Cost-effectiveness Analysis of Lead Poisoning Screening Strategies Following the 1997 Guidelines of the Centers for Disease Control and Prevention FREE

IN THE United States, the mean blood lead (BPb) level in children aged 1 to 5 years has decreased substantially in the past 2 decades, from 0.72 µmol/L (15 µg/dL) from 1976 to 1980,¹ to 0.13 µmol/L (2.7 µg/dL) from 1991 to 1994.² Currently 4.4% of children aged 1 to 5 years have a BPb level of at least 0.48 µmol/L (10 µg/dL).²

With the fall in mean BPb level, the Centers for Disease Control and Prevention (CDC) recently recommended that each state develop regional screening plans for 1- and 2-year-old children based on the local risk for lead exposure.³ Universal BPb screening is recommended for high-risk areas. Targeted risk-based screening is recommended for low-risk areas.

Universal screening can be performed using capillary or venous specimens. Capillary blood sampling is technically easier⁴ and less costly^{5 - 6} than venous blood sampling. However, due to skin contamination, capillary sampling can overestimate true BPb level.⁴

Targeted screening reserves BPb measurements for children considered at risk as judged by a questionnaire. Studies suggest that risk assessment questionnaires may be effective as primary screening devices.^{7 - 8} The CDC risk assessment questionnaire includes questions pertaining to the age of the house in which the child resides, the presence of another child in the home with an elevated BPb level, and whether the child has a playmate who has had lead poisoning.³ This questionnaire has been modified by some researchers to include more questions in an attempt to increase sensitivity.^{9 - 10} As with universal screening, the BPb measurements for those at risk can be performed using venous or capillary blood.

Our purpose was to measure and compare the cost and effectiveness of the following 4 screening strategies: (1) universal screening of venous BPb levels; (2) universal screening of capillary BPb levels; (3) targeted screening of venous BPb levels for those at risk; and (4) targeted screening of capillary BPb levels for those at risk. Costs include those associated with screening, follow-up, and treatment based on the 1997 CDC recommendations.³

Previous studies have compared the costs of BPb screening, using various strategies based on earlier recommendations.^{5 - 6 ,11} We compared cost and effectiveness of each strategy based on current recommendations. Important outcome measures included the proportion of children with elevated BPb levels detected and the cost per case of elevated BPb levels detected.

We constructed decision trees to compare the 4 different screening options (Figure 1). Each tree represents a different screening strategy. The first node represents the true BPb level, which is based on the prevalence of elevated BPb levels. Subsequent nodes represent the classification of the BPb level by true-positive, true-negative, false-positive, and false-negative screening results. This classification depends on the sensitivity and the specificity of the screening test. In practice, a venous BPb assay is assumed to reflect the true BPb level, and therefore is assumed to correctly identify all cases (Figure 1, A). The accuracy of universal screening of capillary blood (Figure 1, B) depends on the characteristics of capillary BPb testing to properly identify cases. Similarly, the accuracy of risk assessment of follow-up venous BPb testing (Figure 1, C) relies on the test characteristics of the screening questionnaire. The accuracy of risk assessment with follow-up capillary BPb testing (Figure 1, D) relies on the test characteristics of the questionnaire and of capillary BPb testing.

The 1997 CDC recommendations³ were incorporated into the model (Table 1). The CDC recommends that the venous BPb levels should be checked in any child found to have a level of at least 0.48 µmol/L (10 µg/dL). This repeated measurement, referred to as diagnostic testing, should be performed even if the initial BPb level was measured using venous blood. We assumed that this diagnostic testing would also involve a detailed history and examination to assess potential risk.

Management of BPb levels from 0.48 to less than 0.72 µmol/L (10 to <15 µg/dL) involves repeated determination of BPb level to ensure that the level is not rising. From 0.72 to less than 0.97 µmol/L (15 to <20 µg/dL), the test is repeated. If the level does not fall, treatment proceeds as if the level is from 0.97 to less than 2.17 µmol/L (20 to <45 µg/dL). Above 0.97 µmol/L (>20 µg/dL), an environmental investigation with lead hazard control is performed. From 2.17 to less than 3.38 µmol/L (45 to <70 µg/dL), oral chelation is begun. At or above 3.38 µmol/L (≥70 µg/dL), intravenous chelation is initiated.

In this model, we assumed that interventions are always successful and need not be repeated. We also assumed that phlebotomy, when needed, is always successful, and that parents can answer all questions on the CDC risk assessment questionnaire. These assumptions minimized cost and biased the model toward the effectiveness of lead screening programs. We also assumed that BPb levels do not change with time, except with intervention. Neither random variation nor regression toward the mean was included in the model. This may have biased costs upward.

Extensive sensitivity analyses were performed to test the robustness of the model and to identify important areas of uncertainty. All modeling was performed using DATA 3.0 (TreeAge Software, Inc, Williamstown, Mass).

This analysis was based on the hypothetical experience of 3 cohorts of 100,0001-year-old children in the United States. These 3 cohorts represented samples from communities with a low, medium, or high prevalence of lead poisoning. The CDC guidelines³ allow states to recommend that 2-year-old children also undergo screening. This study focused only on the experience of 1-year-old children.

The cost-effectiveness analysis was conducted from the perspective of the health care system. The costs of screening, evaluation, and treatment were included. Indirect costs, ie, parental wages lost, transportation costs, and child-care costs incurred due to the screening process, were not included.

Primary outcome measures were the cost per child undergoing screening, the number of cases of BPb levels of at least 0.48 µmol/L (≥10 µg/dL) detected, and the cost per treated case of elevated BPb level. Children with BPb levels greater than 0.72 µmol/L (>15 µg/dL) require more intensive evaluation and treatment than those with BPb levels from 0.48 to less than 0.72 µmol/L (10 to <15 µg/dL). We chose as secondary outcome measures the number of cases detected with BPb levels of at least 0.72 µmol/L (≥15 µg/dL) and the cost per case of detecting these cases.

A child's risk for having an elevated BPb level is based on the prevalence of elevated BPb levels in the community. We evaluated cohorts from low-, medium-, and high-prevalence communities (Table 2). The low-prevalence community was based on the overall distribution of BPb levels in 1- to 2-year-old children in the United States.^{2 ,12} The medium-prevalence community was based on 2 large, metropolitan cities.¹³ The high-prevalence community was based on a former mining town.¹⁴ Large-population studies show that the prevalence of BPb levels greater than 2.17 µmol/L (>45 µg/dL) is rare.^{12 ,15 - 17} To emphasize the potential importance of BPb screening programs, we skewed the medium- and high-prevalence communities to include a small proportion of the populations in this highly elevated range. This assumption could increase the overall cost of any screening strategy that is better able to detect cases.

Although the accuracy of venous BPb levels may vary with the proficiency of the laboratory performing the assay, the model assumed that the sensitivity and specificity of venous BPb tests were 100%.

The sensitivity of capillary BPb tests to detect venous BPb levels of at least 0.48 µmol/L (≥10 µg/dL) increases as the true BPb level increases.¹⁸ The model assumed that if the venous BPb level is from 0.48 to less than 0.72 µmol/L (10 to <15 µg/dL), the sensitivity is 91%,¹⁸ with a possible range of 86% to 95%.⁴ We assumed that the sensitivity of capillary BPb tests to detect venous BPb levels of at least 0.48 µmol/L (≥10 µg/dL) is 100% when the true venous BPb level is at least 0.72 µmol/L (≥15 µg/dL).¹⁸

The baseline estimate of the specificity of capillary BPb screening for a venous BPb level of 0.48 µmol/L (10 µg/dL) was 92%.¹⁸ Since the specificity of capillary BPb tests depends on many factors, including method of finger cleaning, method of sample collection, and type of assay, we evaluated the specificity for a range of 88% to 99%.^{4 ,18}

As with capillary BPb screening, the sensitivity of the risk assessment questionnaire to screen for venous BPb levels of at least 0.48 µmol/L (≥10 µg/dL) increases as true BPb level increases. The model assumed that from 0.48 to less than 0.72 µmol/L (10 to <15 µg/dL), the sensitivity is 75%, with a range of 70% to 90%.^{7 - 9 ,19} We assumed that the sensitivity of the risk-assessment questionnaire to detect venous BPb levels of at least 0.48 µmol/L (≥10 µg/dL) increased by 10% when the true venous BPb level is at least 0.72 µmol/L (≥15 µg/dL).^{8 ,19}

The estimate of the specificity of the questionnaire for a venous BPb level of 0.48 µmol/L (10 µg/dL) was 49%. We evaluated the specificity for a range of 30.0% to 50.0%.^{7 - 9 ,19 - 20}

Table 3 lists the estimates and ranges of the costs of screening, evaluation, and treatment in 1996 dollars. The cost of the risk assessment questionnaire was based on the amount of nursing time needed to administer it. The costs of nurse-only and physician visits were estimated through discussion with an academic pediatric practice and a large, private pediatric practice in North Carolina. The cost of environmental investigation and hazard removal was estimated through discussion with the North Carolina Department of Environmental and Natural Resources, Raleigh. The cost of chelation includes the costs of the drug therapy, required follow-up visits, and laboratory monitoring.

The cost per child undergoing screening, number of cases detected, and cost per case are listed in Table 4. The number of cases detected by each strategy depends on the prevalence of elevated BPb levels. The percentage of cases detected by each strategy is listed inTable 5. Targeted screening detected fewer cases than universal screening. Targeted screening was more sensitive in the identification of cases at least 0.72 µmol/L (≥15 µg/dL) than from 0.48 to less than 0.72 µmol/L (10 to <15 µg/dL).

Regardless of whether a strategy of universal or targeted screening was used, capillary blood testing detected fewer cases from 0.48 to less than 0.72 µmol/L (10 to<15 µg/dL) than venous testing. Above 0.72 µmol/L (≥15 µg/dL), capillary blood testing detected all cases that would have been identified using venous testing.

The cost of screening per child rose with increasing prevalence due to the greater number of cases detected. However, the cost per case fell with increasing prevalence because the overall cost of screening was divided among a greater number of cases. The cost per child undergoing screening was greatest in the high-prevalence population, but the cost per case detected was the least.

Universal capillary BPb screening was more expensive and less effective than universal venous BPb screening for detecting BPb levels of at least 0.48 µmol/L (≥10 µg/dL). Similarly, targeted screening using capillary BPb testing was more expensive and less effective than targeted screening using venous BPb testing. Changing from a strategy of venous to capillary testing would detect fewer cases of greater than 0.48 µmol/L (>10 µg/dL) at greater expense.

Capillary testing increased the cost of screening because of the need to follow up false-positive results. The number of false-positive results fell with increasing prevalence, which narrowed the difference in the cost of screening between venous and capillary testing. Although the cost of venous testing strategies included the cost of treatment of more cases than that of capillary testing, capillary testing strategies were more expensive for the detection of levels greater than 0.48 µmol/L (>10 µg/dL), even in the high-prevalence population. Capillary testing was less expensive only in the high-prevalence population in detecting levels greater than 0.72 µmol/L (>15 µg/dL).

The cost per child undergoing universal screening was greater than that for targeted screening. However, universal screening detected more cases. Table 6 lists the extra cost per case identified incurred by switching from a strategy of targeted screening to universal screening. In the high-prevalence population, the cost per case detected using universal screening was less than that of targeted screening for levels of greater than 0.48 µmol/L (>10 µg/dL).

Sensitivity analysis was performed to evaluate the conclusion that venous BPb testing is superior to capillary BPb testing for detecting levels of at least 0.48 µmol/L (≥10 µg/dL), regardless of whether a strategy of universal or targeted screening is used. Univariate sensitivity analyses were conducted on all variables in the model.

Venous testing detected more cases of at least 0.48 µmol/L (≥10 µg/dL) at less cost per case than capillary testing for universal and targeted screening across all ranges of sensitivity and specificity for capillary BPb testing and risk assessment.

For universal and targeted screening, venous testing was superior across all ranges of costs of capillary phlebotomy, lead assay, physician and nurse-only visits, environmental investigation and hazard removal, and oral and intravenous chelation.

For the low- and medium-prevalence populations, if the cost of venous phlebotomy was above $12 to $13, the cost per case detected using capillary testing would be less than that detected using venous testing. The difference, however, was small even at a level that would create the largest difference. For example, if the cost of venipuncture were $15, the cost per case detected using universal venous testing would be $979 in the low-prevalence population, as compared with $934 using universal capillary screening.

The current CDC guidelines set the threshold for lead poisoning at 0.48 µmol/L (10 µg/dL).² Only universal venous screening will detect all cases of elevated BPb levels of at least 0.48 µmol/L (≥10 µg/dL). In all but high-prevalence settings, this strategy has the highest cost per case detected. Universal capillary screening is less effective than universal venous screening in detecting cases from 0.48 to less than 0.72 µmol/L (10 to <15 µg/dL). Furthermore, the cost per case detected using universal capillary screening is greater than that of universal venous screening. Targeted screening detects fewer cases than universal screening. However, in all but high-prevalence populations, targeted screening has a lower cost per case detected and treated than universal screening.

The CDC recommends universal screening in areas where the prevalence of elevated BPb levels is at least 12%.³ The selection of this cutoff point was a policy decision based on the potential adverse outcome in missed cases. This analysis further supports universal venous screening in high-prevalence populations, as this strategy has a lower cost per case detected.

Strategies for lead poisoning screening in children should be highly sensitive if there are significant consequences of a missed diagnosis. Screening with a risk assessment questionnaire may fail to detect 27% of children with elevated BPb levels. The distribution of elevated BPb levels is such that most of the missed cases are in children with levels of less than 0.72 µmol/L (<15 µg/dL). Although marginally elevated BPb levels may be harmful,²² there is little evidence that interventions are successful in lowering BPb levels in this range.¹⁸ Universal screening would detect all cases; however, the additional cost per case detected above 0.72 µmol/L (≥15 µg/dL) in low- and medium-prevalence populations is large.

Our study supports previous findings that capillary screening with follow-up of abnormal results using venous sampling is more expensive than universal screening using venous specimens alone.⁶ Many providers consider venipuncture time-consuming, and many parents find it unappealing. Capillary screening seems simpler and less expensive than venous screening. However, false-positive results are common and require follow-up with venous BPb screening. Unfortunately, venipuncture can be difficult for some toddlers. The cost of screening based on a recommendation that BPb levels be measured using venous sampling would be a weighted average based on the success rate of venous testing and the rate of children undergoing capillary testing.

Unlike a previous study that examined only the direct costs of screening strategies for BPb poisoning,⁶ this analysis includes the costs of follow-up and treatment. Our model does not include the costs incurred due to the effects of elevated BPb levels, ie, decreased productivity or IQ. These effects are difficult to quantify, and the long-term benefits of treating elevated BPb levels are not known.²³

Our analysis used only direct costs. We suspect that indirect costs would add more substantially to the cost of capillary testing, especially in the low-prevalence population, due to the larger number of false-positive results. A child with false-positive results of BPb testing requires a separate clinic visit for evaluation and repeated BPb level measurement.

Our results support the use of universal screening in high-prevalence populations. In low- and medium-prevalence populations, the benefit of detecting extra cases through universal screening must be weighed against the extra cost of such a screening policy. This analysis also suggests that, when required, BPb levels should be measured using venous specimens. Regardless of whether a strategy of universal or targeted screening is used, the cost per case detected of venous testing is less than that of capillary testing.

Editor's Note: If cost-effective lead screening in both high- and moderate-low–prevalence areas are from Venous, what is from Mars?—Catherine D. DeAngelis, MD

Accepted for publication June 11, 1998.

Reprints: Alex R. Kemper, MD, MPH, University of North Carolina, Campus Box 7225, Chapel Hill, NC 27599-7225 (e-mail: akemper@med.unc.edu).