The controversy over the possible association between magnetic field exposure and childhood leukemia has led several researchers to summarize the literature using meta-analysis. This paper reviews these previous meta-analyses and extends them by adding results from four studies published since the most recent analysis. The analyses include odds ratio calculations based on both dichotomous and continuous exposure models, heterogeneity analysis including subgroup summaries and meta-regression, "leave one out" influence analyses, and publication bias assessments. In addition, there is a review of some of the considerations of the exposure assessments used in the studies and their implications for cross-study comparisons. Finally, the results of the analyses using dichotomous and continuous exposure model are combined with national exposure data to estimate the population attributable risk of childhood leukemia among children in the US. If an association exists, as many as 175-240 cases of childhood leukemia in the US may be due to magnetic field exposure.
Copyright 2001 Wiley-Liss, Inc.