The potential impact of bias in studies of residential exposure to magnetic fields and childhood leukemia.

Bias can have a major impact on the results of epidemiologic studies. In investigations of the possible association between residential exposure to magnetic fields and the occurrence of childhood leukemia, many have raised questions about selection bias, including participation bias and information bias. In this review, the data on these possible sources of bias are summarized and their likely impact is evaluated. Most data suggest that if a bias exists, it is a bias towards the lack of association between exposure to magnetic fields and childhood leukemia. In addition, given the wide variety of study populations and measurement protocols, it is unlikely that a single design flaw has resulted in consistent effects across all studies and can be the sole explanation for the reported associations.