Epidemiologic approaches to assessing the developmental toxicity of lead

A variety of designs have been employed in epidemiologic studies of the developmental morbidity associated with low-level lead exposure. Historically, cross-sectional and retrospective cohort designs have been used most frequently. Despite improvements in their methodological rigor, however, certain design features constrain the inferences such studies can support. These limitations stem from the substantial risk that children's exposure status may be misclassified due to reliance on indices with short averaging times, and an inability to identify either age-related changes in vulnerability or time-dependent aspects of the expression of toxicity (e.g., reversibility). In response to these limitations, several studies were initiated involving repeated measurements of children's lead exposure and development over periods as long as a decade. Although these prospective studies are characterized by an unusual degree of coordination among the investigators, there are differences among them as well, most notably in terms of sample characteristics and patterns of exposure. As a result, the studies should be viewed as complementary rather than simply as replicates of one another. Moreover, like all epidemiologic approaches the prospective design has its own limitations. These include the need to maintain follow-up over a long period of time, as well as the attendant risk of bias in sample attrition, and the need to distinguish developmental effects of lead from psychometric artifacts. The Boston prospective study is used to illustrate both the strengths and weaknesses of the prospective design.