Methodology, inference and causation: environmental lead exposure and childhood intelligence

Kaufman's critique of the literature on the associations between lead exposure and child intelligence raises important methodological and inferential points. We address the concerns he raises regarding measuring known and unknown confounders, statistical modeling, reverse causality and quality control. Mismeasurement of potential confounders of the lead-IQ relationship, such as parenting skills, parental intelligence, maternal smoking during pregnancy, or otitis media can either strengthen or weaken the estimated association between exposure and child intelligence. Despite some variability in design and measurement, a series of comprehensive prospective investigations in varied populations, by different sets of investigators, provided consistent replication; taken together these studies point to the conclusion that lead exposure has adverse consequences for child development, and that the deficits are likely to be small in comparison to the contribution of measured social factors.