EPIDEMIOLOGICAL RESEARCH RELEVANT TO IDENTIFICATION OF HEALTH HAZARDS ASSOCIATED WITH GWEN FIELDS
An extensive, but fragmented, literature deals with the potential health effects of extremely-low-frequency (ELF) magnetic fields (50-60 Hz) encountered in occupational and residential settings. Investigators have been following at least three different avenues of inquiry: occupational hazards, general environmental hazards, and cancer etiology. Research in all those fields has been hampered by technical difficulties in measuring actual exposure to ELF fields. Because of the difficulty in measuring exposure to ELF fields, most investigators have had to study ecological associations of magnetic fields and disease rather than exposure of individuals with and without disease to magnetic fields (case-comparison studies) or disease incidence among individuals with smaller and larger exposures (cohort studies). Another set of measurement problems, those related to morbidity, has led to serious concern about the influence of ascertainment bias on the results of the few studies that looked at diseases other than cancer. Most of those studies have tended to concentrate on cancer end points in population-based registries and control subjects. The entire health-effects literature was reviewed by Coleman and Beral1 and Gordon et al.2 and expanded and updated by Tenforde,3 Theriault,4 and Poole and Trichopoulos.5 This committee endorses the general conclusions of the three 1991 reports, and this chapter summarizes the conclusions and highlights the results of key investigations. It does not provide a comprehensive literature review.
STUDIES OF GENERAL ENVIRONMENTAL EXPOSURE
Most of these studies measured residential exposure in one of two ways: by direct measurement of magnetic fields in houses or multiple-family dwellings or by use of wire codes, with electric-current configurations (high-current or low-current) as indirect measurements of magnetic-field exposure.
The archetypal residential-exposure study was published by Wertheimer and Leeper.6 It used a case-control design and measurements of wiring configuration to consider 344 cancer deaths of newborn to 18-yr-old children recorded in Denver, Colorado, between 1950 and 1973. The dead children were compared with 344 living children of the same ages. The configuration of the wiring around the home of each dead or matched child was coded “high current” or “low current” according to the types of neighboring electrical distribution lines. For example, a home less than 40 m from large-gauge primary wires was coded as “high current”. A major flaw in the study was that wiring configurations of households were not blindly classified.
The address recorded as the place of death was the residence of each dead (case) and matched living (control) child 2 yr before the diagnosis of each dead child. A total of 491 case and 472 control addresses were coded. The homes of 182 of 309 dead children were classified as having high-current configurations, compared with 103 of 369 homes of living children--a relative frequency of 2.11. The relative frequency proved to be rather consistent and significantly different from unity for the results of a number of analyses of subsets of the data, e.g., for birth and death addresses of case and control children 0-5 yr old, for those 6-18 yr old, for children of all ages who had leukemia and their controls, and for children who had nervous system tumors and their controls.
That startling result led others to attempt to replicate the findings. One of the first groups, Fulton et al.7 in Rhode Island, used 119 cases, defined as leukemia patients with onset before age 20 and year of onset between 1964 and 1978. Controls were matched, two to one, by birth year. An attempt was made to use the same scheme as Wertheimer and Leeper6 to code the electrical distribution-line configuration for the residences of all cases and the birth residences of all controls. Fulton et al. coded 198 case and 225 control residences and found no association between leukemia risk and high-current wiring configuration. Wertheimer and Leeper8 later reevaluated the data of Fulton et al. and suggested that the coding scheme was not exactly comparable with their own and that electrical distribution systems are different in Rhode Island and Colorado. Moreover, Wertheimer and Leeper asserted that, if the analysis was restricted to data on children 8 yr old and older, marginally significant increases in risk could be seen.
Shortly thereafter, a Swedish study by Tomenius9 made more direct measureof exposure and coding wiring configurations. Tomenius studied 716 incident cases among people 0-18 yr old, born in Stockholm county, and diagnosed there during 1958-1973. Each case was matched to a control by age, sex, and church district of birth. Birth and diagnosis addresses were determined for cases and controls (for a control, the address when the matching case was diagnosed). A total of 1,129 case and 969 control dwellings were visited by the investigators. “Visible electrical structures” were recorded at each address, and magnetic field strengths were measured. A statistically significant odds ratio (3.7; P < 0.05) for nervous system tumors was observed when dwellings with measured field strengths greater than 0.3 µT were compared with those with field strengths less than 0.3 µT. However, a decreased leukemia risk of 0.3 bordered on significance. There also appeared to be a generally increased risk of cancer (odds ratio, 2.1) for dwellings within 150 m of 200-kV transmission lines. Because many of the dwellings near 200-kV lines were also above 0.3 µT, those results are not wholly independent.
Savitz et al.10,11 conducted a study on childhood cancer in the Denver, CO, area in which the residence at the time of diagnosis was measured for the strength of the electric and magnetic fields and the current configuration of the surrounding electrical
distribution wires was coded using the method of Wertheimer and Leeper.6 Whenever possible, prior residences were also measured and coded. In contrast with the earlier study,6 the assessment and classification of homes were blind. The Savitz et al.11 study included 356 children aged 0-14 yr who were residents of the 1970 Denver Standard Metropolitan Statistical Area and had a cancer diagnosis during 1976-1983. Controls, selected with random-digit dialing, were similar to the cancer children in age, sex, and geographic area of residence. Residential histories were obtained during interviews. Odds ratios of between 1.17 (appliances on) and 1.49 (appliances off) were obtained for homes with measured 60-Hz magnetic fields of greater than 0.25 µT. Associations with wire configurations were much higher. The odds ratio was 2.2 for “very high” current configuration, compared with buried cable, when the residence at the time of diagnosis was considered and rose to 5.22 when the same comparison was made for the residences occupied by cases 2 yr before diagnosis. The last result was statistically significant.
London et al.12 conducted a case-control study of electric and magnetic fields and childhood cancer in Los Angeles. The basic population consisted of 232 cancer children, drawn from the Los Angeles County Cancer Surveillance tumor registry, and 232 controls, some drawn from neighbors of the cancer children and some from random-digit dialing surveys. Attempts were made to match cases and controls by age, sex, and ethnicity, but this was not always possible; in the case of random-digit dialing, no match could be obtained in all cases. When a match could not be obtained readily, matching criteria were relaxed first on ethnicity, then on sex, and last on age. Efforts were made to obtain spot and 24-h readings of electric and magnetic fields in the residences of cases and controls and to characterize the configuration of distribution wiring around the residences with the coding scheme of Wertheimer and Leeper.13 Odds-ratio analyses of 24-h magnetic fields (162 cases and 143 controls) and spot measurements of magnetic fields (140 cases and 109 controls) showed no association of leukemia risk with increased magnetic or electric field strength. The one significant odds ratio (2.15) was observed in a comparison of underground and very-low-current wiring-configuration residences (31 cases and 38 controls) with very-high-current configuration residences (42 cases and 24 controls) on the Wertheimer-Leeper coding scheme. The authors suggested that the results support an association between leukemia risk and wiring codes, but not with either electric or magnetic fields. Their discussion reflects the uncertainty of whether lack of association with magnetic fields is a result of deficiencies in field measurements (measuring the wrong thing) or the observed association of risk with wiring codes is the product of some unidentified confounder.
Wertheimer and Leeper13 also conducted the first of a relatively small series of studies on residential fields and adult cancer. They used a case-control design and four series of cases. The first two series of cases were all people who died of cancer in 1967-1975 in the towns of Longmont and Boulder, Colorado (plus a few cancer survivors living in these towns found through the Colorado cancer registry). The other
two series, in Denver and the Denver suburbs, included both dead and surviving cancer patients. A control matched by age, sex, and year of death was selected for each dead patient; and a living neighborhood control for each survivor. Analysis was restricted to subjects whose history could be traced for at least 4 yr before diagnosis. It included 194 cases and controls in Longmont, 321 in Boulder, 255 in Denver suburbs, and 409 in Denver. The electric-wiring configuration was coded for the residence in which the subject had spent most of the 3-10 yr before diagnosis. Control addresses for Longmont and Boulder were chosen in the same manner for the same 3-10 yr. The addresses chosen by random-digit dialing to recruit the Denver controls were used for measurement purposes. As in the authors' earlier study, there was no blinded coding and classification of homes. Comparison of very-high-current configuration with low-current configuration in all four series showed odds ratios of 1.5-2. The analysis suggested that at least three of the comparisons were statistically significant (the authors used a “C” statistic, based on a sign test, which is less powerful than the more conventional chi-squared test).
Another study of acute nonlymphocytic leukemia was conducted by Severson et al.14 in western Washington. Wiring configurations and direct measurements (short-term and 24 h) of 60-Hz electric and magnetic fields were recorded. The case series consisted of people aged 20-79 yr old, diagnosed with acute nonlymphocytic leukemia during 1981-1984 in King, Pierce, or Snohomish County. The control group, matched by sex and age, was selected from the same three-county area with random-digit dialing. Approximately one control per case was selected. Subjects were interviewed for a complete residential history. All addresses occupied by cases during the 15 yr before diagnosis and on a randomly assigned reference date by control were coded for wiring configuration, and the 60-Hz electric and magnetic fields were measured for the current residence at the time the interview occurred (or for the last residence of dead subjects). A total of 114 cases and 133 controls were interviewed. An analysis of the data showed no association of leukemia risk with wire code (odds ratios, 0.60-1.36). A similar analysis based on measured fields also showed no significant association with leukemia risk. The latter result was criticized by Wertheimer and Leeper15 on the grounds that measurements were made during only 39% of case interviews and 70% of control interviews. They also asserted that the lack of association with wiring configuration was due to a high degree of nondifferential misclassification of exposure.
McDowall16 reported the results of a population-based cohort study in the United Kingdom. The cohort was identified with maps of East Anglia and data from the 1971 British Census to enumerate the population living in the vicinity of “electrical installations.” A total of 2,839 dwellings inhabited by 7,920 people were identified, and this cohort was traced through 1983. A total of 7,631 people were successfully traced, of whom 814 had died during 1971-1983. The overall standard mortality ratio (SMR) was 89, which reflected a low death rate from circulatory disease. Only one cancer SMR--for lung cancer in women--was found to be significantly increased. The observation that leukemia in women and leukemia in both sexes had above-average
SMRs among people living within 15 m of an electrical installation was interesting, but the results were not statistically significant.
In another United Kingdom study, Youngson et al.17 estimated exposure to the field strengths due to maximum load currents from overhead power lines. They correlated the estimates with measurements of magnetic fields made on the premises of home addresses of a sample of people in their study living close to power lines. For analysis, they used estimates of power flowing through the lines during the 5 yr before diagnosis of their cases. Of 1,511 people 15 yr old or older who were diagnosed with hematological malignancy between January 1, 1983, and December 31, 1985, 1,491 resided within the North West Regional Hospital Authority boundaries of England and Wales. Matching controls (by sex, age, and year of diagnosis) were randomly selected from computer lists of all inpatient discharges in the same area. Matched-pair analysis, with conditional logistic regression, showed no statistically significant associations of disease with either magnetic-field exposures or distance of homes from power lines. However, as in many other studies, the odds ratios for subgroups of cases (versus their controls) living within 30 m of overhead transmission lines verge on statistical significance. Of more importance for the present report is the finding of a nonsignificant trend with distance from the power lines, which suggested an increased risk--a relative risk of about 1.3--associated with larger magnetic fields or with proximity to overhead power lines. However, on the basis of both direct and indirect measurements of exposure, the investigators stated that their study provided no consistent evidence that distance from power lines correlates accurately with direct measurements of magnetic-field exposure.
Those selected studies have been described in enough detail to emphasize the overwhelming dilemma faced by investigators who want to answer the question, “Is any health effect associated with exposure to power-frequency magnetic fields?” Conceptually, direct measurements of electric and magnetic fields in homes must be better assays of the exposure of people living in those homes than are territorial distribution wire codes. Furthermore, 24-h measurements must be better than spot measurements of direct exposure. On balance, the more direct and the better the measurement of exposure to magnetic fields, the smaller the likelihood of detecting increased risks of leukemia in either children or adults.
Studies that included all cancer deaths, rather than just leukemia in adults, have had to deal with the additional bias of the types of death involved, because leukemia incidence is overwhelmed by the incidences of the most common and most lethal cancers, such as those of the lung and colon. In this context, the results of the relationship (or lack of relationship) of cancer deaths with measured field exposures are even more difficult to interpret.
In contrast, classification of residential addresses according to wire-code configurations--the least direct and most tenuous measurement of a person's field
exposure--suggests that children with leukemia are more likely (perhaps twice as likely) to be found in homes coded as having high versus low current configurations.
There are at least two possible interpretations of these dichotomous results. Either wire code is a better indicator of long-term exposure to magnetic fields than is more direct measurement of those fields, or wire code is a surrogate of some household exposure other than exposure to the ambient 60-Hz magnetic field. The wire code could also be a surrogate of demographic and lifestyle characteristics of residents in the coded households that are associated with higher risks of leukemia. The former might well be a function of the age of the buildings.
This report addresses a much more limited question: “Do the studies provide any evidence that people living close to GWEN sites will have an increased risk of developing leukemia, other cancers, and other diseases?” The reported studies do not provide any evidence that any measurable risk is likely to be due to GWEN-related exposure.
Occupational studies have used job codes and titles as indicators of worksite exposure to magnetic fields. The general conclusions of a current comprehensive review of studies of Occupational exposures by the National Institute for Occupational Safety and Health18 are endorsed by this committee, including its lack of confidence in occupational codes as adequate indicators of exposure. Because of that lack of confidence and because the results of occupational studies are less relevant than residential studies to the general environmental setting that is central to the present report, those studies are described here in much less detail.
Milham19,20 using proportionate mortality ratio (PMR) inventories, showed excess leukemia associated with the occupation “electrical worker” and excess brain cancer with the occupation “electrician.” Case-control studies have been conducted of brain cancer21,22,23 and leukemia;24,25 they typically report increased odds ratios--1.5 to 3--for such job titles as “electrician” or “welder.” Cohort studies have also been carried out on workers in the electrical industry;26,27,28 these studies have led to reports of excess cancers in many anatomic sites, from melanoma to male breast cancer.
The epidemiological results have generated a number of useful hypotheses; e.g., if ELF magnetic fields interfere with melatonin metabolism, excess breast cancer29 might be expected. However, the literature is of limited value for risk analysis. First, there have been inconsistent increases in cancers of different sites, even though, as in the residential studies, brain cancer and leukemia appear to be possible candidates. More important, the excesses detected are related to occupation, not to exposure. That is, electricians might be at excess risk for brain cancer, but this does not implicate ELF
magnetic-field exposure as the causative factor. However, the hypotheses raised by the studies should be considered further, because some measurements of worker exposure suggest increased exposure to ELF magnetic fields, but the measurements have not yet been made in the context of epidemiological research and cannot be related to specific disease risks.
EPIDEMIOLOGIC STUDIES OF HEALTH EFFECTS OF MICROWAVE EXPOSURE
Another relevant data source, epidemiologic studies of populations exposed to radar and microwave, is rather limited. Two populations, U.S. naval personnel exposed to radar and American embassy personnel in Moscow, dominate the literature. The first group of studies, reviewed by Tenforde and Budinger,30 consist of cohort studies of high-exposure and low-exposure groups. One such study31 followed a high-exposure group of 20,109 Korean War veterans involved in radar repair, compared with a similar group of 20,781 veterans involved in radar operation and considered to have received low exposure. No excesses of either cancer or cataracts (hypothesized to be related to occupational exposure to microwaves) were seen in the high-exposure cohort. Another study of occupational exposures to radar showed a similar lack of effect.32
Szmigielski et al.33 reported a significant elevation of cancer incidence rates among Polish military personnel exposed to radar relative to nonexposed personnel. An elevated cancer rate was observed both for all cancers combined and for specific cancers of the lymphatic and hematopoietic system. Cancer rates were reported as incidence rate per 100,000 per year for the period 1971-1980, and no data were presented on the actual numbers of cases in the exposed and nonexposed groups. The design of this epidemiological study and the methods used for data analysis were not adequately described to permit a judgment on the reliability of the results.
The Moscow embassy studies involved followup of 1,827 persons who had served in the embassy from 1953 to 1976, during which period the embassy was exposed to radiofrequency (RF) power densities of as much as 15 µW/cm2 for 18 h/d. The morbidity and mortality experience of this group was compared with that of a group of 2,561 persons who had served as employees at eight unirradiated East European embassies or consulates during the same period. A variety of health end points were compared, but no significant exposure-related effects were observed.34
Another negative study of cancer associated with microwave radiation was conducted by Selvin et al.35 It examined whether cases of bone, brain, and lymphatic cancer in persons less than 20 yr old in San Francisco from 1973 to 1986 might be clustered around a large microwave-transmission tower. The study was based on 30 cases of bone cancer, 27 cases of brain cancer, and 26 cases of lymphatic cancer. Clustering was shown for lymphatic cancer, but not centered near the tower. No
significant clustering was shown for the other two cancers. Thus, the study was negative.
Because the studies in question are of rather low power and lack precise measurements of exposure and because radar and microwave radiation have frequencies very different from those of GWEN fields, the findings do not offer much insight as to whether the GWEN system poses any risks to public health.
RADIO BROADCAST STATIONS
One study that has direct relevance to RF exposures near GWEN frequencies was performed in the workforce of the Meadowlands Sports Complex (MSC) in New Jersey.36 This study was prompted by a cancer cluster that occurred on the New York Giants football team between 1980 and 1987. Because there were three AM broadcasting stations located at the MSC and five more within 2-3 miles, measurements of electric field strengths were made at MSC; it was found that they were in the top 0.1% of field strengths measured in urban areas in the United States. The root summed square for all AM signals ranged from 0.013-9.63 V/m.
The study involved 147 cancer cases that occurred in the Meadowlands workforce during 1978-1987. The expected numbers of deaths from various types of cancer were calculated with New Jersey mortality data. Two significant proportionate cancer incidence ratios (PCIRs), based on one case each, were observed for male breast cancer and nasopharyngeal cancer. Significantly increased PCIRs were not observed for either brain cancer or leukemia. Although this result does not rule out odds ratios of 2 or less for these cancers, the study had sufficient power to detect odds ratios of 5 or more with a probability of about 90%.
A proportionate-mortality analysis based on 225 deaths and 65 cancer deaths was performed in the study. The overall proportionate mortality ratio (PMR) for all cancers was almost exactly equal to 1. Because all cancer PMRs as small as 1.5 would likely have been detected, that result strongly suggests that no large general cancer increase was present. The generally negative findings of the study support the view that exposure to high AM radio-broadcast field strengths does not result in large excess cancer risks.
Another study on cancer incidence in relation to RF exposure from radio and television broadcast stations was conducted in the state of Hawaii using data from the Hawaii Tumor Registry.37 Cancer incidence rates were examined for nine census tracts with broadcast towers (in Honolulu and Waikiki) and for two tracts without towers (more centrally located on the island of Oahu). The standardized incidence ratio for cancer was significantly elevated for both males and females living in tracts with broadcast towers relative to the general population of Hawaii. No significant elevation
in cancer incidence was observed for the two populations living in census tracts without towers. This study was very weak due to the inability to simultaneously adjust the data for race, age and sex (due to the small sample sizes), or to account for differences in expected cancer rates in rural areas (without towers) relative to urban areas (with towers). In addition, no exposure data were presented and the RF exposure consists of a mixture of AM and FM signals. As a consequence, the results of this study cannot be regarded as conclusive.
AMATEUR RADIO OPERATORS
Milham38,39,40 reported that amateur radio operators in the states of Washington and California have a significantly elevated risk of mortality from acute myeloid leukemia, multiple myeloma and other neoplasms of the lymphoid tissues. Mortality data were obtained from the death certificates of members of the American Radio Relay League whose deaths were reported in the League's magazine. Comparison data were based on the 1976 report of U.S. age-specific white male death frequencies. The overall death rate of the amateur radio operators was significantly lower than that of the general population for all causes of mortality and for all classes of cancer combined. Because Milham's study provided no estimate of exposure to either RF or ELF fields among the amateur radio operators, the results cannot be considered directly relevant to GWEN fields.
Reports of occupational studies have generated useful hypotheses, but so far none has been helpful in specifically testing hypotheses or quantifying risk. It is even harder to conclude how much has been learned about hypothesized risks from the studies of residential exposure. Certainly, the two Wertheimer and Leeper studies6,13 suggest that a consistent hazard is associated with exposure to “high-current” configuration wiring codes. However, it is not clear that high-current configuration wiring codes are closely associated with magnetic-field exposure. Poole and Trichopoulos,5 citing Kaune41 and Barnes,42 point out that the observed correlation coefficient between magnetic field and wiring code rarely exceeds 0.6 and is more often in the range of 0.4 - 0.5. It could be argued that nondifferential misclassification, which reduces the apparent strength of associations between diseases and their causes, is masking a true causal link between magnetic-field strength and leukemia.5 Results of studies that have measured both field strength and wiring codes10,14 do not support the idea that magnetic fields are a clear-cut cause of leukemia, inasmuch as associations of wiring codes with disease are stronger than those for 60-Hz magnetic fields. Youngson et al17 strengthened those observations and emphasized that any association between distance from power lines and hematological cancer is likely to be very small and that
even large observational epidemiological studies will probably not prove that it does or does not exist.
That does not mean that further efforts should not be made to learn why high-current electrical wiring configuration might be an indicator of increased disease risk. Poole and Trichopoulos5 have suggested that poor households are harder to recruit as controls when random-digit dialing is used for this purpose. Those households might be more often associated with high-current configuration (inner-city) wiring. According to their hypothesis, the increased odds ratio would indicate a deficit of high-current configuration wiring among the controls, not an excess among the cases.
A parallel argument might be made to explain the results of the Wertheimer and Leeper6 study. Cases were selected from death certificates, and controls were selected from birth certificates. If poor children with leukemia are more likely to die than the more affluent, an excess of high-current wiring configuration associated with death certificates could reflect that difference. A dramatic change in survival of leukemia victims was in fact occurring at the time of their study, in which 48% of the childhood cancers were leukemias. Death certificates used in the study were collected in 1950 - 1973, a period when death rates for childhood-leukemia death rates fell by about 33%.43 If richer children got the best treatment first, the hypothesis would be strengthened. The investigators in both studies (Savitz; 10 Wertheimer and Leeper6) considered and dismissed confounding by socioeconomic factors as a possible explanation of their findings, but the possibility of confounding factors probably deserves further consideration.
The only two residential studies that showed systematic internal consistency (Wertheimer and Leeper)6,13 were the most likely to be influenced by procedural bias. Coding of wiring configurations was not blind to the identity of case and control subjects, and the choice of control groups was rather arbitrary (two control definitions were used in the first study, three in the second).
With respect to its relevance to the potential health effects of GWEN fields, the literature provides no consistent or even suggestive evidence that household measurements of magnetic-field exposures are associated with increases in rates of hematological or other cancers in children or adults. Whether distance of residences from power lines is a surrogate measure of exposure to other cancer-causing agents is an unanswered question of public-health importance. But, the answer is unlikely to provide any new information that could be used to assess the potential risk associated with GWEN electromagnetic fields.
1. Coleman, M., and V. Beral. 1988. A review of epidemiological studies of the health effects of living near or working with electricity generation and transmission equipment . Int. J. Epid. 17(1) : 1-13.
2. Gordon, I., M. Motika, and T. Nolan. 1990. Epidemiological studies of cancer and powerline frequency electromagnetic fields: a meta analysis. Report 242. The University of Melbourne Statistical Consulting Center, Melbourne, Australia.
3. Tenforde, T. S. 1992. Biological interactions and potential health effects of extremely low-frequency magnetic fields. Ann. Rev. Public Health 13 : 173-196.
4. Theriault, G. 1991. Health effects of electromagnetic radiation on workers: Epidemiological studies. Pp. 93-124 in Proceedings of the Scientific Workshop on the Health Effects of Electric and Magnetic Fields in Workers. P. J. Bierbaum and J. M. Peters, eds. Publ. No. 91-111. Cincinnati, OH : U.S. Department of Health and Human Services.
5. Poole, C., and D. Trichopoulos. 1991. Perspective: Extremely lowfrequency electric and magnetic fields and cancer. Cancer Causes and Control 2 : 267-276.
6. Wertheimer, N., and E. Leeper. 1979. Electrical wiring configurations and childhood cancer. Am. J. Epidemiol. 109 : 273-284.
7. Fulton, J. P., S. Cobb, L. Preble, L. Leone, and E. Forman. 1980. Electrical wiring configuration and childhood leukemia in Rhode Island . Am. J. Epidemiol. 111 : 292-296.
8. Wertheimer, N., and E. Leeper. 1980. Electrical wiring configurations and childhood leukemia in Rhode Island. Am. J. Epidemiol. 292 : 461-462.
9. Tomenius, L. 1986. 50-Hz electromagnetic environment and the incidence of childhood tumors in Stockholm County. Bioelectromagnetics 7 : 191-207.
10. Savitz, D. A. 1987. Case-control study of childhood cancer and residential exposure to electric and magnetic fields. Contractor's Final Report, New York State Power Lines Project, Contract 218217, March 30, 1987.
11. Savitz, D. A., H. Wachtel, F. A. Barnes, E. M. John, and J. G. Tvrdik. 1988. Case-control study of childhood cancer and exposure to 60-Hz magnetic fields. Am. J. Epidemiol. 128(1) : 21-38.
12. London, S. J., D. C. Thomas, J. D. Bowman, E. Sobel, T.-C. Cheng, and J. M. Peters. 1991. Exposure to residential electric and magnetic fields and risk of childhood leukemia. Am. J. Epidemiol. 134 : 923-937.
13. Wertheimer, N., and E. Leeper. 1982. Adult cancer related to electrical wires near the home. Int. J. Epidemiol. 11 : 345-355.
14. Severson, R. K., et al. 1988. Acute nonlymphocyte leukemia and residential exposure to power frequency magnetic fields. Am. J. Epidemiol. 128(1) : 10-20.
15. Wertheimer, N., and E. Leeper. 1989. Fetal loss associated with two seasonal sources of electromagnetic field exposure. Am. J. Epidemiol. 129(1) : 220-224.
16. McDowall, M. N. 1986. Mortality of persons resident in the vicinity of electricity transmission facilities. Br. J. Cancer 53 : 271-279.
17. Youngson, J. H. A. M., A. D. Clayden, A. Myers, and R. A. Cartwright. 1991. A case/control study of adult hematological malignancies in relation to overhead powerlines. Br. J. Cancer 63 : 977-985.
18. National Institute for Occupational Safety and Health. 1991. In Proceedings of the Scientific Workshop on the Health Effects of Electric and Magnetic Fields on Workers, P. J. Bierbaum, and J. M. Peters, eds. Publication No. 91-111. Ohio : U.S. Department of Health and Human Services.
19. Milham, S., Jr. 1982. Mortality from leukemia in workers exposed to electrical and magnetic fields. (Letter to the Editor). N. Engl. J. Med. 307(4) : 249.
20. Milham, S., Jr. 1985. Mortality in workers exposed to electromagnetic fields. Environ. Health Perspec. 62 : 297-300.
21. Lin, R. S., P. C. Dischinger, J. Conde, and K. P. Farrell. 1985. Occupational exposure to electromagnetic fields and the occurrence of brain tumors. J. Occup. Med. 27 : 413-419.
22. Speers, M. A., J. G. Dobbins, and V. S. Miller. 1988. Occupational exposures and brain cancer mortality: a preliminary study of East Texas residents. Am. J. Ind. Med. 13 : 629-638.
23. Savitz, D. A., and L. Feingold. 1989. Association of childhood cancer with residential traffic density. Scan. J. Work Environ. Health 15 : 360-3.
24. Stern, F. B., R. A. Waxweiler, J. J. Beaumont, S. T. Lee, R. A. Rinsky, R. D. Zumwalde, W. E. Halperin, P. J. Bierbaum, P. J. Landrigan, and W. E. Murray, Jr. 1986. A case-control study of leukemia at a naval nuclear shipyard. Am. J. Epidemiol. 123 : 980-992.
25. Flodin, U., M. Fredriksson, Om Axelson, B. Persson, and L. Hardell. 1986. Background radiation, electrical work, and some other exposures associated with acute myeloid leukemia in a case-referent study. Arch. Environ. Health 41 : 77-84.
26. Olin, R., D. Vägerö, and A. Ahlbom. 1985. Mortality experience of electrical engineers. Br. J. Ind. Med. 42 : 211-2.
27. Matanoski, G. M., E. A. Elliott, and P. N. Breysse. 1991. Electromagnetic field exposure and male breast cancer. The Lancet : 337 : 737.
28. Vägerö, D., A. J. Swerdlow, and V. Beral. 1990. Occupational andmalignant melanoma: a study based on cancer registration data in England, Wales and in Sweden. Brit. J. Ind. Med. 47 : 317-325.
29. Stevens, R. G., B. W. Wilson, and L. E. Anderson. 1990. The question of cancer. Pp. 369-370 in Extremely Low Frequency Electromagnetic Fields: The Question of Cancer , B. W. Wilson, R. G. Stevens, and L. E. Anderson, eds. Columbus, OH : Battelle Press.
30. Tenforde, T. S., and Budinger. 1986. Biological effects and physical safety aspects of NMR imaging and in vivo spectroscopy. Pp. 493-548 in NMR in Medicine: Instrumentation and Clinical Applications, S. R. Thomas, and R. L. Dixon, eds. New York, NY : American Association of Physicists in Medicine.
31. Robinette, C. D., C. Silverman, and S. Jablon. 1980. Effects upon health of occupational exposure to microwave radiation (radar). Am. J. Epidemiol. 112 : 39-53.
32. Barron, C. I., and A. A. Baraff. 1958. Medical considerations of exposure to microwaves (radar). JAMA 168 (No. 9) : 1194-1199.
33. Szmigielski, S., M. Blelec, S. Lipski, and G. Sokolska. 1988. Immunologic and cancer-related aspects of exposure to low-level microwave and radiofrequency fields. Pp. 861-925 in Modern Bioelectricity, ed. A. A. Marino, New York : Marcel-Dekker.
34. Lilienfeld, A., J. Tonascia, S. Tonascia, C. Libauer, G. Cauthen, J. Markowitz, and S. Weida. 1978. Evaluation of health status of foreign service and other employees from selected Eastern European posts (unpublished).
35. Selvin, S., J. Schulman, and D. W. Merrill. 1992. Distance and risk measures for the analysis of spatial data: a study of three childhood cancers. Soc. Sci. Med. 34 : 769-777.
36. Kraut, W. T., E. Chan, P. J. Lioy, F. B. Cohen, B. D. Goldstein, and P. J. Landrigan. 1991. Epidemiologic investigation of a cancer cluster in professional football players. Environ. Res. 56 : 131-143.
37. Hawaii Department of Health Environmental Epidemiology Program. 1986. “Cancer incidence in census tracts with broadcasting towers in Honolulu, Hawaii.” Report to the City Council, City and County of Honolulu, Hawaii.
38. Milham, S., Jr. 1985. Silent keys: Leukemia mortality in amateur radio operators. Lancet 1 : 812.
39. Milham, S., Jr. 1988. Increased mortality in amateur radio operators due to lymphatic and hematopoietic malignancies. Am. J. Epidemiol. 127 : 50-54.
40. Milham, S., Jr. 1988. Mortality by license class in amateur radio operators. Am. J. Epidemiol. 128 : 1175-1176.
41. Kaune, W. T., R. G. Stevens, N. J. Callahan, R. K. Severson, and D. B. Thomas. 1987. Residential magnetic and electric fields. Bioelectromagnetics 8 : 315-335.
42. Barnes, F., H. Wachtel, D. Savitz, and J. Fuller. 1989. Use of wiring configuration and wiring codes for estimating externally generated electric and magnetic fields. Bioelectromagnetics 10 : 13-21.