Lead Exposure and Cognitive Development

Publisher Website
Google Scholar
Abstract
To the Editor: We commend Bellinger et al. (April 23 issue) for thoroughness in the reports of their work.1 2 3 4 5 Nevertheless, an unusual configuration of results undermines the inferences drawn. First of all, linear models were used to adjust the cognitive data for 12 covariates; of these, several were related in a curvilinear manner to lead and to mental development as well. Most notably, gestational age was related in a positive curvilinear manner to cognitive scores and to the lead level1 (g=×4 2 = 11.74, P<0.05). Linear representation of curvilinear relations may have resulted in spurious "significant" results due to overprediction . . .