April, 1996
EFFECT OF IN-HOME EDUCATIONAL INTERVENTION
      ON CHILDREN'S BLOOD LEAD LEVELS
               IN MILWAUKEE
         TECHNICAL REPORT
           Technical Programs Branch
       Chemical Management Division (7404)
      Office of Pollution Prevention and Toxics
       U.S. Environmental Protection Agency
              401 M Street, S.W.
            Washington, D.C. 20460

-------
                                DISCLAIMER

      The material in this document has been subject to Agency technical and policy
review and approved for publication as an EPA report. Mention of trade names, products,
or services does not convey, and should not  be interpreted as conveying, official  EPA
approval, endorsement, or recommendation.

-------
                      CONTRIBUTING ORGANIZATIONS

      The study described in  this report was conducted by the U.S. Environmental
Protection Agency  (EPA) and its  contractor QuanTech  and the Milwaukee Health
Department.  The Milwaukee Health Department provided the data and EPA and its
contractor entered the data into a database, analyzed the data, and produced the report.

                                 QuanTech

      Quantech (formerly David  C. Cox & Associates) provided technical assistance
regarding the data management, and was responsible for the statistical analysis, and for
the overall production of the report.

                    U.S. Environmental Protection Agency

      The U.S. Environmental Protection Agency (EPA) funded the analysis of the data
and was  responsible for managing the study, for reviewing study documents,  and for
arranging for the peer review of the final report.  The EPA Project Leader was  Bradley
Schultz.  The EPA Work Assignment Manager and Project Officer was Samuel Brown.
Cindy Stroup and Barbara Leczynski provided valuable assistance.  Janet Remmers,  Dan
Reinhart,  Phil Robinson, and Ben  Lim also provided useful comments.

                        Milwaukee Health Department

      The study could not have been done without the assistance and cooperation of the
Milwaukee Health Department. Major contributors included Amy Murphy, MaryJoGerlach,
Kris White, and Sue Shepeard.

-------
IV

-------
                                  Table of Contents


EXECUTIVE SUMMARY	  xi

1      DESCRIPTION OF STUDY 	  1
      1.1    OBJECTIVE	  1
      1.2    BACKGROUND	  1
      1.3    DATA 	  3
            1.3.1   The Outreach Study Group	  4
            1.3.2   The Outreach Reference Group	  6
            1.3.3   Adjustment of Data for Age and Seasonality Effects	  6
      1.4    PEER REVIEW	  6

2     SUMMARY OF ANALYSIS AND RESULTS  	  9
      2.1    COMPARISONS BETWEEN THE STUDY AND REFERENCE GROUPS  	  9
      2.2    DEFINING THE REFERENCE GROUP 	  16

3 CONCLUSIONS	  25

REFERENCES	  27

APPENDIX A. DATABASE DEVELOPMENT	A-1

APPENDIX B. DATA ADJUSTMENT PROCESS	B-1

APPENDIX C. COST OF IN-HOME EDUCATIONAL VISITS IN MILWAUKEE	C-1

-------
VI

-------
                                          List of Tables
Table 1.       Number of Children by Age Group  	  4
Table 2.       Number of Children by Race/Ethnicity in the Study and Reference groups	  4
Tables.       Number of Initial and Followup Measurements forthe Study and Reference Groups by Time
              Period	  5
Table 4.       Summary Statistics for Changes in Adjusted Blood Lead Levels (ug/dl) for the Study Group
              and Reference Group by Measurement Type and Gender  	  11
Table 5.       Summary Statistics for Adjusted Initial Mean Blood Lead Levels (ug/dl) forthe Study Group
              and Reference Group by Measurement Type and Gender  	  12
Table 6.       Summary Statistics forChanges in Unadjusted Blood Lead Levels (ug/dl) forthe Study and
              Reference Group by Measurement Type and Gender	  13
Table 7.       Summary Statistics for Unadjusted Initial Mean Blood Lead  Levels (ug/dl) for the Study
              Group and Reference Group by Measurement Type and Gender	  14
Table 8.       Analyses of Variance for Changes in Blood Lead Levels Due to Intervention, Gender, and
              Measurement Type 	  15
Table 9.       Summary of Major Results 	  15
Table 10.      Comparison of Summary Statistics for Changes in Unadjusted Blood Lead Levels (ug/dl)
              for Children  who did  not Receive an Outreach Visit but  had Elevated Blood  Lead
              Measurements between 1984-89, 1990-September, 1991, and October, 1991 to 1994	  18
Table 11.      Analyses of Variance for Changes in Blood Lead Levels Due to Intervention, Gender, and
              Measurement Type 	B-4
Table 12.      Summary Statistics forChanges in Blood Lead Levels Adjusted Using Alternative Method
               	  B-5
Table 13.      Summary Statistics for Initial Blood  Lead Levels Adjusted Using Alternative Method	  B-6
                                                VII

-------
VIM

-------
                                           List of Figures
Figure 1.       Frequency Distribution forChanges in Adjusted Blood Lead Levels (ug/dl) of Children in the
              Study of Outreach Intervention  	  9
Figure 2.       Frequency distribution for Male Children: Changes in Adjusted Blood Lead Levels (ug/dl).
               	  19
Figure 3.       Frequency  Distribution for Female  Children: Changes in Adjusted  Blood  Lead Levels
              (ug/dl)	  20
Figure 4.       Outreach Study Cases: Change in Adjusted Blood Lead Levels (ug/dl) by Time Between
              Measurements	  21
Figures.       Reference Group Cases (1990-93): Change in Adjusted Blood Lead Levels (ug/dl) by Time
              Between Measurements	  22
Figure 6.       Outreach Study Cases: Change in Adjusted Blood Lead Levels (ug/dl) by Age	  23
Figure 7.       Reference Group Cases (1990-93): Change in Adjusted Blood Lead Levels (ug/dl) by Age.
               	  24
                                                 IX

-------

-------
                          EXECUTIVE SUMMARY
BACKGROUND
   Education and counseling are relatively inexpensive components of some programsfor
reducing blood lead  levels in  children.  However, these measures have not been
conclusively demonstrated to be effective. The purpose of this study was to determine
whether blood lead levels declined after in-home educational visits. The interventions were
conducted from 1991 to 1993 by regular Milwaukee Health Department staff who went to
homes of children with elevated (20-24 ug/dl) blood lead levels. The in-home educational
visits described hazards associated with childhood lead exposure, and potential sources
of the hazards in the home were identified. The importance of the child's personal hygiene,
the child's nutrition, and overall dust reduction and cleaning practices was also discussed.
The  visits   lasted  about  one hour and  were  performed  by health department
paraprofessionals.  The in-home educational visits are part  of a program designed to
reduce children's lead exposure in Milwaukee where widespread blood lead testing
identifies children with elevated  blood lead levels.  Outreach workers and/or other public
health officials currently attempt to contact each family with children having elevated blood
lead levels.

METHODS

   Data was compiled retrospectively from the Milwaukee Health Department records of
blood lead measurements collected through the blood lead testing program. The analysis
was based on a comparison of changes in blood lead levels for a study group of children
who received the outreach educational interventions versus a reference group of children
who did not receive the educational interventions. Children who moved or whose blood
lead levels may have been affected by a lead paint abatement were eliminated from the
study.  The study group includes all other children who received outreach interventions
between 1991 and 1993, had at least one blood lead measurement between 20-24 ug/dl
before the intervention, and at least one measurement after the intervention. Similarly, the
reference  group  includes children who from 1990 to  early 1994  had at  least one
measurement between 20-24 ug/dl and a followup measurement. Comparisons of the
children whose families received an educational visit with a reference group was important,
because changes  in average  blood lead  level measurements may be caused by
phenomena unrelated to educational intervention.
RESULTS

   Average blood lead levels, adjusted for seasonality and age of the children  in the
Milwaukee outreach intervention program, were about 21% lower after intervention than
before intervention. Blood lead levels in the reference group of non-recipients of outreach
                                      XI

-------
visits also declined, but by about 6%. This difference was statistically significant at a p-
value less than 0.001.  The difference in the average declines in blood lead levels yielded
an estimate of the net effectiveness of outreach educational intervention of 21% - 6% =
15% (with a 95% confidence interval of 8% to 23%). Effectiveness of the educational
intervention did  not depend significantly on a child's age or sex.

DISCUSSION

   The retrospective comparison shows that in-home educational visits may have resulted
in reducing children's  blood lead levels by about  15% more than for a reference group
without interventions.  The validity of this conclusion depends upon whether children who
received the visits were comparable to reference group children whose families were  often
unavailable for outreach visits. Families that were unavailable for outreach visits may have
been more likely to exhibit behavior patterns responsible for the continued elevation of their
children's blood lead levels. Nevertheless, an examination of available data on blood lead
levels and demographics indicated that the study and reference groups were similar and
were comparable for the purposes of determining the beneficial effects of the outreach
educational program.  Educational efforts at doctor's offices and clinics may also  have
contributed to reduced blood  lead levels in both groups, thus having little effect on the net
reduction.

   Total costs of the outreach educational visits were estimated to be  in the range of  $100
per visit.  Blood lead  levels of the  children studied were usually still elevated after the
educational intervention alone.  However,  important declines were observed. Educational
intervention appears to be a useful and inexpensive component of lead exposure reduction
programs.
                                       XII

-------
1  DESCRIPTION OF STUDY
1.1    OBJECTIVE

   Changes in blood lead levels (PbB) following outreach interventions were investigated
using data available from the Milwaukee Health Department on PbB levels through July,
1994. Outreach involves educational visits by Milwaukee Health Department staff to the
homes of children with elevated blood lead levels, usually in the 20-24 ug/dl range.  The
purpose of the analysis was to determine whether blood lead levels declined after these
educational interventions, calculate the magnitude of the change, and identify factors
related to any change.
1.2   BACKGROUND

   There have been previous indications that education and counseling may be effective
components of programs for reducing blood  lead levels in children.  The main focus of
counseling efforts would be to encourage housecleaning, improve personal hygiene and
diet, and discourage hand-to-mouth  behavior.  An early study (Charney, et, al, 1983)
demonstrated that  periodic  wet mopping of rooms to remove dust lead can result in
declines in blood lead levels. However, no study conclusively showed that educational
efforts have been effective (see USEPA, 1995). A study that specifically addressed the
issue of the effectiveness of education was based on data from children living near a lead
smelter in Granite City, Illinois in 1991  (Kimbrough, et, al). After education and counseling
of households with children with slightly elevated blood lead levels, arithmetic mean blood
lead levels decreased from 15.0 to 7.8 ug/dl.  However, since there was no control group
for comparison, the decreases in the blood lead  levels could be at least partially attributed
to other factors such as seasonality and age.

   The data for this study of educational intervention effects is a result of a widespread
blood lead testing program targeting all Milwaukee children between six months and seven
years old.  The program identifies children with  elevated blood  lead levels, so steps can
be taken to reduce lead exposure and lead-related health  impacts. Under the current
program, families of all children are notified of the results and encouraged to obtain
another measurement within a year.  Measurements above 24 ug/dl trigger attempts to
schedule visits by a public health nurse and/or  an Environmental Health  Inspector.  If a
measurement is  from  20-24 ug/dl  (and no previous measurement exceeds  24 ug/dl),
attempts are made  to arrange a visit by an outreach worker.

   An outreach worker is a  public health paraprofessional who received about one full
week of training  followed by an approximately eight-week apprenticeship.  Outreach
workers typically  make only one visit to  a family. During the visit, the outreach worker:

-------
   1) instructs the family on the significance of blood lead measurements;
   2) verifies sources of medical care and encourages followup measurements;
   3) teaches the family about the hazards of lead poisoning, its prevention and control;
   4) teaches the family about  identifying sources  of  lead in  the  home  and other
      environments where the child spends significant time;
   5) identifies  siblings at risk for blood lead poisoning whose blood lead levels should
      also be tested;
   6) surveys the immediate environment for sources of lead exposure;
   7) demonstrates simple and appropriate clean-up measures for reducing obvious lead
      exposures (such as washing and rinsing window sills and taping cardboard over
      cracks in  the wall);
   8) makes note of obvious evidence of environmental lead hazards to be inspected by
      an Environmental Health Inspector if deemed appropriate; and
   9) documents the outreach worker's activities.

From an informal analysis reproduced in Appendix C, total costs of the program have been
estimated to result in average costs of about $100 per outreach visit.

   Outreach interventions were phased  in from 1991 through the first half of 1992.
Outreach was intended as a supplement to an intervention program that already included
visits from Public Health Nurses and Environmental Lead Inspectors to families of children
with blood lead  measurements  exceeding 24 ug/dl. By the summer of 1992, outreach
visits were attempted for all families with children with at least one measurement from 20-
24 ug/dl and no  previous measurement greater than 24 ug/dl.

   The analysis of the Milwaukee Health Department data is based on a retrospective
comparison of changes in pairs of blood  lead  levels for a study group of children who
received the outreach educational interventions versus a reference group of children who
did not receive the educational interventions. Children who moved or whose  blood lead
levels may have been affected by a lead paint abatement, a public health nurse visit, or a
change of address were excluded from the study.  The study group  includes all other
children who received  outreach interventions between 1991 and 1993 and had at least one
blood lead measurement between 20-24  ug/dl before the intervention and at least one
measurement after the intervention.  Similarly, the reference group includes children who
from 1990 to early 1994 had at least one  measurement between 20-24 ug/dl and a
followup measurement.

   A comparison between study and reference groups is important, because "regression
to the means" and time-related trends unrelated to the intervention may result in decreases
(or increases) in observed blood lead levels after intervention. "Regression to the means"
is the phenomenon that first measurements are followed by second measurements which
on average are  closer to the mean.  The phenomenon occurs because high  blood lead
measurements may be an indication of measurement error and short-term fluctuations, as
well as actual high sustained  blood lead levels.  Measurements may tend to be lower after

-------
an intervention, regardless of whether the intervention is truly effective, because only
children with elevated blood lead measurements receive interventions.

   The validity of the comparison of blood lead levels between the study and reference
groups depends on the assumption that the two groups of children were similar with
respect to factors related to blood lead levels. Table 1 demonstrates the similarity of the
age distributions for the two groups of children.  Table 2 shows that the two groups of
children were similar with respect to categories based on race and ethnicity. As Table 3
shows, children in the reference group were generally tested earlier than  children who
received outreach visits. Seventy of the reference group children had blood lead tests
before the start of the outreach program. Families of many of the remaining 156 children
in the reference group could not be contacted after three attempts.  The main analysis
used all 226 children in the reference group tested from 1990 to early 1994. A sensitivity
analysis then tested to see how results would be affected if the reference group had been
restricted to children with  initial  measurements after September, 1991.   Differences
between the reference group and the study group in: 1) the timing of the measurements
and 2) behavioral characteristics (indicated by the apparent refusal of outreach visits to
families of children in the reference group) must be taken into account to properly interpret
the study's results. Nevertheless, the comparison between the study and reference groups
in conjunction with the sensitivity analysis allowed for a relatively unbiased assessment of
the effect of the outreach interventions.
1.3   DATA

   Two sets of data of pairs of childrens' blood lead measurements were created for the
analysis of outreach effectiveness, one for the "study" group and one for the "reference"
group. Blood lead measurements were either based on capillary or venous blood samples.
Venous measurements are thought to be more reliable, since they are based on larger
blood samples, and are thus less sensitive to contamination.  Also, fingers are probably
more likely to be contaminated by lead dust.  For this reason, the comparison between
groups is considered separately for venous and capillary samples.

   Families of children in both groups may have received educational information at their
doctor's office or clinic where the blood sample was taken. Thus, the comparison between
the study children and reference children is designed to measure the additional benefit
attributable to the outreach visits. For both study and reference databases, an attempt was
made to exclude pairs of measurements for children whose blood lead levels may have
changed  between  measurements because of other events  not directly  related to the
outreach intervention. In particular, measurements were excluded for children who moved
between measurement dates, or if an abatement or a follow-up visit by a public health
nurse occurred before the second measurement.

-------
Table 1.    Number of Children by Age Group1.
AGE (years)
0.5 to 1.5
1.5 to 2. 5
2.5 to 3.5
3.5 to 4.5
4. 5 to 5. 5
5. 5 to 7.0
TOTAL
STUDY
12(6%)
46 (25%)
39(21%)
44 (24%)
29(16%)
17(9%)
187
REFERENCE
25 (1 1 %)
56 (25%)
50 (22%)
47(21%)
32(14%)
16(7%)
226
1Age at the time of the initial measurement.
Table 2.    Number of Children by Race/Ethnicity in the Study and Reference groups.
RACE/ETHNICITY
AFRICAN-AMERICAN
ASIAN
NATIVE AMERICAN
WHITE HISPANIC
WHITE NON-HISPANIC
OTHER
UNKNOWN
TOTAL
STUDY
128(68%)
4 ( 2%)
1 (1%)
20 (1 1 %)
13(7%)
2(1%)
19(10%)
187
REFERENCE
102(45%)
6 ( 3%)
0 ( 0%)
18(8%)
10(4%)
9 ( 4%)
81 (36%)
226
  The creation of data sets for both groups is described in greater detail in the next two
sections and Appendix A.
1.3.1  The Outreach Study Group
   Data for the study group includes the last pre-intervention and first post-intervention
blood lead measurements from 187 out of 365 children identified as

-------
Table 3.   Number of Initial and Followup Measurements for the Study and Reference Groups by Time
         Period.
YEAR/QUARTER
1 990/1 stQ
2ndQ
3rdQ
4thQ
1991/1 stQ
2ndQ
3rdQ
4thQ
1992/1 stQ
2ndQ
3rdQ
4thQ
1993/1 stQ
2ndQ
3rdQ
4thQ
1994/1 stQ
2ndQ
TOTAL
STUDY
PRE
INTERVENTION







9
19
32
62
21
17
18
7
2


187
POST
INTERVENTION








4
4
19
25
41
42
29
13
8
2
187
REFERENCE
MEASUREMENT
INITIAL
8
8
8
7
12
8
19
26
17
34
44
19
13
2
1



226
FOLLOWUP
1
3
5
1
4
8
10
11
23
23
43
30
33
23
8



226
receiving outreach visits between 1991  and 1993.  Reasons for excluding the 365-
187=178 children are as follows: 8 had  pre-intervention measurements above 24 ug/dl,
10 had pre-intervention measurements  that never exceeded 20 ug/dl, 29 moved, 25
had post-outreach intervention measurements but none before the outreach visit, 98
had pre-outreach intervention measurements but none after, and 8 were not between 6
months and 7 years old.

-------
1.3.2  The Outreach Reference Group

   Data for the reference group of 226 children includes pairs of consecutive
measurements from children who did not receive outreach visits. A pair of
measurements was eligible for the reference database if: 1) neither measurement
occurred before a nursing intervention, or abatement; 2) the first of the two
measurements was between 20-24 ug/dl; and 3) the first measurement was taken
between 1990 and 1993, and 4) both measurements were taken while the child was
between 6 months and 7 years old.

   Each child's first two measurements (in chronological order) for the same address
from the data set described in  Appendix A were used to create the reference group
database.  328 of the children  had at least two measurements with the first
measurement falling between  20 and 24 ug/dl and no intervention before the second
measurement date. Eighteen  cases with an abatement (removal of lead based paint)
occurring between the measurement dates and 1  case with an invalid recorded date of
measurement were excluded.  In 236 of the remaining cases, both measurements were
after December 31, 1989.  Of these 236 cases, 226 children were between  six months
and seven years old at the time of both measurements.
1.3.3  Adjustment of Data for Age and Seasonality Effects

   1990-93 blood lead level measurements collected by the Milwaukee Health
Department from 13,746 children were shown in an analysis by Pawel, et al (1995) to
depend on age and season.  Results from the analysis were used to adjust the 1990-
1994 blood lead levels in both the reference and study databases for seasonality and
age effects. The adjustments were designed to produce "equivalent" PbB
measurements that would have been obtained in January at age two years. The
seasonality and age effect adjustments were based on data from the same location and
a similar time period as the study and reference group data. Otherwise, the
assessment of outreach intervention effects could have been seriously biased.
Seasonality, for example, may differ substantially by location and time, because
seasonality may depend on factors such as climate (see McCusker,  1979), age
(Cooney, et al, 1989; Baghurst, et al, 1992), and race.  Portables of the adjustment
factors and additional  details, see Pawel, et al, 1995.
1.4   PEER REVIEW

   This study was reviewed independently by members of a peer review panel.
Comments which are important for interpreting the study results or which had an
important impact on the report are discussed below.

-------
   Many of the comments from the reviewers requested additional information on the
characteristics of the study and reference groups. In response, descriptive information
about the groups was added. Exclusion criteria had been used to define the groups in
a way that the two groups would be comparable. Tables were added to the report to
show that the study and reference group children were similar with respect to
categories based on age, race and ethnicity. Some reviewers were concerned that
much of the reference group data was collected before the start of the outreach
program (October, 1991).  The report now includes a sensitivity analysis which showed
that excluding measurements before October, 1991  would not have substantially
changed the results of the study. Section 2.2 was added to provide a detailed
discussion of these and related issues.

   A comment was made that the report did not present data showing that the outreach
visits induced families to make changes in self-protective behavior.  The changes in
behavior would then have resulted in changes in blood lead levels.  Unfortunately, since
the outreach workers would typically make only one visit to a family, data on behavior
changes was unavailable. However, the study did find an association between
conducting the visits and declines  in blood lead levels. Since the study and reference
groups were comparable, changes in  blood lead  levels most likely would have been
similar without the outreach interventions. The difference in the changes in blood lead
levels most likely was due to the information imparted through the outreach
interventions.

   EPA has established a public record for the peer review under administrative record
151.  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  SUMMARY OF ANALYSIS AND RESULTS
2.1    COMPARISONS BETWEEN THE STUDY AND REFERENCE GROUPS

   Figure 1 illustrates greater decreases in adjusted blood lead levels for the 187 children
who received outreach visits than for the 226 children in the reference group. As described
later in this section, an analysis of variance confirmed that the declines in adjusted PbB
values were significantly greater for the study than the reference group.  Except when
noted, the analysis is based upon the adjusted blood lead values.
       Percent of
        Children
            30
            25
            20-
            15-
            10
1
                                                   I
                  -21  -18  -15 -12  -9  -6   -3
                                           12   15  18  21
                                     Children Receiving Outreach Visit, n = 187
                                    Reference Group (No outreach visit), n=226
Figure 1.   Frequency Distribution for Changes in Adjusted Blood Lead Levels (ug/dl) of Children in the
         Study of Outreach Intervention
   Overall changes in mean blood lead values by measurement type and gender for both
the study and reference groups are summarized in Table 4.  Initial measurements are

-------
summarized in Table 5.  Major results are summarized in Table 9.  For children who
received an outreach intervention, average post-intervention PbB measurements were 4.24
ug/dl, or about 21% less than pre-intervention measurements.  The average pre-
intervention measurement was about 20 ug/dl.  The  corresponding 95% confidence
intervals for declines in PbB levels are 3.3 ug/dl to 5.2  ug/dl  and 16% to 26%. For the
reference group, the second measurement was on average 1.18 ug/dl or about 6% less
than the first measurement. The corresponding 95% confidence intervals for children who
did not receive an outreach intervention are 0.2 ug/dl to 2.2 ug/dl and 1% to 10%. The
estimated decrease in PbB attributable to outreach intervention is 3.06 ug/dl = 4.24 ug/dl -
1.18 ug/dl or about a 15% decline from pre-intervention levels.  The corresponding 95%
confidence  intervals  are 1.6 ug/dl to 4.5 ug/dl  and 8% to 23%.

   An  analysis of unadjusted  data would have yielded essentially identical results.
Summary statistics for the unadjusted data are shown in Tables 6 and 7. The average
decrease in the unadjusted means was 4.12 ug/dl for the study  group and 1.24 ug/dl for
the reference group.  The estimated decrease in PbB attributable to outreach intervention
would have been  2.88 ug/dl = 4.12 ug/dl - 1.24  ug/dl, which represents about a 14%
decline from pre-intervention levels.  The corresponding 95%  confidence intervals would
have been 1.5 ug/dl  to 4.3 ug/dl and 7% to 20%.

   The four measurement type codes indicate the method of measurement, capillary or
venous, for the pairs of PbB measurements. Except for the Venous-Capillary permutation
(with only a total of 20 measurements), average net differences between decreases  in
(adjusted) PbB for the reference and study groups were very consistent, between 2.88
ug/dl and 3.65 ug/dl. For both the reference and study groups, declines in PbB were not
statistically  significantly different between males and females.

   Results from the analyses of variance (ANOVA) in Table 8 showed that declines in PbB
were significantly  greater for the study  group than for the  reference group (p<.001);
differences  in  the declines did  not differ significantly by sex or measurement type.
Frequency distributions for changes between measurements in PbB are shown separately
for males and females in Figures 2 and 3.

   The last two rows of Tables 4 and 5 indicate how the results would have changed if the
reference group had been restricted to children with initial measurements after September,
1991. The  estimated decrease attributable to the educational  intervention would have
been somewhat larger, 3.95 ug/dl, since the average change in PbB for children in the
"restricted"  reference group was only about -0.3ug/dl. Nevertheless, the  ANOVA results
would have been essentially unchanged. Declines  in PbB would still not have differed
significantly by sex or blood lead measurement type.

   The relationship between changes in  PbB  and the time between  measurements was
also examined. The downward sloping regression lines shown in Figures 4 and 5 suggest
slightly larger decreases in PbB values between measurements when the time
                                      10

-------
Table 4.    Summary Statistics for Changes in Adjusted Blood Lead Levels (ug/dl) for the Study Group and Reference Group by
           Measurement Type and Gender.
GROUP
STUDY
REFERENCE2
NET2
REFERENCE
From 10/913
NET3
STATISTIC
Mean
S. Deviation
Sample Size
Mean
S. Deviation
Sample Size
Mean
S. Error
Mean
S. Deviation
Sample Size
Mean
S. Error
MEASUREMENT TYPE1
CC
-5.06
5.33
(37)
-1.58
6.63
(49)
-3.48
1.29
-1.13
6.78
(34)
-3.94
1.46
CV
-4.73
5.38
(29)
-1.08
7.40
(21)
-3.65
1.90
-1.69
7.79
(18)
-3.04
2.09
W
-3.86
6.11
(65)
-0.98
6.39
(71)
-2.88
1.07
-1.69
6.70
(54)
-2.17
1.19
VC
-4.59
10.17
(10)
1.88
6.79
(10)
-6.47
3.87
2.97
6.20
(9)
-7.56
3.82
MISSING
-3.72
6.88
(46)
-1.55
8.03
(75)
-2.17
1.37
2.15
6.32
(41)
-5.87
1.42
GENDER
MALE
-3.71
5.80
(107)
0.37
6.54
(123)
-4.08
0.81
0.69
6.81
(95)
-4.40
0.90
FEMALE
-4.97
6.89
(79)
-2.78
7.28
(90)
-2.19
1.09
-1.81
6.76
(61)
-3.16
1.16
MISSING
-2.37
(D
-4.75
7.78
(13)
2.38
8.07
N/A
(0)
N/A
OVERALL
-4.24
6.28
(187)
-1.18
7.11
(226)
-3.06
0.66
-0.29
6.88
(156)
-3.95
0.72
1CC = Both measurements are capillary; CV = First is capillary, second is venous;
VC = First is venous, second is capillary; W = Both are venous.
MISSING = Type of at least one measurement is unknown.
2Reference group includes 226 children; Net = Difference between study and reference group.
3Excludes reference group measurements before 10/91.

-------
       Table 5.    Summary Statistics for Adjusted Initial Mean Blood Lead Levels (ug/dl) for the Study Group and Reference Group by
                  Measurement Type and Gender.

GROUP


STUDY


REFERENCE

REFERENCE
From 10/912


STATISTIC

Mean
S. Deviation
Sample Size
Mean
S. Deviation
Sample Size
Mean
S. Deviation
Sample Size
MEASUREMENT TYPE1

CC
19.2
3.33
(37)
21.3
3.89
(49)
21.7
3.16
(34)

CV
19.6
3.19
(29)
21.2
3.66
(21)
21.3
3.93
(18)

MM
20.2
2.97
(65)
21.0
3.24
(71)
21.9
2.93
(54)

VC
19.5
7.14
(10)
22.7
3.94
(10)
22.7
4.17
(9)

MISSING
20.8
4.90
(46)
21.0
3.78
(75)
20.8
2.99
(41)
GENDER

MALE
19.8
3.78
(107)
21.3
3.52
(123)
21.7
3.37
(95)

FEMALE
20.3
4.12
(79)
20.8
3.38
(90)
21.2
2.90
(61)

MISSING
22.8

(D
22.6
5.67
(13)
N/A
(0)

OVERALL

20.0
3.92
(187)
21.2
3.63
(226)
21.5
3.20
(156)
IV)
       1CC = Both measurements are capillary; CV = First is capillary, second is venous;
       VC = First is venous, second is capillary; W = Both are venous.
       MISSING = Type of one or both of the measurements unknown.
       2Excludes reference group measurements before 10/91.

-------
Table 6.    Summary Statistics for Changes in Unadjusted Blood Lead Levels (ug/dl) for the Study and Reference Group by
           Measurement Type and Gender.
GROUP
STUDY
REFERENCE2
NET
REFERENCE
From 10/913
NET
STATISTIC
Mean
S. Deviation
Sample Size
Mean
S. Deviation
Sample Size
Mean
S. Error
Mean
S. Deviation
Sample Size
Mean
S. Error
MEASUREMENT TYPE1
CC
-4.51
5.06
(37)
-1.41
7.95
(49)
-3.10
1.41
-0.71
7.45
(34)
-3.80
1.52
CV
-5.28
5.74
(29)
-1.38
7.27
(21)
-3.90
1.91
-1.39
7.50
(18)
-3.89
2.06
W
-3.82
5.82
(65)
-1.48
6.39
(71)
-2.34
1.03
-1.22
6.28
(54)
-2.60
1.12
VC
-2.70
10.85
(10)
2.10
9.88
(10)
-4.80
4.64
2.78
10.23
(9)
-5.48
4.84
MISSING
-3.80
6.17
(46)
-1.32
7.01
(75)
-2.48
1.22
1.24
6.00
(41)
-5.04
1.31
GENDER
MALE
-4.07
5.95
(107)
0.04
6.82
(123)
-4.11
0.84
0.84
6.88
(95)
-4.91
0.91
FEMALE
-4.16
6.31
(79)
-2.92
7.28
(90)
-1.24
1.05
-1.95
6.67
(61)
-2.21
1.11
MISSING
-5.00
(D
-1.77
6.36
(13)
-3.23
6.60
N/A
(0)
N/A
OVERALL
-4.12
6.07
(187)
-1.24
7.10
(226)
-2.88
0.65
-0.25
6.91
(156)
-3.87
0.71
1CC = Both measurements are capillary; CV = First is capillary, second is venous;
VC = First is venous, second is capillary; W = Both are venous.
MISSING = Type of at least one measurement is unknown.
2Reference group includes 226 children; Net = Difference between study and reference group.
3Excludes reference group measurements before 10/91.

-------
Table 7.    Summary Statistics for Unadjusted Initial Mean Blood Lead Levels (ug/dl) for the Study Group and Reference Group by
           Measurement Type and Gender.

GROUP


STUDY


REFERENCE

REFERENCE
From 10/912


STATISTIC

Mean
S. Deviation
Sample Size
Mean
S. Deviation
Sample Size
Mean
S. Deviation
Sample Size
MEASUREMENT TYPE1

CC
20.1
2.79
(37)
22.1
1.38
(49)
22.0
1.45
(34)

CV
21.4
1.72
(29)
22.0
1.32
(21)
21.8
1.22
(18)

MM
21.3
2.09
(65)
21.8
1.48
(71)
21.9
1.41
(54)

VC
19.6
5.08
(10)
22.3
1.49
(10)
22.2
1.56
(9)

MISSING
21.3
2.78
(46)
21.5
1.24
(75)
21.6
1.18
(41)
GENDER

MALE
21.1
2.38
(107)
21.8
1.40
(123)
21.8
1.38
(95)

FEMALE
20.9
2.94
(79)
21.9
1.36
(90)
21.9
1.29
(61)

MISSING
24.0

(D
21.2
1.17
(13)
N/A
(0)

OVERALL

21.0
2.63
(187)
21.8
1.38
(226)
21.8
1.34
(156)
1CC = Both measurements are capillary; CV = First is capillary, second is venous;
VC = First is venous, second is capillary; W = Both are venous.
MISSING = Type of one or both of the measurements unknown.
2Excludes reference group measurements before 10/91.

-------
Table 8.    Analyses of Variance for Changes in Blood Lead Levels Due to Intervention, Gender, and
           Measurement Type1.
SOURCE
Intervention2
Sex3
Sex*lntervention4
First Measurement Type5
Second Measurement Type6
DF
1
1
1
1
1
TYPE I SS
838.2
131.3
13.7
73.2
0.1
F-VALUE
20.5
3.2
0.3
1.8
0.0
PR>F
< 0.001
0.08
0.57
0.19
0.97
Observations with missing data for sex and measurement type are excluded.
2Row entries show whether changes in PbB are different for intervention vs. reference groups.
3Row entries show whether changes in PbB (for both intervention and reference groups) differ by sex.
Interaction between sex and intervention type. Shows whether intervention effectiveness differs by sex.
5First measurement is capillary or venous.
6Second measurement is capillary or venous.
Table 9.    Summary of Major Results.  The 95% Confidence Intervals are Enclosed in Brackets.
DATA
Adjusted
Data
Unadjusted
Data
GROUP
Study
Reference
Difference1
Study
Reference
Difference1
DECLINE IN PBB
MEASUREMENTS
[jg/dl
4.24
[3.3, 5.2]
1.18
[0.2, 2.2]
3.06
[1.6,4.5]
4.12
[3.2, 5.0]
1.24
[0.3, 2.2]
2.88
[1.5,4.3]
AVERAGE INITIAL
MEASUREMENT MQ/dl
20.0
21.2
N/A
21.0
21.8
N/A
PERCENT
DECLINE
21
[16,26]
6
[1,10]
15
[8, 23]
20
[15,24]
6
[1,10]
14
[7, 20]
1The difference between the Study group and the Reference group.
                                             15

-------
between measurements is increased. The slopes of the regression lines, -0.20 ug/dl per
month for the study group versus -0.35 ug/dl per month for the reference group were not
significantly different.

   The fit to a model (excluding observations from children with unrecorded sex and age)
that describes the relationship between the change in the adjusted blood lead levels and
the time between measurements reasonably well is:

(1)    Y  = -0.00 - .25*t  for the reference group
            = -2.58 - .25*t  for the study group,
    where t = time between measurements in months, and
         Y  = second measurement (ug/dl) - first measurement (ug/dl).

In the model used to obtain equation 1,  unlike the models used in figures 4 and 5, the
slopes of the equations were assumed to be equal.  Thus, the rate of decrease in blood
lead (in this case estimated to be 0.25 ug/dl per month) were assumed to be the same for
both the reference and study groups. Equation 1 suggests that the outreach interventions
may have caused an initial relatively abrupt  drop in average PbB levels of 2.58  ug/dl,
followed by a decline of about 0.25 ug/dl per month.  An approximate 95% confidence
interval for the initial drop in PbB levels is 1.0 to 4.2 ug/dl. This result is very similar to the
estimates of intervention effectiveness derived from the summary statistics in Table 4.

   Equation 1 is used to obtain only a very approximate characterization of how blood lead
levels decrease after an outreach intervention.  The data contains too much variation to
determine how rapidly or extensively blood lead  levels drop within the first few weeks after
an intervention. The regression estimates are also greatly influenced  by pairs of PbB
measurements with  large time differences between measurements.  The average time
between measurements was about 6.7 months the study group and 6.1 months for the
reference groups.  For several children the time between measurements was greater than
two years.

   Figures 6 and 7 show that there is no substantial relationship between decreases in
PbB and age.  The slopes of the regression  lines (-.22 and .25 ug/dl per month for the
study  and reference groups) are not significantly different from 0.
2.2   DEFINING THE REFERENCE GROUP

   The analysis summarized in the previous section was based upon the assumption that
prior to the visits of the outreach workers, the study and reference groups were similar with
respect to factors associated with blood lead levels. Exclusion criteria for the timing of the
measurements in the reference group were defined to satisfy this assumption. This section
discusses reasons for defining the reference group to include children with initial elevated
(20-24 UQ/dl) measurements from 1990-March, 1994.  The discussion  compares this
choice to two other groups defined by time intervals starting from a) 1984, the beginning

                                       16

-------
of computerized Milwaukee Health Department  records of blood lead  levels, and b)
October, 1991, the beginning of the outreach program.

   A simple choice for a reference  group would have been to  include children with
measurements from 1984.  However, data collected before 1990  was excluded in part
because the  pre-1990  data was representative of only the early participants in the
Milwaukee blood lead testing  program.  These early participants formed only a  small
minority of the children who lived in Milwaukee during the 1980's. The children tended to
live in sectors of the city targeted by the city's program to reduce lead exposure, because
these sectors were thought to  have high concentrations of children with elevated  blood
lead levels. Also, there were concerns that the pre-1990 measurements were far more
variable than measurements from the 1990's.

   Another obvious approach would have been to define the reference group to include
only children  with pairs of measurements  after  the start of the  outreach program in
October, 1991. Unfortunately, many of the families of these later reference group children
were simply unavailable for an outreach visit.  The groups would not be directly comparable
if non-availability was an indicator of family behavior patterns associated with the continued
elevation of children's blood lead levels.

   Defining the reference group to include only children with measurements from 1990
balanced concerns discussed in the previous two paragraphs.  Certainly, the inclusion of
children with measurements before the start of the outreach program raised concerns that
the analysis could have been adversely affected by changing time trends in blood lead
levels.  However, it seemed unlikely  that the time trends would have changed enough
within two years to have had much of an effect.  Also, note that many of the 70 reference
group children (see Table 10) with elevated measurements  before October, 1991 (in
contrast to the other reference group children) would have received an outreach visit if the
program had been already established.

   Table 10 compares changes in unadjusted blood lead levels by time period for children
who did not receive outreach visits. (Measurements before 1990 could not be adjusted for
seasonality). The paired measurements increased before 1990, and decreased during the
early 1990's. At first glance, Table 10 seems to indicate that the blood lead measurements
of children measured after September, 1991 (average change = -0.25 ug/dl)  may have
differed substantively from children measured between 1990 to September, 1991 (average
change = -3.46 ug/dl). However,  much of the difference was attributable to the 41 +34=74
pairs of measurements of unknown (MISSING) measurement type.  In fact, an analysis
(which  accounted for the  possible  effects  of measurement type) of the  remaining
36+115=151 changes in blood lead measurements found no significant (a=.05) difference
in changes in blood  lead levels between the two time periods.  Thus, data in Table 10
indicates that the reference group, which included
                                      17

-------
       Table 10.  Comparison of Summary Statistics for Changes in Unadjusted Blood Lead Levels (ug/dl) for Children who did not Receive
                 an Outreach Visit but had Elevated Blood Lead Measurements between 1984-89, 1990-September, 1991, and October,
                 1991 to 1994.
GROUP
1984-1989
REFERENCE
(1990-9/1991)
REFERENCE
(10/1991-1994)
STATISTIC
Mean
S. Deviation
Sample Size
Mean
S. Deviation
Sample Size
Mean
S. Deviation
Sample Size
MEASUREMENT TYPE1
CC
4.40
5.13
(5)
-3.00
9.06
(15)
-0.71
7.45
(34)
CV
2.33
8.39
(3)
-1.33
6.43
(3)
-1.39
7.50
(18)
W
12.00
11.14
(3)
-2.29
5.83
(17)
-1.22
6.28
(54)
VC
1.00
(1)
-4.00
(D
2.78
10.23
(9)
MISSING
2.44
8.16
(59)
-4.41
6.96
(34)
1.24
6.00
(41)
GENDER
MALE
3.29
6.44
(21)
-2.68
5.98
(28)
0.84
6.88
(95)
FEMALE
4.59
10.34
(17)
-4.97
8.17
(29)
-1.95
6.67
(61)
MISSING
1.91
7.92
(33)
-1.77
6.37
(13)
N/A
(0)
OVERALL
2.96
8.14
(71)
-3.46
7.06
(70)
-0.25
6.91
(156)
oo
       1CC = Both measurements are capillary; CV = First is capillary, second is venous;
       VC = First is venous, second is capillary; W = Both are venous.
       MISSING = Type of one or both of the measurements unknown.

-------
        Percent of
         Children
             30 1
             25-
             20
              15
              10
                    I    Y\W
I
                    -21  -18  -15 -12  -9   -6   -3
          12   15   18  21
                                         Children Receiving Outreach Visit, n=107
                                        Reference Group (No outreach visit), n=123
Figure 2.   Frequency distribution for Male Children: Changes in Adjusted Blood Lead Levels (ug/dl).


measurements from 1990, allowed for a reasonable assessment of the effects of outreach
intervention.  The data also did not provide strong evidence for the existence of rapidly
changing time trends  in actual blood lead levels.
                                           19

-------
        Percent of
          Children
              30 H
               25-
               20-
               15-
               10-
                    J
E_
1
1
I    I   
                     -21  -18  -15  -12  -9   -6   -3   0
                                                 12   15   18   21
                                             Children Receiving Outreach Visit, n= 79
                                            Reference Group (No outreach visit), n= 90
Figure 3.    Frequency Distribution for Female Children: Changes in Adjusted Blood Lead Levels
            (ug/dl).
                                              20

-------
         Change in PbB
          Levels (ug/dl)
                  30 H
                  20-
                  10-
                   o-
                  -10-
                  -20-
                  -30 4
                                     10      15      20      25      30

                                     Time Between Measurements (in months)
35
40
Figure 4.   Outreach Study Cases: Change in Adjusted Blood Lead Levels (ug/dl) by Time
           Between Measurements.
                                           21

-------
         Change in PbB
          Levels (ug/dl)
                  30 H
                  20-
                  10-
                   o-
                  -10-
                  -20-
                  -30 4
                                     10      15      20      25      30

                                     Time Between Measurements (in months)
35
40
Figure 5.   Reference Group Cases (1990-93): Change in Adjusted Blood Lead Levels (ug/dl) by
           Time Between Measurements.
                                           22

-------
         Change in PbB
          Levels (ug/dl)
                   30 H
                   20-
                   10-
                    o-
                  -10-
                  -20:
                  -30 4
                                         2345

                                         Age at First Measurement (in years)
Figure 6.   Outreach Study Cases: Change in Adjusted Blood Lead Levels (ug/dl) by Age.
                                             23

-------
         Change in PbB
          Levels (ug/dl)
                   30 H
                   20-
                   10-
                    o-
                  -10-
                  -20:
                  -30 4
                                        2345

                                        Age at First Measurement (in years)
Figure 7.   Reference Group Cases (1990-93): Change in Adjusted Blood Lead Levels (ug/dl) by
           Age.
                                            24

-------
3  CONCLUSIONS

   Declines in blood lead levels were about 15% greaterfor children who received in-home
educational visits than for a reference group of children  without interventions.   A
comparison between study and reference groups is important in any study of intervention
effectiveness  because of unpredictable changes in  average blood lead levels unrelated to
interventions.  In this study, average blood lead level declines were about 21 % for children
who received  home visits, and 6% for children in the reference group. The average time
between measurements was about six  months.   Results from this retrospective study
indicate that the visits were responsible for a marginal decline of 8% to 23% in blood lead
levels.

    The validity of this conclusion depends on whether the children who received outreach
visits were comparable to the  reference group  children  whose  families were often
unavailable for outreach visits.  Families that were unavailable for outreach visits may have
been more likely to exhibit behavior patterns responsible for the continued elevation of their
children's blood  lead levels.    Nevertheless, an examination of available data  on
demographics and blood lead levels indicated the two groups were comparable.  The two
groups had similar age distributions, racial characteristics, average baseline blood lead
levels,  and average times between initial and followup  blood lead measurements.  A
complicating  factor, common  to retrospective studies,  was the considerable loss of
followup. About  one half of the  children who received an outreach visit did not have a
followup measurement.

   Blood lead measurements were adjusted to control for effects of seasonality and age,
but the seasonality and  age adjustments had very little effect on  the  final results.
Nevertheless, seasonality and age effects may be important factors in analyses  of the
effectiveness  of  other  interventions,  especially when  pre-  and  post-intervention
measurements occur in different seasons of the year, or there are large time lags between
measurements.

   Outreach visits are relatively inexpensive, with total health department costs estimated
to be in the range of $100 per visit.  The results of this  study indicate that educational
interventions  can be a useful  and  inexpensive component of lead exposure reduction
programs.
                                       25

-------
26

-------
                                REFERENCES
Baghurst P, long S, McMichael A, Roberson E, Wigg N, Vimpani G (1992). "Determinants
of blood lead concentrations to age 5 years in a birth cohort study of children living in the
lead smelting city of Port Pirie and surrounding areas." Archives of Environmental Health,
203-210.

Burgoon D, Rust S, Schultz B (1994). "A summary of studies addressing the efficacy of
lead abatement," in Lead in Paint soil, and Dust: Health Risks, Exposure Studies, Control
Measures, Measurement Methods, and Quality Assurance, ASTM, STP 1226, Michael E.
Beard and S.D. Allen Iske Eds., American Society for Testing and Materials, Philadelphia.

Centers for Disease Control (1993). Stellar User Guide.

Charney E, Kessler B, Farfel M, Jackson  D (1983).  "Childhood Lead  Poisoning: A
Controlled Trial of the Effect of Dust-Control Measures on  Blood Lead  Levels." New
England Journal of Medicine, 309:1089-1093.

Cooney GH, Bell A, McBride W, Carter C (1989). "Low-level exposures to lead: the Sydney
Lead Study." Developmental Medicine and Child Neurology, 31:640-649.

Kimbrough R, LeVois M, Webb D (1994). "Management of children with slightly elevated
blood lead levels." Pediatrics, 93: 188-191.

McCusker J (1979). "Longitudinal changes  in  blood lead level in children and their
relationship to season,  age and exposure to paint or plaster." American Journal of Public
Health. 69:348-52.

Pawel D, Foster C, Cox D (1995). "Seasonal Trends in Blood  Lead Levels  in Milwaukee:
Statistical Methodology." Report to the Environmental Protection Agency, 401 MSt., S.W.,
Washington, D.C.

United States Environmental Protection Agency (1995). "Review of Studies Addressing
Lead Abatement Effectiveness." Report EPA 747-R-95-006 (available by calling 1-800-
424-LEAD).
                                      27

-------

-------
APPENDIX A. DATABASE DEVELOPMENT
              A-1

-------
A-2

-------
   The Milwaukee Health Department uses a software package, "STELLAR", as the lead
case tracking system. This system includes four modules: case, address, investigation,
and lab. The case data file contains information pertaining to a child's name, sex, race,
guardian, number of siblings, and status relating to blood  lead levels.  The address
identification file contains the address identifier, street address, county, city, zip code, and
census tract. The investigation file contains information on the address such as year of
construction, inspection date, abatement date, etc. The lab file contains information about
childrens' blood lead levels, the sampling date, the sampling type, date of birth, providers's
name, etc.  All four data files can be linked by the child identifier and address identifier.
The four Milwaukee data files were used in the analysis of the  in-home educational
intervention.

1.  STELLAR Files

   QuanTech received an original set of four ASCII data files from the Milwaukee Health
Department on  September 23, 1992.  These files, which came from files of the lead case
tracking system included: ADDRESS.BAS,  CASE_RCD.BAS, INVEST.BAS, andLAB.BAS.
Two more sets of updated data files were received from the Health Department in August,
1993 and July, 1994. These four files were converted into corresponding SAS files having
the same name but with an SSD  extension.  The ADDRESS.BAS file contained 32,679
observations and 9 variables; the CASE_RCD. BAS file contained 29,333 observations and
56 variables; and the INVEST.BAS file contained 4,628 observations and 47 variables. All
three LAB.BAS files were concatenated bringing it to a total of 75,084 observations and
26 variables. Among the 138 variables in  the four files, only the 46 variables need for the
analysis were included in  the final database.

2.  Transposing Data

   The  LAB file is a  large file  containing  records for all  reported blood  lead level
measurements.  This file also includes the corresponding date the sample was obtained,
the sample type, the address identifier, etc. Investigating the lead levels for each child over
time is the main reason for setting  up this database, so it was important to have one record
per child per address with the respective  measurements, dates,  and types. (A separate
record was kept for each address since moves are major events that may affect blood lead
levels).  The number of measurements  differs by child.  The LAB file was sorted by
CHILDJD (the unique identifier), ADDRJD (the address identifier), and SMPOBTDT (the
date the sample was taken) and transposed by CHILDJD and ADDRJD. This smaller file
contains 52,719 records with 92  variables: CHILDJD, ADDRJD,  PBB1-PBB30, DT1-
DT30,  and TYP1-TYP30.   For each child, PBB1-PBB30 contains up to 30  lead
measurements, DT1-DT30 contains the corresponding dates, and TYP1-TYP30 contains
the corresponding measurement methods (i.e capillary, venous, or unknown).

   The  file  was  transposed in the following  manner.   Suppose  a child  with
CHILD_ID=10001 had three measurements at the same location on  1/15/92, 2/15/95, and
3/15/95. The original LAB file would contain three records with  CHILD JD=10001, with one

                                     A-3

-------
record  per measurement.   The  transposed  file  would  contain  one record for
CHILD_ID=10001 with three measurements (stored in variables PBB1-PBB3), three dates
(DT1 =1/15/92,  DT2=2/15/92,  and DT3=3/15/92), and  three sample  types (stored in
variables TYP1-TYP3). The variables, PBB4-PBB30, DT4-DT30, and TYP4-TYP30 would
be set to missing. For children with measurements at multiple locations, the transposed
file contains multiple records with the corresponding measurements.  For example, if a
child  (CHILD_ID=20001)  had  three  measurements  at  ADDR_ID=100  and  two
measurements at ADDR_ID=200 then the transposed file would contain two records for
CHILDJD 20001.  For the first record CHILD_ID=20001, ADDR_ID=100 and the three
measurements, dates, and types would be stored in variables PBB1-PBB3, DT1-DT3, and
TYP1-TYP3. The second record would have CHILD_ID=20001 and ADDR_ID=200 with
values for two measurements,  dates, and types.

3.  Merging All Files

   Once the LAB file was transposed into a workable format, cases were examined to
determine whether they should be included in the final database. The  specific cases of
interest are children with at least two blood lead measurements (non-FEP measurements)
and children living in the city of Milwaukee.  Children who lived in other cities  were
excluded from the final database since treatment outside Milwaukee may differ.  Only
cases having  more than one blood lead measurement were kept in the transposed
database.  Cases with only PBB measurements = 0 were deleted. (A zero was recorded
in STELLAR whenever only an FEP  measurement was taken).

   Initial analysis of the data showed that 75 - 100 percent of the data fields in the
CASE_RCD.SSD and INVEST.SSD files had missing values. Therefore, it was necessary
to enter and verify data from paper files.  The Milwaukee Health  Department hard copy
files were used three separate  times (corresponding to the three times data was received
by QuanTech). The SAS database was converted into a working data entry file (a Lotus
spreadsheet) using the Translate Utility of Lotus Version 2.2 software.

4.  Child Address

   An important issue in developing this database is the ability to track changes in the
address of a  child.  Cases of most interest  are those children with more than  one
measurement at one particular address. A child with multiple measurements at more than
one location will appear in the database as many times as the number of locations. All
measurements, dates, and types for each child are verified and  based on the findings,
additional  records are added into the Lotus  file when necessary.   A new variable,
MOVEFLAG, was defined to indicate which children had moved.

5.  Data Verification

   Because about 75% of the electronic database contained some missing values (mostly
from the "INVEST.BAS" file), updating and verifying these records was time consuming.

                                     A-4

-------
The process included tracking each case back to the original hardcopy files that had been
used during data entry.  The final database containing 10,624 cases was divided into
eighteen Lotus files and updated one file at a time.  After the first Lotus file (about 500
records) was updated it was decided to update only those records where an abatement
occurred. The initial set of data received in 1992 was completely verified and updated by
August 1993.  The second set of data received  in 1993 was completely verified and
updated by November 1993. The final set of data received in 1994 was completely verified
and updated by February 1995.

6. Checking for Potentially Erroneous Data

   Several  data anomalies associated with this study warranted further investigation,
including the occurrence of two or three consecutive measurements with identical blood
lead levels or extremely high PbB measurements.

   Several records were found in Stellar that contained two or three consecutive identical
blood lead measurements. These measurements were verified several ways. Files that
contained  consecutive duplicate measurements  (same value and same  date) were
compared  to Public Health Nursing  records.   It  was determined  that duplicate
measurements in the Stellar files with the same sample collection date were data entry
errors, and only one of the entries was retained in  the database.

   Stellar files that contained three consecutive identical measurements were referenced
against the original lab sheets stored at the Milwaukee Health Department. They did not
appear to be data entry errors.

   Blood lead measurements that were abnormally high were also investigated. CHILDJD
11772 had a PbB reported of 201 ug/dl. After checking nursing records it was found that
this was an FEP measurement and not a blood lead value. CHILDJD 433 had a PbB of
97 ug/dl. This capillary sample may have been contaminated.  Capillary measurements
may be affected by contamination when proper aseptic techniques are not followed during
the sample collection.
                                     A-5

-------

-------
APPENDIX B. DATA ADJUSTMENT PROCESS
                B-1

-------
B-2

-------
   Blood levels were adjusted through basically a four step process. First, 90th percentiles
were calculated to summarize PbB levels for each semimonthly period from 1990 through
1993. Second, the 90th percentile PbB values were fit to a model so that long-term trends
in PbB could be removed.  Additive seasonal adjustments were then based on moving
averages of the detrended 90th percentile  PbB values.   Finally, multiplicative age
adjustments were calculated as simple ratios of arithmetic means using predefined age
categories.  The result are adjusted  PbB values which approximate the  PbB values that
would have been observed during the first half of January,  1993 had the children been
1.75-2 years old.

   Adjustments factors were not calculated before 1990, because of concerns about the
quality, quantity and  relevance  of pre-1990 Milwaukee blood  screening data.  Thus,
measurements in the reference group database made before 1990 were not adjusted. All
measurements for the study group were made after 1989 and were adjusted.

   A difficult step of the process was detrending the data, because of uncertainty about
the effect on PbB values of procedural  changes in October, 1991.  To allow for a crude
sensitivity analysis, the data were detrended two ways resulting in two sets of adjusted PbB
values. For the first set of adjusted PbB values (linearly detrended adjusted PbB values),
the 90th percentiles were detrended assuming a simple linear decline in PbB values.
Results in the main body of the report were based upon this set of adjusted PbB values.

   The second set, used mainly to check the sensitivity of analysis of variance results to
the method used to generate the  adjustment factors, allowed for a sudden change in PbB
values occurring in October, 1991. An analysis of variance table and summary statistics
based on the second set of  adjusted  PbB values  or "adjusted PbB values using the
procedural correction" are shown in Tables 11 through 13.
                                      B-3

-------
Table 11.   Analyses of Variance for Changes in Blood Lead Levels Due to Intervention, Gender, and
           Measurement Type1.
ADJUSTMENTS MADE WITH PROCEDURAL CORRECTION
SOURCE
Intervention2
Sex3
Sex*lntervention4
First Measurement Type5
Second Measurement Type6
DF
1
1
1
1
1
TYPE I SS
475.5
128.3
17.2
66.9
0.3
F-VALUE
11.8
3.2
0.4
1.7
0.0
PR>F
< 0.001
0.08
0.51
0.20
0.93
Observations with missing data for sex and measurement type were excluded.
2Row entries show whether changes in PbB are different for intervention vs. reference groups.
3Row entries show whether changes in PbB (for both intervention and reference groups) differ by sex.
Interaction between sex and intervention type.  Shows whether intervention effectiveness differs by sex.
5First measurement is capillary or venous.
6Second measurement is capillary or venous.
                                             B-4

-------
       Table 12.    Summary Statistics for Changes in Blood Lead Levels Adjusted Using Alternative Method.
GROUP
STUDY
REFERENCE
NET
REFERENCE
From 10/91
NET
STATISTIC
Mean
S. Deviation
Sample Size
Mean
S. Deviation
Sample Size
Mean
S. Error
Mean
S. Deviation
Sample Size
Mean
S. Error
MEASUREMENT TYPE2
CC
-4.28
5.26
(37)
-1.78
6.86
(49)
-2.50
1.31
-0.80
6.74
(34)
-3.48
1.44
CV
-3.75
5.21
(29)
-0.81
7.23
(21)
-2.94
1.92
-1.10
7.67
(18)
-2.65
2.05
W
-3.15
6.16
(65)
-1.05
6.07
(71)
-2.10
1.05
-1.15
6.61
(54)
-2.00
1.18
VC
-3.63
10.03
(10)
2.30
6.88
(10)
-5.93
3.85
3.44
6.21
(9)
-7.07
3.79
MISSING
-2.92
6.83
(46)
-1.37
8.30
(75)
-1.55
1.39
2.75
6.34
(41)
-5.67
1.41
GENDER
MALE
-2.97
5.80
(107)
0.42
6.59
(123)
-3.39
0.82
1.21
6.76
(95)
-4.18
0.89
FEMALE
-4.08
6.81
(79)
-2.80
7.34
(90)
-1.28
1.09
-1.31
6.74
(61)
-2.77
1.15
MISSING
-1.66
(1)
-4.47
8.01
(13)
2.81
8.31
N/A
(0)
N/A
OVERALL
-3.44
6.28
(187)
-1.14
7.17
(226)
-2.30
0.66
0.22
6.84
(156)
-3.66
0.71
CD
(In
       'Alternative method is based on use of procedural term.
       2CC = Both measurements are capillary; CV = First is capillary, second is venous;
        VC = First is venous, second is capillary; W = Both are venous.
        MISSING = Type of at least one measurement is unknown.

-------
        Table 13.    Summary Statistics for Initial Blood Lead Levels Adjusted Using Alternative Method.
GROUP
INTERVENTION
REFERENCE
REFERENCE
From 10/91
STATISTIC
Mean
S. Deviation
Sample Size
Mean
S. Deviation
Sample Size
Mean
S. Deviation
Sample Size
MEASUREMENT TYPE2
CC
18.7
3.08
(37)
21.3
3.56
(49)
20.9
3.00
(34)
cv
19.0
3.05
(29)
20.6
3.60
(21)
20.4
3.82
(18)
W
19.9
2.88
(65)
20.7
2.79
(71)
21.0
2.85
(54)
VC
18.8
6.64
(10)
22.2
3.96
(10)
21.9
4.04
(9)
MISSING
20.3
4.81
(46)
21.0
3.68
(75)
20.0
2.99
(41)
GENDER
MALE
19.3
3.71
(107)
21.0
3.30
(123)
20.9
3.20
(95)
FEMALE
19.8
3.90
(79)
20.7
3.14
(90)
20.5
2.87
(61)
MISSING
(1)
22.8
5.30
(13)
N/A
(0)
OVERALL
19.5
3.78
(187)
21.0
3.39
(226)
20.7
3.07
(156)
CD
05
        'Alternative method is based on use of procedural term.
        2C = Both measurements are capillary; CV = First is capillary, second is venous;
        VC = First is venous, second is capillary; VV = Both are venous.
        MISSING = Type of one or both of the measurements unknown.

-------
   APPENDIX C. COST OF IN-HOME EDUCATIONAL VISITS IN MILWAUKEE

(This appendix reproduces a note from Brad Schultz to Amy Murphy which describes
        an informal analysis of the costs of outreach visits in Milwaukee.)
                                  c-1

-------
C-2

-------
Note to: Amy Murphy, Milwaukee Health Department
From: Brad Schultz, U.S. Environmental Protection Agency
Subject: An informal analysis of the costs of MHD outreach visits.

   Based on information from the Milwaukee Health Department, we have come up with
some "ballpark" estimates of the cost per visit of the in-home educational visits by the MHD
outreach workers.  In comparison with what other health departments might do, I  would
place the program development activities and implementation on a per visit basis at a fairly
inexpensive level as the health department is using relatively low-skilled para-professionals
for the  in-home outreach educational visits.

   There are two components of this cost per visit estimate: (1) the cost per year of the
outreach worker; and (2) the numbers of visits that they could make if completely devoted
to making in-home  educational visits.

(1) The cost per year of the outreach worker is approximately $30,000 per year, including
salary,  benefits,  car mileage and cellular telephone.  It is estimated that the outreach
worker requires an  extra 10% of supervisory time for the outreach program, and this cost
is 10% of $60,000, or $6,000 per year.  The total cost of the outreach worker is then
$36,000.

(2) I am estimating  that the number of weeks that the outreach worker could make visits
is 52 weeks per year - (6 weeks of vacation/holiday/sick leave) - (1 week full-time training)
= 45 weeks.  I am also estimating two visits per day on average for 3 days of the week and
one visit for the  other two days.  The  other time is used for unsuccessful attempts at
making educational visits, paperwork in the office, and staff meetings. In this scenario, no
time is allocated for activities other than in-home educational  visits and work directly
related to supporting those visits.

   From those two  pieces, I am estimating roughly that there are  45 weeks of visits, times
8 visits per week, for 360 visits per year. The cost of those visits would be $36,000, thus
resulting in an average cost of $100 per visit. My guess is that it is very possible that the
real cost could be twice this amount or half this amount.

   The number of visits is probably somewhat conservative since  it is difficult to completely
remove other duties from the outreach worker time. But it also  reflects the fact that the
majority of visits by the Milwaukee Health Department are cold calls (no appointment), as
eventually a higher percentage of families are reached this way, and since many  of the
families have no phones to call to make an appointment.  On the other hand, there may
be a tendency  to underestimate costs a little since some costs may be overlooked.  But
overall, I have no reason to suspect that the cost per visit estimate is too high or too low.

   Start up costs are very hard to estimate at this point, but  may have been around
$10,000 for the  Milwaukee Health Department.  This includes purchasing equipment,
                                      c-3

-------
management time to learn about and develop the program, deciding what materials to
hand out, and so forth.
                                     c-4

-------
50272-101
REPORT DOCUMENTATION 1 RE^ART74N7R.95.009
4. Title and Subtitle
EFFECT OF IN-HOME EDUCATIONAL INTERVENTION ON CHILDREN'S
BLOOD LEAD LEVELS IN MILWAUKEE
7. Author(s)
Pawel, D.J; Foster, C.; Cox, D.C.
9. Performing Organization Name and Address
QuanTech, Inc.
1911 North Fort Myer Drive, Suite 1000
Rosslyn, Virginia 22209
12. Sponsoring Organization Name and Address
U.S. Environmental Protection Agency
Office of Pollution Prevention and Toxics
Washington, DC 20460
3. Recipient's Accession No.
5. Report Date
April 1996
6.
8. Performing Organization Rept. No.
10. Project/Task/Work Unit No.
1 1 . Contract (C) or Grant (G) No.
68-D3-0004
13. Type of Report & Period Covered
Technical Report
14.
 15. Supplementary Notes

     In addition to the authors listed above, Jill LeStarge of Quantech was a major contributor to the study.
 16. Abstract (Limit: 200 words)

     Education and counseling have recently been recognized as potentially effective, and relatively inexpensive components
     of programs for reducing blood lead levels in children. The purpose of this retrospective analysis was to determine
     whether blood lead levels declined after in-home educational visits by Milwaukee Health Department staff to homes
     of children with elevated blood lead  levels, usually between 20-24 jig/dl.  During the educational visits, hazards
     associated with childhood lead exposure were described, and potential sources of the hazards were identified.  The
     importance of the  child's personal hygiene, nutrition, and overall dust reduction and cleaning practices were also
     discussed. Data were gathered from 1990 to 1994.

     Average blood lead measurements were about 21% lower after intervention than before intervention. Blood lead levels
     in a reference group of children who did not receive the interventions also declined, but by about 6%. The difference
     in the average declines in blood lead levels yielded an estimate of net effectiveness of outreach intervention of 21%-
     6%=15% with  a 95% confidence interval of 8% to 23%.  The educational interventions appear to be a useful
     component of lead exposure reduction programs.
 17. Document Analysis a. Descriptors
     Lead exposure reduction, children, blood lead levels, educational intervention
 18. Availability Statement
19. Security Class (This Report)
    Unclassified
                                                                       20. Security Class (This Page)
                                                                           Unclassified
21. No. of Pages
        58
                                                                                                    22. Price
(SeeANSI-Z39.18)
               OPTIONAL FORM 272 (4-77)
                      (Formerly NTIS-35)

-------