Environmental Lead and Children's Intelligence: A Review of Recent Epidemiological Studies

Abstract

The possibility that environmental lead in the general community could have an effect on children's intelligence, behaviour and performance has provoked consider- able scientific, public and political concern. In the past few years there have been several epidemiological surveys of young children, particularly in Britain and the United States, which have aimed to evaluate whether low level lead exposure in young children is associated with small deficits in IQ and other neuropsychological tests. There are several methodological issues that need to be considered when interpret- ing study findings. These include the possibility of selection bias, difficulties in measuring body lead burden, allowance for the numerous confounding factors which are known to be associated with a child's IQ, the appropriate use of statistical analysis techniques, the inadequate size of many studies and the difficulty in attaching any causal interpretation to statistical associations in cross-sectional surveys. Studies of lead and IQ in the United States have recently been-criticised for serious deficiencies in design and analysis. Accordingly, recent British studies of young children seem likely to provide the most reliable evidence. One pilot study in 1981 of children in East London did show a negative association between blood lead and IQ, but parental IQ was not taken into account. Subsequent British studies in Birmingham, Leeds and London have not found any significant lead-IQ association. In particular, further analysis of the largest study of 402 London children has recently been undertaken. Previous analyses had compared mean IQ in low, medium and high tooth lead groups. A more quantitative approach relating individual IQs and tooth lead estimations gives further insight into the data and reinforces the previous conclusion that there is no evidence that moderately elevated tooth lead levels are associated with lower IQs after allowance for confounding factors. This analysis illustrates the use of 'optimal' regression techniques and the need to define an analysis policy for such data. The inherent weakness of the epidemiological approach in studies of low level environmental hazards affecting large populations needs further emphasis.