|
Health Effects of Low-Intensity Electric and Magnetic Fields (and Scientific Error)
By Dr. Robert K. Adair
A century ago, nearly one fifth of babies born in the United States did not live to see their tenth birthday. Today, we lose less than one in one hundred. As late as the eve of WWII, 20% of us didn't make it to fifty. Today only about 6% pass away before that birthday. Yet, entering this twenty-first century, with death pushed further from our children and ourselves than ever before, we seem to dwell on possible dangers to health more than ever before.
We know that the health we enjoy and cherish follows largely from science -- especially from fundamental biological research and from the applied biology of medicine. Hence we look to scientific research -- and especially biological research -- for the identification and relief of those matters that may still impinge on our health. But, for most non-scientists -- and too many scientists! -- that "look" is based on a flawed understanding of the scientific process. People believe too easily and too much.
Many published results of scientific research are wrong: let's look at a subset of papers on weak electromagnetic fields (EMFs) phenomena as an example.
There are at least 300 papers reporting biological effects of EMFs generated at low frequencies by our power distribution system and at higher frequencies by radio, TV, radar, and communication devices such as cell phones. But there are no such effects. As with many other scientific hypotheses, such as cold fusion and polywater in physics, all of the many positive results showing biological effects of weak EMFs, many published in reputable journals, are wrong.
These "results" can conveniently be separated into two classes, laboratory experiments and epidemiological surveys, different in process, but with similar defects. Epidemiological surveys attempt to identify the effects on health of EMFs by comparing the health of populations that are and are not subject to EMFs. Under this category, I include similar determinations of the effects of EMFs on groups of animals. Laboratory experiments largely attempt to note effects of EMFs on biological functions -- which may, or may not, in turn effect health.
Both sets of studies are subject to errors that follow from faulty processes. Science is difficult; honest, competent scientists can make mistakes. However, much error is from incompetence and some from corruption. Epidemiologist John P. A. Ioannides has published a widely circulated paper titled, "Why Most Published Research Findings are False," in which he considers factors that too often lead to erroneous conclusions.
Epidemiology's Strengths and Weaknesses
Epidemiology is a powerful tool especially important in uncovering environmental insults -- such as smoking -- where the etiology is not completely understood. The results of epidemiological studies are an especially important source of lay evaluations of health effects of EMFs because of the simple form in which the conclusions are presented.
While epidemiology is subject to procedural errors as with other studies, the emphasis on statistical significance is especially strong -- and often a source of misunderstanding if not misrepresentation. The results of epidemiological studies of health effects of EMFs are usually presented as simple relative risk-ratio (RR) numbers together with the probability (P) that the result is accidental. For example, one important study found that 11 children living near a power line source of EMFs became ill with childhood leukemia while the expectation from children who lived away from the lines, with no significant EMFs, was 4.4; thus the risk-ratio, RR = 11/4.4 = 2.5. Those children living near the power lines seemed to have two and one-half times the likelihood of developing leukemia as those who lived away from EMF sources. Elementary statistical theory tells us that the probability that 11 or more children would get leukemia by chance when the expected number is 4.4, is P = 0.0058 = 1/170. Thus, there was only one chance in 170 that 11 cases would show up in the sample accidentally. The results of this study were then widely held to show that power line EMFs were a likely cause of childhood cancer.
But like many other studies, perhaps most such studies, that result was misleading. The authors of the study had reviewed more than 300 different disease possibilities (more than 170!) of which the one described here, and in their published account, was the most significant. When the large number of studies was taken into account, it was more likely than not that in one of the studies -- that is, for some disease or disease manifestation -- 11 cases would be found when only 4 or 5 were expected. Thus the result was probably only a statistical fluctuation and no indication of EMF effects.
Lab Accidents
Often reports of positive results from laboratory experiments stem from similar abuses of statistical methods. The experimenter will examine a large set of possibilities and select for emphasis those that, by chance, mimic positive results.
The flood of poorly conducted over-advertised studies on effects of EMFs that report erroneous results stems from various causes. Scientists today, if ever, cannot proceed exactly as they please. Research requires support -- money. Unavoidably, those who disburse the money must decide where their funds will be most useful. The considerable funding allocated by agencies with missions directed to health must make hard choices. Hence, scientists that present results that suggest that weak EMFs may effect health are much more likely to be funded than those who find no such effects. Also, journals are much more likely to accept papers that report possible biological effects of such fields than papers that report their absence.
These pressures affect the conduct of the science that bears on biological effects of EMFs and the reported outcomes. For the most part experimental science is a complex, subtle, and difficult affair. Things often go wrong. Often good data must be selected from imperfect sets. That selection is, inevitably, subjective. I should add to these faults, excesses of self-promotion. Most scientists work to present their work in a favorable light -- I do myself. As with most advertising, this can blend into disingenuousness. Thus, the pressures that award positive results and dismiss negative results sometimes corrupt that choice and then the ensuing science.
Eliminating Error
Science is a human enterprise beset with all of the frailties of humans. As Mark Twain wrote, "All that I care to know is that a man is a human being -- that is enough for me; he can't be any worse."
Some scientific "errors" are a part of good science. An important part of research leads to the generation by competent scientists, proceeding correctly, of tentative conclusions that are understood to be on shaky ground -- and thus are often "wrong." Scientists sometimes find what might be important clues to the properties of nature that are not definitive but should, properly, be brought to the attention of the community as "hypothesis-generating results," in distinction to the more definitive results that form the solid bases of scientific knowledge. But the results of such science, more often wrong than right, are no guide to policy.
Some have emphasized the importance of the duplication of results. Competent skeptics have never been able to reproduce any of the many reported biological effects of EMFs. Indeed, while the possibility of a simple "duplication" is a necessary condition if a result is to be considered valid, it is not a sufficient condition. An attempt to duplicate errant results may lead to the same erroneous results by duplicating erroneous procedures. It is the establishment of patterns of results -- the many consequences of a hypothesis -- that forms the foundation of science. Science is not a set of discrete results, a compendium of individual facts like a stamp collection, but sets of patterns. Often single experiments generate such patterns. Such experiments need not be duplicated and seldom are. Such duplication would be a waste of resources.
In the weighing of the validity of the interpretation of an experimental result, it is also important -- usually very important -- that the interpretation does not violate well-established patterns. The pattern of nature that we call the "conservation of energy" is such a pattern, firmly established by a plethora of observations (experiments). The probability that this law fails is very, very small. Thus an interpretation of the results of an experiment that requires the violation of the conservation of energy must be considered with that small intrinsic probability in mind. If that interpretation is to be held, the probability that the result is correct must by very, very great. I call this necessary initial disbelief a "Bayesian barrier" after Thomas Bayes, who described this effect quantitatively two and a half centuries ago.
Of course, if one is only looking for a singular result as if it were a special shell in the sand by the sea, one might argue that the failure to find a true effect of EMFs only tells us that no one has looked in the right place. However, there is a strong Bayesian barrier to the possibility of any biological effects of these weak fields. Physicists know that the interaction of electromagnetic fields with matter is fundamentally very simple. The fields exert a force on charged particles proportional to the strength of the field and the magnitude of the charge -- that is all! We also know that at life temperatures, all of these charged particles are subject to ever-changing fields and forces so much larger than the weak fields and weak forces from, for instance, cell phones that the cell phone fields must go unnoticed. So physics gives us a strong Bayesian argument against any possible effect of weak EMFs.
In summary, though individual published research findings, like those which report biological effects of EMFs, are very often wrong -- either because they are badly done, dishonestly conducted, or are well-done hypothesis-generating experiments with uncertain conclusions -- the body of science is almost never wrong. But no concept is admitted into that body except after a process of establishing well-defined patterns by sifting and winnowing. That process has led to the firm conclusion that weak EMFs cannot generate biological effects and, hence, cannot affect health.
Robert Adair, Ph.D., is Sterling Professor of Physics Emeritus at Yale University. He is a member of the National Academy of Sciences, where he has served as Chairman of the Physics Section and Chairman of the Class of Physical Sciences. His researches have addressed nuclear physics, elementary particle physics, biophysics, and sports physics. On a sociologically related but physically distinct topic, see ACSH's pamphlet on Health Effects of Low-Level Radiation.
|
