From the State of California, Division of Environmental and Occupational Disease Control, Department of Health Services, Oakland, CA.
Correspondence: Raymond Richard Neutra, State of California, Division of Environmental and Occupational Disease Control, Department of Health Services, 1515 Clay Street; 17th Floor, Oakland, CA 94612. E-mail: email@example.com.
In the January 2003 issue of Epidemiology, David Savitz1 wrote a provocative editorial about publication priorities for research on electric and magnetic fields (EMF). Savitz suggested that well-done research was not necessarily enough, and that papers should materially advance our scientific understanding of EMF effects. In expressing these views, he was articulating the classic concerns of one important stakeholder in hazards research, namely, our scientific journals.
However, hazards research has many stakeholders. These include persons at risk for the hazard, the people (if any) who produce and profit from the hazard, lawyers and judges, and the general public. As research progresses to the stage of publication, we can add editors, journals, and responsible government agencies to our list of stakeholders. The inevitable divergence of interests among stakeholders can lead them to advocate very different courses of action when it comes to the conduct of research and publication of its results.
As a member of the editorial board of Epidemiology, I understand the journal’s need to intrigue and enlarge the readership with articles that provide something exciting and new. Articles that simply reiterate, in new settings or with different methods, what is already known elsewhere are admittedly boring.
However, when I wear my hat as a public official, with responsibility for doing meta-analyses of a body of evidence, I want to see all of the evidence. This includes the really boring studies with null results that authors and editors alike are tempted not to publish. For years, I have been trying to persuade the U.S. Environmental Protection Agency to support an online journal that could be called The Journal of Boring Policy-Relevant and Null Results. This would provide an outlet for good-quality articles that contribute to the weight of evidence for public policy.
I have pointed out elsewhere that researchers have interests distinctly different from at-risk stakeholders.2 Researchers are obliged to deploy their scarce research years efficiently so as to maximize their chances of successful discoveries. They need to avoid antagonizing their fellow hunters by sending them on wild goose chases. Also, they need to keep the quinquannual grant money flowing. All these forces push research toward safe, incremental projects. Academic research is indeed the art of the soluble. In contrast, the at-risk stakeholders would like research to address their perceived problems, no matter how long, difficult, costly, or academically suicidal the task might be.
Another approach is a utilitarian “most good for the most people at the least cost” approach. A utilitarian analysis might suggest a set of research priorities that differs from either those of distinguished scientists or those of specific advocacy groups. During our California EMF project, we contracted with an expert decision analyst to do a “value-of-information” decision analysis on EMF research policy. His approach was to translate varying degrees of certainty regarding a range of health effects into a rational basis for setting an EMF research budget. The resulting decision tree (which is available on our web site3) does not consider the probability of obtaining research funding, colleague admiration, satisfaction of scientific curiosity, or other outcomes of interest to researchers and editors. It does consider the differential cost of cheap or expensive EMF mitigation, the probability of lawsuits and social strife, the cost of human life, the probability of research breakthroughs, and the expected burden of ill health.
The analysis concluded that if there is more than a 10% certainty that 100 deaths from EMF occur annually nationwide, then an expenditure of $50 million a year on EMF research would be justified. The basis for this hefty EMF research budget is the high value of a human life and the fact that expensive policy decisions would be influenced by the research results.
A rational analysis would probably give an even higher benefit–cost ratio for investments into research on non-EMF causes of early childhood conditions such as birth defects, because such conditions extract an even greater cost in quality-adjusted life-years (QUALYs). A recent panel’s recommendation to defund EMF research entirely4 might be explained by such analyses, as well as by the academic considerations I have mentioned and by the thought that every dollar spent on EMFs is one dollar less for other research topics cherished by the panel’s members.
Decisions are never purely utilitarian. If they were, we might be spending our first trillion dollars of research on birth defects, leaving all other diseases unfunded. Other ethical frameworks must come into play, for example, considerations of distributional and restitutional justice. We spend money on diseases that affect only a few vulnerable people, or on exposures that are unequally and unfairly distributed in the population (as in the case for many environmental exposures such as EMFs) even if it is not necessarily efficient or cost-effective to do so.
Indeed, the principle of “polluter or user pays” could suggest that the EMF research budget not be supported from the usual sources of research funds. Fifty million dollars per year represents an additional 25 cents a year to a typical American utility bill and approximately 1/1000 of the annual profits of utility company shareholders.3
If one accepts the recommendation of a $50 million annual EMF research budget, then one is faced with what to spend it on. The decision tree suggests paying attention to the likelihood of research success, the prior certainty that the disease is linked to EMFs, the QUALYs lost each year to that disease, and the impact of new information on policy. After a rough semiquantitative contemplation of these factors, we made some research recommendations in California.5 We gave high priority to further study of EMFs and sudden cardiac death, despite the lack of clarity about mechanism. We were influenced by the high incidence of this condition, combined with the evidence from Savitz’s own very large study,6 that any meta-analysis dwarfs the 2 other much smaller contradictory studies.7,8 Also, the implied incubation period was short enough that we thought a large well-designed study was feasible. Finally, policy would be more strongly influenced by results showing an effect of EMFs on a high-incidence disease such as sudden cardiac death than by data that upgraded the certainty of an effect on a rare disease such as childhood leukemia.
Policy would be served, too, by improvements in occupational job-exposure matrices to determine whether the confounding factors of intermittent contact currents and shocks explained the utility worker associations with brain cancer and leukemia, because they were alleged to explain9 the consistent10 association between magnetic fields and amyotrophic lateral sclerosis. Both of these lines of research (along with some suggestions for laboratory studies) were given high priority because their results, regardless of outcome, would help decide whether matters of EMF mixture are worth further study and on which component to focus.
Answering these questions might take several rounds of study. The results, if null or inconclusive, would nonetheless be influential to policy formation, whether from a utilitarian or a social justice perspective. Epidemiology might not publish these boring results, but the proposed Journal for Boring Policy-Relevant and Null Results ought to.
Proclivities of scientists aside, the conduct and publication of epidemiologic studies should be as much about policy as about intellectual stimulation.
About the Author
Raymond Richard Neutra is a physician epidemiologist who, after a brief career on the Navajo reservation, has worked in the borderlands of epidemiology and policy in both academia and public agencies. He oversaw a decade-long project in California regarding EMF policy options, including policies on further research. He has written on EMF exposure analysis and effects of EMF on miscarriage and birth defects.
1.Savitz DA. Health effects of electric and magnetic fields:are we done yet? Epidemiology. 2003;14:15–17.
2.Neutra RR. Should one start or continue a line of research? Stakeholders’ interests and ethical frameworks give different answers. Annals AAPSS. 2002;584:125–134.
4.National Research Council. Research on power frequency fields completed under the energy policy act of 1992. 1999. Available from: http://www.nap.edu/catalog/9587.html
. Accessed March 21, 2003.
5.Neutra RR, DelPizzo V, Lee G. An evaluation of the possible risks from electric and magnetic fields (EMFs) from power lines, internal wiring, electrical occupations, and appliances. California Department of Health Services; 2002, Available from http://www.dhs.ca.gov/ehib/emf
6.Savitz DL, Sastre A, Kleckner RC. Magnetic field exposure and cardiovascular disease mortality among electrical workers. Epidemiology. 1998;149:135–142.
7.Kelsh MS, Sahl JD. Mortality among a cohort of electrical utility workers, 1960-1991. Am J Indust Med. 1997;31:534–544.
8.Baris D, Armstrong BG, Deadman J, et al. Mortality study of electrical utility workers in Quebec. Occup Environ Med. 1996;53:25–31.
9.Johansen C, Olsen JH. Risk of cancer among Danish utility workers—a nationwide cohort study. Am J Epidemiol. 1998;147:548–555.
10.Ahlbom A. Neurodegenerative diseases, suicide and depressive symptoms in relation to EMF. Bioelectromagnetics. 2001;Suppl 5:S132–S143.
© 2003 Lippincott Williams & Wilkins, Inc.