Kemper AR, Cohn LM, Fant KE, Dombkowski KJ, Hudson SR. Follow-up Testing Among Children With Elevated Screening Blood Lead Levels. JAMA. 2005;293(18):2232-2237. doi:10.1001/jama.293.18.2232
Author Affiliations: Child Health Evaluation
and Research Unit, Division of General Pediatrics, University of Michigan,
Ann Arbor (Drs Kemper and Dombkowski and Mss Cohn and Fant); and Childhood
Lead Poisoning Prevention Program, Michigan Department of Community Health,
Lansing (Ms Hudson).
Context Follow-up testing after an abnormal screening blood lead level is a
key component of lead poisoning prevention.
Objectives To measure the proportion of children with elevated screening lead levels
who have follow-up testing and to determine factors associated with such care.
Design, Setting, and Participants Retrospective, observational cohort study of 3682 Michigan Medicaid-enrolled
children aged 6 years or younger who had a screening blood lead level of at
least 10 μg/dL (0.48 μmol/L) between January 1, 2002, and June 30, 2003.
Main Outcome Measure Testing within 180 days of an elevated screening lead level.
Results Follow-up testing was received by 53.9% (95% confidence interval [CI],
52.2%-55.5%) of the children. In multivariate analysis adjusting for age,
screening blood lead level results, and local health department catchment
area, the relative risk of follow-up testing was lower for Hispanic or nonwhite
children than for white children (0.91; 95% CI, 0.87-0.94), for children living
in urban compared with rural areas (0.92; 95% CI, 0.89-0.96), and for children
living in high- compared with low-risk lead areas (0.94; 95% CI, 0.92-0.96).
Among children who did not have follow-up testing, 58.6% (95% CI, 56.3%-61.0%)
had at least 1 medical encounter in the 6-month period after the elevated
screening blood lead level, including encounters for evaluation and management
(39.3%; 95% CI, 36.9%-41.6%) or preventive care (13.2%; 95% CI, 11.6%-14.8%).
Conclusions The rate of follow-up testing after an abnormal screening blood lead
level was low, and children with increased likelihood of lead poisoning were
less likely to receive follow-up testing. At least half of the children had
a missed opportunity for follow-up testing. The observed disparities of care
may increase the burden of cognitive impairment among at-risk children.
In 1997, the Centers for Disease Control and Prevention (CDC) changed
the recommendation for childhood lead poisoning prevention from near-universal
testing of all children to targeted testing based on the risk of lead exposure.1,2 This change was motivated by the decrease
in the prevalence of lead poisoning because of the success of primary prevention
strategies, such as the removal of lead from paint and gasoline.3 The
CDC directed states to develop plans for lead testing according to local risk.1 Testing was also recommended for children according
to the results from a standardized risk-assessment questionnaire and for those
enrolled in public-assistance programs (eg, Medicaid; the Supplemental Food
Program for Women, Infants, and Children).1
An expected benefit of switching to risk-based lead testing, also referred
to as lead screening, was to allow greater health care resources to be directed
to individuals at greatest risk for lead poisoning. In 1997 and again in 2002,
the CDC outlined the role of child health care providers after an elevated
screening blood lead level (≥10 μg/dL [0.48 μmol/L]): all elevated
screening blood lead test results require diagnostic confirmation, and because
capillary sampling has been associated with false elevations, only venous
blood should be used for confirmation.1,4 The
urgency for confirmatory testing varies according to the initial level, from
3 months for levels 10 to 19 μg/dL (0.48-0.82 μmol/L) to emergently
for children with levels of at least 70 μg/dL (3.38 μmol/L). Once the
level is confirmed, repeated testing is recommended, with a frequency ranging
from as soon as possible for those with levels of at least 45 μg/dL (2.17 μmol/L)
to 3 months for those with levels from 10 to 14 μg/dL (0.48-0.67 μmol/L)
to ensure that the blood lead level is not increasing and, if applicable,
is responding to intervention.4 Of note, any
repeated testing that occurs after a 6-month break is considered to be a screening
test, regardless of the previous lead level, and would therefore require confirmatory
testing and subsequent repeated testing as necessary.
Lead poisoning prevention is a collaborative effort between primary
care clinicians and public health agencies. Primary care clinicians should
ensure that children are appropriately screened for lead poisoning as part
of routine preventive care and then receive follow-up testing and care as
necessary. State and local public health departments provide and coordinate
services for children identified with lead poisoning (eg, environmental investigation,
lead abatement). In some communities, public health departments also offer
blood lead testing, usually for children who do not have a regular source
of medical care.
Because of the harm of even modest elevations in blood lead level,5 significant efforts have been made to improve screening
among at-risk children. However, screening is effective only with appropriate
follow-up care. No previous population-based study has evaluated the care
that children receive after having an elevated screening blood lead level.
To begin to understand the care provided to children after an elevated
screening blood lead level, we chose to focus on one component of care: follow-up
blood lead testing. We based our study in Michigan because this state has
a reporting mechanism for all blood lead levels, regardless of result, and
compared with other states, Michigan has a high number of children with lead
poisoning.6 We chose to study Medicaid-enrolled
children because they are at high risk for lead poisoning7 and
because demographic and health care use data are available for these children.
We performed a retrospective cohort study of children aged 6 years and
younger who had an elevated blood lead level (≥10 μg/dL [0.48 μmol/L])
between January 1, 2002, and June 30, 2003, in Michigan and who were continuously
enrolled in Michigan Medicaid during the 180-day period after the elevated
blood lead level. Because we were interested in newly identified cases of
lead poisoning, we excluded children who had an elevated blood lead level
reported in 2001.
For each child, we identified the first elevated blood lead level during
this 18-month period. We then identified any other blood lead testing during
the subsequent 180 days. We chose 180 days because blood lead testing after
a 6-month break is considered to be a new screening test and because follow-up
blood testing, regardless of the initial blood lead level, should occur earlier.1 All medical encounters during this 180-day period
were identified to determine missed opportunities for follow-up testing.
This study was approved by the University of Michigan Medical School
institutional review board, which waived informed consent for this retrospective
Demographic, enrollment, and encounter data were obtained from Medicaid
program administrative files and were linked to blood lead results collected
by the Michigan Department of Community Health (MDCH). Each laboratory in
Michigan has been required since 1997 to report all blood lead results to
the MDCH. The laboratories supply identifying information about each individual
tested (eg, name, address, birth date, Medicaid number), collection date,
blood lead level result, and the method of specimen sampling (eg, venous,
capillary). These data are entered into an electronic file that is subsequently
linked through a complex algorithm to other data sets maintained by the state,
including the Medicaid program files.
Bull Services conducted an internal study in 2002 commissioned by MDCH
that found the linkage process across all data sets to be more than 99% accurate,
with 0.4% false matches and 0.3% false nonmatches (written communication,
Tom Rothan, June 2004. This study was undertaken to test the accuracy of the
match for purposes of overall calibration of the Unique Client Identifier
system. Bull Services Inc believes the study was accurate for that purpose.
It was based on a sampling of data and reflected the data sets at the time
of the study . The results of the study were not intended as a guarantee
or warranty of accuracy for any selected matching process using Unique Client
Identifier at that time or in the future).
The main outcome measures of this study were the proportion of children
who had at least 1 follow-up test during the 180 days after an elevated screening
blood lead level and the number of missed opportunities among those children
who did not have any other follow-up testing. We determined missed opportunities
by using claims data, classifying encounter types according to Current Procedural Terminology code.8 Medical
encounters were classified as visits for evaluation and management (99201-5,
99211-5, 99354-5), preventive care (99381-3, 99391-3), emergency care (99281-5),
consultation (99241-5), and inpatient care (99221-3, 99231-6, 99251-5, 99261-3,
99291-9, 99346-7). We also evaluated the relationship between the screening
blood lead level and the first follow-up test result.
Certain demographic factors are associated with the risk of lead poisoning,
including age, race or ethnicity, urban or rural status, and local risk of
lead exposure.7,9 We hypothesized
that children with increased likelihood of having elevated blood lead levels
(eg, younger children, nonwhite children, children living in urban areas or
in communities with a high risk of lead exposure) would also have a greater
likelihood of follow-up testing after an elevated screening level. We also
hypothesized that there would be differences in follow-up testing rates across
local public health department catchment areas. Although there is variation
in the proportion of children with elevated blood lead levels across the catchment
areas, all local public health departments in Michigan share responsibility
with private practitioners in coordinating services for children with lead
poisoning. Finally, we hypothesized that follow-up would be greater among
children who had an initial capillary sample or who had higher initial blood
In our analysis, we dichotomized race or ethnicity as non-Hispanic white
and Hispanic or nonwhite according to classification by parents on Medicaid
enrollment forms. Address in the calendar year of the screening test was used
to classify urban or rural status, lead-exposure risk, and health department
Urban residence was classified according to metropolitan statistical
areas (MSAs), as defined by the US Census Bureau.10 Each
MSA is formed around an urbanized area of 50 000 or more inhabitants
and includes adjacent communities if they are economically or socially integrated
to the urbanized area. Each MSA is composed of 1 or more counties. In Michigan,
26 of the 83 counties are classified as being in an MSA.
Children were considered to have a high risk of lead exposure according
to Michigan’s targeted screening plan, which categorizes ZIP code areas
by the incidence of lead poisoning, the stock of older houses, and the proportion
of children living in poverty.11 In cases of
incomplete address information, we used the ZIP code from the following or
preceding calendar year in our data set for risk classification. We performed
a sensitivity analysis to test the validity of this assumption by reanalyzing
the data, omitting children with missing ZIP code data.
There are 45 local health departments in Michigan. To evaluate the effect
of health department, we compared the rates of follow-up testing in the 2
local health departments that had the largest number of children with elevated
screening blood lead levels with that of the other local health departments.
Other independent variables were the blood sample type (ie, capillary,
venous) and the value of the screening blood lead level. We categorized blood
lead level to reflect recommended treatment: 10 to 19 μg/dL (0.48-0.92 μmol/L)
(follow-up lead monitoring and education), 20 to 44 μg/dL (0.97-2.13 μmol/L)
(as per lower levels plus environmental investigation and abatement, and neurodevelopmental
monitoring), and at least 45 μg/dL (2.17 μmol/L) (as per lower levels
plus chelation therapy).1,4 Throughout,
to convert blood lead levels to μmol/L, multiply values by 0.0483.
Confidence intervals (CIs) were based on a normal distribution for continuous
variables and on a binomial distribution for categorical variables. We used
3 measures to evaluate the association between each independent variable and
likelihood of follow-up testing: the proportion of children at each level
of the variable that had follow-up testing, the unadjusted relative risk (RR)
of follow-up testing, and the adjusted RR of follow-up testing. Modified Poisson
regression was used to determine the adjusted RRs and their CIs.12 Variables
were also compared with Pearson χ2 test for categorical variables
or t test for continuous variables. Observations
with missing data were excluded from bivariate and regression analyses. All
reported P values and CIs are 2-sided. P<.05 was considered to indicate statistical significance. Stata
8.2 software (StataCorp, College Station, Tex) was used for all analyses.
There were 5175 Medicaid-enrolled children who had a blood lead level
of at least 10 μg/dL (0.48 μmol/L) between January 1, 2002, and June
30, 2003. Of these, 3682 (71.2%) did not have an elevated blood lead level
during 2001 and were therefore included in this analysis.
The demographic characteristics of these children are listed in Table 1. For all but 148 of the children (96.0%),
we used ZIP code data from the calendar year of the screening test. For the
remainder, we used ZIP code data for the other year during the study period.
One- and 2-year-old children accounted for slightly more than half of
the children with elevated blood lead levels. Most children were Hispanic
or nonwhite, lived in urban areas, and had high risk of lead exposure.
Race and ethnicity and risk of lead exposure were clustered by urban
or rural residence. Compared with rural areas, urban areas had a greater proportion
of Hispanic or nonwhite children (88.8% vs 21.0%; P<.001)
and a greater proportion of children with high risk of lead exposure (96.1%
vs 84.9%; P<.001).
Most of the children lived within districts served by either of 2 local
public health departments, both serving urban areas but on opposite sides
of the state. One served the area in which 67.0% of the children lived, and
the other served the area in which 14.0% of the children lived.
The screening test was based on a capillary sample for 1543 (41.9%)
of the children, a venous sample for 2138 (58.1%) of the children, and unknown
for 1 child. The mean blood lead level did not vary according to blood sample
type (capillary, 14.7 μg/dL; venous, 14.4 μg/dL; P = .11). Table 2 lists
the categorized distribution of blood lead levels stratified by blood sample
type; differences in the distribution were not statistically significant (P = .39).
Overall, 53.9% (95% CI, 52.2%-55.5%) had follow-up testing within 180
days of their elevated blood lead screening test, with a mean of 68.5 days
(95% CI, 66.3-70.6 days). The mean number of days before the first follow-up
test was shorter for capillary (51.5 days; 95% CI, 48.5-54.4 days) than for
venous screening tests (83.7 days; 95% CI, 80.8-86.6 days) and for higher
screening blood lead levels (10-19 μg/dL: 73.2 days [95% CI, 70.8-75.6
days]; 20-44 μg/dL: 49.2 days [95% CI, 44.4-54.1 days]; ≥45 μg/dL:
10.0 days [95% CI, 5.9-14.0 days]).
Most follow-up tests were done with venous samples (n = 1789;
90.2%), including 88.4% of the screening tests that used capillary samples
(n = 829). Mean follow-up blood lead levels were 3.6 μg/dL (95%
CI, 3.0-4.2 μg/dL) lower than the screening blood lead level. The mean
change was greater for capillary (6.6 μg/dL; 95% CI, 6.0-7.1 μg/dL)
compared with venous screening tests (3.0 μg/dL; 95% CI, 2.6-3.3 μg/dL).
On follow-up testing, 47.5% (95% CI, 45.2%-50.0%) of the children still
had elevated blood lead levels. Children with screening tests using venous
blood compared with capillary blood were more likely to have an elevated lead
level on follow-up testing (60.1% vs 33.4%; P<.001).
Regardless of blood sample type, higher screening levels were associated with
a greater likelihood of an elevated follow-up blood lead level (10-14 μg/dL:
32.8%; 15-19 μg/dL: 64.6%; ≥20 μg/dL: 77.8%; P<.001).
Table 1 lists the proportion,
unadjusted RR, and adjusted RR of follow-up testing by each of the independent
variables. Although higher screening levels were associated with increased
rates of follow-up testing, not all children in the highest category, at least
45 μg/dL, had follow-up testing. Children who had screening with capillary
blood or who had a higher screening blood lead level had a greater likelihood
of follow-up testing.
The likelihood of follow-up testing decreased with increasing age after
2 years (P<.001). The likelihood of follow-up
testing was lower for Hispanic or nonwhite children (P<.001),
for children with urban residence (P<.001), and
for children with high lead-exposure risk (P = .003).
Children living within the area served by the first local public health department
had a lower likelihood of follow-up testing than those served by other health
departments (P<.001). In contrast, children served
by the second local public health department had a greater likelihood of follow-up
testing than other health departments (P<.001)
(Table 1). The association between follow-up
and these demographic factors persisted after multivariate adjustment.
Omitting cases with missing ZIP code data in the year of testing had
no significant effect on the overall rate of follow-up testing, the proportion
of children in low- or high-risk areas for lead exposure who had follow-up
testing, the unadjusted risk of follow-up testing by lead-exposure risk, or
any of the adjusted RRs for follow-up testing.
Among individuals who did not have follow-up testing, 58.6% (95% CI,
56.3%-61.0%) had at least 1 medical encounter during the 180 days after the
elevated screening blood lead level, with a mean of 2.3 (95% CI, 2.1-2.4)
encounters among those who had any subsequent encounters. The most common
type of medical encounter was for evaluation and management (39.3%; 95% CI,
36.9%-41.6%); however, 13.2% (95% CI, 11.6%-14.8%) had at least 1 preventive
care visit, and 26.7% (95% CI, 24.6%-28.8%) had an emergency department visit.
Outpatient consultations (2.6%; 95% CI, 1.9%-3.4%) and hospitalizations (2.4%;
95% CI, 1.7%-3.1%) were rare. Among individuals who did not have follow-up
testing, 11.4% (95% CI, 10.0%-12.9%) had an emergency department visit as
their only medical encounter in the 180 days after the initial elevated blood
This is the first population-based study, to our knowledge, of follow-up
after an elevated screening blood lead level. Although we cannot comment on
other interventions that these children may have received for their elevated
blood lead level, follow-up testing is the cornerstone of lead poisoning management
and an essential component of secondary prevention.1,4 We
found that nearly half of the children in this study had no follow-up testing
6 months after an elevated screening blood lead level result. Furthermore,
those children with the greatest risk of lead poisoning according to demographic
factors, including nonwhite children, those living in urban areas or in communities
with a high risk of lead exposure, and those living in the local public health
department catchment area with the greatest number of elevated screening blood
lead levels, were the least likely to have follow-up testing. Multivariate
modeling demonstrated that these effects are independent; the more demographic
risk factors a child had, the less likely the child was to receive follow-up
testing. These findings suggest a lack of connection between federal efforts
to eliminate childhood lead poisoning13 and
current lead screening practices.
The lack of follow-up testing is likely to have a significant clinical
effect. Even modestly elevated blood lead levels have been associated with
intellectual impairment.5 Nearly half of the
individuals with follow-up testing had persistently elevated blood lead levels.
We suspect that the proportion of children with persistently elevated levels
may be even higher in those without follow-up testing because of their greater
risk of lead poisoning. The differential pattern of follow-up testing may
further disadvantage minority children.
Our study has several limitations. We are unable to determine the cause
of the low rate of follow-up testing or its inequitable pattern. Our findings
could be biased by inaccuracies in the Medicaid enrollment files, including
classification of race and ethnicity. We classified children’s residence
according to a single address and did not consider the effect of changing
residences. Our classification of urban or rural status does not allow us
to understand neighborhood-level effects. Finally, we are unable to specify
the site of screening or follow-up testing.
Under the current system, primary care providers are responsible for
follow-up testing as part of the care provided within the medical home, with
local health departments primarily coordinating treatment for children with
confirmed lead poisoning. Loss of medical follow-up does not itself account
for the low rate of follow-up testing. More than half of the children with
no follow-up testing had medical encounters in the 6 months after their elevated
screening blood lead level result. However, at least 10% of these encounters
were outside of the primary care setting, where there may be no knowledge
of the elevated screening level and follow-up lead testing is unlikely to
occur. To minimize loss to follow-up because of poor information sharing,
New York City has recently integrated blood lead test results into their immunization
registry.14 A similar approach has been proposed
Information-related barriers are unlikely to solely account for the
observed disparities. We suspect that elevated screening blood lead levels
in children perceived to be at low risk may attract extra attention. In contrast,
care may be less aggressive in high-risk populations if lead poisoning is
not considered unusual or if resources for optimal care (eg, environmental
investigation, lead abatement) are insufficient. Inadequate guideline adherence
is not unique to childhood lead poisoning prevention.16- 18 Future
research is needed to understand the specific barriers to optimal care for
children with elevated screening blood lead levels and to clearly define the
responsibilities of public and private health care practitioners.
Childhood lead poisoning is common, affecting 2% of US children aged
1 through 5 years.6 Furthermore, Medicaid-enrolled
children have a 3-fold greater risk.7 Current
federal plans call for the elimination of childhood lead poisoning by 2010,13 primarily through secondary prevention.1,4 In
this first population-based study of the outcomes of screening, we found that
half of Medicaid-enrolled children with an elevated blood lead level have
no follow-up testing, and those children at greatest risk of having an elevated
blood lead level are less likely to receive follow-up testing. Because each
state handles lead poisoning prevention differently, we do not know whether
these results are generalizable to other states. We hope that our findings
lead other states to perform similar assessments. To maximize cognitive development
in these children, it is crucial to improve follow-up and to understand and
develop interventions to overcome these unexpected disparities in care.
Corresponding Author: Alex R. Kemper, MD,
MPH, MS, 6E18 300 N Ingalls Bldg, Ann Arbor, MI 48109-0456 (firstname.lastname@example.org).
Author Contributions: Dr Kemper had full access
to all of the data in the study and takes responsibility for the integrity
of the data and the accuracy of the data analysis.
Study concept and design: Kemper, Cohn, Fant,
Acquisition of data: Kemper, Cohn, Dombkowski.
Analysis and interpretation of data: Kemper,
Cohn, Fant, Dombkowski, Hudson.
Drafting of the manuscript: Kemper.
Critical revision of the manuscript for important
intellectual content: Cohn, Fant, Dombkowski, Hudson.
Statistical analysis: Kemper, Dombkowski.
Obtained funding: Kemper.
Administrative, technical, or material support:
Financial Disclosures: None reported.
Funding/Support: This work was funded by the
Michigan Department of Community Health.
Role of the Sponsor: The Michigan Department
of Community Health participated in the design of this project, provided the
data, and reviewed and approved the manuscript.