SUMMARY AND CONCLUSIONS
This chapter presents the committee's assessment of the risks to human health from exposure to electric and magnetic fields. Assessment of the risk from exposure to a chemical or physical agent usually begins from a supposition—born of observation—that some hazard exists; for example, that cigarette smoking is linked to cancer and lung disease or that animal tests with chemicals of concern show excess cancer. If data are sufficient, the assessor may extrapolate these data and arrive at a quantitative estimate of the magnitude of the risk; for example, that one in seven lifetime smokers of a pack of cigarettes a day will contract lung cancer or that a specific dose-response relationship exists in at least one animal species and can be used by analogy to estimate human-health risk. That process is called risk assessment. The task of risk assessment is made more accurate and compelling if the epidemiologic and laboratory data are consistent in positively identifying the suspected agent of disease as the causal factor; a good estimate of the exposure is possible; a dose-response relationship is evident in the data (that is, if one smokes more cigarettes, one is more likely to get lung cancer); and a biologically plausible explanation exists for the relationship between the action of the presumed causal agent and the observed effect.
In previous chapters, the committee reviewed the data concerning the possible relationship between exposure to electric and magnetic fields and biologic effects in both in vitro and in vivo biologic systems and in epidemiologic studies. The conclusions reached regarding these data are summarized at the beginning of Chapters 3, 4, and 5. The set of studies that investigated the relationship between
childhood leukemia and residential proximity to particular configurations of external electric wires has captured the interest and concern of the public, scientists, and government officials. The committee carefully considered the reports of weak positive associations and found them consistent and not explainable by other known factors. This chapter considers the use of those data for estimating the nationwide childhood cancer risk from residential exposure to magnetic fields.
The conclusions in this chapter about the possible risks of exposure to electric and magnetic fields relate mainly to cancer, as is typical in risk assessments. At the end of the chapter, the committee comments on what the data might say about any possible association between exposure to magnetic fields and other health-effect end points, such as adverse effects on reproduction or on learning ability.
Risk assessment is a method designed to evaluate human-health risk (NRC 1983; Wartenberg and Chess 1992). Together with information about costs, benefits, and sociopolitical concerns, risk assessments are used to formulate policies to reduce risks, a process called risk management. The basic tenet of risk assessment is that data on health effects detected in small populations exposed to high concentrations of suspected hazardous agents, usually chemicals, can be extrapolated to predict health effects in large populations exposed to lower concentrations of the same agent. Thus, if workers in a factory are exposed to a high concentration of the solvent toluene and no one develops cancer after follow-up testing for a sufficient number of years, toluene can be inferred to be less carcinogenic than another substance known to produce cancer in workers. Furthermore, if community residents are exposed to toluene at a low concentration, it can be inferred that they will not incur a substantial increased risk of cancer either.
In concept, risk assessment is meant to be an objective approach to risk evaluation for making informed public policy. Although components of the method allow for some flexibility, permitting risk assessors to accommodate a range of circumstances, many federal agencies have explicit guidelines for carrying out risk assessment. Thus, some federal officials believe that little room is left for subjective judgments if agency guidelines are followed. In practice, however, many staunch supporters of risk assessment acknowledge the imprecision of the technique, recognizing that it entails a considerable measure of subjective judgment.
Risk assessment is a four-stage process (NRC 1983). The goals of the first stage, known as hazard identification, are to catalogue situations in which an agent can pose a risk to human health and to predict all possible adverse health effects. It is meant to address a hazardous agent regardless of the amount present
and to characterize all possible health end points. At a later stage, specific agents or health effects can be excluded if they are deemed to be unimportant.
The second stage of risk assessment is dose-response assessment—determining how much exposure to a given hazardous agent is harmful to public health. For carcinogens, the results of this stage of risk assessment typically are reported as the ''cancer potency." That is, risk assessors fit a mathematical equation to the experimental data to describe how the risk of disease increases with the amount of an agent to which a person is exposed and report a standard index of that rate of increase.
The third stage of risk assessment is exposure assessment: investigators estimate the amount of a given agent a typical person is likely to encounter. These data can be obtained in many ways, including direct measurement, historical records of releases, and extrapolation of similar processes occurring in other locations and adjusted for the specifics of the current study. The final result is a combination of where a person spends time and what activities a person engages in coupled with the source, concentration, and movement of the studied agent.
In the fourth stage of risk assessment, risk characterization, information from the three other stages is combined into a single overall estimate of risk. For the agent identified in the first stage, the assessor determines in stage two if a dose-response relationship exists, and in stage three, quantifies the exposure. That exposure is then multiplied by the potency estimated in stage two to derive a predicted risk of cancer. In this final stage, a quantitative risk assessment is made; that is, a numerical estimate of the magnitude of human risk.
The assessments from all these stages are combined with other insights, such as the presence or absence of a biologic explanation for the relationship between exposure and effect, to reach a judgment about the overall concern warranted by exposure to an agent. The strength of individual studies and of groups of studies is weighted by assessors when making risk-assessment judgments. More weight might be ascribed to studies that have been replicated in different laboratories and to studies that are well designed and carefully performed.
In the case of exposure to magnetic fields, the most compelling data available are derived from human studies rather than animal experiments. The most reliable positive laboratory data come from studies of mammary-cell tumors in test animals (Beniashvili et al. 1991; Mevissen et al. 1993; Löscher et al. 1993). However, tumors occurred only when the animals were treated with a chemical carcinogen or ionizing radiation before exposure to the magnetic field, and even then, the tests did not show a classic exposure-effect relationship. Given that exposure to electric and magnetic fields is not genotoxic and that this type of exposure is qualitatively and mechanistically different from both chemical exposure and ionizing-radiation exposure, the committee felt it was premature to use those data further.
Using human-response data from epidemiologic studies of exposure to low field strengths presents a somewhat different situation for evaluation and extrapolation
than that typically encountered in risk assessment. In terms of more accurate and precise estimates, the human data obviate the need to extrapolate from animals and, if the studies identified earlier in this report are used, extrapolation from high experimental doses to low environmental exposures is unnecessary because typical environmental exposures were used in those studies. In terms of sources of error that are not taken into account, the committee has added considerations of variation from person to person, which is controlled for genetically in animal studies; variation of dose and inaccurate measurement of dose, which are prescribed and carefully assessed in animal studies; and failure to adjust for each subject's characteristics (e.g., ethnicity or socioeconomic status) or exposure to other agents (e.g., ionizing radiation) that are known to modify the effect of exposure or directly cause the disease under study. Adjusting for these effects is not a part of traditional risk assessment, although failure to do so might result in inaccurate estimates of risk.
After reviewing the available data, the committee does not believe it is appropriate to perform a complete assessment of the risks of exposure to power-frequency electric and magnetic fields through the four formal stages described above. The committee believes that the data are too uncertain to result in a meaningful analysis. Some members of the committee considered the data so inconclusive as to preclude any attempt at risk assessment. Others believe that, even though a quantitative prediction of risk is technically possible, the interpretation of the resulting risk number would be problematic and likely lead to misinterpretations. However, many persons are concerned about the possible risks of exposure to residential electric and magnetic fields, and in light of its examination of the entire body of evidence, the committee recognizes the need to contribute to a better understanding of the risks by assessing them in a more limited context. That is attempted below. These observations might help put the risk into a context that can be considered for possible personal actions or government policies. The formal framework of risk assessment is used to the extent possible.
The conclusions presented in previous chapters concerning the possibility of risk from exposure to electric and magnetic fields are the following:
There is no evidence of effects of electric-and magnetic-field exposure on bacterial DNA. Negative results from such tests are generally accepted to imply that DNA was not damaged by the agent to which the bacterial DNA was exposed.
Cancer is generally assumed to be initiated because DNA has been damaged; therefore, findings of effects, or no effects, in this type of test might be important (subject to verification by tests in other biologic systems) in identifying an agent as hazardous.
There is no consistent or convincing evidence of effects of electric-and magnetic-field exposure on cell systems cultured in the laboratory that might imply a human-health effect at typical environmental exposure levels. Although exposure to electric and magnetic fields at 50-60 Hz has been shown to induce changes in cultured cells, the exposure conditions exceed ordinary human residential exposure by factors of 1,000 to 100,000 times. In studies at exposure levels similar in magnitude to residential exposure, no effects have been reported and replicated.
Results from tests performed using mammalian cells cultured in the laboratory have broader implications than studies using bacteria. The cell systems are closer in their complexity to those of humans, and measurements of effects on organelles, such as chromosomes, that are present in humans can be performed.
There is no consistent or convincing evidence of effects of power-frequency electric-and magnetic-field exposure on whole animals that might imply a human-health effect at typical environmental exposure levels. Laboratory animals have shown neurobehavioral and neuroendocrine changes in response to electric and magnetic fields. However, these responses are not convincing evidence of adverse effects for humans, and they occur at exposure levels well above the human experience. Results obtained on whole animals, when the animals tested respond in a way analogous to the way humans might respond, are important in determining risk for human beings.
There is a moderately consistent, statistically significant association between wire codes, an indirect measurement of electric-and magnetic-field exposure, and childhood leukemia. Average magnetic fields measured in the home after diagnosis of disease have not been found to be associated with excess childhood cancers.
Studies have not identified the factors that explain the association between wire codes and childhood leukemia. Wire codes are not strong predictors of exact magnetic-field strengths in the home, though they do distinguish high versus low field strengths reasonably well.
There is a moderately consistent, statistically significant association between indirect measurements of occupational exposure to magnetic fields and both leukemia and brain cancer. The associations have been primarily with job titles that are expected to provide a greater than average exposure to magnetic fields; in some more recent studies, measurements have also been made of contemporary magnetic fields. Although the magnitude of the risks vary from study to study, most studies show increases in those two cancers.
Evidence from epidemiologic studies is the most important class of data when performing a risk assessment. The studies involve humans, and the results of the studies can be applied directly. In other types of studies, even when the results are positive, the findings are indirectly related to humans and must be extrapolated from the tested biologic system to humans.
In general, hazardous substances exert their effects in proportion to the amount of exposure received—for example, two packs of cigarettes per day should be more hazardous than one pack per day.
In reviewing the data, the committee perceived no clear indication that exposures to electric and magnetic fields of different magnitudes could be related to variations in responses in systems tested in laboratory studies. The epidemiologic data, however, are open to varying interpretations. Some study results have been interpreted as showing a weak dose response; that is, some analyses indicate that those groups of subjects with higher rates of disease are more highly exposed. Nonetheless, the committee concludes that electric-and magnetic-field-exposure data overall do not provide sufficiently convincing evidence of a dose-response relationship to use these data to develop a mathematical model.
This finding is important because a demonstration of a dose-response relationship can serve as a strong confirmation that any effect that might have occurred is real and not an artifact of the experiment.
The epidemiologic data that support a dose-response relationship are sparse, and the nature of the studies and the data published preclude rigorous analysis.
It is essential when assessing risk to ascertain whether humans or test species have been exposed to the putative causative agent and to quantify the extent of their exposure. There is no doubt that humans are exposed to electric and magnetic fields in their daily lives. In fact, exposure is so universal and unavoidable that even a very small proven adverse effect of exposure to electric and magnetic fields would need to be considered from a public-health perspective: a very small adverse effect on virtually the entire population would mean that many people are affected.
Extensive data are available about human exposure to 60-Hz magnetic fields and, to a lesser extent, to electric fields in various environments, including the residential environment. Surrogates for magnetic-field exposures, such as wire codes, have been used in some epidemiologic studies. Typical values for U.S. populations, such as the percentage of homes in various exposure categories, are presented in Chapter 2.
It is critical to the understanding of our risk assessment to recognize that the epidemiologic studies showing an association between wire codes and childhood leukemia do not establish an association between directly measured electric and magnetic fields and disease, because wire codes have not been validated as an appropriate indirect measure of the fields.
The conclusions reached in previous chapters of this report and the summary risk assessments made above indicate that the data on the effects of exposure to electric and magnetic fields on biologic systems are either negative or so uncertain that making such an estimate would be injudicious and misleading, though a theoretic risk assessment could be performed.
Biologic Mechanism of Action
As mentioned above, scientists understand cancer to be associated with damage to DNA. Thus, if an agent is found to damage DNA, the carcinogenic potential of that agent has been established as biologically plausible. That is an example of a general principle—in attempting to assess the risks of an agent, one should examine whether some biologically plausible means exists for that agent to cause disease.
No biologically plausible explanation for a putative relationship between exposure to electric and magnetic fields from power lines and an adverse effect in biologic systems has been proposed, tested, and shown to be valid.
The occurrence of a copromotional effect of exposure to very-low-frequency electric and magnetic fields, which has been reported in cell systems and animals, deserves consideration. Cancer is thought to occur after an initiator (a chemical or physical agent) has damaged DNA and started a process leading to the disease. A copromoter might not have the ability to initiate the cancer process, but the risk of cancer can increase when the test biologic system is subjected to a copromoter after the system has been exposed to an initiator.
Overall Conclusions for Risk Assessment
The body of evidence, in the committee's judgment, has not demonstrated that exposure to power-frequency electric and magnetic fields is a human-health hazard. However, some epidemiologic data support an association between surrogate measurements of magnetic fields and an increased risk of childhood leukemia. Further research for understanding the various ways of measuring exposure and their possible association with adverse health outcomes in model and human systems will be needed to resolve the uncertainty.
The committee's overall conclusion is based on the weight of the evidence after review and analysis of biologic data at the molecular, cellular, and whole-animal level that are considered relevant in assessing the possibility that exposure to electric and magnetic fields in the environment causes cancer in humans. First, in vitro studies were found to observe biologic effects at field strengths that are 1,000 or more times greater than would be experienced in residential situations. Even then, the results are inconsistent and often not reproducible. The demonstration
of in vitro effects does not necessarily imply potential adverse human-health risk. Animals see because a neural effect occurs; light enters the eye, strikes the retina, and the brain forms messages on the basis of information it receives. Yet, the effect of vision is not a deleterious human-health effect, even though an effect, indeed, has been demonstrated. For example, although tests have shown that many animals can detect electric fields at 1 to 5 kV/m, there has been no evidence that these effects imply a health hazard.
Exposure to electric and magnetic fields is not genotoxic at any level of exposure. After considering the effects of exposure to electric and magnetic fields on laboratory animals, the committee concludes that the effects of such fields have no consistent pattern as a direct carcinogen. Evidence, not yet replicated, does show that exposure to electric and magnetic fields combined with exposure to high concentrations of known carcinogens increases the number of rat tumors and accelerates their appearance. Such evidence does not identify electric and magnetic fields as a probable carcinogen.
Finally, after analyzing the epidemiologic data, the committee concludes that the evidence of an association between exposure to electric and magnetic fields and cancer is not convincing, although residential wire codes have been associated with cancer. In addition to these data, no plausible biophysical mechanism has been identified that would suggest that the action of electric and magnetic fields is carcinogenic.
The data at different biologic complexities, taken in total, do not provide convincing evidence that electric and magnetic fields experienced in residential environments are carcinogenic. No tests or studies can prove that an agent is not carcinogenic at some dose level, in combination with some other biologic agent, or for some sensitive populations of humans. All that can be stated is that, under the exact experimental conditions of an extremely large number of studies, exposure to power-frequency electric and magnetic fields at environmental strengths does not produce patterns of data similar to those found for other agents that have been shown to be carcinogens. Possibly, such fields can be hypothesized to act as a nongenotoxic cocarcinogen or to act through hormonal pathways to suppress protective molecules, such as melatonin. Although such hypotheses must be tested carefully when scientific justification exists to do so, the overall conclusion that must be drawn from all the data in the studies of cells, lower animals, and humans is that the data are negative or inconclusive. Electric and magnetic fields are neither genotoxic in cells, nor a direct carcinogen in animals, nor associated conclusively with cancer in exposed humans.
The association between residential proximity to high-wire-code configurations and increased rates of childhood leukemia remains unexplained, as do the associations between occupational exposures and leukemia and brain cancer. Positive human epidemiologic data are the strongest evidence in evaluating any human-health risk. The associations for childhood leukemia have been shown to be statistically reliable and robust findings that must be considered carefully in
drawing conclusions about overall risk. Uncertainty is introduced because the associations found with wire codes are not found with measured fields, and that raises serious questions about the interpretation of the positive findings and their use in quantitative modeling. The human epidemiologic studies stand alone as suggesting possible adverse health effects, and the results themselves indicate small risks (e.g., relative risk of 1.5) relative to other adverse exposures that epidemiologists consider.
Other Possible Human-Health Effects
The committee was asked to examine not only data for cancer but also data from studies investigating a possible association between exposure to electric and magnetic fields and health effects related to reproduction and development and neurologic effects expressed as learning or behavioral disorders. Although fewer studies of these effects have been conducted, the committee concludes that no studies to date have shown an association between exposure to residential electric and magnetic fields and an adverse human-health effect.
The following conclusions are based on studies with animals:
There is no convincing evidence of an association between exposure to power-frequency electric and magnetic fields and reproductive or developmental effects.
There is no convincing evidence of an adverse neurobehavioral effect in association with exposure to residential electric and magnetic fields.
The following conclusions are based on studies in humans:
There is no indication of an adverse human-health impact from exposure to power-frequency electric and magnetic fields, although there is some evidence for electric-and magnetic-field-induced neuroendocrine changes.
The data from epidemiologic studies on exposure to electric and magnetic fields and adverse pregnancy outcomes do not support the existence of an association. Epidemiologic studies have not been performed with the specific aim of determining the existence of neurobiologic effects, and no statements can be made regarding the occurrence of this human-health effect.
There is no consistent or convincing evidence of adverse effects of power-frequency electric and magnetic fields on reproduction or development in human studies. However, the number and quality of the epidemiologic studies is limited, making any inferences tentative.