United States
Environmental Protection
Agency
Office of Pollution
Prevention and Toxics
Washington, DC 20460
EPA747-R-99-002
March 1999
Lead Exposure Associated with
Renovation and Remodeling
Activities: Phase III
Wisconsin Childhood Blood-Lead Study
-------
EPA 747-R-99-002
March 1999
LEAD EXPOSURE ASSOCIATED WITH RENOVATION
AND REMODELING ACTIVITIES: PHASE III
WISCONSIN CHILDHOOD BLOOD-LEAD STUDY
Prepared by
BATTELLE
505 King Avenue
Columbus, Ohio 43201-2693
for
Technical Branch
National Program Chemicals Division
Office of Pollution Prevention and Toxics
U.S. Environmental Protection Agency
Washington, D.C. 20460
-------
U.S. EPA DISCLAIMER
This report was prepared under contract to an agency of the United States
Government. Neither the United States Government nor any of its employees,
contractors, subcontractors, or their employees makes any warranty, expressed or
implied, or assumes any legal liability or responsibility for any third party use of or
the results of such use of any information, apparatus, product, or process disclosed
in this report, or represents that its use by such third party would not infringe on
privately owned rights.
Publication of the data in this document does not signify that the contents
necessarily reflect the joint or separate views and policies of each sponsoring agency.
Mention of trade names or commercial products does not constitute endorsement or
recommendation for use.
BATTELLE DISCLAIMER
This is a report of research performed for the United States Government by
Battelle. Because of the uncertainties inherent in experimental or research work,
Battelle assumes no responsibility or liability for any consequences of use, misuse,
inability to use, or reliance upon the information contained herein, beyond any
express obligations embodied in the governing written agreement between Battelle
and the United States Government.
-------
ACKNOWLEDGMENTS
Five institutions participated in designing and conducting the Wisconsin Childhood
Blood-Lead Study. The State of Wisconsin Bureau of Public Health, in cooperation with the
University of Wisconsin-Madison, conducted the retrospective study. Key personnel were
Dr. Marty Kanarek (Principal Investigator) and Tiffany Taha of the Department of Preventive
Medicine at the University of Wisconsin and Joe Schirmer of the State of Wisconsin. The
University of Wisconsin-Madison was responsible for human subjects approval, preparing the
database, and coordinating between the University and the State of Wisconsin. Dr. Kanarek
designed the study and questionnaire and provided overall guidance. Joe Schirmer developed the
initial questionnaire. Larry Hanrahan, Wisconsin Bureau of Public Health, helped design the
study; and Barbara Haley, U.S. Department of Housing and Urban Development, helped develop
the questionnaire.
Telephone interviews were conducted by the University of Wisconsin Extension Survey
Research Lab. Lead personnel were Nancy Davenport and Stig Nelson. The Survey Research
Lab was also responsible for quality assurance. Battelle assisted in designing the study and
questionnaire, conducted the data analysis, and prepared this report. Drs. Nancy McMillan and
Ron Menton were the key Battelle personnel involved.
The Environmental Protection Agency (EPA) funded the study through the University of
Wisconsin. Dan Reinhart was the key EPA contact. Battelle's efforts were funded by EPA
under Contract Number 68-D5-0008, Work Assignments 1-14, 2-14, 3-20, and 4-05. Dan
Reinhart was the Work Assignment Manager.
This report benefitted significantly from comments and suggestions provided by four peer
reviewers. The reviewers were Mr. John Rogers (Westat) and Drs. Janie Gittleman (NIOSH),
Bruce Lanphear (Children's Hospital Medical Center), and Dori Reissman (CDC).
in
-------
TABLE OF CONTENTS
Page
ACKNOWLEDGMENTS iii
EXECUTIVE SUMMARY vii
1.0 INTRODUCTION AND BACKGROUND 1
1.1 PEER REVIEW 2
2.0 STUDY DESIGN 5
2.1 ELIGIBILITY REQUIREMENTS 5
2.2 SAMPLE SELECTION 5
2.3 QUESTIONNAIRE DESIGN AND IMPLEMENTATION 6
2.4 STUDY LIMITATIONS 8
3.0 DATA SUMMARY AND STATISTICAL ANALYSIS 10
3.1 DATA SUMMARY 11
3.1.1 Available Blood-Lead Concentration Data 11
3.1.2 Characteristics of Sampled Children 14
3.1.3 R&R Exposure Characterization 16
3.2 DATA QUALITY 24
3.2.1 Blood-Lead Registry and Interview Data 25
3.2.2 Assessment of Nonresponse and Selection Bias 26
3.2.3 Assessment of Skin Contamination 29
3.3 STATISTICAL ANALYSIS 32
3.3.1 Univariate Statistical Analyses 33
3.3.2 Multivariate Statistical Analyses 37
3.3.3 Statistical Results 49
4.0 CONCLUSIONS 51
5.0 REFERENCES 52
APPENDIX A: EXPLORATORY ANALYSIS OF WISCONSIN CHILDHOOD BLOOD-LEAD STUDY
QUESTIONNAIRE DATA A-1
APPENDIX B: LOGISTIC REGRESSION BACKGROUND B-1
APPENDIX C: TEST-WISE SIGNIFICANT RESULTS C-1
APPENDIX D. UNIVARIATE REGRESSION OF CONSTRUCTED R&R VARIABLES D-1
LIST OF TABLES
Table 2-1. Sample Selection Summary 7
Table 2-2. Summary of Questionnaire 8
Table 3-1. Summary Statistics for Medical Assistance, Race, Gender, Method of Blood Draw,
Date Samples Collected, and Age of Child 12
Table 3-2. R&R Activity Exposure Summary 18
IV
-------
Table 3-3a. Inside Painting Specific Exposure Summary 20
Table 3-3b. Other Repairs (Windows) Specific Exposure Summary 21
Table 3-3c. Carpets and Floors (Repair or Replacement) Specific Exposure Summary 22
Table 3-4. Summary of Demographic Information for Participants and Nonparticipants ... 28
Table 3-5. Comparison of Participation Between Cases and Controls 30
Table 3-6. Six Capillary Blood-Lead Concentration Measurements with Possible Skin
Contamination Bias 31
Table 3-7. Assessment of Skin Contamination of Capillary Blood Samples for Six Age
Groups of Children 31
Table 3-8. Significant (Based on the Holm Procedure) Univariate Odds Ratio Estimates
and Confidence Intervals3 35
Table 3-9. Variables Considered for the Baseline Logistic Regression Model 38
Table 3-10. Odds Ratio Estimates and Confidence Intervals for Baseline Logistic Regression
Model 41
Table 3-11. Renovation and Remodeling Variables Used to Assess Relative Risk 42
Table 3-12. Unconditional Odds Ratios from Logistic Regression with Single R&R Variables . 44
Table 3-13. Conditional Odds Ratios from Logistic Regressions with Single R&R Variables
(and Any Other R&R Variables Nesting Them) 45
Table 3-14. Conditional Odds Ratios from Logistic Regression with Multiple R&R Variables . 47
Table 3-15. Breakdown of Type of Specific R&R Activity by Type of Individual Who
Did Work 48
LIST OF FIGURES
Figure 3-1. Trend Plot to Assess the Seasonality of Blood-Lead Concentrations in the
Registry Database 17
Figure 3-2. Odds Ratio Comparing Odds of an Elevated Blood-Lead Concentration for a
Child of Age Specified by the x-Axis to a 1-Year-Old Child's Odds 40
Figure 3-3. Odds Ratio Comparing Odds of an Elevated Blood-Lead Concentration for a
Child in a Household with Number of Residents Specified by the x-Axis to a
Child in a Household with Three Residents 40
-------
This page intentionally blank.
VI
-------
EXECUTIVE SUMMARY
Lead poisoning is considered a serious threat to health in the United States. Since the use
of lead in gasoline has been phased out, exposure to lead is now primarily from lead-based paint,
particularly for children and construction workers. Federal programs undertaken to mitigate
exposure to lead-based paint have focused on deteriorated paint and methods of abatement. As a
result, the potential for exposure of workers and building occupants to lead resulting from
renovation and remodeling (R&R) conducted with no abatement intent has not been evaluated.
To address this potential, the United States Congress directed the U.S. Environmental
Protection Agency (EPA) to
• Determine the extent to which persons engaged in various types of residential R&R
activities are exposed to lead
• Determine the extent to which persons engaged in various types of R&R activities
disturb lead and create a lead-based paint hazard (on a regular or occasional basis) to
building occupants or other exposed individuals.
To meet these objectives, the EPA launched a series of studies in 1993 collectively
known as the R&R study. The first of these studies was the Environmental Field Sampling
Study (EFSS or Phase I). In this study, lead levels in settled dust and in breathing zone air
resulting from a variety of R&R activities (window replacement, sanding, paint preparation,
demolition, cutting painted wood, etc.) were measured. These lead measurements were assumed
to indicate the potential for lead exposure to construction workers and to building occupants as a
result of R&R activities.
The second study, the Worker Characterization and Blood-Lead Study (WCBS or
Phase II), examined the relationship between the blood-lead concentrations of construction
workers and their R&R activities, work habits, medical histories, hobbies, etc.
In general, the results of Phases I and II indicated that, for some R&R activities, airborne
lead levels within workers' breathing zones often exceeded the Occupational Safety and Health
Administration's permissible exposure limit, but the blood-lead concentrations of workers that
regularly engage in these activities were not likely to be elevated (^ 10 /ug/dL). The implications
for building occupants (especially children), however, were not as clear. The potential for
exposure to occupants was characterized in the EFSS (Phase I) by measuring lead levels in dust
that was generated by various R&R activities. The results of Phase I indicated that most R&R
activities have a potential to disturb substantial amounts of lead and that occupants could be
exposed to lead if appropriate containment and cleanup precautions are not employed. However,
much uncertainty remained concerning the extent to which this potential for lead exposure
translated to an actual internal dose for occupants (especially children).
Because children represent the population that is most likely to be sensitive to lead
exposure from R&R activities, it became imperative that the EPA acquire additional data to
assess the impact of residential R&R on children. Therefore, a third study, the Wisconsin
vii
-------
Childhood Blood-Lead Study was conducted to determine the impact of residential R&R on the
blood-lead concentrations of children occupying the residences.
The Wisconsin Childhood Blood-Lead Study (Phase III) was a retrospective case-control
study designed to systematically examine the association between R&R activities and elevated
blood-lead (EBL) levels (* 10 (ig/dL) among children. The primary objective of the study was to
compare the incidence of R&R activities in the residences of children with EBLs to R&R
activities in the residences of children without EBLs. Another objective of the study was to
determine if specific R&R activities were more prevalent in households with an EBL child than
in households with a non-EBL child. The study targeted children under the age of six who were
included in the Wisconsin Bureau of Public Health's blood-lead registry. These children resided
in communities other than Milwaukee and Racine and were screened between March 1996 and
December 1996.
To meet these objectives, telephone interviews were conducted with the parents or
guardians of 3,654 children under the age of six about R&R activities in their residences.
Responses during the telephone interview, as well as the child's blood-lead concentration as
recorded in the Wisconsin Bureau of Public Health's blood-lead registry, were used to assess
whether R&R, hi general, increases a child's risk of having an elevated blood-lead concentration.
The telephone questionnaire and blood-lead information were also used to determine if specific
R&R activities were associated with an increased risk of an elevated blood-lead concentration.
The questionnaire was designed to determine whether R&R activities such as inside
painting, outside painting, carpet and floor repair or replacement, and other repairs (such as
window repair) were conducted in the residences of the children in the study. Although it is
difficult to generalize the information beyond the study population, the frequency of general and
specific R&R activities in residences populated by young children is a valuable product of this
study. While the main objective of this study was to investigate the relationship between
incidences of R&R activities and EBL children, a dose-response type relationship, this
information is most useful when the degree of exposure is also known. In this case, if a slight
increase in risk due to childhood lead exposure as a result of R&R activity is detected, the
implications are best considered in light of the numbers of children likely to be exposed.
Analysis of the exposure data related to these activities revealed that at least one R&R
activity had been conducted in 67.2 percent of the residences in the previous 12 months. Inside
painting occurred hi 50 percent of residences and outside painting, carpet and floor repair or
replacement, or other repairs occurred in 20 percent. Some form of surface preparation was
involved in 42.3 percent of R&R activities. Approximately 65 percent of the outside painting
involved surface preparation, compared to 15 percent to 30 percent for each of the other types of
activities. Most surface preparation involved hand scraping or sanding. Heat guns were used
7 percent of the time, and chemical paint removers were used 13.6 percent of the time. Surface
preparation for painting was usually performed by the owner of the residence, the building
superintendent, or apartment staff, while preparation for carpet and floor repair or replacement or
other repairs was usually performed by paid professional contractors.
VIII
-------
Univariate and multivariate logistic regression models developed from the questionnaire
and blood-lead data collected for this study were used to
1. Determine if the incidence of R&R in residences was associated with an increased
risk of an EBL (;>10 ug/dL).
2. Determine if specific R&R activities were associated with an increased risk of a child
having an EBL.
The study demonstrated that residential R&R is associated with an increased risk of an
EBL in children. The study also demonstrated that specific R&R activities are associated with
an increase in the risk of an EBL in children. In particular, painting inside or outside, removing
paint (using open flame torch, using heat guns, using chemical paint removers, and wet
scraping\sanding), preparing surfaces by sanding or scraping, and living in a home when R&R
work was done significantly increased the risk of EBLs. An EBL was more likely when a
relative or friend not living in the household did the R&R work. In addition, the more rooms
involved in the total R&R project, the more likely a child was to have an EBL. Any type of
R&R work in the kitchen increased the odds of an EBL.
Overall, these results agree with those from earlier phases of the R&R Study — R&R
activities that disturb lead-based paint increase the risk of exposure to occupants. For example,
children living in a residence while R&R was conducted were 1.3 times more likely to have
EBLs than children who did not live in a residence while R&R was conducted.
Further, the study has identified specific R&R activities and other conditions (such as age
and type of residence) that are associated with increased risk to children as a result of lead
exposure. This information can be used to develop regulations that focus on particular R&R
activities (e.g., using a heat gun to remove paint), groups of persons (e.g., a household member
other than the head of household or spouse) who perform the activities, and the other conditions
(e.g., adult exposure, age of child) that significantly increase the risk to children. The results of
this study concerning activities associated with increased risk (e.g., using a heat gun to remove
paint) also can be combined with the worker profile results from Phase II to perform an overall
assessment of the worker groups or situations where interventions are needed to reduce exposure
from R&R.
IX
-------
This page intentionally blank.
-------
1.0 INTRODUCTION AND BACKGROUND
Lead poisoning is considered a serious threat to health in the United States. Since the use
of lead in gasoline has been phased out, exposure to lead is now primarily from lead-based paint
(CDC, 1991). To address the potential for this type of exposure, in 1992, the United States
Congress enacted the Residential Lead-Based Paint Hazard Reduction Act (Public Law
102-550), which required the U.S. Environmental Protection Agency (EPA) to conduct a study of
lead exposure associated with renovation and remodeling activities (R&R study). In particular,
paragraph (2) of Section 402(c) states:
The Administrator shall conduct a study of the extent to which persons engaged in
various types of renovation and remodeling activities in target housing, public buildings
constructed before 1978, and commercial buildings are exposed to lead in the conduct of
such activities or disturb lead and create a lead-based paint hazard on a regular or
occasional basis.
The overall objectives of the R&R study were to
• Determine the extent to which persons engaged in various types of residential R&R
activities are exposed to lead
• Determine the extent to which persons engaged in various types of R&R activities
disturb lead and create a lead-based paint hazard (on a regular or occasional basis) to
building occupants or other exposed individuals.
To meet these objectives, a series of studies, known collectively as the R&R Study, was
launched in 1993 by EPA:
• Phase I: The Environmental Field Sampling Study (EFSSX TheEFSSwas
conducted to measure the airborne lead levels and lead levels in settled dust resulting
from R&R activities (EPA, 1997a, and EPA, 1997b).
• Phase II: The Worker Characterization and Blood-Lead Study (WCBSX TheWCBS
was designed to collect questionnaire and blood-lead data that could be used to assess
the relationship between R&R activities and lead exposure for the R&R workers
conducting these activities (EPA, 1997a, and EPA, 1997c).
In Phase I, lead levels in settled dust and in breathing zone air that resulted from a variety
of R&R activities (window replacement, sanding, paint preparation, demolition, cutting painted
wood, etc.) were measured. These environmental lead measurements were assumed to indicate
the potential for lead exposure to R&R workers and to building occupants resulting from R&R
activities. In Phase II, the relationship between blood-lead levels among R&R workers and
variables such as their work activities, work habits, medical histories, and hobbies, was
examined.
-------
In general, Phases I and II indicated that, for some R&R activities, airborne lead levels
within workers' breathing zones often exceeded the Occupational Safety and Health
Administration's permissible exposure limit, but worker blood-lead concentrations were not
likely to be elevated. The implications for building occupants (especially children), however,
were not as clear. The potential for exposure to occupants was characterized in Phase I by
measuring lead levels in settled dust that was generated or released by various R&R activities.
Phase I indicated that most R&R activities have the potential to disturb substantial amounts of
lead to which occupants could be exposed if appropriate containment and cleanup precautions
were not employed. However, much uncertainty remained concerning the extent to which this
potential for lead exposure translated to an actual internal dose for occupants (especially
children).
Because children represent the population that is most likely to be sensitive to lead
exposure from R&R activities, it became imperative that the EPA acquire additional data to
assess the impact of residential R&R on children. Therefore, a third study, the Wisconsin
Childhood Blood-Lead Study was conducted to determine the impact of residential R&R on the
blood-lead concentrations of children occupying the residences.
The Wisconsin Childhood Blood-Lead Study (Phase III) was designed to systematically
examine the association between R&R activities and elevated blood-lead levels (EBL
^ 10 ug/dL) among children. To overcome the major obstacles associated with collecting
children's blood samples for a prospective study (problems with human subjects approval,
parental permission, low participation rates, etc.), Phase III used the Wisconsin Bureau of Public
Health's registry of children's blood-lead measurements.
The primary objective of the study was to compare the incidence of R&R activities in the
residences of children with an EBL to incidence of R&R activities in the residences of those
without an EBL. Furthermore, the study was designed to determine if specific R&R activities
were more prevalent in residences of EBL children than in residences of children who did not
have EBLs. To meet these objectives, telephone interviews were conducted with the parents or
guardians of 3,654 EBL and non-EBL children about R&R activities in their residences. The
study targeted children under the age of six who resided in communities other than Milwaukee
and Racine and whose blood-lead concentrations were screened between March 1996 and
December 1996.
1.1 PEER REVIEW
This report has benefirted significantly from comments and suggestions provided by four
external peer reviewers. Each of the reviewers recommended publishing the report after minor
revisions. Comments which had a significant impact on the report or which aid in interpreting
the study results are discussed below.
In response to comments by a number of the reviewers, more details regarding the study
design were provided. In particular, certain design decisions which placed limitations on the
study results were emphasized. Because recall bias was a concern for several reviewers a table
was added (Table 2-1) to summarize time delays between blood sampling and administration of
-------
the questionnaire. The questionnaire used in this study was designed to facilitate recall, and
therefore, minimize recall bias.
The requirement that all study participants have known medical assistance status is a
limitation of this study that was noted by multiple reviewers. Prior to performing the study, it
was hypothesized that the impact of R&R activities on EBLs may be different between the two
medical assistance populations. If this hypothesis were true, separating these two populations
would have been instrumental in determining the relationship between R&R activities and EBLs.
Since the Wisconsin registry did not contain numbers of medical assistance and non-medical
assistance children typical of Wisconsin in general, representativeness of the study data with
respect to medical assistance status was not affected by this decision. The effect of medical
assistance status on incidence of EBLs was accounted for by incorporation of the medical
assistance status variable in a baseline model of EBL incidence prior to consideration of the
effect of R&R activities. Thus, study results are believed to be applicable to both medical
assistance and non-medical assistance populations. However, the limitation inherent in the
decision to only include children with known medical assistance status is that a large number of
children, possibly many with EBLs, were excluded from the study.
At the request of several reviewers, more detail on the Wisconsin blood-lead registry was
provided. In particular, quality control during blood sample collection and analysis was an issue
for the reviewers. Quality control of analysis is regulated by CLIA, the clinical laboratory
improvement amendments of 1988. However, quality control of the sample collection is up to
the sample collector. Several Wisconsin state organizations work with sample collectors to
inform them of quality control issues that may arise.
In regard to the Wisconsin blood-lead registry, another issue of concern among the
reviewers was how representative the registry is of the state population. A comparison of the
racial distribution across the state (excluding Milwaukee and Racine counties) to the racial
distribution selected for this study revealed that minorities are likely over represented in the
registry. This finding does not limit the main study conclusions as exact representativeness of
the data with respect to an explanatory variable is not essential for determination of the
relationship between another explanatory variable (R&R activities) and a response (incidence of
EBL).
Based on the comments of one reviewer, a new multivariate analysis of the study data
was added to the report. This analysis demonstrates the impact of simultaneously considering
many R&R activities to determine those most associated with increased odds of an EBL. Results
from this analysis were very consistent with analyses considering R&R activities individually.
At least one reviewer was concerned that there appeared to be no difference in risk
between R&R activities performed in pre- and post-1980 homes. Additional discussion of the
statistical analysis used to determine this finding was added to the report to allay this concern.
There are many possible reasons for this finding, including:
1. Age of residence was an interview reported variable, and, thus, may have been
reported inaccurately for some homes.
-------
2. Only the manufacturing of lead-based paint was banned in 1978. Residences may
have been painted with old lead-based paint.
3. R&R activities in residences may increase risk due to childhood lead exposure by
stirring up lead contaminated dust from sources other than lead-based paint.
EPA has established a public record for the peer review under administrative record
AR-209. The record is available in the TSCA Nonconfidential Information Center, which is
open from noon to 4 PM Monday through Friday, except legal holidays. The TSCA
Nonconfidential Information Center is located in Room NE-B607, Northeast Mall,
401 M Street SW, Washington, D.C.
-------
2.0 STUDY DESIGN
The Wisconsin Childhood Blood-Lead Study was a retrospective case-control study of
the association between EBLs in children and the incidence of R&R activities in their residences.
Blood-lead concentration data were obtained from screening data recorded in the Wisconsin
Bureau of Public Health's blood-lead registry. A subset of children were selected from the
registry for participation in the study based on several eligibility requirements (Section 2.1). All
eligible children with a blood-lead concentration greater than or equal to 7 ^g/dL were selected
into the study. Random selection was used to select among the eligible children with blood-lead
concentrations less than 7 ug/dL. Sample selection (Section 2.2) was a sequential process
consisting of separate samples drawn monthly from screening data compiled between the months
of March and December 1996. Information on R&R activities in participating children's
residences was obtained by administering a questionnaire (Section 2.3) over the telephone. To
the extent feasible, this study collected interview data as soon as possible once each month's
blood screening data was available in the registry.
2.1 ELIGIBILITY REQUIREMENTS
The sampling frame consisted of children under the age of six who were screened for
blood-lead concentrations from March 1996 to December 1996. Only one child per household
was eligible for the study. All eligible children with a blood-lead concentration greater than or
equal to 7 ug/dL were selected into the study. Random selection was used to select among the
eligible children with blood-lead concentrations less than 7 |ig/dL.
All the children selected were from communities outside Milwaukee and Racine.
Children from these cities were omitted for two reasons. First, these cities have a higher
incidence of lead poisoning compared to the state as a whole. This higher incidence of lead
poisoning may be a result of a greater density of older, deteriorated housing, which would make
it more difficult to isolate the effect of lead exposure from renovation activities on the children.
Second, the health departments in these communities manage their data independently, which
would have resulted in delays in contacting the families directly. In designing the study, it was
hoped that the possibility of finding an effect on blood-lead concentrations resulting from R&R
would be maximized by studying a large, non-inner city population.
In Wisconsin, blood-lead screening is mandatory for children receiving medical
assistance (Medicaid) and voluntary for others. As a result, the proportion of families receiving
medical assistance in the Wisconsin blood-lead registry may be higher than the proportion of
such families in the state. Therefore, separate sampling frames were constructed for families on
medical assistance and for families not on medical assistance. Known medical assistance status
was required for inclusion of a child into the study.
2.2 SAMPLE SELECTION
Using screening data collected from March 1996 to December 1996, a sample was drawn
from the Wisconsin blood-lead registry for each month. Monthly samples were selected
approximately sequentially; i.e., the May sample was selected before the June sample. (Because
-------
sample sizes achieved each month were smaller than anticipated, samples for the months of
March and April were added to the study after the May and June samples were completed.) The
sampling frame for each month contained all eligible children whose first blood-lead sample was
collected in that month, and who did not have another household member already included in the
study. When first blood-lead data were collected during the same month for more than one child
in the same family, one child was randomly selected to be eligible for the study.
Initially, 100 percent sampling was conducted for cases (children with a blood-lead
concentration greater than or equal to 10 ug/dL), and a random sample was drawn from the
controls (children with a blood-lead concentration less than 10 ug/dL). For each sampling frame,
however, the number of children with an EBL was smaller than expected. Therefore, the number
of cases selected, even with complete sampling, was smaller than anticipated. To compensate for
this, the study design was modified prior to selecting the August sample to include 100 percent
sampling for children with a blood-lead concentration between 7 and 10 ug/dL. Thus, before the
August sample was collected eligible children from the months of March through July with
blood-lead concentrations between 7 and 10 ug/dL were added to the study.
Random selection of children with blood-lead concentration less than 7 ug/dL was
performed separately for medical assistance and non-medical assistance children. Monthly
samples for March though September targeted 450 children with blood-lead concentration less
than 10 ug/dL from each group. Monthly samples for October through December targeted
250 children with blood-lead concentration less than 7 ng/dL from each group. Some months,
one of the groups did not have enough children to meet the target. For that group, 100 percent
sampling was performed.
Table 2-1 provides information regarding the monthly samples selected for this study. In
Table 2-1 the month that each sample was selected is noted. To minimize the time between
blood sampling and interview data collection, the sample drawn for each month was immediately
sent to Survey Research Labs to begin telephone interviewing. On average, there was a 4-month
delay between blood screening and administration of the questionnaire. In general, the lag time
was smaller toward the end of the study. The change in numbers of low blood-lead
concentration children targeted in the monthly sampling that occurred between September and
October can be clearly seen in Table 2-1.
2.3 QUESTIONNAIRE DESIGN AND IMPLEMENTATION
The limited resources available to conduct this study in an area as geographically diverse
as the State of Wisconsin precluded in-person interviews as a method for obtaining the
questionnaire information. Further, to obtain information quickly (thus reducing the potential for
recall bias), and to increase the participation rate, a telephone interview approach was adopted
rather than a mail questionnaire. (Groves et al., 1988, discusses the advantages and
disadvantages of telephone survey methods.)
-------
Table 2-1. Sample Selection Summary
Month
May
June
April
March
July
August
September
October
November
December
Sample
Selection Date
September 1996
September 1996
October 1996
October 1996
October 1996
January 1997
January 1997
January 1997
February 1997
February 1997
Number of
Children
Eligible
1458
1328
1286
1153
1528
1499
1405
1240
1005
914
Children Selected
PbB<7 ug/dL
855
767
709
686
805
843
745
493
486
474
7ug/dUPbB<10ug/dL
134
75
104
93
120
132
114
91
66
63
PbB;>10 ug/dL
85
71
92
76
121
113
79
61
64
41
With any survey requesting information on events that occurred in the past, one potential
for bias is due to the inherent difficulty of recalling past activities. Activities conducted in the
recent past are more easily recalled than activities conducted in the more distant past. Therefore,
delays between blood-lead sampling and administering the questionnaire could bias results
(i.e., R&R activities occurring further in the past may not be "remembered" or participants may
provide less reliable information) As noted in the previous section, the time delay between
blood-lead sampling and questionnaire administration was, on average, four months. The
questionnaires used in this study were designed to facilitate recall and, therefore, minimize this
source of bias. For example, in addition to asking participants when and for how long (hours or
days) they performed an activity, the participants were also simply asked whether or not they
performed that activity (easier questions to recall and therefore less prone to recall bias).
An attempt was made to contact by telephone the parent or guardian of each child
selected for the study. Because phone numbers and addresses were missing for some children, it
was not possible to contact a parent or guardian for each selected child. If the parent or guardian
was contacted, a telephone questionnaire (see Appendix A) was administered. The telephone
interview collected data on, among other things, specific R&R activities that may have been
related to the child's blood-lead concentration. Table 2-2 summarizes the type of information
elicited by the questionnaire.
The questionnaire was pretested using 13 respondents. The interviews averaged
13.4 minutes. As a result of the pretest, questions about the occupations and hobbies of adults in
the household, eating habits, painting preparations, and painting done inside the home were
added to the questionnaire, which was tested on an additional 32 people. Obtaining this
additional information added an average of 4.6 minutes to the interview.
-------
Most of the respondents in the study were interviewed in English; but, because a
substantial number of Hmong and Spanish families were expected in the sample, the interview
was translated into Hmong and Spanish to ensure consistent questioning. Interviews were
conducted in Hmong with 103 families, or 2.8 percent of the interviewees, in Spanish with 44, or
1.2 percent of the interviewees, and in English with 3,508, or 96.0 percent of the interviewees.
Table 2-2. Summary of Questionnaire
Type of Information
Child's Residence
Duration of Residence
R&R Activities
Blood Sampling
Household Information
Adult Occupations
Hobbies
Income and Address
Rationale
Includes questions on age of dwelling, if buying or renting, and
presence of peeling paint. This information may be related to child's
blood-lead concentration.
Includes questions on whether child lived in current home longer than
1 2 months. This information was used to ensure that information
regarding child's residence (i.e., age of residence) was accurate at
time of blood collection.
Includes questions about the conduct of specific R&R activities in the
past 1 2 months, such as how was the work done, who did it, and
where it was done. Targeted activities were interior painting, exterior
painting, window repair, carpet removal, and wall repair.
Includes questions on how many and why blood samples were taken.
This information was used to help ensure that blood-lead
concentration was not reduced due to medical treatment.
Includes questions on number of household residents, education
attained by parent or guardian, and family dining habits. This
information may be related to lead exposures.
Includes questions about occupations that may be confounded with
lead exposures in the household.
Includes questions about hobbies that may be confounded with lead
exposures in the household.
Includes questions on income and child's address. Income and
neighborhood may be associated with child's blood-lead concentration.
2.4 STUDY LIMITATIONS
As with all studies, there are limitations associated with study design decisions.
Limitations of this study include
1. Restriction to only registry children with known medical assistance status
2. Differential selection probabilities among the low blood-lead concentration children
(<7 ug/dL) across the study months
3. Exclusion of Milwaukee and Racine children
4. Recall bias due to the time delay between blood-lead sampling and questionnaire
administration
5. Exclusion of possible confounding variables from the questionnaire.
-------
Each of the above limitations is due to the inevitable trade-off between the practical issues
associated with implementing a study and the desire to obtain the best data possible.
The study was restricted to include only children with known medical assistance status
because, prior to performing the study, it was hypothesized that the impact of R&R activities on
EBLs may be different between the two medical assistance populations. If this hypothesis were
true, separating these two populations would have been instrumental in determining the
relationship between R&R activities and EBLs. Since the Wisconsin registry did not contain
numbers of medical assistance and non-medical assistance children typical of Wisconsin in
general (screening of medical assistance children is mandatory), representativeness of the study
data with respect to medical assistance status was not affected by this decision. However, the
limitation inherent in the decision to only include children with known medical assistance status
is that a large number of children, possibly many with EBLs, were excluded from the study.
Differential selection probabilities among the low blood-lead concentration children
(<7 ug/dL) across the study months adversely affects the ability to detect seasonality effects in
the data. While this limitation is unfortunate, interviewing low blood-lead concentration children
during the last months of the study was restricted to boost the number of EBLs in the study.
Having only a small percentage of the study children in the elevated group reduces the power to
detect the effect of R&R activities on the incidence of EBLs.
Exclusion of Milwaukee and Racine children from the study has the unfortunate side
effect of reducing the number of EBL children eligible for the study. However, a number of
practical concerns, as discussed in Section 2.1, dictated this decision.
As with any survey requesting information on events that occurred in the past, one
potential for bias in the relationship between the conduct of R&R activities and children's
blood-lead levels is due to inherent difficulties associated with recalling activities that occurred
in the past. This limitation is described in Section 2.3.
The extensive telephone interview employed in this study required about 18 minutes per
interview. A variety of information (adult occupations, hobbies, household information) was
solicited during this interview in addition to the R&R activity data. However, other variables
that affect the relationship between R&R activities and incidence of EBLs may not have been
obtained. Possible candidates include use of traditional medicines or use of ceramics in cooking
and serving food. These variables are particularly important among the Hmong and Spanish
households surveyed.
In spite of the above limitations, this study provides valuable data for estimating the
effect of R&R activities on the incidence of EBLs among a diverse population of children.
-------
3.0 DATA SUMMARY AND STATISTICAL ANALYSIS
This study evaluated the impact of residential R&R on children's blood-lead
concentrations by comparing the incidence of R&R activities in residences with an EBL child to
R&R in residences with a non-EBL child. Specifically, analyses were conducted to
• Determine if the incidence of R&R activities in residences was associated with an
increased risk of an EBL
• Determine if specific R&R activities were associated with an increased risk of a child
having an EBL.
Based on the data collected, exposure of residents to R&R activities in study homes was also
characterized. This information provides an assessment of the portion of residences affected by
hazards due to R&R activities. This chapter contains a discussion of the data collected and
characterization of exposure to R&R activities (Section 3.1), the potential problems with the data
(Section 3.2), and the analysis of the data to determine the impact of R&R on the probability of
an elevated blood-lead concentration in children (Section 3.3).1
The data, summarized in Section 3.1, includes information on 8,651 Wisconsin children
residing outside of Milwaukee and Racine whose first blood-lead concentration measurement
was collected (and recorded in the Wisconsin Bureau of Public Health's blood-lead registry)
between March and December of 1996. Interview data was collected for approximately
42 percent (3,654) of these children and is reported in Section 3.1.
A number of data quality concerns, specifically quality control procedures used during
the collection of blood-lead registry and interview data, are discussed (Section 3.2.1). Because
nonresponse and sampling bias are of considerable concern when interview data are collected,
nonresponse and sampling bias are also discussed (Section 3.2.2). Skin contamination was a
concern in this study because the Wisconsin blood-lead registry consists primarily of
measurements collected by the capillary method. Confirmatory venous blood-lead
measurements (taken when a child's first blood-lead measurement is very high) indicated the
possibility of skin contamination for four samples (Section 3.2.3).
Data collected in this study were analyzed using logistic regression, a statistical method
used to explain the relationship between a dichotomous (yes/no) response variable and a suite of
explanatory variables (McCullagh and Nelder, 1989). A number of social and demographic
variables are known to be associated with elevated blood-lead concentrations (Bomschein et al.,
1985; Brody et al., 1994; and Lanphear et al., 1996). Analysis were conducted to examine the
impact of these variables prior to considering the effects due to R&R. A baseline logistic
regression model was developed to explain the odds of an elevated blood-lead concentration as a
function of such social and demographic variables as the child's age, the age of the residence,
Elevated blood-lead concentrations are defined as ;> 10 ng/dL (CDC, 1994) and nonelevated blood-lead
concentrations as less than 7 ng/dL.
10
-------
and family income. R&R variables were then entered into this model and their statistical
significance assessed. Conclusions on the magnitude and significance of particular R&R
variables, calculated in this manner, are found in Section 3.3.2. Appendix B provides
background information on logistic regression, including an explanation of why this method was
used to analyze data from this study.2
3.1 DATA SUMMARY
Between March and December of 1996, the first blood-lead concentration measurements
of approximately 50,000 children residing outside of Milwaukee and Racine were recorded in the
Wisconsin Bureau of Public Health's blood-lead registry. Multiple blood-lead measurements
were reported for some of these children, generally as confirmatory measurements when the first
measurement was elevated. (See CDC, 1991, for childhood blood-lead screening
recommendations.) A total of 8,651 of these children were selected to participate in this study.
Interviews were conducted for 3,654 of the 8,651 children. Four children who were interviewed
had blood-lead concentrations that were believed to be mistaken and were dropped from the
study (see Section 3.2.3).
This study consists of data on the remaining 3,650 children. Some analyses presented in
this report used the larger database consisting of all 8,651 children selected to the study. This
larger database is referred to as the registry database. Analyses based on the registry database are
clearly indicated.
3.1.1 Available Blood-Lead Concentration Data
The Wisconsin blood-lead registry contained data on the following variables:
• Medical assistance status (Medicaid)
• Race
• Gender
• Blood sample collection technique
• Month that first blood-lead sample was collected
• Age
• Blood-lead group (number of low, medium, and high observations)
• Blood-lead concentration.
The first nine and the last columns of Table 3-1 present a summary of the data available for each
of these variables. Appendix A shows similar descriptive statistics for data resulting from the
interview responses. In addition to counts of children by levels of demographic variables and
blood-lead group, geometric means and geometric standard deviations are calculated by levels of
demographic variables.
Since the individual observations were selected into the study based on blood-lead concentration, using
blood-lead concentration as a response variable was not appropriate. Logistic regression of the
categorical high/low blood-lead group variable is the appropriate analytical procedure.
11
-------
Table 3-1. Summary Statistics for Medical Assistance, Race, Gender, Method of Blood
Draw, Date Samples Collected, and Age of Child
Medical
Assistance
Race
Gender
Blood
Sample
Collection
Technique
Month Blood
Collected
Age
Response
Category
Medical
Assistance
Other
Caucasian
African
American
Asian
Native
American
Unknown
Male
Female
Unknown
Venous
Capillary
March
April
May
June
July
August
September
October
November
December
Less than 1 yr
1 year old
2-year old
3-year old
4-year old
5-year old
6-year old
Number of Observations for
Low(PbB<7pg/dL),
Medium (7sPbB<10 pg/dL),
and High (PbBi100g/dL)
Blood-Lead Concentration
(n * 3650)
Low
(n » 2999)
1127
1872
2368
56
121
49
405
1506
1481
12
2901
98
337
346
427
362
414
403
365
239
65
41
362
923
435
418
420
424
17
Medium
(n « 366)
113
253
276
10
19
2
59
181
183
2
355
11
34
49
57
27
54
49
45
37
3
11
29
131
48
69
44
44
1
High
(n » 285)
113
172
203
12
29
3
38
170
114
1
275
10
28
46
33
26
55
36
26
24
8
3
31
99
51
46
38
19
1
Blood-Lead Concentration
Geometric Mean
Estimate
2.62
2.41
2.43
3.36
3.14
222
2.67
2.51
2.47
3.51
2.50
2.22
2.77
2.81
2.60
2.03
2.76
2.23
227
2.47
2.49
3.18
2.01
2.76
2.62
2.90
2.40
1.91
0.54
Confidence
Interval1
(2.40,2.85)
(2.24,2.59)
(228.2.59)
(2.48,4.55)
(2.47.3.99)
(1.37,3.60)
(2.34,3.06)
(2.32,2.71)
(229.2.67)
(2.19.5.61)
(2.36.2.64)
(1.56.3.17)
(2.41.3.19)
(2.42.326)
(227,2.97)
(1.66.2.48)
(2.41,3.16)
(1.87,2.65)
(1.91,2.71)
(2.03,3.01)
(1.72.3.61)
(2.31,4.37)
(1.64.2.47)
(2.52,3.02)
(2.28.3.01)
(2.56,327)
(2.05,2.79)
(1.56.2.33)
(0.02,122)
Geometric Standard
Deviation
Estimate
2.60
2.66
2.63
2.62
3.00
222
2.43
2.73
2.53
1.90
2.63
2.91
2.40
2.83
2.44
2.63
2.69
2.75
2.61
2.66
2.46
2.18
2.81
2.61
2.64
2.47
2.59
2.67
5.07
Confidence
Interval1
(2.41,2.80)
(2.50.2.83)
(2.49,2.78)
(1.99,3.46)
(2.41,3.73)
(1.50,328)
(2.15,2.73)
(2.55.2.93)
(2.36.2.70)
(124.2.93)
(2.50.2.76)
(2.15.3.95)
(2.11.2.71)
(2.47.324)
(2.17.2.75)
(223.3.10)
(2.38.3.04)
(2.37.3.19)
(225,3.03)
(224,3.16)
(1.78,3.39)
(1.63.2.92)
(2.37,3.34)
(2.41,2.82)
(2.34,2.99)
(221.2.75)
(227,2.96)
(227.3.14)
(0.64,40.0)
Log Odds Difference1
Estimate
-0.09
0.92*
1.03'
-0.34
0.09
-0.38-
•0.30
-0.07
0.13
0.60
0.05
-0.02
0.60
020
-0.03
0.32
0.52
-0.23
0.09
0.03
-0.17
-0.87'
-0.60
Standard
Error
0.13
0.33
022
0.60
0.18
0.13
1.04
0.34
0.63
0.62
0.62
0.63
0.62
0.62
0.63
0.64
0.71
021
0.18
0.19
020
026
1.03
P-Value
0.4910
0.0050
0.0001
0.5743
0.6250
0.0025
0.7713
0.8274
0.8400
0.3342
0.9303
0.9765
0.3322
0.7487
0.9661
0.6183
0.4612
02949
0.6247
0.8914
0.3942
0.0007
0.5614
Percent of
Ob«.Wlth
Censored
Blood-Lead
67.8%
70.7%
70.5%
59.0%
60.4%
77.8%
68.9%
68.8%
70.6%
60.0%
69.6%
71.4%
66.9%
63.9%
69.4%
76.4%
65.8%
72.7%
732%
70.0%
71.1%
63.6%
75.4%
66.0%
67.6%
65.1%
71.3%
78.0%
89.5%
Note: Shaded cells (and asterisk) indicate statistically significant results.
1 The log odds difference compares the odds (and hence the risk) of a child having an EBL concentration
between one category and a reference category within each variable. The reference groups in the table are
the row with blanks in the log odds difference column. For example, the reference group for "Age" is
"1 year old". A negative log odds difference indicates that the category has lower risk than the reference
category, and vice versa for positive log odds difference. For example, a log odds difference of -0.87
indicates that 5-year olds have significantly lower risk than 1 year olds. See Appendix B for more details on
log odds differences.
2 The confidence intervals for the geometric mean and geometric standard deviations are calculated by
constructing the approximate confidence interval (± 2 standard errors) for the logarithm of each statistic and
then taking the exponential of the confidence bounds.
12
-------
Geometric means and standard deviations are often used instead of their arithmetic
counterparts when the data are known to have positive values and the upper limit of the values is
unknown. The data in such cases are often assumed to follow a lognormal distribution, and the
geometric mean and standard deviation are natural parameters for describing the distribution (in
the same way the arithmetic mean and standard deviation are natural parameters for describing
the normal (Gaussian) distribution). The use of the geometric measures to describe the
distribution of blood-lead concentrations is common in the literature (e.g., Brody et al.,1994).
The geometric means and geometric standard deviations of blood-lead concentration
reported in Table 3-1 (and Appendix A) were calculated using censored data techniques.
Censored data techniques are useful when, for some observations, exact values are not reported.
For example, when chemically analyzing a sample for some substance present only in trace
amounts, the reported result can be less than the limit of detection (LOD). Thus, the only real
information obtained from the analysis is that the value is less than the LOD. Such a data point
is referred to as left censored. A right censored data point is one for which all that is known is
that data point is above some value. Using censored data techniques allows inexactly reported
data (i.e., values only known to be in some interval) to be appropriately considered.
According to the Wisconsin Bureau of Public Health, for many of the laboratories
reporting blood-lead measurements to the state registry, 4 ug/dL is the limit of detection for
blood-lead concentrations. Therefore, no blood-lead concentrations less than 4 ug/dL are in the
registry, and blood-lead concentrations reported as 4 |ig/dL were assumed to be some value less
than or equal to 4 ug/dL (left censored at 4 ug/dL). The distribution of blood-lead
concentrations was assumed to be lognormal, and the method of maximum likelihood was used
to estimate the geometric means and geometric standard deviations. The assumption of
lognormality can be a limitation of the censored data approach. However, a lognormal
distribution has been extensively shown to be reasonable for blood-lead concentrations. Use of
censored data techniques was critical because, the percent of blood-lead concentrations that were
censored, listed in the last column of Table 3-1, generally is somewhere between 60 percent and
80 percent, except for cases when the number of observed blood-lead concentrations in the
category is low (e.g., when "Age" = "6 years old").
Six blood-lead concentration measurements in the study database were greater than
40 Hg/dL. (Eleven additional children in the registry database had first blood-lead
concentrations >40 ug/dL.) All these samples were collected using the capillary technique,
which can result in erroneous blood-lead measurements because of skin contamination
(CDC, 1991). Follow-up measurements for each of the six children in the study database
indicated that two had elevated blood-lead concentrations and four did not. The two
observations for which follow-up blood-lead measurements indicated an elevated blood-lead
concentration were treated as right censored at 40 jig/dL for estimating of the geometric means
and geometric standard deviations. The four children for whom subsequent blood-lead
measurements were <10 ug/dL were deleted from the study database. This is discussed in more
detail in Section 3.2.3.
Geometric means and geometric standard deviations of blood-lead concentration based
on data derived from this study should be considered with care. Because this study was a
13
-------
retrospective study (children were differentially selected for the study based on their blood-lead
concentrations), these values do not characterize blood-lead concentrations in Wisconsin or even
in the registry. Every child from the Wisconsin blood-lead registry with an elevated blood-lead
concentration was a candidate for the study, while only a fraction of the children with blood-lead
concentrations <7 ug/dL were included. Thus, geometric mean blood-lead concentrations
reported based on this study may be higher than those based on the full registry.
An adjustment was made to the study database before the data summaries presented in
Appendix A were prepared or any statistical analysis begun. The adjustment was made to
remove R&R activities that occurred after blood sampling.
For each type of R&R activity about which questions were asked (inside painting, outside
painting, other repairs, and carpets and floors), the respondent was also asked when these
activities began and ended. Because of the time delay between blood sampling and the
interviews, some R&R activities that did not begin until after the blood sampling were included.
Therefore, if the start date of an activity was after the blood sampling date, then that activity was
treated as if it did not occur. For example, if the inside painting start date (Question 21) was
after the blood sampling date, then the responses to questions 10 through 25 were changed to
"no" or "missing" as appropriate in the study database. Many start dates were missing from the
interview data. This was interpreted to mean that the R&R activity began before the blood
sampling.
In 432 (of 3,654) cases, some R&R activity occurred after blood collection. However,
266 of these had either another R&R activity before blood collection or one with a missing start
date. Thus, only 166 (432 minus 266) fewer children were considered to have had any exposure
to R&R as a result of this filtering of R&R activity data.
For homes constructed after the 1978 ban on lead-based paint, R&R activities are not
expected to disturb any lead. Thus, R&R activities in these residences may not impact lead
exposures in the same manner as those conducted in pre-1980 homes. However, in this report,
R&R activities in post-1980 residences were treated identically to R&R activities in pre-1980
homes. Some statistical analysis were performed attempting to identify differences in risk due to
lead exposure between R&R activities performed in pre- versus post-1980 homes (see
Section 3.3.2), but available data did not support the differences.
3.1.2 Characteristics of Sampled Children
Below are some key characteristics of children represented in the study database
(n=3650):
• Blood-lead concentrations
— 82 percent of children had blood-lead concentrations ^6 ug/dL (not elevated)
— 10 percent of children had blood-lead concentrations of 7, 8 or 9 fig/dL (not
included)
— 8 percent of children had blood-lead concentrations * 10 ug/dL (elevated).
14
-------
• Blood sample collection technique
— 97 percent of blood samples were collected via the capillary technique (which
sometimes results in skin contamination).
• Medical assistance status
— 37 percent of children received medical assistance.
• Race
— 78 percent of children were Caucasian.
• Year residence built
— 25 percent of children lived in residences built after 1978 (R&R activities in these
homes should result in lower lead exposure).
• Tenure at current residence
— 22 percent of children lived in their current residences less than 1 year.
• Peeling paint
— <4 percent of children resided in residences with "a lot of peeling paint"3 (inside)
— <8 percent of children resided in residences with "a lot of peeling paint"3
(outside).
• Prevalence of R&R activities
— in >40 percent of homes some interior painting occurred within the last
12 months; and, in most cases (nearly 90 percent), it was done by someone living
at the home
— in 12 percent of homes some exterior painting occurred within the last 12 months
— in 22 percent of homes windows were repaired or new windows were installed
— in 22 percent of homes carpets were repaired or replaced.
Child's age at blood sample collection
— 12 percent of children were less than 1 year old
— 32 percent of children were 1 year old
— 15 percent of children were 2 years old
— 15 percent of children were 3 years old
— 14 percent of children were 4 years old
— 13 percent of children were 5 years old
— <0.5 percent of children were 6 years old.
Because the definitions of lead-based paint hazards in the HUD guidelines (HUD, 1995) or the §403 Risk
Analysis (EPA, 1997) were too complicated for use during a telephone interview, "a lot of peeling paint"
was the terminology used in the questionnaire. See, for example, Question 6 in Appendix A.
15
-------
• Occupational or hobby related exposure
— 31 percent of study children may have been exposed to lead as a result of the
occupation of a household member
— 43 percent may have been exposed to lead as a result of a hobby of a household
member.
• Household income
— 43 percent of children lived in households with annual incomes above $30,000
— 57 percent of children lived in households with annual incomes below $30,000.
• Respondents education (most respondents were female)
— 52 percent of respondents had a high school education or less
— 33 percent of respondents had some education beyond high school
— 16 percent of respondents were college graduates.
Because of the extended period (March to December) over which the blood samples of
children in this study were collected, blood-lead concentrations were examined to assess possible
seasonality. Previous work (EPA, 1995, and EPA, 1996) has indicated that there is a tendency
for children's blood-lead concentrations to be higher in the summer (warmer months) and lower
in the winter (colder months), possibly due to different activity patterns. Because longitudinal
blood-lead concentration data for individual children included in this study were not available,
blood-lead concentration trends were examined at an aggregate level. Figure 3-1 presents the
geometric mean blood-lead concentration in the registry database plotted as a function of sample
collection month with pointwise confidence intervals. Blood-lead concentrations appear to be
lower in the summer than the winter, which is inconsistent with previously reported work (EPA,
1995, and EPA, 1996). However, since the confidence intervals suggest that any evidence of a
seasonal trend in the blood-lead concentrations of children in the study is fairly weak, this effect
was not pursued further. Additionally, fewer nonelevated blood-lead concentration children were
sampled in October, November, and December, likely leading to the slight increase in blood-lead
concentrations observed in the data for those months.
3.1.3 R&R Exposure Characterization
Information on the frequency of general and specific R&R activities in residences
populated by young children is a valuable product of this study. While the main objective of this
study was to investigate the relationship between incidences of R&R activities and EBL children
(a dose-response type relationship), this information is most useful when the degree of exposure
is also known. In this case, if a slight increase in risk due to childhood lead exposure as a result
of R&R activity is detected, the implications must be considered in light of the numbers of
children likely to be exposed.
16
-------
3.4
2.3 H
Mar Apr May Jun Jul Aug Sep
Month of Blood Collection
Oct
Nov
Dec
Figure 3-1. Trend Plot to Assess the Seasonality of
Blood-Lead Concentrations in the Registry
Database
Tables 3-2 and 3-3 list the frequencies and percentages summarizing the responses to
each question in the R&R activities portion of the questionnaire (Questions 10 through 71 in
Appendix A). The R&R activities portion is subdivided into four parts, with each part relating to
one of four distinct types of R&R activities that may increase lead contamination in homes:
"inside painting," "outside painting," "other repairs (windows)," and "carpets and floors (repair
or replacement)." An examination of the questionnaire shows that each of the four parts contains
a set of similar questions concerning details common to all four types of R&R activity, such as
the method of surface preparation carried out for the activity. The first column in Table 3-2 lists
these questions, accompanied by the question numbers identifying the four interview questions
related to each type of activity (i.e., inside painting, outside painting, other repairs, carpets and
floors). The question numbers are in the same order as the activities listed in the columns
(e.g., Q19 related to using chemical paint removers for inside painting). The remaining questions
from the R&R activities portion of the questionnaire are unique to only one of the four activities
and can be found in the first column of Tables 3-3a (inside painting), 3-3b (other repairs), and
3-3c (carpets and floors) with their respective question numbers.
17
-------
Table 3-2. R&R Activity Exposure Summary
Question
Was the R&R activity carried out in
the last 12 months?
Q10, 028, 039, 054
Was the surface prepared by
methods such as sanding or
scraping?
Q1 3,029,043,057
Was hand scraping or hand sanding
used?
Q14, Q30, Q44, Q58
Was power sanding, grinding or
sandblasting used?
Q15, 031,045,059
Was an open llama torch used?
Q16, 032, Q47, 060
Were heat guns used?
Q17, 033, Q48, Q61
Was washing, wet scraping, wet
sanding or water blasting used?
Q18, O34, 049, 062
Response
Yes
No
Don't Know
Yes
No
Don't Know
Yes
No
Don't Know
Yes
No
Don't Know
Yes
No
Don't Know
Yes
No
Don't Know
Yes
No
Don't Know
R&R Activity
(n-3654)
Inside Painting
Frequency
1858
1786
10
541
1241
76
482
42
17
166
356
19
6
519
16
44
476
21
160
352
29
Percent
50.8%
48.9%
0.3%
29.1%
66.8%
4.1%
89.1%
7.8%
3.1%
30.7%
65.8%
3.5%
1.1%
95.9%
3.0%
8.1%
88.0%
3.9%
29.6%
65.1%
5.4%
Outside Painting
Frequency
703
2924
27
445
225
33
413
22
10
64
359
22
5
427
13
24
402
19
111
310
24
Percent
19.2%
80.0%
0.7%
63.3%
32.0%
4.7%
92.8%
4.9%
2.2%
14.4%
80.7%
4.9%
1.1%
96.0%
2.9%
5.4%
90.3%
4.3%
24.9%
69.7%
5.4%
Other Repairs (Windows)
Frequency
805
2839
10
125
630
50
114
9
2
33
89
3
2
120
3
11
111
3
33
89
3
Percent
22.0%
77.7%
0.3%
15.5%
78.3%
6.2%
91.2%
7.2%
1.6%
26.4%
71.2%
2.4%
1.6%
96.0%
2.4%
8.8%
88.8%
2.4%
26.4%
71.2%
2.4%
Carpets and Floors (Repair
or Replacement)
Frequency
816
2825
13
206
537
73
163
35
8
64
133
9
1
196
9
5
190
11
38
158
10
Percent
22.3%
77.3%
0.4%
25.2%
65.8%
8.9%
79.1%
17.0%
3.9%
31.1%
64.6%
4.4%
0.5%
95.1%
4.4%
2.4%
92.2%
5.3%
18.4%
76.7%
4.9%
Any R&R Activity
Frequency
2456
1179
19
1038
1274
144
937
72
29
288
705
45
11
991
36
73
917
48
292
687
59
Percent
67.2%
32.3%
0.5%
42.3%
51.9%
5.9%
90.3%
6.9%
2.8%
27.7%
67.9%
4.3%
1.1%
95.5%
3.5%
7.0%
88.3%
4.6%
28.1%
66.2%
l_ 5.7%
00
-------
Table 3-2. (Continued)
Question
Were chemical paint removers used?
Q19, Q35, Q50, Q63
Who did the work?
(Head or Spouse)
Q20-1, Q36-1, Q51-1, Q64-1
Who did the work?
(Other Person in Household)
020-2, 036-2, Q51-2. Q64-2
Who did the work?
(Relative/Friend Not in Household)
Q20-3, Q36-3, Q51-3, Q64-3
Who did the work?
(Owner, Building Superintendent, Apt.
Staff)
Q20-4, Q36-4, Q51-4, 0.64-4
Who did the work?
(Paid Professional Contractor)
020-5, 036-5, Q51-5, Q64-5
Did anyone in household live at home
while work was being done?
Q22, Q38, Q53, Q66
Response
Yes
No
Don't Know
Yes
No
Don't Know
Yes
No
Don't Know
Yes
No
Don't Know
Yes
No
Don't Know
Yes
No
Don't Know
Yes
No
Don't Know
R&R Activity
Inside Painting
Frequency
96
428
17
421
119
1
76
464
1
67
473
1
54
486
1
51
489
1
458
83
0
Percent
17.7%
79.1%
3.1%
77.8%
22.0%
0.2%
14.0%
85.8%
0.2%
12.4%
87.4%
0.2%
10.0%
89.8%
0.2%
9.4%
90.4%
0.2%
84.7%
15.3%
0.0%
Outside Painting
Frequency
24
389
32
305
135
5
36
404
5
29
411
5
73
367
5
53
387
5
423
22
0
Percent
5.4%
87.4%
7.2%
68.5%
30.3%
1.1%
8.1%
90.8%
1.1%
6.5%
92.4%
1.1%
16.4%
82.5%
1.1%
11.9%
87.0%
1.1%
95.1%
4.9%
0.0%
Other Repairs (Windows)
Frequency
14
107
4
93
31
1
6
118
1
12
112
1
5
119
1
24
100
1
115
10
0
Percent
11.2%
85.6%
3.2%
74.4%
24.8%
0.8%
4.8%
94.4%
0.8%
9.6%
89.6%
0.8%
4.0%
95.2%
0.8%
19.2%
80.0%
0.8%
92.0%
8.0%
0.0%
Carpets and Floors (Repair
or Replacement)
Frequency
28
167
11
139
65
2
30
174
2
23
181
2
14
190
2
45
159
2
155
51
0
Percent
13.6%
81.1%
5.3%
67.5%
31.6%
1.0%
14.6%
84.5%
1.0%
11.2%
87.9%
1.0%
6.8%
92.2%
1.0%
21.8%
77.2%
1.0%
75.2%
24.8%
0.0%
Any R&R Activity
Frequency
141
841
56
762
267
9
121
908
9
112
917
9
130
900
8
155
874
9
918
120
0
Percent
13.6%
81.0%
5.4%
73.4%
25.7%
0.9%
11.7%
87.5%
0.9%
10.8%
88.3%
0.9%
12.5%
86.7%
0.8%
14.9%
84.2%
0.9%
88.4%
11.6%
0.0%
-------
Table 3-3a. Inside Painting Specific Exposure Summary
Inside Painting
(n = 3654)
Question
Was the R&R activity carried out in
the last 1 2 months?
Q10
Was any work done to repair broken
plaster or damaged walls in the
room/rooms you painted in the last
1 2 months?
Q11
Was any work done where old walls
were taken down or moved, while
working in your home?
Q12
Was the surface prepared by
methods such as sanding or
scraping?
Q13 CYes'toQIO)
In how many rooms in your home
was this work done?
Q23'('Yes'toQ13)
Was any work done in the kitchen?
Q24{'Yes'toQ13)
Was any work done in the
bathroom?
025 ('Yes' to Q1 3)
Response
Yes
No
Don't Know
Yes
No
Don't Know
Yes
No
Don't Know
Yes
No
Don't Know
None
1
2
3
4
Yes
No
Don't Know
Yes
No
Don't Know
Frequency
1858
1786
10
1
1005
2636
12
395
3252
7
541
1241
76
8
1
205
119
67
141
222
318
1
239
301
1
Percent
50.8%
48.9%
0.3%
27.5%
72.2%
0.3%
10.8%
89.0%
0.2%
29.1%
66.8%
4.1%
.
0.2%
38.5%
22.3%
12.6%
26.5%
41.0%
58.8%
0.2%
44.2%
55.6%
0.2%
20
-------
Table 3-3b. Other Repairs (Windows) Specific Exposure Summary
Other Repairs (Windows)
(n = 3654)
Question
Was the R&R activity carried out in
the last 1 2 months?
Q39
Was any work done to repair broken
plaster or damaged walls while
repairing or putting in the windows?
Q40 {'Yes' to Q39)
Was any work done where old walls
were taken down or moved while
repairing or putting in the windows?
Q41 ('Yes' to Q39)
Was there any painting or were any
surfaces prepared for paint with the
installation of the new windows?
042 ('Yes' to Q39)
Response
Yes
No
Don't Know
Yes
No
Don't Know
Yes
No
Don't Know
Yes
No
Don't Know
Frequency
805
2839
10
1
153
637
14
109
691
5
235
561
9
Percent
22.0%
77.7%
0.3%
19.0%
79.2%
1 .7%
13.5%
85.8%
0.6%
29.2%
69.7%
1.1%
21
-------
Table 3-3c. Carpets and Floors (Repair or Replacement) Specific Exposure Summary
Carpets and Floors (Repair or Replacement)
(n = 3654)
Question
Was the R&R activity carried out in
the last 1 2 months?
Q39
As a result of the work that was
done in your home, were dust and
dirt spread to .... ?
Q71
Do you have any wall-to-wall
carpeting in your home?
055 ('Yes' to Q54)
Did you replace or remove a wall-to-
wall or other large carpet in the last
1 2 months?
Q56 ('Yes' to Q54)
Response
Yes
No
Don't Know
Only in the Work Area
In the Room next to the
Work
Through the House
No Dirt or Dust
Generated
Don't Know
Yes
No
Don't Know
Yes
No
Don't Know
Frequency
816
2825
13
1206
903
264
335
857
89
639
176
1
543
269
4
Percent
22.3%
77.3%
0.4%
36.9%
10.8%
13.7%
35.0%
3.6%
78.3%
21.6%
0.1%
66.5%
33.0%
0.5%
22
-------
The complexity of the tables reflect the complexity of the skip patterns used in the
questionnaire. To illustrate how the setup in Table 3-2 works, take as an example the first row
(which is subdivided further into three subrows in all except the first column) and find the
number "703" in the fifth column of the first subrow. The first level label for the column is
"Frequency" and the second level label is "Outside Painting." The first column labeled
"Question" for that entry contains the question "Was the R&R activity carried out in the last
12 months?" and the second column labeled "Response" contains the response "Yes." Therefore,
"703" is the frequency of people who responded "Yes" to the R&R question that asked whether
any painting outside the house was carried out within the last 12 months. To pinpoint the actual
question in the questionnaire, go back to the first column, where four question numbers ("Q10,
Q28, Q39, Q54") are listed in the same order of R&R activity type as the columns. Since
"Outside Painting" is the second R&R activity type in the columns, the question of whether any
painting outside the house was carried out within the last 12 months was asked in Question 28.
The percentage "19.2%" in the column to the right of "703" refers to the same 703 who answered
"Yes" out of the 3,654 (which equals 703+2924 +27, the sum of the frequency of all possible
responses in the first row) who responded to Question 28.
The numbers in the last two columns of Table 3-2, however, are slightly different from
the others, in that they summarize the responses across the four types of activities using an
amalgamation of the responses. Take again, for example, the first rows. If a respondent
answered "Yes" to any of Q10, Q28, Q39 or Q54, then the amalgamated response was set to
"Yes." If a respondent answered "No" to all of Q10, Q28, Q39 or Q54, then it was set to "No."
Otherwise, the response was set to "Don't know." This amalgamation can be interpreted as a
response to the question "Was any R&R activity carried out in the last 12 months?," hence the
label "Any R&R Activity" in the second level of the column labels.
The double lines between some rows of Table 3-2 take into account the skip patterns in
the questionnaire, where only those respondents who answered "Yes" to a question above the
double line were allowed to answer the questions below the double line. For example, the
second and fourth rows are separated by a double line. Only those 541 people who were
responded "Yes" to Question 13 were asked to respond to Question 15 (notice that
541=166+356+19). Hence, the percentages below the double lines are conditional on answers to
previous questions. For example, the percentage of people who responded "Yes" to Question 15
is 30.7 percent (166/541).
Table 3-2 reveals the pattern of R&R activities for study participants. Based on all
participants, at least one of the four R&R activities had been conducted in 67.2 percent of the
residences in the last 12 months, with around 50 percent conducting some inside painting and
approximately 20 percent each conducting outside painting, other repairs, and carpets and floors
(repair or replacement) (row 1). Of the 2,456 houses where at least one of the four R&R
activities was carried out, some form of surface preparation (row 2) was performed in
42.3 percent. Approximately 65 percent of outside painting involved some surface preparation,
compared to 15 percent to 30 percent for each of the other activities. This difference is not too
surprising since outside painting tends to require more paint removal than the other activities. As
for the type of surface preparation techniques employed, the conditional percentages do not vary
much within the rows (rows 3 to 8). Therefore, if any surface preparation was carried out, the
23
-------
type of R&R activity does not appear to affect the surface preparation technique used for the
activity. Most surface preparation involved some sort of hand scraping or hand sanding
(80 percent to 90 percent), whereas the use of open flame torches was rare (1 percent to
2 percent). Heat guns were used 7 percent of the time, and chemical paint remover was used
13.6 percent of the time.
The conditional percentages for the type of individual who carried out the surface
preparation also did not change much within the rows; although owner, building superintendent,
and apartment staff were more likely to have been involved in any surface preparation for inside
or outside painting, whereas paid professional contractors were more likely to have been
involved in the other two R&R activities, which tend to require more skilled labor (rows 9 to 13).
The head of household or spouse was involved in surface preparation work approximately
70 percent of the time, whereas others were involved 10 percent to 15 percent of the time.
Finally, some person was usually living at the home while surface preparation work was
being done. The percentages for inside painting and carpets and floors (repair or replacement),
which occurred inside the house, were lower than the percentages for outside painting and other
repairs (window), which involved more outside work (row 14).
Tables 3-3a (inside painting), 3-3b (other repairs), and 3-3c (carpets and floors) contain
similar statistics for the questions unique to each R&R activity. Additional nonunique questions
from Table 3-2 are duplicated in some rows so that Tables 3-3a through 3-3c are set up similarly
to Table 3-2. The double lines separate the questions into components as before, where
participants were required to respond to a question in a component only if they gave a "Yes"
response to a particular question in the previous component. The parentheses next to the
question number contain the particular response and interview question. Note, for example, that
the frequencies in row 4 of Table 3-3a, where the question was responded to only by those
individuals who answered "Yes" to Question 10, add up to the frequency of "Yes" responses in
row 1(541+ 1241+76 = 1858).
Table 3-3a shows that very little of the painting activities required any nonsurface
preparation activities such as repairs to broken plaster or damaged walls (row 2) and taking down
or moving walls (row 3). In the homes where surface preparation was carried out for inside
painting, most were jobs involving one to three rooms, with just over a quarter involving more
(row 5). The results from Table 3-3b indicate that in homes where windows were either repaired
or replaced, broken plaster or walls were repaired 19 percent of the time (row 2), walls were
taken down or moved 13.5 percent of the time (row 3), and painting or surface preparation for
painting was carried out 29.2 percent of the time (row 4). Finally, of the homes where carpets
and floors were repaired or replaced, 78.3 percent had wall-to-wall carpeting and 66.5 percent
had wall-to-wall or other large carpet replaced or removed (rows 3 and 4 in Table 3-3c).
3.2 DATA QUALITY
Data from two sources are analyzed in this report: blood-lead registry data from
Wisconsin children outside of Milwaukee and Racine and interview data collected specifically
for this study from a subset of these same children. The blood-lead registry data reflect a
24
-------
continuing problem with reporting race and ethnicity in Wisconsin. (Ethnicity data are not
analyzed in this report due to these problems.) For example, of the 3,650 children in this study,
ethnicity was unknown for 97 percent of children. Medical assistance status was also unknown
for a portion of the children in the registry. This is not reflected in Table 3.1 because one of the
criteria for a child being in this study was known medical assistance status. (This variable was
used in the study design. See Section 2.0.)
Although this study was designed to separately assess the impact of R&R activities on
blood-lead concentrations for children receiving medical assistance and children not receiving
medical assistance, the results reported in Table 3-1, Section 3.3, and Appendices A, C, and D
are based on both groups. Analysis of the preliminary data set, containing approximately
one-third of the full data set, indicated that similar relationships were observed for both groups.
Other potential data quality concerns include quality control procedures employed
during collection of blood-lead registry and questionnaire data, possible nonresponse and
sampling biases, and skin contamination in the blood-lead registry data. Section 3.2.1 discusses
measures for ensuring the quality of the blood-lead registry and questionnaire data. Section 3.2.2
discusses nonresponse and sampling biases. Section 3.2.3 addresses skin contamination.
3.2.1 Blood-Lead Registry and Interview Data
Elevated blood-lead concentration (^lOug/dL) has been a reportable condition in
Wisconsin since 1979. In 1993 the law changed to require that all blood-lead tests be reported
for children under 6 years old. The Wisconsin State Health Department has encouraged
laboratories to report blood-lead concentration, but the law required a person who screens to
report results of testing of blood-lead concentrations of children under six. The law also requires
any person including doctors, nurses, hospital administrators, directors of clinical laboratories
and health officers who diagnoses lead poisoning (^lOug/dL) to report it regardless of the age of
the person tested.
The state department of health and family services is considering issuing rules to clarify
reporting responsibilities, but as of today only the statute ensures compliance with reporting. In
practice, this is a lab based system. Labs report to the health department. The health department
staff contact labs and doctors offices to obtain any missing demographic information and pass on
information about children with elevated levels to local health departments for follow-up.
Quality control of analysis is regulated by CLIA, the clinical laboratory improvement
amendments of 1988. All commercial labs must participate in a HCFA approved proficiency
testing program such as the programs operated by the University of Wisconsin and the College of
American Pathologists.
Quality control of the sample collection is up to the sample collector. In practice, the
state department of health and family services, the state lab of hygiene, and local health agencies
work with health care providers to inform them of quality control issues that arise, especially
with capillary specimens.
25
-------
A number of measures were used to ensure interview data quality. When frequency
counts were run, a separate file was created for codes that were out of the range for the variable.
Errors were corrected using the original interviews. Consistency errors were infrequent. In some
cases, field sections were checked to clarify a problem, and sometimes the respondent was
recontacted to solve the problem. Finished codebooks and frequency runs were reviewed to
make sure that all data corrections were made.
3.2.2 Assessment of Nonresponse and Selection Bias
Whenever a sample is selected, there is a potential for bias. For example, bias can be
introduced if careful interviewing practices are not followed. In this study, the interviewers were
specifically trained in administering telephone interviews. Further, the interviewers followed a
pre-designated script to collect the information from and to recruit participants. Training and
following a set script minimizes the potential for interviewer bias. Nonresponse is also a
potential source for bias in the sample estimates.
In general, bias in sample estimates is caused by differences between the population
parameters of participants and nonparticipants and, therefore, can be investigated by comparing
characteristics of participants in a study to those of nonparticipants (Thompson, 1992). Due to
the mandatory screening of medical assistance children, it is unlikely that the Wisconsin
blood-lead registry is representative of the children of Wisconsin in many demographic features.
For example, the 1990 census reported that 95 percent of Wisconsin children outside of
Wisconsin and Racine counties are Caucasian. Only 67 percent of the children selected for the
study were Caucasian and only 78 percent of those participating. The registry sample had larger
percentages of African-Americans, Asians, and Native Americans than percentages reported by
the census. However, some of the difference between the census and registry sample racial
distributions may be due to the large percentage of unknown race children reported in the registry
sample (17%) as compared to the census (<1%).
Although incomplete reporting, i.e., the large percentage of unknown race children, is a
problem for any study, the data selected for this study does not need to be representative of a
particular group of children to achieve the primary study objectives. The primary purpose of this
study was to establish a dose-response type relationship between incidence of general and
specific R&R activities and EBLs. This dose-response relationship is assumed to be applicable
to many similar populations.
In this nonresponse assessment, "participants" are children selected for inclusion and
participating in the study. "Nonparticipants" are children who were selected for inclusion in the
study, but did not participate for some reason (e.g., the child's guardian refused to be interviewed
or study personnel were unable to reach the child's guardian). However, even extreme
differences between participants and non-participants may not cause severe bias as long as the
participation rate is the same between children with blood-lead levels below 7 ng/dL (controls)
and children with blood-lead levels £ 10 ug/dL (cases). Therefore, to investigate bias it is
important to examine the participation rate among the cases to that among the controls. This
does not mean that comparisons between characteristics of participants and nonparticipants are
without merit; these comparisons will serve to indicate the overall representativeness of the
26
-------
selected sample of children compared to the registry population. In this section, we present the
results of both types of comparisons beginning with comparing the characteristics of the
interviewed children to the sampled population.
A limited amount of information on the characteristics of study participants was
available in the Wisconsin Bureau of Public Health's blood-lead registry. Besides blood-lead
concentrations, this information included the gender of the child, race of the child and parent,
child's birthday, and the child's age when the first blood-lead sample was collected. Table 3-4
presents a summary of demographic information for participants and nonparticipants. The last
column in the table presents the p-value for a statistical test comparing characteristics of
participants and nonparticipants. For the two continuous variables (child's age and blood-lead
concentration), the p-value corresponds to a t-test comparing the mean levels between
participants and nonparticipants. For the categorical characteristics, the p-value corresponds to a
likelihood ratio chi-square statistic comparing the proportions of participants and nonparticipants
in the levels of the characteristic.
With the exception of gender, characteristics of study participants were statistically
different from those of nonparticipants. However, in practical terms, these differences may not
be meaningful. For example, study participants were significantly older than nonparticipants;
but they were, on average, only 0.137 years or approximately 1.6 months older (95 percent
confidence interval from 0.81 to 2.5 months). The geometric mean blood-lead concentrations for
participants were significantly lower than that of nonparticipants; on average, 1.26 times lower,
with a 95 percent confidence interval of 1.18 to 1.35 times. The proportion of children with
blood-lead concentrations £ 10 jig/dL was also significantly smaller for participants than
nonparticipants. It can be concluded that fewer EBL children could be located or agreed to
participate in the study (nonresponse bias).
27
-------
Table 3-4. Summary of Demographic Information for Participants and Nonparticipants
Variable
Description
Sample Size
Age (Years)
Blood-Lead
Concentration (fjgldL)
Blood-Lead
Concentration Group
Gender of Child
Race of Child
Race of Parent
25th Percentile
Mean (Standard Error)
75th Percentile
25th Percentile
Geometric Mean1"1
(Log Standard Error)
75th Percentile
<10//g/dL
>10//g/dL
Female
Male
Unknown
Asian
African-American
Native-American
Caucasian
Unknown101
Asian
African-American
Hispanic
Native-American
Caucasian
Unknown
Nonparticipants
5,004
1.1
2.55 (0.022)
3.9
1.68
3.11
(0.019)
5.75
89.73%
10.27%
47.96%
51.3%
0.70%
9.19%
9.37%
3.24%
59.29%
18.90%
9.19%
9.37%
8.05%
3.24%
59.17%
10.97%
Participants
3,654
1.1
2.69 (0.028)
4.1
1.27
2.47
(0.028)
4.80
92.09%
7.91%
48.74%
50.85%
0.41%
4.63%
2.13%
1.48%
78.02%
13.74%
4.60%
2.13%
2.71%
1.48%
77.94%
11.14%
p-value'"
0.0001
0.0002
0.001
0.171
0.001
0.001
(a) This p-value corresponds to testing for significant differences between the study participants and nonparticipants. The
p-value is based upon a t-test for continuous variables and a x2 test for categorical variables.
(b) Geometric means and log standard errors were calculated accounting for censored values. Blood-lead concentrations
reported as 4 i/g/dL or lower were considered to be left-censored, and values reported at or above 40 //g/dL were
considered to be right-censored. There were 3,095 left-censored and 13 right-censored values for nonparticipants.
There were 2,540 left-censored and 6 right-censored values for participants.
(c) A distinct identifier identifying Hispanic children is not present in the Wisconsin registry. Therefore the Unknown
category contains these children.
Greater differences were found when comparing the race of participants and
non-participants. In particular, a larger percentage of participants were Caucasian. This result
was true for both parents and their children. Thus, it will be important to consider race when
interpreting the results of this study.
Potential bias in the statistical results presented in Section 3.3 were investigated by
comparing the participation rate among cases to that among controls. Unfortunately, we were
only able to investigate the potential bias for those variables included in the registry database.
28
-------
Bias in the R&R results could not be fully investigated. The potential for bias was determined
by comparing the distribution of participation across the levels of a particular demographic
factors changed with blood-lead group. Using log-linear models, these comparisons were
performed by investigating the three-way interaction between the factor of interest, participation,
and blood-lead group. A significant three-way interaction indicates that the effect of the
demographic variable on blood-lead is confounded with participation. Table 3-5 presents
frequency tables that formed the basis for the tests, as well as the p-value associated with the
significance of the interaction term. The results of these tests indicate that the effect of a child's
age on blood-lead is confounded with participation. Therefore, the age of the child needs to be
included in the logistic regression models so that this source of variability can be accounted for.
However, care needs to be taken when interpreting the model estimates for this variable, because
of its potential confounding with participation.
3.2.3 Assessment of Skin Contamination
Six blood-lead concentrations were >40 ug/dL in the study database. (Eleven additional
children in the registry database had blood concentrations >40 ug/dL.) All six measurements
were based on capillary samples (which can result in skin contamination of a sample) and had
one or two follow-up blood-lead measurements. Table 3-6 presents first, second, and third
blood-lead concentration measurements for these six children. In all cases presented, second and
third blood-lead samples were collected within a month of the first sample, making them
reasonable follow-up blood-lead measurements. [One month is the longest time span
recommended by CDC for confirming an elevated capillary blood-lead result (CDC, 1991).]
The four children whose first blood-lead concentrations were greater than 40, but for
whom second and third measurements indicated nonelevated blood-lead concentrations (1, 2, 4,
and 6), were deleted from the study database. The two children for whom the confirmatory
blood-lead measurements supported a diagnosis of elevated blood-lead concentration were not
deleted from the database.
Blood-lead concentration data from the blood-lead registry database were examined for
possible biases due to blood sampling technique (i.e., capillary vs. venipuncture). Table 3-7
presents, for six age groups of children, confidence intervals for the ratio of capillary to
venipuncture geometric mean and logarithmic standard deviation blood-lead concentrations. The
blood-lead screening data available in the registry were collected primarily by the capillary
method. Thus, the sample size for the venous calculations is small. Based on Table 3-7, it can
be concluded that blood-lead concentration measurements taken via the capillary method were
not, on average, significantly higher than those taken via venipuncture for any age group. In fact,
it appears that, for children less than 1 year of age, there is a tendency for venipuncture
blood-lead measurements to be higher than capillary measurements.
29
-------
Table 3-5. Comparison of Participation Between Cases and Controls
Variable
Gender of Child
Race of Child
Race of Parent
Age of Child
Description
Female: % Participants
Total No. of Records
Male: % Participants
Total No. of Records
Unknown: % Participants
Total No. of Records
Asian % Participants
Total No. of Records
African- % Participants
American Total No. of Records
Native- % Participants
American Total No. of Records
Caucasian % Participants
Total No. of Records
Unknown"1 % Participants
Total No. of Records
Asian % Participants
Total No. of Records
African- % Participants
American Total No. of Records
Hispanic % Participants
Total No. of Records
Native- % Participants
American Total No. of Records
Caucasian % Participants
Total No. of Records
Unknown % Participants
Total No. of Records
Less than % Participants
1 Year Total No. of Records
1 Year Old % Participants
Total No. of Records
2 Years Old % Participants
Total No. of Records
3 Years Old % Participants
Total No. of Records
4 Years Old % Participants
Total No. of Records
5 Years Old % Participants
Total No. of Records
6 Years Old % Participants
Total No. of Records
Children
with
PbB*10
(Cases)
34.3%
341
37.4%
457
20.0%
5
26.1%
111
12.4%
97
37.5%
8
45.2%
458
29.5%
129
25.5%
110
12.4%
97
20.3%
59
37.5%
8
45.1%
452
39.0%
77
49.2%
63
37.6%
266
34.2%
152
37.0%
127
31.4%
121
28.2%
71
33.3%
3
Children with
PbB<7
(Controls)
43.8%
3378
43.7%
3349
33.3%
36
29.2%
414
15.4%
364
25.3%
194
49.8%
4760
35.8%
1131
29.2%
414
15.4%
364
19.4%
371
25.3%
194
49.8%
4757
43.6%
763
38.6%
937
43.5%
2123
44.3%
981
40.1%
1042
40.8%
1030
59.4%
714
47.2%
36
P-value
0.629
0.867
0.920
0.0004
(a) A distinct identifier identifying Hispanic children is not present in the Wisconsin registry. Therefore the Unknown category
contains these children.
30
-------
Table 3-6. Six Capillary Blood-Lead Concentration Measurements with Possible Skin
Contamination Bias
Table 3-7.
Child
1
2
3
4
5
6
Blood-Lead Concentration Measurement (i/g/dL)
Rrst
58
61
62
76
85
140
Second
8
4
59
4
21
8
Third
*
4
29
*
17
*
* Elevated blood-lead measurements in the shaded rows were assumed not to be
from contamination. Measurements in nonshaded rows were assumed due to skin
contamination and were removed from the study database.
Assessment of Skin Contamination of Capillary Blood Samples for Six Age
Groups of Children*
Age Group
(years)"
<1
1-2
2-3
3-4
4-5
5-6
Age Group
(years)
<1
1-2
2-3
3-4
4-5
5-6
Capillary
Sample Size
1048
2639
1250
1318
1248
821
Geometric
Mean
1.76
3.05
3.22
3.19
2.80
2.33
Logarithmic
Standard
Deviation
0.0057
0.00070
0.0014
0.0012
0.0018
0.0041
Geometric Mean Ratio
(Capillary /Venipuncture)
Estimate
0.45
0.93
0.75
0.77
0.75
0.96
95% Confidence Interval
[0.32, 0.62]c
[0.69, 1 .24]
[0.52, 1.08]
[0.58, 1.02]
[0.56, 1.01]
[0.60, 1.55]
Venipuncture
Sample Size
31
89
41
38
39
52
Geometric
Mean
3.93
3.28
4.30
4.15
3.73
2.42
Logarithmic
Standard
Deviation
0.021
0.021
0.032
0.019
0.021
0.053
Logarithmic Standard Deviation Ratio
(Capillary /Venipuncture)
Estimate
1.67
0.94
0.97
1.29
1.40
1.18
95% Confidence Interval
[1.09, 2.55]°
[0.72, 1.24]
[0.68, 1 .38]
[0.89. 1.87]
[0.94. 2.09]
[0.74, 1 .88]
(a) Results calculated using censored data techniques. Blood-lead concentrations of 4 pg/dL were treated as
left-censored at 4 //g/dL and blood-lead concentrations > 40 |/g/dL were treated as right-censored at 40 /yg/dL.
(b) The categorization scheme for the age group categories is:
< 1: 0 months s age < 12 months
1-2: 12 months z age < 24 months
2-3: 24 months s age < 36 months
etc.
(c) Statistically significant results at the 5 percent significance level.
31
-------
There may be some selection factors that relate to type of blood sampling method.
CDC 1991 states that elevated blood-lead results obtained on capillary specimens are
presumptive and must be confirmed using venous blood and that children at the highest risk
should be given the highest priority for screening. Thus, physicians are more likely to perform
initial venous screenings on high risk children. Non-random selection of children for initial
venous screenings may be the reason venipuncture blood-lead measurements are higher than
capillary measurements among children less than 1 year of age. This does not preclude the
possibility of skin contamination affecting individual blood-lead concentration measurements,
but suggests that, in general, blood-lead concentration measurements taken by the capillary
method should not be adjusted downward.
The variation in blood-lead concentrations, as measured by the ratio of logarithmic
standard deviations, was statistically different between the two groups only for children less than
1 year of age. Variability in blood-lead concentration measurements among children less than
1 year of age was larger among those children sampled by the capillary method. This could be
because children less than 1 year of age who receive venous blood-lead screening are mostly
alike in that they are at high risk of lead exposure.
3.3 STATISTICAL ANALYSIS
Children were selected into this study based on their observed blood-lead concentrations.
Since the purpose of this study was to assess the impact of R&R activities on children's
blood-lead concentrations, an analysis methodology that respects the fact that observations were
chosen based on the value of the response variable must be used. One such inference mechanism
is the odds ratio. The odds ratio is the natural form of inference based on logistic regression.
The data collected in this study were analyzed using logistic regression. Logistic
regression is a statistical method used to explain the relationship between a dichotomous
response variable and a suite of explanatory variables. The dichotomous response variable
analyzed is an indicator of elevated or nonelevated concentration. The explanatory variables
provide information on social and demographic factors affecting children (e.g., household
income), as well as on R&R activities conducted in children's residences (e.g., surface
preparation for indoor painting). Appendix B discusses logistic regression and the odds ratio in
detail.
Two types of logistic regression analyses were conducted. First, univariate regressions
were performed for each explanatory variable. Both blood-lead registry variables and
questionnaire responses were analyzed. Only six blood-lead registry variables were available.
The statistical significance for each of the blood-lead registry variables was assessed and is
presented in Table 3-1. Because of the large number of interview questions (and possible
responses), the univariate logistic regression results based on the interview questions were
considered together to assess statistical significance. Univariate logistic regression results are
presented in Section 3.3.1. When the univariate logistic regressions were carried out for every
single question in the questionnaire, the research question being asked was "Are any of the
variables related to incidence of elevated blood-lead?," which is, in effect, the universal
hypothesis. In this stage of the analysis, the Holm procedure, a modification of the Bonferonni
32
-------
method (Holm, 1979) was used to adjust for the many simultaneous inferences being made.
Later, in the multivariate analysis, specific questions such as "Is there an increased risk due to
any R&R activity?," were each considered independently, without adjustment for multiple
comparison (Savitz et al., 1995).
Second, because it is known that a number of social and demographic factors affect
children's blood-lead concentrations (Bornschein et al., 1985; Brody et al., 1994; Lanphear et al.,
1996), multivariate logistic regression analyses were performed to assess the significance of
certain R&R variables after accounting for known effects. To account for known factors
affecting blood-lead concentration, a baseline (multivariate) logistic regression model was
developed to explain the risk of an elevated blood-lead concentration. This baseline model, as
well as the results from introducing R&R variables into the baseline model, are presented in
Section 3.3.2.
3.3.1 Univariate Statistical Analyses
Univariate logistic regression was used to assess the odds of an elevated blood-lead
concentration among various groups of children. A separate logistic regression analysis was
performed for each interview question and blood-lead registry variable. Odds ratios were
estimated to compare the odds of elevated blood-lead concentrations for children in each
response category.
An odds ratio is a fraction, with the odds for one group of children in the denominator
and the odds for a second group in the numerator (see Appendix B). The odds are the probability
that a subject has some condition over the probability that a subject does not have the condition.
If the odds are about the same for the two groups, then the ratio is close to 1. An odds ratio
statistically greater than 1 indicates that the second group of children (numerator group) is at
higher risk for the specified condition than the first group (denominator group). For these
analyses, the odds are the probability of a child having an elevated blood-lead concentration
relative to the probability of a child not having an elevated blood-lead concentration. The two
groups of subjects are defined by their responses to a particular question or values of a registry
variable. For example, the odds ratio for males versus females could be estimated.
In preliminary statistical analysis, logistic regression using two possible definitions of
nonelevated blood-lead concentrations was performed. For the first statistical analysis,
blood-lead concentrations 210 ^ig/dL were considered elevated, but only those blood-lead
concentrations <7 ug/dL were considered not elevated.4 Subjects with blood-lead concentrations
between 7 and 10 |Ag/dL were not considered. For the second statistical analysis, all blood-lead
concentrations <10 ug/dL were considered not elevated. Elevated blood-lead concentration was
always defined as ^ 10 jig/dL based on the Centers for Disease Control and Prevention definition
(CDC, 1994).
The cutoff for nonelevated blood-lead concentrations was chosen arbitrarily based on the study design,
which dictated 100 percent sampling of children with blood-lead concentrations >7 ug/dL.
33
-------
The second definition of not elevated (<10 ug/dL) is appealing because it allows all
observations to be used in the analysis and, thus, increases its statistical power by increasing the
sample size. However, the sample size of the nonelevated group is much larger than that of the
elevated group by either definition. Because excluding medium blood-lead concentrations
(7 ug/dL - 10 ug/dL) may magnify the difference between high and low groups, the first
definition may permit better detection of the covariates that affect a child's risk of an elevated
blood-lead concentration. Preliminary analysis suggested that the results would be similar based
on both definitions of nonelevated blood-lead concentration. Therefore, the first definition of
<7 fig/dL was used for nonelevated blood-lead concentration in all further analysis.
Blood-lead concentrations £ 10 ug/dL were considered elevated.
Columns 10,11 and 12 of Table 3-1 present univariate logistic regression results for
each of the blood-lead registry variables. Column 10 is the estimated log odds difference, 11 the
standard error of the log odds difference, and 12 a p-value indicating statistical significance. The
log odds difference is the logarithm of the odds ratio. Thus, exponentiating the log odds
difference yields the odds ratio.
For each of the six variables for which results are presented in Table 3-1, there is one
level with no values recorded in columns 10,11 and 12. This was the reference group for the log
odds differences reported for the other levels of the variable. For example, no log odds
difference values are reported for the "Caucasian" level of the race variable. Thus, each of the
other levels ("African-American," "Asian," "Native-American," and "Unknown") of log odds
differences compare that level to "Caucasian."
Based on the logistic regression results reported in Table 3-1, African-Americans and
Asians are estimated to be e0'92 = 2.5 and e1'03 = 2.8, respectively, more likely to have elevated
blood-lead concentrations than Caucasians. Males are more likely than females to have elevated
blood-lead concentrations (e40'34 = 1.4), and 1 year olds are more likely than 5-year olds (e40'87 =
2.4). Other results were not statistically significant.
A complete presentation of the logistic regression results from the questionnaire data is
provided in Appendix A. Of the variables in Appendix A, Appendix C lists all variables
significant at the 10 percent level. Of the variables in Appendix C, Table 3-8 lists those
significant at a 10 percent experiment-wise level based on the Holm multiple comparison
procedure.
When many tests of significance are conducted, some statistically significant results
might be obtained due to chance, even if there were no "real" effects. Thus, for 181
simultaneous tests, 18 significant results could be expected at the 10 percent significance level,
even if there were no responses related to the probability of elevated blood-lead concentrations.
The Holm procedure, a modification of the Bonferroni procedure (Holm, 1979), was
used to protect against making errors due to random chance (assigning significance when there is
no real effect, "false positive") while simultaneously making 181 inferences, i.e., interpreting
181 tests of significance. This procedure yielded an overall "experiment-wise" confidence level
of 90 percent. That is, when determining significance, all 181 tests are considered together,
34
-------
controlling the probability of making a "false positive" claim simultaneously across all 181 tests.
Table 3-8 presented questions that were significant at the 90 percent confidence level using the
Holm procedure.
Table 3-8. Significant (Based on the Holm Procedure) Univariate Odds Ratio Estimates and
Confidence Intervals"
All Children with Blood-Lead Concentrations <7//g/dL or *10//g/dL
n = 3,288
0.17
Q4C
0.1
Q6
Q1A
Q2
Q2
Q4B
Q4B
Q1A
Q6
Q13
Q1
Q3
Q11
Question
Heat guns used (inside)
Year residence was built
Rent or own current residence
Very little, some, or a lot of
peeling paint (inside)
Type of building (current)
Year current residence was built
Year current residence was built
Type of building (previous)
Type of building (previous)
Type of building (current)
Very little, some, or a lot of
peeling paint (inside)
Surface prepared by sanding or
scraping (inside)
Rent or own current residence
Has the tested child lived in
(current) home longer than
1 2 months
Any repair to broken plaster or
damaged walls in the room/rooms
painted
Higher Risk
Group
Yes
Pre-1940
Fee Free"
A lot of
Peeling Paint
Apartment/
Condo with 4
or Fewer Units
Pre-1940
Pre-1940
Apartment/
Condo with 4
or Fewer Units
Duplex
Duplex
Some Peeling
Paint
Yes
Rent
No
Yes
Lower Risk
Group
No
Post- 1980
Rent
None or Very
Little Peeling
Paint
Single Family
Home
Post-1 980
1970-79
Single Family
Home
Single Family
Home
Single Family
Home
None or Very
Little Peeling
Paint
No
Own
Yes
No
Odds Ratio
Estimate
3.456
3.32
2.974
2.691
2.612
2.586
2.226
1.859
1.804
1.522
1.462
1.419
1.405
1.336
1.31
95% Confidence Interval
[1.584, 7.538]
[1.105, 9.974]
[1.209,7.316]
[1.665,4.349]
[1.786, 3.819]
[1.699,3.935]
[1.377, 3.597]
[0.923, 3.743]
[0.951, 3.421]
[1.105, 2.096]
[1.041, 2.054]
[1.03, 1.954]
[1.083, 1.822]
[1.01, 1.768]
[0.99, 1 .733]
(a) These question responses are significant using the Holm's procedure at a 10 percent experiment-wise
significance level.
(b) "Fee free" means the respondent and child live free (no rent). For example, they live with grandparents.
35
-------
However, the Holm procedure conservatively corrects for the possibility of "false
positives" by possibly creating "false negatives"5 (Savitz et al., 1995). Therefore, test-wise
significant results were preserved and are presented in Appendix C. Based on what is already
known about lead exposure and elevated blood-lead concentration, some of these results may
represent useful information.
To save time during the telephone interview, the questionnaire was designed so that
groups of questions regarding a particular R&R activity would be skipped if the response to the
first question was negative (see Section 3.1.3). Thus, for questions such as "Were heat guns
used?," a large number of values are missing. Those missing values indicate that heat guns were
not used because paint removal did not occur at all. Therefore, for the logistic regression
analyses reported in Appendix A and summarized in Table 3-8 (and Appendices C and D),
missing values such as these were treated as "No" responses.
In Table 3-8, "Higher Risk Group" and "Lower Risk Group" refer to the odds of having
an elevated blood-lead concentration and indicate the group more or less likely to have elevated
blood-lead concentrations, respectively. For example, the table can be read "Children in the
higher risk group are odds ratio times more likely than children in the lower risk group to have
elevated blood-lead concentrations." Specifically, children who live in homes where heat guns
were used (to prepare a surface for interior painting) are 3.45 times more likely to have elevated
blood-lead concentrations than children in homes where heat guns were not used. Here "not
used" can mean either heat guns were not used while preparing a surface for interior painting or
no interior painting was done.
Table 3-8 presents 95 percent confidence intervals for each of the significant odds ratios
given. Some of these intervals include "1" because this is a more stringent confidence level than
that used for determining whether a question was significant. Odds ratios presented in Table 3-8
were sorted by magnitude.
Three R&R-related questions had (experiment-wise) statistically significant odds ratios.
The ratio was significant for using heat guns (inside), preparing surfaces by sanding or scraping
(inside), and repairing broken plaster or damaged walls (interior). Odds ratios were also
(experiment-wise) significant for a number of demographic, housing-related, and
occupational/hobby questions. As seen in Appendix C, the following additional R&R-related
questions were test-wise significant:
• The number of rooms where work was done
• Whether windows were repaired or new windows put in
• Whether work was done in the bathroom
• Whether work was done in the kitchen
• Whether heat guns were used (floor surface or covering disturbed)
• Whether carpet was replaced or removed or painted floors refinished
• Who did the (carpet or floor) work (owner, building superintendent, staff)
' Not assigning significance when there is a real effect is a "false negative."
36
-------
• Whether chemical paint removers were used (inside)
• Whether washing, wet scraping, wet sanding, or water blasting were used (outside)
• Whether power sanding, grinding, or sandblasting was used (outside)
• Whether chemical paint removers were used (outside)
• Whether there was any outside painting
• Whether an open flame torch was used (outside)
• Whether heat guns were used (outside)
• Who did the (outside) work (relative/friend not in household)
Given the conservative nature of the adjustments used to account for simultaneous inference, it is
likely that some of these results may be of public health significance. Recall that considering
only the experiment-wise statistically significant results and not the test-wise statistically
significant results increases the chance of not detecting significance when there is a real effect
(false negatives). The experiment-wise results do, however, properly ensure against the chance
of detecting significance when there is no real effect (false positive).
3.3.2 Multivariate Statistical Analyses
The basic approach for the multivariate statistical analysis consisted of the following
steps:
1. Combine responses from various social, demographic, and other non-R&R variables,
based on the results of the univariate analyses, for use in a multivariate logistic
regression model explaining blood-lead concentration
2. Construct a baseline logistic regression model of the probability of a child's
blood-lead concentration being greater than or equal to 10 ug/dL as a function of the
social, demographic, and other non-R&R-related variables
3. Define composite R&R variables that characterize R&R activities based on the
questionnaire responses
4. Assess whether R&R activities are associated with a significant increase in the
probability that blood-lead concentrations will be greater than or equal to 10 ug/dL by
introducing the R&R variables into the baseline model.
Baseline Model. Steps 1 and 2 of this analysis yielded a baseline logistic regression
model. The variables considered in constructing the baseline model are detailed in Table 3-9.
These variables were constructed from a combination of data provided from the blood-lead
registry and the questionnaire. The variables considered for the baseline model can be divided
into three types: 1) variables defining the type and condition of the child's residence,
2) demographic factors describing the child and the child's family, and 3) variables identifying
known sources of lead exposure affecting the child.
37
-------
Table 3-9. Variables Considered for the Baseline Logistic Regression Model
Variable
Levels
Basis
Type and Condition of Residence
Peeling Paint Inside or Outside'
Home Age
Rent or Own
Type of Residence
Yes/no
<1960; *1960
Rent/Own
Single family; duplex; apt. with four
or fewer units; apt. with five or
more units; mobile home/trailer;
don't know/refused
Q6 or Q7 of questionnaire
Q.2 or Q4c of questionnaire
Q1 of questionnaire
Q1a or Q4b of questionnaire
Demographic Factors
Medical Assistance Status
Income
Number of People
Education Level
Age
Medical Assistance/Nonmedical
Assistance
< 30,000; ^ 30,000
1, 2,3, ...
High school or less; some education
beyond high school; college
graduate
Quadratic and linear effects of age
were considered.
Blood-lead registry variable
Q335 of questionnaire
Q333 of questionnaire
Q333d of questionnaire
Age in days was calculated from
the blood-lead registry variables
for samples called on date and
birthdate.
Other Known Lead Exposures
Adult Exposure*
Yes/no
Q306 - Q322 or Q324 - Q332 of
questionnaire
(a) If the answer to any of the following questions was yes, this indicator variable was set to 1.
Because 22 percent of the children in this study moved during the year prior to then-
telephone interviews (see Appendix A, Question 3) and because telephone interviews were
collected several months after blood sampling, dates of blood sampling were compared to the
reported move date for each of the 809 families who moved. The child's residence during the
month of blood collection was used to determine the home age and type of residence variables.
Peeling paint and rent-versus-own questions were not asked for the respondent's previous
residence. Thus, these variables were set according to current residence responses for all
children. Three hundred sixty-eight families moved after the month of blood collection.
Backwards variable selection was used to select the baseline logistic regression model
using the variables provided in Table 3-9 (Neter and Wasserman, 1974). Three variables
describing the type and condition of the child's residence were significant in the baseline model:
age of home, type of residence, and peeling paint. Five demographic variables describing the
child and the child's family were significant in the model: education level, medical assistance
status, family income, number of household residents, and child's age (both the linear and
38
-------
quadratic effects). Adult exposure, the only variable indicating possible exposure from a source
other than lead-based paint, was also significant in the model.
Income and medical assistance status are related variables. However, because children
were selected into the study with their medical assistance status taken into consideration and
income is believed to be a more reliable data response, both terms were kept in the model. Older
housing was defined to be pre-1960 housing. This decision was made based on the univariate
analysis of the age of house variable provided in Appendix A (Q2). This analysis suggested that
only post-1970 housing could be statistically distinguished from pre-1940 housing in terms of
the odds of an elevated blood-lead concentration. In order to achieve a more even distribution of
observations between the older and newer housing groups, 1960-1970 homes were added to the
newer group.
Table 3-10 presents estimated odds ratios and corresponding 95 percent confidence
intervals for each of the categorical baseline model variables. Figures 3-2 and 3-3 present plots
of estimated odds ratios for a child's age and number of household residents. In the odds ratios
depicted in Figures 3-2 and 3-3, the reference group for the numerator of the odds ratio is
specified by the x-axis value. The reference group for the denominator of the odds ratio is held
constant in these figures. For Figure 3-2, the reference group for the denominator is 1-year-old
children. Figure 3-2 shows that a child's risk of having an elevated blood-lead concentration
level peaks at around age 2. By age 4, the risk is lower than it is for 1 year olds. Note how the
confidence interval narrows as age increases due to the negative correlation between the
estimated effects of the linear and quadratic terms. In Figure 3-3, the reference group for the
denominator is children living in a home with three residents. Figure 3-3 indicates that more
people living in a child's home increases the risk of an elevated blood-lead concentration. Odds
ratios are provided rather than parameter estimates for the baseline model to aid in interpreting
the model.
Addition of R&R Variables into Baseline Model. The next step was to construct R&R
variables from questionnaire responses for incorporation into the baseline model. Following
that, the R&R variables were added as explanatory variables to the baseline model, and the
significance of the added variables was assessed.
Both individual R&R variables, based on questionnaire responses from a single question,
and composite R&R variables, based on responses from multiple questions, were constructed.
The composite R&R variables generally correspond to the variables constructed for the "Any
R&R Activity," column of the "R&R Activity Exposure Summary" table (Table 3-2), and the
individual R&R variables generally correspond to the variables constructed for Table 3-3. As
noted in the discussion of these tables, most of the composite variables (particularly the specific
surface preparation activities) exhibited similar exposure patterns across all four types of R&R
activity. Use of the composite R&R variable rather than the individual question responses in the
multivariate analysis assumes that the effect of activities like "use of a heat gun" on blood-lead
concentration was the same any time a heat gun was used to remove paint, regardless of whether
39
-------
1.9 •
1.8 •
1.7 -
1.6 -
1.5 •
1.4 •
1.3 -
0.8 -
0.7 •
0.6 -
0.5 -
0.4
0.3
0.2 -
0.1
_»« Ratio
ower g5% Confidence Bound
ppez 95% Confidence Bound
3
AGE
Figure 3-2. Odds Ratio Comparing Odds of an Elevated Blood-Lead Concentration for a
Child of Age Specified by the x-Axis to a 1-Year-Old Child's Odds
2.8 -i
2.7
2.6
2.5 :
2.4 :
2.3 -
2.2
2.1 :
02.0 •
*jl.9 •
Si.8
•§1.7
01.6 ^
1.5
1.4 ^
1.3
1.2
1.1
1.0
9 :
8
rf5f Confidence
r 95% Conzldence
Jound
Sound
45678
Number of Household Residents
10
Figure 3-3. Odds Ratio Comparing Odds of an Elevated Blood-Lead Concentration for a
Child in a Household with Number of Residents Specified by the x-Axis to a
Child in a Household with Three Residents
40
-------
Table 3-10. Odds Ratio Estimates and Confidence Intervals for Baseline Logistic
Regression Model
Variable
Adult Exposure
Education Level
Home Age
Income
Medical Assistance Status
Peeling Paint Indicator
Type of Residence
Higher Risk
Group
Yes
High School or
less
High School or
less
Older
< $30,000
No
Yes
Single
Mobile
Apt.s Four
Units
Duplex
Lower Risk Group
no
College Graduate
or More
Some College
Newer
> $30,000
Yes
no
Apt. 2 Five Units
Single
Single
Single
Odds Ratio
Estimate
1.318
1.385
1.305
2.104
1.901
1.360
1.487
2.127
1.166
1.658
1.864
95% Confidence
Interval
[1.055, 1.646]
[0.999, 1.919]
[1.032, 1.651]
[1.632, 2.713]
[1.492, 2.422]
[1.078, 1.715]
[1.191, 1.857]
[1.130,4.003]
[0.749, 1.817]'
[1.098, 2.504]
[1.371, 2.534]
(a) Not a statistically significant odds ratio.
it occurred during interior or exterior painting, other repairs, or carpets and floor repair or
replacement, or any combination of the four R&R activity types. Table 3-11 presents the R&R
variables constructed, as well as the questions that provided the basis for the variables.
The constructed R&R variables for the multivariate analysis were coded a little
differently from those constructed for Tables 3-2 and 3-3. If respondents who were not asked a
particular question because of the skip patterns in the questionnaire, the R&R variable for the
question was coded as "No" instead of "Missing." (Questions with ordinal responses such as the
"number of rooms in which work was done" had the respective R&R variable set to 0 instead of
"Missing"). This coding change resulted in more observations being included in the multivariate
analysis, where any observation with an explanatory variable set to "Missing" would have been
excluded in the logistic regression.
Five sets of logistic regression models were created using the constructed R&R variables
hi increasing order of complexity. In the first set, each R&R variable in Table 3-11 was the
single exploratory variable in a set of univariate logistic regressions. This univariate analysis
was carried out as a preliminary step hi the multivariate analysis for exploring the effect of the
R&R variables. In the second set of models, each R&R variable was individually included as an
explanatory variable in the baseline model (i.e., each model could only have one R&R
explanatory variable). The third set of models was developed from the second set by including
all the other R&R variables that "nest" the single R&R variable already in the model. A variable
is said to "nest" another if the value of the former restricts the value of the latter. In the case of
R&R variables, "nesting" occurs due to the skip patterns in the questionnaire, where the response
to one R&R question can determine whether another question is skipped (remembering that
41
-------
variables were coded "No" or 0 if the question was skipped). The fourth set consisted of a single
model that was created by including all the R&R variables in Table 3-11 as additional
explanatory variables in the baseline model. A restricted form of stepwise backward elimination
was carried out on this model, where explanatory variables could be eliminated only if they were
insignificant variables that did not nest another variable in the model. In the final set of models,
a single interaction term between an R&R variable and an age-of-house variable (pre- versus
post-1980) was added to the baseline model to test whether a ban on lead paint in 1978 reduced
the association between incidence of R&R activities and elevated blood-lead concentrations in
children. Note that all sets of models except the first use the baseline model as a basis on which
to build the models. Hence, in the interpretations to follow, the R&R effect in the last four sets
of models should always be implicitly qualified with the expression "taking into account the
effect of non-R&R factors on the risk of elevated blood-lead concentrations in children."
Table 3-11. Renovation and Remodeling Variables Used to Assess Relative Risk
Variable
Indicator of R&R Activity
Any R&R Work
Inside Painting
Window Repair or Replacement
Inside or Outside Painting
Prepared Surface
PrejDared Surface for Inside Painting
Hand Sanding or Scraping
Power Sanding, Grinding, Sandblasting
Open Flame Torch
Heat Gun
Washing, Wetscraping, Water Blasting
Chemical Paint Removers
Levels
Yes/No
Yes/No
Yes/No
Yes/No
Yes/No
Yes/No
Yes/No
Yes/No
Yes/No
Yes/No
Yes/No
Yes/No
Basis'
Q10, Q28, Q39. Q54
Q10
Q39
Q10, Q28
Q13.Q29.Q42.Q57
Q13
Q14, Q30, Q44, Q58
Q15, Q31, Q45, Q59
Q16, Q32, Q47, Q60
Q17, Q33, Q48, Q61
Q18, Q34, Q49, Q62
Q19, Q35, Q50, Q63
Who Did the Work?
Head of the Household or Spouse
Other in Household
Relative or Friend Not in Household
Owner or Apartment Staff
Professional
Yes/No
Yes/No
Yes/No
Yes/No
Yes/No
Q20_1. Q36J, Q51_1,
Q20_2, Q36_2, Q51_2,
Q20_3, Q36_3, Q51_3,
Q20_4, Q36_4, Q51_4,
Q20_5, Q36_5, Q51_5,
Q64_1
Q64_2
Q64_3
Q64_4
Q64_
Lived in Home While R&R Was Done
Number of Rooms
R&R Work in Kitchen
Yes/No
0-14
Yes/No
Q22, Q30, Q53, Q66
023
Q24
(a) If the answer to any of the following questions was yes, this indicator variable was set to 1.
42
-------
Appendix D presents univariate logistic regression results from the first set of models and
some descriptive statistics for each of the R&R variables described in Table 3-11. Significant
R&R activity results were indicated for:
• Conducting any R&R work
• Painting done inside or outside
• Preparing surfaces
• Power sanding, grinding, sandblasting
• Using heat guns
• Washing, wetscraping, waterblasting
• Using chemical paint removers
• Who did the work (owner or apartment staff)
• Living in the home while R&R work was done
• Number of rooms
• Work done in kitchen.
Table 3-12 lists the p-values of the estimated coefficients, the odds ratios, and confidence
intervals of the odds ratios for the R&R variables from the second set of models. The odds ratios
in this set of models compare the odds of a child having elevated blood-lead concentrations
when a particular R&R activity occurred against the odds of a child having elevated blood-lead
concentrations when the particular R&R activity did not occur. To illustrate, take the case where
"Open Flame Torch" was included as the single R&R explanatory variable. The odds ratio
compares the risk of a child having elevated blood-lead concentrations when an open flame torch
was used against the risk of a child having elevated blood-lead concentrations when no open
flame torch was used, which included the case when no surface preparation (or, for that matter,
any R&R activity) was carried out. The reason for calling these odds ratios "unconditional" will
become clear when the "conditional" odds ratios are defined.
The significant results in Table 3-12 indicate that R&R activities of some kind
significantly increase the odds of a child having elevated blood-lead concentration, although this
result appears to be partially driven by the effect of "Outside Painting," as evidenced by the
significance of "Inside or Outside Painting" and the insignificance of "Inside Painting." Many of
the particular R&R activities lead to significantly increased risk. The odds are higher when a
greater number of rooms are worked on. Results also indicate that the odds of an elevated
blood-lead concentration is higher if R&R work is either not done or conducted by someone that
is not an "Other in household" than the odds if R&R is conducted by an "Other in household."
The odd feature of this result is that no R&R work done is a subset of the higher risk group. This
last result is the only statistically significant result in this study that indicates that an R&R
activity of any type conducted by any individual could reduce the risk due to lead exposure. The
univariate logistic regression also showed that someone not in the household carrying out the
work reduced the odds of an EBL, although the result was not statistically significant
(Appendix D).
Table 3-13 lists the p-values of the estimated coefficients, the odds ratios, and the
confidence intervals of the odds ratios of all the R&R variables from the third set of models,
43
-------
Table 3-12. Unconditional Odds Ratios from Logistic Regression with Single R&R
Variables
Variable '
Any R&R Work
Inside Painting
Window Repair or
Replacement
Inside or Outside Paintina
Prepared Surface
Prepared Surface for Inside
Paintina
Hand Sanding or Scraping
Power Sanding, Grinding,
Sandblasting
Open Flame Torch
Heat Gun
Washing, Wetscraping,
Water Blasting
Chemical Paint Removers
Hiaher Risk
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Lower Risk
No
No
No
No
No
No
No
No
No
No
No
No
P-Value
0.0220
0.9267
0.4652
0.0116
0.0038
0.0645
0.1158
0.1035
0.0101
<0.0001
0.0092
0.0046
Odds Ratio
1.309*
1.010
1.095
1 .322*
1.430*
1.325
1.226
1.372
4.883*
4.597*
1.625*
1.969*
Confidence
Interval
(1.035,1.656)
(0.814,1.252)
(0.855,1.402)
<1. 060.1. 649)
(1.117,1.830)
(0.977,1.796)
(0.946,1.588)
(0.930,2.025)
(1.423,16.759)
(2.715,7.782)
(1.119,2.360)
(1.220,3.176)
Who Did the Work?
Head of the Household or
Spouse
Other in Household
Relative or Friend Not in
Household
Owner or Apartment Staff
Professional
Yes
No
Yes
Yes
Yes
No
Yes
No
No
No
0.1696
0.0355
0.0015
0.4787
0.1195
1.214
3.000*
2.231*
1.244
1.490
(0.915,1.611)
(1.055.8.531)
(1.344,3.705)
(0.672,2.305)
(0.893,2.486)
Lived in Home While R&R
Was Done
Number of Rooms
R&R Work in Kitchen
Yes
1b
Yes
No
Ob
No
0.0163
0.0007
0.0243
1.365*
1.119*
1.569*
(1.054.1.769)
(1.047,1.197)
(1. 052.2.340)
Note: Shaded area (and asterisk) indicates statistically significant results.
(a) See Table 3-11 for variable definitions.
(b) "Number of Rooms" was included as an ordinal variable. The risk groups were chosen for Illustration.
More generally, the odds ratio between n+k and n rooms is 1.119k.
44
-------
Table 3-13. Conditional Odds Ratios from Logistic Regressions with Single R&R Variables
(and Any Other R&R Variables Nesting Them)
Variable a
Any R&R Work
Inside Painting
Window Repair or
Replacement
Inside or Outside Paintinq
Prepared Surface b
Prepared Surface for Inside
Paintina °
Hand Sanding or Scraping d
Power Sanding, Grinding,
Sandblasting
Open Flame Torch d
Heat Gun d
Washing, Wetscraping,
Water Blasting d
Chemical Paint Removers d
Higher Risk
Yes
Yes
Yes
Yes
Yes
Yes
No
Yes
Yes
Yes
Yes
Yes
Lower Risk
No
No
No
No
No
No
Yes
No
No
No
No
No
P-Value
0.0220
0.9267
0.4652
0.0116
0.0152
0.0349
0.0085
0.8079
0.0280
0.0000
0.2101
0.0507
Odds Ratio
1.309*
1.010
1.095
1.322*
1.383*
1.419*
2.385*
1.055
3.870*
4.091*
1.313
1.654
Confidence
Interval
(1.035,1.656)
(0.814,1.252)
(0.855,1.402)
f 1.060.1. 6491
(1.059,1.806)
(1.018,1.976)
(1.232,4.614)
(0.678,1.643)
(1.129,13.264)
(2.345,7.137)
(0.850,2.029)
(0.988,2.769)
Who Did the Work?
Head of the Household or
Spouse d
Other in Household d
Relative or Friend Not in
Household d
Owner or Apartment Staff d
Professional d
No
No
Yes
No
Yes
Yes
Yes
No
Yes
No
0.1132
0.0065
0.0250
0.8726
0.5364
1.439
4.160*
1.831*
1.053
1.184
( 0.909,2.279)
(1.460.11.853)
(1.067,3.141)
(0.553,2.004)
(0.686,2.041)
Lived in Home While R&R
Was Done d
Number of Rooms e
R&R Work in Kitchen e
No
1f
Yes
Yes
0(
No
0.8720
0.0023
0.1668
1.049
1.178*
1.470
(0.577,1.907)
(1.058,1.311)
(0.842.2.568)
Note: Shaded area (and asterisk) indicates statistically significant results.
(a) See Table 3-11 for variable definitions.
(b) Odds ratio conditional on "Any R&R Work" = 'Yes'.
(c) Odds ratio conditional on "Inside Painting" = 'Yes'.
(d) Odds ratio conditional on "Prepared Surface" = 'Yes1.
(e) Odds ratio conditional on "Prepared Surface for Inside Painting" = 'Yes'.
(f) "Number of Rooms" was included as an ordinal variable. The risk groups were chosen for illustration.
More generally, the odds ratio between n+k and n rooms is 1.178k.
45
-------
which adds a single R&R variable and any other R&R variables nesting that particular variable
to the baseline model. (Recall that, a variable is said to "nest" another if the value of the former
restricts the value of the latter. In the case of R&R variables, "nesting" occurs due to the skip
patterns in the questionnaire, where the response to one R&R question can determine whether
another question is skipped.) By including these nested R&R variables, the models are taking
the skip patterns into account. The odds ratio in this set of models compares the odds of a child
having an EBL when a R&R event occurred against the odds of a child having an EBL when
some alternative R&R activity (or activities) occurred instead. To illustrate, take the case in
which "Open Flame Torch" and all the R&R variables that nest it ("Any R&R Work" and
"Prepared Surface") were included as explanatory variables. The skip pattern in the
questionnaire ensured that only those respondents who answered "Yes" to "Any R&R Work"
and "Prepared Surface" could answer the question concerning whether they used an open flame
torch. The odds ratio therefore compares the odds of a child having elevated blood-lead
concentrations when an open flame torch was used to prepare the surface against the odds of a
child having elevated blood-lead concentrations when some method(s) other than an open flame
torch was (were) used instead to prepare the surface. Alternatively, the "Open Flame Torch"
odds ratio could be interpreted as comparing the risk of using an open flame torch against not
using an open flame torch, given that at least some form of surface preparation was carried out.
This explains the use of "conditional" to describe the odds ratios from the third set of models.
The results in Table 3-13 are fairly consistent with the results in Table 3-12. In
Table 3-13, the odds of an EBL were significantly higher by a factor of 1.419 if the surface was
prepared for inside painting, given that some inside painting was carried out; but in Table 3-12
the unconditional odds for the same variable bordered on insignificant. The unconditional odds
ratio for "Hand Sanding or Scraping" in Table 3-12 (1.226) is insignificant but much lower than
the unconditional odds ratio for "Prepared Surface" (1.430). This explains the difference in the
conditional odds ratios of Table 3-13, where the odds are significantly lower if hand sanding or
scraping is carried out, given that the surface is prepared. Note that the results for the first
four rows are the same for Tables 3-12 and 3-13 since the R&R variables in those rows are not
nested by any other R&R variables.
Table 3-14 lists the p-values of the estimated coefficients, the odds ratios, and the
confidence intervals of the odds ratios for all the R&R variables from the model created by
including all the R&R variables from Table 3-11 as additional explanatory variables in the
baseline model and eliminating the insignificant R&R variables using stepwise backward
elimination. The backward elimination was restricted so that only those R&R variables that did
not nest another significant R&R variable could be eliminated at each step.
Note that some insignificant R&R variables were retained in the final model due to the
restriction in the backward elimination that accounted for the skip patterns. The final model is
fairly consistent with the results from the previous two sets of models with single R&R variables.
"Any R&R Work," which was significant in the analysis of single R&R variables, becomes
insignificant, probably due to including the two painting variables that are closely associated
with the variable. The significance of "Inside Painting" lowering the risk is misleading since the
effect is negated by the "Inside or Outside Painting" variable. "Prepared Surface for Inside
Painting" is insignificant probably due to its close association with "Prepared Surface."
46
-------
Table 3-14. Conditional Odds Ratios from Logistic Regression with Multiple R&R
Variables
Variable '
Any R&R Work
Inside Painting
Inside or Outside Paintina
Prepared Surface b
Prepared Surface for Inside
Paintina c
Hand Sanding or Scraping d
Heat Gun d
Other in Household d
Relative or Friend Not in
Household d
Number of Rooms e
HiaherRisk
Yes
No
Yes
Yes
No
No
Yes
No
Yes
1f
Lower Risk
No
Yes
No
No
Yes
Yes
No
Yes
No
Of
P-Value
0.8518
<0.0001
0.0005
0.0099
0.1189
0.0026
0.0001
0.0060
0.0028
0.0014
Odds Ratio
1.039
2.432*
2.517*
2.537*
1.629
2.795*
4.138*
4.456*
2.339*
1.194*
Confidence
Interval
(0.690,1.565)
(1.605,3.685)
M. 484.4.268)
(1.232,5.224)
(0.871,3.048)
(1.411,5.538)
(2.269,7.548)
(1.502,13.225)
(1.324,4.131)
M. 069.1. 334)
Note: Shaded area (and asterisk) indicate statistically significant results.
(a) See Table 3-11 for variable definitions.
(b) Odds ratio conditional on "Any R&R Work" = 'Yes'.
(c) Odds ratio conditional on "Inside Painting" = 'Yes'.
(d) Odds ratio conditional on "Prepared Surface" = 'Yes'.
(e) Odds ratio conditional on "Prepared Surface for Inside Painting" = 'Yes'.
(f) "Number of Rooms" was included as an ordinal variable. The risk groups were chosen for illustration.
More generally, the odds ratio between n+k and n rooms is 1.194*.
The surprising result that was consistent across all the multivariate models is that the odds
of a child having an elevated blood-lead concentration is significantly lower if the surface
preparation was carried out by someone in the household other than the head of household or
spouse. Compared to professionals who had (insignificantly) higher odds, the non-head-of-
households might be expected to be less careful about the spread of lead from R&R work. There
was some concern that the unexpected results obtained for work done by some household
member other than the head of household may be due to the type of R&R activity performed or
the size of the job. Table 3-15 presents information on the number and percentage of specific
R&R activities performed by various types of individuals.
Based on Table 3-15, it can be concluded that all types of individuals use the various R&R
techniques considered with similar frequencies. It appears that professionals and owners
undertake larger jobs than those living in the household and relatives and that different sized jobs
are tackled by different methods. However, since the odds ratio in the final model with multiple
R&R variables accounts for the differences in the number of rooms, there is no obvious
indication of why work performed by a household member other than the head of household (or
spouse) should decrease the odds of an elevated blood level concentration. No conclusive
resolution to this counter-intuitive result could be produced.
47
-------
Table 3-15. Breakdown of Type of Specific R&R Activity by Type of Individual Who Did
Work
R&R Activity
Any R&R Work3
Inside Painting"
Window Repair or
Replacement"
Inside or Outside Painting"
Prepared Surface"
Prepares Surface for Inside
Paintina0
Hand Sanding or Scraping"
Power Sanding, Grinding,
or Blasting"
Open Flame Torch"
Heat Gun"
Washing, Wetscraping,
Water Blasting"
Chemical Paint Removers"
Lived in Home While R&R
Done"
Number of Rooms
Who Did the Work
Head of
Household
or Spouse
(n=488)
488
(100%)
417
(85%)
169
(35%)
469
(96%)
488
(100%)
281
(67%)
461
(94%)
146
(30%)
6
(1%)
43
(9%)
160
(33%)
80
(16%)
452
(93%)
2.9
Other in
Household
(n=69)
69
(100%)
59
(86%)
28
(41%)
63
(91%)
69
(100%)
49
(93%)
65
(94%)
23
(33%)
-
3
(4%)
25
(36%)
13
(19%)
64
(93%)
3.4
Relative or
Friend not in
Household
(n=71)
71
(100%)
59
(83%)
30
(42%)
68
(96%)
71
(100%)
47
(80%)
68
(96%)
35
(49%)
-
5
(7%)
27
(38%)
13
(18%)
59
(83%)
3.3
Owner or
Apartment
Staff
(n=52)
52
(100%)
31
(60%)
17
(33%)
51
(98%)
52
(100%)
18
(58%)
47
(90%)
10
(19%)
-
3
(6%)
7
(13%)
6
(12%)
40
(77%)
4.3
Professional
(n=70)
70
(100%)
60
(86%)
36
(51%)
69
(99%)
70
(100%)
33
(55%)
64
(91%)
24
(34%)
4
(6%)
5
(7%)
17
(24%)
10
(14%)
59
(84%)
3.6
Number ol
Rooms
2.8
2.9
3.2
2.9
2.9
2.9
2.9
3.4
5.3
3.5
3.2
4.2
2.8
a. Percentage in parentheses is in relation to n in column labels.
b. Percentage in parentheses Is in relation to frequency in "R&R Work" row.
c. Percentage in parentheses is In relation to frequency in "Inside Painting" row.
d. Percentage in parentheses is in relation to frequency in "Prepared Surface" row.
For the final set of models, a single interaction between each R&R variable and an age of
house variable was included as an additional explanatory variables in the baseline model. The
significance of the interactions was examined in an attempt to identify differences in risk due to
48
-------
lead exposure between R&R activities performed in pre- and post-1980 homes. When the age of
house variable was defined as either pre- or -post-1980, the interaction term in all the models
were insignificant. Hence, the available data did not provide any evidence that the 1978 ban on
lead-based paint affected the odds of elevated blood-lead concentrations in children as a result of
R&R activities. Similar analysis using a three-level age of house variable (pre-1960, 1960-1980,
and post-1980) produced the same results. There are many possible reasons for this finding:
1. Age of residence was an interview reported variable, and, thus, may have been
reported inaccurately for some homes.
2. Only the manufacturing of lead-based paint was banned in 1978. Residences may
have been painted with lead-based paint using old paint.
3. R&R activities in residences may increase risk due to childhood lead exposure by
stirring up lead-contaminated dust from sources other than lead-based paint.
3.3.3 Statistical Results
Although medical assistance status was a design variable in this study, it was not treated
as such in the analysis of the data. Table 3-1 indicates the univariate relationship between
medical assistance status and odds of an elevated blood-lead concentration. This relationship is
not statistically significant. Medical assistance status is, however, a significant factor in the
multivariate baseline logistic regression model. Its inclusion in this model accounts for the effect
of medical assistance status in the multivariate analysis. In preliminary analysis, univariate
regression results were calculated separately for children receiving medical assistance and
children not receiving medical assistance, as might normally be done for a design variable. The
results (the effects of other demographic and R&R variables on odds of an elevated blood-lead
concentration) were similar for both medical assistance statuses; and, thus, a decision was made
to pool the data across medical assistance status. The fact that medical assistance status was
accounted for in the multivariate model is one reason for preferring the results based on the
multivariate analysis to the univariate results.
The multivariate results reported are based on the statistical significance of composite
R&R activity variables when introduced into a baseline logistic regression model. The
composite R&R variables were constructed from responses to multiple interview questions. The
baseline model accounted for the effects of a number of housing factors, demographics, and other
lead exposures. This approach identified more statistically significant effects than univariate
analysis of individual question responses (Section 3.3.1) for two reasons:
1. Variability due to other known lead exposure factors was reduced by using a baseline
model when assessing the impact of an R&R activity.
2. The overall incidence of the composite R&R activity variables was higher. This
makes it easier to assess the effect of a variable on the odds of an elevated blood-lead
concentration.
49
-------
The analysis documented in this report demonstrated a statistically significant
relationship between R&R activities and the odds of elevated blood-lead concentrations.
Specifically, children residing in residences in which some R&R activity was conducted in the
last 12 months were estimated to have odds of an elevated blood-lead concentration 1.3 times
greater than children residing in residences where no R&R was conducted. One reason for this
increase in odds appeared to be due to the significant increase in odds in residences where
outside painting was carried out. When paint removal using a heat gun was performed at a
residence, the odds of an elevated blood-lead concentration were highly significant (4.6 times
greater) than if the work was not performed. (The odds were over 4 times greater when
compared to the case when some other type of surface preparation was carried out and to the case
when the effects of other R&R factors were taken into account). Conversely, hand sanding and
scraping appeared to have significantly smaller odds given that some alternative form of surface
preparation was carried out. Increasing the number of rooms in which surface preparation was
carried out for inside painting also increased the risk of EBL in children. Finally, a relative or
friend not in the household increased the risk of elevated blood-lead concentrations in children
significantly, possibly due to the lack of care taken in preventing the spread of lead in the house
during the R&R activities. The significant and unexpected reduction in odds when someone in
the household other than the head or spouse carried out the R&R activity could not be explained.
50
-------
4.0 CONCLUSIONS
This study demonstrated that general residential R&R is associated with an increased risk
of EBLs in children and that specific R&R activities are also associated with an increase in the
risk of EBLs in children. In particular, removing paint (using open flame torches, using heat
guns, using chemical paint removers, and wet scraping/sanding) and preparing surfaces by
sanding or scraping significantly increased the risk of EBLs. Overall, these results agree with
those from earlier phases of the R&R Study—R&R activities that disturb lead-based paint
increase the risk of exposure to occupants. Additionally, children living in a residence while
R&R was conducted were 1.3 times more likely to have EBLs than children who did not live in a
residence while R&R was conducted.
The study also characterized the exposure of residents to R&R activities. At least one
R&R activity such as inside painting, outside painting, carpet and floor repair or replacement, or
other repairs (e.g., window repair) had been conducted in 67.2 percent of the study residences in
the previous 12 months. Some form of surface preparation was involved in 42.3 percent of R&R
activities. Most surface preparation involved hand scraping or sanding. Heat guns were used for
surface preparation 7 percent of the time, and chemical paint removers were used 13.6 percent of
the time.
The results of this study point toward a continuing need to educate parents and guardians
about avoiding more risky R&R activities. Parents or guardians also could be educated about the
positive benefits of relocation while R&R is being performed in a residence.
Further, the study has identified specific R&R activities and other conditions (such as age
and type of residence) that are associated with increased risk to children as a result of lead
exposure. This information can be used to develop regulations that focus on particular R&R
activities (e.g., using a heat gun to remove paint), the groups of persons (e.g., a household
member other than the head of household or spouse) who perform the activities, and the other
conditions (e.g., adult exposure, age of child) that significantly increase the risk to children. The
results of this study concerning activities associated with increased risk (e.g., using a heat gun to
remove paint) also can be combined with the worker profile results from other phases of the
R&R study to perform an overall assessment of the worker groups or situations where
interventions are needed to reduce exposure from R&R.
51
-------
5.0 REFERENCES
Bornschein, R. L., Succop, P. A., Dietrich, R. N., Clark, C. S., Que Hee, S., and Hammond, P.B.,
(1985) "The Influence of Social and Environmental Factors on Dust Lead, Hand Lead, and
Blood-Lead Levels in Young Children." Environmental Research 38:108-118.
Brody, D. J., Pirkle, J. L., Kramer, R. A., Flegal, K. M., Matte, T. D., Gunter, E. W., and
Paschal, D. C., (1994) "Blood-Lead Levels in the US Population: Phase I of the Third National
Health and Nutritional Examination Survey (NHANES III, 1988 to 1991)." Journal of the
American Medical Association 272(4):277-283.
CDC (see Centers for Disease Control)
Centers for Disease Control (1991) "Preventing Lead Poisoning in Young Children: A
Statement by the Centers for Disease Control." Public Health Service, Centers for Disease
Control and Prevention, U.S. Department of Health and Human Services, October 1991.
Centers for Disease Control (1994) "Plan and Operation of the Third National Health and
Nutrition Examination Survey, 1988-94." National Center for Health Statistics, Centers for
Disease Control and Prevention, U.S. Department of Health and Human Services, DHHS
Publication No. (PHS), 94-1308.
EPA (see U.S. Environmental Protection Agency)
Groves, R. M., Bieme, P. B., Lyberg, L. E., Massy, J. T., Nicholls, W. L., Waksberg, J., (1988)
Telephone Survey Methodology. John Wiley & Sons, New York.
Holm, Sture (1979) "A Simple Sequentially Rejective Multiple Test Procedure." Scandinavian
Journal of Statistics, 6, 65-70.
Lanphear, B. P., Weitzman, M., Winter, N. L., Eberly, S., Yakir, B., Tanner, M., Emond, M., and
Matte, T. D., (1996b) "Lead-Contaminated House Dust and Urban Children's Blood-Lead
Levels." American Journal of Public Health 86 (10): 1416-1421.
McCullough, P., and Nelder, J. A., (1989) Generalized Linear Models. Chapman and Hall,
London.
Neter, J., and Wassemian, W., (1974) Applied Linear Statistical Models, Homewood, IL:
Richard D. Irwin, Inc.
Savitz and Olshan (1995) "Multiple Comparisons and Related Issues in the Interpretation of
Epidemiologic Data." American Journal of Epidemiology 142( 9).
Thompson, S., (1992) Sampling. John Wiley & Sons.
52
-------
U.S. Environmental Protection Agency (1994) "Reducing Lead Hazards When Remodeling
Your Home." Office of Pollution Prevention and Toxics, U.S. Environmental Protection
Agency, EPA 747-R-94-002, April 1994.
U.S. Environmental Protection Agency (1995) "Seasonal Rhythms of Blood-Lead Levels."
Office of Pollution Prevention and Toxics, U.S. Environmental Protection Agency, EPA 747-R-
94-003, September 1995.
U.S. Environmental Protection Agency (1996) "Seasonal Trends in Blood-Lead Levels in
Milwaukee." Office of Pollution Prevention and Toxics, U.S. Environmental Protection Agency,
EPA 747-R-95-010, August 1996.
U.S. Environmental Protection Agency (1997a) "Lead Exposure Associated With Renovation
and Remodeling Activities: Summary Report." Office of Pollution Prevention and Toxics,
U.S. Environmental Protection Agency, EPA 747-R-96-005, May 1997.
U.S. Environmental Protection Agency (1997b) "Lead Exposure Associated With Renovation
and Remodeling Activities: Environmental Field Sampling Study." Office of Pollution
Prevention and Toxics, U.S. Environmental Protection Agency, EPA 747-R-96-007, May 1997.
U.S. Environmental Protection Agency (1997c) "Lead Exposure Associated With Renovation
and Remodeling Activities: Worker Characterization and Blood-Lead Study." Office of
Pollution Prevention and Toxics, U.S. Environmental Protection Agency, EPA 747-R-96-006,
May 1997.
U.S. Environmental Protection Agency (1997) "Lead Exposure Associated With Renovation and
Remodeling Activities: Phase IV, Worker Characterization and Blood-Lead Study of R&R
Workers Who Specialize in Renovation of Old or Historic Buildings." Office of Pollution
Prevention and Toxics, U.S. Environmental Protection Agency, Draft Final Report,
January 29, 1998.
U.S. Department of Housing and Urban Development (1995) "Guidelines for the Evaluation and
Control of Lead-Based Paint Hazards in Housing." Office of Lead-Based Paint Abatement and
Poisoning Prevention.
53
-------
This page intentionally blank.
54
-------
APPENDIX A:
EXPLORATORY ANALYSIS OF WISCONSIN CHILDHOOD
BLOOD-LEAD STUDY QUESTIONNAIRE DATA
A-1
-------
Exploratory Analysis of Wisconsin Childhood Blood-lead
Study Questionnaire Data
Skip Patterns
To save time during the telephone interviews, groups of questions regarding a particular R&R
activity were skipped if the response to the first question was negative (see Section 3.1.3). The
table on the next page shows the pattern of skipping questions.
Most of the questions concerning R&R activities were introduced with the phrase "In the last 12
months...." as in the first cell of the table on the next page. Reading across the table gives the
type of R&R activity that fills in the blank for the question number indicated in parentheses.
The first column gives the specific question, and the two columns below each activity type
specify the question number and who answered the question.
For example, for "any painting or surfaces prepared for paint INSIDE home?," the second
column indicates that this is question 10, and column three indicates that everyone answered that
question. Further down the first column, for the question "surface prepared by sanding/scraping .
..?," column two indicates that this is question 13, and column three indicates that anyone who
answered "Yes" to question 10 answered question 13. Likewise for the question
"Handsanding/handscraping used?" (question 14), anyone who answered "Yes" to question 13
answered question 14. This pattern can be followed down the table for a specific R&R activity
and across the table for types of R&R activities.
A-2
-------
In the last 12 mos was any of work done?
Any painting or surfaces prepared for paint INSIDE home
Any painting or surfaces prepared for paint OUTSIDE home
Any windows repaired or new windows put in
Any carpets replace/repaired or painted floors refinished
Repair broken plaster or damaged walls in the rooms
painted
Old walls taken down or moved
Any painting or surfaces prepared for paint with the
installation of windows
Do you have wall-to-wall carpet
Repair/replace wall-to-wallor other large carpet
Surface prepared by sanding/scraping off old paint
(adhesive, chemicals o.t. primer)
Handsanding/handscraping
Powersanding/grindlng/sandblasting
Open flame torch
Heatguns
Wetscraping/wet sanding/water blasting
Chemical paint removers
Who did the work
Did anyone live there while the work was being done?
Number of rooms where work was done
Work in Kitchen
Work in Bathroom
Were dust and dirt spread
Inside Painting (q 10)
Question
Number
q10
q11
q12
q13
q14
q15
q16
q17
q18
q19
q20 1-
q20_5
q22
q23
q24
q25
Who
Answered
Everyone
Everyone
Everyone
YestoqIO
Yestoq13
Yes to q1 3
Yes to q1 3
Yes to q1 3
Yes to q1 3
Yes to q1 3
Yes to q1 3
Yes to q1 3
Yestoq13
Yes to q1 3
Yestoq13
Outside Painting (q28)
Question
Number
q28
q29
q30
q31
q32
q33
q34
q35
q36 1-
q36_5
q38
Who
Answered
Everyone
Yes to q28
Yes to q29
Yes to q29
Yes to q29
Yes to q29
Yes to q29
Yes to q29
Yes to q29
Yes to q29
Windows repaired or replaced
(q39)
Question
Number
q39
q40
q41
q42
q43
q44
q45
q47
q48
q49
q50
q51_1-q51_5
q53
Who
Answered
Everyone
Yes to q39
Yes to q39
Yes to q39
Yes to q39
Yes to q43
Yes to q43
Yes to q43
Yes to q43
Yes to q43
Yes to q43
Yes to q43
Yes to q43
Carpets replace/removed or
>ainted floors refinished (q54)
Question
Number
q54
q55
q56
q57
q58
q59
q60
q61
q62
q63
q64_1-q64_5
q66
q67
Who
Answered
Everyone
Yes to q54
Yes to q54
yes to q56
Yes to q57
Yes to q57
Yes to q57
Yes to q57
Yes to q57
Yes to q57
Yes to q57
Yes to q57
Everyone
-------
HOUSING
Ql: First, I would like to ask you some questions about your housing. Do you own or rent
your home?
1. RENTING
2. OWN
3. FEE FREE*
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
3
8
•
Number of
PbB <7
/yg/dL
1245
1719
21
13
1
7 s PbB
< 10
//g/dL
169
188
5
3
1
PbB * 10
/yg/dL
141
139
7
2
0
Blood-Lead
Concentration
Geometric
Mean
2.67
2.30
3.61
3.78
5.92
Geometric
Standard
Deviation
2.67
2.66
2.57
2.97
1.18
Log-Odds Ratio
Estimate
-0.34
1.08
Standard
Error
0.13
0.45
P-value
0.0062
0.0147
* "Fee free" means the respondent and child live free (no rent). For example, they live with grandparents.
A-4
-------
Qla: What kind of building do you live in now? Is it a single family home, a duplex, an
apartment or condominium with 4 or fewer units, an apartment or condominium with 5
or more units, or a mobile home or trailer?
1. SINGLE FAMILY HOME
2. DUPLEX
3. APARTMENT/CONDOMINIUM WITH 4 OR FEWER UNITS
4. APARTMENT/CONDOMINIUM WITH 5 OR MORE UNITS
5. MOBILE HOME OR TRAILER
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
3
.
Number of
PbB <7
//g/dL
2327
457
200
15
7 <: PbB
< 10
//g/dL
268
54
41
3
PbB 2 10
//g/dL
187
56
42
0
Blood-Lead
Concentration
Geometric
Mean
2.35
2.53
3.78
4.18
Geometric
Standard
Deviation
2.64
2.74
2.36
1.56
Log-Odds Ratio
Estimate
0.42
0.96
Standard
Error
0.16
0.19
P-value
0.0087
0.0001
Q2: About what year was your residence built? (Ifyou don't know for sure, please guess in
which decade it was built.)
ENTER COMPLETE YEAR, E.G. 1888, d OR r
Response
Category
Pre 1940
1 940-49
1950-59
1 960-69
1970-79
Post 1 980
•
Number of
PbB <7
//g/dL
1048
162
181
179
398
638
393
7 i PbB
< 10
//g/dL
154
16
25
24
41
41
65
PbB k 10
//g/dL
123
13
15
14
21
29
70
Blood-Lead
Concentration
Geometric
Mean
2.95
2.11
2.26
2.56
2.25
1.70
3.02
Geometric
Standard
Deviation
2.49
2.63
2.79
2.53
2.49
2.81
2.69
Log-Odds Ratio
Estimate
-0.38
-0.35
-0.41
-0.80
-0.95
Standard
Error
0.30
0.29
0.29
0.24
0.21
P-value
0.2105
0.2222
0.1666
0.0010
0.0001
A-5
-------
Q3: Has {01} lived in your current home longer than 12 months? (If no, then all questions
apply to current home as well as any other one lived in during the past year.)
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
•
Number of
PbB <7
A/g/dL
2371
627
1
7 s PbB
< 10
jug/dl_
258
108
0
PbBi 10
/yg/dL
210
74
1
Blood-Lead
Concentration
Geometric
Mean
2.36
2.96
4.79
Geometric
Standard
Deviation
2.66
2.53
2.73
Log-Odds Ratio
Estimate
0.29
Standard
Error
0.14
P-value
0.0439
Q4: In what moth and year did your family move into your current home?
ENTER MONTH AND YEAR E.G. 11/85
Response
Category
Pre 1 980
1980-1989
1990
1991
1992
1993
1994
1995
1996
1997
.
Number of
PbB <7
jug/dL
83
399
166
178
208
305
408
574
627
29
22
7 <; PbB
< 10
//g/dL
11
35
8
16
25
39
51
58
108
9
6
PbB* 10
//g/dL
7
42
14
8
14
27
36
53
74
3
7
Blood-Lead
Concentration
Geometric
Mean
2.92
2.00
1.96
2.05
2.22
2.57
2.69
2.33
2.98
3.28
3.29
Geometric
Standard
Deviation
2.16
3.02
2.86
2.54
2.63
2.57
2.47
2.68
2.54
2.45
2.94
Log-Odds Ratio
Estimate
0.22
-0.00
-0.63
-0.23
0.05
0.05
0.09
0.34
0.20
Standard
Error
0.43
0.48
0.53
0.48
0.44
0.43
0.42
0.41
0.72
P-value
0.6026
1 .0000
0.2388
0.6390
0.9127
0.9164
0.8288
0.4151
0.7776
A-6
-------
Q4b: What kind of building do you live in before your current residence?
1. SINGLE FAMILY HOME
2. DUPLEX
3. APARTMENT/CONDOMINIUM WITH 4 OR FEWER UNITS
4. APARTMENT/CONDOMINIUM WITH 5 OR MORE UNITS
5. MOBILE HOME OR TRAILER
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
3
4
5
8
Number of
PbB <7
//g/dL
283
106
76
106
55
0
2373
7 <. PbB
< 10
//g/dL
42
20
13
18
13
2
258
PbB* 10
//g/dL
28
19
14
8
3
0
213
Blood-Lead
Concentration
Geometric
Mean
2.82
3.24
3.79
2.47
2.77
8.00
2.36
Geometric
Standard
Deviation
2.50
2.73
2.21
2.58
2.52
1.00
2.67
Log-Odds Ratio
Estimate
0.59
0.62
-0.27
-0.60
Standard
Error
0.32
0.35
0.42
0.63
P-value
0.0619
0.0773
0.5159
0.3408
Q4c: In what year was your previous residence build? (Ifyou don't know for sure, please guess
in which decade it was built)
ENTER COMPLETE YEAR, E.G. 1888. D or R
Response
Category
Pre 1 940
1940-49
1950-59
1960-69
1 970-79
Post 1 980
refused
•
Number of
PbB <7
//g/dL
190
28
38
41
82
105
5
2510
7 <; PbB
< 10
//g/dL
35
4
7
5
17
17
0
281
PbB :> 10
//g/dL
24
7
4
5
5
4
2
234
Blood-Lead
Concentration
Geometric
Mean
3.26
2.76
2.50
1.78
2.21
3.12
3.54
2.42
Geometric
Standard
Deviation
2.46
3.21
2.76
3.91
2.76
1.95
3.00
2.65
Log-Odds Ratio
Estimate
0.68
-0.18
-0.04
-0.73
-1.20
Standard
Error
0.47
0.57
0.52
0.51
0.55
P-value
0.1505
0.7485
0.9462
0.1525
0.0304
A-7
-------
Q5: Does (tested child) live in your household most of the time?
LYES
2. NO
3. ALL OF THE TIME (VOL)*
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
3
Number of
PbB <7
jt/g/dL
2565
23
408
3
7 * PbB
< 10
fjgldL
309
0
57
0
PbB z 10
fjgldL
234
3
48
0
Blood-Lead
Concentration
Geometric
Mean
2.50
1.25
2.47
2.76
Geometric
Standard
Deviation
2.60
3.96
2.80
1.05
Log-Odds Ratio
Estimate
0.36
0.25
Standard
Error
0.62
0.17
P-value
0.5627
0.1282
Q6: Would you say that there is no or very little peeling paint, some peeling paint or a lot of
peeling paint inside your home.
1. NONE OR VERY LITTLE PEELING PAINT (surface intact)
2. SOME PEELING PAINT (less than 2 sq ft)
3. A LOT OF PEELING PAINT (involves than 2 sq ft)
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
3
Number of
PbB <7
//g/dL
2537
353
104
5
7 <; PbB
< 10
fjg/dL
288
64
14
0
PbB a 10
fjgldL
219
46
24
0
Blood-Lead
Concentration
Geometric
Mean
2.33
3.05
3.92
2.76
Geometric
Standard
Deviation
2.70
2.56
2.37
1.05
Log-Odds Ratio
Estimate
0.41
0.98
Standard
Error
0.17
0.24
P-value
0.0312
0.0001
* "(Vol)" indicates that this response was volunteered and was not originally listed as a response to the question.
A-8
-------
Q7: Would you say that there is no or very little peeling paint, some peeling paint or a lot of
peeling paint on the outside of your home.
1. NONE OR VERY LITTLE PEELING PAINT (surface intact)
2. SOME PEELING PAINT (less than 2 sq ft)
3. A LOT OF PEELING PAINT (involves more than 2 sq ft)
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
3
.
Number of
PbB <7
j/g/dL
2327
457
200
15
7 ± PbB
< 10
fJQ/dl
268
54
41
3
PbB * 10
fjg/dl
190
57
42
0
Blood-Lead
Concentration
Geometric
Mean
2.33
2.50
3.78
4.18
Geometric
Standard
Deviation
2.69
2.80
2.36
1.56
Log-Odds Ratio
Estimate
0.42
0.94
Standard
Error
0.16
0.19
P-value
0.0079
0.0001
A-9
-------
INSIDE PAINTING
We are trying to learn what people are doing to keep up their homes or to make major changes
to them. First, I -would like to start by asking you about any work that has been done or is being
done inside your home or apartment.
Q10: During the past 12 months, was there any painting or were any surfaces prepared for
paint inside your home?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
Number of
PbB <7
fjg/dl
1181
1810
8
7 s PbB
< 10
fjg/dl
163
203
0
PbB 2 10
jug/dL
121
162
2
Blood-Lead
Concentration
Geometric
Mean
2.65
2.39
0.88
Geometric
Standard
Deviation
2.61
2.64
6.62
Log-Odds Ratio
Estimate
-0.14
Standard
Error
0.13
P-value
0.2828
Qll: Was any work done to repair broken plaster or damaged walls in the room/rooms you
painted in the last 12 months?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
.
Number of
PbB <7
//g/dL
635
2353
10
1
7 s. PbB
< 10
fjg/dl
101
265
0
0
PbB 2 10
fjg/dL
74
209
2
0
Blood-Lead
Concentration
Geometric
Mean
2.93
2.36
2.62
2.76
Geometric
Standard
Deviation
2.49
2.68
2.87
1.05
Log-Odds Ratio
Estimate
-0.27
Standard
Error
0.14
P-value
0.0566
A-10
-------
Q12: Was any work done where old walls were taken down or moved, while working in your
home?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
Number of
PbB <7
//g/dL
279
2717
3
7 <: PbB
< 10
//9/dL
48
317
1
PbB * 10
jjg/dL
33
249
3
Blood-Lead
Concentration
Geometric
Mean
2.92
2.43
6.84
Geometric
Standard
Deviation
2.47
2.65
2.01
Log-Odds Ratio
Estimate
-0.26
Standard
Error
0.20
P-value
0.1922
ASK Q's 13 THROUGH 27 IF INSIDE OF HOME WAS PAINTED.
Q13: Was the surface prepared for the new paint, by methods such as sanding or scraping off
the old paint?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
•
Number of
PbB <7
//g/dL
412
1023
65
1499
7 s PbB
< 10
//g/dL
76
129
4
157
PbB -i. 10
A/g/dL
53
89
7
140
Blood-Lead
Concentration
Geometric
Mean
3.26
2.31
2.32
2.34
Geometric
Standard
Deviation
2.44
2.76
2.90
2.69
Log-Odds Ratio
Estimate
-0.35
Standard
Error
0.16
P-value
0.0309
*Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
A-11
-------
How was the surface prepared for the new paint?
Q14: Was hand scraping or hand sanding used?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
Number of
PbB <7
/yg/dL
374
27
11
2587
7 <. PbB
< 10
fjg/dl
68
7
1
290
PbBk 10
fjg/dl
40
8
5
236
Blood-Lead
Concentration
Geometric
Mean
3.19
4.25
4.59
2.33
Geometric
Standard
Deviation
2.35
2.69
2.83
2.72
Log-Odds Ratio
Estimate
-0.14
Standard
Error
0.18
P-value
0.4479
*Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
Q15: Was power sanding, grinding, or sandblasting used?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
Number of
PbB <7
fjg/dL
128
270
14
2587
7 <; PbB
< 10
fjgldl
22
51
3
290
PbB* 10
fjg/dl
16
35
2
236
Blood-Lead
Concentration
Geometric
Mean
3.20
3.29
3.51
2.33
Geometric
Standard
Deviation
2.56
2.39
2.30
2.72
Log-Odds Ratio
Estimate
-0.28
Standard
Error
0.27
P-value
0.3115
* Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
A-12
-------
Q16: Was an open flame torch used?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
Number of
PbB <7
//g/dL
2
399
11
2587
7 s PbB
< 10
A/g/dL
3
71
2
290
PbB ^ 10
fjg/dl
1
49
3
236
Blood-Lead
Concentration
Geometric
Mean
6.05
3.23
3.62
2.33
Geometric
Standard
Deviation
1.79
2.44
2.64
2.72
Log-Odds Ratio
Estimate
-1.66
Standard
Error
1.23
P-value
0.1769
*Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
Ql 7: Were heat guns used?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
.
Number of
PbB <7
fjg/dL
28
370
14
2587
7 <. PbB
< 10
fjg/dL
7
67
2
290
PbB 2 10
fjg/dl
9
39
5
236
Blood-Lead
Concentration
Geometric
Mean
4.40
3.17
4.86
2.33
Geometric
Standard
Deviation
2.92
2.34
2.41
2.72
Log-Odds Ratio
Estimate
-1.24
Standard
Error
0.39
P-value
0.0014
*Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
A-13
-------
Q18: Was washing, wet scraping, wet sanding, or water blasting used?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
Number of
PbB <7
//g/dL
119
269
24
2587
7 s PbB
< 10
yug/dL
24
49
3
290
PbB* 10
/yg/dL
17
34
2
236
Blood-Lead
Concentration
Geometric
Mean
3.55
3.19
2.68
2.33
Geometric
Standard
Deviation
2.29
2.51
2.34
2.72
Log-Odds Ratio
Estimate
-0.41
Standard
Error
0.27
P-value
0.1220
* Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
Q19: Were chemical paint removers used?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
.
Number of
PbB <7
pg/dL
70
331
11
2587
7 s. PbB
< 10
//g/dL
14
60
2
290
PbB* 10
//g/dL
12
37
4
236
Blood-Lead
Concentration
Geometric
Mean
3.83
3.13
4.54
2.33
Geometric
Standard
Deviation
2.48
2.39
2.65
2.72
Log-Odds Ratio
Estimate
-0.61
Standard
Error
0.32
P-value
0.0575
* Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
A-14
-------
Q20: Who did the work?
1. HEAD OR SPOUSE
2. OTHER PERSON IN HH
3. RELATIVE/FRIEND NOT IN HH
4. OWNER, BUILDING SUPERINTENDENT, APT STAFF
5. PAID PROFESSIONAL CONTRACTOR
ENTER ALL THAT APPLY SEPARATED BY SLASHES, d OR r, a FOR ALL
(FOR EACH ITEM, THE ANSWER WILL BE:
1. YES
2. NO
8. DON'T KNOW
9. REFUSED)
Response
Category
1
2
8
Number of
PbB <7
Jjg/dL
324
87
1
2587
7 <. PbB
< 10
fjg/dL
61
15
0
290
PbB 2 10
A/g/dL
36
17
0
236
Blood-Lead
Concentration
Geometric
Mean
3.20
3.56
2.76
2.33
Geometric
Standard
Deviation
2.41
2.49
1.05
2.72
Log-Odds Ratio
Estimate
-0.16
Standard
Error
0.19
P-value
0.3916
*Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
2.
Response
Category
1
2
8
Number of
PbB <7
fjg/dl
63
348
1
2587
7 s. PbB
< 10
fjg/dl
8
68
0
290
PbB 2 10
Jjg/dl
5
48
0
236
Blood-Lead
Concentration
Geometric
Mean
2.96
3.33
2.76
2.33
Geometric
Standard
Deviation
2.33
2.45
1.05
2.72
Log-Odds Ratio
Estimate
0.20
Standard
Error
0.47
P-value
0.6724
'Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
A-15
-------
3.
Response
Category
1
2
8
Number of
PbB <7
/jg/dl
48
363
1
2587
7 s PbB
< 10
fjg/di.
11
65
0
290
PbB 2 10
fjgldl
8
45
0
236
Blood-Lead
Concentration
Geometric
Mean
3.64
3.22
2.76
2.33
Geometric
Standard
Deviation
2.26
2.46
1.05
2.72
Log-Odds Ratio
Estimate
-0.56
Standard
Error
0.39
P-value
0.1482
*Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
4.
Response
Category
1
2
8
Number of
PbB <7
ua/di
39
372
1
2587
7 s PbB
< 10
fjg/dl
8
68
0
290
PbB 2 10
fjg/dl
7
46
0
236
Blood-Lead
Concentration
Geometric
Mean
3.48
3.25
2.76
2.33
Geometric
Standard
Deviation
2.48
2.43
1.05
2.72
Log-Odds Ratio
Estimate
-0.63
Standard
Error
0.42
P-value
0.1273
* Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
5.
Response
Category
1
2
8
•
Number of
PbB <7
fjg/dl
42
369
1
2587
7 * PbB
< 10
fjg/dL
4
72
0
290
PbBi 10
fjg/dL
5
48
0
236
Blood-Lead
Concentration
Geometric
Mean
2.21
3.37
2.76
2.33
Geometric
Standard
Deviation
3.10
2.39
1.05
2.72
Log-Odds Ratio
Estimate
-0.21
Standard
Error
0.48
P-value
0.6532
'Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
A-16
-------
Q21: When was this job started and when was it completed? (If more than one job was done,
then record the start of the first job and the end of the last one.)
ENTER MONTH AND YEAR FOR EACH ON SEPARATE LINES d OR r
ENTER n FOR COMPLETION DATE IF WORK IS STILL IN PROGRESS
Q22: Did anyone in your household live in the home while the work was being done?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
.
Number of
PbB <7
//g/dL
351
61
2587
7 s PbB
< 10
//g/dL
64
12
290
PbBi 10
//g/dL
43
10
236
Blood-Lead
Concentration
Geometric
Mean
3.30
3.07
2.33
Geometric
Standard
Deviation
2.38
2.77
2.72
Log-Odds Ratio
Estimate
-0.28
Standard
Error
0.17
P-value
0.1135
*Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
Q23: In how many rooms in your home was this work done?
ENTER NUMBER OF ROOMS dORr OR 0 FOR NONE
Response
Category
1
2
3
= >4
None
.
Number of
PbB <7
//g/dL
161
94
55
98
0
2591
7 * PbB
< 10
//g/dL
31
13
9
18
1
294
PbB* 10
//g/dL
13
12
3
25
0
236
Blood-Lead
Concentration
Geometric
Mean
3.20
2.80
3.15
3.87
7.00
2.34
Geometric
Standard
Deviation
2.19
2.72
2.32
2.56
1.00
2.72
Log-Odds Ratio
Estimate
0.46
-0.39
1.15
0.12
Standard
Error
0.42
0.66
0.37
0.30
P-value
0.2764
0.5518
0.0016
0.6842
A-17
-------
Q24: Was any of this work done in the kitchen?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
Number of
PbB <7
fjg/dl
155
256
1
2587
7 z PbB
< 10
vg/dL
39
37
0
290
PbB ^ 10
fjg/dl
28
25
0
236
Blood-Lead
Concentration
Geometric
Mean
3.33
3.24
5.00
2.33
Geometric
Standard
Deviation
2.68
2.26
1.00
2.72
Log-Odds Ratio
Estimate
-0.68
Standard
Error
0.22
P-value
0.0017
*Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
Q25: Was any of this work done in the bathroom?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
Number of
PbB <7
^g/dL
182
229
1
2587
7 <. PbB
< 10
//g/dL
31
45
0
290
PbB t 10
ug/dL
26
27
0
236
Blood-Lead
Concentration
Geometric
Mean
3.42
3.12
5.00
2.33
Geometric
Standard
Deviation
2.40
2.48
1.00
2.72
Log-Odds Ratio
Estimate
-0.43
Standard
Error
0.22
P-value
0.0527
* Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
A-18
-------
OUTSIDE PAINTING
Q28: During the past 12 months, was there any painting or were any surfaces prepared for
paint on the outside of your home?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
Number of
PbB <7
//g/dL
340
2640
19
7 <; PbB
< 10
fjg/dl
49
313
4
PbB 2 10
/yg/dL
49
232
4
Blood-Lead
Concentration
Geometric
Mean
2.74
2.45
3.55
Geometric
Standard
Deviation
2.72
2.62
2.57
Log-Odds Ratio
Estimate
-0.49
Standard
Error
0.17
P-value
0.0031
Q29: Was the surface for this outside job prepared for the new paint, by methods such as
sanding or scraping off the old paint?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
•
Number of
PbB <7
fjg/dL
358
186
24
2431
7 <. PbB
< 10
/yg/dL
45
21
6
294
PbB 2 10
fjg/dl
42
18
3
226
Blood-Lead
Concentration
Geometric
Mean
2.58
2.04
3.86
2.47
Geometric
Standard
Deviation
2.76
2.98
2.40
2.65
Log-Odds Ratio
Estimate
-0.23
Standard
Error
0.18
P-value
0.1925
•Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
A-19
-------
How was the surface prepared for the new paint?
Q30: Was hand scraping or hand sanding used?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
Number of
PbB <7
//g/dL
331
19
8
2641
7 <; PbB
< 10
A/g/dL
43
1
1
321
PbBk 10
//g/dL
39
2
1
247
Blood-Lead
Concentration
Geometric
Mean
2.56
3.10
2.27
2.46
Geometric
Standard
Deviation
2.79
2.15
3.14
2.66
Log-Odds Ratio
Estimate
-0.23
Standard
Error
0.18
P-value
0.2057
*Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
Q31: Was power sanding, grinding, or sandblasting used?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
.
Number of
PbB <7
//g/dL
47
294
17
2641
7 <; PbB
< 10
//g/dL
8
35
2
321
PbB a 10
//g/dL
9
30
3
247
Blood-Lead
Concentration
Geometric
Mean
3.30
2.40
3.32
2.46
Geometric
Standard
Deviation
2.38
2.87
2.46
2.66
Log-Odds Ratio
Estimate
-0.71
Standard
Error
0.37
P-value
0.0553
* Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
A-20
-------
Q32: Was an open flame torch used?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
.
Number of
PbB <7
pg/dL
3
348
7
2641
7 s PbB
< 10
pg/dL
0
42
3
321
PbBi 10
fjg/dl
2
37
3
247
Blood-Lead
Concentration
Geometric
Mean
5.25
2.47
5.33
2.46
Geometric
Standard
Deviation
2.17
2.80
1.98
2.66
Log-Odds Ratio
Estimate
-1.95
Standard
Error
0.91
P-value
0.0332
*Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
Q33: Were heat guns used?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
•
Number of
PbB <7
yug/dL
14
331
13
2641
7 <; PbB
< 10
A/g/dL
1
41
3
321
PbBi 10
fJQ/dL
9
30
3
247
Blood-Lead
Concentration
Geometric
Mean
4.77
2.47
4.03
2.46
Geometric
Standard
Deviation
3.22
2.67
2.20
2.66
Log-Odds Ratio
Estimate
-1.93
Standard
Error
0.43
P-value
0.0001
'Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
A-21
-------
Q34: Was washing, wet scraping, wet sanding, or water blasting used?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
•
Number of
PbB <7
fjg/dL
84
257
17
2641
7 <; PbB
< 10
fjg/dl
13
29
3
321
PbB 2. 10
fjg/dl
14
24
4
247
Blood-Lead
Concentration
Geometric
Mean
2.83
2.40
3.79
2.46
Geometric
Standard
Deviation
2.67
2.84
2.33
2.66
Log-Odds Ratio
Estimate
-0.58
Standard
Error
0.30
P-value
0.0506
*Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
Q35: Were chemical paint removers used?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
Number of
PbB <7
fjg/dl
17
317
24
2641
7 z PbB
< 10
fjg/dl
1
41
3
321
PbB* 10
//g/dL
6
31
5
247
Blood-Lead
Concentration
Geometric
Mean
3.90
2.46
3.25
2.46
Geometric
Standard
Deviation
2.28
2.78
2.78
2.66
Log-Odds Ratio
Estimate
-1.32
Standard
Error
0.48
P-value
0.0057
* Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
A-22
-------
Q36: Who did the work?
1. HEAD OR SPOUSE
2. OTHER PERSON IN HH
3. RELATIVE/FRIEND NOT IN HH
4. OWNER, BUILDING SUPERINTENDENT, APT STAFF
5. PAID PROFESSIONAL CONTRACTOR
ENTER ALL THAT APPLY SEPARATED BY SLASHES, d OR r, a FOR ALL
(FOR EACH ITEM, THE ANSWER WILL BE:
1. YES
2. NO
8. DON'T KNOW
9. REFUSED)
1.
Response
Category
1
2
8
.
Number of
PbB <7
fjg/dL
246
108
4
2641
7 <; PbB
< 10
fjg/dl
30
14
1
321
PbB i 10
fjg/dl
29
13
0
247
Blood-Lead
Concentration
Geometric
Mean
2.56
2.67
2.07
2.46
Geometric
Standard
Deviation
2.75
2.79
2.28
2.66
Log-Odds Ratio
Estimate
-0.22
Standard
Error
0.21
P-value
0.2868
*Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
2.
Response
Category
1
2
8
•
Number of
PbB <7
/ig/dL
32
322
4
2641
7 s PbB
< 10
fjg/dl
2
42
1
321
PbB 2 10
/jg/dl
2
40
0
247
Blood-Lead
Concentration
Geometric
Mean
1.30
2.70
2.07
2.46
Geometric
Standard
Deviation
3.69
2.71
2.28
2.66
Log-Odds Ratio
Estimate
0.44
Standard
Error
0.73
P-value
0.5492
* Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
A-23
-------
3.
Response
Category
1
2
8
•
Number of
PbB <7
//g/dL
19
335
4
2641
7 t PbB
< 10
jug/dL
5
39
1
321
PbBk 10
A/g/dL
5
37
0
247
Blood-Lead
Concentration
Geometric
Mean
4.43
2.45
2.07
2.46
Geometric
Standard
Deviation
1.99
2.85
2.28
2.66
Log-Odds Ratio
Estimate
-1.01
Standard
Error
0.51
P-value
0.0452
* Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
4.
Response
Category
1
2
8
Number of
PbB <7
jjgldl
59
295
4
2641
7 <; PbB
< 10
fjg/dl
6
38
1
321
PbB ;> 10
fJQ/dl
8
34
0
247
Blood-Lead
Concentration
Geometric
Mean
2.62
2.59
2.07
2.46
Geometric
Standard
Deviation
2.87
2.74
2.28
2.66
Log-Odds Ratio
Estimate
-0.35
Standard
Error
0.38
P-value
0.3617
•Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
5.
Response
Category
1
2
8
Number of
PbB <7
^g/dL
41
313
4
2641
7 * PbB
< 10
jjg/dl
5
39
1
321
PbB a 10
fjg/dl
7
35
0
247
Blood-Lead
Concentration
Geometric
Mean
2.64
2.59
2.07
2.46
Geometric
Standard
Deviation
3.01
2.72
2.28
2.66
Log-Odds Ratio
Estimate
-0.58
Standard
Error
0.41
P-value
0.1599
*Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
A-24
-------
Q37: When was this job started and when was it completed (If more than one job was done,
then record the beginning date of the first job and the ending date of the last one)?
ENTER MONTH AND YEAR FOR EACH ON SEPARATE LINES d OR r
ENTER n FOR COMPLETION DATE IF WORK IS STILL IN PROGRESS
Q38: Did anyone in your household live in the home while the work was being done?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
Number of
PbB <
7 ^g/dL
339
19
2641
7 <; PbB
< 10
Jjg/dl
43
2
321
PbB;> 10
pg/dL
41
1
247
Blood-Lead
Concentration
Geometric
Mean
2.54
3.32
2.46
Geometric
Standard
Deviation
2.81
1.90
2.66
Log-Odds Ratio *
Estimate
-0.26
Standard
Error
0.18
P-value
0.1442
*Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
A-25
-------
OTHER REPAIRS
Next I would like to ask you about repairs and major changes to the rest of your home during the
past 12 months.
Q39: Were any windows repaired or new windows put in?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
Number of
PbB <
7 pg/dL
528
2466
5
7 s PbB
< 10
^g/dL
86
278
2
PbB a 10
;/g/dL
65
217
3
Blood-Lead
Concentration
Geometric
Mean
3.10
2.34
4.64
Geometric
Standard
Deviation
2.43
2.68
2.78
Log-Odds Ratio
Estimate
-0.34
Standard
Error
0.15
P-value
0.0245
Q40: Was any work done to repair broken plaster or damaged walls while repairing or putting
in the windows?
l. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
Number of
PbB <
7 /yg/dL
128
497
9
2365
7 s PbB
< 10
pg/dL
13
83
2
268
PbBi 10
pg/dL
12
57
3
217
Blood-Lead
Concentration
Geometric
Mean
2.55
3.04
3.61
2.33
Geometric
Standard
Deviation
2.46
2.52
2.69
2.72
Log-Odds Ratio*
Estimate
0.02
Standard
Error
0.31
P-value
0.9458
'Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
A-26
-------
Q41: Was any work done where old walls were taken down or moved, while repairing or
putting in the windows ?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
Number of
PbB <
7//g/dL
90
540
4
2365
7 <; PbB
< 10
jug/dL
10
87
1
268
PbBk 10
fjg/dl
9
64
0
216
Blood-Lead
Concentration
Geometric
Mean
2.64
2.99
3.64
2.33
Geometric
Standard
Deviation
2.73
2.51
1.84
2.72
Log-Odds Ratio*
Estimate
-0.04
Standard
Error
0.36
P-value
0.9174
*Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
Q42: Was there any painting or were any surfaces prepared for paint with the installation of
the new windows?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
Number of
PbB <
7 jjg/dL
182
445
7
2365
7 * PbB
< 10
Jjg/dL
37
61
0
268
PbB * 10
/jg/dl
16
55
2
216
Blood-Lead
Concentration
Geometric
Mean
3.00
2.94
3.30
2.33
Geometric
Standard
Deviation
2.36
2.57
3.69
2.72
Log-Odds Ratio*
Estimate
0.09
Standard
Error
0.27
P-value
0.7301
*Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
A-27
-------
Q43: Did the -window work involve preparing surfaces for paint, by methods such as sanding,
scraping, using heat guns or the use of chemicals other than prime painting?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
Number of
PbB <
7 /yg/dL
96
502
36
2365
7
-------
Q45: Was power sanding, grinding, or sandblasting used?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
Number of
PbB <
7 /yg/dL
23
71
2
2903
7 * PbB
< 10
//g/dL
8
11
0
347
PbB 2 10
//g/dL
2
7
1
279
Blood-Lead
Concentration
Geometric
Mean
3.19
2.92
2.22
2.45
Geometric
Standard
Deviation
2.26
2.39
5.56
2.69
Log-Odds Ratio*
Estimate
0.10
Standard
Error
0.74
P-value
0.8918
*Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
Q46: On which surfaces was power sanding, grinding, or sandblasting used?
1. EXTERIOR WALLS
2. TRIM/EAVES
3. PORCHES
4. DOORS / WINDOWS
5. ROOF
(delete)
ENTER ALL THAT APPLY SEPARATED BY SLASHES, d OR r, a FOR ALL
A-29
-------
Q47: Was an open flame torch used?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
Number of
PbB <7
^g/dL
0
94
2
2903
7 s PbB
< 10
pg/dL
1
18
0
347
PbB* 10
//g/dL
1
8
1
279
Blood-Lead
Concentration
Geometric
Mean
9.80
2.91
2.22
2.45
Geometric
Standard
Deviation
1.22
2.35
5.56
2.69
Log-Odds Ratio
Estimate
Standard
Error
P-value
*Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
Q48: Were heat guns used?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
.
Number of
PbB <7
//g/dL
7
87
2
2903
7 <: PbB
< 10
A/g/dL
3
16
0
347
PbBi 10
//g/dL
1
8
1
279
Blood-Lead
Concentration
Geometric
Mean
4.55
2.82
2.22
2.45
Geometric
Standard
Deviation
1.89
2.42
5.56
2.69
Log-Odds Ratio
Estimate
-0.40
Standard
Error
1.07
P-value
0.7104
'Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
A-30
-------
Q49: Was washing, wet scraping, wet sanding, or water blasting used?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
Number of
PbB <7
A/g/dL
21
73
2
2903
7 s PbB
< 10
yug/dL
9
10
0
347
PbB * 10
jug/dL
3
6
1
279
Blood-Lead
Concentration
Geometric
Mean
3.56
2.86
2.22
2.45
Geometric
Standard
Deviation
2.48
2.24
5.56
2.69
Log-Odds Ratio
Estimate
-0.40
Standard
Error
0.62
P-value
0.5191
* Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
Q50: Were chemical paint removers used?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
Number of
PbB <7
//g/dL
10
83
3
2903
7 i PbB
< 10
pg/dL
2
17
0
347
PbB 2 10
//g/dL
2
7
1
279
Blood-Lead
Concentration
Geometric
Mean
3.39
3.04
1.20
2.45
Geometric
Standard
Deviation
3.02
2.21
7.10
2.69
Log-Odds Ratio
Estimate
-0.74
Standard
Error
0.78
P-value
0.3434
*Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
A-31
-------
Q51: Who did the work?
1. HEAD OR SPOUSE
2. OTHER PERSON IN HH
3. RELATIVE/FRIEND NOT IN HH
4. OWNER, BUILDING SUPERINTENDENT, APT STAFF
5. PAID PROFESSIONAL CONTRACTOR
ENTER ALL THAT APPLY SEPARATED BY SLASHES, d OR r, a FOR ALL
(FOR EACH ITEM, THE ANSWER WILL BE:
1. YES
2. NO
8. DON'T KNOW
9. REFUSED)
1.
Response
Category
1
2
8
.
Number of
PbB <7
fjg/dl
68
28
0
2903
7 <. PbB
< 10
fjg/dl
17
1
1
347
PbB 2 10
fjg/dl
8
2
0
279
Blood-Lead
Concentration
Geometric
Mean
3.35
1.49
7.00
2.45
Geometric
Standard
Deviation
2.30
3.20
1.00
2.69
Log-Odds Ratio
Estimate
-0.20
Standard
Error
0.38
P-value
0.5891
*Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
2.
Response
Category
1
2
8
Number of
PbB <7
/jg/dl
6
90
0
2903
7 n PbB
< 10
jjg/dl
0
18
1
347
PbB* 10
fjg/dL
0
10
0
279
Blood-Lead
Concentration
Geometric
Mean
3.48
2.90
7.00
2.45
Geometric
Standard
Deviation
1.47
2.49
1.00
2.69
Log-Odds Ratio
Estimate
Standard
Error
P-value
'Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
A-32
-------
3.
Response
Category
1
2
8
•
Number of
PbB <7
j/g/dL
11
85
0
2903
7 z PbB
< 10
fjg/dl
0
18
1
347
PbB* 10
fjg/dL
1
9
0
279
Blood-Lead
Concentration
Geometric
Mean
0.38
3.15
7.00
2.45
Geometric
Standard
Deviation
5.67
2.36
1.00
2.69
Log-Odds Ratio
Estimate
0.06
Standard
Error
1.05
P-value
0.9554
*Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
4.
Response
Category
1
2
8
•
Number of
PbB <7
/yg/dL
5
91
0
2903
7 $ PbB
< 10
fjg/dL
0
18
1
347
PbB 2 10
fjg/dl
0
10
0
279
Blood-Lead
Concentration
Geometric
Mean
2.48
2.95
7.00
2.45
Geometric
Standard
Deviation
1.82
2.45
1.00
2.69
Log-Odds Ratio
Estimate
20.03
Standard
Error
0.00
P-value
*Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
5.
Response
Category
1
2
8
•
Number of
PbB <7
fjg/dl
22
74
0
2903
7 * PbB
< 10
jjg/dl
1
17
1
347
PbB 2 10
^g/dL
1
9
0
279
Blood-Lead
Concentration
Geometric
Mean
1.70
3.20
7.00
2.45
Geometric
Standard
Deviation
2.90
2.36
1.00
2.69
Log-Odds Ratio
Estimate
0.76
Standard
Error
1.02
P-value
0.4609
*Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
A-33
-------
Q52: When was this job started and when was it completed? (If more than one job was done,
then enter the start date of the first job and the ending date of the last one.)
ENTER MONTH AND YEAR FOR EACH ON SEPARATE LINES d OR r
ENTER n FOR COMPLETION DATE IF WORK IS STILL IN PROGRESS
Q53: Did anyone in your household live in your home while the work was being done?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
.
Number of
PbB <7
fjg/dL
86
10
2903
7 s PbB
< 10
A/9/dL
19
0
347
PbB* 10
//g/dL
10
0
279
Blood-Lead
Concentration
Geometric
Mean
3.22
2.76
2.45
Geometric
Standard
Deviation
2.34
1.05
2.69
Log-Odds Ratio
Estimate
-0.19
Standard
Error
0.34
P-value
0.5683
* Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
A-34
-------
CARPETS AND FLOORS
Q54: Were any carpets replaced or removed or painted floors refinished in the last 12 months?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
Number of
PbB <7
A/g/dL
517
2472
10
7 s PbB
< 10
//g/dL
91
272
3
PbBi 10
fjg/dL
65
220
0
Blood-Lead
Concentration
Geometric
Mean
2.94
2.38
3.49
Geometric
Standard
Deviation
2.54
2.66
1.79
Log-Odds Ratio
Estimate
-0.35
Standard
Error
0.15
P-value
0.0206
Q55: Do you have any wall-to-wall carpeting in your home?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
Number of
PbB <7
j/g/dL
499
144
1
2355
7 * PbB
< 10
fjq/dl
79
23
0
264
PbB* 10
jug/dL
61
9
0
215
Blood-Lead
Concentration
Geometric
Mean
2.74
2.99
2.76
2.41
Geometric
Standard
Deviation
2.67
2.14
1.05
2.66
Log-Odds Ratio
Estimate
-0.67
Standard
Error
0.37
P-value
0.0694
A-35
-------
Q56: Did you replace or remove a wall-to-wall or other large carpet in the last 12 months?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
.
Number of
PbB <7
/yg/dL
432
209
3
2355
7 <. PbB
< 10
pg/dL
61
40
1
264
PbB* 10
//g/dL
50
20
0
219
Blood-Lead
Concentration
Geometric
Mean
2.59
3.14
2.51
2.38
Geometric
Standard
Deviation
2.74
2.26
2.13
2.71
Log-Odds Ratio
Estimate
-0.22
Standard
Error
0.16
P-value
0.1868
*Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
A-36
-------
READ EF FLOOR SURFACE OR COVERING DISTURBED
Q57: Was the surface for this carpet or flooring job prepared by methods such as sanding or
scraping off old paint or adhesive?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
.
Number of
PbB <7
/yg/dL
165
417
62
2355
7 <; PbB
< 10
fjg/dl
21
76
5
264
PbB 2 10
fjg/dl
20
44
6
219
Blood-Lead
Concentration
Geometric
Mean
2.15
3.03
2.49
2.38
Geometric
Standard
Deviation
3.14
2.40
2.54
2.71
Log-Odds Ratio
Estimate
-0.24
Standard
Error
0.25
P-value
0.3182
*Log-odds ratios were calculated-assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
How was the surface prepared for the carpet or flooring?
Q58: Was hand scraping or hand sanding used?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
.
Number of
PbB <7
/yg/dL
132
26
7
2834
7 <. PbB
< 10
fjg/dl
17
4
0
345
PbB 2 10
Jjg/dl
14
5
1
269
Blood-Lead
Concentration
Geometric
Mean
2.21
2.64
0.15
2.49
Geometric
Standard
Deviation
2.86
3.59
18.4
2.65
Log-Odds Ratio
Estimate
-0.10
Standard
Error
0.29
P-value
0.7241
* Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
dc this R&R activity were not asked this question.
A-37
-------
Q59: Was power sanding, grinding, or sandblasting used?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
.
Number of
PbB <7
/;g/dL
51
106
8
2834
7 <; PbB
< 10
/yg/dL
7
14
0
345
PbB ^ 10
fjgldl
6
13
1
269
Blood-Lead
Concentration
Geometric
Mean
2.86
1.95
0.11
2.49
Geometric
Standard
Deviation
2.82
3.17
20.2
2.65
Log-Odds Ratio
Estimate
-0.20
Standard
Error
0.44
P-value
0.6396
* Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
Q60: Was an open flame torch used?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
.
Number of
PbB <7
A/9/dL
1
157
7
2834
7 <. PbB
< 10
A/g/dL
0
21
0
345
PbBi 10
fjg/dl
0
18
2
269
Blood-Lead
Concentration
Geometric
Mean
2.76
2.22
1.00
2.49
Geometric
Standard
Deviation
1.05
3.04
6.84
2.65
Log-Odds Ratio
Estimate
Standard
Error
P-value
*Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
A-38
-------
Q61: Were heat guns used?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
.
Number of
PbB <7
//g/dL
3
153
9
2834
7 <; PbB
< 10
//g/dL
0
21
0
345
PbBi 10
//g/dL
2
16
2
269
Blood-Lead
Concentration
Geometric
Mean
1.53
2.34
0.51
2.49
Geometric
Standard
Deviation
42.0
2.77
10.6
2.65
Log-Odds Ratio
Estimate
-1.94
Standard
Error
0.91
P-value
0.0336
*Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
Q62: Was washing, wet scraping, wet sanding, or water blasting used?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
.
Number of
PbB <7
//g/dL
30
127
8
2834
7 * PbB
< 10
//g/dL
6
15
0
345
PbB 2 10
//g/dL
2
16
2
269
Blood-Lead
Concentration
Geometric
Mean
2.10
2.27
0.64
2.49
Geometric
Standard
Deviation
2.92
3.02
9.86
2.65
Log-Odds Ratio
Estimate
0.37
Standard
Error
0.73
P-value
0.6163
*Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
A-39
-------
Q63: Were chemical paint removers used?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
Number of
PbB <7
jug/dL
22
133
10
2834
7 $ PbB
< 10
X/g/dL
4
17
0
345
PbBi 10
jt/g/dL
2
17
1
269
Blood-Lead
Concentration
Geometric
Mean
1.98
2.30
0.84
2.49
Geometric
Standard
Deviation
4.43
2.86
5.62
2.65
Log-Odds Ratio
Estimate
0.06
Standard
Error
0.74
P-value
0.9370
*Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
Q64:
Who did the work?
1. HEAD OR SPOUSE
2. OTHER PERSON IN HH
3. RELATIVE/FRIEND NOT IN HH
4. OWNER, BUILDING SUPERINTENDENT, APT STAFF
5. PAID PROFESSIONAL CONTRACTOR
ENTER ALL THAT APPLY SEPARATED BY SLASHES, d OR r, a FOR ALL
(FOR EACH ITEM, THE ANWER WILL BE:
1. YES
2. NO
8. DON'T KNOW
9. REFUSED)
1.
Response
Category
1
2
8
Number of
PbB <7
fJQ/dl
110
53
2
2834
7 <: PbB
< 10
//g/dL
18
3
0
345
PbB* 10
fjg/dL
11
9
0
269
Blood-Lead
Concentration
Geometric
Mean
2.31
1.87
2.76
2.49
Geometric
Standard
Deviation
3.00
3.47
1.05
2.65
Log-Odds Ratio
Estimate
-0.04
Standard
Error
0.32
P-value
0.9067
'Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
A-40
-------
2.
Response
Category
1
2
8
.
Number of
PbB <7
fjg/dl
27
136
2
2834
7 * PbB
< 10
//g/dL
2
19
0
345
PbBi 10
//g/dL
1
19
0
269
Blood-Lead
Concentration
Geometric
Mean
2.22
2.22
2.76
2.49
Geometric
Standard
Deviation
2.29
3.22
1.05
2.65
Log-Odds Ratio
Estimate
0.96
Standard
Error
1.02
P-value
0.3455
*Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
3.
Response
Category
1
2
8
.
Number of
PbB <7
//g/dL
18
145
2
2834
7 * PbB
< 10
//g/dL
2
19
0
345
PbBk 10
//g/dL
3
17
0
269
Blood-Lead
Concentration
Geometric
Mean
1.83
2.21
2.76
2.49
Geometric
Standard
Deviation
3.65
3.08
1.05
2.65
Log-Odds Ratio
Estimate
-0.55
Standard
Error
0.63
P-value
0.3788
"Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
4.
Response
Category
1
2
8
•
Number of
PbB <7
//g/dL
9
154
2
2834
7 * PbB
< 10
//g/dL
2
19
0
345
PbB 2 10
//g/dL
3
17
0
269
Blood-Lead
Concentration
Geometric
Mean
4.17
2.04
2.76
2.49
Geometric
Standard
Deviation
2.58
3.17
1.05
2.65
Log-Odds Ratio
Estimate
-1.25
Standard
Error
0.67
P-value
0.0624
*Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
A-41
-------
5.
Response
Category
1
2
8
•
Number of
PbB <7
/ig/dL
42
121
2
2834
7 s PbB
< 10
jug/dL
1
20
0
345
PbB* 10
/vg/dL
2
18
0
269
Blood-Lead
Concentration
Geometric
Mean
1.45
2.49
2.76
2.49
Geometric
Standard
Deviation
2.77
3.08
1.05
2.65
Log-Odds Ratio
Estimate
0.71
Standard
Error
0.73
P-value
0.3265
* Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
Q65: When was this job started and when was it completed? (If more than one job was done,
then enter the start date of the first job and the ending date of the last one.)
ENTER MONTH AND YEAR FOR EACH ON SEPARATE LINES d OR r
ENTER n FOR COMPLETION DATE IF WORK IS STILL IN PROGRESS
Q66: Did anyone in your household live in the home while the work was being done?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
.
Number of
PbB <7
//g/dL
129
36
2834
7 * PbB
< 10
//g/dL
14
7
345
PbB* 10
pg/dL
12
8
269
Blood-Lead
Concentration
Geometric
Mean
1.86
3.14
2.49
Geometric
Standard
Deviation
3.22
2.85
2.65
Log-Odds Ratio
Estimate
0.04
Standard
Error
0.31
P-value
0.9048
'Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
A-42
-------
Q71: As a result of the work that was done in your home were dust and dirt spread to 1. Only
the work area 2. The rooms next to the work area 3. Throughout the house or 4. No dirt
or dust was generated?
1. ONLY IN THE WORK AREA
2. IN THE ROOMS NEXT TO THE WORK
3. THROUGHOUT THE HOUSE
4. NO DIRT OR DUST GENERATED
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
3
4
8
•
Number of
PbB <7
pg/dL
725
211
266
706
70
1021
7 s PbB
< 10
;/g/dL
101
29
43
91
7
95
PbB 2 10
fjg/dL
77
24
26
60
12
90
Blood-Lead
Concentration
Geometric
Mean
2.63
2.75
2.89
2.35
3.07
2.23
Geometric
Standard
Deviation
2.67
2.73
2.39
2.74
2.61
2.70
Log-Odds Ratio
Estimate
0.07
-0.08
-0.20
Standard
Error
0.25
0.24
0.15
P-value
0.7809
0.7270
0.1712
*Log-odds ratios were calculated assuming that missing values were 'No' response because people who didn't
do this R&R activity were not asked this question.
A-43
-------
TESTING FOR BLOOD LEAD
Q200: Now I'd like to ask you about your child's blood test for lead. Was your child tested
because you asked for it, or because your nurse or doctor told you that your child
needed the test?
1. I ASKED FOR IT
2. DR/NURSE/CLINIC RECOMMENDED IT
3. WIC RECOMMENDED
4. SCHOOL REQUIRED
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
3
4
8
Number of
PbB <7
/yg/dL
976
1241
385
163
232
2
7 i PbB
< 10
j/g/dL
132
136
44
22
32
0
PbB i 10
yug/dL
83
126
53
12
11
0
Blood-Lead
Concentration
Geometric
Mean
2.72
2.31
2.51
2.36
2.63
2.76
Geometric
Standard
Deviation
2.48
2.79
2.83
2.58
2.22
1.05
Log-Odds Ratio
Estimate
0.18
0.48
-0.14
Standard
Error
0.15
0.19
0.32
P-value
0.2302
0.0095
0.6524
Q200a: If someone else recommended it, was he or she from either the WIC program or from a
health department?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
•
Number of
PbB <7
pg/dL
1438
405
178
978
7 <. PbB
< 10
//g/dL
160
50
24
132
PbB i 10
/yg/dL
145
44
13
83
Blood-Lead
Concentration
Geometric
Mean
2.45
2.15
2.23
2.71
Geometric
Standard
Deviation
2.67
2.93
2.69
2.48
Log-Odds Ratio
Estimate
0.07
Standard
Error
0.18
P-value
0.6805
A-44
-------
Q200b: Has the children has a subsequent blood test?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
Number of
PbB <7
//g/dL
150
2761
88
7 <. PbB
< 10
//g/dL
26
327
13
PbB* 10
//g/dL
139
138
8
Blood-Lead
Concentration
Geometric
Mean
5.81
2.43
2.59
Geometric
Standard
Deviation
2.70
2.37
2.66
Log-Odds Ratio
Estimate
-2.92
Standard
Error
0.15
P-value
0.0001
Q201: How many blood lead tests has {01} had in the past 12 months?
1. ONE
2. TWO
3. THREE
4. FOUR OR MORE
8. DON'T KNOW
9. REFUSED
(delete this question)
Q201a: In -which month and year was the first blood test?
ENTER MONTH/YEAR, E.G. 12/96 d OR r
A-45
-------
Q201b: For this test did they take the blood from the finger, the arm or someplace else?
1. FINGER (capillary)
2. ARM (venous)
3. SOMEPLACE ELSE(specify:
4. FOOT
5. TOE
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
3
4
5
8
.
Number of
PbB <7
//g/dL
2377
179
6
42
82
313
0
7 s PbB
< 10
//g/dL
296
19
0
5
3
42
1
PbB 2 10
/jg/dL
179
85
0
1
6
14
0
Blood-Lead
Concentration
Geometric
Mean
2.52
4.02
3.65
2.36
2.09
2.34
7.00
Geometric
Standard
Deviation
2.47
3.26
1.30
2.11
2.47
2.36
1.00
Log-Odds Ratio
Estimate
1.84
-19.8
-1.15
-0.03
Standard
Error
0.15
29E3
1.01
0.43
P-value
0.0001
0.9995
0.2565
0.9467
Q201c: What was the lead level?
ENTER NUMBER FROM 0 TO 100 d OR r
Response
Category
0-4
5-9
10-19
unknown
refused
•
Number of
PbB <7
fjg/dL
245
93
3
2561
9
88
7 * PbB
< 10
jjg/dl
24
9
0
286
4
43
PbBi 10
//g/dL
21
14
1
141
1
107
Blood-Lead
Concentration
Geometric
Mean
1.98
2.53
1.20
2.40
3.61
7.41
Geometric
Standard
Deviation
2.88
3.01
7.10
2.41
2.45
2.21
Log-Odds Ratio
Estimate
0.56
1.36
Standard
Error
0.37
1.18
P-value
0.1238
0.2485
A-46
-------
Q221: What is (TESTED CHILD) 's birthdate?
ENTER MONTH, DAY AND YEAR E.G. 11/18/92 d OR r
(Analysis is based on the year of birth)
Response
Category
90
91
92
93
94
95
96
.
Number of
PbB <7
//g/dL
202
408
419
396
482
954
115
23
7
-------
HOUSEHOLD INFORMATION
Q300: How many persons live in your household... counting all adults and children and
including yourself ?
ENTER #, d OR r
(Q300 -Q300m are not in data set)
So that we can make the correct reference to everyone there, please tell me just the first name (or
some other way to identify those in the household), RELATIONSHIP TO {01}, sex, and age of all
persons who live in your household. Let's start with you
Name Relationship Sex Age Delete
300d 300e 300f 300g 300h
Q300i: Correct number of people in household.
ENTER NUMBER
Q3001: Did respondent supply correct number of people in household?
1. Yes
2. No
Q300m: I've listed you, and (READ NAMES). Have I missed anyone who usually lives there
but is now away from home?
1. YES, NEED TO ADD SOMEONE TO ROSTER (PRESS 2 OR 3 TIMES AS
NECESSARY)
2. NO, TABLE IS CORRECT AS IS
3. I NEED TO CORRECT OR CHANGE A CELL
4. I NEED TO DELETE ONE PERSON FROM THE ROSTER
I would like to ask a few questions about {01}'s eating habits.
A-48
-------
Q302: Does your family usually sit down together for one or more meals per day?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
Number of
PbB <7
A/g/dL
2853
145
1
7 <. PbB
< 10
//g/dL
346
20
0
PbB 2 10
//9/dL
266
19
0
Blood-Lead
Concentration
Geometric
Mean
2.48
2.68
2.76
Geometric
Standard
Deviation
2.63
2.62
1.05
Log-Odds Ratio
Estimate
0.34
Standard
Error
0.25
P-value
0.1773
Q303: How often does {01} usually sit at the dining or kitchen table for meals? Would you say
always, sometimes, seldom or never?
1. ALWAYS
2. SOMETIMES
3. SELDOM
4. NEVER
6. TOO YOUNG
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
3
4
6
8
Number of
PbB <7
//g/dL
2184
643
73
44
53
2
7 <: PbB
< 10
//g/dL
255
80
12
15
4
0
PbB* 10
//g/dL
187
73
10
6
9
0
Blood-Lead
Concentration
Geometric
Mean
2.40
2.64
3.22
3.76
2.09
2.76
Geometric
Standard
Deviation
2.62
2.68
2.43
2.04
3.38
1.05
Log-Odds Ratio
Estimate
0.28
0.47
0.47
Standard
Error
0.15
0.35
0.44
P-value
0.0519
0.1740
0.2922
A-49
-------
Q304:
How often does {01} wash his/her hands before eating? Would you say always,
sometimes, seldom or never?
1. ALWAYS
2. SOMETIMES
3. SELDOM
4. NEVER
6. TOO YOUNG
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
3
4
6
8
Number of
PbB <7
fjg/dl
1225
1441
200
73
55
5
7 <: PbB
< 10
fjgfdl
151
180
28
4
3
0
PbB* 10
fjg/dL
110
146
14
7
7
1
Blood-Lead
Concentration
Geometric
Mean
2.47
2.49
2.69
2.57
1.72
2.91
Geometric
Standard
Deviation
2.66
2.64
2.43
2.40
3.31
2.21
Log-Odds Ratio
Estimate
0.12
-0.25
0.07
Standard
Error
0.13
0.29
0.41
P-value
0.3608
0.3967
0.8721
Q305:
How often does {01} use a plate when eating? Would you say always, sometimes,
seldom or never?
1. ALWAYS
2. SOMETIMES
3. SELDOM
4. NEVER
6. TOO YOUNG
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
3
4
6
8
Number of
PbB <7
ug/dl
2414
353
52
101
76
3
7 * PbB
< 10
pg/dL
321
30
7
5
3
0
PbB* 10
fjg/dL
228
32
9
8
8
0
Blood-Lead
Concentration
Geometric
Mean
2.55
2.47
2.15
1.79
1.69
2.76
Geometric
Standard
Deviation
2.60
2.60
3.43
2.79
3.28
1.05
Log-Odds Ratio
Estimate
-0.04
0.61
-0.18
Standard
Error
0.20
0.37
0.37
P-value
0.8351
0.0994'
0.6378
A-50
-------
ADULT OCCUPATIONS
Now I would like to ask a series of questions about your occupation or the occupation of any
other adult who lives in the household.
Q306: First, in the past 12 months has any adult held a job doing paint removal including
scraping and sanding?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
Number of
PbB <
7 /yg/dL
222
2768
8
1
7 z PbB
< 10
j/g/dL
44
321
1
0
PbB * 10
A/g/dL
25
259
1
0
Blood-Lead
Concentration
Geometric
Mean
3.16
2.43
1.64
2.76
Geometric
Standard
Deviation
2.31
2.67
3.04
1.05
Log-Odds Ratio
Estimate
-0.19
Standard
Error
0.22
P-value
0.4013
Q307: (In the past 12 months, has any adult held a job in) Home remodeling and repair?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
•
Number of
PbB <
7 A/g/dL
284
2709
5
1
7 <. PbB
< 10
fjgldl
49
317
0
0
PbB a 10
fjgldl
29
255
1
0
Blood-Lead
Concentration
Geometric
Mean
2.95
2.44
1.05
2.76
Geometric
Standard
Deviation
2.36
2.67
4.21
1.05
Log-Odds Ratio
Estimate
-0.08
Standard
Error
0.21
P-value
0.6923
A-51
-------
Q308: (In the past 12 months, has any adult held a job in ) Plumbing?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
.
Number of
PbB <
7 /ig/dL
117
2877
4
1
7 <. PbB
< 10
fjg/dl
22
344
0
0
PbB 2 10
fjg/dl
14
269
2
0
Blood-Lead
Concentration
Geometric
Mean
2.97
2.47
4.47
2.76
Geometric
Standard
Deviation
2.46
2.63
4.30
1.05
Log-Odds Ratio
Estimate
-0.25
Standard
Error
0.29
P-value
0.3948
Q309: (In the past 12 months, has any adult held a job in) Building demolition?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
.
Number of
PbB <
7 fjg/dl
104
2886
8
1
7 <. PbB
< 10
fjg/dl
22
344
0
0
PbB a 10
ug/dl
16
266
3
0
Blood-Lead
Concentration
Geometric
Mean
3.30
2.45
3.65
2.76
Geometric
Standard
Deviation
2.30
2.65
2.79
1.05
Log-Odds Ratio
Estimate
-0.51
Standard
Error
0.28
P-value
0.0635
A-52
-------
Q310: (In the past 12 months, has any adult held a job in) Welding?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
Number of
PbB <
7 /yg/dL
297
2691
10
1
7 <; PbB
< 10
A/g/dL
43
323
0
0
PbB 2 10
fjg/dL
29
256
0
0
Blood-Lead
Concentration
Geometric
Mean
2.76
2.47
2.76
2.76
Geometric
Standard
Deviation
2.44
2.66
1.05
1.05
Log-Odds Ratio
Estimate
-0.03
Standard
Error
0.21
P-value
0.8990
Q311: (In the past 12 months, has any adult held a job in ) A battery manufacturing plant?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
•
Number of
PbB <
7 /yg/dL
19
2975
4
1
7 <; PbB
< 10
//g/dL
2
363
1
0
PbB ^ 10
//g/dL
2
283
0
0
Blood-Lead
Concentration
Geometric
Mean
2.34
2.49
3.64
2.76
Geometric
Standard
Deviation
2.95
2.63
1.84
1.05
Log-Odds Ratio
Estimate
-0.10
Standard
Error
0.75
P-value
0.8920
A-53
-------
Q312: (In the past 12 months, has any adult held a job in ) The salvage of batteries or
radiators?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
Number of
PbB <
7 fjg/dl
34
2960
4
1
7 s PbB
< 10
fjg/dl
4
360
2
0
PbB 2 10
;/g/dL
10
275
0
0
Blood-Lead
Concentration
Geometric
Mean
3.98
2.47
4.36
2.76
Geometric
Standard
Deviation
2.53
2.63
1.77
1.05
Log-Odds Ratio
Estimate
-1.15
Standard
Error
0.37
P-value
0.0016
Q313: (In the past 12 months, has any adult held a job in ) Ship building or repair?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
.
Number of
PbB <
7 //g/dL
21
2977
0
1
7 s PbB
< 10
//g/dL
4
361
1
0
PbB 2 10
/vg/dL
3
281
1
0
Blood-Lead
Concentration
Geometric
Mean
3.54
2.48
10.4
2.76
Geometric
Standard
Deviation
2.28
2.64
1.15
1.05
Log-Odds Ratio
Estimate
-0.41
Standard
Error
0.62
P-value
0.5041
Q314: (In the past 12 months, has any adult held a job in ) Other lead-related industry work?
(delete)
A-54
-------
Q3J5: (In the past 12 months, has any adult held a job in) Smelter or foundry work?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
Number of
PbB <
7 //g/dL
94
2888
16
1
7 <: PbB
< 10
//9/dL
18
345
3
0
PbB 2 10
//g/dL
14
270
1
0
Blood-Lead
Concentration
Geometric
Mean
3.72
2.44
3.27
2.76
Geometric
Standard
Deviation
2.23
2.66
2.25
1.05
Log-Odds Ratio
Estimate
-0.47
Standard
Error
0.29
P-value
0.1126
Q316: (In the past 12 months, has any adult held a job in ) Oil refinery -work?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
•
Number of
PbB <
7 //g/dL
9
2989
0
1
7 i PbB
< 10
//g/dL
1
364
1
0
PbBk 10
//g/dL
0
284
1
0
Blood-Lead
Concentration
Geometric
Mean
3.72
2.48
8.94
2.76
Geometric
Standard
Deviation
1.51
2.64
1.12
1.05
Log-Odds Ratio
Estimate
Standard
Error
P-value
A-55
-------
Q317: (In the past 12 months, has any adult held a job in ) Auto body work?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
Number of
PbB <
7 pg/dL
157
2835
6
1
7 i PbB
< 10
/yg/dL
34
332
0
0
PbB* 10
fjg/dl
16
269
0
0
Blood-Lead
Concentration
Geometric
Mean
3.05
2.46
2.89
2.76
Geometric
Standard
Deviation
2.33
2.66
1.42
1.05
Log-Odds Ratio
Estimate
-0.07
Standard
Error
0.27
P-value
0.7914
Q318: (In the past 12 months, has any adult held a job in ) Glass work?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
.
Number of
PbB <
7 fjgldl
53
2943
2
1
7 <. PbB
< 10
jug/dL
11
354
1
0
PbBi 10
fjg/dl
5
280
0
0
Blood-Lead
Concentration
Geometric
Mean
2.97
2.48
2.88
2.76
Geometric
Standard
Deviation
2.52
2.64
2.62
1.05
Log-Odds Ratio
Estimate
0.01
Standard
Error
0.47
P-value
0.9857
A-56
-------
Q319: (In the past 12 months, has any adult held a job in) A chemical plant?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
Number of
PbB <
7 //g/dL
43
2951
4
1
7 <. PbB
< 10
//g/dL
4
361
1
0
PbB i 10
//g/dL
2
283
0
0
Blood-Lead
Concentration
Geometric
Mean
3.35
2.47
3.67
2.76
Geometric
Standard
Deviation
1.84
2.65
1.59
1.05
Log-Odds Ratio
Estimate
0.72
Standard
Error
0.73
P-value
0.3189
Q320: (In the past 12 months, has any adult held a job in ) Sandblasting?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
.
Number of
PbB <
7 //g/dL
97
2894
7
1
7 * PbB
< 10
//g/dL
21
344
1
0
PbB i 10
//g/dL
6
278
1
0
Blood-Lead
Concentration
Geometric
Mean
2.94
2.48
1.06
2.76
Geometric
Standard
Deviation
2.41
2.64
6.22
1.05
Log-Odds Ratio
Estimate
0.44
Standard
Error
0.43
P-value
0.3007
A-57
-------
Q321: (In the past 12 months, has any adult held a job in ) Other lead related occupations?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
Number of
PbB <
7 y^g/dL
65
2888
45
1
7 <. PbB
< 10
fjg/dl
12
345
9
0
PbBi 10
//g/dL
13
266
6
0
Blood-Lead
Concentration
Geometric
Mean
3.21
2.46
3.34
2.76
Geometric
Standard
Deviation
2.77
2.64
2.12
1.05
Log-Odds Ratio
Estimate
-0.78
Standard
Error
0.31
P-value
0.0125
Q322: (Has any one in the household worked at another lead related occupation in the past 12
months?) What jobs were they?
ENTER EACH ADULT OCCUPATION AND INDUSTRY ON SAME LINE, d
ORr,
(Response was either 'other' or 'missing')
Response
Category
0
1
2
3
4
5
6
7
8
9
998
.
Number of
PbB <
7 jug/dL
5
11
8
2
6
5
3
11
4
5
1
2938
7 ± PbB
< 10
//g/dL
3
2
1
0
1
0
0
3
0
1
0
355
PbB * 10
jug/dL
1
3
2
0
1
1
1
3
0
1
0
272
Blood-Lead
Concentration
Geometric
Mean
4.80
3.74
3.04
2.76
3.07
0.44
4.30
4.05
3.32
0.35
6.00
2.48
Geometric
Standard
Deviation
1.89
2.18
2.58
1.05
2.73
10.6
2.21
2.56
1.35
69.5
1.00
2.63
Log-Odds Ratio
Estimate
0.31
0.22
-23.8
-0.18
0.00
0.51
0.31
-23.8
-0.00
Standard
Error
1.27
1.35
23 E4
1.54
1.55
1.59
1.27
16E4
1.55
P-value
0.8077
0.8688
0.9999
0.9057
1 .0000
0.7483
0.8077
0.9999
1 .0000
A-58
-------
HOBBIES
Now I would like to ask a series of questions about hobbies of anyone who lives in the household.
Q324: Within the past 12 months, has anyone in the household removed paint or varnish from
furniture?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
Number of
PbB <
7 /yg/dL
334
2664
1
7 <; PbB
< 10
fjg/dl
48
318
0
PbB 2 10
/yg/dL
36
248
1
Blood-Lead
Concentration
Geometric
Mean
2.85
2.44
4.96
Geometric
Standard
Deviation
2.43
2.66
3.35
Log-Odds Ratio
Estimate
-0.15
Standard
Error
0.19
P-value
0.4347
Q325: (Within the past 12 months, has anyone in the household) Soldered pipes or repaired
plumbing?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
Number of
PbB <
7 /yg/dL
465
2529
5
7 <. PbB
< 10
/yg/dL
66
298
2
PbB* 10
/yg/dL
54
231
0
Blood-Lead
Concentration
Geometric
Mean
2.76
2.44
2.56
Geometric
Standard
Deviation
2.57
2.65
2.43
Log-Odds Ratio
Estimate
-0.24
Standard
Error
0.16
P-value
0.1319
A-59
-------
Q326: (Within the past 12 months, has anyone in the household) Joined pieces of stained
glass with solder?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
Number of
PbB <
7 //g/dL
12
2986
1
7 s PbB
< 10
/yg/dL
3
363
0
PbBi 10
fjg/dl
2
283
0
Blood-Lead
Concentration
Geometric
Mean
4.70
2.47
5.00
Geometric
Standard
Deviation
1.67
2.65
1.00
Log-Odds Ratio
Estimate
-0.56
Standard
Error
0.77
P-value
0.4613
Q327: (Within the past 12 months, has anyone in the household) Painted pictures or jewelry
with artists paint?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
Number of
PbB <
7 fjg/dL
214
2774
11
7 <. PbB
< 10
//g/dL
29
336
1
PbB * 10
Hg/dl
22
263
0
Blood-Lead
Concentration
Geometric
Mean
2.52
2.49
2.92
Geometric
Standard
Deviation
2.66
2.64
1.64
Log-Odds Ratio
Estimate
-0.08
Standard
Error
0.23
P-value
0.7282
A-60
-------
Q328: (Within the past 12 months, has anyone in the household) Glazed pottery or ceramic
objects?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
Number of
PbB <
7 /yg/dL
94
2903
2
7 <: PbB
< 10
fjg/dL
8
358
0
PbBi 10
fjg/dL
7
277
1
Blood-Lead
Concentration
Geometric
Mean
2.06
2.50
2.04
Geometric
Standard
Deviation
2.84
2.63
7.15
Log-Odds Ratio
Estimate
0.25
Standard
Error
0.40
P-value
0.5321
Q329: (Within the past 12 months, has anyone in the household) Performed auto maintenance
or body repair near the house?
(REFERS TO BODY WORK, BATTERY WORK OR RADIATOR REPAIR - DOES
NOT INCLUDE OIL CHANGES, LUBRICATION, ETC.)
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
.
Number of
PbB <
7 fjg/dl
655
2342
1
1
7 * PbB
< 10
fjg/dl
90
275
1
0
PbB* 10
fjgldl
71
213
1
0
Blood-Lead
Concentration
Geometric
Mean
2.79
2.40
7.82
6.00
Geometric
Standard
Deviation
2.56
2.65
3.37
1.00
Log-Odds Ratio
Estimate
-0.18
Standard
Error
0.14
P-value
0.2229
A-61
-------
Q330: (Within the past 12 months, has anyone in the household) Molded lead into bullets,
sinkers or other objects?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
Number of
PbB <
7 //g/dL
40
2956
3
7 s PbB
< 10
//g/dL
4
362
0
PbB * 10
//g/dL
6
279
0
Blood-Lead
Concentration
Geometric
Mean
1.94
2.50
2.76
Geometric
Standard
Deviation
3.60
2.62
1.05
Log-Odds Ratio
Estimate
-0.46
Standard
Error
0.44
P-value
0.2949
Q331: Has any one in the household worked at another lead related hobby in the past 12
months?
1. YES
2. NO
8. DONTKNOW
9. REFUSED
Response
Category
1
2
8
Number of
PbB <
7 //g/dL
45
2939
15
7 * PbB
< 10
//g/dL
9
355
2
PbB* 10
//g/dL
7
274
4
Blood-Lead
Concentration
Geometric
Mean
3.52
2.47
3.27
Geometric
Standard
Deviation
2.59
2.63
2.43
Log-Odds Ratio
Estimate
-0.51
Standard
Error
0.41
P-value
0.2131
A-62
-------
Q332: What hobbies were they?
ENTER ALL HOBBIES AND PERSON NUMBER'S ON SEPARATE LINES, d
ORr,
(The following analysis is based on the number of hobbies)
Response
Category
0
1
2
3
4
5
6
98
Number of
PbB <
7 //g/dL
6
6
7
15
6
1
2
2
2954
7 s PbB
< 10
/ig/dL
0
1
2
3
1
1
1
0
357
PbB i
10
//g/dL
0
1
2
4
0
0
0
0
278
Blood-Lead
Concentration
Geometric
Mean
2.89
3.17
4.41
4.22
3.02
4.38
5.33
4.27
2.47
Geometric
Standard
Deviation
1.42
2.36
3.97
2.32
1.72
1.67
1.50
1.45
2.63
Log-Odds Ratio
Estimate
Standard
Error
P-value
A-63
-------
BACKGROUND
Now we just have a few background questions to help us to interpret the results of this study.
Q333: How many people live in your household...counting all adults and children and
including yourself?
ENTER #,d OR r
Response
Category
<=2
3
4
5
6
>=7
.
Number of
PbB <
7 jJQldl
111
602
1046
704
293
241
2
7 & PbB
< 10
pg/dL
18
75
119
64
52
38
0
PbB * 10
fjg/dl
10
46
89
65
33
42
0
Blood-Lead
Concentration
Geometric
Mean
2.91
2.45
2.30
2.28
2.91
3.25
2.76
Geometric
Standard
Deviation
2.46
2.53
2.66
2.81
2.47
2.54
1.05
Log-Odds Ratio
Estimate
-0.16
-0.06
0.02
0.22
0.66
Standard
Error
0.36
0.35
0.35
0.38
0.37
P-value
0.6509
0.8696
0.9448
0.5545
0.0746
A-64
-------
Q333a: What was your marital status?
1. DIVORCED
2. WIDOWED
3. SEPARATED
4. NEVER BEEN MARRIED
5. MEMBER OF AN UNM
6. MARRIED
8. DON'T KNOW/NOT SURE
9. REFUSED
Response
Category
1
2
3
4
5
6
8
.
Number of
PbB <
7 //g/dL
224
10
102
311
110
2235
2
5
7 s PbB
< 10
fjg/dl
28
5
13
45
13
261
0
1
PbB 2 10
fjg/dl
23
1
10
31
18
206
0
0
Blood-Lead
Concentration
Geometric
Mean
2.52
3.82
2.39
2.70
2.87
2.41
2.76
1.77
Geometric
Standard
Deviation
2.59
1.96
2.83
2.68
2.63
2.68
1.05
2.41
Log-Odds Ratio
Estimate
-0.03
-0.05
-0.03
0.47
-0.11
Standard
Error
1.07
0.40
0.29
0.34
0.23
P-value
0.9803
0.9073
0.9182
0.1649
0.6399
A-65
-------
Q333d: What was the highest grade or year in school that (you have /{:B} has) completed?
0. 8TH GRADE OR LESS
1. 9TH-11TH GRADE
2. H.S. GRADUATE OR HAS G.E.D.
3. SOME TECHNICAL SCHOOL OR VOCATIONAL TRAINING
4. TECHNICAL SCHOOL GRADUATE
5. SOME COLLEGE OR ASSOCIATE DEGREE
6. COLLEGE GRADUATE
7. POST GRADUATE OR PROFESSIONAL DEGREE
9. DON'T KNOW / REFUSED
Response
Category
0
1
2
3
4
5
6
7
.
Number of
PbB <
7 fjg/dl
67
244
1188
254
153
595
364
126
8
7 s PbB
< 10
/vg/dL
12
40
157
30
17
61
42
6
1
PbBi 10
UgldL
23
38
118
24
15
36
20
11
0
Blood-Lead
Concentration
Geometric
Mean
4.06
3.39
2.55
2.54
2.76
2.20
2.16
1.39
2.59
Geometric
Standard
Deviation
2.67
2.37
2.65
2.59
2.63
2.54
2.53
3.50
1.81
Log-Odds Ratio
Estimate
-0.79
-1.24
-1.29
-1.25
-1.74
-1.83
-1.37
Standard
Error
0.30
0.26
0.32
0.36
0.30
0.33
0.40
P-value
0.0080
0.0001
0.0001
0.0006
0.0001
0.0001
0.0006
A-66
-------
Q333p: What was the highest grade or year in school that (your partner/spouse have /{:B} has)
completed?
0. 8TH GRADE OR LESS
1. 9TH-11TH GRADE
2. H.S. GRADUATE OR HAS G.E.D.
3. SOME TECHNICAL SCHOOL OR VOCATIONAL TRAINING
4. TECHNICAL SCHOOL GRADUATE
5. SOME COLLEGE OR ASSOCIATE DEGREE
6. COLLEGE GRADUATE
7. POST GRADUATE OR PROFESSIONAL DEGREE
9. DON'T KNOW/REFUSED
Response
Category
0
1
2
3
4
5
6
7
.
Number of
PbB <
7 //g/dL
76
200
995
155
150
340
299
119
665
7 s PbB
< 10
//g/dL
11
24
134
17
15
34
32
7
92
PbB 2. 10
//g/dL
19
30
85
11
10
34
20
10
70
Blood-Lead
Concentration
Geometric
Mean
3.34
3.18
2.62
2.30
2.31
2.13
1.90
1.39
2.62
Geometric
Standard
Deviation
2.69
2.58
2.52
2.63
2.35
2.85
2.96
3.69
2.67
Log-Odds Ratio
Estimate
-0.51
-1.07
-1.26
-1.32
-0.92
-1.32
-1.09
Standard
Error
0.32
0.28
0.40
0.42
0.31
0.35
0.42
P-value
0.1134
0.0001
0.0018
0.0015
0.0034
0.0001
0.0090
A-67
-------
INCOME
Q335: Is your household income more or less than $30,000 per year?
1. MORE THAN 30,000
2. LESS THAN 30,000
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
Number of
PbB <
7 fjg/dl
1299
1606
65
29
7 i PbB
< 10
Jjgldl
139
213
11
3
PbB * 10
fjg/dl
83
193
8
1
Blood-Lead
Concentration
Geometric
Mean
2.12
2.78
2.54
2.54
Geometric
Standard
Deviation
2.67
2.59
2.68
2.18
Log-Odds Ratio
Estimate
0.63
Standard
Error
0.14
P-value
0.0001
Q335a: Thank you very much for your help with these questions. I have some information for
you on lead poisoning prevention. Would you like me to send it to you?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
8
•
Number of
PbB <
7 fjg/dL
1537
1457
2
3
7 i PbB
< 10
ug/dL
188
178
0
0
PbB 2 10
fjg/dL
150
135
0
0
Blood-Lead
Concentration
Geometric
Mean
2.58
2.39
4.15
2.76
Geometric
Standard
Deviation
2.59
2.69
1.23
1.05
Log-Odds Ratio
Estimate
-0.05
Standard
Error
0.12
P-value
0.6758
A-68
-------
Q335b: The address we have for you is {Ok} Is that correct?
1. YES
2. NO
8. DON'T KNOW
9. REFUSED
Response
Category
1
2
Number of
PbB <
7 //g/dL
162
29
2808
7 s PbB
< 10
//g/dL
17
3
346
PbB 2 10
//g/dL
18
2
269
Blood-Lead
Concentration
Geometric
Mean
2.37
3.53
2.46
Geometric
Standard
Deviation
2.80
1.87
2.68
Log-Odds Ratio
Estimate
-0.48
Standard
Error
0.77
P-value
0.5368
Q336: What is your correct address?
ENTER COMPLETE ADDRESS INCLUDING ZIP CODE ON SEPARATE LINES
A-69
-------
This page intentionally blank.
A-70
-------
APPENDIX B:
LOGISTIC REGRESSION BACKGROUND
B-1
-------
This page intentionally blank.
B-2
-------
APPENDIX B: LOGISTIC REGRESSION BACKGROUND
Logistic regression (McCullagh and Nelder, 1989) is a commonly used method for
analyzing the data from a retrospective case control study. In this study, children were selected
based on their observed blood-lead concentrations. Since the purpose of this study was to assess
the impact of renovation and remodeling activities on children's blood-lead concentrations, an
analysis methodology that respects the fact that observations were chosen based on the value of
the response variable must be used.
One inference mechanism useful for retrospective, binomial data is the odds ratio. An
odds ratio is a fraction with the odds for one group of children in the denominator and the odds
for a second group in the numerator. The odds for a group are the probability that a subject in
that group has some condition over the probability that a subject does not have the condition. In
this study, the odds of interest are the probability of an elevated blood-lead concentration divided
by the probability of a nonelevated blood-lead concentration. Groups are defined by responses to
particular questions. An odds ratio statistically greater than 1 indicates that the numerator group
of children is at higher risk as a result of lead exposure.
In a prospective study a group of children with a specified risk factor are first identified
and then blood-lead concentrations are measured. In a retrospective study, risk factors are
determined for a group of children whose blood-lead concentrations were previously measured.
The odds ratio for a retrospective study:
. _ Pr(RR | Case) I Pr(No RR \ Case)
V Retrospective Study p,.^ ( Controi) / pr^No gR \ Control)
is identical (in interpretation) to that for a prospective study:
. _ Pr(Case \ RR) I Pr(Control \ RR)
Prospective Study pr(Case \ No RR) I Pr(Cantrol \ No RR)
where
. _ Pr(RR | Case) I Pr(No RR \ Case)
Retrospective Study ^^ ( Control) / jy^ jyj | Control)
is identical (in interpretation) to that for a prospective study:
where
Pr(RR|Case) = the conditional probability of renovation and remodeling
having been performed in a house with a child whose
blood-lead concentration exceeds 10 ug/dL,
B-3
-------
Pr(No RR|Case) = the conditional probability of no renovation and remodeling
having been performed in a house with a child whose
blood-lead concentration exceeds 10 ug/dL,
Pr(RR| Control) = the conditional probability of renovation and remodeling
having been performed in a house with a child whose
blood-lead concentration does not exceed 10 ug/dL, and
Pr(no RR|Control) = the conditional probability of no renovation and remodeling
having been performed in a house with a child whose
blood-lead concentration does not exceed 10 ug/dL.
Cases = Residences containing children with a blood-lead
concentration exceeding 10 ug/dL.
The odds ratio is used to test whether the incidence of elevated blood-lead concentrations
in homes in which renovation and remodeling occurred is greater than that in which renovation
and remodeling did not occur. For example, an odds ratio of 2 may be interpreted as the odds of
a child's blood-lead concentration exceeding 10 ug/dL increases twofold with the conduct of
renovation and remodeling.
The odds ratio is the natural form of inference based on logistic regression. Logistic
regression models the probability of an event in the following form
where x,, x2,..., Xp are predictors of the event and p(xl5 x2,..., x,,) is the probability of the event
given the values of the predictor variables. Alternatively, the logistic regression model can be
written as
..... . B§+prt+pA+_+| . (i)
Therefore, the logistic regression model assumes that the log-transformed odds ratio at one set of
predictor variables is a linear function of the parameters P0, PI} P2,..., Pp.
Log odds differences are calculated to compare the odds at one set of predictor variables
to the odds at another set of predictor variables by differencing the appropriate linear functions of
the parameters,
.( p(x,,x2,...,x ) "i ( p(yi,y2>...,yp)
Iny = ln\ — - ln\ t—
\i-p(x1,x2,...,xp)) \l-(yi,y2,-,yp)
B-4
-------
Odds ratios, i|j, are calculated by exponentiating the log odds difference. To compare the odds
for two groups that are different only in one predictor variable, for example, the log odds
difference is equal to the parameter associated with that predictor multiplied by the difference in
the predictor variable between the two groups. The odds ratio is calculated by exponentiating
that value.
B-5
-------
This page intentionally blank.
B-6
-------
APPENDIX C:
TEST-WISE SIGNIFICANT RESULTS
C-1
-------
This page intentionally blank.
C-2
-------
APPENDIX C: TEST-WISE SIGNIFICANT RESULTS
This appendix presents all the univariate logistic regressions for which the test-wise
significance level was less than 10 percent, but the result failed to qualify for significance with
the Holm Bonferroni correction. Results are sorted by magnitude of the odds ratios under the
restriction that odds ratios based on a single question remain together.
Table C-1. Odds Ratio Estimates and 95 percent Confidence Intervals, Test-Wise
Significant Question Responses
All Children with Blood-Lead Concent
n = 3.28
Q32
Q61
Q33
0.20 1b
Q333d
Q333d
0.333d
0.333d
Q333d
Q333d
Q333d
Q35
Q64_4
Q23
Q312
0.3 6_3
Q221
Question
Open flame torch (outside)
Heat gun (carpet & floor)
Heat gun (outside)
Blood drawn from finger, arm or
someplace else
Highest grade completed by
respondent
Highest grade completed by
respondent
Highest grade completed by
respondent
Highest grade completed by
respondent
Highest grade completed by
respondent
Highest grade completed by
respondent
Highest grade completed by
respondent
Chemical paint remover (outside)
Owner, staff did work (carpet &
floor)
In how many rooms
Adult occupation in salvage of
batteries or radiators
Relative not in household did
work (outside)
Date of birth for tested child
Hiaher Risk Group
Yes
Yes
Yes
Arm (venous)
8th Grade or less
8th Grade or less
8th Grade or less
8th Grade or less
8th Grade or less
8th Grade or less
8th Grade or less
Yes
Yes
*4
Yes
Yes
95
rations <7//g/dL or i10//g/dL
8
Lower Risk Grouo
No
No
No
Finger (capillary)
Some College
Tech. Grad.
College Grad.
Post Grad.
H.S. Grad or GED
Some Tech or
Vocational
9th- 11th Grade
No
No
1
No
No
90
Odds Ratio
Estimate
7.029
6.959
6.89
6.297
5.697
3.49
6.234
3.935
3.456
3.633
2.203
3.743
3.49
3.158
3.158
2.746
1.896
95% Confidence
Interval
[1.139, 43.38]
[1.127, 42.948]
[2.915, 16.281]
[8.499, 4.665]
[3.127, 10.381]
[1.699, 7.171]
[3.222, 12.061]
[1.768, 8.758]
[2.054, 5.812]
[1.916, 6.89]
[1.209, 4.015]
[1.433, 9.777]
[0.914, 13.33]
[1.507, 6.619]
[1.507, 6.619]
[0.99, 7.614]
[3.743, 0.961]
C-3
-------
Table C-1a. (Continued)
All Children with Blood-Lead Concentrations <7//g/dL or *10//g/dL
n = 3.288
Q221
Q221
Q221
Q321
Q31
Q24
Q55
Q333
Q335
Q19
Q305
Q34
Q309
Q28
Q200
Q25
Q54
Q303
Question
Date of birth for tested child
Date of birth for tested child
Date of birth for tested child
Adult occupation in other lead
related occupation
Power sanding, grinding or
sandblasting (outside)
Any work in kitchen
Any wall to wall carpet in home
How many people live in
household
Household Income
Chemical paint remover (inside)
How often does the tested child
use a plate
Washing, wet scraping, wet
sanding or water blasting
(outside)
Adult occupation in building
demolition
Any outside painting
Who asked for blood test
Any work in bathroom
Carpets replaced or removed or
painted floors refinished
Does the tested child sit at the
table
Hiaher Risk Group
93
94
92
Yes
Yes
Yes
Yes
*7
It 30,000.
Yes
Seldom
Yes
Yes
Yes
Wisconsin
recommended
Yes
Yes
Sometimes
Lower Risk Grouo
90
90
90
No
No
No
No
*2
gt 30,000.
No
Always
No
No
No
I asked for it
No
No
Always
Odds Ratio
Estimate
2.34
2.435
2.117
2.181
2.034
1.974
1.954
1.935
1.878
1.84
1.84
1.786
1.665
1.632
1.616
1.537
1.419
1.323
95% Confidence
Interval
[4.807, 1.139]
[4.904, 1.209]
[4.349, 1.03]
[1.174, 4.055]
[0.97, 4.263]
[1.271, 3.065]
[0.932, 4.096]
[4.055, 0.923]
[2.484, 1.419]
[0.97, 3.49]
[3.857, 0.878]
[0.98, 3.254]
[0.951, 2.915]
[1.162, 2.293]
[2.363, 1.105]
[0.99, 2.387]
[1.051, 1.916]
[1.786, 0.98]
C-4
-------
APPENDIX D:
UNIVARIATE REGRESSION OF
CONSTRUCTED R&R VARIABLES
D-1
-------
This page intentionally blank.
D-2
-------
Univariate Regression of Constructed R&R Variables
Any R&R Work
Was any renovation and remodeling work done in the last 12 months, including interior painting
or exterior painting or were windows replaced, walls moved, carpets replaced, or floors
resurfaced?
1. YES
2. NO
Response
Category
1
2
.
Number of
PbB<7
Mg/dL
1836
1151
12
7 s PbB <
10
Mg/dL
251
113
2
PbB* 10
Mg/dL
187
92
6
Blood-Lead
Concentration
Geometric
Mean
2.61
2.28
4.27
Geometric
Standard
Deviation
2.63
2.61
2.99
Log-Odds Difference*
Estimate
-0.48
Standard
Error
0.16
P-value
0.0024
Inside Painting
During the last 12 months, was there any painting or were any surfaces prepared for paint inside
the home?
.1. YES
2. NO
Response
Category
1
2
.
Number of
PbB<7
Mg/dL
1371
1620
8
7 * PbB <
10
M9/dL
181
185
0
PbB a 10
Mg/dL
128
155
2
Blood-Lead
Concentration
Geometric
Mean
2.54
2.45
0.88
Geometric
Standard
Deviation
2.63
2.63
6.62
Log-Odds Difference
Estimate
-0.14
Standard
Error
0.14
P-value
0.3222
D-3
-------
Other Repairs (Window)
Were any windows repaired or new windows put in?
1. YES
2. NO
Response
Category
1
2
Number of
PbB<7
ug/dL
592
2402
5
7 <; PbB <
10
ug/dL
93
271
2
PbB * 10
|jg/dL
68
214
3
Blood-Lead
Concentration
Geometric
Mean
3.00
2.35
4.64
Geometric
Standard
Deviation
2.46
2.68
2.78
Log-Odds Difference
Estimate
-0.25
Standard
Error
0.17
P-value
0.1283
Inside or Outside Painting
During the last 12 months, was there any painting or were any surfaces prepared for paint inside
or outside the home?
1. YES
2. NO
Response
Category
1
2
.
Number of
PbB<7
ug/dL
1535
1451
13
7 <; PbB <
10
ug/dL
203
161
2
PbB i 10
ug/dL
155
125
5
Blood-Lead
Concentration
Geometric
Mean
2.55
2.43
2.51
Geometric
Standard
Deviation
2.65
2.58
4.22
Log-Odds Difference
Estimate
-0.31
Standard
Error
0.14
P-value
0.0304
D-4
-------
Prepared Surface
Were any surfaces prepared for interior painting, exterior painting, window installation/wall
repair or carpet replacement, or floor resurfacing by methods such as sanding, scraping, —
of chemicals other than prime painting?
or use
1. YES
2. NO
Response
Category
1
2
Number of
PbB<7
Mg/dL
527
2348
124
7 £ PbB <
10
(jg/dL
84
266
16
PbB ;> 10
Mg/dL
66
204
15
Blood-Lead
Concentration
Geometric
Mean
2.87
2.38
2.98
Geometric
Standard
Deviation
2.60
2.65
2.51
Log-Odds Difference
Estimate
-0.51
Standard
Error
0.16
P-value
0.0019
Prepared Surface for Inside Painting
Were any surfaces prepared for interior painting by methods such as sanding, scraping, or use
of chemicals other than prime painting?
1. YES
2. NO
Response
Category
1
2
.
Number of
PbB<7
Mg/dL
283
2651
65
7 * PbB <
10
ug/dL
48
314
4
PbB i 10
Mg/dL
34
244
7
Blood-Lead
Concentration
Geometric
Mean
3.22
2.41
2.32
Geometric
Standard
Deviation
2.40
2.66
2.90
Log-Odds Difference
Estimate
-0.39
Standard
Error
0.21
P-value
0.0627
D-5
-------
Hand Sanding or Scraping
Was hand scraping or hand sanding used to prepare surfaces for inside painting, outside
painting, window installation or wall repair, or when carpets were replaced or floors
resurfaced?
1. YES
2. NO
Response
Category
1
2
.
Number of
PbB<7
ug/dL
481
2503
15
7 <; PbB <
10
ug/dL
76
289
1
PbB* 10
|jg/dL
54
227
4
Blood-Lead
Concentration
Geometric
Mean
2.84
2.42
2.41
Geometric
Standard
Deviation
2.50
2.65
4.21
Log-Odds Difference
Estimate
-0.32
Standard
Error
0.17
P-value
0.0671
Power Sanding, Grinding, Sandblasting
Was power sanding, grinding, or sandblasting used to prepare the surface for interior painting,
exterior painting, window installation/wall repair, carpet replacement, or floor resurfacing?
1. YES
2. NO
Response
Category
1
2
.
Number of
PbB<7
ug/dL
146
2829
24
7 <: PbB <
10
ug/dL
29
334
3
PbB * 10
ug/dL
22
260
3
Blood-Lead
Concentration
Geometric
Mean
3.33
2.44
2.59
Geometric
Standard
Deviation
2.47
2.64
2.85
Log-Odds Difference
Estimate
-0.68
Standard
Error
0.25
P-value
0.0063
D-6
-------
Open Flame Torch
Was an open flame torch used to prepare surfaces for inside painting, outside painting, window
installation or wall repair, or for carpet replacement or floor resurfacing?
1. YES
2. NO
Response
Category
1
2
Number of
PbB<7
|jg/dL
6
2978
15
7 <. PbB <
10
ug/dL
0
363
3
PbB 2 10
ug/dL
2
278
5
Blood-Lead
Concentration
Geometric
Mean
3.04
2.48
3.62
Geometric
Standard
Deviation
2.78
2.63
2.75
Log-Odds Difference
Estimate
-1.54
Standard
Error
0.82
P-value
0.0604
Heat Guns
Were heat guns used to prepare surface for inside painting, outside painting, window installation
or wall repair, or for carpet replacement or floor resurfacing?
LYES
2. NO
Response
Category
1
2
.
Number of
PbB<7
Mg/dL
33
2947
19
7 * PbB <
10
Mg/dL
5
358
3
PbB i 10
ug/dL
13
266
6
Blood-Lead
Concentration
Geometric
Mean
4.38
2.47
3.44
Geometric
Standard
Deviation
2.56
2.62
3.10
Log-Odds Difference
Estimate
-1.58
Standard
Error
0.36
P-value
0.0001
D-7
-------
Washing, Wetscraping, Waterblasting
Was washing, wet scraping, wet sanding, or water blasting used to prepare surfaces for inside
painting, outside painting, window installation or wall repair, or for carpet replacement or floor
resurfacing?
1. YES
2. NO
Response
Category
1
2
Number of
PbB<7
Mg/dL
144
2824
31
7 s PbB <
10
Mg/dL
30
333
3
PbB 2 10
Mg/dL
22
257
6
Blood-Lead
Concentration
Geometric
Mean
3.23
2.45
2.43
Geometric
Standard
Deviation
2.44
2.64
3.26
Log-Odds Difference*
Estimate
0.00
-0.53
Standard
Error
0.00
0.27
P-value
0.0465
Chemical Paint Removers
Were chemical paint removers used to prepare for inside painting or for outside painting, when
windows or walls were replaced/moved, when carpets were replaced, or when floors were
resurfaced?
LYES
2. NO
Response
Category
1
2
.
Number of
PbB<7
Mg/dL
73
2904
22
7 s. PbB <
10
Mg/dL
12
352
2
PbB i 10
Mg/dL
12
266
7
Blood-Lead
Concentration
Geometric
Mean
3.49
2.47
2.52
Geometric
Standard
Deviation
2.62
2.61
4.18
Log-Odds Difference
Estimate
-0.78
Standard
Error
0.33
P-value
0.0190
D-8
-------
Who did the work for interior painting, exterior painting, window work, or carpet or floor work?
l.YES
2. NO
Head of Household or Spouse
Response
Category
1
2
Number of
PbB<
7 ug/dL
394
2605
7 <. PbB
<10
ug/dL
66
300
PbB * 10
ug/dL
41
244
Blood-Lead
Concentration
Geometric
Mean
2.96
2.41
Geometric
Standard
Deviation
2.41
2.68
Log-Odds Difference
Estimate
-0.23
Standard
Error
0.19
P-value
0.2329
Other (than head) Person in Household
Response
Category
1
2
Number of
PbB<
7 ug/dL
64
2935
7s PbB
<10
ug/dL
6
360
PbB 2 10
ug/dL
3
282
Blood-Lead
Concentration
Geometric
Mean
2.21
2.48
Geometric
Standard
Deviation
2.18
2.64
Log-Odds Difference
Estimate
-0.45
Standard
Error
0.59
P-value
0.4483
Friend or Relative not in Household
Response
Category
1
2
Number of
PbB<
7 ug/dL
54
2945
7s PbB
<10
ug/dL
10
356
PbB 2 10
ug/dL
13
272
Blood-Lead
Concentration
Geometric
Mean
3.39
2.47
Geometric
Standard
Deviation
2.52
2.63
Log-Odds Difference
Estimate
-0.59
Standard
Error
0.41
P-value
0.1494
D-9
-------
Who did the work for interior painting, exterior painting, window work, or carpet or floor work?
(Continued)
Owner or Apartment Staff
Response
Category
1
2
Number of
PbB<
7 Mg/dL
64
2935
7s PbB
<10
Mg/dL
11
355
PbB i 10
Mg/dL
12
273
Blood-Lead
Concentration
Geometric
Mean
2.78
2.49
Geometric
Standard
Deviation
3.11
2.62
Log-Odds Difference
Estimate
-0.79
Standard
Error
0.35
P-value
0.0237
Professional
Response
Category
1
2
Number of
PbB<
7 Mg/dL
81
2918
7s PbB
<10
M9/dL
6
360
PbBi 10
ug/dL
10
275
Blood-Lead
Concentration
Geometric
Mean
1.94
2.50
Geometric
Standard
Deviation
3.35
2.62
Log-Odds Difference
Estimate
-0.43
Standard
Error
0.36
P-value
0.2258
D-10
-------
Live in Home While R&R was Done
Did anyone live in your household while the work (inside painting, outside painting, window
work or floor work) was being done?
1. YES
2. NO
Response
Category
1
2
Number of
PbB<7
Mg/dL
457
2542
7 <: PbB <
10
ug/dL
71
295
PbBi 10
ug/dL
55
230
Blood-Lead
Concentration
Geometric
Mean
2.87
2.42
Geometric
Standard
Deviation
2.56
2.65
Log-Odds Difference
Estimate
-0.42
Standard
Error
0.17
P-value
0.0151
Number of Rooms
In how many rooms in your home was this work done?
ENTER NUMBER OF ROOMS d OR r OR 0 FOR NONE
Response
Category
0
1
2
3
*4
Number of
PbB<7
ug/dL
2720
95
63
34
87
7 <. PbB <
10
ug/dL
322
13
10
8
13
PbB 2. 10
ug/dL
251
6
7
3
18
Blood-Lead
Concentration
Geometric
Mean
2.41
2.85
2.99
3.71
3.56
Geometric
Standard
Deviation
2.66
2.25
2.47
2.32
2.49
Log-Odds Difference *
Estimate
0.13
Standard
Error
0.05
P-value
0.0089
* Number of Rooms was entered as a continuous variable. Hence the log-odds difference compares n+1 to n rooms. (See
Appendix B for details.)
D-11
-------
R&R Work in Kitchen
Was any of this (indoor painting and surface preparation) work done in the kitchen?
1. YES
2. NO
Response
Category
1
2
Number of
PbB<7
pg/dL
117
2881
1
7 <: PbB <
10
Mg/dL
27
339
0
PbB * 10
pg/dL
20
265
0
Blood-Lead
Concentration
Geometric
Mean
3.18
2.46
5.00
Geometric
Standard
Deviation
2.69
2.63
1.00
Log-Odds Difference*
Estimate
-0.66
Standard
Error
0.28
P-value
0.0164
D-12
-------
50272-101
REPORT DOCUMENTATION
PAGE
1. REPORT NO.
EPA-747-R-99-002
3. Recipient's Accession No.
4. Title and Subtitle
"Lead Exposure Associated with Renovation and Remodeling Activities:
Phase III, Wisconsin Childhood Blood-Lead Study"
5. Report Date
March 1999
6.
7. Author(s)
Nancy McMillan, Ron Menton, Kensuke Shirakawa, Abi Katz-Stein, Ben Pierce,
and Ying-Liang Chou
8. Performing Organization Rept. No.
9. Performing Organization Name and Address
Battelle Memorial Institute
505 King Avenue
Columbus, Ohio 43201-2693
10. Project/Task/Work Unit No.
G003470-05
11. Contract(C) or Grant(G) No.
(C) 68-D5-0008
12. Sponsoring Organization Name and Address
U.S. Environmental Protection Agency
Office of Pollution Prevention and Toxics (7401)
401 M Street, S.W.
Washington, D.C. 20460
13. Type of Report & Period Covered
Final Report
14.
15. Supplementary Notes
16. Abstract (Limit 200 words)
In 1992, the United States Congress enacted the Residential Lead-Based Paint Hazard Reduction Act, which required
(Section 402(c)) the U.S. Environmental Protection Agency to conduct a study of lead exposure associated with
renovation and remodeling activities (R&R Study). This report documents Phase II of that study. Phase III is a
retrospective study of the association between elevated blood lead (EBL) concentrations (2lO//g/dL) in children and the
incidence of R&R activities in their residences.
A subset of children were selected for participation from the Wisconsin Bureau of Public Health's blood lead registry.
Information on R&R activities in participating children's residences was obtained by telephone interview. The study
included 3,654 children. Logistic regression was used to determine if incidence of general or specific R&R activities in
residences was associated with increased risk of an EBL.
Based on statistical analysis, children residing in residences in which some R&R activity was conducted were estimated
to have odds of an EBL 1.3 times greater than children residing in residences where no R&R was conducted. When paint
removal using a heat gun was performed, the odds of an EBL was 4.6 times greater than if the work was not performed.
Other significant associations were found.
17. Document Analysis
a. Descriptors: Blood-lead concentration, childhood health risks, lead-based paint, lead hazard, renovation and remodeling,
elevated blood-lead concentration
b. Identifiers/Open-Ended Terms: Section 402, Renovation and Remodeling Study
c. COSATI Field/Group
18. Availability Statement
Release Unlimited
19. Security Class (This Report)
Unclassified
20. Security Class (This Page)
Unclassified
21. No. of Pages
156
22. Price
(See ANSI-239.18)
OPTIONAL FORM 272 (4-77)
(Formerly NTIS-35)
Department of Commerce
------- |