Lead Poisoning: Presentation to Conference on Health Disparities

1
Lead Poisoning Presentation to Conference on
Health Disparities

• Stan Kaplowitz
• Professor of Sociology
• MSU

2
Objectives

• 1) To improve the identification of children and
  residential areas
• that are of highest BLL risk and should be a top
  
• priority for
• Blood Lead Level testing and
• medical and environmental intervention
• 2) To develop methods that are generalizable and
  can be readily used by medical and public health
  professionals
• using characteristics of the childs family and
  their neighborhood

3
Possible Approaches

• Various methods have been used to identify those
  at high risk for EBLL, with varying degrees of
  accuracy
• Universal testing, under current reimbursement
  mechanisms, is cost-prohibitive and a difficult
  sell to physicians and
• does not find lead poisoning until it has already
  developed

4
My Approach vs. Bob Scotts

• The presentation by Bob Scott focused on a visual
  picture of lead poisoning
• I will give a numerical view looking at data from
  the entire state of MI (or at least many areas of
  it) and developing a prediction equation

5
CDC Risk Assessment Questions (1997)

• Does the child now, or in the recent past, live
  in or often visit a house built before 1950 with
  peeling or chipping paint? This could include a
  day care, preschool, or home of a relative.
• Does the child now, or in the recent past, live
  in or often visit a house built before 1978 that
  has been remodeled within the last year?
• Does the child have a brother or sister (or
  playmate) with lead poisoning? This instrument
  (and other similar ones) has been found to
  have only modest predictive value.

6
MDCH Screening Method (1998)
 

• Contains the three CDC questions and adds
• Does the child live with an adult whose job or
  hobby involves lead? (list of jobs/hobbies
  provided) and
• Does the childs family use any home remedies
  that may contain lead? (list of remedies
  provided)
• Recommended testing of children for lead
  poisoning if the caregiver answers "yes" or "I
  don't know" to any of the five questions above
  OR
• Medicaid-enrolled
• OR
• Lives in a high BLL risk ZIP code

7
High Risk ZIP Code

• One that meets any of the following criteria
• 12 or greater incidence of lead poisoning among
  children ages 12 to 36 months in 2000
• 27 or greater pre-1950s housing
• a combination of percentage of pre-1950's
  housing and percentage of children under age six
  living in poverty
• In Michigan, half of all ZIP codes were called
  high BLL risk by MDCH
• including many fairly affluent ZIP codes

8
Purposes of this Research Program
     

• Perform a quantitative evaluation of the
  predictive value of the MDCH risk assessment
  method
• Create a new method that
• has greater predictive validity
• is easy to use
• Received funding from CDC in 2000 to start this
  project

9
Methods

• Examined the relationship of BLL to two kinds of
  data
• self-report questionnaire
• neighborhood socio-demographics from the 2000
  census (e.g., data on race, income, age of
  housing within the census block group)

10

Methods Self-report questionnaire

• n 3376 for whom we have a good address with
  block group and who are between 8.5 months and 72
  months (the recommended ages for BLL testing
• collected in 30 pediatric clinics in Michigan
  from October 2001 through December 2002 see map
  for geography based on the MDCH questionnaire,
  but it was changed in two ways
• 1) added some questions, in the hope they would
  prove to be useful predictors of BLL

11
2) simplified the sentence structure

• Old Does the child now, or in the recent past,
  live in or often visit a house built before 1950
  with peeling or chipping paint? This could
  include a day care, preschool, or home of a
  relative.
• Our survey Please think about the current place
  that the child lives and any previous place that
  the child has lived. If the child has lived in
  more than one place before the current residence,
  please use as previous place the previous place
  that the child lived in longest. Please circle,
  or fill in, the response which fits each place.
• Did it have peeling or chipping paint?
• Please think about any house, day care center, or
  preschool that the child has regularly visited.
  Did any of them have peeling or chipping paint?

12
(No Transcript)
13
Methods

• Measuring Socio-demographic Predictors of BLL
• each residence address was geo-coded with the
  census 2000 block group
• from the census, socio-demographic
  characteristics of residents were associated with
  each block group
• in analyzing data we logarithmically transform
  BLL to meet necessary statistical assumptions of
  regression

14
Results Survey Data

15
Comments

• Clearly, a method based largely on block groups
  and characteristics is more predictive than the
  old methods because
• A block group is much smaller and more
  homogeneous than a ZIP code
• We are treating the risk in an area as continuous
  not dichotomous

16
Comments (Cont.)

• 3) The low R-square can be explained by the
  following
• those tested are mostly well above average risk
  in the state
• children have not spent all of their time in and
  around their current block group

17
The Predicted Value of Ln(BLL -.5)

• Blockgroup Score .414(SIB_EBLL)
  .192(PLYMT_EBLL) .066(ADT_EBLL)
• .091(ANYLDPIP) .262(INPTPRE50HP)
• -.079( INCOME PRE50HP) .117(MEDICAID)
  .206(BLACK) .
• Notes
• 1) coefficients in bold are different from 0
  at p lt.001.
• 2) All independent variable above range from 0
  to 1 except for INCOME, whose range is 0 to 6.
• 3) A coefficient of .414 means that a
  one unit increase in this variable multiplies
  expected BLL by 1.51. A coefficient of .206 means
  that one unit increase in this variable
  multiplies expected BLL by 1.24.

18
Summary

• Controlling for all other predictors, each of the
  following indicates increased risk
• Being Black (African-American)
• Having a sibling with Elevated BLL
• Being eligible for Medicaid insurance
• Living in a house with peeling paint
• Lower family income
• (for both of the above esp. if most houses in the
  block group were built before 1950)
• Having an adult family member or playmate with
  elevated BLL
• Drinking water from lead pipes

19
MDCH Database of BLL Tests (19982001)

• contains the residence address of the child,
  Medicaid status, and race
• analysis was restricted to the test results
  analyzed of those children who were one year old
  at the time tested (one test per child)
• After removing cases with missing data on
  Medicaid Status or Race, we were left with more
  than 45,000 cases

20
Prediction Equation

• Ln (BLL-.5) .019 .326Black .322Medicaid
  1.098OLPVGM
• .556PRE50HP .489EDUC12ND 184RENTPCT
  -.468POVPCT . 198BLACKPCT .365LATINPCT
  .165INNRCITY -.484(MedicaidOLPVGM)
• This is the version developed in 2004 and being
  updated

21
Comments

• The most important single predictor of a
  neighborhoods BLL risk is
• OLPVGM ?( PRE50HP POVPCT)
• R-sq is approx 30. Such a figure is quite good
  for social science data predicting an outcome.

22
Using the MSU-MDCH Risk Assessment Method

• Web site http//midata.msu.edu/bll/
• Web-based risk assessment tool asks user to
  enter
• the childs address
• second address, if the child has lived at more
  
• than one address
• answers to the individuallevel questions

23
(No Transcript)
24
(No Transcript)
25
The MSU-MDCH Risk Assessment Web Site

• The site outputs
• an estimate of the childs probability of EBLL
• the probability of EBLL in the total MI data
• a recommendation as to whether a BLL test is
  needed and information about BLL risk factors

26
Future Work Toward Improvements in Prediction

• Validating Previous findings on BLL data from
  2002-2005
• Effects of Lead Emissions in the Soil on BLL
• EPA website allows one to search multiple
  environmental databases and obtain a list of
  facilities that emitted lead including the name,
  street address with ZIP code, amount of and kinds
  of emissions (into air, land, drinking water, or
  waste stream). (This data was collected in prior
  years).
• The analysis of those cases that are on Medicaid
  is funded by a Medicaid Match Grant.

27

• Measuring distances from motor vehicle emissions
• maps of traffic volume on major roads in 1975
  were used to identify a set of high traffic roads
  and intersections
• distance from the residence group to the nearest
  road or intersection
• volume of traffic
• This data was also collected in prior years

28

• These data allow us to compute the proximity of
  residence to
• high traffic roads (relevant from the days of
  leaded gasoline)
• industrial emissions containing lead
• We are now taking into account wind patterns at
  each area to predict the amount of lead in the
  soil from such emissions in each area.

29
Additional Data to be Linked to BLL Soon and
Improve Prediction

• Linking MDCH Vital Records data base with the BLL
  data base
• obtaining data on education, ethnic ancestry,
  and birth country of the childs parents
• Linking BLL data base to data on type of service
  lines (lead vs. non-lead)
• Start with Lansing and Grand Rapids and see how
  much of an effect the type of lines has on BLL

30
Later Possibilities

• In many cities, the tax assessor database
  provides, for many residences, information useful
  for predicting BLL risk
• a) year in which the structure was built
• b) whether it is owned or rented
• Paint highest in lead content was the most
  expensive.
• We shall locate those areas once inhabited by the
  wealthy and are now inhabited by the poor
• To do so, we plan to use the National Historic
  Geographic Information Systems Project