Research Needs and Research Agenda
Power-frequency electric and magnetic fields of the strength found in residences have not been shown to constitute a threat to public health that would warrant an adjustment in national research policy. However, within the range of funding that might be available for issues relating to the biologic effects of electric and magnetic fields, certain avenues of research could be pursued to resolve uncertainties that remain in the epidemiologic and laboratory findings.
Epidemiologists have found an apparent weak association between childhood leukemia and wire codes, an index describing the neighborhood electric-power distribution system. Wire codes were used because they are considered an index of long-term exposure to electric and magnetic fields. Yet, subsequent attempts to replace estimates based on wire codes with measurements of electric-and magnetic-field exposure made after cancer was diagnosed have resulted in a weakening, not a strengthening, of the association. This difference in results could be resolved by further studies aimed at tracing the association, if it persists, either to some specific property of the ambient fields or to some completely different source of the disease.
From the beginning, epidemiologic evidence has been at the forefront of the concerns with possible health effects of magnetic fields. In the committee's review of the epidemiologic data, the research was found to show patterns
indicative of associations but also to show some notable inconsistencies. Additional epidemiologic studies might resolve these concerns.
Various researchers have reported anecdotally that the most highly exposed individuals often seem to drive the results of an epidemiologic study. That is, a relatively small number of subjects typically have notably higher exposures than others in the study, and these higher exposures are associated with an increased risk of cancer. The obvious inference from that finding is that conducting another study focusing on the individuals with higher exposures would be most informative. If the risk factors for exposure to electric and magnetic fields operate the same as most chemical risk factors, then individuals with the highest exposures would have the highest risks, and studies of them would have the greatest statistical power and would best address issues of dose-response relationships. Therefore, it might be particularly fruitful to consider the study of children in very-high-current-configuration homes.
Two contradictions remain: Why is no association with childhood leukemia found with spot measurements of magnetic-field strength after cancer diagnosis, even though the association is found with all the other exposure metrics, and why is no consistent dose-response relationship found in these data? With respect to the spot measurements of magnetic fields, if the results of the other exposure metrics are not spurious, two obvious potential explanations exist. First, the surrogate exposure metrics (wire codes, distances from power lines, and calculated fields) might be indicators for the true risk factor and the true risk factor is not related to magnetic-field strength. A number of such plausible risk factors have been investigated, but none has explained the observed association. Second, other surrogate magnetic-field exposure metrics might be more biologically relevant measurements of magnetic-field exposure than spot measurements. These other surrogate metrics might be more representative of long-term integrated averages of magnetic-field strength or of some other aspect of magnetic-field exposure that is related to the cause of the disease (e.g., peak field strength, field variability, frequency of transients above a given field strength, or time above a specific threshold value). What would best advance understanding of this issue are studies that address the inconsistencies, not just more studies similar to those that have been done. The committee's sample-size analyses show that a study would be unlikely to change markedly the existing pattern of results unless hundreds if not thousands of leukemia cases were included in the study.
Another research approach would be to consider different exposure regimes, an approach that might enable investigators to untangle the conflicting results. Therefore, a study of children with high exposure from sources other than outdoor distribution lines might be fruitful. People with homes near high-voltage transmission lines have high exposures and are different from most of the populations studied so far in the United States. Although such studies have been conducted in the Nordic countries, there have been very few cases because of the small populations involved. Much more high-frequency variation (i.e., more on and
off switching), reflecting the usage patterns of the source, is likely to be shown in distribution-line sources than in high-voltage lines, which impart much more consistent fields.
The areas of possible research parallel the issues discussed in this report. Those areas can be organized on the basis of the arrows shown in Figure 5-1 (see Chapter 5). Each of the items noted in Figure 5-1 is subject to uncertainty and could be clarified either by building upon completed studies or by developing entirely new studies of the issues. Countering the claim that epidemiologic studies have gone as far as they can in addressing the potential role of exposure to electric and magnetic fields in cancer etiology, it is quite likely that, at least in the near term, only further epidemiologic research can more strongly implicate or exonerate magnetic fields. That is not to argue for simply conducting more studies to reach consensus, but rather to design studies, some of a purely methodologic nature, that can address the specific gaps in our understanding.
Wire Codes and Childhood Cancer
Although the committee has concluded that an association probably exists between living in homes with high wire codes and childhood cancer, the epidemiologic evidence is not entirely persuasive for two reasons: potential control-selection bias and imprecision due to the modest number of homes within the high-wire-code categories.
Potential for selection bias could be mitigated by conducting studies in settings in which the roster of eligible controls is more readily defined (e.g., identifying and selecting controls at the approximate time that cases are identified) or in settings where a complete roster of the population is available as a sampling frame for controls. An alternative is to scrutinize the control selection procedures used in completed studies with additional analyses and possibly to acquire new data to evaluate the potential for such bias. In broader terms, the generic problem of identifying suitable control groups for case-control studies in the United States could be addressed, and the hypothesized deficiencies associated with random digit dialing need to be identified.
The limited precision of the wire-code categories of greatest interest, including the very-high-current configuration defined by Wertheimer and Leeper (1979), suggests identifying locations in which these categories are more prevalent than found in previous studies. Perhaps a systematic effort to estimate the prevalence of high-wire-code homes in different geographic locations could guide investigators in selecting a more optimal location.
Wire Codes and Confounders
The correlates of wire codes, such as age of home and sociodemographic characteristics of the home occupants, are not well understood. An effort to characterize
the neighborhood and personal correlates of wire codes would help in postulating and testing potential confounders that could account for the link between wire codes and childhood cancer. Completed studies could be examined more systematically than they have been if descriptions were given of the geographic areas and personal attributes associated with different wiring configurations.
Confounders and Childhood Cancer
As part of a broader research agenda, more knowledge of the causes of childhood cancer would be of great benefit in evaluating the role of exposure to electric and magnetic fields. Delineation of risk factors allows control for them as potential confounders. Well-defined risk factors also would provide clues regarding mechanisms for possible biologic activity of magnetic fields (e.g., the critical time periods for cancer etiology), which could then be tested. Obviously, the desire for a better understanding of the causes of childhood cancer extends far beyond the scope of the health effects of electric and magnetic fields.
Wire Codes and Magnetic Fields
In addition to the more engineering-based questions regarding sources of magnetic fields in the environment and the role of power lines as a source (see following section on exposure and physical interactions), the empirical relationships between wire codes and human exposure are a concern. Clearer understanding of the relationship of exterior power lines to patterns of magnetic fields in homes, time spent in homes by occupants, and additional sources of magnetic fields, such as appliances, could be gained; more broadly, an understanding of the sources of human exposure and the role played by wire codes and specific aspects of electric power lines could be gained. Determinants of human exposure could be understood more broadly to assess whether wire codes, though weakly predictive of average magnetic fields in homes, are more strongly predictive of some other parameter of exposure.
Magnetic Fields and Childhood Cancer
The relationship between magnetic fields and childhood cancer is the single issue of greatest uncertainty and importance in regard to future research. Two distinct research avenues for extending knowledge can be defined.
Improved Studies of Measured Residential Magnetic Fields and Childhood Cancer
As noted previously, incomplete coverage of homes in past studies calls the largely negative results into question. Combining more advanced methods of
characterizing home exposure (integration over time and consideration of grounding currents) with complete coverage of cases and controls could yield important insights into the possibility that magnetic fields account for the wire-code and cancer association.
Studies of Magnetic Fields from Sources Other Than Power Lines
Evidence that sources of fields other than those from power lines are related to childhood cancer would provide convergent evidence and strongly bolster the hypothesis that it is actually the magnetic fields associated with wire codes that influence childhood cancer risk. Other sources of exposure to electric and magnetic fields in the home, such as appliances, or sources outside the home, such as those encountered in schools, could have different potential confounders and other methodologic issues, and observed associations would be unlikely to suffer from the same validity concerns as the associations of power lines and cancer in previous studies.
Both epidemiologic and biologic studies rely on an accurate determination of the exposure variables. Characterization of the environmental fields provides a basis to test associations of epidemiologic variables with the measured field parameters and provides information on relevant features of the fields for use in biologic studies. Additional refinements in engineering techniques and in field characterization could be considered an integral part of any research agenda.
Instrumentation and Transient Currents
Instrumentation is needed that can measure more rapid changes in magnetic-field strength (changes on the order of 0.1 sec). Currently, instrumentation can capture a sample of the magnetic field every few seconds, but in biologic experiments, cells can respond to more rapid changes. For example, the ear can distinguish events that are separated by more than about 0.1 sec; it fuses events that occur more rapidly. The ability to measure magnetic-field change on the order of 0.1 sec could be used to test some of the more recent hypotheses of magnetic-field interactions. In addition, before epidemiologic or laboratory studies can be designed to test hypotheses involving transient currents, a standard way to produce and measure transient currents must be developed.
Realistic models are needed to evaluate induced electric currents and fields at the cellular and subcellular levels. An intermediate step could involve the development of better models of the body, such as the incorporation of anisotropies; the complexity of today's models is generally limited to inhomogeneities in tissues.
Studies are needed to better understand the correlation between wire codes and magnetic fields for different utility-service areas; this work is needed before multi-city epidemiologic studies (using wire codes) can be undertaken and to better interpret the studies already completed. Such data might also be useful for extending meta-analyses to more meaningful exposure estimates.
Grounding System Currents
Grounding system current is one of the three major sources of residential magnetic fields. It seems reasonable to assume that ground currents will be greater in areas with higher wire codes or near substations (the ultimate destinations of ground return currents). How grounding system currents vary as a function of distance from substations and how these might correlate with wire-code categories are questions that could be studied.
Contemporary Versus Historical Exposures
Studies of the correlation between personal-exposure measurements (or time-weighted-average measurements) taken days, months, and years apart might contribute to an understanding of the relationship between contemporary magnetic-field strengths and magnetic fields measured at the time of disease etiology.
Wire Codes Versus Contemporary and Historical Magnetic-Field Measurements
The stability of wire codes (they change little with time over reasonably long periods) suggests that wire codes should correlate as well with contemporary spot (or time-weighted-average) measurements as they would with historical (prediagnosis) measurements. However, several trends complicate that hypothesis, and studies could examine the relationships. For example, many regions tend to increase the current demand over time, causing power lines to become more heavily loaded. Eventually, the power lines in a particular area might have to be upgraded to handle the load; thus, the wire configuration is changed, and the
cycle of increased loading begins again. If the wire configurations remain unchanged over the course of a particular epidemiologic study, contemporary magnetic-field measurements might be somewhat higher than historical measurements, and the absolute magnetic-field cut points between wire-code categories would be somewhat higher than they would have been at the time of disease etiology. This trend causes no problem for the long-term validity of wire codes if the relative discriminating ability of wire codes is consistent, a point which could be examined.
There is a tendency, seen in the 1,000-homes study (EPRI 1993a) and elsewhere, for older homes to have higher wire codes, higher power-line fields and higher total fields. Possible reasons for such a trend could be that most homes built over 50 years ago and 7.1% of homes built 30-50 years ago have knob and tube wiring (which can cause higher in-home fields); none of the homes built less than 30 years ago had knob and tube wiring (EPRI 1993a). In addition, homes seem to accumulate a larger number of wiring errors as they age; wiring practices in older homes might have permitted more wiring irregularities; or power lines might be more fully loaded in older neighborhoods.
Assuming that the trend is correct, it could be interpreted to mean that in the historical period, when homes were newer, a greater spread in the magnetic fields was represented by the wiring codes. Therefore, wire codes might have somewhat greater discriminating power when applied to a historical period rather than a current period, a point which should be examined.
A serious barrier to acceptance of a possible weak connection between human health and exposure to extremely-low-frequency electric and magnetic fields in residences is the absence of a plausible physical mechanism to account for such a connection. Because any biologic effect of exposure to electric or magnetic fields at residential field strengths is, at best, at the margins of detectability, laboratory experiments to identify such a mechanism at those field strengths appear not to be feasible. However, at higher magnetic-field strengths, there appear to be genuine biologic effects, which are accessible to experiment. Mechanistic studies at high field strengths, accompanied by attention to dose-response relationships, might illuminate the situation at lower field strengths.
Laboratory experiments must be designed with appropriate positive and negative controls, conducted under blind or double-blind conditions as appropriate, and described in sufficient detail to permit independent replication. The committee found that a large fraction of published reports failed to meet those criteria.
The areas of biologic research that seem most likely to be productive include the following:
Bone healing and other therapeutic applications. The clinical application of electric-and magnetic-field exposure at relatively high field strengths for bone healing is of limited but real benefit. Additional research could clarify the molecular mechanisms involved. Better understanding of these mechanisms might allow design of electric and magnetic-field treatments for other conditions, such as osteoporosis, and lend insight into understanding possible interactive mechanisms that might result in adverse health effects of such exposure.
Characterization of the dose-response relationship for in vitro effects. Reproducible effects have been documented in cultured cells for electric-and magnetic-field-induced changes in calcium flux and gene expression, but only at very high magnetic-field flux densities or high electric-field strengths. When a robust effect can be observed in such studies, special effort should be made to define the change in effect as a function of the strength of the applied fields. The shape of the relationship between the field strength and the biologic effect must be established precisely to permit extrapolation and use in predicting effects at lower strengths. Exposure-response studies could also be extended to characterize the effects of the frequency of the applied fields and the effects of transient currents on the biologic response.
Signal-transduction events. Further replication and validation studies could be carried out to investigate the apparent effects of magnetic fields on signal-transduction events, such as Ca2+ flux, protein-kinase cascades, and membrane-receptor activities. These pathways are important in both normal and neoplastic cell proliferation and differentiation, and possible effects of magnetic fields on these pathways might be related to the observed copromotion activity of exposure to magnetic fields in animal studies.
Gene expression. Previously reported effects of magnetic fields on gene expression (e.g., changes in differentiation markers of bone cells and changes in signal-transduction effects) could be investigated further. Studies of putative direct effects of magnetic fields on transcriptional events could have low priority.
Biophysical mechanisms. Research could be directed at plausible biophysical mechanisms to explain the observed in vitro and in vivo effects at relatively high magnetic-field strengths (e.g., 1-10 G). The possible role of transient currents on electric-and magnetic-field-induced effects also could be examined.
Cocarcinogenesis. There are unreplicated data in animal studies that reported increased tumor incidence when magnetic fields were applied in combination with chemical carcinogens. Those data require replication, and if replication reinforces the reported positive results, these observations should be pursued in detailed experiments. Such experiments should focus on the dose-response relationships of magnetic-field exposures, the interacting exposures, and the temporal relationships between the different exposure conditions. Positive results could be tested in different cell or animal systems to determine whether the response is peculiar to specific biologic systems.
Magnetic-field-exposed initiated animals. Early studies that suggested an influence of magnetic-field exposure on mammary-cancer development could be extended in initiated animals. Studies should include a rigorous investigation of mammary-tumor development in exposed animals. Additionally, the possible changes observed in relevant hormonal factors in magnetic-field-exposed animals should be investigated to examine potential mechanisms related to mammary cancer.
The characteristics of electric and magnetic fields generated by the production, transmission, and use of electric energy and the possible effects of these fields on biologic systems have been the subject of extensive research for the past two decades. Due to the uncertainties and inconsistencies in the results of much of this work, Congress passed legislation in 1992 to fund an enhanced program of scientific research in this area. The Energy Policy Act of 1992 established a 5-year program of enhanced study of the characteristics of environmental exposure to power-frequency electric and magnetic fields and enhanced study of the in vitro and in vivo biologic responses to exposure to low-strength 60-Hz magnetic fields. This program is now in its third year of research and is focusing on the replication of experiments considered important to understanding the mechanism by which such fields might interact with the living system. This program is an important part of the research strategy for resolving the issues related to the possible biologic effects of magnetic-field exposure.
The work supported by the Energy Policy Act of 1992 is not anticipated to answer all the questions regarding the possible health effects of exposure to electric and magnetic fields by the program's end in 1997. Following the enhanced study supported by the Energy Policy Act of 1992, research in engineering, dosimetry, biology, and epidemiology could be initiated based on the scientific merit of the proposed work and could follow leads to plausible mechanisms that have been uncovered in previous studies. Continued research is important, however, because the possibility that some characteristic of the electric or magnetic field is biologically active at environmental strengths cannot be totally discounted. If ongoing or future research should uncover evidence of potential mechanisms that could lead to such a result, research should be continued to follow those leads and address that possibility.