D Wartenberg1. 1. Department of Environmental and Community Medicine, University of Medicine and Dentistry of New Jersey-Robert Wood Johnson Medical School, Piscataway, USA. dew@eohsi.rutgers.edu
Abstract
OBJECTIVES: This article uses meta-analysis methodology to examine the statistical consistency and importance of random variation among results of epidemiologic studies of residential magnetic field exposure and childhood leukemia. METHODS: A variety of meta-analytic statistical methods were applied to all available studies combined and on sub-groups of studies chosen by exposure characteristics. Sample sizes and fail-safe n's were calculated to determine the robustness of results and the potential role of publication bias. RESULTS: Most studies show elevated but not statistically significant odds ratios. Results for exposures assessed by wire codes, distance, and/or historically reconstructed fields are relatively consistent, homogeneous, and positive, while those for direct magnetic field measurements are consistent, homogeneous, and marginally protective. Several unpublished studies, or a single unpublished study with several hundred subjects, would be needed to nullify the observed data. CONCLUSIONS: The observed results identify a consistent risk that cannot be explained by random variation. The data supporting magnetic fields as the principal risk factor are suggestive but inconsistent. Additional studies using innovative designs that focus on highly exposed children offer the most hope of untangling this issue.
OBJECTIVES: This article uses meta-analysis methodology to examine the statistical consistency and importance of random variation among results of epidemiologic studies of residential magnetic field exposure and childhood leukemia. METHODS: A variety of meta-analytic statistical methods were applied to all available studies combined and on sub-groups of studies chosen by exposure characteristics. Sample sizes and fail-safe n's were calculated to determine the robustness of results and the potential role of publication bias. RESULTS: Most studies show elevated but not statistically significant odds ratios. Results for exposures assessed by wire codes, distance, and/or historically reconstructed fields are relatively consistent, homogeneous, and positive, while those for direct magnetic field measurements are consistent, homogeneous, and marginally protective. Several unpublished studies, or a single unpublished study with several hundred subjects, would be needed to nullify the observed data. CONCLUSIONS: The observed results identify a consistent risk that cannot be explained by random variation. The data supporting magnetic fields as the principal risk factor are suggestive but inconsistent. Additional studies using innovative designs that focus on highly exposed children offer the most hope of untangling this issue.
Authors: J Michaelis; J Schüz; R Meinert; M Menger; J P Grigat; P Kaatsch; U Kaletsch; A Miesner; A Stamm; K Brinkmann; H Kärner Journal: Cancer Causes Control Date: 1997-03 Impact factor: 2.506
Authors: E Petridou; D Trichopoulos; A Kravaritis; A Pourtsidis; N Dessypris; Y Skalkidis; M Kogevinas; M Kalmanti; D Koliouskas; H Kosmidis; J P Panagiotou; F Piperopoulou; F Tzortzatou; V Kalapothaki Journal: Int J Cancer Date: 1997-11-04 Impact factor: 7.396
Authors: E P Washburn; M J Orza; J A Berlin; W J Nicholson; A C Todd; H Frumkin; T C Chalmers Journal: Cancer Causes Control Date: 1994-07 Impact factor: 2.506