Eliminating childhood lead toxicity in Australia: a call to lower the intervention level

To the Editor: Taylor and colleagues1 suggest lowering the National Health and Medical Research Council intervention level for blood lead level (BLL) because two reports indicate that health effects occur at lower BLLs.2,3 Such action may be unwarranted as these effects could be artefacts of suspect methodology that was missed by peer reviewers.

One of the reports, a monograph by the US National Toxicology Program, summarised negative associations between health outcomes and lead measurements from about 600 studies, but did not critically analyse individual studies or convincingly establish causation at low BLLs.2 It ignored contradictory positive associations between IQ and BLL for Mexico City,3 Yugoslavia4 and Boston,5 which show that factors other than lead influence child IQ (eg, parental IQ, home environment, nutrition and other contaminants). It did not examine whether statistical interaction of these factors had been adequately controlled for.

The other was a report on an international analysis that pooled seven non-randomised prospective studies, including the three mentioned above.3 The analysis deleted the critical lowest 5% of lead exposures; assumed the effect of lead was the same at any IQ; ignored possible distortion of the Yugoslav data by abbreviated IQ tests; and used BLL measurements from different age groups (resulting in some BLLs probably being only half their earlier peaks). The model included only four covariates and no interaction terms, despite some factors potentially affecting lead exposure and confounding results. The arbitrary adoption of a log–linear model produced increasingly steep trends at lower BLLs, whereas the best-fit model for the depleted dataset was linear for BLL < 15 μg/dL.

Predictions based on these studies are likely to significantly exaggerate the effects of low BLL. Changes to the reference BLL should await more thorough statistical analysis of existing data. We recommend that well controlled studies be used to establish whether BLLs < 10 μg/dL cause adverse health effects.