Objective. Lead poisoning is a well-recognized public health concern for children living in the United States. In 1992, Health Care Financing Administration (HCFA) regulations required lead poisoning risk assessment and blood lead testing for all Medicaid-enrolled children ages 6 months to 6 years. This study estimated the prevalence of blood lead levels (BLLs) ≥10 μg/dL (≥0.48 μmol/L) and the performance of risk assessment questions among children receiving Medicaid services in Alaska.
Design. Measurement of venous BLLs in a statewide sample of children and risk assessment using a questionnaire modified from HCFA sample questions.
Setting. Eight urban areas and 25 rural villages throughout Alaska.
Patients. Nine hundred sixty-seven children enrolled in Medicaid, representing a 6% sample of 6-month- to 6-year-old Alaska children enrolled in Medicaid.
Outcome Measure(s). Determination of BLL and responses to verbal-risk assessment questions.
Results. BLLs ranged from <1 μg/dL (<0.048 μmol/L) to 21 μg/dL (1.01 μmol/L) (median, 2.0 μg/dL or 0.096 μmol/L). The geometric mean BLLs for rural and urban children were 2.2 μg/dL (0.106 μmol/L) and 1.5 μg/dL (0.072 μmol/L), respectively. Six (0.6%) children had a BLL ≥10 μg/dL; only one child had a BLL ≥10 μg/dL (11 μg/dL or 0.53 μmol/L) on retesting. Children whose parents responded positively to at least one risk factor question were more likely to have a BLL ≥10 μg/dL (prevalence ratio = 3.1; 95% confidence interval = 0.4 to 26.6); the predictive value of a positive response was <1%.
Conclusions. In this population, the prevalence of lead exposure was very low (0.6%); only one child tested (0.1%) maintained a BLL ≥10 μg/dL on confirmatory testing; no children were identified who needed individual medical or environmental management for lead exposure. Universal lead screening for Medicaid-enrolled children is not an effective use of public health resources in Alaska. Our findings identify an example of the importance in considering local and regional differences when formulating screening recommendations and regulations, and continually reevaluating the usefulness of federal regulations. lead poisoning, child health services, mass screening, government regulations, Medicaid.
Childhood lead poisoning is a well-recognized public health concern for children living in the United States (US). In 1991, the Centers for Disease Control and Prevention (CDC) lowered the childhood blood-lead level of concern to 10 micrograms per deciliter (μg/dL) (0.48 μmol/L), and recommended a protocol for virtual universal screening of young children.1 In 1992, in response to these recommendations, the Health Care Financing Administration (HCFA) began requiring blood-lead testing for children enrolled in Medicaid.2 The regulation required testing for such children at 6 to 12 months and at 2 years of age, and lead-exposure risk assessment with more frequent testing for those children who were determined by the assessment to be at high risk for exposure to lead. The initial regulation included an exemption for children living in communities that could demonstrate the absence of a childhood lead problem. In 1993, this exemption was removed, requiring Medicaid providers to screen all children enrollees regardless of their potential for exposure.3
Children in the US are not uniformly at risk for lead poisoning. The National Health and Nutrition Examination Surveys III4found that during 1988 through 1991, the prevalence of elevated blood-lead levels (BLLs) varied from as low as 5% among non-Hispanic white children living outside of urban central cities to as high as 35% among non-Hispanic poor black children living in central city areas. Clinic-based studies have estimated prevalences from 0.7%5,6 to more than 50%.7 In this paper, we present results of a systematic, statewide evaluation of BLLs and HCFA risk-assessment questions in Alaska children enrolled in the Medicaid program.
Alaska is geographically the largest state, with approximately one-sixth the land area of the US, and a population density of less than one person per square mile (US average = 70 persons per square mile).8 According to US census data, in 1990, of 550 043 Alaska residents, 41% lived in the city of Anchorage, 28% lived in one of 16 other cities with populations of ≥2500 persons and the remaining 31% were located throughout rural Alaska, mainly in villages inaccessible by road and hundreds of miles from medical facilities.
Many of the sources commonly associated with elevated BLLs in children1 are not frequently found in Alaska, which has a low-population density, a limited road system, and where 52% of public9 and >38% of all housing8 has been built since 1980 (lead paint ceased to be used for residential purposes after 1978). Although the state has maritime industry, a battery factory, and one lead mine (another is located near the border in Canada), routine testing by pediatricians over the past two decades10 and targeted community screening in villages nearest to lead mines11,12 have identified very few children with a BLL ≥10 μg/dL. Targeted screening of children in communities with municipal water systems exceeding Environmental Protection Agency standards for lead13 also found only a very small number of children with a BLL ≥10 μg/dL.
We determined the sampling design using March 1993 state Medicaid data. At that time, of Alaska’s nearly 67 000 children ages 6 months through 6 years, 16 120 children (24%) were enrolled in Medicaid. Enrolled children were those meeting all eligibility requirements who were certified to receive Medicaid services. Because of the difficulty in performing venipuncture for young children, and since BLLs peak at around 2 years of age,1 we offered testing only to children ≥2 years old.
Communities were identified from state Medicaid enrollment lists and state population reports.8 Because of the extreme remote nature, small size and difficult access to health care of many of Alaska’s rural villages, we used different sampling methods for children living in more populated or “urban” and less populated or “rural” communities. Urban and rural communities were defined based on the number of Medicaid-enrolled children in each community. Twenty-five very small communities with fewer than 20 Medicaid-enrolled children 2 to 6 years of age were not considered for sampling due to limited resources.
We included as “urban” all communities that had ≥35 Medicaid-enrolled children 2 or 3 years of age. Eight communities met this definition, all had a population of ≥2500 persons. Based on the number of urban Medicaid-enrolled children age 6 months to 6 years and using an estimated prevalence of BLL ≥10 μg/dL of 2.5% with an upper limit of 5.0% and a 99% confidence interval around this estimate, we determined the needed urban sample size to be 481 children.16 Only children 2 or 3 years of age were included in the urban sample. Using Medicaid enrollment lists for the month before testing, we selected a systematic random sample of 2- and 3-year-old children from each of the eight communities distributed according to the total number of 1- to 6-year-old children in each community. Anticipating low response rates, we invited 1414 urban children for BLL testing.
We defined “rural” as all non-urban communities with ≥20 Medicaid-enrolled children 2 to 6 years of age. Sixty-eight communities, including the nine with populations of ≥2500 persons not included in the urban sample, met this definition. Using calculations similar to those above and substituting the number of rural Medicaid-enrolled children, we determined that we would need to test 462 rural children. Because of poor accessibility to most remote communities in the state, the expense of travel, and limited numbers of children in each village, rather than selecting a random sample of children from each of the 68 communities, we targeted 25 communities and offered testing to all Medicaid-enrolled children 2 to 6 years of age in each of these communities. Twelve communities were selected because 1991 Women, Infants, and Children (WIC) hemoglobin (HgB) screening detected anemia (ie, HgB < 11.0 g/dL for children <2 years of age or HgB < 11.2 g/dL for children 2 to 3 years of age) in ≥40% of children 6 months to 3 years of age who were tested. One community was selected because of its proximity to a lead mine. The remaining 12 communities were selected to insure that the sample tested would approximate the geographic distribution of the state’s rural population of 1- to 6-year-old children. A total of 825 rural children were invited for testing.
Testing was conducted during September 1993 through March 1994 by a mobile field team. Two weeks before each community visit, letters were sent to parents and guardians explaining the program and inviting them to bring their children to a local clinic for testing. When possible, each parent or guardian was contacted by telephone 1 or 2 days before the team’s visit. In some rural villages, local health-care workers contacted parents by personal visit, telephone, or radio to remind them that the team was in town. For a portion of children who did not appear when they were to be tested, if time was available, a home visit was made. Home visits were not attempted in Anchorage. Demographic information was not available for children not tested. We used information from children for whom a home visit was made as a surrogate for the non-responder population.
Following informed consent, parents were asked five risk assessment questions selected from the eight questions suggested in the 1992 HCFA regulation (Fig 1). We did not use three of the suggested questions: one asked about proximity to a major highway, and another to industry likely to release lead—these questions pertained to few, if any, children in Alaska. As few children in Alaska had previously been routinely tested for lead we also did not ask the suggested question concerning playmates’ blood lead history. Two of the five selected questions were modified to fit the Alaska population, and one question was reworded but not modified. A child was considered to be at “high risk” for lead exposure if a parent answered “yes” to one or more risk assessment questions. A sixth question pertaining to the child’s tendency toward eating non-food substances was asked, but was analyzed separately from the HCFA suggested questions.
Skin was cleansed vigorously with alcohol swabs and blood was collected by venipuncture in 3-mL, lead-free EDTA tubes. Tubes were kept refrigerated and hand-carried to the state public health laboratory in Anchorage, where they were packaged and sent to ESA Laboratories (identification of this laboratory does not imply endorsement by the US Public Health Service or the Department of Health and Human Services) in Bedford, Massachusetts for analysis. Tests were performed using graphite furnace atomic absorption with Zeeman background correction.
If a child’s BLL measured ≥10 μg/dL, the child was retested, consenting household members were tested, and a limited environmental investigation, including examining the condition of house paint, water sources, play areas, and hobbies and employment of household adults, was performed.
Only BLLs obtained from initial venipuncture were used for analysis. BLLs reported by the laboratory to be <1 μg/dL (0.048 μmol/L) were assigned the value of 0.5 μg/dL (0.024 μmol/L). The mean BLL for urban children was weighted by the number of Medicaid-enrolled children in each community. Univariate and bivariate analyses were performed using Epi-Info version 5.16
We estimated the laboratory cost to screen all Medicaid-enrolled children in Alaska; a cost of $20 per test for either a venous or finger-stick sample was used17,18 (personal communication, ESA Laboratories). The estimate did not include the cost of personnel, travel, shipping, and medical follow-up.
Of 2239 children who were invited to be tested, 967 (43%) were tested (Table 1). All urban children (n = 397) and 44% of the rural children (n = 570) tested were 2 or 3 years of age; in total, 66% of all children tested were 2 or 3 years of age. The response rate was higher for the rural sample (69%), where 78% of the children sampled were Alaska Native, than for the urban sample (29%), where 25% of the children sampled were Alaska Native and 49% were white. Although 42% of the Medicaid-enrolled children in the state were Alaska Native, this group represented 60% of those tested (Table 2); white children were under represented in the group tested (28%) compared with the Medicaid-enrolled population (41% were white). The geographic distribution of children tested approximated that of all children enrolled in Medicaid.
All children with an initial BLL ≥10 μg/dL had a lower level on repeat testing. Only one child, representing 0.1% of all children tested, had a persistent BLL ≥10 μg/dL (11 μg/dL or 0.53 μmol/L). None of 20 persons who lived in the households of the six retested children had BLLs ≥10 μg/dL; no sources of lead exposure were found for any of the children. All six of the children with initially elevated BLLs came from rural communities; one child was from a community selected due to having a high rate of anemia. The geometric mean BLL for children living in the village near the lead mine was 1.6 μg/dL (0.077 μmol/L).
Of 967 children tested, 856 (89%) had come electively to the clinics for testing; home visits were completed for 101 children who did not appear for testing at the clinic. There were no significant differences in sex or age between those who came to the clinics and those who were visited at home, although the proportion of children tested who were white was greater for home visits (41%) than for clinic visits (26%). The difference in geometric mean BLLs between children tested in clinics (1.9 μg/dL or 0.091 μmol/L) and those tested in their homes (1.8 μg/dL or 0.086 μmol/L) was not significant.
No single risk-exposure assessment question was significantly associated with a BLL ≥10 μg/dL (Table 3). No correlation existed between the number of positive responses and geometric mean BLL. Using only the five risk-exposure assessment questions derived from those recommended by HCFA, and defining a child to be at high risk with a positive response to one or more questions, 592 children (61%) were at high risk. The prevalence ratio, or the prevalence of a BLL ≥10 μg/dL among children identified as high risk compared with those not identified as high risk, was 3.1 (95% confidence interval = 0.4, 26.6); the predictive value of one or more positive responses was 0.8% (Table 3). Also, as two questions from the HCFA regulation had been modified, we assessed the performance of the three unmodified questions. Using this method, 346 (36%) of the children tested were classified as being at high risk; children whose parents gave a positive response to one or more of these questions were not more likely than those who did not respond positively to have a BLL ≥10 μg/dL (prevalence ratio: 1.8; 95% confidence interval = 0.4, 8.8), and the predictive value of a positive response to any of these questions was 0.9%. The sixth question, not derived from the HCFA regulation, did not identify any children with an initial BLL ≥10 μg/dL.
The estimated laboratory cost of implementing testing for every Medicaid-enrolled child in Alaska twice by 6 years of age, using a prevalence of BLL ≥10 μg/dL of 0.6%, would be $153 960 per year, or $3347 per child identified with an initial BLL ≥10 μg/dL.
In this study of a statewide sample of Medicaid-enrolled children, including those living in communities with high rates of childhood anemia or those living near a lead mine, we found the prevalence of BLLs ≥10 μg/dL to be <1%, and we did not find any children for whom clinical intervention for exposure to lead should be initiated.1 There were no differences in blood-lead levels by age, sex, race, community, urban or rural status, community prevalence of childhood anemia, or proximity to a lead mine. These findings are consistent with previous results in Alaska from targeted community screening activities and surveillance by pediatricians. All six children with an initial BLL ≥10 μg/dL had a lower BLL when retested. As retesting was performed within 1 month of the initial testing and the technique and laboratory were identical, we believe that the lower BLLs were due to regression to the mean.19
The HCFA sample questions were not useful for identifying children with BLLs ≥10 μg/dL in this population. Three questions taken directly from the eight suggested by HCFA classified nearly 40% of the children tested as “high risk,” at a minimum requiring a second blood-lead test before 6 years of age; the predictive value of risk testing using these three questions was <1%. Adding modified versions of two additional HCFA questions did not improve the predictive value of this tool. The lack of usefulness of the questions is most likely related to the extremely low prevalence of BLLs ≥10 μg/dL in this population and unique social and demographic factors; other studies have shown the HCFA questions to be useful in identifying children at high risk in higher prevalence areas, or areas where there are children with BLLs far above the recommended level of concern.20,21
Our study was limited by several factors. First, the response rate was low in urban communities. As information on non-responders was not available, we evaluated as surrogates for non-responders those children who required a home visit for venipuncture; comparison of BLLs of children who required a home visit and those who presented electively for testing suggested that a higher response rate would most likely have resulted in similar findings. Second, children <2 years of age were not studied. Recent data from the National Health and Nutrition Examination Surveys III found that the geometric mean BLL was highest among persons 1 to 2 years of age or >50 years of age.22However, compared to 1 to 2 year olds, children 3 to 5 years of age had only a slightly lower geometric mean BLL (0.17 μg/dL vs 0.19 μg/dL for 1 to 2 year olds) and a slightly lower proportion with a BLL ≥10 μg/dL (7.3% vs 11.5% for 1 to 2 year olds).22 Since 30% of our study population consisted of 2 year olds and the differences between 1 to 2 year olds and 3 to 5 year olds are relatively small, we would expect substantially the same results if children <2 years of age had been included in the study. Furthermore, we found no variation in either the geometric mean BLL or the prevalence of BLL ≥10 μg/dL between the 2 and 3 year olds; lack of variation by age along with the very low levels and low prevalence of elevated BLL in these young children suggests that it is unlikely that we would have found significantly higher BLLs if we had tested children <2 years of age. Third, since testing was conducted during the fall, winter, and spring, it is possible that a somewhat greater number of children would have had BLLs ≥10 μg/dL if testing was done during the summer.23
Currently, public health nurses are the major providers of HCFA mandated examinations for Medicaid-enrolled children living in the more remote sections of the state; these nurses do not routinely perform and most are not trained in pediatric venipuncture. Implementing universal screening using venipuncture in rural Alaska would require either: (1) training nurses in pediatric venipuncture, or (2) sending a traveling team to each village, as was done in this study. Universal screening by finger-stick blood sampling would be simpler as it would require less training; however, children with a BLL ≥10 μg/dL would need to have a follow-up specimen collected by venipuncture.
We estimated the cost for laboratory services alone of implementing a universal screening program in Alaska to be more than $3300 per child identified with an initial BLL ≥10 μg/dL. The necessary addition of personnel, provider training, travel, and follow-up testing, particularly in the more remote areas of the state, would greatly increase the cost of any blood-lead screening program.
There is considerable controversy over the usefulness of universal blood-lead screening for young children.24 Using today’s standards, more than 88% of US children 1 to 5 years of age in the late 1970s had BLLs above the recommended level of concern; as of 1988 through 1991, the proportion of children 1 to 5 years of age who had a BLL ≥10 μg/dL decreased to approximately 9%.22 As the prevalence of lead exposure among children has declined, so has the usefulness of universal screening as a prevention tool. This statewide study failed to identify any children who would benefit from intervention for exposure to lead. Additionally, all HCFA recommended risk assessment questions, including those most likely to identify children with BLLs ≥10 μg/dL exhibited a very low predictive value in this low prevalence population. Our findings suggest there are a very few, if any, Medicaid-enrolled children in Alaska who would benefit from either HCFA mandated lead-exposure risk assessment or blood lead testing. Children presenting with signs or symptoms of lead poisoning or who are suspected to be at high risk of having an elevated BLL based on their specific circumstances, however, should continue to be tested.
HCFA regulations were developed in response to 1991 CDC recommendations for lead poisoning risk assessment and testing. These recommendations and regulations were initially intended to address the high proportion of children, particularly low-income children living in inner-cities, who were exposed to lead. These screening requirements do not appear to benefit the Medicaid-enrolled children living in Alaska. Money spent screening such children could have a much greater impact if used for more common pediatric public health problems such as infant mortality, child abuse, teenage pregnancy, maternal substance abuse, or adolescent suicide. Although results from Alaska may be generalizable only to some US children, they strongly suggest that the current screening regulations are wasteful and not useful in all communities. Although CDC recommended exempting blood lead testing in communities demonstrating a low prevalence of lead exposure, HCFA does not allow such an exemption. Furthermore, no mechanisms have been recommended by CDC or HCFA for states or regions to appropriately and efficiently assess the extent of childhood lead exposure. The forthcoming challenge will be to develop ways of assessing risk in individual groups, localities, and regions, determining the most effective methods of identifying children at highest risk, and developing appropriate methods for responding to the difficult problem of lead poisoning in children. Unless exemptions are reinstated for low-risk groups, universal testing of children in low-risk areas will continue to divert health care resources into activities providing little or no benefit.
The authors would like to acknowledge the contributions of the following persons: Diane Ingle, Elaine McKenzie, Connie Ozment, Mindy Schloss, Margaret Smith, Jan Sullivan, State of Alaska Department of Health and Social Services; Peter Briss, John Horan, Betsy Thompson, Centers for Disease Control and Prevention.
- Received May 29, 1996.
- Accepted September 4, 1996.
Reprint requests to (M.B.) Section of Epidemiology, Alaska Department of Health and Social Services, PO Box 240249, Anchorage, AK 99524.
- US =
- United States •
- CDC =
- Centers for Disease Control and Prevention •
- HCFA =
- Health Care Financing Administration •
- BLL =
- blood lead level •
- WIC =
- Supplemental Program for Women, Infants, and Children •
- HgB =
- ↵Centers for Disease Control. Preventing Lead Poisoning in Young Children. Atlanta, GA: Centers for Disease Control; 1991
- ↵State Medicaid Manual. Early and Periodic Screening, Diagnosis, and Treatment (EPSDT). Juneau, AK: State Medicaid Manual; 1992
- ↵State Medicaid Manual. Early and Periodic Screening, Diagnosis, and Treatment (EPSDT). Juneau, AK: State Medicaid Manual; 1993
- Kazal LA,
- Bullock PW
- Wiley JF,
- Bell LM,
- Rosenblum LS,
- Nussbaum J,
- Tobin R,
- Henretig FM
- ↵State of Alaska. Alaska Population Overview: 1990 Census and Estimates. Anchorage, AK: 1991
- ↵Communications. Anchorage, AK: Alaska Housing Finance Corporation; 1995
- ↵Gessner B. Internal Report, State of Alaska Section of Epidemiology. Anchorage, AK: 1995
- ↵Lee C. Final Report. Health Hazard and Risk Assessment from Exposure to Heavy Metals in Ore in Skagway, Alaska. Anchorage, AK: State of Alaska Section of Epidemiology, October 23, 1989
- ↵Middaugh J. Final Report. Assessment of lead levels in Noatak and Kivalina. October 1990. Anchorage, AK: State of Alaska Section of Epidemiology, May 1, 1991
- ↵Bathony B, Williams J. Effects of Elevated Water Lead Levels on the Blood Lead Levels of Residents of the North Slope Borough, Alaska. Barrow, AK: North Slope Borough Health and Social Services Agency (undated)
- Middaugh J. Memorandum for the Record. Lead in Birch Creek Drinking Water. Anchorage, AK: State of Alaska Section of Epidemiology, July 28, 1980
- Beller M. Final Report. Blood Lead Testing in Port Graham, Alaska. Anchorage, AK: State of Alaska Section of Epidemiology, April 27, 1994
- ↵Dean AG, Dean JA, Burton AH, Dicker RC. Epi Info, Version 5: A Word Processing, Database, and Statistics Program for Epidemiology on Microcomputers. Atlanta, GA: Centers for Disease Control; 1990
- ↵Brown BW, Hollander M. Statistics: a Biomedical Introduction. New York, NY: John Wiley & Sons; 1977
- Schaffer SJ,
- Szilagyi PG,
- Weitzman M
- Nordin JD,
- Rolnick SJ,
- Griffin JM
- Needleman HL
- Piomelli S
- Schoen EJ. Blood lead screening: the argument against it. Am Fam Physician. 1993;1371–1372
- Reigart JR. Blood lead screening: the argument for it. Am Fam Physician. 1993;1369–1371
- Harvey B
- Chisholm J,
- Goldstein G,
- Cory-Slechta D,
- et al.
- Harvey B
- Copyright © 1997 American Academy of Pediatrics