Epidemiologic Studies of Health Effects of Electromagnetic Fields

David Savitz

Concern with potential health effects of electromagnetic fields, a form of nonionizing radiation in the power-frequency range (50-60 Hz)1 provides an interesting illustration of the evolution of an interesting and controversial topic in environmental epidemiology.  It was driven, from the beginning, by epidemiology as the sole empirical foundation and largely fueled by public concern and regulatory considerations rather than by supportive laboratory research.   To appreciate the flavor of the story, it has to begin with the highly unconventional, seminal paper of Wertheimer and Leeper (1979), which singlehandedly put the issue on the table.  A series of studies followed over the next 15 years, growing in scale and quality, but with one promised “definitive” study after another failing to put the controversy to rest, including a large, well-designed study from the US National Cancer Institute (Linet et al., 1997).  Over time, the focus came to rest on childhood leukemia as the primary concern, and particularly on the impact of being in the uppermost end of the magnetic field exposure distribution.  Given this interest in a rare disease and a rare exposure, even the largest studies generated imprecise results.  Two pooled analyses were conducted using broadly similar approaches (Ahlbom et al., 2000; Greenland et al., 2000) and came to a very similar conclusion:  no support for an association up to a relative high magnetic field level (0.3 or 0.4 uT), and statistically clear indications of a positive association above that level.  With subsequent studies, the evidence remains essentially where it was 15 years ago (Ahlbom et al., 2001; Kheifets et al., 2010).  While there are those who may well argue that this unexplained association of elevated magnetic field exposure and childhood leukemia provides an important clue to the environmental etiology of disease, research on this topic has gradually declined, not due to a clear conclusion or lack of interest but rather because advancing knowledge calls for a unique and elusive set of optimal conditions – large, well-defined population data resources for conducting case-control studies, and high prevalence of elevated exposure from accurately measured sources.  An implicit consensus has emerged, however, that acknowledges that the possibility of adverse effects and justifies reasonable, modest efforts to minimize exposure are warranted.  Without some major breakthrough in exposure assessment or mechanistic understanding of the etiologic process, new epidemiologic studies are not likely to markedly advance knowledge. 

1Research on power frequency fields is quite distinct from radiofrequency fields, a different form of nonionizing radiation, and the basis for concerns with cell phones and cancer. 

Leave a Reply