Housing Units (HUs) with Deteriorated and
Significantly 

		Deteriorated Lead-Based Paint (LBP) by Construction Year	4-9

Table 4.5	Distribution of Paint Lead Loading by Location in the Building
4-11

Table 4.6	Distribution of Paint Lead Loading by Location in the Building
and

		Construction Year	4-12

Table 4.7	Estimated Empirical Distribution Parameters of Paint Lead
Loadings by

		Interior Component Types	4-13

Table 4.8	Estimated Empirical Distribution Parameters of Paint Lead
Loadings by

		Exterior Component Types	4-13

Table 4.9	Percentage of Components with Lead-Based Paint by Component
Type

		And HU Age	4-14

Table 4.10	Estimated Empirical Distribution Parameters of Paint Lead
Loadings by Room

		Type	4-14

Table 4.11	Comparison of the Prevalence of Lead-Based Paint to the 1990
LBP Survey	4-15

Table 4.12	Amount of LBP by Painted Component	4-18

Table 5.1	Prevalence of Housing Units with a Dust Lead Hazard Somewhere
in the Home	5-2

Table 5.2	Distribution of Maximum Dust Lead Loadings in Housing Units by
Surface	5-5

Table 5.3	Distribution of Average Dust Lead Loadings in Housing Units by
Surface	5-7

Table 5.4	Distribution of Dust Lead Loading by Room and Surfaces	5-9

Table 5.5a	Maximum Floor Dust Lead Loading by Year of Construction	5-11

Table 5.5b	Maximum Window Sill Dust Lead Loading by Year of Construction
5-12

Table 5.5c	Maximum Window Trough Dust Lead Loading by Year of
Construction	5-13

Table 5.6a	Maximum Floor Dust Lead Loadings by Household Income	5-14

Table 5.6b	Maximum Window Sill Lead Dust Loadings by Household Income
5-15

Table 5.6c	Maximum Window Trough Lead Dust Loadings by Household Income
5-19

Table 5.7	Estimated Empirical Distribution Parameters of Dust Lead
Loadings by

		Surface Types	5-19

Table 5.8	Association Between Dust Lead Hazards and Presence and
Condition of

		Interior Lead-Based Paint	5-20

Table 5.9	Association Between Floor Dust Lead Loadings and Presence of
Interior 

		Lead-Based Paint	5-21

Table 5.10	Association Between Window Sill Dust Lead Loadings and
Presence of Interior 

		Lead-Based Paint	5-22

Table 6.1	Distribution of Maximum Bare Soil Sample Lead Concentrations
(All Sampled

		Locations)	6-3

Table 6.2	Distribution of Maximum Bare Soil Sample Lead Concentrations
(All Sampled

		Locations)	6-4

Table 6.3	Estimated Empirical Distribution Parameters of Soil Lead
Concentrations by

		Sample Site	6-6

Table 6.4	Association Between Bare Soil Lead Concentration and Presence
of

		Significantly Deteriorated Exterior LBP, All Sampled Locations	6-7

Table 6.5	Distribution of Maximum Soil Lead Concentrations in
Children’s Play Areas	6-9

Table 6.6	Distribution of Maximum Soil Lead Concentration in
Children’s Play Areas,

		by Construction Year	6-10

Table 6.7	Association Between Bare Soil Lead Concentration and Presence
of

		Significantly Deteriorated Exterior LBP, in Children’s Play Areas
6-11

Table 6.8	Distribution of Maximum Bare Soil Lead Concentrations in the 

		Rest of the Yard	6-12

Table 6.9	Distribution of Maximum Bare Soil Lead Concentrations in the
Rest of the

		Yard, by Construction Year	6-13

Table 6.10	Comparison of the Prevalence of Lead-Contaminated Bare Soil
in the

		National Survey and the 1990 LBP Survey	6-14

Table 7.1	Completion Rates by 1990 Census Block Group Characteristics
7-6

Table 7.2	Mean Percents for Completed and Noncompleted Housing Units by
1990

		Census Block Group Characteristics	7-7

Table 7.3	Comparison of “Easy-to-Recruit” Respondents Versus
“Hard-to-Recruit”

		Respondents by Reported Housing Characteristics	7-8

Table 7.4	Completions in Play Area Subsample	7-10

List of Figures

Figure 3.1	Prevalence of Housing Units (HUs) with Significant LBP
Hazards	3-4

Figure 4.1	Prevalence of Lead-Based Paint by Selected Housing Unit
Characteristics	4-2

Figure 4.2	Prevalence of HUs with LBP by Construction Year & Presence of
Children < 6	4-3

Figure 4.3	Prevalence of Lead-Based Paint by Location in the Building
4-7

Figure 4.4	Presence and Condition of LBP by Construction Year	4-10

Figure 5.1	Dust Lead Hazards by Construction Year	5-2

Figure 5.2	Dust Lead Hazards by Household Income	5-3

Figure 5.3	Distribution of Maximum Floor Dust Lead Loadings	5-4

Figure 5.4	Distribution of Maximum Window Sill Dust Lead Loadings	5-4

Figure 5.5	Association Between Dust Lead Loading on Floors and Window
Sills	5-10

Figure 5.6	Box Plots for Dust Lead Loadings by Room	5-17

Figure 5.7	Box Plots for Dust Lead Loadings by Surface	5-18

Figure 5.8	Association Between Dust Lead Hazards and Condition of
Interior 

		Lead-Based Paint	5-20

Figure 6.1	Distribution of Maximum Bare Soil Lead Concentrations	6-2

Figure 6.2	Box Plots for Lead in Soil Samples by Sample Site	6-5

Figure 6.3	Association Between Lead in Bare Soil and Significantly
Deteriorated Exterior

		Lead-Based Paint	6-8

Figure 6.4	Distribution of Maximum Bare Play Area Soil Lead
Concentrations	6-9

EXECUTIVE SUMMARY

The National Survey of Lead and Allergens in Housing (NSLAH) was
conducted under the sponsorship of the Department of Housing and Urban
Development (HUD) and the National Institute of Environmental Health
Sciences (NIEHS) to assess children's potential household exposure to
lead and allergens.  The NSLAH measured the levels of lead in dust,
soil, and paint, the prevalence of hazardous levels of lead, and levels
and patterns of various indoor allergens in dust in homes.  Volume I
includes the findings for lead hazards, and describes lead levels in
dust, soil, and paint in the nation’s housing by age, type,
geographical location, and exposed populations.  This Executive Summary
refers to the standards of HUD’s “Lead Safe Housing Rule” (24 CFR
35 Subparts B-R) on Federally-owned and -assisted housing with
thresholds from the EPA "Rule: Identification of Dangerous Levels of
Lead" (40 CFR Part 745).

Results: Extent of Lead-Based Paint and Lead-Based Paint Hazards in
Housing

An estimated 38 million homes (40 percent of all homes) in the United
States have lead-based paint somewhere in the building.  Of these, 20
million homes have lead-based paint present on both interior and
exterior surfaces, 9 million homes have lead-based paint only on the
interior, and another 9 million homes have lead-based paint only on the
exterior.

Although a large number of homes have lead-based paint, most of them
have relatively small surface areas of it.  The average home with
lead-based paint has an estimated 259 square feet of interior lead-based
paint and 996 square feet of exterior lead-based paint.  For comparison,
a room 10 feet by 12 feet with an 8 foot ceiling has a wall area of 352
square feet.

An estimated 24 million (26 percent) homes have significant lead-based
paint hazards somewhere in the building or on the premises; this is
consistent with earlier HUD estimates of 24 million homes.  Based on the
HUD Lead Safe Housing Rule, a home is said here to have a significant
lead-based paint hazard if one or more of the following conditions
exists:  lead-based paint with deterioration larger than de minimis
levels specified in the Lead Safe Housing Rule, dust lead loadings at or
above specified thresholds on floors or window sills; bare soil in
children’s play areas above specified thresholds; or more than 9
square feet of bare soil in the rest of the yard with lead
concentrations at or above specified thresholds.

Of the 16.4 million homes with one or more children under age 6, an
estimated 4.2 million (25 percent) have significant lead-based paint
hazards.  Of all 4.8 million homes with household incomes under $30,000
and one or more children under age 6, an estimated 1.2 million (25
percent) have significant lead-based paint hazards.  Thus, one in four
homes with young children among the residents have significant
lead-based paint hazards.  Table ES.1 summarizes these basic estimates
of the prevalence of lead-based paint and significant lead-based paint
hazards.

Table ES.1	Summary Estimates of Prevalence of Lead-Based Paint and
Lead-Based Paint Hazards

Housing Unit Characteristica	

Number of Housing Units (millions)	Number of Housing Units with
Lead-Based Paint

(millions)	Number of Housing Units with Significant Lead-Based Paint
hazards

(millions)

Total housing units	95.7	37.9	24.0

One or more children under age 6	16.4	5.3	4.2

One or more children under age 6, less than $30,000/year household
income	4.8	1.4	1.2

a “Housing units” include permanently occupied, noninstitutional
housing units in which children are permitted to live.

Of the 24 million homes with significant lead-based paint hazards, an
estimated 15 million have interior dust lead hazards, 14 million have
deteriorated lead-based paint at or above de minimis levels and 6
million have soil lead hazards, as displayed in Table ES.2.

Table ES.2	Type of Lead-Based Paint Hazard

Type of Hazard	Number of Housing Units

(millions)	Percentage of Housing Unitsa

	Estimate	95% CIb	Estimate	95% CI

Significantly Deteriorated lead-based paint	13.63	10.9 – 16.3	14	11-17

Interior lead-contaminated dust	15.47	13.0 – 18.0	16	14-19

Lead-contaminated soil	6.46	3.1 – 9.8	7	3-10

Any lead-based paint hazard	24.03	21.3 – 26.7	25	22-28

a	All percentages are calculated with total housing units (95,688,000)
as the denominator; percentages may not total 100% due to rounding. 

b	95% CI = 95 percent confidence interval.  

Dust lead levels above the Lead Safe Housing Rule’s standards are
associated with the presence of interior lead-based paint.  An estimated
29 million homes have some interior lead-based paint, of which 39
percent have dust lead levels above the Lead Safe Housing Rule’s
standards.  In contrast, only 6 percent of the 67 million homes without
interior lead-based paint have dust levels above the Lead Safe Housing
Rule’s standards.

Soil lead levels above the Lead Safe Housing Rule’s standards are
associated with the presence of deteriorated exterior lead-based paint. 
Homes with deteriorated exterior lead-based paint above de minimis
levels are six times more likely to have soil lead hazards than are
homes free of such deteriorated lead-based paint.

This most recent HUD survey shows that the number of housing units with
lead-based paint has declined from 64 million in 1990 to 38 million ten
years later.  On the other hand, the prevalence of homes with
deteriorating lead-based paint increased slightly, from 19% to 22%.

Survey Design and Methodology

The principal lead-related purpose of the NSLAH was to develop a
scientific description of the existing lead levels in paint, dust, and
soil in the nation’s housing. Additional objectives were to obtain
data to:  (1) estimate the number and percent of homes with dust and
soil lead levels above selected thresholds;  (2) identify sources of
lead in dust in housing, e.g., paint and soil;  (3) permit future
analyses of lead hazard control strategies and costs, e.g., quantities
of deteriorated painted surfaces; and (4) permit future analyses for
regulation, policy, and guidance that minimize regulatory and program
implementation burden.

The target population included approximately 96 million homes, of the
112 million total homes in the nation, including single- and
multi-family buildings and manufactured housing units, e.g., mobile
homes and trailers.  Homes built in all age categories were included. 
Vacant housing, group quarters, and hotels and motels were excluded for
reasons of difficulties gaining cooperation.  Housing where children
were not permitted to live, e.g., elderly care facilities, were excluded
because the primary interest was in children’s exposure to lead. 
Thus, 16 million units out of 112 million total units were excluded from
this survey.

The main field survey was conducted in 1998-1999, with an augmentation
of the soil sampling in 2000 to include additional play area samples.  A
nationally-representative sample of 1,984 homes was drawn from 75
clusters (each a metropolitan statistical area (MSA) or a group of
counties that have a minimum population of 15,000) called primary
sampling units (PSUs). A total of 831 eligible homes were recruited and
completed the survey.

Four rooms were randomly selected for environmental sample collection
and testing from each of four room types:  kitchen, common living areas,
bedrooms (preferably those occupied by children), and other rooms.  In
each of these four rooms floor, window sill, and window trough dust
samples were collected, painted surfaces were measured for lead content,
and the condition of painted surfaces was assessed.  Outside the
building, soil samples were taken and exterior painted surfaces were
tested.  A floor dust sample was collected in the interior common area
of multi-family buildings.

Measurements of lead in paint were made by State- or EPA-certified
lead-based paint inspectors using an XRF analyzer and a protocol based
on HUD's 1997 Guidelines’ inspection procedure.  The instrument model
used does not require making substrate corrections, nor have an
inconclusive range, both of which involve destructive sampling of
painted surfaces.  One XRF reading was made per painted component in
each room, approximately in the center of a randomly selected quadrant
of the total building component surface area.

Single wipe dust samples were collected by the technique described in
ASTM E 1728-95. Floor dust samples were collected in the center of the
largest open floor area in the room.  The floor samples in the major
entrance and interior common area were collected approximately six
inches away from the center of the doorway.  One-square-foot templates
were used for floor samples.  Window sill and trough samples were
collected from a random, openable window in each selected room.  The
entire area was wiped for window sill and trough samples (up to two
square feet).  All dust samples were analyzed by flame atomic absorption
spectrophotometry, using the Environmental Protection Agency’s
(EPA’s) SW-846 method 3050 digestion method and the American
Industrial Hygiene Association’s Environmental Lead Laboratory
Accreditation Program (ELLAP) proficiency testing procedures.  The
laboratory that performed the analysis was NLLAP – recognized for dust
lead analyses.

Soil sampling was conducted in accordance with core sampling procedures
described in ASTM E 1727-95.  Only the top one-half inch of each soil
core, i.e., that portion most accessible to children, was included in
the sample.  Where necessary, grass or leaf covering was gently removed
before taking the core.  Soil samples were taken outside the building at
the major entrance, and along the dripline and mid-yard area of two
sides of the building.  Soil samples were collected from children’s
play areas in a subsample of 375 homes.  Soil samples were analyzed by
inductively-coupled plasma atomic emission spectroscopy, using the
SW-846 digestion method and the ELLAP proficiency testing procedures.
The laboratory that performed the analysis was NLLAP – recognized for
dust lead analyses.

1.  INTRODUCTION

The National Survey of Lead and Allergens in Housing (NSLAH) was
conducted under the sponsorship of the Department of Housing and Urban
Development (HUD) and the National Institute of Environmental Health
Sciences (NIEHS) to assess children's potential household exposure to
lead and allergens, i.e., to estimate the levels of lead in dust, soil,
and paint, the prevalence of hazardous levels of lead, and levels and
patterns of various indoor allergens (allergy-inducing substances) in
dust in homes.  Combining the goals of HUD and NIEHS into a single
survey saved significant public funds, reduced the survey response
burden on the public, and substantially reduced the time required to
obtain the data needed by both agencies for their ongoing primary and
secondary prevention activities.

This report, Volume I, includes the findings for lead hazards, and
describes lead levels in dust, soil, and paint in the nation’s housing
by age, type, geographical location, and exposed populations.  In
addition, the report estimates the number and percent of homes with dust
and soil lead levels above selected thresholds, especially thresholds in
HUD’s Lead Safe Housing Rule (24 CFR Part 35 Subparts B-R,
Requirements for Notification, Evaluation and Reduction of Lead-Based
Paint Hazards in Federally Owned Residential Property and Housing
Receiving Federal Assistance, effective September 15, 2000), and EPA's
Final Rule: Identification of Dangerous Levels of Lead (40 CFR 745),
effective March 6, 2001.

1.1	Background

Lead is a toxin that affects the central nervous system and is
particularly damaging to the developing nervous system of young children
and fetuses.  High blood lead levels can result in convulsions, mental
retardation, and even death. Research has shown that even low lead
levels can have serious health consequences.  These include reduced
intelligence and short-term memory, slower reaction times, poorer
hand-eye coordination, reduced height, hearing problems, and numerous
behavioral problems.

Although there are many sources of lead in the environment, including
drinking water, food, emissions from gasoline combustion, and industrial
emissions, it is clear that lead-based paint (LBP) plays a major role in
high blood lead levels among children today.  Research indicates that
dust and soil are the most significant pathways for children's lead
exposure, and that LBP is the major important source of household dust
lead.,

The fundamental purpose of the Residential Lead-Based Paint Hazard
Reduction Act of 1992, which is Title X of the Housing and Community
Development Act of 1992 (P.L. 102-550) is prevention, i.e., to find and
mitigate LBP hazards in homes before children are poisoned.  In Sections
1051 and 1052(10), Congress required HUD to conduct research on risk
reduction strategies from household-based lead exposure, and to assess
the effectiveness of lead hazard evaluation activities, respectively. 
The NSLAH was undertaken to provide current information needed for
regulatory and policy decisions and for assessment of the effectiveness
of lead hazard reduction strategies, e.g., lead information disclosure
during housing sale or lease transactions and certification of LBP
professionals.

In 1989-1990, HUD sponsored a national survey of LBP in housing
(referred to as the 1990 LBP Survey).  The primary objective of that
survey was to estimate the prevalence of LBP in housing - not to address
the presence of lead-based hazards in the housing.  Since 1990, there
have been advances in the understanding of the sources and pathways of
lead transport and exposure, advances in the protocols for collecting
samples of paint and dust for lead contamination, and improvements in
the understanding of the susceptibility of children to the effects of
lead exposure in the intervening years.  For example, it is now
understood that lead-contaminated house dust from LBP is most often the
primary lead hazard for children.,,  Also, HUD now recommends, and HUD
and EPA require in their regulations, the use of dust wipe sampling as
opposed to the vacuum sampling employed in the 1990 LBP Survey.  In
addition, the 1990 survey data are dated because the housing stock has
continued to evolve as older houses are renovated, repaired, and/or
demolished.  Thus, it may not serve as an appropriate estimate for
evaluation of current LBP hazard reduction strategies.  Finally, the
1990 LBP Survey excluded certain housing categories such as manufactured
housing, and housing built after 1979 (i.e., after the Consumer Product
Safety Commission’s ban on lead-containing residential paint went into
effect).

The NSLAH has updated HUD's 1990 LBP Survey and will enable an
assessment of progress in making the U.S. housing stock lead-safe. 
Further, it provides current baseline information needed for regulatory
and policy decisions and for assessment of the effectiveness of lead
hazard reduction strategies currently under development.

1.2	Survey Objectives

HUD’s principal lead-related purpose for the NSLAH was to develop a
scientific description of the existing lead levels in dust, soil, and
paint in the Nation’s housing.  In addition, the survey of lead
hazards in homes collected data to:

Estimate the number and percent of homes with dust and soil lead levels
above selected thresholds.

Identify likely sources of lead in dust in housing, e.g., paint and
soil.

Permit future analyses of lead hazard control strategies and costs,
e.g., quantities of deteriorated painted surfaces.

Permit future analyses for regulation, policy, and guidance that
minimize regulatory and program implementation burden.

In order to meet the survey objectives, a nationally-representative
sample of 1,984 housing units (HUs) was drawn from 75 clusters called
primary sampling units (PSUs).  A general three-stage sample design was
utilized to accomplish these goals as efficiently as possible.  A total
of 831 eligible HUs were recruited into the survey.  In each recruited
HU, samples of dust and soil were collected and painted surfaces were
tested.  (See Volume II: Design and Methodology for details on design
and data collection protocols.)

1.2.1	Technical Note on the Survey Weights and Estimation Procedure

Paint lead measurements, dust samples, and general area soil samples
were collected from all 831 homes in the sample.  In contrast, data on
the presence of children’s play areas and play area soil samples were
collected from a nationally representative statistical subsample of 375
homes in the sample.  Consequently, two sets of survey sampling weights
have been developed:  one for the full sample of 831 surveyed homes; and
one for the subsample of 375 homes with play area soil lead data.  The
construction of these weights is described in Volume II.  Both sets of
weights are unbiased; they will both produce unbiased national estimates
to characterize the target population of all 95.7 million occupied
housing units in the U.S. where children are permitted to live. 
However, the play area subsample, being smaller, produces wider
confidence intervals than the full sample.  To minimize the impact of
this fact and produce the most efficient possible estimates, the
following procedure was adopted.  Data from all 831 homes were used to
estimate paint, dust, and non-play-area soil lead hazards; then data
from the 375 home subsample were used to estimate the added number of
homes with only play area soil lead hazards.  Further details on this
estimation procedure are provided in Appendix D.

1.3	Report Organization

The report for the NSLAH consists of two volumes:  Volume I presents the
major lead hazard findings.  Volume II presents the survey design and
methodology.  

There are seven chapters in Volume I, including this introduction. 
Descriptions of each chapter are as follows:

Chapter 2 describes the population surveyed and compares the survey
population to Current Population Survey (CPS) and American Housing
Survey (AHS) populations.

Chapter 3 presents the estimates of the prevalence of significant LBP
hazards in housing, based on the findings presented in Chapters 4, 5,
and 6 for paint, dust, and soil, respectively.  The association between
lead in each of the matrices (paint, dust, soil) is presented, as well
as the prevalence of lead-related occupations and hobbies among housing
residents.

Chapter 4 presents the estimates of the prevalence and amount of LBP
and deteriorated LBP in housing, including paint lead loadings in
housing. Relevant estimates are compared with the findings of the 1990
LBP Survey.

Chapter 5 presents the estimates of the prevalence of lead-contaminated
dust in housing, including the dust lead loadings and the association
between interior dust lead and LBP condition.  Relevant estimates are
compared with the findings of the 1990 LBP Survey.

Chapter 6 presents the estimates of the prevalence of residential soil
lead, including soil lead concentrations and the association between
soil lead and exterior LBP condition.  Relevant estimates are compared
with the findings of the 1990 LBP Survey.

Chapter 7 examines the quality of the data and the resulting quality of
projected national estimates.  The chapter addresses nonresponse rates
and provides a summary of field data collection quality control
activities.  Classification bias due to measurement error is discussed
in Appendix C. 

2.  Surveyed housing population

The National Survey of Lead and Allergens in Housing (NSLAH) population
included the national housing stock of permanently-occupied,
noninstitutional housing units (HUs), including multi-family buildings
and manufactured HUs, i.e., mobile homes and trailers.  Homes built in
all age categories in all 50 states and the District of Columbia were
included.  Homes built before 1978 were included to update and expand
upon the findings of the 1990 LBP Survey.  Homes built in 1978 or after
were included to verify the assumption that newer homes have minimal
lead hazards, based on the 1978 ban of lead-based paint (LBP) for
residential use.  Vacant housing, group quarters, hotels and motels,
military bases, and short-term housing were excluded because of
difficulties gaining cooperation and are consistent with exclusions
under the Lead Safe Housing Rule.  Housing where children were not
permitted to live, e.g., elderly care facilities, were excluded because
the primary interest of the survey was in children’s exposure to lead
and allergens.  However, a home was not excluded simply because a child
was not currently residing in the home at the time of the survey.  With
these exclusions, the eligible national housing stock consisted of
approximately 96 million housing units.

A nationally-representative sample of 1,984 HUs was drawn from 75
clusters called primary sampling units (PSUs).  A total of 831 eligible
HUs were recruited and completed the survey.  Table 2.1 presents the
national estimates for selected characteristics of the survey
population, including geographic region, year of construction, degree of
urbanization, presence of children under age 6 and age 18, tenure,
income, poverty, government support, race, and ethnicity.  All estimates
presented are weighted national estimates as discussed in Volume II. 
Chapter 7 and Appendix C of this volume presents an extensive discussion
of the potential effect of nonresponse bias.

One important measure of the representativeness of the NSLAH is to
examine how the distributions of the housing characteristics,
socioeconomic and demographic factors compare to national distributions.
 National distributions were obtained from the 1997 American Housing
Survey (AHS) and the 1998 and 1999 Current Population Surveys (CPS). 
The weighted percent distribution of the NSLAH sample by race,
ethnicity, income, presence of a child under 18, Census region, year of
construction, single family vs. multi-family, metropolitan status, and
tenure (owned vs. rented) were compared with that of the available AHS
and CPS data.

The 95 percent confidence intervals for the NSLAH sample estimate were
found to contain the AHS or CPS estimate for most of the available
statistics.  Slight differences in estimates were observed as follows:

The 1997 AHS estimate of 35 percent of all homes being in the South is
slightly lower than the survey estimate of 36 percent to 39 percent of
all homes being in the South.

The 1997 AHS estimate of 37 percent of all homes having children under
age 18 is slightly lower than the survey estimate of 38 percent to 39
percent of all homes.

The 1998 CPS estimate of 26 percent of all household incomes falling in
the $0-19,999 range is slightly higher than the survey estimate of 17
percent to 24 percent of all households in this income range.

The 1998 CPS estimate of 85 percent of all households not in poverty is
slightly higher than the survey estimate of 77 percent to 83 percent of
all households not in poverty.

A few items should be noted which affect the comparability of the
estimates and may explain the above observed differences.  The first is
that the target population for the National Survey excludes housing that
excludes children, whereas the AHS and CPS estimates include such
housing.  Second, the cut-offs used for defining urbanization were not
exactly the same for the NSLAH and the CPS.  The NSLAH used a cut-off of
2,100,000 while the CPS used 2,500,000.  Third, for the income and
poverty comparisons, the CPS uses the family size and income to
determine poverty status, whereas the NSLAH has used household size and
income.  The NSLAH estimates of any particular income class are also
deflated due to the 9 percent of respondents whose income level is
unknown.  Similarly, there are 6 percent of respondents whose poverty
status is unknown.  Fourth, race and ethnicity are based on the youngest
household member for the NSLAH, while the AHS bases these data on the
first householder identified over the age of 18. Fifth, the most recent
AHS available is for 1995 and 1997, thus the AHS data are slightly out
of date as compared to the NSLAH. Finally, there is a small amount of
sampling error in both the AHS and CPS estimates, and in the standard
error estimates for all three surveys.  The standard error for both AHS
estimates (percent of homes in the south, percent of homes with children
under age 18) is approximately 0.25%.  The difference between the NSLAH
and AHS estimates is still statistically significant for both variables.
 CPS standard errors are of a comparable magnitude to those for the AHS
(similar sample sizes and designs) and also do not affect the
statistical significance of the results.

Table 2.1	Characteristics of the National Survey Population, with
Comparisons to the American Housing Survey (AHS) and the Current
Population Survey (CPS)

HU Characteristic	National Survey Estimates	HUs in sample	AHS

(1997)	CPS

(1998-99)c

	Estimate (000)	Estimate (%)a	Lower 95% CIb (%)	Upper 95% CI (%)



	Total Housing Unitsd	95,688	100%

	831



Region:







	Northeast	19,290	20%	19%	22%	155	20%

	Midwest	22,083	23%	22%	24%	196	24%

	South	35,474	37%	36%	39%	277	35%

	West	18,841	20%	18%	21%	203	21%

	Construction year:







	1978-1998	29,774	31%	30%	32%	220	30%

	1960-1977	27,874	29%	28%	30%	267	30%

	1940-1959	20,564	21%	20%	23%	186	20%

	Before 1940	17,476	18%	17%	20%	158	20%

	Region by Construction year:







	Northeast	19,290	20%	19%	22%	155	20%

	1978-1998	4,358	5%	3%	6%	30	3%

	1960-1977	3,754	4%	3%	5%	30	5%

	1940-1959	4,261	5%	4%	5%	36	4%

	Before 1940	6,917	7%	6%	8%	59	7%

	Midwest	22,083	23%	22%	24%	196	24%

	1978-1998	4,801	5%	4%	6%	41	6%

	1960-1977	6,283	7%	6%	7%	55	7%

	1940-1959	5,899	6%	5%	7%	47	5%

	Before 1940	5,101	5%	5%	6%	53	6%

	South	35,474	37%	36%	39%	277	35%

	1978-1998	14,447	15%	14%	17%	95	14%

	1960-1977	11,261	12%	11%	12%	96	12%

	1940-1959	6,320	7%	6%	7%	57	6%

	Before 1940	3,445	4%	3%	4%	29	4%

	West	18,841	20%	17%	21%	203	21%

	1978-1998	6,169	6%	5%	8%	54	8%

	1960-1977	6,536	7%	6%	7%	85	7%

	1940-1959	4,124	4%	3%	6%	47	4%

	Before 1940	2,013	2%	1%	3%	17	2%

	Urbanization:





	1999

MSA equal to or above

  2 million population	26,814	28%	24%	32%	276

30%

MSA below 2 million

  Population	45,753	48%	43%	53%	417

47%

Non-MSA	23,121	24%	19%	30%	138

23%



Table 2.1	Characteristics of the Survey Population, with Comparisons to
the American Housing Survey (AHS) and the Current Population Survey
(CPS) (continued)

HU Characteristic	National Survey Estimates	HUs in sample	AHS

(1997)	CPS

(1998-99)3

	Estimate (000)	Estimate (%)	Lower 95% CI (%)	Upper95% CI (%)



	One or more children under age 6:	16,402	17%	15%	19%	184	17%

	Refusal/Don’t Knowe	352



5



One or more children under age 18:	36,994	39%	38%	39%	398	37%

	Refusal/Don’t Know	290



3



Housing Unit Type:







	Single family	82,651	86%	84%	89%	705	88%

	Multi-family	13,037	14%	11%	16%	126	12%

	Tenure:





	1999

Owner-occupied	66,232	69%	65%	73%	539

67%

Renter-occupied	29,074	30%	27%	34%	289

33%

Refusal/Don’t Know	381



3



Household Income

  ($30,000):





	1998

Less than $30,000/year	33,830	35%	30%	41%	309

40%

Equal to or more than

  $30,000/year	56,111	59%	54%	63%	482

60%

Refusal/Don’t Know	5,747



40



Household Income

  ($20,000):





	1998

$0-19,999/year	19,359	20%	17%	24%	189

26%

$20-39,999/year	25,855	27%	23%	31%	228

27%

$40-59,999/year	19,316	20%	16%	25%	152

19%

Equal to or more than

  $60,000/year	22,890	24%	20%	28%	203

28%

Refusal/Don’t Know	8,268



59



Government Support:







	Government support	4,809	5%	3%	7%	54



No Government support	86,070	90%	88%	92%	733



Refusal/Don’t Know	4,809	5%

	44



Poverty:





	1998

In poverty	13,221	14%	11%	16%	137

15%

Not in poverty	76,336	80%	77%	82%	651

85%

Refusal/Don’t Know	6,130	6%

	43





Table 2.1	Characteristics of the Survey Population, with Comparisons to
the American Housing Survey (AHS) and the Current Population Survey
(CPS) (continued)

HU Characteristic	National Survey Estimates	HUs in sample	AHS

(1997)	CPS

(1998-99)3

	Estimate (000)	Estimate (%)	Lower 95% CI (%)	Upper95% CI (%)



	Poverty by Urbanization:







	MSA equal to or above 2 million population	26,814	28%	24%	32%	276



In poverty	2,962	3%	2%	4%	35



Not in poverty	22,005	23%	19%	27%	226



Refusal/Don’t Know	1,847	2%

	15



MSA below 2 million population	45,753	48%	43%	53%	417



In poverty	6,996	7%	5%	9%	75



Not in poverty	35,786	37%	34%	41%	323



Refusal/Don’t Know	2,971	3%

	19



Non-MSA	23,121	24%	19%	30%	138



In poverty	3,264	3%	2%	5%	27



Not in poverty	18,544	19%	14%	25%	102



Refusal/Don’t Know	1,313	1%

	9



Race:







	White	77,005	80%	78%	83%	622	83%

	African American	10,365	11%	9%	13%	116	12%

	Otherf	6,571	7%	5%	8%	77	6%

	Refusal/Don’t Know	1,746	2%

	16



Ethnicity:







	Hispanic/Latino	7,434	8%	6%	10%	86	9%

	Not Hispanic/Latino	87,008	91%	88%	93%	736	91%

	Refusal/Don’t Know	1,246	1%

	9



a	All percentages are calculated with total housing units (95,688) as
the denominator.  Percentages may not total 100% due to rounding.

b	CI = 95% confidence interval for the estimated number or percent.

c	Current Population Survey (CPS) data was taken from either 1998 or
1999 CPS, as indicated by the boldface year at the top of each section
of the column.

d	“Housing units” include permanently occupied, noninstitutional
housing units in which children are permitted to live.

e	Refusals and “don’t know” responses by survey respondents.

f	“Other” race includes Asian, American Indian or Alaskan Native,
Native Hawaiian or other Pacific Islander, and more than one race.



3.  LEAD-BASED PAINT (LBP) HAZARDS IN HOUSING

Chapter 3 presents the estimates of the prevalence of lead-based paint
(LBP) hazards in housing, based on the findings presented in Chapters 4,
5, and 6.  The associations between lead in each of the matrices (paint,
dust, soil) are presented.  In addition, the prevalence of lead-related
behaviors, occupations, and hobbies among housing residents is
presented.  No comparison is made with the dust lead findings of the
1990 LBP Survey because the concept of hazard from the earlier survey is
not comparable to the definitions in use today.  The effect of
measurement error on the estimates is discussed in Appendix C.

3.1	Definition of Significant Lead-Based Paint (LBP) Hazards

Under Title X, a LBP hazard is defined as “any condition that causes
exposure to lead from lead-contaminated dust; bare, lead contaminated
soil; LBP that is deteriorated; or LBP present on accessible surfaces,
friction surfaces, or impact surfaces.”  Because the data collection
protocol of the NSLAH was not designed as a conventional risk assessment
due to the statistical and operational considerations discussed below,
this report defines significant lead-based paint (LBP) hazards based on 
the HUD Lead Safe Housing Rule (24 CFR 35.1320) and EPA Lead Hazards
Identification Rule (40 CFR 745.65).  If any of the following situations
exist at any location in a home, a significant LBP hazard exists in the
home under this definition:

Deteriorated LBP - LBP with deterioration larger than the de minimis
levels per Section 35.1350(d) of the Lead Safe Housing rule -
deterioration of more than 20 square feet (exterior) or 2 square feet
(interior) of LBP on large surface area components (walls, doors), or
damage to more than 10% of the total surface area of interior small
surface area components (window sills, baseboards, trim).  LBP is
defined as any paint or other surface coating (e.g., varnish, lacquer,
or wallpaper over paint) that contains lead equal to or greater than 1.0
mg/cm2); or

Lead-contaminated dust - Dust on floors with greater than or equal to 40
µg/ft2 lead, dust on window sills with greater than or equal to 250
µg/ft2 lead; or

Lead-contaminated bare soil – More than 9 square feet of bare soil
with a lead concentration greater than or equal to 1,200 ppm, or 400 ppm
for bare soil in an area frequented by a child under the age of 6 years.

It is important to note the distinctions between the significant LBP
hazards defined here, and the LBP hazards as defined in the EPA Lead
Hazards Identification Rule, which does not specify de minimis levels
for deteriorated LBP and non-play area bare soils.  In addition, the
significant LBP hazards for dust and soil as used in this report,
although applying the same numerical thresholds of the EPA Lead Hazards
Identification Rule, are based on the maximum dust lead and soil lead
measurements, while the EPA thresholds are based on the average dust
lead and soil lead measurements.

The NSLAH used a statistically-based sampling strategy in order to
obtain statistically valid national estimates. The NSLAH took fewer
measurements in a home than a full risk assessment because of the
statistically-based strategy of maximizing the number of housing units
tested (to yield the most representative national prevalence estimates
with the resources available), and HUD's intention to minimize the
burden on occupants of the housing units being tested. The subset of
rooms and components within each home was randomly selected and
evaluated in the NSLAH so that all work - including allergen sampling -
could be completed in a 2-to-3 hour period in each home. For the same
reasons, the survey did not assess chewable surfaces, friction surfaces,
or impact surfaces as required for conventional risk assessments. In
summary, the statistical nature, operational constraints, and objectives
of the NSLAH  required  defining significant LBP hazards in the manner
shown above. 

Under this definition, a home is described as having significant LBP
hazards somewhere in the home if any significant LBP hazard exists at
any location in the home. All tables and figures on housing units with
significant LBP hazards in this report use this definition.

3.2	Prevalence of Significant Lead-Based Paint Hazards in Housing

An estimated 24 million ((3 million) or 25 percent ((3%) of housing
units in the United States have significant LBP hazards. Figure 3.1
presents the number of housing units with significant lead-based paint
hazards by selected characteristics, including the presence of a
resident child under six years of age, housing unit age, measures of
household income, race, and ethnicity.  Table 3.1 presents the number
and percentage of housing units with significant lead-based paint
hazards by a much wider range of housing unit and household
characteristics. 

Homes in Northeastern and Midwestern states are more likely to have
significant LBP hazards than homes in Southern or Western states.  An
estimated 40 percent ((10%) of homes in the Northeast have significant
LBP hazards, while the estimates are 17 percent ((4%) and 15 ((6%)
percent for homes in the South and West, respectively.  Older homes are
more likely to have significant LBP hazards than newer homes.  An
estimated 8 percent ((4%) of homes built between 1960 and 1977 have
significant LBP hazards, but the percentage increases to 43 percent
((10%) for homes built between 1940 and 1959, and to 68 percent ((11%)
for homes built before 1940.  A similar pattern of results was found for
homes with children under age 6 categorized by age of construction.

Figure 3.1	 Prevalence of Housing Units (HUs) with Significant LBP
Hazards

 Table 3.1	Prevalence of Housing Units with Significant Lead-Based
Paint (LBP) Hazards, by 

Selected Characteristics

HUD Lead Safe Housing Rule:  Significant LBP Hazardsa

Characteristic	All HUs

(000)b	No. of HUs with Significant LBP Hazards (000)	Percent of HUs with
Significant LBP Hazards (%)c	HUs in Sample



Estimate	Lower

95% CId	Upper

95% CI	Estimate	Lower

95% CI	Upper

95% CI

	Total Occupied HUs 	95,688	24,026	21,307	26,746	25%	22%	28%	831

Region:









Northeast	19,290	7,679	5,748	9,611	40%	30%	50%	155

Midwest	22,083	7,250	6,402	8,097	33%	29%	37%	196

South	35,474	6,191	4,964	7,419	17%	14%	21%	277

West	18,841	2,906	1,856	3,956	15%	10%	21%	203

Construction Year:









1978-1998	29,774	1,042	169	1,915	3%	1%	6%	220

1960-1977	27,874	2,340	1,445	3,235	8%	5%	12%	267

1940-1959	20,564	8,826	6,720	10,933	43%	33%	53%	186

Before 1940	17,476	11,818	10,045	13,591	68%	57%	78%	158

Region by Construction Year:









Northeast









HUs built 1978-1998	4,358	76	0	225	2%	0%	5%	30

HUs built 1960-1977	3,794	1,478	348	2,609	39%	9%	69%	31

HUs built 1940-1959	4,221	3,089	2,179	3,999	73%	52%	95%	35

HUs built before 1940	6,917	5,957	5,187	6,728	86%	75%	97%	59

Midwest









HUs built 1978-1998	4,801	533	0	1,134	11%	0%	24%	41

HUs built 1960-1977	6,283	1,771	872	2,670	28%	14%	42%	55

HUs built 1940-1959	5,899	4,785	4,011	5,559	81%	68%	94%	47

HUs built before 1940	5,101	4,658	3,888	5,429	91%	76%	100%	53

South









HUs built 1978-1998	14,447	1,197	0	2,436	8%	0%	17%	95

HUs built 1960-1977	11,261	1,914	1,216	2,612	17%	11%	23%	96

HUs built 1940-1959	6,320	3,431	2,329	4,532	54%	37%	72%	57

HUs built before 1940	3,445	3,065	2,676	3,453	89%	78%	100%	29

West









HUs built 1978-1998	6,169	225	0	473	4%	0%	8%	54

HUs built 1960-1977	6,536	1,414	816	2,011	22%	12%	31%	85

HUs built 1940-1959	4,124	2,866	1,715	4,017	69%	42%	97%	47

HUs built before 1940	2,013	1,437	376	2,498	71%	19%	100%	17

Urbanization









MSA >= 2 million population	26,814	6,793	4,978	8,609	25%	19%	32%	276

MSA =< 2 million population	45,753	10,232	8,171	12,293	22%	18%	27%	417

Non-MSA	23,121	7,001	3,848	10,153	30%	17%	44%	138

Table 3.1	Prevalence of Housing Units with Significant Lead-Based Paint
(LBP) Hazards, by Selected Characteristics  (continued)

HUD Lead Safe Housing Rule:  Significant LBP Hazards1

Characteristic	All HUs

(000)2	No. of HUs with Significant LBP Hazards (000)	Percent of HUs with
Significant LBP Hazards (%)3	HUs in Sample



Estimate	Lower

95% CI4	Upper

95% CI	Estimate	Lower

95% CI	Upper

95% CI

	One or More Children Under Age 6:









All HU ages	16,402	4,155	2,948	5,363	25%	18%	33%	184

HUs built 1978-1998	5,847	<58g	-	-	<1g	-	-	56

HUs built 1960-1977	5,098	469	0	940	9%	0%	18%	61

HUs built 1940-1959	3,055	1,732	1,088	2,375	57%	36%	78%	40

HUs built before 1940	2,401	1,955	1,190	2,720	81%	50%	113%h	27

Housing Unit Type:









Single family	82,651	21,584	18,974	24,194	26%	23%	29%	705

Multi-family	13,037	2,442	1,208	3,676	19%	9%	28%	126

Tenure:









Owner-occupied	66,232	15,305	13,191	17,419	23%	20%	26%	539

Renter-occupied	29,074	8,721	6,583	10,859	30%	23%	37%	289

Refusal/Don’t Knowe	381





	3

Household Income:









Less than $30,000/year	33,830	12,007	9,336	14,679	35%	28%	43%	309

Equal to or more than

  $30,000/year	56,111	10,464	8,250	12,678	19%	15%	23%	482

Refusal/Don’t Know	5,747





	40

One or More Children Under Age 6:









All Income Categories	16,402	4,155	2,948	5,363	25%	18%	33%	184

Less than $30,000/year	4,791	1,201	600	1,801	25%	13%	38%	61

Equal to or more than

  $30,000/year	11,236	2,860	1,763	3,957	25%	16%	35%	117

Refusal/Don’t Know	375





	6

Government Support:









Government support	4,809	805	275	1,335	17%	6%	28%	54

No government support	86,070	22,198	19,252	25,144	26%	22%	29%	733

Refusal/Don’t Know	4,809





	44

Poverty:









In Poverty	13,221	4,976	3,458	6,494	38%	26%	49%	137

Not in Poverty	76,336	16,576	13,598	19,555	22%	18%	26%	651

Refusal/Don’t Know	6,130





	43

Poverty by Urbanization:









MSA equal or above 2

  Million population









In poverty	2,962	1,205	735	1,674	41%	25%	57%	35

Not in poverty	22,005	7,758	5,957	9,559	35%	27%	43%	226

MSA below 2 million

  Population









In poverty	6,996	3,795	2,248	5,341	54%	32%	76%	75

Not in poverty	35,786	12,455	9,722	15,188	35%	27%	42%	323



Table 3.1	Prevalence of Housing Units with Significant Lead-Based Paint
(LBP) Hazards, by Selected Characteristics  (continued)

HUD Lead Safe Housing Rule:  Significant LBP Hazards1

Characteristic	All HUs

(000)2	No. of HUs with Significant LBP Hazards (000)	Percent of HUs with
Significant LBP Hazards (%)3	HUs in Sample



Estimate	Lower

95% CI4	Upper

95% CI	Estimate	Lower

95% CI	Upper

95% CI

	Non-MSA









In poverty	3,264	1,362	310	2,414	42%	9%	74%	27

Not in poverty	18,544	8,684	5,071	12,297	47%	27%	66%	102

Refusal/Don’t Know if in

  Poverty	6,131





	43

Race:









White	77,005	19,089	16,475	21,703	25%	21%	28%	622

African American	10,365	2,969	1,807	4,131	29%	17%	40%	116

Otherf	6,571	1,496	672	2,321	23%	10%	35%	77

Refusal/Don’t Know	1,746





	16

Ethnicity:









Hispanic/Latino	7,434	2,399	1,235	3,564	32%	17%	48%	86

Not Hispanic/Latino	87,008	21,196	18,674	23,719	24%	21%	27%	736

Refusal/Don’t Know	1,246





	9

a Significant LBP hazard as defined in text and HUD Lead Safe Housing
Rule.

b “Housing units” include permanently occupied, noninstitutional
housing units in which children are permitted to live.

c All percentages are calculated with the “All HUs” column in each
row used as the denominator.

d CI = 95% confidence interval for the estimated number or percent.

e Refusals and “don’t know” responses by survey respondents.

f “Other” race includes Asian, American Indian or Alaskan Native,
Native Hawaiian or other Pacific Islander, and more than one race.

g No 1978-1988 housing units with one or more children < 6 years old in
this sample have lead-based paint hazards.

h Upper 95% CI value > 100% reflects uncertainty in number of housing
units in first data column.

More homes with lower income occupants have significant LBP hazards than
homes where occupants have higher incomes.  An estimated 35 percent
((8%) of households with less than $30,000/year income have significant
LBP hazards, compared with 19 percent ((4%) of households in the
$30,000/year or above income level.

Government-supplied housing may have fewer lead-based paint hazards than
housing without Government support.  An estimated 17 percent of
Government-supported housing had significant lead-based paint hazards,
compared to 26 percent of housing without Government support.

Table 3.2 presents the number of homes with significant LBP hazards by
location in the building–either interior or exterior, or both.

Table 3.2	Prevalence of Significant Lead-Based Paint (LBP) Hazards by
Location in the Building

HUD Lead Safe Housing Rule:  Significant LBP Hazards

LBP Hazard Location	Number of HUsa (000)	Percent of HUsb	HUs in Sample 

	Estimate	Lower

95% CIc	Upper

95% CI	Percent	Lower

95% CI	Upper

95% CI

	Interior only	8,823	6,439	11,207	9%	7%	12%	63

Both Interior and Exterior	8,869	6,634	11,104	9%	7%	12%	99

Exterior only	6,334	4,741	7,928	7%	5%	8%	48

Anywhere	24,026	21,307	26,746	25%	22%	28%	210

No Significant LBP Hazard	71,661	68,498	74,825	75%	72%	78%	621

Total HUs	95,688

	100%

	831

a “Housing units” include permanently occupied, noninstitutional
housing units in which children are permitted to live.

b All percentages are calculated with total housing units (95,688) as
the denominator.  Percentages may not total 100% due to rounding.

c CI = 95% confidence interval for the estimated number or percent.

Table 3.3 presents data for the presence of significant LBP hazards in
homes by type of hazard, for all homes in the National Survey’s target
population and for homes with one or more children under the age of 6
years.  The percentages for each “All HUs” row of Table 3.3 show the
percent of all HUs with the component of significant LBP hazard, while
the percentages in each “HUs w/Child Under 6” row of Table 3.3 show
the percent of all HUs with a child under age 6 with that component of
significant LBP hazard. 

Table 3.4 breaks out the data in Table 3.3 for all HU’s by
construction year.  All three components of significant lead-based paint
hazards show the same pattern of substantially greater prevalence of
hazards among older housing units, especially pre-1940 homes.  For all
housing unit ages, soil lead hazards have the lowest prevalence. 

Table 3.3	Prevalence of Significant Lead-Based Paint (LBP) Hazards in
Housing Units with a Child Under 6 Years of Age by Type of Hazard

HUD Lead Safe Housing Rule: Significant LBP Hazards

Type of Hazard	Number of HUsa (000)	Percent of HUsb (%)

	Estimate	Lower

95% CIc	Upper

95% CI	Estimate	Lower

95% CI	Upper

95% CI

Significantly Deteriorated Lead Based Paint







All HUs	13,634	10,928	16,341	14%	11%	17%

HUs w/ Child Under 6	2,519	1,715	3,324	15%	10%	20%

Interior Lead Dust







All HUs	15,468	12,982	17,954	16%	14%	19%

HUs w/ Child Under 6	2,634	1,587	3,681	16%	10%	22%

Lead Contaminated Soil







All HUs	6,460	3,122	9,799	7%	3%	10%

HUs w/ Child Under 6	1,511	0	3,108	9%	0%	19%

Any LBP Hazard







All HUs	24,026	21,306	26,746	25%	22%	28%

HUs w/ Child Under 6	4,155	2,948	5,363	25%	18%	33%

a “Housing units” include permanently occupied, noninstitutional
housing units in which children are permitted to live.

b Percentages are calculated with total housing units (95,688) or with
housing units with a child under age 6 (16,402) as the denominator, as
applicable.

c CI = 95% confidence interval for the estimated number or percent.

Table 3.4 	Prevalence of Significant Lead-Based Paint (LBP) Hazards in
Housing Units by Type of Hazard and Housing Unit Age

HUD Lead Safe Housing Rule: Significant LBP Hazards

Type of Hazard	Number of HUsa (000)	Percent of HUsb (%)

	Estimate	Lower

95% CIc	Upper

95% CI	Estimate	Lower

95% CI	Upper

95% CI

Significantly Deteriorated Lead Based Paint







Built 1978-1998	83	0	240	0%	0%	1%

Built 1960-1977	610	91	1,128	2%	0%	4%

Built 1940-1959	5,190	3,367	7,013	25%	16%	34%

Built Before 1940	7,752	6,029	9,475	44%	34%	54%

Interior Lead Dust







Built 1978-1998	959	98	1,821	3%	0%	6%

Built 1960-1977	1,943	1,173	2,714	7%	4%	10%

Built 1940-1959	4,665	3,181	6,150	23%	15%	30%

Built Before 1940	7,735	5,982	9,489	44%	34%	54%

Lead Contaminated Soil







Built 1978-1998	0	0	0	0%	0%	0%

Built 1960-1977	130	0	263	0%	0%	1%

Built 1940-1959	2,562	0	3,830	12%	0%	19%

Built Before 1940	3,867	254	4,741	22%	1%	27%

Any LBP Hazard







Built 1978-1998	1,042	169	1,914	3%	1%	6%

Built 1960-1977	2,340	1,445	3,235	8%	5%	12%

Built 1940-1959	8,826	6,364	10,932	43%	31%	53%

Built Before 1940	11,818	9,688	13,591	68%	55%	78%

a “Housing units” include permanently occupied, noninstitutional
housing units in which children are permitted to live.

b Percentages are calculated with housing units built in that time
period as the denominator.

c CI = 95% confidence interval for the estimated number or percent.

3.3	Prevalence of Lead-Related Occupations or Hobbies

Table 3.5 presents the number and percent of all U.S. households in
which at least one occupant engages in a lead-related occupation or
hobby.  Data for home cleanliness and clutter categories are also
presented.  Table 3.6 presents the same data for homes with significant
LBP hazards.  As described below, all of these estimates are likely to
overstate the number of households that may have contributions to lead
in dust from these behaviors.

An estimated 24 percent ((3%) of households report that at least one
occupant engages in a listed lead-related occupation, e.g., construction
or renovation work, lead industry work, automotive repair or radiator
work, or firing range work.  This may be elevated above the actual
estimate of people who may bring lead home due to the fact that many
people in the construction business only work with new construction or
in projects which that do not involve disturbing LBP, dust-lead hazards
or soi6l-lead hazards.  Similarly, a person working in a lead-related
industry may have an administrative position and have no lead exposure
at all.

An estimated 41 percent ((5%) report that at least one occupant engages
in a potentially lead-related hobby at home, e.g., making bullets or
sinkers, furniture or home renovation, stained glass, pottery, or making
jewelry.  Among homes with significant LBP hazards, 31 percent ((8%)
report that at least one occupant engages in a potentially lead-related
hobby.  These may be high estimates of the number of homes where lead
dust is actually generated since some of these people may generally or
always use lead-free materials in their hobbies.  

Home cleanliness has been associated with lead dust levels.  About half
of all homes were found to be clean (59%) and organized (43%), using the
criteria in this survey.  Homes with significant LBP hazards in their
interiors were generally somewhat less clean and organized; the
percentage of clean homes is 36%, 23% lower than the general population,
and the percentage of organized homes is 28%, 15% lower than the general
population.

Table 3.5	Prevalence of Housing Units with Selected Lead-Related
Characteristics

Lead Related Behavior	Number of HUs (000)a	Percent of HUs (%)b	HUs in
Sample

	Estimatec	Lower

95% CId	Upper

95% CI	Estimate	Lower

95% CI	Upper

95% CI

	Lead Related Occupation	22,673	19,732	25,615	24%	21%	27%	203

Lead Related Hobby	39,281	35,020	43,543	41%	36%	46%	347

Cleanliness







	House Appears Clean	56,058	51,887	60,228	59%	54%	63%	462

Some Evidence of Cleaning	25,347	21,417	29,277	26%	22%	31%	237

No Evidence of Cleaning	9,646	7,577	11,714	10%	8%	12%	86

Clutter







	Clutter Organized	41,158	37,650	44,666	43%	40%	46%	347

Average Amount of Clutter	38,601	35,663	41,539	40%	37%	43%	336

No Organization	11,045	8,859	13,231	12%	9%	14%	100

Total HUs	95,688





831

a “Housing units” include permanently occupied, noninstitutional
housing units in which children are permitted to live.

b All percentages are calculated with total housing units (95,688) as
the denominator.  Percentages may not total 100% due to rounding.

c Estimates are based on the full weighted sample.

d CI = 95% confidence interval for the estimated number or percent.

Table 3.6	Prevalence of Selected Lead-Related Characteristics in Homes
with Significant Interior LBP Hazards

Lead Related Behavior	Number of HUs (000)a	Percent of HUs (%)b	HUs in
Sample

	Estimate	Lower

95% CIc	Upper

95% CI	Estimate	Lower

95% CI	Upper

95% CI

	Lead Related Occupation	4,990	3,195	6,785	28%	18%	38%	45

Lead Related Hobby	5,485	4,043	6,926	31%	23%	39%	57

Cleanliness







	House Appears Clean	6,360	4,030	8,691	36%	23%	49%	60

Some Evidence of Cleaning	6,757	5,221	8,294	38%	30%	47%	63

No Evidence of Cleaning	3,258	1,646	4,870	18%	9%	28%	28

Clutter







	Clutter Organized	5,041	3,052	7,029	28%	17%	40%	44

Average Amount of Clutter	7,899	6,386	9,412	45%	36%	53%	77

No Organization	3,436	1,911	4,961	19%	11%	28%	30

Total HUs	17,692





162

a “Housing units” include permanently occupied, noninstitutional
housing units in which children are permitted to live.

b All percentages are calculated with total housing units with
significant interior LBP hazard (18,827) as the denominator. 
Percentages may not total 100% due to rounding.

c CI = 95% confidence interval for the estimated number or percent.

4.  LEAD-BASED PAINT (LBP) IN HOUSING

Chapter 4 presents estimates of the prevalence, location, and amount of
lead-based paint (LBP) and deteriorated LBP in housing, including paint
lead loadings in housing.  Relevant estimates are compared with the
findings of the 1990 LBP Survey.

Under the HUD Lead Safe Housing Rule (24 CFR 35 Subparts B-R), LBP is
defined as any paint or other surface coating (e.g., varnish, lacquer,
or wall-paper over paint) that contains lead equal to or greater than
1.0 mg/cm2.  The estimates for deteriorated LBP and significantly
deteriorated LBP are presented in Section 4.2.  Under the Lead Safe
Housing Rule, LBP is considered to be deteriorated if there is any
deterioration.  It is considered to be significantly deteriorated if the
deterioration exceeds the de minimis thresholds given in the definition
of a significant LBP hazard presented in Chapter 3.

4.1	Prevalence of Lead-Based Paint

An estimated 38 million ((3 million) or 40 percent ((4%) of housing
units (HUs) in the United States have LBP on either the interior or
exterior painted surfaces, or both.  Figures 4.1 and 4.2 present the
percentage of housing units with LBP by selected characteristics,
including presence of children under age 6, geographic region, housing
unit type, tenure, race, and year of construction.  Table 4.1 presents
the number and percentage of housing units with LBP by a much wider
range of housing unit and household characteristics.

The data suggest that homes in northeastern and midwestern states are
more likely to have LBP than homes in southern or western states.  An
estimated 55 percent and 53 percent of homes in the Northeast and
Midwest have LBP, while the estimates are 27 percent and 32 percent for
homes in the South and West, respectively.  This finding can be
explained by the fact that the housing stock in the Northeast and
Midwest is older.

Figure 4.1	Prevalence of LBP by Selected Housing Unit Characteristics

Figure 4.2	Prevalence of HUs with LBP by Construction Year & Presence of
Children <6

As expected, older homes are more likely to have LBP than newer homes. 
An estimated 24 percent (+6%) of homes built between 1960 and 1977 have
LBP, but the percentage increases to 69 percent (+9%) for homes built
between 1940 and 1959, and to 87 percent (+5%) for homes built before
1940.  Table 4.1 indicates that 5 to 8 percent of homes built after 1977
have LBP.  This estimate is likely to be somewhat high, for two reasons.
 First, housing unit age is based on the residents’ reports.  Some
residents, especially renters, may have reported their homes as being
newer than they actually are. There were a few homes in the survey in
which an examination of all of the data collected from a home presented
a picture that suggested that the home may be older than reported by the
respondent.  Even so, in no case was the respondent’s reported age
overridden in the analyses.  Second, most of the homes in the sample
built after 1977 that had LBP had only one or two components measured
with LBP.  In approximately one-third of these homes, the substrate of
the one painted component in the home may have interfered with the
reading (these included painted metal, ceramic, or stone substrates
which may contain lead, but paint was not scraped to measure bare
substrate for this purpose).

Table 4.1	Prevalence of Lead-Based Paint (LBP) by Selected Housing Unit
(HU)

Characteristics

HU Characteristic	All HUs

(000)	Number of HUs with LBP (000)	Percent of HUs with LBP (%)a	HUs in
Sample



Estimate	Lower

95% CIb	Upper

95% CI	Estimate	Lower

95% CI	Upper

95% CI

	Total Housing Unitsc	95,688	37,897	34,521	41,272	40%	36%	43%	831

Region:









Northeast	19,290	10,600	8,306	12,895	55%	46%	64%	155

Midwest	22,083	11,748	10,546	12,950	53%	48%	59%	196

South	35,474	9,607	7,762	11,451	27%	22%	32%	277

West	18,841	5,942	4,747	7,137	32%	25%	38%	203

Construction Year:









1989-1998	10,378	514	-	1,444	5%	0%	14%	61

1978-1988	19,397	1,517	549	2,485	8%	3%	13%	159

1960-1977	27,874	6,577	4,875	8,280	24%	18%	30%	267

1940-1959	20,564	14,171	12,203	16,139	69%	60%	77%	186

Before 1940	17,476	15,117	13,532	16,702	87%	82%	91%	158

Region by Construction Year:









Northeast









HUs built 1978-1998	4,358	76	0	225	2%	0%	5%	30

HUs built 1960-1977	3,794	1,478	348	2,609	39%	9%	69%	31

HUs built 1940-1959	4,221	3,089	2,179	3,999	73%	52%	95%	35

HUs built before 1940	6,917	5,957	5,187	6,728	86%	75%	97%	59

Midwest









HUs built 1978-1998	4,801	533	0	1,134	11%	0%	24%	41

HUs built 1960-1977	6,283	1,771	872	2,670	28%	14%	42%	55

HUs built 1940-1959	5,899	4,785	4,011	5,559	81%	68%	94%	47

HUs built before 1940	5,101	4,658	3,888	5,429	91%	76%	100%	53

South









HUs built 1978-1998	14,447	1,197	0	2,436	8%	0%	17%	95

HUs built 1960-1977	11,261	1,914	1,216	2,612	17%	11%	23%	96

HUs built 1940-1959	6,320	3,431	2,329	4,532	54%	37%	72%	57

HUs built before 1940	3,445	3,065	2,676	3,453	89%	78%	100%	29

West









HUs built 1978-1998	6,169	225	0	473	4%	0%	8%	54

HUs built 1960-1977	6,536	1,414	816	2,011	22%	12%	31%	85

HUs built 1940-1959	4,124	2,866	1,715	4,017	69%	42%	97%	47

HUs built before 1940	2,013	1,437	376	2,498	71%	19%	100%	17

Urbanization:









MSA equal or above 2

  million population	26,814	9,681	7,550	11,812	36%	30%	42%	276

MSA below 2 million

  Population	45,753	17,390	14,026	20,754	38%	32%	44%	417

Non-MSA	23,121	10,826	7,458	14,193	47%	35%	59%	138



Table 4.1	Prevalence of Lead-Based Paint (LBP) by Selected Housing Unit
(HU)

Characteristics (continued)

HU Characteristic	All HUs

(000)	Number of HUs with LBP (000)	Percent of HUs with LBP (%)a	HUs in
Sample



Estimate	Lower

95% CIb	Upper

95% CI	Estimate	Lower

95% CI	Upper

95% CI

	One or More Children Under Age 6:









All HU Ages	16,402	5,328	4,048	6,609 	32%	26%	39%	184

HUs built 1978-1998	5,847	202 	0	436 	3%	0%	7%	56

HUs built 1960-1977	5,098 	876 	416 	1,337 	17%	8%	26%	61

HUs built 1940-1959	3,055 	1,997 	1,341 	2,654 	65%	44%	87%	40

HUs built before 1940	2,401 	2,253 	1,426 	3,079 	94%	59%	100%	27

Housing Unit Type:









Single family	82,651	34,081	30,874	37,289	41%	37%	45%	705

Multi-family	13,037	3,815	2,470	5,160	29%	20%	39%	126

Tenure:









Owner-occupied	66,232	25,172	22,400	27,943	38%	35%	41%	539

Renter-occupied	29,074	12,409	9,538	15,281	43%	35%	50%	289

Refusal/Don’t Knowd	381





	3

Household Income:









Less than $30,000/year	33,830	15,007	11,604	18,411	44%	37%	52%	309

Equal to or more than

  $30,000/year	56,111	20,815	17,745	23,885	37%	32%	42%	482

Refusal/Don’t Know	5,747





	40

One or More Children Under Age 6:









All Income Categories	16,402	5,328	4,048	6,609	32%	26%	39%	184

	Less than $30,000/year	4,791	1,375	784	1,965	29%	16%	41%	61

	Equal to or more than

	  $30,000/year	11,236	3,820	2,579	5,061	34%	23%	45%	117

	Refusal/Don’t Know	375





	6

Government Support:









Government support	4,809	1,741	678	2,805	36%	16%	56%	54

No government support	86,070	33,871	30,681	37,062	39%	36%	43%	733

Refusal/Don’t Know	4,809





	44

Poverty by Urbanization:









MSA equal or above 2

  Million population









In poverty	2,962	1,205	735	1,674	41%	25%	57%	35

Not in poverty	22,005	7,758	5,957	9,559	35%	27%	43%	226

MSA below 2 million

  Population









In poverty	6,996	3,795	2,248	5,341	54%	32%	76%	75

Not in poverty	35,786	12,455	9,722	15,188	35%	27%	42%	323

Non-MSA









In poverty	3,264	1,362	310	2,414	42%	9%	74%	27

Not in poverty	18,544	8,684	5,071	12,297	47%	27%	66%	102

Refusal/Don’t Know if in

  Poverty	6,131





	43

Table 4.1	Prevalence of Lead-Based Paint (LBP) by Selected Housing Unit
(HU)

Characteristics (continued)

HU Characteristic	All HUs

(000)	Number of HUs with LBP (000)	Percent of HUs with LBP (%)a	HUs in
Sample



Estimate	Lower

95% CIb	Upper

95% CI	Estimate	Lower

95% CI	Upper

95% CI

	Race:









White	77,005	30945	28,037	33853	40%	37%	44%	622

African American	10,365	4,228	2,767	5,689	41%	30%	52%	116

Othere	6,571	1,913	1,015	2,811	29%	17%	41%	77

Unknown	1,746





	16

Ethnicity:









Hispanic/Latino	7,434	3,329	2,044	4,614	45%	31%	59%	86

Not Hispanic/Latino	87,008	33,830	30,436	37,223	39%	35%	42%	736

Refusal/Don’t Know	1,246





	9

a All percentages are calculated with the “all HUs” on the left most
column of each row as the denominator.

b CI = 95% confidence interval for the estimated number or percent.

c “Housing units” include permanently occupied, noninstitutional
housing units in which children are permitted to live.

d Refusals and “don’t know” responses by survey respondents.

e “Other” race includes Asian, American Indian or Alaskan Native,
Native Hawaiian or other Pacific Islander, and more than one race.

An estimated 5.3 million ((1.3 million) or 32 percent (( 6%) of homes
with children under the age of 6 years have lead-based paint, although
children are eligible to live in all of the estimated 38 million homes
with lead-based paint.  Homes with children under age 6 showed the same
relationship between HU age and the presence of LBP as do all homes.

The differences among LBP prevalence by urbanization, single family
versus multi-family housing, occupant status, household income, race,
ethnicity, and poverty crossed with urbanization do not appear to be
significant in that the confidence intervals overlap.  Likewise, there
were no differences in LBP prevalence when one or more children under
age 6 was crossed by construction year.

Figure 4.3 and Table 4.2 present the number of homes with LBP by
location in the building – either interior or exterior, or both. 
About one-half of homes with lead-based paint have it on both interior
and exterior surfaces (21% of all homes, or 53% of homes with LBP
anywhere in the building).

Table 4.2	Prevalence of Lead-Based Paint (LBP) by Location in the
Building

LBP Location	Number of HUsa with LBP (000)	Percent of HUs with LBP (%)b
HUs in Sample

	Estimate	Lower

95% CIc	Upper

95% CI	Estimate	Lower

95% CI	Upper

95% CI

	Interior Only	8,609	6,102	11,116	9%	6%	12%	77

Both Interior and Exterior	20,260	17,961	22,558	21%	19%	24%	181

Exterior Only	9,028	6,535 	11,521	9%	7%	12%	80

Subtotal – LBP anywhere

  in Building	37,897	34,521	41,272	40%	36%	43%	338

No LBP in Building	57,791 	54,624	60,959	60%	57%	64%	493

Total HUs	95,688 

	100%

	831

a	“Housing units” include permanently occupied, noninstitutional
housing units in which children are permitted to live.

b	All percentages are calculated with total housing units (95,688) as
the denominator.  Percentages may not total 100% due to rounding.

c	CI = 95% confidence interval for the estimated number or percent.

Figure 4.3	Prevalence of Lead Based Paint (LBP) by Location in the
Building

4.2	Prevalence of Deteriorated Lead-Based Paint

Although there are many homes with LBP, the condition of the paint is
important in determining whether a hazard exists.  Except during
renovations, maintenance, or other activities that could disturb it,
intact lead-based paint is believed to pose little immediate risk to
occupants.  However, significantly deteriorated lead-based paint may
present an immediate danger to occupants, especially to young children. 
Table 4.3 presents the number and percentage of HUs with deteriorated
LBP and significantly deteriorated LBP by location in the building -
either interior or exterior, or both.

Table 4.3	Prevalence of Deteriorated and Significantly Deteriorated
Lead-Based Paint (LBP) by Location in the Building

Deteriorated LBP

Location	Number of HUsa with Deteriorated LBP (000)	Percentb of HUs with
Deteriorated LBP(%)	HUs in Sample

	Estimate	Lower 95% CIc	Upper 95% CI	Estimate	Lower 95% CI	Upper 95% CI

	Interior Only	4,180	2,851	5,509	4%	3%	6%	39

Both Interior and Exterior	6,236	4,661	7,811	7%	5%	8%	62

Exterior Only	7,009	4,922	9,097	7%	5%	10%	61

Total with Deteriorated LBP	17,425	14,816	19,735	18%	15%	21%	162

No Deteriorated LBP	78,263	75,953	80,572	82%	79%	84%	669

TOTAL	95,688

	100%

	831

Significantly Deteriorated LBP 

Location	Number of HUsa with Significant Deteriorated LBP (000)	Percentb
of HUs with Significant Deteriorated LBP(%)	HUs in Sample

	Estimate	Lower 95% CIc	Upper 95% CI	Estimate	Lower 95% CI	Upper 95% CI

	Interior Only	2,629	1,692	3,566	3%	2%	4%	28

Both Interior and Exterior	3,487	2,132	4,842	4%	2%	5%	34

Exterior Only	7,518	5,357	9,679	8%	6%	10%	65

Total with Significantly Deteriorated LBP	13,634	10,928	16,341	14%	11%
17%	127

No Significantly Deteriorated LBP	82,053	79,347	84,760	86%	83%	89%	704

TOTAL	95,688

	100%

	831

a	“Housing units” include permanently occupied, noninstitutional
housing units in which children are permitted to live.

b	Percentages are calculated with total housing units (95,688) as the
denominator.  Percentages may not total 100% due to rounding.

c	CI = 95% confidence interval for the estimated number or percent.

An estimated 17 million ((2 million) or 18 percent ((3%) of housing
units in the United States have deteriorated LBP.  The deteriorated LBP
is only on the exterior for approximately 40 percent of the homes with
deteriorated LBP.  An estimated 14 million ((3 million) or 14 percent
((3%) of housing units in the United States have significantly
deteriorated LBP.  Roughly 55 percent of these homes have significant
deterioration only on exterior surfaces.  Twenty percent of these homes
have the significantly deteriorated LBP only on interior surfaces.

Table 4.4 presents the number and percentage of housing units with
deteriorated and significantly deteriorated LBP by construction year. 
The data suggest that older homes are more likely to have deteriorated
LBP than newer homes.  Only 3% of homes built between 1960 and 1977 have
deteriorated LBP, but the percentage increases to 32% for homes built
between 1940 and 1959, and to 56% for homes built before 1940.  Only 2%
of homes built between 1960 and 1977 have significantly deteriorated
LBP, but the percentage increases to 25% for homes built between 1940
and 1959, and to 44% for homes built before 1940.

Table 4.4	Distribution of Housing Units (HUs) with Deteriorated and
Significantly Deteriorated Lead-Based Paint (LBP) by Construction Year

Deteriorated LBP

Construction Year	Total HUs (000)b	Number of HUs with

Deteriorated LBP (000)	Percent of HUs with

Deteriorated LBP (%)a



Estimate	Lower 95% CIc	Upper 95% CI	Estimate	Lower 95% CI	Upper 95% CI

1978-1998	29,774	139	0	330	0%	0%	1%

1960-1977	27,874	910	235	1,586	3%	1%	6%

1940-1959	20,564	6,510	4,603	8,418	32%	22%	41%

Before 1940	17,476	9,866	8,111	11,620	56%	46%	66%

Total HUs	95,688	17,425	15,222	19,628	18%	16%	21%

Significantly Deteriorated LBP

Construction Year	Total HUs (000)b	No. of HUs with Significantly
Deteriorated LBP (000)	Percent of HUs with Significantly Deteriorated
LBP (%)a



Estimate	Lower 95% CIc	Upper 95% CI	Estimate	Lower 95% CI	Upper 95% CI

1978-1998	29,774	83	0	238	0%	0%	1%

1960-1977	27,874	610	97	1,122	2%	0%	4%

1940-1959	20,564	5,190	3,387	6,993	25%	16%	34%

Before 1940	17,476	7,752	6,048	9,456	44%	35%	54%

Total HUs	95,688	13,635	9,893	16,582	14%	10%	17%

a	Percentages may not total 100% due to rounding.

b	“Housing units” include permanently occupied, noninstitutional
housing units in which children are permitted to live.

c	CI = 95% confidence interval for the estimated number or percent.

Figure 4.4 is a bar chart that summarizes the above survey data on
deteriorated and significantly deteriorated LBP by construction year. 
It graphically displays the downward trends in the prevalence of LBP in
homes and of damaged LBP in homes.

Figure 4.4.  Presence and Condition of LBP by Construction Year

 

4.3	Paint Lead Loadings in Housing

Table 4.5 presents the distribution of the highest lead paint loading by
location in the building for selected thresholds: 0.3, 0.6, 0.8, 1.0,
1.3, 4.0, and 10.0 mg/cm2.  By definition, paint with less than 1.0
mg/cm2 is not LBP; thus, the first four categories represent paint that
is considered not to be LBP.  The majority of the surfaces did not
contain lead-based paint: in 70 percent of HUs the highest interior
readings, and in 69 percent of HUs the highest exterior readings, were
below 1.0 mg/cm2. Fourteen percent of HUs had at least one paint sample
with 10 mg/cm2 or more of lead-based paint.



Table 4.5	Distribution of Paint Lead Loading by Location in the Building

Maximum Paint	Interior (% HUs)a	Exterior (% HUs)	Anywhere (% HUs)

Lead Loading in HUc	Estimate	Lower

95% CIb	Upper

95% CI	Estimate	Lower

95% CI	Upper

95% CI	Estimate	Lower

95% CI	Upper

95% CI

GT 0 mg/cm2 	100%	100%	100%	100%

	100%



GE 0.3 mg/cm2	51%	47%	55%	42%	39%	46%	62%	58%	67%

GE 0.6 mg/cm2	37%	34%	39%	35%	32%	38%	47%	44%	50%

GE 0.8 mg/cm2	31%	29%	34%	32%	29%	36%	42%	39%	45%

GE 1.0 mg/cm2	30%	27%	33%	31%	27%	34%	40%	36%	43%

GE 1.3 mg/cm2	26%	24%	29%	29%	26%	33%	36%	33%	40%

GE 4.0 mg/cm2	17%	14%	20%	18%	15%	22%	24%	20%	27%

GE 10.0 mg/cm2	9%	7%	12%	10%	8%	13%	14%	11%	17%

TOTAL HUs	100%

	100%

	100%



a	All percentages are calculated with total housing units (95,688) as
the denominator.  “Housing units” include permanently occupied,
noninstitutional housing units in which children are permitted to live.

b	CI = 95% confidence interval for the estimated number or percent.

c	GT equals “greater than.”  GE equals “greater than or equal
to.”

Table 4.6 presents the distribution of paint lead loadings by location
in the building and construction year for the selected thresholds.  This
clearly demonstrates the effectiveness of the reduction from 1940 to
1980 in the amount of lead added to commercial residential paint.  An
estimated 87 percent of housing reported as built before 1940 had at
least one paint lead measurement somewhere in the house at 1.0 mg/cm2,
or above.  This decreased to 24 percent of housing reported as built
between 1960 and 1977, and to 7 percent of housing reported as built
since 1978.  The same pattern holds for very high lead levels, with 55
percent of pre-1940 homes having some lead above 10 mg/cm2 but only 1
percent for post-1977 housing.

Table 4.6	Distribution of Paint Lead Loading by Location in the
Building and Construction Year

Percent of HUs with LBPa,b

Largest Paint Lead Loading in the Housing Unit	Year of Construction

	1978-1998	1960-1977	1940-1959	Before 1940	Subtotal

Interior

GT 0 mg/cm2 	100%	100%	100%	100%	100%

GE 0.3 mg/cm2	23%	39%	77%	91%	51%

GE 0.6 mg/cm2	9%	21%	59%	83%	37%

GE 0.8 mg/cm2	6%	16%	48%	80%	31%

GE 1.0 mg/cm2	4%	16%	46%	79%	30%

GE 1.3 mg/cm2	3%	12%	41%	72%	26%

GE 4.0 mg/cm2	1%	6%	19%	60%	17%

GE 10.0 mg/cm2	1%	2%	7%	38%	9%

Exterior

GT 0 mg/cm2 	100%	100%	100%	100%	100%

GE 0.3 mg/cm2	11%	31%	69%	81%	42%

GE 0.6 mg/cm2	7%	18%	64%	76%	35%

GE 0.8 mg/cm2	4%	16%	61%	73%	32%

GE 1.0 mg/cm2	3%	13%	59%	72%	31%

GE 1.3 mg/cm2	3%	11%	56%	71%	29%

GE 4.0 mg/cm2	0%	6%	28%	56%	18%

GE 10.0 mg/cm2	0%	2%	10%	41%	10%

Anywhere in Building

GT 0 mg/cm2 	100%	100%	100%	100%	100%

GE 0.3 mg/cm2	30%	57%	89%	95%	62%

GE 0.6 mg/cm2	15%	31%	80%	89%	47%

GE 0.8 mg/cm2	10%	26%	70%	88%	42%

GE 1.0 mg/cm2	7%	24%	69%	87%	40%

GE 1.3 mg/cm2	5%	18%	65%	84%	36%

GE 4.0 mg/cm2	1%	10%	34%	73%	24%

GE 10.0 mg/cm2	1%	3%	14%	55%	14%

a	“Housing units” include permanently occupied, noninstitutional
housing units in which children are permitted to live.

b	GT equals “greater than.”  GE equals “greater than or equal
to.”

Tables 4.7 and 4.8 present selected parameters of the distributions of
paint lead loadings by interior and exterior component types.  Tables
4.7 and 4.8 do not include the geometric means and standard deviations
because of the large number of zero XRF readings.  All of the
distributions in Tables 4.7 and 4.8 are right-skewed and cannot be
fitted by normal distributions.  A better model would be the lognormal
distribution.  Appendix C presents a discussion of fitting models to
these data.

Table 4.7	Estimated Empirical Distribution Parameters of Paint Lead
Loadings by Interior Component Types

	Walls, Floors, Ceilings	Windows	Doors	Trim	Other

	mg/cm2	mg/cm2	mg/cm2	mg/cm2	mg/cm2

Arithmetic Mean	0.2	0.9	0.5	0.5	0.4

Arithmetic Standard Deviation	1.5	3.4	2.5	2.3	2.6

25th Percentile	0.0	0.0	0.0	0.0	0.0

Median	0.0	0.0	0.0	0.0	0.0

75th  Percentile	0.0	0.1	0.0	0.0	0.0

90th Percentile	0.1	1.3	0.4	0.2	0.2

95th Percentile	0.3	6.0	1.9	1.4	0.5

Number of Readings	14,876	5,513	4,596	2,578	2,686



Table 4.8	Estimated Empirical Distribution Parameters of Paint Lead
Loadings by Exterior Component Types

	Walls	Windows	Doors	Trim	Porch	Other

	mg/cm2	mg/cm2	mg/cm2	mg/cm2	mg/cm2	mg/cm2

Arithmetic Mean	0.9	2.5	1.2	1.1	1.1	1.6

Arithmetic Standard Deviation	3.4	6.2	4.4	4.5	3.9	5.1

25th Percentile	0.0	0.0	0.0	0.0	0.0	0.0

Median	0.0	0.0	0.0	0.0	0.0	0.0

75th  Percentile	0.1	1.7	0.1	0.1	0.1	0.3

90th Percentile	1.9	7.7	2.4	1.8	2.2	3.3

95th Percentile	4.8	15.3	7.4	5.4	7.8	10.7

Number of Readings	2,008	781	1,398	446	698	250



Table 4.9 summarizes the data on LBP by component type from a different
perspective; it presents the percentage of components with LBP by
component type and housing unit age.  Table 4.9 shows the expected
trends and differences:  exterior components are more likely to be
leaded than interior components; interior walls are least likely to have
LBP; and older homes have more LBP than newer homes.

Table 4.9	Percentage of Components with Lead-Based Paint by Component
Type and HU Age

Component Type	Year of Construction

	All Years	1978-1998	1960-1977	1940-1959	Before 1940

Interior	Walls, Floors Ceilings	2%	0%	1%	2%	7%

	Windows	9%	1%	2%	6%	21%

	Doors	7%	0%	1%	7%	22%

	Trim	5%	0%	2%	4%	15%

	Other	4%	0%	1%	2%	12%

	All Components	30%	4%	16%	46%	79%

Exterior	Walls	14%	0%	9%	18%	34%

	Windows	25%	0%	12%	30%	41%

	Doors	15%	2%	5%	29%	33%

	Trim	11%	3%	8%	16%	24%

	Porch	15%	1%	7%	25%	28%

	Other	18%	0%	8%	37%	37%

	All Components	31%	3%	13%	59%	72%



Table 4.10 presents selected parameters of the distributions of paint
lead loading by interior rooms and for all exterior surfaces. Paint lead
loadings are higher on exterior surfaces than on interior surfaces. 
Differences between average paint lead loadings in interior rooms are
small compared to differences between component types within rooms (see
Table 4.7).

Table 4.10	Estimated Empirical Distribution Parameters of Paint Lead
Loadings by Room Type

	Kitchen	Living Room	Bed Room	Other Room	Exterior

	mg/cm2	mg/cm2	mg/cm2	mg/cm2	mg/cm2

Arithmetic Mean	0.5	0.4	0.4	0.4	1.3

Arithmetic Standard Deviation	2.3	2.3	2.3	2.3	4.4

25th Percentile	0	0	0	0	0

Median	0	0	0	0	0

75th  Percentile	0.1	0	0	0	0.2

90th Percentile	0.4	0.2	0.2	0.2	2.7

95th Percentile	1.7	0.8	1	0.8	7.8

Number of Readings	7127	7375	8053	7387	5581



4.4	Comparison of Prevalence of Lead-Based Paint (LBP) to the 1990 LBP
Survey

Table 4.11 compares the prevalence of LBP and deteriorated LBP found in
the NSLAH (restricted to pre-1980 construction) with the prevalence
found in the 1990 LBP Survey, which was similarly restricted.  The NSLAH
shows fewer total homes built before 1980.  This is partly due to the
fact that housing where children could not live was excluded from the
current survey.  There has also been a loss of homes built before 1980
due to demolition.

 

Table 4.11	Comparison of the Prevalence of Lead-Based Paint1 to the 1990
LBP Survey

	1990 LBP Survey	Current NSLAH

(Pre-1980 HUs)

	Number

(000)	Percent

(%)	Number

(000)	Percent

(%)

Total HUs Built Before 1980	77,177	100%	68,756	100%

	HUs with LBP	64,059	83%	34,195	50%

	Interior LBP	48,986	63%	26,184	38%

	Exterior LBP	56,495	73%	27,373	40%

HUs with Deteriorated LBPa	14,354	19%	14,962	22%

	Interior Deteriorated LBP	5,596	7%	7,281	11%

	Exterior Deteriorated LBP	9,657	13%	11,784	17%

a	Deteriorated LBP is defined in the 1995 HUD Guidelines (Guidelines for
the Evaluation and Control of Lead-Based Paint Hazards in Housing, HUD
1539-LBP, Washington, D.C.) for the NSLAH, and as 5 square feet or more
of deteriorated LBP for the 1990 survey.



A lower percentage of pre-1980 homes were found with LBP during the
current NSLAH (50% versus 83% found in 1990).  This was not unexpected
because there has been renovation, remodeling, demolition, and paint
removal activities in the intervening years.

The number of homes with LBP was expected to decrease between 1990 and
1999, primarily due to demolition, renovation and remodeling of older
homes.  For homes built around the same time, the percentage of homes
with LBP was also expected to decrease between 1990 and 1999, primarily
due to renovation and remodeling of older homes.  However, the
difference between the 1990 and 1999 survey estimates of the number and
percentage of homes with LBP is greater than might be expected from
these sources.  Factors that might explain the differences include:

Demolition.  Demolition of older homes reduces the number of homes with
LBP over time.  

Renovation, Remodeling, and Remediation.  Removal of surfaces with LBP
during renovation and remodeling reduces both the number and percentage
of homes with LBP.  Covering (encapsulating) surfaces with LBP reduces
the quantity of lead directly accessible to occupants and reduces the
XRF reading on the surface.  The lower XRF readings result in fewer
homes classified as having LBP.  Although additional coats of paint are
expected to reduce XRF readings, the effect of painting is expected to
be small.

XRF Models.  XRF measurements from the 1999 survey are more precise than
those from the 1990 survey.  The less precise XRF instrument used in the
1990 survey was more likely to misclassify a surface as having LBP and
thus increases the estimated number of surfaces with LBP.

XRF Calibration Procedures.  The calibration of the XRF instrument used
for the 1990 survey was checked at paint loadings of 0.6 and 3.0 mg lead
per sq.cm.  The 1990 XRF readings were recalibrated based on these
checks.  The calibration of the XRF instrument used for the 1999 survey
was checked at paint loadings of 0.0 and 1.0 mg lead per sq.cm.  The
difference in the calibration procedures and the assumptions required
for the re-calibration of the 1990 XRF readings have an unknown effect
of the estimated number of homes with LBP.

Sample Design.  The two surveys used slightly different criteria and
sampling methods for selecting the sampled homes.  The primary
difference being that the 1999 survey excluded homes where children were
excluded, some of which may have been included in the 1990 survey.

Within-home Data Collection Procedures.  The two surveys had different
criteria and procedures for within-home data collection.  The 1999
survey collected measurements in more rooms within the sampled homes
than the 1990 survey (4 to 6 rooms vs. 2 rooms).  Within the sampled
rooms the 1999 survey also collected more measurements.  The larger
number of measurements in the 1999 survey would tend to increase the
number of homes classified as having LBP, other factors being equal,
contrary to the observed decrease.  

Approximately the same percentage of pre-1980 homes with LBP had
deteriorated LBP in both surveys (22% versus 19% found in 1990).  The
slight increase in the percentage of homes with deteriorated LBP was
expected, because these homes are now nine years older.  In addition,
the definitions of deteriorated LBP were different for the two studies. 
In the 1990 LBP Survey, deteriorated interior LBP was defined as more
than 5 square feet of deteriorated interior lead-based paint, with a
similar definition for exterior lead-based paint.  Table 4.11 defines
deteriorated LBP for the NSLAH according to the 1995 Guidelines
(Guidelines for the Evaluation and Control of Lead-Based Paint Hazards
in Housing).  For these reasons, Table 4.11 does not attempt to present
comparisons with respect to significantly deteriorated LBP; the concepts
of deterioration above de minimis levels were different in 1990 and
1999.

4.5	Area of Lead-Based Paint in Housing

Table 4.12 presents estimates of the areas of surfaces with LBP by
architectural component type.  An estimated 7.4 billion square feet of
painted interior surfaces are covered with LBP.  This represents 2
percent of the area of painted interior surfaces in all homes.  Although
2 percent of paint on walls, floors, and ceilings is lead-based, the
area of these LBP-coated components accounts for 67 percent of all
interior surfaces with LBP. Conversely, paint on window and door system
components is more likely to contain LBP, but the total surface area of
LBP on these components is only 21 percent of the area of all interior
painted surfaces.

An estimated 29.2 billion square feet of painted exterior surfaces are
covered with LBP.  This represents 22 percent of the area of painted
exterior surfaces in all homes.  Wall siding accounts for most (67%) of
the surface area of LBP.

Although a large number of homes have LBP, most of them have relatively
small areas of LBP.  The average home with LBP has 259 square feet of
interior LBP and 996 square feet of exterior LBP. 

Table 4.12	Area of LBP by Painted Component

Component	National Total Amount of LBP	Average Amount LBP Per Housing
Unit with LBP

	Millions of sq. ft.	Percent of All Paint on Component	(square feet)

INTERIOR:



	Wall, Floor, Ceiling	4,993	2%	173

Window	687	9%	24

Door	911	6%	32

Trim	499	5%	17

Cabinets, Chimney, Beams	388	2%	13

TOTAL	7,448	2%	259

EXTERIOR:



	Wall	26,706	18%	912

Window	365	28%	12

Door	446	14%	15

Trim	556	12%	19

Porch	1,086	21%	37

TOTAL	29,159	22%	996

5.  DUST LEAD IN HOUSING

Chapter 5 presents estimates of the prevalence of lead-contaminated dust
in housing, including the dust lead loadings and the association between
interior dust lead and exterior LBP condition.  No comparison is made
with the dust lead findings of the 1990 LBP Survey because the vacuum
technique employed in the earlier study is not comparable to the wipe
technique used in the NSLAH.

5.1	Prevalence of Dust Lead in Housing

Table 5.1 presents the prevalence of all homes and homes with one or
more children under 6 years of age with a dust lead hazard somewhere in
the home, as defined by the HUD Lead Safe Housing Rule (24 CFR 35
Subparts B-R).  The HUD Lead Safe Housing Rule defines a dust lead
hazard as greater than or equal to 40 µg/ft2 lead on floors or 250
µg/ft2 lead on window sills.  There is no longer a hazard level defined
for dust lead on window troughs.  The earlier HUD 1995 Guidelines
considered lead in dust to be a hazard when dust on floors had greater
than 100 µg/ft2 lead, dust on window sills had greater than 500 µg/ft2
lead, or dust on window troughs had greater than 800 µg/ft2 lead.  As
discussed in Section 3.1 a housing unit is considered to have a dust
lead hazard if a dust lead hazard exists at any of the sampled locations
in the housing unit.  In contrast, the EPA Lead Hazards Identification
Rule considers a housing unit to have a dust lead hazard if the average
of the samples exceeds the relevant 

Using the HUD Lead Safe Housing Rule definition of dust lead hazard, an
estimated 16 percent (+2%) of all homes have a dust lead hazard
somewhere in the home, and 3 percent (+1%) of all homes have both a
child under 6 years of age and a dust lead hazard.

Figures 5.1 and 5.2 present the percentage of homes with dust lead
hazards by surface – floor and window sill – and by housing unit age
and household income.  The expected gradient with housing unit age is
seen in Figure 5.1.  Both figures show that window sills have the
greater prevalence of dust lead hazards.

Table 5.1	Prevalence of Housing Units with a Dust Lead Hazard Somewhere
in the Home

HU Categorya	Number of HUs (000)	Percent of HUs(%)b

	Estimate	Lower

95% CIc	Upper

95% CI	Estimate	Lower

95% CI	Upper

95% CI

HUs with Lead Dust Hazard 	15,468	12,982	17,954	16%	14%	19%

HUs with Children Under 6 Years and Lead Dust Hazard 	2,634	1,586	3,682
16%	11%	20%

a	“Housing units” include permanently occupied, noninstitutional
housing units in which children are permitted to live.

b	All percentages are calculated with total housing units (95,688) or
HUs with resident children under age 6 (16,402) as the denominator.

c	CI = 95% confidence interval for the estimated number or percent.

Figure 5.1	Dust Lead Hazards by Construction Year

Figure 5.2	Dust Lead Hazards by Household Income

5.2	Dust Lead Loadings in Housing

Figures 5.3 and 5.4 present the distribution of maximum (within housing
unit) dust lead loadings on floors and window as histograms. Tables 5.2
and 5.3 present information on maximum and average dust loadings,
respectively, by surface.

Table 5.2 presents the distribution of maximum dust lead loadings by
surface (floor, window sill, and window trough) for all U.S. homes in
the target population, for selected threshold values.  As with Table
5.1, the estimates are based on the maximum dust lead loading in the
home for the particular surface.

Only an estimated 6 percent of all homes have maximum floor dust lead
loadings above the HUD Lead Safe Housing Rule standard of 40 µg/ft2. 
More homes have a window sill lead dust hazard than have a floor dust
hazard.  An estimated 14 percent of all homes have sill dust lead
loadings above the HUD Lead Safe Housing Rule standard hazard of 250
µg/ft2.

Figure 5.3		Distribution of Maximum Floor Dust Lead Loadings

Figure 5.4		Distribution of Maximum Window Sill Dust Loadings

Table 5.2	Distribution of Maximum Dust Lead Loadings in Housing Units by
Surface

Maximum Dust Lead Loading in HU (µg/sq ft)	Number of HUs (000)a	Percent
of HUs (%)b

	Estimate	Lower

95% CId	Upper

95% CI	Estimate	Lower

95% CI	Upper

95% CI

Floors:c

LT LODe	38,369	34,302	42,436	40%	36%	44%

GE LOD	57,196	53,147	61,244	60%	56%	64%

GE 5	28,200	24,920	31,481	30%	26%	33%

GE 10	15,964	13,141	18,787	17%	14%	20%

GE 20	8,989	6,871	11,108	9%	7%	12%

GE 40	5,495	3,770	7,220	6%	4%	8%

GE 100 	2,426	1,470	3,382	3%	2%	4%

Missingf	123

	0%



Window Sills:







LT LOD	9,602	7,326	11,879	10%	8%	13%

GE LOD	82,134	78,850	85,418	86%	83%	88%

GE 125	20,338	17,590	23,085	21%	19%	24%

GE 250	13,439	11,516	15,362	14%	12%	16%

GE 500 	9,042	7,136	10,949	10%	8%	12%

No sill present in HUg	2,221	848	3,594	2%	1%	4%

Missingf	1,731

	2%



Window Troughs:







LT LOD	374	0	799	0%	0%	1%

GE LOD	72,638	67,107	78,169	76%	70%	82%

GE 800	21,210	16,489	25,931	22%	17%	27%

No trough present in HU	7,318	5,176	9,459	8%	5%	10%

Missingf	15,358

	16%



a	“Housing units” include permanently occupied, noninstitutional
housing units in which children are permitted to live.

b	All percentages are calculated with total housing units (95,688) as
the denominator.

c	Floors include both carpeted and uncarpeted floors.

d	CI = 95% confidence interval for the estimated number or percent.

e	LT equals “less than.”  GE equals “greater than or equal to.” 
LOD equals “limit of detection.”

f 	Missing means that the floor, sill, or trough was present, but that
no lead value is available (either the sample was not collected, e.g.,
due to inaccessibility or respondent refusal, or the laboratory did not
submit a value).

g 	No sill/trough present” means that there was no sill or trough in
the HU, e.g., windows were flush with the wall, or awning windows were
installed.

Table 5.3 presents the distribution of average dust lead loadings by
surface (floor, window sill, and window trough) for all U.S. homes in
the target population, for selected threshold values.  The average dust
loading for each surface was determined by simply adding floor, window
sill, or window trough dust loadings for each room sampled in each HU
and dividing by the number of rooms sampled (unweighted average).  

The same trends are observed in Table 5.3 for average dust lead loadings
as for the distribution of maximum dust lead loadings in Table 5.2. 
However, Table 5.3 shows that fewer homes have carpeted floor dust lead
hazards than uncarpeted floor dust hazards (i.e., carpeted floors have
lower dust lead loadings – as indicated by the results of the wipe
sampling employed in the survey).  In fact, the regression modeling
suggested that lead loadings on carpeted surfaces were approximately 25
percent lower than smooth and cleanable surfaces and 75 percent lower
than uncarpeted surfaces that were not smooth and cleanable (see Figure
C.7)

Table 5.3	Distribution of Average Dust Lead Loadings in Housing Units
by Surface

Average Dust Lead Loading

in HU (µg/sq ft)	Number of HUs (000)a	Percent of HUs (%)b

	Estimate	Lower

95% CIc	Upper

95% CI	Estimate	Lower

95% CI	Upper

95% CI

Floors (Uncarpeted):







LT LODd	51,252	47,180	55,323	54%	49%	58%

GE LOD	40,529	36,366	44,693	42%	38%	47%

GE 5	17,291	13,855	20,727	18%	14%	22%

GE 10	8,512	6,348	10,676	9%	7%	11%

GE 20	4,843	3,263	6,423	5%	3%	7%

GE 40	2,449	1,414	3,484	3%	2%	4%

GE 100 	966	239	1,694	1%	0%	2%

No Uncarpeted	3,907

	4%



Floors (Carpeted):







LT LODd	66,628	63,563	69,694	70%	67%	73%

GE LOD	21,356	18,700	24,012	22%	20%	25%

GE 5	5,806	4,073	7,540	6%	4%	8%

GE 10	2,374	1,488	3,261	3%	2%	3%

GE 20	1,368	674	2,061	1%	1%	2%

GE 40	298	0	634	0%	0%	1%

GE 100 	59	0	190	0%	0%	0%

No Carpeted	7,704

	8%



Window Sills:







LT LOD	12,800	10,201	15,399	13%	11%	16%

GE LOD	78,936	75,462	82,410	83%	80%	85%

GE 125	13,875	11,717	16,033	15%	12%	17%

GE 250	8,287	6,636	9,938	9%	7%	10%

GE 500 	4,900	3,611	6,190	5%	4%	7%

No sill present in HUe	2,221	848	3,594	2%	1%	4%

Missingf	1,731

	2%



Window Troughs:







LT LOD	663	52	1,274	1%	0%	1%

GE LOD	72,349	66,714	77,985	76%	70%	81%

GE 800	16,395	12,827	19,964	17%	13%	21%

No trough present in HU	7,318	5,176	9,459	8%	5%	10%

Missing	15,358

	16%



a	“Housing units” include permanently occupied, noninstitutional
housing units in which children are permitted to live.

b	All percentages are calculated with total housing units (95,688) as
the denominator.

c	CI = 95% confidence interval for the estimated number or percent.

d	LT equals “less than.”  GT equals “greater than.” LOD equals
“limit of detection.”

e	“No sill/trough present” means that there was no sill or trough in
the HU, e.g., windows were flush with the wall, or awning windows were
installed.

f	Missing means that the floor, sill, or trough was present, but that no
lead value is available (either the sample was not collected, e.g., due
to inaccessibility or respondent refusal, or the laboratory did not
submit a value).

Table 5.4 presents the distribution of dust lead loadings by room type
and surface for selected threshold values.  The vast majority of floors
had undetectable levels of dust lead.  About 1 percent of rooms had dust
lead levels above the HUD Lead Safe Housing Rule standard for floors. 
Five percent of rooms had dust lead levels above the HUD Lead Safe
Housing Rule standard for window sills.

Figure 5.5 presents a scatter plot of the window sill dust lead loadings
against the floor dust lead loadings.  Each point in Figure 5.5
represents the floor and window dust lead loadings from the same room in
the same housing unit.  The figure shows evidence of a moderate positive
correlation between the dust lead loadings on these two neighboring
surfaces.

Table 5.4	Distribution of Dust Lead Loading by Room and Surfaces

Dust Lead 	Kitchens	Living Rooms	Bedrooms	Other Rooms

Loading

(µg/sq ft)a	Number of Rooms (000)b	Percent of Rooms (%)c	Number of
Rooms (000)	Percent of Rooms (%)	Number of Rooms (000)	Percent of Rooms
(%)	Number of Rooms (000)	Percent of Rooms (%)

Floorsd









LT LODe	63,244	66%	98,433	78%	170,153	78%	231,924	73%

GE LOD	31,633	33%	26,732	21%	45,915	21%	84,438	27%

GE 5	14,062	15%	9,863	8%	19,340	9%	34,291	11%

GE 10	5,568	6%	4,287	3%	10,612	5%	18,933	6%

GE 20	2,571	3%	1,824	1%	4,175	2%	10,626	3%

GE 40	712	1%	880	1%	1,593	1%	7,477	2%

GE 100 	335	0%	170	0%	1,354	1%	785	0%

Missingf	488	1%	1,737	1%	1,002	1%	2,594	1%

Total Rooms	95,365	100%	126,902	100%	217,069	100%	318,956	100%

Window Sills

LT LOD	23,001	24%	29,378	23%	38,394	18%	44,841	14%

GE LOD	50,393	53%	72,154	57%	140,748	65%	141,008	44%

GE 125	7,037	7%	10,937	9%	20,393	9%	23,472	7%

GE 250	4,329	5%	6,731	5%	11,075	5%	17,156	5%

GE 500 	2,455	3%	4,386	4%	6,661	3%	10,096	3%

Missing	3,870	4%	5,462	4%	17,993	8%	16,321	5%

No Sills	18,102	19%	19,907	16%	19,934	9%	116,785	37%

Total Rooms	95,365	100%	126,902	100%	217,069	100%	318,956	100%

Window Troughs

LT LOD	3,293	4%	4,125	3%	5,211	2%	6,657	2%

GE LOD	49,962	52%	64,827	51%	122,752	57%	119,236	37%

GE 800	9,249	10%	12,795	10%	23,268	11%	21,487	7%

Missing	16,542	17%	27,537	22%	57,738	27%	58,044	18%

No Trough	25,568	27%	30,413	24%	31,368	15%	135,019	42%

Total Rooms	95,365	100%	126,902	100%	217,069	100%	318,956	100%

a	In this table, maximum loading is not applicable as only one dust
sample was collected from each surface in each room.

v	“Housing units” include permanently occupied, noninstitutional
housing units in which children are permitted to live.

c	All percentages are calculated with total rooms of that type as the
denominator.

d	Floors include both carpeted and uncarpeted floors.

e	LT equals “less than.”  GE equals “greater than or equal to.” 
LOD equals “limit of detection.”

f	Missing means that the floor, sill, or trough was present, but that no
lead value is available (either the sample was not collected, e.g., due
to inaccessibility or respondent refusal, or the laboratory did not
submit a value).



Figure 5.5	Association Between Dust Lead Loading on Floors and Window
Sills

Tables 5.5a through 5.5c present the distributions of floor, window
sill, and window trough dust lead loadings, respectively, by selected
thresholds and by year of construction.  It is evident that older homes
have considerably more dust lead than newer homes.  The percentage of
homes over the HUD Rule standard of 40 µg/ft2 for floor dust (Table
5.5a) increases from less than 1 percent for post-1977 homes to 16
percent for pre-1940 homes.  The percentage of homes over the sill dust
standard (Table 5.5b) steadily increases from 4 percent for post-1977
homes to 14 percent for 1940-1959 homes to 40 percent for pre-1940
homes.

Table 5.5a	Maximum Floor Dust Lead Loading by Year of Construction

Maximum Floor Dust	Year of Construction

Lead Loading(µg/sq ft)a	1978-1998	1960-1977	1940-1959	Before 1940

	Number (000)	Percent

(%)b	Number (000)	Percent

(%)	Number (000)	Percent

(%)	Number (000)	Percent

(%)

LT LODc	Number HUsd	17,487	59%	13,006	47%	5,667	28%	2,208	13%

	Lower 95% CIe	15,646	53%	10,272	38%	4,076	20%	782	4%

	Upper 95% CI	19,329	65%	15,740	56%	7,257	35%	3,635	21%

GE LOD	Number HUs	12,241	41%	14,868	53%	14,820	72%	15,268	87%

	Lower 95% CI	10,419	35%	12,454	44%	12,770	65%	13,008	79%

	Upper 95% CI	14,064	47%	17,281	62%	16,869	80%	17,527	96%

GE 5	Number HUs	3,233	11%	4,968	18%	8,753	43%	11,245	64%

	Lower 95% CI	2,285	8%	3,567	13%	7,060	35%	9,635	55%

	Upper 95% CI	4,181	14%	6,370	23%	10,446	50%	12,855	73%

GE 10	Number HUs	1,153	4%	2,488	9%	4,938	24%	7,386	42%

	Lower 95% CI	370	1%	1,607	6%	3,447	17%	5,802	33%

	Upper 95% CI	1,935	7%	3,369	12%	6,428	31%	8,970	52%

GE 20	Number HUs	97	0%	1,112	4%	2,784	14%	4,996	29%

	Lower 95% CI	0	0%	516	2%	1,283	6%	3,759	22%

	Upper 95% CI	267	1%	1,708	6%	4,286	21%	6,234	35%

GE 40	Number HUs	97	0%	588	2%	1,967	10%	2,843	16%

	Lower 95% CI	0	0%	216	1%	718	4%	1,989	11%

	Upper 95% CI	267	1%	961	4%	3,215	16%	3,698	21%

GE 100	Number HUs	97	0%	280	1%	935	5%	1,114	6%

	Lower 95% CI	0	0%	0	0%	121	1%	587	3%

	Upper 95% CI	267	1%	640	2%	1,750	9%	1,642	9%

Missing	Number HUs	0	0%	0	0%	77	0%	0	0%

a	Floors include both carpeted and uncarpeted floors.

b	All percentages are calculated with total housing units (95,688) as
the denominator.

c	LT equals “less than.”  GE equals “greater than or equal to.”
LOD equals “limit of detection.”

d	“Housing units” include permanently occupied, noninstitutional
housing units in which children are permitted to live.

e	CI = 95% confidence interval for the estimated number or percent.

Table 5.5b	Maximum Window Sill Dust Lead Loading by Year of
Construction

Maximum Floor Dust	Year of Construction

Lead Loading(µg/sq ft)a	1978-1998	1960-1977	1940-1959	Before 1940

	Number (000)	Percent

(%)b	Number (000)	Percent

(%)	Number (000)	Percent

(%)	Number (000)	Percent

(%)

LT LODc	Number HUsd	6,196	21%	1,924	7%	1,074	5%	408	2%

	Lower 95% CIe	4,270	14%	826	3%	277	1%	0	0%

	Upper 95% CI	8,122	27%	3,021	11%	1,871	9%	835	5%

GE LOD	Number HUs	21,823	73%	24,729	89%	18,779	91%	16,803	96%

	Lower 95% CI	19,833	68%	22,996	84%	16,956	87%	15,103	93%

	Upper 95% CI	23,814	79%	26,462	94%	20,602	96%	18,503	99%

GE 125	Number HUs	1,806	6%	4,097	15%	5,407	26%	9,028	52%

	Lower 95% CI	578	2%	2,444	9%	3,954	19%	7,196	42%

	Upper 95% CI	3,033	10%	5,749	21%	6,860	33%	10,861	61%

GE 250	Number HUs	1,029	4%	1,755	6%	3,712	18%	6,943	40%

	Lower 95% CI	139	1%	1,086	4%	2,556	12%	5,476	31%

	Upper 95% CI	1,919	7%	2,424	9%	4,867	24%	8,410	48%

GE 500	Number HUs	447	2%	747	3%	2,869	14%	4,980	29%

	Lower 95% CI	0	0%	274	1%	1,779	9%	3,712	21%

	Upper 95% CI	1,024	3%	1,219	4%	3,959	19%	6,247	36%

Missing	Number HUs	299	1%	851	3%	361	2%	220	1%

No sills	Number HUs	1,456	5%	371	1%	349	2%	45	0%

	Lower 95% CI	456	2%	0	0%	0	0%	0	0%

	Upper 95% CI	2,456	8%	762	3%	730	4%	143	1%

a	Missing means that the sill was present, but that no lead value is
available (either the sample was not collected, e.g., due to
inaccessibility or respondent refusal, or the laboratory did not submit
a value).  “No sill present” means that there was no sill in the HU,
e.g., windows were flush with the wall, or awning windows were
installed.

b	All percentages are calculated with total housing units (95,688) as
the denominator.

c	LT equals “less than.”  GT equals “greater than or equal to.” 
LOD equals “limit of detection.”

d	“Housing units” include permanently occupied, noninstitutional
housing units in which children are permitted to live.

e	CI = 95% confidence interval for the estimated number or percent.

Table 5.5c	Maximum Window Trough Dust Lead Loading by Year of
Construction

Maximum Window Trough

Dust Lead Loading (µg/sq ft)a	Year of Construction

	1978-1998	1960-1977	1940-1959	Before 1940

	Number (000)	Percent

(%)b	Number (000)	Percent

(%)	Number (000)	Percent

(%)	Number (000)	Percent

(%)

LT LODc	Number HUsd	280	1%	94	0%	0	0%	0	0%

	Lower 95% CIe	0	0%	0	0%	0	0%	0	0%

	Upper 95% CI	681	2%	235	1%	0	0%	0	0%

GE LOD	Number HUs	20,969	70%	20,319	73%	16,406	80%	14,943	86%

	Lower 95% CI	17,718	60%	17,730	64%	14,638	73%	13,192	80%

	Upper 95% CI	24,221	81%	22,909	82%	18,174	86%	16,694	91%

GE 800	Number HUs	2,252	8%	3,788	14%	6,286	31%	8,883	51%

	Lower 95% CI	1,032	4%	2,504	9%	4,500	21%	7,084	41%

	Upper 95% CI	3,473	12%	5,072	18%	8,073	40%	10,683	61%

Missing	Number HUs	4,184	14%	5,885	21%	2,966	14%	2,322	13%

No troughs	Number HUs	4,341	15%	1,576	6%	1,191	6%	210	1%

	Lower 95% CI	2,380	8%	732	3%	368	2%	0	0%

	Upper 95% CI	6,301	21%	2,419	9%	2,014	10%	451	3%

a	Missing means that the trough was present, but that no lead value is
available (either the sample was not collected, e.g., due to
inaccessibility or respondent refusal, or the laboratory did not submit
a value).  “No trough present” means that there was no trough in the
HU, e.g., windows were flush with the wall, or awning windows were
installed.

b	All percentages are calculated with total housing units of that age as
the denominator.

c	LT equals “less than.”  GT equals “greater than or equal to.”
LOD equals “limit of detection.”

d	“Housing units” include permanently occupied, noninstitutional
housing units in which children are permitted to live

e	CI = 95% confidence interval for the estimated number or percent.

Tables 5.6a through 5.6c present floor, window sill, and window trough
dust loading distributions by household income level.  There are some
significant differences shown in the table.  A greater percent of homes
in the higher income level have lower lead dust loadings.  For example,
47 percent of higher income homes have lead dust loadings on floors
below the limit of detection, while only 29 percent of homes in the
lower income level have loadings below the limit of detection.  Only 1
percent of homes in the higher household income level exceed 100 (g/ft2
floor dust lead loading, while 5 percent of lower income level homes
exceed this loading.  Similar trends are observed for window sill and
trough dust loadings.

Although the data have not been presented, there were no apparent
differences between urbanization categories and dust lead loadings.

Table 5.6a	Maximum Floor Dust Lead Loadings by Household Income

g/sq ft)a	Household Income

	Less than $30,000/year	Equal to or above $30,000/year

	Number (000)	Percent (%)b	Number (000)	Percent (%)

LT LODc	Number HUsd	9,747	29%	26,508	47%

	Lower 95% CIe	6,944	23%	22,360	42%

	Upper 95% CI	12,550	35%	30,657	53%

GE LOD	Number HUs	24,038	71%	29,525	53%

	Lower 95% CI	19,932	65%	26,065	47%

	Upper 95% CI	28,143	77%	32,985	58%

GE 5	Number HUs	13,364	40%	13,215	24%

	Lower 95% CI	10,562	32%	10,859	20%

	Upper 95% CI	16,166	47%	15,571	27%

GE 10	Number HUs	8,276	25%	6,792	12%

	Lower 95% CI	6,219	19%	5,052	9%

	Upper 95% CI	10,332	30%	8,532	15%

GE 20	Number HUs	4,282	13%	4,135	7%

	Lower 95% CI	3,117	9%	2,782	5%

	Upper 95% CI	5,447	17%	5,488	10%

GE 40	Number HUs	2,819	8%	2,170	4%

	Lower 95% CI	1,710	5%	924	2%

	Upper 95% CI	3,927	12%	3,415	6%

GE 100	Number HUs	1,637	5%	435	1%

	Lower 95% CI	728	2%	57	0%

	Upper 95% CI	2,546	8%	813	1%

Missing	Number HUs	46	0%	77	0%

a	Missing means that the sill or trough was present, but that no lead
value is available (either the sample was not collected, e.g., due to
inaccessibility or respondent refusal, or the laboratory did not submit
a value).

b	All percentages are calculated with total housing units in that income
class as the denominator.

c	LT equals “less than.”  GT equals “greater than or equal to.” 
LOD equals “limit of detection.”

d	“Housing units” include permanently occupied, noninstitutional
housing units in which children are permitted to live.

e	CI = 95% confidence interval for the estimated number or percent.

g/sq ft)a	Household Income

	Less than $30,000/year	Equal to or Above $30,000/year

	Number (000)	Percentb	Number (000)	Percent

LT LODc	Number HUsd	1,448	4%	7,111	13%

	Lower 95% CIe	376	1%	5,291	9%

	Upper 95% CI	2,520	7%	8,931	16%

GE LOD	Number HUs	29,948	89%	47,597	85%

	Lower 95% CI	24,656	84%	42,665	81%

	Upper 95% CI	35,241	93%	52,528	89%

GE 125	Number HUs	10,322	31%	8,865	16%

	Lower 95% CI	7,909	26%	6,896	13%

	Upper 95% CI	12,735	36%	10,835	19%

GE 250	Number HUs	7,671	23%	4,772	9%

	Lower 95% CI	5,776	18%	3,611	7%

	Upper 95% CI	9,565	28%	5,933	11%

GE 500	Number HUs	4,395	13%	3,893	7%

	Lower 95% CI	2,943	9%	2,773	5%

	Upper 95% CI	5,846	17%	5,014	9%

Missing	Number HUs	1,137	3%	594	1%

No sills	Number HUs	1,297	4%	809	1%

	Lower 95% CI	250	1%	151	0%

	Upper 95% CI	2,345	7%	1,466	3%

a	Missing means that the sill or trough was present, but that no lead
value is available (either the sample was not collected, e.g., due to
inaccessibility or respondent refusal, or the laboratory did not submit
a value).  “No sill/trough present” means that there was no sill or
trough in the HU, e.g., windows were flush with the wall, or awning
windows were installed.

b	All percentages are calculated with total housing units in that income
class as the denominator.

c	LT equals “less than.”  GT equals “greater than or equal to.”
LOD equals “limit of detection.”

d	“Housing units” include permanently occupied, noninstitutional
housing units in which children are permitted to live.

e	CI = 95% confidence interval for the estimated number or percent.

Table 5.6c	Maximum Window Trough Lead Dust Loadings by Household Income

g/sq ft)a	Household Income

	Less than $30,000/year	Equal to or Above $30,000/year

	Number (000)	Percentb	Number (000)	Percent

LT LODc	Number HUsd	59	0%	315	1%

	Lower 95% CIe	0	0%	0	0%

	Upper 95% CI	180 	1%	723	1%

LT 800	Number HUs	25,985	77%	42,253	75%

	Lower 95% CI	21,486	70%	37,300	69%

	Upper 95% CI	30,485	84%	47,205	81%

GE 800	Number HUs	9,449	28%	10,623	19%

	Lower 95% CI	6,742	20%	8,067	15%

	Upper 95% CI	12,156	36%	13,179	23%

Missing	Number HUs	5,624	17%	8,788	16%

No troughs	Number of HUs	2,162	6%	4,755	9%

	Lower 95% CI	869	3%	3,009	5%

	Upper 95% CI	3,456	10%	6,501	12%

a	Missing means that the sill or trough was present, but that no lead
value is available (either the sample was not collected, e.g., due to
inaccessibility or respondent refusal, or the laboratory did not submit
a value).  “No sill/trough present” means that there was no sill or
trough in the HU, e.g., windows were flush with the wall, or awning
windows were installed.

b	All percentages are calculated with total housing units in that income
class as the denominator.

c	LT equals “less than.  “GE equals “greater than or equal to.”
LOD equals “limit of detection.”

d	“Housing units” include permanently occupied, noninstitutional
housing units in which children are permitted to live.

e	CI = 95% confidence interval for the estimated number or percent.

Figure 5.6 shows the distribution of the dust lead loadings by room
type: kitchen, common living area, bedroom, and other room.    For these
four room types, both floor dust and window dust have been combined. 
Dust loadings for the main entrance are also shown in the figure, but
these are only floor loadings since no data were collected from main
entrance windows.  Figure 5.7 shows the distribution of the dust lead
measurements by surface and carpet.  In both figures, the distributions
are extremely right-skewed.  Except for troughs, none of the boxes
extend above 40 µg/ft2, which means that the 75th percentile is less
than 40 µg/ft2.  However, there are dust lead loadings well above
10,000 µg/ft2.  On troughs, they extend above 100,000 µg/ft2.

Table 5.7 presents selected parameters of the distributions of dust lead
loadings by surface types, corresponding to the boxplots in Figure 5.7. 
Table 5.7 also presents geometric means and standard deviations.  The
distributions in Table 5.7 are all right-skewed, so that they are not
normally distributed.  A better model would be the lognormal
distribution.  Appendix C includes a discussion of distributional models
for these data.

Figure 5.6	Box Plots for Dust Lead Loadings by Room

a	The first four plots include floor and window loadings.  Main entrance
only includes floors since no data were collected from main entrance
windows.

b 	The box in the box plot represents the middle 50 percent of the data;
the bottom of the box gives the 25th percentile; the top of the box
gives the 75th percentile, and the horizontal line inside the box gives
the median or 50th percentile.  The vertical lines extending from the
top and bottom of the box reach to the largest and smallest
observations, respectively, except for outliers.

Figure 5.7	Box Plots for Dust Lead Loadings by Surface

a 	The box in the box plot represents the middle 50 percent of the data;
the bottom of the box gives the 25th percentile; the top of the box
gives the 75th percentile, and the horizontal line inside the box gives
the median or 50th percentile.  The vertical lines extending from the
top and bottom of the box reach to the largest and smallest
observations, respectively, except for outliers.

Table 5.7	Estimated Empirical Distribution Parameters of Dust Lead
Loadings by Surface Types

	Floors	Window Sills	Window Troughs

	(g/ft2	(g/ft2	(g/ft2

Arithmetic Mean	13.6	194.9	1,990.9

Arithmetic Standard Deviation	483.5	1682.7	12,086.5

Geometric Mean 	1.1	9.4	96.4

Geometric Standard Deviationa	3.8	9.3	14.4

25th Percentile	.375	2.0	18.0

Median	0.9	8.3	89.1

75th  Percentile	2.0	37.13	462.0

90th Percentile	6.0	172.8	2,824.2

95th Percentile	13.2	524.9	6,974.6

Number of Samples	3,894	2,302	1,607

a	The geometric standard deviation is computed as exp(s), where s is the
arithmetic standard deviation of the natural logarithms of the loadings
(see, e.g., Gilbert, R. O. (1987) Statistical Methods for Environmental
Pollution Monitoring, Van Nostrand Reinhold Company New York).

b	For this table, values below 0.375 were set equal to 0.375, which is
¼ of the limit of detection.

5.3	Association between Interior Dust Lead Hazards and Interior LBP
Condition

Figure 5.8 presents the prevalence of interior dust lead hazards in
relation to the condition of the interior LBP.  Table 5.8 presents
further details on these estimates.  Dust lead hazards are more likely
to exist in homes with deteriorated LBP.  An estimated 61 percent (+20%)
of homes with significantly deteriorated LBP have lead dust hazards,
while only 33 percent (+7%) of homes with LBP in good condition have
lead dust hazards.  Only 6 percent (+2%) of homes with no interior LBP
have lead dust hazards.  Although it appears from the data that the
presence of LBP, especially significantly deteriorated LBP, contributes
to higher dust lead hazard, there are additional sources of lead in the
environment to account for dust lead in homes with no lead-based paint.

Table 5.8 also allows one to compare the relative risks (95 percent
confidence intervals on that risk are not shown in the table) of
interior lead dust hazards associated with different paint conditions. 
The presence of significantly deteriorated LBP makes a house 1.8 (+0.5)
times as likely to have an interior lead dust hazard compared to a house
where the LBP is in good condition, and 10.0 (+1.9) times as likely as a
house without LBP.  Even a house with LBP in good condition is 5.4
(+0.8) times as likely to have interior lead dust hazards as one without
any LBP.

Paint

Table 5.8	Association Between Dust Lead Hazards and Presence and
Condition of Interior Lead-Based Paint

Interior Dust Lead Hazards	No LBP on Interior or Exterior	No Interior
LBP	Interior LBP in Good Condition	Significantly Deteriorated Interior
LBP

	Number (000)	Percent

(%)	Number (000)	Percent

(%)	Number (000)	Percent

(%)	Number (000)	Percent

(%)

None Present	Estimatea	55,105	95%	62,752	94%	15,244	67%	2,389	39%

	Lower 95% CIb	51,893	90%	60,141	90%	12,633	56%	1,565	26%

	Upper 95% CI	58,318	100%	65,363	98%	17,855	78%	3,213	53%

Some Present	Estimate	2,686	5%	4,068	6%	7,508	33%	3,727	61%

	Lower 95% CI	1,372	2%	2,584	4%	6,024	26%	2,505	41%

	Upper 95% CI	4,001	7%	5,552	8%	8,992	40%	4,949	81%

Total HUs

57,791	100%	66,820	100%	22,752	100%	6,116	100%

a	Estimate is either the number of permanently occupied,
noninstitutional housing units (000) in which children are permitted to
live, or the percentage of total housing units.

b	CI = 95% confidence interval for the estimated number or percent.

Tables 5.9 and 5.10 present data on the association between floor and
sill dust lead loadings and the presence of interior lead-based paint. 
Both tables show a positive relationship between the presence of
interior lead-based paint and dust lead levels.  Housing units with some
interior LBP are twice as likely to have floor dust, and three times as
likely to have window sill dust, above hazardous levels compared to
housing units without LBP.

Table 5.9	Association Between Floor Dust Lead Loadings and Presence of
Interior Lead-Based Paint

Maximum Floor Dust Lead 

Loading (g/sq ft)a	Interior Lead Based Paint

	None Present	Some Present

	Number (000)	Percent (%)b	Number (000)	Percent (%)

LT LODc	Number HUsd	32,873	34%	5,496	6%

	Lower 95% CIe	29,475	31%	3,529	4%

	Upper 95% CI	36,270	38%	7,463	8%

GE LOD	Number HUs	33,824	35%	23,372	24%

	Lower 95% CI	30,086	31%	19,997	21%

	Upper 95% CI	37,561	39%	26,748	28%

GE 5	Number HUs	12,242	13%	15,958	17%

	Lower 95% CI	9,835	10%	13,372	14%

	Upper 95% CI	14,650	15%	18,544	19%

GE 10	Number HUs	6,003	6%	9,961	10%

	Lower 95% CI	4,238	4%	7,615	8%

	Upper 95% CI	7,768	8%	12,307	13%

GE 20	Number HUs	2,493	3%	6,496	7%

	Lower 95% CI	1,606	2%	4,661	5%

	Upper 95% CI	3,381	4%	8,331	9%

GE 40	Number HUs	1,666	2%	3,830	4%

	Lower 95% CI	993	1%	2,378	3%

	Upper 95% CI	2,338	3%	5,281	6%

GE 100	Number HUs	1,122	1%	1,304	1%

	Lower 95% CI	419	0%	654	1%

	Upper 95% CI	1,825	2%	1,954	2%

Missing	Number HUs	123



	a	Missing means that the sill or trough was present, but that no lead
value is available (either the sample was not collected, e.g., due to
inaccessibility or respondent refusal, or the laboratory did not submit
a value).

b	All percentages are calculated with total housing units as the
denominator.

c	LT equals "less than." GE equals "greater than or equal to." LOD
equals "limit of detection."

d	"Housing units" include permanently occupied, noninstitutional housing
units in which children are permitted to live.

e	CI = 95% confidence interval for the estimated number or percent.

Table 5.10	Association Between Window Sill Dust Lead Loadings and
Presence of Interior Lead-Based Paint

Maximum Window Sill  Dust Lead 

Loading (g/sq ft)a	Interior Lead Based Paint

	None Present	Some Present

	Number (000)	Percent (%)b	Number (000)	Percent (%)

LT LODc	Number HUsd	8,728	9%	874	1%

	Lower 95% CIe	6,544	7%	187	0%

	Upper 95% CI	10,912	12%	1,562	2%

GE LOD	Number HUs	54,454	57%	27,680	29%

	Lower 95% CI	51,195	54%	24,515	26%

	Upper 95% CI	57,713	60%	30,845	32%

GE 125	Number HUs	6,553	7%	13,785	14%

	Lower 95% CI	5,043	5%	11,177	12%

	Upper 95% CI	8,063	8%	16,393	17%

GE 250	Number HUs	3,434	4%	10,005	11%

	Lower 95% CI	2,193	2%	8,096	9%

	Upper 95% CI	4,675	5%	11,914	13%

GE 500	Number HUs	1,498	2%	7,544	8%

	Lower 95% CI	651	1%	5,770	6%

	Upper 95% CI	2,346	3%	9,318	10%

No sill present in HUa	Number HUs	2,043	0%	178	0%

Missing	Number HUs	1,595

136

	a	Missing means that the sill was present, but that no lead value is
available (either the sample was not collected, e.g., due to
inaccessibility or respondent refusal, or the laboratory did not submit
a value).

b	All percentages are calculated with total housing units as the
denominator.

c	LT equals "less than." GE equals "greater than or equal to." LOD
equals "limit of detection."

d	"Housing units" include permanently occupied, noninstitutional housing
units in which children are permitted to live.

e	CI = 95% confidence interval for the estimated number or percent.

6.  RESIDENTIAL SOIL LEAD

Chapter 6 presents estimates of the prevalence of soil lead by lead
concentration, and the association between soil lead concentration and
exterior lead-based paint condition.  Relevant estimates are compared
with the findings of the 1990 LBP Survey.  The prevalence of soil lead
hazards in housing is presented in Chapter 3.  Since soil lead hazards
include lead in soil in children’s play areas, most estimates in this
Chapter are based on the subsample of 375 homes with data on the
presence or absence of children’s play areas in the yards and on the
extent of soil lead hazards in children’s play areas.  A few tables in
this chapter are independent of the presence or absence of
lead-contaminated soil in children’s play areas; such tables are
therefore based on the full sample and have a footnote indicating that
the basis is the full sample.

6.1	Prevalence of Residential Soil Lead Over All Sampled Locations

A composite soil sample was collected at each of five sites on the
property of each dwelling unit: 1) near the most commonly used entrance,
2) the dripline and 3) the mid-yard line of the wall with the main
entrance, and 4) the dripline and 5) mid-yard line of a second,
randomly-selected wall.  The main entrance sample was a composite sample
of two cores from the main entrance area.  The dripline and mid-yard
samples on each wall were composite samples from three locations along
the length of the sample site.  In addition, soil samples were collected
from children’s play areas for a subsample of homes.  At each of these
homes, up to four samples were collected from children’s play
equipment, when present; otherwise, one sample was collected from an
area of the yard identified as being where children play.  The tables in
this section and in Section 6.2 are based on the lead concentration data
from all of these soil samples, referenced as “all sampled
locations.”  Section 6.3 then presents data on soil lead
concentrations in children’s play areas, while Section 6.4 presents
corresponding data for the “rest of the yard,” i.e., areas not
identified as children’s play areas.

Figure 6.1 and Table 6.1 present the distribution of maximum bare soil
lead concentrations.  Figure 6.1 presents a histogram of the
distribution, while Table 6.1 presents the number and percentage of HUs
by selected soil lead concentration thresholds:  0, 20, 50, 200, 400,
1,200, 1,600, 2,000, and 5,000 ppm.  Figure 6.1 and Table 6.1 include
bare soil from all sampled locations, both play areas and all other
locations.  An estimated 58 percent ((8%) of homes have soil lead levels
above the limit of detection.  An estimated 10 percent ((4%) of homes
have soil lead levels above 400 ppm, and 7% ((4%) have soil lead levels
above 1,200 ppm.  Only 3 percent of homes were found to have soil lead
above 2,000 ppm. The maximum soil values for each HU have been used in
Table 6.1.  The effect of using the average soil lead for each HU would
drive the distribution towards the lower thresholds, i.e. more homes
have lower average soil lead concentrations.

Care is to be exercised in comparing the tables in this Chapter with
Table 3.3, which tabulate the prevalence of soil lead hazards.  The
tables in this chapter present the distribution of residential soil lead
concentrations at selected locations, while soil lead hazards are
defined in terms of soil lead concentrations in two types of locations. 
Specifically, a housing unit is defined to have a soil lead hazard if
soil lead concentrations exceed 400 ppm in play areas or exceed 1,200
ppm in the rest of the yard.  Thus, the tables in this section do not
directly compare with Table 3.3.

Figure 6.1	Distribution of Maximum Bare Soil Lead Concentrations

Table 6.1	Distribution of Maximum Bare Soil Sample Lead Concentrations
(All Sampled Locations)

Bare Soil Lead	Number of HUs (000)a	Percent of HUs (%)b

	

Estimate	Lower

95% CIc	Upper

95% CI	

Estimate	Lower

95% CI	Upper

95% CI

GEd0 ppm	77,888	72,054	83,722	81%	75%	88%

GE 20 ppm	55,114	47,348	62,881	58%	50%	66%

GE 50 ppm	40,023	31,365	48,680	42%	33%	51%

GE 200 ppm	15,299	11,626	18,971	16%	12%	20%

GE 400 ppm	9,996	6,398	13,594	10%	7%	14%

GE 1,200 ppm	6,271	2,733	9,809	7%	3%	10%

GE 1,600 ppm	3,900	1,670	6,131	4%	2%	6%

GE 2,000 ppm	3,124	827	5,420	3%	1%	6%

GE 5,000 ppm	1,580	0	3,309	2%	0%	4%

No Bare Soil	15,413	9,789	21,037	16%	10%	22%

No Soil	2,242	330	4,154	2%	0%	4%

Missinge	145	0	527	0%	0%	1%

a	“Housing units” include permanently occupied, noninstitutional
housing units in which children are permitted to live.

b	All percentages are calculated with total housing units (95,688) as
the denominator.

c	CI = 95% confidence interval for the estimated number or percent.

d	GE equals “greater than or equal to.” 

e	“Missing” means that soil was present, but that no lead value is
available (either the sample was not collected, e.g. due to
inaccessibility or respondent refusal, or the laboratory did not submit
a value).  “No soil” means that there was no soil on the property to
sample. 

Table 6.2 presents the number and percentage of housing units by
construction year for selected soil lead concentration thresholds for
bare soil only, again for all sampling locations. As seen above for all
soil, as the soil lead threshold increases, the number of newer homes
meeting the criteria for bare soil decreases faster than the older
homes.  In fact, practically no newer homes (1960-1998) have lead in
bare soil above 1,200 ppm, and less than 4 percent have soil lead above
400 ppm.  Thus, the data suggest that older homes have higher bare soil
lead levels than new homes.  Since the amount of lead added to
commercial residential paint declined from 1940 to 1980, these
observations are not unreasonable, and have been reported by others. 

Table 6.2	Distribution of Maximum Bare Soil Sample Lead Concentration
by Construction Year (All Sampled Locations)

Bare Soil Lead Concentration	Number of HUs (000)a	Percent of HUs (%)b

	Before 1940	1940 - 1959	1960 – 1977	1978 - 1998	Before 1940	1940 -
1959	1960 - 1977	1978 - 1998

GEc 0 ppm	12,015	16,843	23,185	25,845	69%	82%	83%	87%

GE 20 ppm	12,015	15,404	17,345	10,350	69%	75%	62%	35%

GE 50 ppm	11,193	12,789	10,437	5,603	64%	62%	37%	19%

GE 200 ppm	7,243	6,073	1,793	190	41%	30%	6%	1%

GE 400 ppm	5,148	3,736	1,111	0	30%	18%	4%	0%

GE 1,200 ppm	3,386	2,886	0	0	19%	14%	0%	0%

GE 1,600 ppm	2,006	1,894	0	0	12%	9%	0%	0%

GE 2,000 ppm	1,320	1,804	0	0	8%	9%	0%	0%

GE 5,000 ppm	1,106	475	0	0	6%	2%	0%	0%

Missingd	145	0	0	0	1%	0%	0%	0%

No Bare Soil	4,313	2,762	4,613	3,724	25%	13%	17%	13%

No Soil 	1,003	939	95	205	6%	5%	0%	1%

Total	17,476	20,544	27,893	29,774	100%	100%	100%	100%

a	“Housing units” include permanently occupied, noninstitutional
housing units in which children are permitted to live.

b	All percentages are calculated with total housing units of that age as
the common denominator.

c	GE equals “greater than or equal to.”

d	Missing means that soil was present, but that no lead value is
available (usually due to inaccessibility or respondent refusal).  “No
soil” means that there was no soil on the property to sample. 

Figure 6.2 presents a series of boxplots showing the distribution of
soil lead concentrations at each of the seven sample sites.  The last
two boxplots represent only the data collected during the supplemental
data collection from play areas.  As with paint lead and dust lead
loadings, soil lead concentrations are extremely skewed with over
three-fourths of the samples under 200 ppm, but some samples are well
above 1,000 ppm.

Table 6.3 presents selected parameters of the distributions of soil lead
concentrations by sample sites, corresponding to the first five boxplots
in Figure 6.2.  Table 6.3 also presents geometric means and standard
deviations.  As with the distributions of paint lead loadings and dust
lead loadings, the distribution of soil lead concentrations is
right-skewed.  Thus, a normal distribution would not be a suitable model
for the distribution.  A lognormal distribution would be a more suitable
distribution.  Appendix C includes a discussion of modeling these data.

Figure 6.2	Box Plots for Lead in Soil Samples by Sample Site

a 	The box in the box plot represents the middle 50 percent of the data;
the bottom of the box gives the 25th percentile; the top of the box
gives the 75th percentile, and the horizontal line inside the box gives
the median or 50th percentile.  The vertical lines extending from the
top and bottom of the box reach to the largest and smallest
observations, respectively, except for outliers.

b	“Wall 1”  refers to the main entrance wall and “Wall 2” refers
to the second, randomly-selected wall.

Table 6.3	Estimated Empirical Distribution Parameters of Soil Lead
Concentrations by Sample Site

	Main Entry	Wall 1 Dripline	Wall 2 Dripline	Wall 1 Midyard	Wall 2
Midyard

	ppm	ppm	ppm	ppm	ppm

Arithmetic Mean	234.8	242.9	404.1	87.3	123.4

Arithmetic Standard Deviation	1,094.4	817.6	1,612.7	194.7	360.4

Geometric Mean 	43.3	44.5	49.0	28.1	29.9

Geometric Standard Deviationa	5.6	5.8	6.8	4.3	4.9

25th Percentile	12.1	11.4	10.8	7.8	8.5

Median	40.2	38.8	40.3	27.0	29.1

75th  Percentile	133.4	130.7	165.4	76.3	74.2

90th Percentile	433.5	553.5	712.5	209.0	277.0

95th Percentile	1,005.8	1,110.8	1,444.5	411.3	538.8

Number of Samples	707	704	704	723	728

a	The geometric standard deviation is computed as exp(s), where s is the
arithmetic standard deviation of the natural logarithms of the
concentrations (see, e.g., Gilbert, R. O. (1987) Statistical Methods for
Environmental Pollution Monitoring, Van Nostrand Reinhold Company New
York).  For the calculations of the geometric mean and standard
deviation, zero and negative values were set to 5.

6.2	Association between Bare Soil Lead and Exterior Paint Condition

Table 6.4 and Figure 6.3 show the association between bare soil lead
concentration and the condition of the exterior LBP.  Higher bare soil
lead concentrations occur for homes with deteriorated LBP.  An estimated
56 percent ((8%) of homes with intact or minimally-deteriorated LBP have
bare soil lead above 20 ppm, while 73 percent ((19%) of  homes with
significantly deteriorated LBP have bare soil levels above 20 ppm.  Only
4 and 2 percent of homes free of significantly deteriorated LBP have
bare soil lead levels above 1,200 and 2,000 ppm, respectively, while 24
and 13 percent of homes with significantly deteriorated LBP have bare
soil lead levels above 1,200 and 2,000 ppm, respectively.

Table 6.4	Association Between Bare Soil Lead Concentration and Presence
of Significantly Deteriorated Exterior LBP (All Sampled Locations)

Bare Soil Lead	Housing Units without

Any Exterior LBPa,b	Housing Units without

Significantly Deteriorated Exterior LBPa,b	Housing Units with

Significantly Deteriorated Exterior LBPa,b

	

Percent	Lower

95% CIc	Upper

95% CI	

Percent	Lower

95% CI	Upper

95% CI	

Percent	Lower

95% CI	Upper

95% CI

GEd0 ppm	83%	78%	88%	83%	77%	88%	73%	55%	92%

GE 20 ppm	49%	41%	56%	56%	48%	63%	73%	54%	92%

GE 50 ppm	28%	20%	36%	38%	30%	47%	67%	51%	83%

GE 200 ppm	5%	1%	9%	13%	9%	17%	39%	19%	58%

GE 400 ppm	3%	0%	5%	8%	5%	11%	30%	11%	49%

GE 1,200 ppm	1%	0%	3%	4%	2%	7%	24%	7%	41%

GE 1,600 ppm	1%	0%	3%	2%	1%	4%	17%	4%	30%

GE 2,000 ppm	1%	0%	2%	2%	0%	4%	13%	2%	24%

GE 5,000 ppm	0%	0%	0%	1%	0%	2%	8%	0%	17%

Missinge	0%	0%	0%	30%	0%	0%	1%	0%	5%

No Bare Soil	14%	10%	19%	15%	11%	20%	22%	3%	41%

No Soil 	3%	0%	6%	2%	0%	4%	4%	0%	9%

Total 	100%

	100%

	100%



a	“Housing units” include permanently occupied, noninstitutional
housing units in which children are permitted to live.

b	The denominators for the percentages are 66,032,000 HUs without any
exterior LBP, 84,215,000 HUs without significantly deteriorated exterior
LBP, and 11,473,000 HUs with significantly deteriorated exterior LBP.

c	CI = 95% confidence interval for the estimated number or percent.

d	GE equals “greater than or equal to.” 

e	Missing means that soil was present, but that no lead value is
available (usually due to inaccessibility or respondent refusal).  “No
soil” means that there was no soil on the property to sample. Figure
6.3	Association Between Lead in Bare Soil and Deteriorated Exterior
Lead-Based Paint

6.3	Prevalence of Bare Soil Lead in Children’s Play Areas

Table 6.5 presents the number and percentage of housing units with bare
soil lead in children’s play areas above selected concentration
thresholds: 0, 20, 50, 200, 400, 1,200, 1,600, 2,000, and 5,000 ppm. 
Figure 6.4 presents the point estimates as a histogram.  An estimated 51
percent ((6%) of homes have bare soil lead levels in play areas above
the limit of detection. An estimated five percent ((3%) of homes with
play areas have soil lead levels above 400 ppm, while an estimated two
percent ((2%) of homes have play area soil lead levels above 2,000 ppm. 


Where more than one soil sample was collected from children’s play
areas at a home, the maximum soil value for housing unit has been used
in Table 6.5.  The effect of using the average soil lead for each
housing unit would drive the distribution towards the lower thresholds,
i.e. more homes would have lower soil lead concentrations.

Figure 6.4	Distribution of Maximum Bare Play Area Soil Lead
Concentration

Table 6.5	Distribution of Maximum Soil Lead Concentrations in
Children’s Play Areas

	Number of HUs (000)a	Percent of HUs (%)b

	Bare Play Area Soil Lead	

Estimate	Lower

95% CIc	Upper

95% CI	

Estimate	Lower

95% CI	Upper

95% CI	HUs in Sample









	GEd0 ppm	76,404	69,826	82,982	80%	73%	87%	294

GE 20 ppm	49,019	42,946	55,092	51%	45%	58%	209

GE 50 ppm	28,878	25,828	31,929	30%	27%	33%	127

GE 200 ppm	10,849	7,899	13,800	11%	8%	14%	101

GE 400 ppm	4,856	2,096	7,616	5%	2%	8%	84

GE 1,200 ppm	2,493	458	4,529	3%	1%	5%	82

GE 1,600 ppm	2,078	92	4,063	2%	0%	4%	80

GE 2,000 ppm	1,777	0	3,871	2%	0%	4%	77

GE 5,000 ppm	380	0	1,231	0%	0%	1%	1

No play area	12,368	6,659	18,077	13%	7%	19%	53

Missinge	6,916	1,862	11,969	7%	2%	13%	23

Total	95,688

	100%

	375

a	“Housing units” are permanently occupied, noninstitutional
residential units in which children are permitted to live.

b	All percentages are calculated with total housing units (95,688) as
the denominator.

c	CI = 95% confidence interval for the estimated number or percent.

d	GE equals “greater than or equal to.” 

e	Missing means that soil was present, but that no lead value is
available (usually due to inaccessibility or respondent refusal).  “No
soil” means that there was no soil on the property to sample.

Table 6.6 presents the number and percentage of housing units with bare
soil lead concentration in children’s play areas by selected
thresholds, by housing unit construction year.  In general, as the soil
lead threshold increases, the number of homes meeting the criteria
decreases as the housing unit age increases.  In fact, nearly all newer
homes (1960-1998) have bare play area soil lead below 400 ppm.  Thus,
the data suggest that older homes have higher bare play area soil lead
concentrations than new homes.  Since the amount of lead added to
commercial residential paint declined from 1940 to 1980, these
observations are not unreasonable, and have been reported by others. 

Table 6.6	Distribution of Maximum Soil Lead Concentrations in
Children’s Play Areas, by Construction Year

	Number of HUs (000)a	Percent of HUs (%)b

Bare Soil Lead Concentration	Before 1940	1940 – 1959	1960 – 1977
1978 – 1998	Before 1940	1940 – 1959	1960 - 1977	1978 – 1998

GEc0 ppm	14,641	15,953	22,536	23,275	84%	78%	81%	78%

GE 20 ppm	14,552	13,074	13,238	8,155	83%	64%	48%	27%

GE 50 ppm	12,562	8,920	4,733	2,664	72%	43%	17%	9%

GE 200 ppm	6,508	2,804	320	1,217	37%	14%	1%	4%

GE 400 ppm	3,325	1,469	62	0	19%	7%	0%	0%

GE 1,200 ppm	1,498	995	0	0	9%	5%	0%	0%

GE 1,600 ppm	1,082	995	0	0	6%	5%	0%	0%

GE 2,000 ppm	872	905	0	0	5%	4%	0%	0%

GE 5,000 ppm	380	0	0	0	2%	0%	0%	0%

No Play Area	613	102	3,410	2,790	13%	22%	7%	13%

Missingd	2,222	4,489	1,947	3,710	4%	1%	12%	9%

Total	17,476	20,544	27,893	29,774	100%	100%	100%	100%

a	“Housing units” are permanently occupied, noninstitutional housing
units in which children are permitted to live.

b	All percentages are calculated with total housing units of that age as
the common denominator.

c	GE equals “greater than or equal to.”

d	Missing means that soil was present, but that no lead value is
available (usually due to inaccessibility or respondent refusal).  “No
bare soil” means that there was no bare soil in children’s play
areas on the property to sample. 

Table 6.7 shows the association between bare play area soil lead
concentration and the condition of the exterior LBP.  Higher bare soil
lead concentrations occur in play areas for homes with significantly
deteriorated LBP.  An estimated 46 percent ((7%) of homes with intact or
minimally-deteriorated LBP have bare play area soil lead above 20 ppm,
while 89 percent ((11%) of homes with significantly deteriorated LBP
have bare soil levels above 20 ppm.  Only three percent ((3%) of homes
with intact or minimally-deteriorated LBP have bare play area soil lead
levels above 400 ppm, while 18 percent ((12%) of homes with
significantly deteriorated LBP have bare soil lead levels above 400 ppm.

Table 6.7	Association Between Bare Soil Lead Concentration and Presence
of Significantly Deteriorated Exterior LBP, in Children’s Play Areas

Bare Play Area Soil Lead	Housing Units without Significantly
Deteriorated Exterior LBP (%)a,b	Housing Units with Significantly
Deteriorated Exterior LBP (%)a,b

	

Estimate	Lower

95% CIc	Upper

95% CI	

Estimate	Lower

95% CI	Upper

95% CI

GEd0 ppm	79%	71%	86%	90%	79%	100%

GE 20 ppm	46%	39%	53%	89%	77%	100%

GE 50 ppm	24%	20%	29%	73%	55%	91%

GE 200 ppm	8%	4%	12%	35%	17%	53%

GE 400 ppm	3%	1%	6%	18%	6%	30%

GE 1,200 ppm	1%	0%	3%	13%	2%	24%

GE 1,600 ppm	1%	0%	3%	11%	0%	23%

GE 2,000 ppm	1%	0%	2%	10%	0%	23%

GE 5,000 ppm	0%	0%	0%	3%	0%	11%

No Play Areas	14%	7%	20%	7%	0%	17%

Missinge	8%	2%	14%	3%	0%	7%

a	“Housing units” include permanently occupied, noninstitutional
housing units in which children are permitted to live.

b	Percentages are calculated with the number of HUs with and without
significantly deteriorated LBP, 11,472 and 84,216, respectively, as the
denominators.

c	CI = 95% confidence interval for the estimated number or percent.

d	GE equals “greater than or equal to.” 

e	Missing means that soil was present, but that no lead value is
available (usually due to inaccessibility or respondent refusal).  “No
soil” means that there was no soil on the property to sample.

6.4	Prevalence of Soil Lead in the Rest of the Yard

Table 6.8 presents the number and percentage of housing units with soil
lead in the rest of the yard -- i.e., areas not identified as
children’s play areas -- above selected concentration thresholds: 0,
50, 200, 400, 1,200, 1,600, 2,000, and 5,000 ppm.  An estimated 75
percent ((6%) of homes have bare soil lead levels in the rest of the
yard above the limit of detection.  An estimated 10 percent ((4%) of
homes have soil lead levels above 1,200 ppm in the rest of the yard,
while an estimated 6 percent ((2%) of homes have soil lead levels above
2,000 ppm.  

Table 6.8	Distribution of Maximum Soil Lead Concentrations in the Rest
of the Yard

Soil Lead Concentration	Number of HUs (000)a	Percent of HUs (%)b	HUs in
Sample

	

Estimate	Lower

95% CIc	Upper

95% CI	

Estimate	Lower

95% CI	Upper

95% CI

	GEd 0 ppm	90,348	87,218	93,479	94%	91%	98%	323

GE 20 ppm	71,537	65,843	77,230	75%	69%	81%	258

GE 50 ppm	54,250	48,525	59,976	57%	51%	63%	162

GE 200 ppm	27,353	25,438	29,267	29%	27%	31%	116

GE 400 ppm	19,709	16,109	23,310	21%	17%	24%	70

GE 1,200 ppm	9,939	6,435	13,444	10%	7%	14%	53

GE 1,600 ppm	6,220	4,034	8,407	7%	4%	9%	49

GE 2,000 ppm	5,905	3,727	8,083	6%	4%	8%	38

GE 5,000 ppm	2,987	1,548	4,427	3%	2%	5%	17

No Soil/No Bare Soile	2,310	366	4,254	2%	0%	4%	9

Missinge	3,029	693	5,365	3%	1%	6%	12

Total	95,688

	100%

	375

a	“Housing units” are permanently occupied, noninstitutional
residential units in which children are permitted to live.

b	All percentages are calculated with total housing units (95,688) as
the denominator.

c	CI = 95% confidence interval for the estimated number or percent.

d	GE equals “greater than or equal to.” 

e	Missing means that soil was present, but that no lead value is
available (usually due to inaccessibility or respondent refusal).  “No
soil” means that there was no soil in the rest of the yard to sample. 
“No bare soil” means there was no bare soil in the rest of the yard.

Table 6.9 presents the number and percentage of housing units with soil
lead concentration in the rest of the yard by selected thresholds, and
by housing unit construction year.  In general, as the soil lead
threshold increases, the number of homes meeting the criteria decreases
as the housing unit age increases.  In fact, nearly all (91%) homes
built between 1960 and 1977 have bare soil lead concentrations below 400
ppm in the rest of the yard.

Table 6.9	Distribution of Maximum Soil Lead Concentrations in the Rest
of the Yard, by Construction Year

Bare Soil Lead Concentration	Number of HUs (000)a	Percent of HUs (%)b

	Before 1940	1940 - 1959	1960 - 1977	1978 - 1998	Before 1940	1940 –
1959	1960 - 1977	1978 – 1998

GEc0 ppm	15,858	19,372	26,950	28,168	91%	94%	97%	95%

GE 20 ppm	15,858	18,947	24,151	12,581	91%	92%	87%	42%

GE 50 ppm	15,350	17,492	13,509	7,900	88%	85%	48%	27%

GE 200 ppm	13,051	9,524	4,430	348	75%	46%	16%	1%

GE 400 ppm	11,255	5,960	2,410	84	64%	29%	9%	0%

GE 1,200 ppm	6,326	2,926	686	0	36%	14%	3%	0%

GE 1,600 ppm	4,245	1,289	686	0	24%	6%	3%	0%

GE 2,000 ppm	3,929	1,289	686	0	23%	6%	3%	0%

GE 5,000 ppm	1,891	865	231	0	11%	4%	1%	0%

No Soil/No Bare Soile	1,211	939	160	0	7%	5%	1%	0%

Missingd	407	233	783	1,606 	2%	1%	3%	5%

Total	17,476	20,544	27,893	29,774	100%	100%	100%	100%

a	“Housing units” are permanently occupied, noninstitutional housing
units in which children are permitted to live.

b	All percentages are calculated with total housing units of that age as
the common denominator.

c	GE equals “greater than or equal to.”

d	Missing means that soil was present, but that no lead value is
available (usually due to inaccessibility or respondent refusal). 

e	“No soil” means that there was no soil in the rest of the yard to
sample.  “No bare soil” means there was no bare soil in the rest of
the yard.

6.5	Comparison of Prevalence of Soil Lead to the 1990 LBP Survey

Table 6.10 compares the prevalence of soil lead found in the NSLAH with
the prevalence of soil lead found in the 1990 LBP Survey.  The estimate
of homes reported for the NSLAH have been limited to those with 500 ppm
soil lead or greater and to homes built before 1980 for comparability to
the 1990 LBP Survey protocols and findings (see Appendix B for a
comparison of the protocols for the two studies).  Statistical
comparison shows no significant difference in the prevalence of soil
above or below 500 ppm in the two studies.

Table 6.10	Comparison of the Prevalence of Lead-Contaminated Bare Soil
in the NSLAH and the 1990 LBP Survey

	1990 LBP Survey 

(pre-1980 HUs)	Current NSLAH

(pre-1980 HUs)

	Number

(000)	Percent

(%)	Number

(000)	Percent

and (CI) a (%)

HUs with Bare Soil Lead 

Above 500 ppm	15,699	20%

	15,909	23%

(19-27%)

HUs with Bare Soil Lead 

Equal to or Below 500 ppm	61,478	80%

	50,290	73%

(70-77%)

No Bare Soil	--	--	2,557	4%

Total	77,177	100%	68,756	100%

a CI = 95% confidence interval for the estimated percent.

7.  QUALITY OF THE NATIONAL SURVEY DATA

Chapter 7 examines the quality of the data and the resulting quality of
projected national estimates.  The greatest source of error (in the
statistical sense) in the NSLAH estimates is sampling error – as
discussed in Volume II, Chapter 2.  This chapter addresses two
additional important potential sources of error – nonresponse bias and
measurement bias – and discusses their effects on the national
estimates of the prevalence of lead-based paint (LBP), lead in dust, and
lead in soil.

The chapter concludes with a summary of the data collection quality
assurance activities, including results of telephone verification, field
team audits, field dust and soil quality control samples, laboratory
performance on dust and soil quality control samples, and paint testing
quality control.

7.1	Statistical Concepts and Terminology

There are two broad types of error in survey estimates: sampling error
and nonsampling error:

Sampling error:  Sampling error arises from surveying a random sample
rather than a complete census of all housing units (HUs).  It is a
function of the sample size and sample design.  Different samples of the
same size drawn using the same sample design will yield varying
estimates of the population parameters.  This variation about the true
population parameter is the sampling error.

Nonsampling error:  Nonsampling errors arise from a number of sources,
including differential response rates from different demographic groups,
types of HUs, and geographical areas; unknown differences between the
respondents and nonrespondents; differences between the sample frame and
the target population; some types of processing and data reduction
techniques; and classification bias due to measurement error inherent in
XRF and laboratory instrumentation and variation in a measured parameter
across a surface and among rooms. 

Throughout the report, the term weight has been used in conjunction with
the sampled HUs, rooms, and surfaces.  It is important that these terms
be understood.

Housing Unit (HU) weight:  The HU weight is the number of HUs in the
target population that a single HU in the survey represents.  The weight
is calculated by taking the inverse of the probability of selection for
that unit.  Thus, if the probability of selection is 0.01, the sample
weight is 100.  With multi-stage samples, the overall probability of
selection is the product of the conditional probabilities of selection
at each stage.  HU weights for this survey reflect nonresponse
adjustments and post-stratification to the 1997 American Housing Survey
(AHS) housing unit totals by Census region, HU age category, and
presence of children under age 18. 

Room weight:  The room weight is the number of rooms in the target
population that a single room in the survey represents.  Room weights
were determined by dividing the post-stratified HU weights by the
probability of room selection based on the inventory of all rooms in
each HU.  A nonresponse adjustment was then made to account for
noncompleted rooms. A room was only considered to be complete if some
environmental samples and data were collected in the room.   

Component weight:  The component weight is the number of components in
the target population that a single component tested in the survey
represents.  For most lead samples, the component weight equaled the
nonresponse-adjusted room weight.  There were two exceptions: 1) XRF
measurements on windows and doors, and 2) window dust samples.  For
these components, a sample of one door or window per room was selected. 
To complete component weights for these components, the
nonresponse-adjusted room weights were divided by the component
probability of selection, i.e., the inverse of the total number of doors
or windows in the room. 

7.2	Potential for Nonresponse Bias

The objective of the nonresponse analysis was to estimate the potential
impact of survey nonresponse on the estimated prevalence of LBP in
housing.  To accomplish this, three analyses were conducted.  First, the
weighted distribution of the NSLAH sample was compared and found to be
comparable with the AHS and CPS (this analysis was presented in Chapter
2).  Second, an analysis of completion rates was performed to look for
correlates with nonresponse. Third, the survey estimates for the
“hard-to-recruit” and HUs that initially refused (proxies for the
nonrespondents) were compared with estimates from HUs that were
relatively easy to recruit and had no history of refusal.

7.2.1	Analysis of Completion and Response Rates

An analysis of completion rates was conducted for the entire sample of
1,984 fielded HUs.  The analysis looked at the relationship between
completion rates and factors such as age of housing, race, ethnicity,
geographic location, income, tenure, and presence of multi-family
housing.  Given that these data were generally not available at the
individual HU level for the noncompletes, data for the block group to
which the HU belonged was used for the entire sample, with the exception
of HU age.  The housing unit age reported by respondents was used when
available, which was the case for approximately one-half of the HUs. 
The source of the block group data was the 1990 Census.  Completion
rates were also compared for HUs receiving an advance letter addressed
to "Current Resident" and HUs receiving a letter addressed to a
household member.  

The completion rates for the analysis were calculated as the weighted
proportion of the HUs sampled that completed both screener and data
collection, or else were found to be ineligible.  Ineligibles were
considered to be screener completes in the sense that their eligibility
status was determined during the screener.  For a large proportion of
the sample (39%), eligibility could not be determined, usually because
contact could not be made.  Most nonresponse occurred at the screener
stage and resulted in unknown eligibility status.  Of the 1,984 HUs
sampled, 831 completed both the screener and data collection and 229
were found to be ineligible.  Of the remaining 924 that did not complete
the data collection, 149 were eligible and another 775 were of unknown
eligibility.  Thus, there were 1,060 completes and 924 noncompletes
overall. The overall unweighted completion rate for the survey was 53.4
percent; the overall weighted completion rate was 53.1 percent.

Formulas for unweighted screener and data collection completion rates
are given below, along with the eligibility rate, refusal rate, and
overall response rate.   The overall completion rate is calculated as
the product of the completion rate at each stage. 

 

 



The overall response rate measures the response among eligible HUs.  The
number of eligible nonrespondents must be estimated because eligibility
cannot usually be established for households that don’t complete the
screener.  It is assumed that the eligibility rate among HUs whose
eligibility is unknown is the same as for HUs that did complete the
screener.  The eligibility rate among HUs that completed a screener was
81 percent.  The refusal rate is the rate of refusal among HUs where
contact was established.  (Numerous attempts were made to gain
cooperation.  Among respondents, two-thirds cooperated on the first or
second attempt.  While the average number of attempts to complete was
2.6, some housing units didn’t cooperate until the eleventh attempt.) 
HUs that were vacant or couldn't be located, that had no one at home,
that were in a locked, gated community where access couldn't be gained,
or that couldn't otherwise be contacted were subtracted from the total
sample size of 1,984 in the denominator.  The overall response rate for
the survey was 51.7 percent.

 

 

 

 



Overall completion rates by Census Division and block group
characteristics such as race/ethnicity, housing age, and type of housing
are presented in Table 7.1.  Mean percents for several characteristics
by completion status are given in Table 7.2.  For housing age, the
predominant building age category for the block group was assigned to
the HU, except when HU age was available from survey respondents.  It is
important to keep in mind that (with the exception of HU age) the
characteristics apply only to the block group in which the HU is
located, and may not apply to the HU itself.  The associations between
completion status and these characteristics were tested using chi-square
and t-tests that take into account the HU weights and the survey design.
 

Significant associations were found between completion status and
percents Hispanic and African American, percent below the federal
poverty level, housing age, and tenure (see Table 7.2).  Different
completion rates for groups within a category cannot always be
explained, but could be due to factors such as the use of incentives,
attitudes of householders, etc.  In Table 7.1, the highest response
rates were found among HUs in block groups with 30 percent or more
Hispanic population, 30 percent or more in poverty, and in newer housing
(1978 or later).  The differences in response rates are substantial for
these characteristics, ranging from 9 to 14 percentage points or more
(possibly a function of the $200 incentive having greater value for
these households).  This means that if the lead outcome variables are
correlated with any of these characteristics (such as housing age), a
greater potential for nonresponse bias would exist had these
characteristics not been used in adjusting for nonresponse at the
screener and data collection stages.  Nonresponse adjustment factors
were calculated within cells defined by the block group percent of
low-income population, percent of Hispanic or African American
population, and percent of pre-1940 and pre-1960 housing.  In addition,
the nonresponse-adjusted HU weights were poststratified to 1997 AHS
housing unit totals by Census region, housing unit age, and presence of
a child under 18.  Use of the final adjusted weights in all analyses
therefore greatly reduces this potential bias in estimates for the
national housing stock. 

The presence or absence of a household member name on the advance letter
was also significantly correlated with completion rates.  For HUs
located in higher income block groups (i.e. fewer than 30 percent of the
households are below the federal poverty level), the response rate for
HUs with a name on the advance letter was significantly lower (p = .01)
than for HUs receiving a letter addressed to "Current Resident" (48% vs.
56%).  For HUs located in lower income block groups (i.e. more than 30
percent of the households are below the federal poverty level), however,
the completion rates were not significantly different.  

Response rates for the play areas sample were also calculated using the
above formula.  Weighted and unweighted response rates were examined in
the same manner, but given the smaller sample size the characteristics
were restricted to Census Region, building age category, and soil lead. 
The national response rate for the play areas survey was 79.5 percent
weighted and 78.0 percent unweighted (higher than the national survey
since we were returning to previously willing HUs).  For each subset,
the response rate remained between 71 and 85 percent. None of the
response rate differences observed for the play areas survey were
statistically significant at the alpha = .10 level.

Table 7.1	Completion Rates by 1990 Census Block Group Characteristics



Unweighted	Weighted	P-Value for 

	1990 Census

Block Group Characteristic	Number of HUs	Completion

Rate	Completion

Rate	Chi-square testa

	Census Division





	New England	102	52%	53%	0.42

	Middle Atlantic	289	53%	49%



East North Central	289	49%	46%



West North Central	181	55%	55%



South Atlantic	326	52%	53%



East South Central	144	53%	54%



West South Central	208	56%	58%



Mountain	136	57%	58%



Pacific	309	56%	57%



% Hispanic Population





	Less than 30% Hispanic	1832	52%	52%	0.005

	30% or more Hispanic	152	65%	65%



% African American Population





	Less than 30% African American	1731	52%	53%	0.08

	30% or more African American	253	60%	59%



% Population in Poverty





	Less than 30% in Poverty	1796	53%	52%	0.04

	30% or more in Poverty	188	60%	61%



% Multi-family Housing Units (5+)





	Less than 30% Multi-family HUs	1522	53%	53%	0.42

	30% or more Multi-family HUs	462	56%	55%



Age of Housing Unitb





	Pre-1940	456	48%	45%	<0.001

	1940 – 1959	363	59%	60%



1960 – 1977	806	47%	45%



1978 and later	359	70%	74%



a	Chi-square test between completion rate and housing characteristic -
takes into account the HU weights and sample design.

b	Reported HU age used when available for HUs completing screener. 

Table 7.2	Mean Percents for Completed and Noncompleted Housing Units by
1990 Census Block Group Characteristics 

1990 Census Block Group Characteristic	Mean of Characteristic	P-value
for t-test between completes and noncompletesa

	Completes	Noncompletes

	Percent HUs Pre-1940b	18%	25%	0.001

	Percent Population below Poverty Level	12%	11%	0.080

	Percent Multi-Family HUs (5+)	17%	16%	0.535

	Percent Multi-Family HUs (20+)	7%	7%	0.908

	Percent HUs Owned	63%	66%	0.042

	Percent Population: African American	11%	8%	0.006

	Percent Population: Hispanic	8%	5%	0.005

	a	The test takes into account the HU weights and sample design.

b	HU age reported by respondents was used when available for HUs
completing screener. 

7.2.2	Comparison of “Hard-to-Recruit” Versus “Easy-to-Recruit”
HUs

Lead measurements were not available for the HUs whose occupants refused
to participate in the survey or who could not be contacted, so it is not
possible to know how their participation would have changed the
estimates of lead prevalence in housing.  However, if the HUs that
initially refused but later cooperated (or those HUs requiring several
attempts to complete the screener) are similar to the survey
nonrespondents, they may be considered as a proxy group for the
nonrespondents.  If they are significantly different, this may indicate
the likely direction of the nonresponse bias in the lead prevalence
estimates.  

The initial refusals were compared with HUs with a history of no
refusal.  HUs requiring 4 or more attempts to complete the screener were
also compared with those requiring 3 or fewer attempts.  The number of
attempts was split at 3-4 for two reasons. Two-thirds of respondents
cooperated on the first or second attempt and the average number of
attempts to complete was 2.6.  The comparisons were made for four key
statistics and by the housing characteristics recorded by the
interviewer.  The statistics were 1) the presence of LBP anywhere in the
home, 2) the presence of deteriorated LBP anywhere in the home, 3)
presence of LBP hazard anywhere in the home, and 4) the presence of a
soil lead hazard.  The housing characteristics were the respondent's
race/ethnicity, household income, tenure, building age, and presence of
a child under 18.  Comparison of whether or not the household had
initially refused (or required more than three attempts to obtain a
completed screener) and these characteristics is presented in Table 7.3.
 

Table 7.3	Comparison of “Easy-to-Recruit” Respondents Versus
“Hard-to-Recruit” Respondents 

National Survey Estimates	Initial Refusal

(%)	p-valuea	More than 3 Attempts to Complete Screener (%)	p-valuea

Presence of LBP anywhere in HU





Yes	9.2	0.60	17.3	0.67

No	10.0

15.1

	Presence of Significantly Deteriorated LBP  anywhere 





Yes	6.0	0.05	12.8	0.29

No	10.4

16.7

	Presence of Significant LBP Hazard in HU





Yes	5.6	0.0007	15.8	0.84

No	11.6

16.5

	Presence of Soil-Lead Hazard





Yes	8.7	0.85	6.2	0.29

No	9.9

16.7

	Ethnicity





Hispanic	10.7	0.78	9.6	0.14

Non-Hispanic	9.7

16.8

	Race





White	10.2	0.38	16.9	0.12

African American	6.7

11.7

	Asian, Pacific Islander, Hawaiian, 	8.9

14.4

	American Indian, Other





Presence of Child under 18





Yes	9.9	0.85	15.1	0.31

No	9.5

16.6

	Year of Construction





Pre-1940	7.2	0.22	18.6	0.18

1940 – 1959	6.1

11.5

	1960 – 1977	12.1

13.3

	1978 or later	11.4

20.2

	Tenure (moved up to be w/hsg info)





Owned	11.2	0.03	15.8	0.15

Rented	6.3

16.8

	Type of Housing





Single-family	10.3	0.02	15.4	0.58

Multi-family	5.8

19.9

	Household Income





Less than $30,000	4.7	<0.0001	12.3	0.03

$30,000 or More	11.7

17.8

	Household Income





Less than $20,000	4.7	0.05	11.1	0.02

$20,000 - $39,999	6.8

15.6

	$40,000 - $59,999	10.9

16.8

	$60,000  and over	13.5

18.9

	Poverty 





Below Poverty Level	4.6	0.003	9.9	0.005

At or Above Poverty Level	9.8

17.0

	a	Chi-square test of association between Ease of Recruitment and
Reported Housing Characteristics.

Table 7.3 shows that poorer households were much less likely to have an
initial refusal, and required fewer attempts to obtain a completed
screener.  This is consistent with earlier analysis showing higher
response rates for lower income households.  It could be that the
monetary incentive was more effective among lower income households in
improving response rates, or that higher income HUs were more difficult
to contact.

Table 7.3 also shows that households with LBP, significantly
deteriorated LBP, significant LBP hazard, or soil lead hazard were just
as likely to require more than three attempts to complete the screener
as those without these lead characteristics.  While households with LBP
or soil hazards were just as likely to initially refuse as other
households, those with significantly deteriorated LBP or LBP hazards
were significantly more likely to initially refuse than those without
these characteristics.  This could be due to the fact that more lower
income homes participated in the survey and lower income homes are more
likely to have these characteristics. 

The nonresponse analysis shows that the households that responded were
more likely to be located in densely Hispanic and low income areas, and
to be renters as opposed to home owners.  They are also more likely to
live in newer housing (post-1977). 

If the initial refusals who agreed to cooperate are representative of
nonrespondents, there would be a potential for bias in unweighted
estimated prevalence of HUs with deteriorated LBP or LBP hazards.  This
means that there would be a potential for bias in the estimated
prevalence of HUs with an LBP hazard.  However, this was partially
corrected by using race/ethnicity, low-income indicators, and building
age in making nonresponse adjustments to the HU weights.  The weighting
adjustments do not eliminate nonresponse bias completely, but they do
reduce it when variables that are correlated with both the response
rates and propensity to have a lead hazard are used in the nonresponse
adjustments.  Comparison of the nonresponse-adjusted and poststratified
weighted distribution of housing from the National Survey with the AHS
and CPS in Table 2.1 show that the weighted National Survey sample
matches the national housing distribution closely.  These results
suggest that there is probably not a serious nonresponse bias in the
weighted estimates of lead hazard prevalence.

Completion Rates in the Play Area Subsample

In the play area subsample, 40 of the 75 PSU’s were selected and all
previously completed homes were selected for revisiting.  There were 481
homes in the 40 PSU’s.  Table 7.4 displays the yield in this
subsample.  In the play area subsample, 78 percent of the homes were
completed, either with the collection of a play area soil sample, or the
determination that one was not necessary.

Table 7.4.	Completions in Play Area Subsample

Result	Number of HUs	Percent of HUs

Complete, no soil sample needed	178	37%

Complete, new soil sample collected	197	41%

Total completes	375	78%

No contact made with respondent	54	11%

Refusal	12	2%

Demolished	1	0%

Vacant	27	6%

Other	12	2%

Total Housing Units	481	100%



7.3	Quality of Field Data Collection and Analysis

Quality assurance was integrated into all components of the study,
including a defensible study design, experienced project personnel,
utilization of well-planned, detailed and tested protocols for all
aspects of data collection, thorough study-specific training of
experienced field staff, electronic sample and data management, and
ongoing communication between individuals responsible for each stage of
the study.  These procedures are described in detail in Volume II,
Chapter 6.  

Four types of replicate sampling were conducted to estimate measurement
error:  replicate XRF testing of one random component per room,
replicate dust sampling of one surface per home, replicate soil sampling
at a different sample site at every third home, and replicate room
sampling at a subset of homes.  The analyses utilizing these replicate
data and resultant measurement error estimates are presented in Appendix
C.

This section summarizes the results of the various activities focussed
at ensuring quality of the field data collection and laboratory analysis
of the environmental samples. 

7.3.1	Field Data Collection

A number of procedures were instituted to ensure quality of the field
data collection, including a manual edit of all data and samples by the
field team, review by the Field Supervisor upon return of the data to
Westat headquarters, and reconciliation of any errors with the field
team prior to submission of any samples to the laboratory.  In addition,
random telephone verification and field team audits were conducted; dust
sample material screens were analyzed; and dust blanks and spike samples
and blind soil reference samples were included in the sample stream.

	Telephone Verification of Data Collection

The Field Director contacted a random subsample of 82 (10%) households
by telephone to verify the team’s activities and conduct and to
validate selected information from the data forms.  No field problems
were identified by this process.

In addition to the random verification process, a number of respondents
and potential respondents utilized the toll-free phone number, or the
HUD phone number, to ask questions, verify the survey, and express
concerns.  All questions or concerns were answered or addressed by the
Field Director or HUD.

	Random Field Audits

The QA Officer or designee, and HUD and NIEHS representatives, conducted
random field audits at 31 households to verify that the protocols were
followed and data collection was accurate and complete.  In addition to
the field audits, the QA Officer conducted 17 telephone audits to
ascertain the team members’ understanding of the protocols, especially
when more than two months had elapsed between assignments.  Problems
noted during these audits were corrected directly with the individual
team members.  In addition, the results of audits were immediately
relayed to the Field Office.  As appropriate, all field staff were
notified by memo of any issues identified with the protocols.  

	Lead Dust Wipe Sample Collection

Lead Dust Wipe Materials Screens

The purpose of a materials screen was to verify that the various
sampling supplies to be used in the field did not have lead
contamination.  Two screens were prepared and analyzed for every lot of
wipe materials and sample tubes before being used in the study.  The
analyses showed that all material screens had below 1.5 µg lead.

Field Blank Wipes

One field blank wipe was prepared for each HU at a specified random
sample location where another wipe sample was collected.  All field
blanks were below 50 µg lead/wipe, as specified in HUD Guidelines. 
Most field blanks (98+% of all field blanks) had lead levels below the
detection limit for the analytical run (approximately 3.5 µg
lead/wipe).  Of 15 field blanks with lead values above 3.5 µg
lead/wipe, only one blank had more than 20 µg lead/wipe; this wipe had
43 µg lead and was collected in a home with elevated window sill and
trough lead dust levels.  Data from this home were used in the survey
since the blank was still below the 50 µg guideline.

Reference (Spike) Sample Dust Wipes

Reference wipe samples were made in advance of the fieldwork by placing
a known quantity of National Institute of Standards and Technology’s
Standard Reference Material (NIST SRM) 1579a on the same wipe material
used in the study.  The reference wipes were labeled like a regular
sample so that the laboratory was blinded to fact that these were
quality control samples.  The Field Office inserted one reference wipe
sample with each group of 50 samples before sending samples to the
laboratory.  A total of 206 reference wipes, ranging from 21 to 516 µg
lead/wipe, were submitted over the course of the study.  The average
reference sample recovery was 96% (range from 83% to 115%) with a
standard deviation of + 5.25 percent.  With a few exceptions, all sample
recoveries were within the HUD Guidelines acceptable range of 80 to 120
percent.  The laboratory was requested to re-analyze those batches with
values outside acceptable limits.  Recoveries were acceptable on the
second run (see Section 6.7 of Volume II) and the second set of data was
used for the entire batch of samples.

	Soil Quality Control Samples

Reference Soil Samples

Reference soil samples were purchased before the fieldwork began.  The
Field Office labeled and included one reference sample with each group
of samples from every three households (approximately one sample in
ten).  A total of 83 soil reference samples were submitted to the
laboratory.  The average recovery was 104% (range 84% to 121%) with a
standard deviation of + 8.04 percent.  Two analytical batches contained
a reference sample with a recovery of 121%, exceeding the control limits
of 80 to 100 percent.  The laboratory was requested to re-analyze these
batches.  Recoveries were acceptable on the second run (see Section 6.9,
Volume II) and the second set of data was used for the entire batch of
samples.

Two types of reference soil samples were used:  urban soil provided by
University of Cincinnati (640, 3,132, and 6,090 ppm lead), and NIST SRM
2709 (San Joacquin soil, 18.9 ppm lead) and SRM 2711 (Montana soil,
1,162 ppm lead).  It is interesting, but not unexpected, that the
average recovery of 89% for the NIST Montana soil was lower than the
average of 105% for the urban soils.  Lead in urban soils tends to be
from more leachable sources (i.e. paint, past automobile emissions,
industrial facilities).

7.3.2	Laboratory Quality Control Samples 

Each laboratory provided quality assurance procedures during the
selection and qualification process.  These approved procedures
(outlined in the National Survey’s Protocol and Sample Design Report,
June, 1999) were adhered to for all study samples.  In general, the
laboratories performed instrumental and duplicate quality control
analyses, as required by ASTM E 1613-94 and the American Industrial
Hygiene Association’s Environmental Lead Laboratory Accreditation
Program (ELLAP) Quality Manual and Policies, to ensure that the original
calibration solutions were accurate, the instruments were properly
zeroed, instrumental drift was not excessive, and carryover between
samples did not occur.  These included duplicate injections of the same
sample, method blanks, and spiked samples at a minimum frequency of five
percent of the samples.

7.3.3	Laboratory Selection Quality Assurance

The laboratories used for analysis of dust and soil samples,
respectively, were recognized by the EPA under its National Lead
Laboratory Accreditation Program (NLLAP) for those analyses throughout
the laboratory qualification and performance phases of the National
Survey.  This recognition provided assurance of the quality of
laboratory performance of lead analyses and reporting.  In addition, the
laboratories were accredited by the American Industrial Hygiene
Association; this accreditation provided a separate assurance of the
quality of laboratory management and performance of environmental
analyses and reporting.

7.4	Paint Testing Quality Assurance

Calibration of the XRF analyzer was performed before and after testing
in every home.  In no case was the instrument used if the calibration
criteria were not met, i.e. the analyzer read 0.0 mg/m2 on the 0.0 film
and between 0.9 and 1.2 mg/m2 on both the front and back of the 1.0
film.  In addition, the average of three readings on the front of the
1.0 film was between 0.9 and 1.2 mg/m2.



APPENDIX A

Selected Additional Tables and Figures

Figure A.1	Prevalence of Significant LBP Hazards, Soil Lead Threshold
at 2,000 ppm and at 1,200 ppm

 

Table A.1	Prevalence of Housing Units with Significant Lead-Based Paint
(LBP) Hazards, by Selected Characteristics, for Alternative Soil Lead
Threshold of 2000 ppm

HUD Lead Safe Housing Rule:  Significant LBP Hazardsa

Characteristic	All HUs

(000)b	No. of HUs with Significant LBP Hazards (000)	Percent of HUs with
Significant LBP Hazards (%)c	HUs in Sample



Estimate	Lower

95% CId	Upper

95% CI	Estimate	Lower

95% CI	Upper

95% CI

	Total Occupied HUs 	95,688	25,517	20,410	30,623	27%	21%	32%	375

Region:









Northeast	19,290	8,260	5,904	10,616	43%	31%	55%	95

Midwest	22,083	7,606	5,691	9,521	34%	26%	43%	102

South	35,474	6,082	3,161	9,003	17%	9%	25%	111

West	18,841	3,569	1,003	6,135	19%	5%	33%	67

Construction Year:









1978-1998	29,774	425	0	1,222	1%	0%	4%	88

1960-1977	27,874	2,843	317	5,370	10%	1%	19%	111

1940-1959	20,564	10,501	7,996	13,006	51%	39%	63%	97

Before 1940	17,476	11,747	8,771	14,723	67%	50%	84%	79

Urbanization









MSA >= 2 million population	26,814	     6,793 	     4,978 	8,609	25%	19%
32%	276

MSA =< 2 million population	45,753	    10,232 	     8,171 	12,293	22%
18%	27%	417

Non-MSA	23,121	     7,001 	     3,848 	10,153	30%	17%	44%	138

One or More Children Under Age 6:









All HU ages	16,402	5,652	3,758	7,546	34%	23%	46%	83

HUs built 1978-1998	5,847	364	0	1,127	6%	0%	19%	25

HUs built 1960-1977	5,098	371	0	987	7%	0%	19%	20

HUs built 1940-1959	3,055	2,662	1,554	3,770	87%	51%	123%	22

HUs built before 1940	2,401	2,255	502	4,007	94%	21%	167%	16

Housing Unit Type:









Single family	82,651	23,204	17,794	28,614	28%	22%	35%	319

Multi-family	13,037	2,313	0	5,143	18%	0%	39%	56

Occupant Status:









Owner-occupied	62,232	16,013 	12,647	19,380	26%	20%	31%	254

Renter-occupied	29,074	9,503 	5,906	13,100	33%	20%	45%	119

Refusal/Don’t Knowe	381





	2

Household Income:









Less than $30,000/year	33,830	13,998	8,938	19,057	41%	26%	56%	145

Equal to or more than

  $30,000/year	56,111	10,060	7,151	12,970	18%	13%	23%	211

Refusal/Don’t Know	5,747





	19

One or More Children Under Age 6:









All Income Categories	16,402	5,652	3,758	7,546	34%	23%	46%	83

Less than $30,000/year	4,791	1,646	0	3,460	34%	0%	72%	28

Equal to or more than

  $30,000/year	11,236	4,006	1,797	6,214	36%	16%	55%	52

Refusal/Don’t Know	375





	3

Table A.1	Prevalence of Housing Units with Significant Lead-Based Paint
(LBP) Hazards, by Selected Characteristics, for Alternative Soil Lead
Threshold of 2000 ppm (continued)

HUD Lead Safe Housing Rule:  Significant LBP Hazardsa

Characteristic	All HUs

(000)b	No. of HUs with Significant LBP Hazards (000)	Percent of HUs with
Significant LBP Hazards (%)c	HUs in Sample



Estimate	Lower

95% CId	Upper

95% CI	Estimate	Lower

95% CI	Upper

95% CI

	Government Support:









Government support	4,809	822	10 	1,634	17%	0%	34%	25

No government support	86,070	23,571	19,114	28,028	27%	22%	33%	327

Refusal/Don’t Know	4,809





	23

Poverty:









In Poverty	13,221	5,053	2,542	7,565	38%	19%	57%	54

Not in Poverty	76,336	18,669	14,396	22,942	24%	19%	30%	300

Refusal/Don’t Know	6,130





	21

Race:









White	77,005	19,164	14,972	23,356	25%	19%	30%	285

African American	10,365	2,317	665	3,969	22%	6%	38%	45

Otherf	6,571	2,631	0	5,734	40%	0%	87%	35

Refusal/Don’t Know	1,746





	10

Ethnicity:









Hispanic/Latino	7,434	3,635	700	6,569	49%	9%	88%	31

Not Hispanic/Latino	87,008	20,841	16,734	24,948	24%	19%	29%	337

Refusal/Don’t Know	1,246





	7

a	Significant LBP hazard as defined in text and HUD Lead Safe Housing
Rule.

b	“Housing units” include permanently occupied, noninstitutional
housing units in which children are permitted to live.

c	All percentages are calculated with the “All HUs” column in each
row used as the denominator.

d	CI = 95% confidence interval for the estimated number or percent.

e	Refusals and “don’t know” responses by survey respondents.

f	“Other” race includes Asian, American Indian or Alaskan Native,
Native Hawaiian or other Pacific Islander, and more than one race.

Table A.2	Prevalence of Significant Lead-Based Paint (LBP) Hazards by
Location in the Building, for Alternative Soil Lead Threshold of 2000
ppm

HUD Lead Safe Housing Rule:  Significant LBP Hazards

LBP Hazard Location	Number of HUsa (000)	Percent of HUsb	HUs in Sample 

	Estimate	Lower

95% CIc	Upper

95% CI	Percent	Lower

95% CI	Upper

95% CI

	Interior only	10,861	6,865	14,857	11%	7%	16%	47

Both Interior and Exterior	7,965	4,891	11,040	8%	5%	12%	39

Exterior only	6,690	3,635	9,745	7%	4%	10%	29

Anywhere	25,517	20,440	30,594	27%	21%	32%	115

No Significant LBP Hazard	70,171	65,094	75,248	73%	68%	79%	260

Total HUs	95,688

	100%

	375

a	“Housing units” include permanently occupied, noninstitutional
housing units in which children are permitted to live.

b	All percentages are calculated with total housing units (95,688) as
the denominator.  Percentages may not total 100% due to rounding.

c	CI = 95% confidence interval for the estimated number or percent.

Table A.3	Prevalence of Significant Lead-Based Paint (LBP) Hazards in
Housing Units with a Child Under 6 Years of Age by Type of Hazard, for
Alternative Soil Lead Threshold of 2000 ppm

HUD Lead Safe Housing Rule: Significant LBP Hazards

Type of Hazard	Number of HUsa (000)	Percent of HUsb (%)

	Estimate	Lower

95% CIc	Upper

95% CI	Estimate	Lower

95% CI	Upper

95% CI

Significantly Deteriorated Lead Based Paint







All HUs	14,124	10,666	17,582	15%	11%	18%

HUs w/ Child Under 6	3,521	1,576	5,467	21%	10%	33%

Interior Lead Dust







All HUs	16,794	12,169	21,420	18%	13%	22%

HUs w/ Child Under 6	3,637	1,636	5,638	22%	10%	34%

Lead Contaminated Soil







All HUs	5,572	2,487	8,657	6%	3%	9%

HUs w/ Child Under 6	1,419	0	3,057	9%	0%	19%

Any LBP Hazard







All HUs	25,517	20,410	30,623	27%	21%	32%

HUs w/ Child Under 6	5,652	3,758	7,546	34%	23%	46%

a	“Housing units” include permanently occupied, noninstitutional
housing units in which children are permitted to live.

b	Percentages are calculated with total housing units (95,688) or with
housing units with a child under age 6 (19,577) as the denominator, or
as applicable.

c	CI = 95% confidence interval for the estimated number or percent.

Table A.4a	Distribution of Single Family Housing Units (SFHUs) with
Deteriorated or Significantly Deteriorated Paint, by Construction Year

Deteriorated Paint

Construction Year	Total SFHUs (000)a	Number of SFHUs with

Deteriorated Paint (000)	Percent of SFHUs with

Deteriorated Paint (%)b



Estimate	Lower 95% CIc	Upper 95% CI	Estimate	Lower 95% CI	Upper 95% CI

1978-1998	26,349	5,231	2,628	7,835	20%	10%	30%

1960-1977	22,657	7,992	5,573	10,410	35%	25%	46%

1940-1959	18,497	11,987	9,643	14,331	65%	52%	77%

Before 1940	15,148	11,518	10,049	12,986	76%	66%	86%

Total SFHUs	82,651	36,728	27,893	45,562	44%	34%	55%

Significantly Deteriorated Paint



Construction Year	Total SFHUs (000)	No. of SFHUs with Significantly
Deteriorated Paint (000)	Percent of SFHUs with Significantly
Deteriorated Paint (%)



Estimate	Lower 95% CI	Upper 95% CI	Estimate	Lower 95% CI	Upper 95% CI

1978-1998	26,349	3,403	1,629	5,176	13%	6%	20%

1960-1977	22,657	6,606	4,291	8,922	29%	19%	39%

1940-1959	18,497	9,426	7,240	11,611	51%	39%	63%

Before 1940	15,148	9,365	7,527	11,203	62%	50%	74%

Total SFHUs	82,651	28,800	20,686	36,913	35%	25%	45%

a	“Single family housing units” include permanently occupied,
noninstitutional single family housing units in which children are
permitted to live.

b	Percentages are calculated with total housing units (95,688) or with
housing units with a child under age 6 (19,577) as the denominator, or
as applicable.

c	CI = 95% confidence interval for the estimated number or percent.



Table A.4b	Distribution of Single Family Housing Units (SFHUs) with
Deteriorated or Significantly Deteriorated Unleaded Paint, by
Construction Year

Deteriorated Unleaded Paint

Construction Year	Total SFHUs (000)a	Number of SFHUs with

Deteriorated Unleaded Paint (000)	Percent of SFHUs with

Deteriorated Unleaded Paint (%)b



Estimate	Lower 95% CIc	Upper 95% CI	Estimate	Lower 95% CI	Upper 95% CI

1978-1998	26,349	5,149	2,525	7,773	20%	10%	29%

1960-1977	22,657	7,081	4,873	9,289	31%	22%	41%

1940-1959	18,497	5,748	4,076	7,420	31%	22%	40%

Before 1940	15,148	2,824	1,828	3,819	19%	12%	25%

Total SFHUs	82,651	20,802	13,303	28,301	25%	16%	34%

Significantly Deteriorated Unleaded Paint



Construction Year	Total SFHUs (000)2	No. of SFHUs with Significantly
Deteriorated Unleaded Paint (000)	Percent of SFHUs with Significantly
Deteriorated Unleaded Paint (%)1



Estimate	Lower 95% CI3	Upper 95% CI	Estimate	Lower 95% CI	Upper 95% CI

1978-1998	26,349	3,320	1,521	5,119	13%	6%	19%

1960-1977	22,657	5,997	3,752	8,241	26%	17%	36%

1940-1959	18,497	4,658	3,107	6,209	25%	17%	34%

Before 1940	15,148	2,250	1,274	3,225	15%	8%	21%

Total SFHUs	82,651	16,225	9,654	22,795	20%	12%	28%

a	“Single family housing units” include permanently occupied,
noninstitutional single family housing units in which children are
permitted to live.

b	Percentages are calculated with total housing units (95,688) or with
housing units with a child under age 6 (19,577) as the denominator, or
as applicable.

c	CI = 95% confidence interval for the estimated number or percent.



Table A.4c	Distribution of Single Family Housing Units (SFHUs) with
Deteriorated and Significantly Deteriorated Lead-Based Paint (LBP) by
Construction Year

Deteriorated LBP

Construction Year	Total SFHUs (000)2	Number of SFHUs with

Deteriorated LBP (000)	Percent of SFHUs with

Deteriorated LBP (%)1



Estimate	Lower 95% CI3	Upper 95% CI	Estimate	Lower 95% CI	Upper 95% CI

1978-1998	26,349	83	0	238	0%	0%	1%

1960-1977	22,657	910	242	1,579	4%	1%	7%

1940-1959	18,497	6,239	4,319	8,159	34%	23%	44%

Before 1940	15,148	8,694	7,107	10,281	57%	47%	68%

Total SFHUs	82,651	15,926	11,668	20,257	19%	14%	25%

Significantly Deteriorated LBP

Construction Year	Total SFHUs (000)2	No. of SFHUs with Significantly
Deteriorated LBP (000)	Percent of SFHUs with Significantly Deteriorated
LBP (%)1



Estimate	Lower 95% CI3	Upper 95% CI	Estimate	Lower 95% CI	Upper 95% CI

1978-1998	26,349	83	0	238	0%	0%	1%

1960-1977	22,657	610	97	1,122	3%	0%	5%

1940-1959	18,497	5,002	3,193	6,811	27%	17%	37%

Before 1940	15,148	7,115	5,605	8,626	47%	37%	57%

Total SFHUs	82,651	12,809	8,894	16,797	15%	11%	20%

a	“Single family housing units” include permanently occupied,
noninstitutional single family housing units in which children are
permitted to live.

b	Percentages are calculated with total housing units (95,688) or with
housing units with a child under age 6 (19,577) as the denominator, or
as applicable.

c	CI = 95% confidence interval for the estimated number or percent.

Table A.5	Estimated Number of Housing Units with Lead-Based Paint on
Windows 

	Number of Housing Units with LBP Windows (000)	HUs in Sample

HU Characteristic	Estimate	Lower 95% CI	Upper 95% CI

	All Housing Units	18,222	15,688	20,757	165

Construction Year:





1978-1998	290	0	711	2

1960-1977	2,092	1,251	2,933	20

1940-1959	4,733	3,202	6,264	41

Before 1940	11,108	9,298	12,917	102

Occupied by Children Under Age 6	2,826	1,661	3,991	32

Household in Poverty	3,547	2,324	4,771	36

Household in Poverty, Resident Children Under Age 6	569	169	969	7



Table A.6	Estimated Number of Windows with Lead-Based Paint and in
Housing Units with

		Lead-Based Paint on Windows 

	Number of Windows with Lead-Based Paint (Average/HU)a	Number of Painted
Windows 

(Average/HU)	Total Number of Windows (Average/HU)

HU Characteristic	Estimate	Lower 95% CI	Upper 95% CI	Estimate	Lower 95%
CI	Upper 95% CI	Estimate	Lower 95% CI	Upper 95% CI

All Housing Units	7.7	6.2	9.3	13.9	12.1	15.7	15.0	13.3	16.8

Construction Year:









	1978-1998	3.5	0.0	7.9	7.2	3.9	10.4	7.2	3.9	10.4

1960-1977	5.6	3.4	7.8	14.0	8.1	20.0	14.4	8.5	20.3

1940-1959	6.5	4.2	8.7	12.8	10.6	15.0	14.0	12.0	16.1

Before 1940	8.7	6.9	10.6	14.5	12.5	16.5	15.7	13.8	17.7

Occupied by Children Under Age 6	12.7	8.0	17.5	20.4	14.5	26.3	21.8	16.2
27.4

Household in Poverty	4.3	2.7	6.0	7.9	5.9	9.9	9.4	8.1	10.7

Household in Poverty, Resident Children Under Age 6	4.8	0.7	8.8	9.8	7.8
11.9	9.8	7.8	11.9

a Entries are average number of windows per housing units, averaged over
the housing units with LBP on windows.

Table A.7	Distribution of Maximum Soil Sample (Bare and Covered) Lead
Concentrations, All Sampled Locations

Bare Soil Lead	Number of HUs (000)a	Percent of HUs (%)b

	

Estimate	Lower

95% CIc	Upper

95% CI	

Estimate	Lower

95% CI	Upper

95% CI

GEd0 ppm	93,111	91,090	95,131	97%	95%	99%

GE 20 ppm	74,621	68,764	80,478	78%	72%	84%

GE 50 ppm	56,266	49,840	62,693	59%	52%	66%

GE 200 ppm	29,234	28,240	30,228	31%	30%	32%

GE 400 ppm	20,390	17,032	23,748	21%	18%	25%

GE 1,200 ppm	11,145	7,580	14,709	12%	8%	15%

GE 1,600 ppm	7,426	5,289	9,563	8%	6%	10%

GE 2,000 ppm	6,809	4,535	9,084	7%	5%	10%

GE 5,000 ppm	2,987	1,548	4,427	3%	2%	5%

No Bare Soil	335	0	960	0%	0%	1%

No Soil	2,242	330	4,154	2%	0%	4%

Missinge	95,688





	a	“Housing units” include permanently occupied, noninstitutional
housing units in which children are permitted to live.

b	All percentages are calculated with total housing units (95,688) as
the denominator.

c	CI = 95% confidence interval for the estimated number or percent.

d	GE equals “greater than or equal to.” 

e	“Missing” means that soil was present, but that no lead value is
available (either the sample was not collected, e.g. due to
inaccessibility or respondent refusal, or the laboratory did not submit
a value).  “No soil” means that there was no soil on the property to
sample. 

Table A.8	Distribution of Maximum Soil Sample (Bare and Covered) Lead
Concentrations by Construction Year, All Sampled Locations

Soil Lead Concentration	Number of HUs (000)a	Percent of HUs (%)b

	Before 1940	1940 - 1959	1960 - 1977	1978 - 1998	Before 1940	1940 - 1959
1960 - 1977	1978 - 1998

GEc  0 ppm	16,328	19,605	27,608	29,569	93%	95%	99%	99%

GE 20 ppm	16,328	19,279	25,238	13,776	93%	94%	91%	46%

GE 50 ppm	15,820	17,670	14,092	8,684	91%	86%	51%	29%

GE 200 ppm	13,314	9,950	4,495	1,476	76%	48%	16%	5%

GE 400 ppm	11,613	6,283	2,410	84	67%	31%	9%	0%

GE 1,200 ppm	6,536	3,922	686	- 	37%	19%	3%	0%

GE 1,600 ppm	4,455	2,284	686	- 	26%	11%	3%	0%

GE 2,000 ppm	3,929	2,194	686	- 	23%	11%	3%	0%

GE 5,000 ppm	1,891	865	231	- 	11%	4%	1%	0%

Missingd	145	- 	190	- 	1%	0%	1%	0%

No soil 	1,003	939	95	205	6%	5%	0%	1%

Total	17,476	20,544	27,893	29,774	100%	100%	100%	100%

a	“Housing units” include permanently occupied, noninstitutional
housing units in which children are permitted to live.

b	All percentages are calculated with total housing units of that age as
the common denominator.

c	GE equals “greater than or equal to.”

d	“Missing” means that soil was present, but that no lead value is
available (usually due to inaccessibility or respondent refusal).  “No
soil” means that there was no soil on the property to sample.



APPENDIX B

Comparison of Protocols

for the 

HUD 1990 Survey of Lead-Based Paint (LBP) in Housing

and the

HUD National Survey of Lead and Allergens in Housing

Comparison of Protocols for the HUD 1990 Survey of Lead-Based Paint
(LBP) in Housing

and the HUD National Survey of Lead and Allergens in Housing

Area	

HUD National Survey of Lead-Based Paint in Housing	

HUD National Survey of Lead and Allergens in Housing



Types/numbers of housing units selected for the survey and whose data
were available to the TSCA Section 403 risk analysis	

284 housing units selected from occupied, permanent, non-institutional
housing in the 48 coterminous states built prior to 1980 and having the
potential for containing children.  (These units were all
privately-owned.  While publicly-owned units were also selected for the
survey, data for these units are not considered in this summary.)	

831 housing units selected in a three-stage stratified random sample
from occupied, permanent, non-institutionalized housing having the
potential for containing children.  75 primary sampling units (PSUs).



Breakdown of selected units by year built	

Pre-1940: 27%

1940-1959: 31%

1960-1979: 42%

Post-1979: 0%	

Pre-1940: 18%

1940-1959: 22%

1960-1977: 29%

Post-1977: 31%



Dates of environmental sampling	

November 1989 to March 1990	

August 1998 to February 1999, and from July to August 1999



Selecting rooms for environmental sampling	

Telephone household interview provided information on rooms. One room
was selected for sampling in each of the following strata:

Wet room -- rooms containing plumbing (e.g., kitchen, bathroom, laundry
room, utility room)

Dry room -- all rooms not classified as wet rooms

Main entryway (floor dust samples only)	

Room Inventory Form from the Screening/Recruiting Questionnaire was used
to obtain information on rooms.  One room was randomly selected for
sampling in each of the following four strata:

Kitchen

Common living area (e.g., living room, den, family room)

Bedroom in which one or more children aged 17 years or younger regularly
slept, or any regularly-occupied bedroom if no such children lived in
the unit (occasionally, two such bedrooms were selected)

Other random room among the remaining rooms in the housing unit.  (Note:
 Two rooms were randomly selected from this stratum if the stratum
contained at least six rooms.)

Main entry (floor dust only)

Interior common area (multi-family dwellings, floor dust only)



Method of assigning sampling weights	

Weights reflect the various stages of sampling.  Total of the sampling
weights equaled the estimated number of housing units with children
under age 7 years (13,912,000, as estimated by the 1987 AHS).  Total of
the sampling weights within a given census region equaled the estimated
number of units with children under age 7 years in the census region.	

Weights reflect the various stages of sampling.  Total of the sampling
weights within a given census region equals the estimated number of
units in the census region.



Method for taking dust samples for lead analysis	

Blue Nozzle vacuum (a few wipe samples were also collected).  Sampling
house dust for lead:  basic concepts and literature review. (1995). EPA
747-R-95-007.	

Wipes, collected in accordance with ASTM E1728-95, Practice for the
field determination of settled dust samples using wipe sampling methods
for lead determination by atomic absorption spectrometry techniques.



Number and location of floor-dust samples per room	

One sample from each selected room (location not dictated in the
protocol)	

One sample from each selected room, generally taken from the center of
the largest open area of the room.



Window sill/trough dust sampling approach	

A window was selected within each selected room according to a ranking
scheme.  Sampling was performed from both the sill and trough of the
selected window until enough dust was collected or until the entire sill
or trough was vacuumed.	

Entire sill and trough sampled from a random window in the selected
room.  Trough definition included sliders. 



Number and location of sill and trough dust samples per room	

One sample from the sill and one sample from the trough of the selected
window in the selected wet room and dry room	

One sample from the sill and one sample from the trough of the selected
window in each selected room



Method of analyzing dust samples	

Graphite Furnace Atomic Absorption Spectroscopy (GFAA) (with EPA SW-846
digestion method) 	

Flame Atomic Absorption Spectrometry (FAAS) using NIOSH method 7082

Digestion method: modification of EPA SW-846 Method 3050 or ASTM ES
36-94 (hot-plate digestions utilizing nitric acid and/or perchloric acid
and/or hydrogen peroxide).  Method same as used in proficiency testing
within the Environmental Lead Laboratory Accreditation Program (ELLAP).



Soil sampling approach	

One composite sample of up to 3 core samples (the latter two taken
within 20 inches of the first), each taken at a depth of 10 cm, was
collected at each of the following locations: entryway, drip-line, and
remote area (i.e., an area halfway between the unit and its property
boundary, or within 25 feet of the unit, whichever was less). 	

Two sides of the unit were selected for soil sampling: the side
containing the major entryway (Wall 1) and a second, randomly-selected
side (Wall 2).  Samples were collected from the top 0.5 inches of soil
at the following three sites:

Main entry - a single sample from Wall 1

Foundation/drip-line - one sample from each of Walls 1 and 2, each
sample being a composite of 3 core subsamples taken within 3 feet of the
foundation

Mid-yard area - one sample from each of Walls 1 and 2, each sample being
a composite of 3 core subsamples taken midway between the drip-line and
boundary of the housing unit property.

Play Area – one composite sample from bare soil under each unit of
fixed play equipment.

Soil samples were collected in accordance with core sampling procedures
based on ASTM E1727-95 (described in the HUD Guidelines and in EPA’s
Residential Sampling for Lead: Protocols for Leaded Dust and Soil
Sampling). Samples were collected from bare soil when possible.  If no
bare soil existed, samples were collected from covered surfaces if
possible. 



Method of analyzing soil samples	

ICP-AES (with SW-846 digestion method)	

ICP-AES using NIOSH method 7082

Digestion method: modification of SW-846 Method 3050 or ASTM ES 36-94
(hot-plate digestions utilizing nitric acid and/or perchloric acid). 
Method same as used in proficiency testing within the Environmental Lead
Laboratory Accreditation Program (ELLAP).



Handling dust-lead and soil-lead measurements below the detection limit	

As log-transformed lead amounts are reported in the database,  only
positive measurements are represented.  No indication is given as to
when data may have been truncated due to being below detection limits.	

The final results as reported by the instrument are recorded in the
database (i.e., not-detected results are not censored), along with
detection limits.



Method for taking paint-lead measurements	

Spectrum analyzer XRF instrument (single 60-second spectrum reading
measurement using a 40 millicurie cobalt source).  Measurements were
adjusted to statistically correct for measurement bias.	

Spectrum analyzer XRF analyzer (full-period readings with a 20-second
minimum in accordance with the applicable HUD-approved Performance
Characteristic Sheet .)



Approach to selecting interior painted components for paint-lead
measurements	

Painted surfaces were categorized into the following four strata:

Walls/ceilings/floors

Metal substrate

Non-metal substrate

Other surfaces

Five painted components were selected randomly for testing in each of
the selected wet and dry rooms, one from each stratum along with a fifth
selected randomly from among all strata.  In addition, up to two
purposive measurements were taken from paint anywhere in the unit that
may be suspected to contain lead.	

The following painted components were measured for lead in each selected
room:

All four major walls

Ceiling

Door of major entryway

Window selected for dust sampling

Baseboard

Floor

Up to two other painted surfaces:  Technician choice based on surfaces
containing deteriorated paint or friction areas.



Approach to selecting exterior painted components for paint-lead
measurements	

Painted surfaces were categorized into the following four strata:

Wall (randomly-selected)

Metal substrate within the selected wall

Non-metal substrate within the selected wall

Other surfaces within the selected wall

Five painted components were selected randomly for testing from the side
of the unit containing the selected wall, one from each stratum along
with a fifth selected randomly from among all strata.  In addition, up
to two purposive measurements were taken from paint anywhere on the
exterior of the unit that may be suspected to contain lead.	

Painted siding was measured for lead levels on each exterior wall.  In
addition, the following painted components were measured for lead on a
random wall:

Miscellaneous trim (2 measurements)

Window

Door of major entryway – wall independent.

Porch and railing – wall independent

Up to two other painted surfaces





APPENDIX C

Correcting for Classification Bias Due to Measurement Error 

Correcting for Classification Bias Due to Measurement Error

Homes were classified as having LBP and lead hazards based on the XRF
readings of paint and the analysis of dust and soil samples.  Random
variation associated with instrument or laboratory measurement, sample
collection, and random selection of sampling locations, can induce a
classification bias resulting in a bias in the estimated prevalence of
HUs with LBP and lead hazards.  In addition, paint and dust measurements
were made in a sample of rooms – not all rooms.  Under this protocol,
it is possible for a home to have LBP or a LBP dust hazard in the
unsampled rooms and non-lead-based paint and/or no dust-lead hazard in
the sampled rooms.  Such HUs would be incorrectly classified as not
having LBP and/or LBP hazards (false negatives).  For this report,
measurement error refers to the combined effect of instrument or
laboratory measurement variation, sampling variation, spatial variation,
and the incomplete sampling of rooms on the important survey estimates. 


The specific procedures, equations, and justification for the
measurement error correction are presented in Appendix C of Volume II. 
The findings for the measurement error analyses for paint, dust, and
soil are discussed below in Sections 7.3.1 through 7.3.3 below.  The
measurement error adjusted values (lead loading or concentration) have,
to the extent possible, the same distribution as the true lead loading
or lead concentration values, without the effect of measurement error. 
The measurement error corrected values are a weighted average of the
observed measurements and predicted values from a regression model. 
Calculating the weights requires modeling the magnitude of the
measurement error variance and the regression error variance.  Replicate
measurements were used to estimate the measurement error variance.  

Note that this analysis used preliminary data from the main survey.  The
analysis did not include data from the follow-on play area survey.  In
addition, minor revisions or corrections to the data may have been made
since the analysis was performed.  As a result, summary statistics from
the measurement error analysis may not agree with those in other
sections of the report.  Nevertheless, the effects of measurement error
are expected to similar in the preliminary and final data.

C.1	Measurement Error - Paint XRF Measurements

XRF readings to measure paint lead loading were taken on painted
surfaces within the sampled rooms and on accessible exterior surfaces on
two sides of the building to assess the lead loading in paint.  The
measurement error adjustment procedures were first applied to the
interior XRF readings.  Because the measurement error adjusted XRF
values were very close to the original interior XRF readings, it was
considered unnecessary to apply a measurement error adjustment to the
exterior XRF readings.

For the measurement error analysis for paint lead measurements, the
objective was to determine the number and percentage of homes with LBP. 
A surface with LBP is a surface with an average lead loading across the
surface of 1.0 mg/cm2 or greater when tested with an XRF analyzer.  The
instrument calculates its best internal estimate of the lead loading,
which it rounds to the nearest 0.1 mg/cm2 for display.  Therefore, to
classify surfaces in an equivalent manner, the error corrected XRF
readings were rounded to the nearest tenth unit for classifying
surfaces.  The classification of surfaces as having or not having LBP is
used to classify homes as having or not having LBP.

The measurement error adjustment procedure assumes that the distribution
of the true paint lead loading around the average for similar homes and
the distribution of the measurement error are normal.  This assumption
provided a relatively poor description of the XRF readings due to the
presence of many zero XRF readings and long tails in the XRF
distribution (the measurement error adjustment results are expected to
be sensitive values in the tails of the distribution).  As a result, the
measurement error adjustment for paint lead loadings should be
considered approximate, at best.  Additional research into the
measurement error of the XRF instrument and the distribution of paint
lead loading across components would be required to improve the
measurement error adjusted values.  

Figure C.1, which shows a scatter plot of replicate XRF readings, helps
to illustrate the distribution of the XRF readings.  The original XRF
reading is on the horizontal axis and the replicate reading is on the
vertical axis.  Three situations occur. The first is where both XRF
readings are zero (72.3 percent of the replicate pairs) - these pairs
provide essentially no information about measurement error.  The second
case is where one of the two readings is zero (12.4 percent of the
replicate pairs) - these pairs fall on the axes.  Lastly, there is the
case where both XRF readings are non-zero (15.3 percent of the replicate
pairs) – these generally fall on the diagonal in Figure C.1. For these
points, the differences between the replicate XRF readings have an
approximately normal distribution as assumed by the measurement error
correction procedure. 

The XRF variation among all surfaces within a home increases with the
average XRF reading within the home.  This suggests that the measurement
error will also increase with the paint lead loading on a component. 
Surfaces with lead loading near zero will have relatively precise
measurements and we expect that large differences between replicate
readings when one reading is zero will be relatively rare.  However,
there are many surfaces for which one reading is zero and the other
paired reading is relatively large (may be due to two different parts of
a component having very different paint lead loadings).  

Figure C.1	Original and Replicate XRF Readings on the Same Component

Note: The axes use the transformed scale used for analysis (see Equation
9 in Appendix C,  Volume II).  

For the measurement error adjustment, homes in which all XRF readings
were zero or negative (162 HUs) were assumed to have no paint lead.  The
remaining homes were used in the measurement error analysis.  For those
homes, the model used to predict paint lead loadings had factors for
housing unit ID, room type, year of construction, and interactions of
year of construction by substrate, component, and percent deteriorated
paint.  A preliminary analysis suggested that the relationship between
paint deterioration and XRF reading was not linear.  To make the
relationship closer to linear, the cube of the percent deterioration was
used in the model.  A further analysis of the relationship would be
necessary to provide a better model.

The regression analysis predicts the approximate median of the interior
paint lead loading. Figure C.2 shows the relative median paint lead
loading estimated from regression, by construction year category and
component substrate.  Figure C.3 shows the relative differences in
median paint lead loading associated with combinations of construction
year and region of the country.  Figure C.4 and 7.5 shows the relative
differences in the median paint lead loading versus component type, room
type, component condition, metro status, presence of pets, overall home
cleanliness, and the number of days the air conditioning was used in the
last month. Because the assumptions behind the measurement error
analysis provide a poor description of the XRF measurements, the
patterns illustrated in Figures 7.2 to 7.5 should be considered
suggestive of, rather than descriptive of, patterns in the population.  

Figure C.2	Predicted Relative Median Paint Lead Loading by Construction
Year of the Home and Substrate (based on regression model)

Figure C.2 shows that interior paint lead loadings are highest in the
oldest homes and lowest in the newest homes.  The patterns for paint
lead loading on specific substrates also generally decrease with the age
of the home.  The paint lead loadings are highest on wood surfaces in
the oldest homes.  For components with drywall, paneling, metal,
plaster, wallpaper, and wood substrates, the median paint lead loading
is low (less than 0.05 mg/cm2) for homes built since 1960.  However,
median paint lead loading on other substrates (such as brick, concrete,
stone, and vinyl) are higher for homes built in the 1960 to 1977 period.
 

Figure C.3 shows that the highest predicted median paint lead loading is
found in homes built before 1960 in the northeast US and in homes built
before 1940 in the southern US.  

Figure C.3	Predicted Relative Median Paint Lead Loading by Construction
Year of the Home and Region (based on regression model)

Figure C.4 shows that higher paint lead loadings are generally found on
doors, window jambs and window sashes.  There are relatively small but
significant differences in the paint lead loading among rooms of
different types.  After accounting for other factors, paint lead
loadings are generally higher in children’s bedrooms and in kitchens. 


Figure C.4	Predicted Relative Median Paint Lead Loading by Component
Type and Room Type (based on regression model)

Figure C.5 shows that higher paint lead loadings are generally found on
components that were judged to be in fair condition as opposed to intact
or poor condition.  However, relatively few surfaces were judged to be
in fair condition.  Small but significant differences were associated
with metro status (lower paint lead loading in non-MSAs), presence of
pets (lower in homes with pets), overall cleanliness, and days using air
conditioning in the last month.

Figure C.5	Predicted Relative Median Paint Lead Loading by Component
Condition, Metro Status, Presence of Pets, Overall Cleanliness, and Air
Condition Use in the Last Month (based on regression model)

As part of the exploratory analysis to identify the model for the data,
such additional variables as tenure and race were considered and not
found to be significant predictors.

The measurement error adjustment procedure identified approximately 30
percent of the XRF readings as outliers.  Outliers are those
measurements that are more than 2.5 standard deviations above or below
the mean of the non-outliers.  In this case, the measurement error
procedure makes a conservative adjustment.  The non-outlier data were
assumed to have a normal error distribution, however, the distribution
of the residuals had a tight distribution in the center and very long
tails, inconsistent with the normal distribution assumption.  The
resulting estimate of the error variance was very small, resulting in
almost no estimated effect of measurement error.  Using the measurement
error procedure in Appendix C (Volume II), the measurement error
adjusted paint lead loadings are essentially equal to the XRF readings. 
Additional work would be required to develop a better model for the XRF
readings and other assumptions might provide a significantly different
assessment of the effects of measurement error. 

The tentative results from the measurement error analysis are that the
interior XRF readings provide a reasonable measure of the paint lead
loading for many surfaces.  For the remaining surfaces that look like
outliers when judged relative to a normal distribution, the effect of
measurement error is difficult to assess.  We believe that similar
conclusions are likely to apply to exterior surfaces; however, no
measurement error adjustment was attempted for the exterior surfaces.  

C.2	Measurement Error – Dust Lead Measurements

Dust samples were taken at all surveyed homes from the floor at the main
entrance and from the floors, window sills, and window troughs of the
sampled rooms.  Separate measurement error adjustments were performed
for the floor and window dust samples.  

Floor Dust Lead Loading

The model for predicting floor dust lead loading included factors for
household ID (as a class variable), surface characteristics at the
sample location, and the interaction of the type of floor cover and year
of construction.  The measurement error adjustment procedures excluded
three percent of the floor dust measurements as outliers.  The
regression analysis predicts the approximate median of the floor dust
lead loading.  

Figure C.6 shows the relative median floor dust lead loading estimated
from regression, by construction year category and type of floor cover
in the sampled room.  The predicted median floor dust lead loadings were
highest for the oldest homes and decrease for homes constructed more
recently.  The floor lead loading was generally higher in rooms with no
floor covering than in rooms with wall-to-wall carpets.  Rooms with some
floor covering had floor dust loadings similar to or somewhat greater
than in rooms with wall-to-wall carpets, and lower than rooms with no
floor covering.  The floor lead loadings for the unknown floor cover
category represent primarily the floor dust lead loadings for the main
entrance (94% of the data in the unknown category, as this information
was not recorded).  The dust lead loading at the main entrance was
similar to dust lead loading from other rooms that have no floor
covering.  

Figure C.6	Predicted Relative Median Floor Dust Lead Loadings by
Construction Year Category and Floor Cover in the Sampled Room (based on
regression model)

Figure C.7 shows the relative median floor dust lead loading estimated
from regression, by home cleanliness, household income, and surface
sampled.  Dust lead loadings were on average higher in lower income
homes, and in homes that show no evidence of cleaning (a small minority
of all homes).  Dust lead loadings were also higher on hard surfaces
that were not smooth and cleanable (a small minority of all surfaces),
perhaps because these surfaces are more difficult to clean or perhaps
because they collect dust faster between cleanings.  Carpets had lower
lead loading than smooth and cleanable surfaces, using the wipe sampling
methods employed.  

Figure C.7	Predicted Relative Median Floor Dust Lead Loadings by
Cleanliness, Household Income, and Surface Condition (based on
regression model)

As part of the exploratory analysis to identify the model for the data,
additional variables were considered and not found to be significant
predictors.  In particular, the type of room (bedroom, bathroom,
kitchen, etc.) and presence of pets were not significant.  

Figure C.8 shows the cumulative distribution of the maximum floor dust
lead loading at homes.  The figure shows three cumulative distributions:

The maximum of the floor dust lead loading measurements at a home
(bottom thin black line).  

The maximum measurement error adjusted floor dust lead loading across
the sampled rooms (top gray line).

The maximum measurement error adjusted floor dust lead loading across
all rooms, sampled and unsampled (middle thick black line).  

The difference between the first (bottom thin black) and second (top
gray) curves shows the effect of the measurement error adjustment on the
classification of the measured surfaces.  The difference between the
second (top gray) and the third (middle thick black) curves illustrates
the effect of random selection of rooms on the classification of homes. 
Measurement error tends to increase the number of homes, and incomplete
sampling of rooms tends to decrease the number of homes, classified as
having maximum floor dust lead loading above a selected value.  For the
floor data, these effects partially cancel out so that the number of
homes classified as having floor lead over a specified value using
either the maximum adjusted lead loading value or the maximum observed
measurements is similar.  The difference between the first (bottom thin
black) and the third (middle thick black) curves illustrates the
combined effect of doing both the measurement-error adjustment and
unsampled-room adjustment.

Using the measurement error corrected values, an estimated four percent
of homes (about 4 million homes) have floor dust lead loadings of 40
g/ft2 or more in one or more rooms.  This is about one percent fewer
homes than estimated using the actual floor dust measurements.

Figure C.8	Cumulative Distribution of the Maximum Floor Dust Lead
Loading for Homes

Window Dust Lead Loading

Window dust samples were taken from randomly selected windows in the
sampled rooms within surveyed homes.  For the measurement error
correction, one objective was to estimate the number of rooms having or
not having average window sill dust lead loading and average window
trough dust lead loading less than a selected value.  A second objective
was to estimate the number of homes with average room window sill and
window trough dust lead loading less than a selected value in all rooms.
 The first objective required estimating lead loadings for the missing
values and adjusting for measurement error.  The second objective also
required adjusting for the unknown lead loading in the unsampled rooms. 


Scatter plots of the data showed that the log transformed window sill
and trough measurements are linearly related, with the ratio of the
trough to sill lead loading being roughly constant at about 11.  The
proximity of the sill to the trough suggests that the similar factors
may affect the lead loading at these two locations.  Therefore, the
analysis below used both the window sill and window trough dust lead
loadings in the same model for calculating the measurement error
adjustment.  Because the log transformation was used, the model assumes
that the same factors predict the relative sill and trough lead loadings
and that the ratio of the sill to trough lead loading was constant
across homes.  Although a separate measurement error adjustment could
have been performed separately for sill and trough measurements, this
combined approach was expected to provide a similar adjustment.  The
measurement error adjustment procedures excluded three percent of the
floor dust measurements as outliers.

Two models were fit to the data.  One was used for predicting window
dust lead loading in homes with at least one window lead loading
measurement.  A second model was used to predict window dust lead
loading for homes with no window dust lead loading measurements in the
data files.  The predictors in each model are presented in Table C.1.  

Table C.1	Regression Model Used for Different Categories of Homes

Category of homes	Number of homes	Factors for predicting window dust
lead loading

I. Homes with no window dust samples	17	Sample location (sill or
trough), year of construction, tenure, race category of occupants,
overall household cleanliness as judged by the interviewer, and room
type.

II. Homes with at least one window dust sample	808	Sample location (sill
or trough), household ID (as a class variable), surface characteristics
at the sample location, whether the window has a vinyl mini-blind, and
room type.



Figure C.9 shows the relative differences in median window dust lead
loading associated with different levels of sample location (sill or
trough), year of construction, tenure, race category, overall household
cleanliness, surface characteristics, room type, and presence of vinyl
mini-blinds. 

Figure C.9	Predicted Median Window Dust Lead Loading by Various Factors
(based on regression model)

 

The predicted median window dust lead loading is much higher in the
window trough than on the window sill.  Lead loading is highest for the
oldest homes and lower for homes constructed more recently.  The window
lead loading is generally higher for rented homes than in private owned
homes and higher in homes with African American residents and in homes
with residents of other races.  The small number of homes that have no
apparent indication of cleaning have higher window dust lead loadings
than homes that were classified as showing some evidence of cleaning or
appearing clean.  Although most rooms had similar window dust lead
loading measurements, measurements were somewhat higher in adult
bedrooms and “Other” rooms (rooms not classified as kitchen, living
room, or bedroom).  Samples from window surfaces that were not smooth
and cleanable had higher lead loading than samples from smooth and
cleanable surfaces.  Finally, windows with vinyl mini-blinds had
somewhat higher dust lead loading than windows without vinyl
mini-blinds.  Differences for other window coverings were not
statistically significant.

Figure C.10 shows the cumulative distribution of the maximum window sill
dust lead loading within homes.  The figure shows four cumulative
distributions:

The maximum of the window dust lead loading measurements within a home
(thin black line).  

The maximum of the window dust lead loading measurements or predicted
values for missing data (also a thin black line).

The maximum measurement-error-adjusted window dust lead loading across
the sampled rooms (thick gray line).

The maximum measurement-error-adjusted window dust lead loading across
all rooms, sampled and unsampled (thick black line).  

For various reasons, such surfaces being inaccessible, data were not
available for some window surfaces.  The first curve assumes that the
lead loading on all surfaces with missing data is negligible.  As a
result of the measurement error adjustment, predicted values are
available for all surfaces.  For surfaces with no data, the predicted
values were used.  The second curve, to the right of the first curve,
shows the cumulative distribution of the best estimate of the lead
loading on all window surfaces, i.e., the measurement on surfaces with
data and the predicted values on surfaces with no measurements.  The
predicted values are referred to as imputed values.  The imputation
generally affects the lower portion of the distribution.

The third curve shows the distribution of the measurement error adjusted
estimates.  Since measurement error will generally increase the maximum
within-home measurements, the effect of the measurement error correction
is to slightly reduce estimated maximum within-home window dust lead
loading.  The fourth curve shows the cumulative distribution of the
measurement error corrected measurements after accounting for the
incomplete sampling of rooms.  These values will be referred to as the
adjusted values.  The primary difference between the maximum of the
observed window sill dust lead loading measurements and the final
adjusted maximum in each home is due to the incomplete sampling of
rooms.

g/ft2 in at least one room.  This is about five percent more homes
than estimated using the actual measurements.  A predicted 22 percent of
homes (21 million homes) have a window trough dust lead loading of at
least 800 g/ft2 in at least one room.  This is about 20 percent more
than estimated using the actual measurements.  

Figure C.10	Cumulative Distribution of the Maximum Within-Home Window
Sill Dust Lead Loading 

Figure C.11 shows the cumulative distribution of the maximum window
trough dust lead loading within homes.  The figure shows four cumulative
distributions for the measurements (thin upper thin black line), the
measurements plus imputed values (lower thin black line), the
measurement error corrected values for the sampled rooms (gray line) and
the final adjusted values that account for the incomplete sampling of
rooms

Because there were many surfaces for which window trough data could not
be obtained, there are relatively large differences between the
distributions for the measurements and the measurements with imputed
values.  As with the window sill lead loading, the measurement error
adjustment for the available measurements makes only a small difference
compared to the effect of missing data and the incomplete sampling of
rooms.   

Figure C.11	Cumulative Distribution of the Maximum Within-Home Window
Trough Dust Lead Loading 

C.3	Measurement Error – Soil Lead Measurements

Soil samples were taken at all surveyed homes at the major entrance and
along the dripline and midyard of two sides of the home, if soil was
present.  For the soil measurement error correction, the objective was
to determine the number and percentage of homes with average soil lead
concentrations above a selected value at the three sample sites -
midyard, dripline, and main entrance.  Note that this analysis was
performed on preliminary data, before the play area survey samples were
collected.

Table C.2 describes the regression model used for different categories
of homes.

For Category I homes, there was no soil at any of the three sample
sites, and so no model was required.  This situation occurred if all
areas of the yard were covered with concrete, asphalt, or rock.  Also, a
specific sample site could have no soil if the site did not exist (for
example, the dripline and the property line coincided so that there was
no midyard sample site). 

For Category II homes, at least one of the sample sites had soil, but
there were no soil lead measurements for the home.  This situation could
arise if the respondent denied permission to collect soil samples, or if
adverse conditions existed, such as an ongoing storm, frozen ground, the
presence of a dog, or large rocks mixed in with the soil. The
measurement error corrected soil lead concentrations are the predicted
values from the regression model.

For Category III homes, all soil measurements were equal to zero.  While
the sample detection limit was determined to be 20 ppm, the laboratory
provided an estimate of the soil lead for all samples with levels below
the detection limit.  However, negative laboratory estimates were
reported as zero. The measurement error corrected soil lead
concentrations are assumed to be less than the detection limit. 

For Category IV homes, some or all of the soil measurements were equal
to a non-zero value (i.e., a value at or more than the detection limit).
 These values were used to predict soil lead concentrations for the
category IV and category II homes.

Table C.2	Regression Model Used for Different Categories of Homes

Category of homes	Number of homes	Model for measurement error corrected
soil concentrations

I. Homes with no soil at any sample site	45	Not applicable.

II. Homes with soil, but no soil measurements	8	Log transformed soil
lead concentration = a mean for each combination of construction year
category and soil sample location and a mean for each combination of
construction year category and region.  

III. Homes with all soil lead measurements equal to zero (therefore no
within-home variation)	33	All measurements assumed to be less than the
detection limit.

IV. Homes with soil lead measurements, some or all of which are non-zero
745	Log transformed soil lead concentration = a mean for each
combination of construction year category and soil sample location, a
mean for each combination of construction year category and region, and
a mean for each home



The regression analysis predicts the approximate median of the soil lead
concentrations.   Figures C.12 and C.13 show the relative median soil
lead concentration for homes within each construction year category by
sample location and region.  As expected, older homes have higher soil
lead concentrations.  Concentrations are also higher on average in the
northeast region and lower in the western region of the country.  Soil
lead concentrations are higher on average at the dripline sample
location and lower at the midyard sample location.  The measurement
error adjustment procedures excluded three percent of the floor dust
measurements as outliers.

As part of the exploratory analysis to identify the model for the data,
additional variables were considered.  In particular, orientation of the
side of the building (north, east, south, or west) and ground cover
(bare soil, grass, ivy, moss, mulch, and other or unknown) were
examined.  Neither the side of the house on which the samples were taken
nor the ground cover was a significant predictor of soil lead
concentrations, after adjusting for the effects of year of construction,
region, and sample location.

Figure C.12	Predicted Relative Median Soil Lead Concentrations by
Construction Year Category and Sample Location (based on regression
model)

Figure C.13	Predicted Median Soil Lead Concentrations by Construction
Year Category and Region (based on regression model)

Figure C.14 shows the cumulative distribution of the maximum soil lead
concentration at homes.  The figure shows three cumulative
distributions:

The maximum of the soil lead measurements at a home (bottom thin black
line).  

The maximum within each home of the average soil lead measurements at
the entrance, dripline, and midyard (middle gray line).

The maximum measurement error corrected soil lead measurements (top
thick black line).  

Figure C.14	Cumulative Distribution of the Maximum Soil Lead
Concentration for All Homes 

Note:  Play area soil sample results were not included in the analysis
for Figure C.14. Thus, data may vary slightly from that presented in
Chapter 6.

The difference between the first and second curves shows the effect of
taking replicate soil measurements (i.e. more than one sample for a
given sample location at a home).  As more measurements are taken, it is
more likely that a sample will be taken which has an unusually high
measurement (either due to laboratory variation or due to sampling a
small area with a locally high concentration).  The second curve is most
comparable to the third curve.  The difference between the second and
the third curve is due to the measurement error correction.  The fact
that the third (measurement error corrected) curve is to the left of the
other curve is because, after correcting for measurement error, the
percentage of homes judged to have soil lead concentrations above any
selected value is decreased.

Using the measurement error corrected values, an estimated 5.6 percent
of homes with soil (about 5.3 million homes) have soil lead
concentrations above 1,200 ppm in one or more of the sampling locations
(dripline, midyard, and main entrance).  This is about two million fewer
homes than estimated using the observed soil lead measurements.  Even
fewer homes have maximum soil lead concentrations over 1,200 ppm on bare
soil.  



APPENDIX D

Calculation of Soil Related Estimates 

Calculation of Soil Related Estimates

In the main survey, soil samples were collected near the main entrance,
along the drip line and in the mid-yard on the side of the house with
the main entrance, and along the dripline and in the mid-yard on a
second randomly selected side of the home.  In a follow-on survey of
children's play areas involving a subset of the homes in the main
survey, residents were asked where children play around the home.  If
there was play equipment at the home or soil samples were not previously
collected in children's play areas, additional soil samples were
collected at the home in a follow-up visit.  In the play area survey,
soil samples were collected from below play equipment and from areas
designated by the respondent as children's play areas.  The main survey
has data from 831 homes.  The follow-on survey has information on 375
homes.  Details of the survey procedures are described in the Design and
Methodology report.

The tables in this report are based on data from the 831 homes whenever
possible.  However, for estimates involving soil in children's play
areas, data was only available from 375 homes.  Therefore, estimates not
involving soil were based on the main survey data.  Estimates involving
only soil were based on the play area survey data.  For estimates that
involve both soil data and data from other locations, such as the number
of homes with significant lead-based paint (LBP) hazards, estimates from
the main survey were combined with estimates from the play area survey,
as described below.

To construct estimates of significant LBP hazards using as much of the
data as possible the following procedure was used.  Estimates of the
number of homes with significant LBP hazards due to paint, dust, and
soil sources other that play soil were calculated from the 831 homes
from the main survey.  Estimates of the number of homes with significant
LBP hazards due only to soil hazards in play areas were calculated from
the 375 homes in the play area survey.  These two estimates were added
together to obtain the combined estimate of the number of homes with
significant LBP hazards from any source.

For example, the estimated number of homes with a significant LBP hazard
is 24,026 thousand.  This is the sum of 23,312 thousand homes with paint
hazards, dust hazards, or soil lead hazards in non-play areas (based on
data from 831 homes) and 714 thousand homes with significant LBP hazards
in play areas only (based on data from 375 homes).  

The variance of the combined estimate depends on the portion of the
estimate contributed by the two data sources, the variance of each
estimate, and the correlation between the estimates.  Calculation of the
variance of the combined estimate is difficult because the correlation
is difficult to estimate.  However, the estimate from the main survey
contributes 97% of the combined estimate.  As a result, the variance of
the combined estimate is primarily due to uncertainty in the estimate
from the main survey.  Therefore, the confidence intervals in the tables
involving lead hazards, are based on the confidence intervals calculated
from the main survey.  The resulting confidence interval will tend to
underestimate the true variance by a small amount.  Thus, the confidence
intervals for the number of homes are calculated as follows:

The confidence interval and estimate for the number of homes with
significant LBP hazards in all but play soil are calculated from the
main survey data.

The estimate for the number of additional homes with a significant LBP
hazards in only the play soil is calculated from the play survey data.

The estimate from the play survey data is added to the estimate from the
main survey to get the combined estimate, and

The estimate from the play survey data is added to the confidence limits
for the estimate from the main survey to approximate the confidence
limits for the combined estimate.

 Title 24 of the Code of Federal Regulations, Part 35 was issued
September 15, 1999 in Volume 64 of the Federal Register, pages
50140-50231, and became effective September 15, 2000. It implemented
sections 1012 and 1013 of the Residential Lead-Based Paint Hazard
Reduction Act of 1992, which is Title X of the Housing and Community
Development Act of 1992 (P.L. 102-550).  A copy is available on the
Internet at www.hud.gov/offices/lead.

 Issued January 5, 2001 in Volume 66 of the Federal Register, pages
1206-1240, and became effective on March 6, 2001.

 The 95 percent confidence intervals for the estimates are presented in
the main body of the report.

 President's Task Force on Environmental Health Risks and Safety Risks
to Children, 2000.  Eliminating Childhood Lead Poisoning: A Federal
Strategy Targeting Lead-Based Paint Hazards (Report and Appendix). 
Washington, D.C.:U.S. Department of Housing and Urban Development and
U.S. Environmental Protection Agency.

 The de minimis levels for LBP deterioration are in Section 35.1350(d)
of the Lead Safe Housing Rule.  These levels are:  deterioration of more
than 20 square feet (exterior) or 2 square feet (interior) of LBP on
large surface area components (walls, doors) or deterioration of more
than 10% of the total surface area of interior small surface area
components types (window sills, baseboards, trim).  These are the same
levels used in the U.S. Environmental Protection Agency’s lead hazard
standards rule implementing the Toxic Substance Control Act’s Section
403.

 The floor and window sill dust lead loading thresholds are dust on
floors with greater than or equal to 40 µg/ft2 lead and dust on window
sills with greater than or equal to 250 µg/ft2 lead.  They are in the 
HUD Lead Safe Housing rule and in the EPA Rule Identification of
Dangerous Levels of Lead; 40 CFR Part 745, January 5, 2001.

 The thresholds for bare, lead-contaminated soil are more than 9 square
feet of bare soil with a lead concentration greater than or equal to
1,200 ppm lead, or 400 ppm for bare soil in an area frequented by a
child under the age of 6 years.  These thresholds are in the HUD Lead
Safe Housing Rule.

 HUD 1539-LBP.  Guidelines for the Evaluation and Control of Lead-Based
Paint Hazards in Housing, Chapter 7.  U.S. Department of Housing and
Urban Development, Washington, D.C.

 ASTM E 1728-95. (1995b). Standard practice for the field collection of
settled dust samples using wipe sampling methods for lead determination
by atomic spectrometry techniques.  American Society for Testing and
Materials, West Conshohocken, PA.

 ASTM E 1727-95.  Standard practice for the field collection of soil
samples for lead determination by atomic spectrometry techniques. 
American Society for Testing Materials, West Conshohocken, PA.

 National Academy of Sciences (1993).  Measuring lead exposure in
infants, children, and other sensitive populations.  National Academy
Press, Washington, DC.

 Bornschein, R., Hammond, P.B., Dietrich, et al.  (1985a).  The
Cincinnati prospective study of low-level lead exposure and its effects
on child development: Protocol and status report.  Environ. Res.
38:4-18.

 Bornschein, R., Succop, P., Dietrich, et al.  (1985b).  The influence
of social and environmental factors on dust lead, hand lead, and blood
lead levels in young children.  Environ. Res. 38:108-118.

 In Title X of the Housing and Community Development Act of 1992 (P.L.
102-550), the term “lead-based paint hazard” means any condition
that causes exposure to lead from lead-contaminated dust,
lead-contaminated soil, lead-contaminated paint that is deteriorated or
present in accessible surfaces, friction surfaces, or impact surfaces
that would result in adverse human health effects as established by the
appropriate Federal agency.

 Clark, S., Bornschein, R., Succop, P., et al. (1985).  Conditions and
type of housing as an indicator of potential environmental lead exposure
and pediatric blood lead levels. Environ. Res. 38, pp. 46-53.

 Centers for Disease Control and Prevention (1991).  Preventing Lead
Poisoning in Young Children. Centers for Disease Control and Prevention,
U.S. Department of Health and Human Services, Atlanta, GA.

 Lanphear, B.P., et al. (1995).  The Relation of Lead-Contaminated House
Dust and Blood Lead Levels Among Urban Children, Final Report. Report to
the U.S. Department of Housing and Urban Development.

 A PSU is a metropolitan statistical area (MSA), county, or cluster of
counties that have a minimum population of 15,000 and do not cross
Census region boundaries.

 Throughout the volume, the concepts of lead loading and lead
concentration are used.  Lead in paint and dust are reported as loadings
(mass of lead per unit area of surface), while lead in soil is reported
as a concentration (mass of lead per unit mass of soil).  For paint,
lead loading is the number of milligrams of lead per square centimeter
of painted surface (mg/cm2). For dust, lead loading is the number of
micrograms of lead per square foot of wiped surface ((g/ft2).  Soil is
reported as the number of micrograms of lead per gram of soil ((g/g),
equivalent to parts per million (ppm).

 A PSU is a metropolitan statistical area (MSA), county, or cluster of
counties that have a minimum population of 15,000 and do not cross
Census region boundaries.

 Cross comparisons of two variables, e.g., Region by Construction Year
and Poverty by Urbanization, result in cells containing 30 or fewer HUs.
 Caution is recommended in the interpretation of these results.

 While we were most interested in children under the age of six years
for lead, we were also interested in all children up to age 18 for
allergen exposures. Comparative data from the AHS are only available for
children under age 18 in the household.  For the same reason,
post-stratification for this survey was based on children under age 18.

 If AHS or CPS data are not listed in Table 2.1, e.g., for Government
Support, they were not available in these sources.

 Family size includes all related people living in a housing unit. 
Household size includes all people living the housing unit, whether or
not they are related to each other.  Thus household size tends to be
larger than family size. The 1999 CPS reports an average household size
of 3.17 and an average family size of 2.62.

 American Housing Survey 1997, Appendix D,
http://www.census.gov/hhes/www.housing/ahs/meth.html.

 The cut-off used with survey data was actually metropolitan statistical
areas (MSAs) of 2,100,000 while the available CPS cut-off is 2,500,000. 
This difference in definitions explains the slight discrepancy in the
findings.

 Intact LBP present on accessible surfaces, friction surfaces, or impact
surfaces were not included in the definition of LBP hazard for the
estimates presented in this report because this information was not
specifically collected for each component.

 Window trough dust is not considered in the definition of a LBP hazard
under the HUD Lead Safe Housing Rule.

 All confidence intervals are at the 95% level for the estimated number
or percent.

  In the interpretation of the data by housing unit age, it is important
to keep the source of the data in mind.  Residents were asked the year
their home was constructed.  If a resident could not report the exact
year, he/she was asked to report the construction year in ranges: 
1978-1998, 1960-1977, 1946-1959, 1940-1945, and 1939 or before.  Over
40% of respondents provided the construction year of their home in this
secondary manner.  For the purposes of the data analyses in this report,
the midpoints of the ranges were assigned as the year of construction,
viz., 1988, 1968, 1953, 1943, and 1922, respectively.  The year 1922 is
not the midpoint of the pre-1940 range; it is the median construction
year for pre-1940 housing, according to the 1995 AHS.  For these
reasons, housing unit age is reported in the four ranges given in Table
3.1; finer breakdowns should be interpreted cautiously.

 A respondent was considered to be engaged in a lead-related occupation
if they worked on any of the following activities in the previous six
months:  building demolition, paint removal (including sanding or
scraping), plumbing, sandblasting, battery manufacturing or salvage
work, explosive or ammunition work, foundry work, glass work, lead
smelter work, oil refinery work, car radiator repair, making or splicing
cable, work at a firing range or police work, welding or torch cutting,
or other lead-related industry work.

 A respondent was counted as engaged in a potentially lead-related
activity at home if they had participated in any of the following
activities in the previous six months:  make bullets or fishing sinkers,
paint cars or bicycles, reload bullets, target shoot, hunt, remove paint
from any part of the house, remove paint from furniture, sand or paint
any part of the house, solder pipes or metal, solder electronic parts,
use artists’ paint (jewelry, pictures), work with stained glass, or
work with pottery or glazes.  However, painting and renovation work on
homes built after 1978 were excluded from these estimates.

 NCLSH (1998).  Evaluation of the HUD Lead-Based Paint Hazard Control
Grant Program, Fifth Interim Report, March 1998, Prepared for the U.S.
Department of Housing and Urban Development by the National Center for
Lead-Safe Housing.

 All confidence intervals are at the 95% level for the estimated number
or percent.

 In the interpretation of the data by housing unit age, it is important
to keep the source of the data in mind.  Residents were asked the year
their home was constructed.  If a resident could not report the exact
year, he/she was asked to report the construction year in ranges: 
1978-1998, 1960-1977, 1946-1959, 1940-1945, and 1939 or before.  Over
40% of respondents provided the construction year of their home in this
secondary manner.  For the purposes of the data analyses in this report,
the midpoints of the ranges were assigned as the year of construction,
viz., 1988, 1968, 1953, 1943, and 1922, respectively.  The year 1922 is
not the midpoint of the pre-1940 range; it is the median construction
year for pre-1940 housing, according to the 1995 AHS.  For these
reasons, housing unit age is reported in most tables in the four ranges
given in Table 3.1; finer breakdowns should be interpreted cautiously.

 The XRF analyzer used for the study corrected for substrate
interferences, but any actual lead in the substrate would be measured.

 It should be noted that the 1 percent of homes built in 1978-1998 with
a maximum paint lead loading above 10.0 mg/cm2 may be a result of
respondent error in reporting the date of home construction.  The year
reported by the respondent was never overridden. 

 A comparison of the protocols for the two surveys is presented in
Appendix B.

 For comparison, a room 10 feet by 12 feet with an 8 foot ceiling has a
wall area of 352 square feet and a combined wall, ceiling and floor area
of 592 square feet.

 The maximum lead dust loading on any surface tested (floor, window
sill, and window trough) in the home was used to determine whether a
dust lead hazard existed.  

 All confidence intervals are at the 95 percent level for the estimated
number or percent.

 The average analytical detection limit for each wipe sample was 3.5 (g.
 While detection limits for each surface are area dependent, this
corresponds to a detection limit of 3.5 (g/ft2 for a one square foot
floor sample, 7 (g/ft2 for a typical 3 inch by 24 inch sill sample, or 8
(g/ft2 for a typical 1 inch by 18 inch trough sample.

 From Table 5.2a, the percent of homes with sill dust loadings above 250
 µg/ft2 equals [100% (all homes) - 82% (homes with lead in sill dust
below 250 (g/ft2) - 2% (homes with missing data) - 2% (homes with no
sills)] = 14%. 

 For averaging floor samples, only carpeted floor samples and uncarpeted
floor samples were combined for the respective average (carpeted or
uncarpeted).

 The regression modeling also suggested that homes with no evidence of
cleaning had higher floor and window dust lead loadings (see Figure C.7
and Figure C.9).  Higher window dust lead loadings were suggested for
rented homes as compared to owned homes, surfaces that were not smooth
and cleanable as compared to smooth surfaces, and windows with vinyl
mini-blinds as compared to those without vinyl mini-blinds (see Figure
C.9).

 Dust lead loading data is presented in box plot form.  Each box plot
shows a univariate data distribution, for example, the dust samples
collected from a specific location (e.g., entrance floor). The box in
the box plot represents the middle 50 percent of the data; the bottom of
the box gives the 25th percentile; the top of the box gives the 75th
percentile, and the horizontal line inside the box gives the median or
50th percentile.  The vertical lines extending from the top and bottom
of the box reach to the largest and smallest observations, respectively,
except for outliers.

 The sample limit of detection for this study was determined to be 20
ppm by testing four distinct soil types from among the study samples in
accordance with EPA SW 840 Method 3050 procedures. 

 All confidence intervals are at the 95 percent level for the estimated
number or percentage.

  Even fewer homes will have greater than 9 square feet of bare soil
above 400 ppm or 2,000 ppm (soil lead hazard as specified in the Lead
Safe Housing Rule).

 Francek, M.  (1992.)  Soil lead levels in a small town environment: A
case study from Mt. Pleasant, Michigan.  Environmental Pollution 76. pp.
251-257.

 Soil lead concentration data is presented in box plot form.  Each box
plot shows a univariate data distribution, for example, the soil samples
collected from a specific location (e.g., main entry). The box in the
box plot represents the middle 50 percent of the data; the bottom of the
box gives the 25th percentile; the top of the box gives the 75th
percentile; and the horizontal line inside the box gives the median or
50th percentile.  The vertical lines extending from the top and bottom
of the box reach to the largest and smallest observations, respectively,
except for outliers.

 The sample limit of detection for this study was determined to be 20
ppm by testing four distinct soil types from among the study samples in
accordance with EPA Method 3050 procedures.

 Francek, M.  (1992.)  Soil lead levels in a small town environment: A
case study from Mt. Pleasant, Michigan.  Environmental Pollution 76. pp.
251-257.

 The sample limit of detection for this study was determined to be 20
ppm by testing four distinct soil types from among the study samples in
accordance with EPA Method 3050 procedures. 

 At the time the 1990 LBP survey was conducted, 500 ppm was the
guideline in EPA's Interim Guidance on Establishing Soil Lead Cleanup
Levels at Superfand Sites, September 7, 1989 (OSWER Directive
#9355-4-02).

 Another source of error in the survey is response bias, i.e., how
correct was the information provided by the respondents?  Significant
information obtained from respondents included year of construction (HU
age) and age of children.  These data were not verified by other means
and are thus associated with an unknown amount of error.  However, the
overall distribution of HU age and age of children reported by survey
respondents were consistent with study expectations (see Volume II,
Chapter 2), indicating no systematic bias in these responses.

 In the data collection completion rate, the ineligible cases are
included in both numerator and denominator.  If the ineligible cases are
not included, the data collection completion rate drops from 90 to 88
percent.  

 EPA’s National Lead Laboratory Accreditation Program (NLLAP) requires
wipes to be used in the field to have less than 5 (g lead/wipe.

 Personal communication with Sandy Roda, Director, Hematology and
Environmental Laboratory, University of Cincinnati.

 Rounding to the nearest tenth unit for classifying surfaces is
equivalent to classifying a surfaces as having LBP if the average lead
loading across the surface is 0.95 or greater.

 Whatever the explanation for the pattern in the data, the patterns are
not consistent with the assumptions behind the measurement error
adjustment.  Since no consistent pattern was found that would provide a
better model, the measurement error adjustment as described in Appendix
C (Volume II) was applied to the data.  The results provide one estimate
of the effect of measurement error on the assessment of the number of
homes with LBP.  However, other assumptions or other modeling approaches
may provide very different estimates of the magnitude of the effect of
measurement error.

 Care should be taken when comparing these data to tables presented in
Chapter 4. For example, a home was included in Table 4.1 if any one
component had LBP.  Values plotted in Figures 7.2 through 7.5 represent
the predicted median paint loadings in the home (i.e., for all
components).  Thus, while Table 4.1 shows that 53% of homes in the
Midwest have LBP, Figure C.3 shows that the median lead loading for all
components is lowest for homes in the Midwest.  This could be due to
there being fewer components with LBP or the lead content of individual
components could be lower in Midwest homes.

 These results suggest that cleaning activity will generally affect dust
lead loading.  

 The average of the available midyard and average of available dripline
samples were used for the measurement error analysis, regardless of
which side of the house they were collected.  That is, separate
assessments were not made for midyard and dripline areas on each side of
the home.

National Survey of Lead	Final Report, Vol. I

and Allergens in Housing	Analysis of Lead Hazards

ES- PAGE   x 

WESTAT		October 31, 2002

National Survey of Lead	Final Report, Vol. I

and Allergens in Housing	Analysis of Lead Hazards

WESTAT	1-  PAGE  6 	October 31, 2002

WESTAT	2-  PAGE  6 	October 31, 2002

WESTAT	3-  PAGE  4 	October 31, 2002

National Survey of Lead	Final Report, Vol. I

and Allergens in Housing	Analysis of Lead Hazards

WESTAT	4-  PAGE  17 	October 31, 2002

WESTAT	5-  PAGE  22 	October 31, 2002

National Survey of Lead	Final Report, Vol. I

and Allergens in Housing	Analysis of Lead Hazards

 PAGE   

WESTAT	6-  PAGE  14 	October 31, 2002

National Survey of Lead	Final Report, Vol. I

and Allergens in Housing	Analysis of Lead Hazards

7- PAGE   14 

WESTAT		October 31, 2002

7- PAGE   1 

WESTAT		January 9, 2001

National Survey of Lead	Final Report, Vol. I

and Allergens in Housing	Analysis of Lead Hazards

WESTAT	A-  PAGE  10 	October 31, 2002

National Survey of Lead	Final Report, Vol. I

and Allergens in Housing	Analysis of Lead Hazards

WESTAT	B-  PAGE  1 	October 31, 2002

National Survey of Lead	Final Report, Vol. I

and Allergens in Housing	Analysis of Lead Hazards

WESTAT	B-  PAGE  5 	October 31, 2002

WESTAT	B-  PAGE  2 	April 18, 2001

National Survey of Lead	Final Report, Vol. I

and Allergens in Housing	Analysis of Lead Hazards

WESTAT	C-  PAGE  1 	October 31, 2002

National Survey of Lead	Final Report, Vol. I

and Allergens in Housing	Analysis of Lead Hazards

WESTAT	D-  PAGE  2 	October 14, 2002

 

 

 

 

 

 

 

 

 

 

 

 

0

200

400

600

800

1000

1200

Window sill

Window trough

1750-1939

1940-1959

1960-1977

1978-1998

Privately owned

Publically owned

Rented

   African American

Other/Unknown

White

Appears clean

 Some evidence of cleaning

No evidence of cleaning

Unknown

Adult bedroom

Bathroom

Child bedroom

Kitchen

Living room

Other

Unknown

Smooth and cleanable

Not smooth/cleanable

No vinyl mini-blinds

Vinyl mini-blinds

Unknown

Relative Median Dust Lead Loading, 



g/ft

2

 (see text)

Sample 

location

Construction 

date

Tenure

Race

 category

Overall Household 

Cleanliness

Room type

Sample 

surface

Vinyl mini-

blinds

