The Ground Wave Emergency Network (GWEN) is a nationwide system of radio transmitters and receivers intended to ensure adequate communication between command authorities and land-based strategic nuclear forces in the event of a nuclear attack on the United States mainland. The full GWEN system would consist of about 86 ground stations that would communicate in both the low-frequency (LF) radioband (150-175 kHz) and the ultrahigh-frequency (UHF) radioband (225-400 MHz) and 12 ground stations that would transmit only in the UHF band. The peak radiated power of the LF and UHF transmissions would be about 3,200 W and 20 W, respectively. Those power levels are modest, in comparison with those of most radio and television stations that broadcast at similar frequencies. During peacetime, the GWEN network would operate in a maintenance mode, each LF transmitter broadcasting about 1% of the time and each UHF transmitter turned on about 2% of the time.
A partial GWEN system involving about 60 broadcast sites is being constructed and tested. In support of the decision to begin construction, the U.S. Air Force completed an environmental impact statement (EIS) on the GWEN system in 1987. The EIS concluded that the impact of GWEN LF and UHF emissions on the health of people living near GWEN sites was likely to be negligible, because exposures would be smaller than safety guidelines established by scientific standard-setting organizations. The guidelines were based on a scientific consensus that harmful biological effects were not likely to occur at exposures smaller than those required to heat tissue measurably. Since publication of the EIS, new issues have arisen concerning the possible health effects of electromagnetic fields (EMFs). Most notable is the increase in epidemiological evidence of an association between cancer risk and exposure to extremely-low-frequency (ELF) fields that are too weak to cause tissue heating. In 1990, Congress asked the Air Force to postpone completion of the GWEN system, pending an evaluation of the new evidence on the biological effects of EMFs. The Air Force then asked the National Research Council to convene an independent scientific committee to address the issue. The Committee on Assessment of the Possible Health Effects of Ground Wave Emergency Network was formed in the Research Council's Board on Radiation Effects Research, and this is the final report of that committee. The report includes a description of the physical nature of GWEN fields and the electrical coupling of GWEN fields to humans, an analysis of current scientific literature bearing on the biological effects of GWEN fields, and evaluations of possible related human-health hazards and options for risk management. A particular effort has been made to interpret the controversial epidemiological data which suggest that a possible link exists between EMF exposure and cancer risk. The conclusions reached by the GWEN committee on this and other health-related aspects of exposure to EMFs are summarized in the
following sections of the Executive Summary. The literature review upon which this report is based extends to February 1992.
DESCRIPTION OF GWEN SYSTEM
GWEN is a radio communication system designed to withstand the damaging effects of electromagnetic-pulse energy surges produced by high-altitude nuclear detonations and other ionospheric disturbances. It uses ground waves, rather than the Earth's ionosphere, as a communication pathway, and it has three types of stations: input/output (I/O), receive-only (RO), and relay nodes (RNs). I/O stations, which use UHF, are capable of both sending and receiving messages. ROs would only receive messages transmitted through I/Os. RNs would provide continuous relay links between I/Os and ROs. The final operating system would consist of 12 I/Os, 130 ROs, and 123 RNs.
The I/Os would broadcast UHF signals in the 225- to 400-MHz band with an effective power of 20 W and a duty cycle of 0.024. A network of RNs would broadcast LF signals in the 150- to 175-kHz range modulated with minimum shift keying with a duty cycle of 0.014; the peak broadcasting power would be 2,000-3,200 W.
At 300 m from an RN, the peak LF electric and magnetic field values are 5 V/m and 165 µG, respectively. The values decrease to 0.2 V/m and 9 µG at a distance of 4 km. RNs would be located primarily in rural areas so surrounding population densities are typically low. It is estimated that the numbers of people living within 300 m and 4 km of all RNs in a completed GWEN system are roughly 80 and 37,000, respectively. For comparison, measurements of electric fields from AM radio broadcasts (535-1605 kHz) made at random outdoor locations in 15 U.S. metropolitan areas show that about 22 million people are exposed to fields stronger than 0.28 V/m and 1.3 million to fields stronger than 1.0 V/m.
The LF RNs and the UHF transmitters of GWEN emit signals that span more than a factor of 1,000 in frequency within the radiofrequency (RF) band. The LF fields emitted by the GWEN system are only weakly absorbed by the human body, and their interactions are best characterized by the current density and induced electric field, rather than by the power deposited in tissue (which depends on the square of the induced electric field). Because the wavelength and depth of penetration of LF fields are much larger than the maximum dimension of the human body, these fields interact in a manner that is more similar to power-frequency fields in the ELF band than to microwaves. In judging the potential biological effects of LF fields from GWEN transmitters, the committee has therefore analyzed both the substantial literature related to ELF fields and the literature on studies conducted with much higher frequencies in the RF band (primarily microwaves). The committee found only a few biological studies on humans or laboratory animals
conducted at frequencies in the LF band that were comparable to the transmitted signals from the GWEN system. Conclusions on the potential effects of LF fields from GWEN RNs have therefore been inferred from available information on the biological interactions of fields oscillating at both lower and higher frequencies, with recognition of the uncertainties associated with extrapolation across such a broad frequency range.
The UHF transmitters in the GWEN system emit signals that are close in frequency to FM radio transmitters and UHF television transmitters. The primary interaction mechanism of these fields involves electrical interactions with dipolar molecules in tissue, and the radiation is absorbed as a function of the depth of penetration inside the body. The potential biological effects of the UHF fields have been assessed by reviewing the literature on responses of humans and laboratory animals to microwave radiation of 300-30,000 MHz.
The maximum induced currents in the body of a standing adult human optimally coupled to the ground (i.e., barefoot) were calculated with an anatomically based electrical model. For the maximum GWEN LF fields at the perimeter of a transmitter site, the calculated current passing through the feet to the ground was 2.89 mA, which is one-thirtieth the maximum permissible current, according to the Institute for Electrical and Electronic Engineers (IEEE) RF-protection guide. The peak current density is induced in the ankles in such an exposure and equals 700 mA/m2 (70 µA/cm2). In the brain and chest, the peak induced current densities are, respectively, 35 and 80 mA/m2. The peak whole-body specific absorption rate (SAR) resulting from exposure to the UHF signal transmitted by the GWEN system is 97.9 µW/kg, on the basis of calculations for a grounded, anatomically based model of a standing man. That SAR is lower by a factor of more than 10,000 than the maximum permissible levels for partial body exposures given in the IEEE radiofrequency protection guide.
Consideration was given to both thermal and nonthermal mechanisms through which the maximum fields induced in tissue could influence biological functions. The rate of energy deposition in tissue by GWEN fields was far below the rates that could produce a biologically important temperature rise. Similarly, on the basis of electrophysiological models, it was concluded that the fields induced in tissue were too low to significantly alter the transmembrane potential (i.e., the calculated change in potential was estimated to be less than 10 µV). Nonthermal mechanisms of interaction of GWEN fields with living cells that might proceed under conditions in which the electrical properties of the cell membrane are not substantially altered were also considered. In general, they depend on a large ELF content of the applied fields, for example, through subharmonic-frequency content or LF amplitude modulation of the RF carrier wave. Apart from the pulse-group repetition frequency of approximately 1 Hz, the GWEN fields lack any ELF components, and it was concluded that such interactions are highly unlikely.
Indirect coupling of the LF fields from GWEN transmitters to persons in contact with ungrounded metallic objects, such as cars, was also considered. That form of coupling can lead to perception, shock, pain, and (at very high field levels) burns. From an analysis of the LF fields at the perimeter of a GWEN site, it was concluded that there would be no risk of electrical shocks or burns outside the site. A similar conclusion was drawn for the UHF fields emitted by GWEN transmitters.
ORGAN AND TISSUE SYSTEMS
The response of tissues to the RF fields emitted by GWEN LF RNs was assessed from a comprehensive review of literature on both ELF and microwave field effects. Potential effects of UHF fields were judged primarily on the basis of the biological literature on the effects of microwave exposures. Both thermal and nonthermal responses to RF radiation were reviewed in the context of possible effects of GWEN fields.
Tissue heating from exposure to RF radiation produces a series of physiological responses characteristic of stress, with a resulting activation of thermoregulatory mechanisms mediated through the neuroendocrine and cardiovascular systems. The threshold SAR for reproducible physiological and behavioral changes associated with tissue heating is approximately 1 W/kg, which greatly exceeds the SAR from GWEN fields. A detailed analysis of the literature on RF-field effects on major organ and tissue systems revealed no adverse responses of the nervous, ocular, endocrine, immune, hematologic, or cardiovascular systems to subthermal energy absorption. Similar conclusions were drawn for reproduction and development in mammals exposed to RF fields.
The responses of tissues to low-intensity, amplitude-modulated RF (AM-RF) fields and to sinusoidal and pulsed ELF fields were also reviewed in the context of possible physiological effects of LF fields from the GWEN system. Although changes in calcium ion binding to nerve cell surfaces and in lymphocyte immune functions have been reported after in vitro exposure to AM-RF fields, no such effects have been established for the weak LF fields produced by the GWEN system. Similarly, a variety of tissue responses to relatively high-intensity ELF fields have been reported, but there is no convincing evidence for adverse health effects of induced tissue currents comparable to those produced by GWEN fields. The effects of pulsed and sinusoidal ELF fields on reproduction and fetal development have been studied in both mammals and nonmammals. Although developmental abnormalities have reportedly occurred in avian embryos after exposure to pulsed magnetic fields with ELF repetition rates, there is no convincing evidence of teratogenic effects of high-intensity ELF fields in developing mammals.
CELLULAR AND SUBCELLULAR EFFECTS
It is difficult to draw firm conclusions with respect to GWEN emissions on the basis of available published data on the biological effects of EMFs, primarily because most of the information pertains to high-level fields with frequencies either higher than GWEN LF frequencies by a factor of about 1,000 (microwaves) or lower than GWEN frequencies by a factor of about 1,000 (ELF). However, a few papers related to bone healing have reported in vitro effects of 60-kHz electric fields. The GWEN UHF emissions oscillate at frequencies closer to the microwave region about which many biological effects data have been published. Positive microwave effects have been reported only after exposures considerably larger than those produced by GWEN emissions. The committee 's review of cellular and subcellular data concentrated on ELF fields and bone-healing pulsed fields, because some of the relevant dose metrics were comparable with, or lower than, those associated with GWEN population exposures and also involved end points perceived to be important in risk assessment.
The more consistent low-level effects are related to changes in gene expression, alterations in Ca2+ balance at the cellular level, and changes in the activity of some enzymes. None of those effects has been clearly related to a particular health problem. Potential genotoxic effects of EMFs have also been addressed in numerous studies. Assays for mutagenicity have been consistently negative, but there have been some controversial results in the studies of chromosomal aberration. Most of the studies showing chromosomal clastogenic effects of EMFs have not been corroborated, and it can therefore be concluded that the data available do not support a direct genotoxic effect of EMFs. Because the genotoxicity studies on EMFs have been predominantly negative, it has been suggested that possible carcinogenic effects might be due to tumor promotion, rather than initiation. Several studies are currently attempting to determine whether EMFs act as a promoter or copromoter in tests with well-established in vitro and in vivo carcinogenicity models.
There are few experimental studies on the possible carcinogenic effect of EMFs with animal models and in vitro systems. In the case of ELF, a few negative results have been reported. At microwave frequencies, several studies have been published, some with positive indications of carcinogenic activity, including a dose-response curve for neoplastic transformation in vitro. All those studies were conducted at higher exposure levels than would be associated with GWEN.
Direct evidence of biological effects of EMFs comes from three primary sources--laboratory studies of human volunteers, epidemiological studies of ELF-exposed populations, and epidemiological studies of populations exposed to AM radio frequencies or to radar or microwaves.
Studies of human volunteers have documented a variety of effects, including cutaneous perception, which is manifested at thresholds of 30 - 200 mA; phosphenes, which are visual effects resulting from induced currents in the retina; pacemaker interference; microwave auditory effects; circadian-rhythm alterations; behavioral changes; heart-rate changes; blood-chemistry changes; and changes in bone growth. The relevance of those observations to the GWEN system is for the most part small, because the exposures were at a much higher field level than would be associated with GWEN fields or at very different frequencies from those encountered in GWEN fields. The bone-growth changes are perhaps of some relevance, because the envelope of effective power and frequency might include expected GWEN emissions. However, the effects observed are restricted to accelerated healing of bone fractures that otherwise failed to heal, and the effects, if any, on healthy tissue are unknown.
Epidemiological studies of ELF-exposed populations can be divided into residential and occupational studies. Several of the dozen or so completed residential studies have reported excesses of childhood or adult cancer, particularly brain cancer and leukemia. The strength of those findings is limited by possible biases in the selection of study populations and by ambiguities in retrospectively assessing the magnetic-field exposure of study subjects. Excess cancer risk seems to be more strongly correlated with the configuration of residential power lines than with measured magnetic field levels in subjects ' homes. That implies either that ELF-field exposure is not causally related to cancer risk, that measurements made in the present are worse indicators of past exposure than is power line configuration, or that biologically relevant aspects of field exposure are not captured by the measurements made to date.
A number of retrospective epidemiological studies have also reported excess cancer risk among workers thought, by virtue of their job titles, to work in unusual electromagnetic environments. Recent measurements of magnetic-field exposures in these electrically related occupations show that, as a group, such workers are indeed exposed to higher fields. However, correlations with cancer risk have generally proved stronger for job title than for measured field, suggesting either that the exposure measurements made to date do not capture the risk-relevant aspects of field exposure or that findings of elevated risk arise from some unknown confounder or from biases in the selection of comparison populations. Most studies of electrical workers have not controlled for possible exposures to hazardous chemical agents. Cancer rates noted in proportionate mortality studies might be higher simply because mortality from chronic diseases other than cancer is lower in electrically related occupations than in other comparison groups (i.e. the “healthy worker” effect).
Taken as a whole, the epidemiological literature indicates that there are not enough data to identify unequivocally, let alone to quantify, ELF-related cancer risks. Moreover, differences in frequencies and other characteristics between ELF and GWEN fields convinced the committee that the ELF data were of limited usefulness in assessing potential health risks from GWEN.
Epidemiological studies of populations exposed to electromagnetic fields from radio and television broadcasting are very few. One occupational study of the effect of exposures at AM radio frequencies found no elevated cancer risks among the study population, but the statistical power of that study to detect all but the shortest latency cancers was small. A study of proportionate cancer incidence in Hawaii reported elevated risks in census tracts near radio and television broadcast towers, but the study did not adjust for race, age, sex, socio-economic status, or urban-rural effects on cancer rates.
The last data source, epidemiological studies of radar- and microwave-exposed populations, is sparse. Two populations, U.S. naval personnel exposed to radar and American embassy personnel in Moscow, dominate the literature. Neither has shown any adverse health effects, but both lack well-defined exposure data. Because the studies in question are of low statistical power and radar and microwave radiation have frequencies very different from those of GWEN, the findings do not offer much insight as to whether the GWEN system poses any risks to public health.
The calculated SARs for LF and UHF radiation for a person standing at the fence of a GWEN site are 0.001 and 0.0001 W/kg, respectively, compared with an IEEE radiation-protection guideline of 0.08 W/kg. The combined SAR of 0.0011 W/kg at the fence of a GWEN site would have a body-heating effect equivalent to raising the environmental temperature by 0.02°C. A temperature increase of that magnitude should not cause any health or environmental concerns.
Assessing risks from exposures to low levels of any environmental agent is an uncertain enterprise. Epidemiological and animal studies are limited in statistical power to risk levels that are well above those that society often considers important. Judgments about the amount of human exposure needed to produce very small risks must rely on extrapolations from higher levels, and often on extrapolations from effects measured in other species. Uncertainties in risk assessment are particularly troublesome for exposures to subthermal levels of nonionizing radiation, because the four usual steps in risk assessment are not complete:
Hazard identification. The hazard is not clearly identified and still very controversial.
Exposure characterization. There is no accepted exposure metric, and there are only categories of exposure of a proxy nature.
Dose-effect relationships. Because there is no agreed on exposure metric, very few exposure measurements, and no clearly identified hazard, dose-effect relationships cannot be established.
Risk characterization. Without quantitative population exposure assessments, and without dose-effect relationships, risks to any population cannot be established.
We concluded, therefore, that a traditional risk assessment could not be done. Consequently, we established upper bounds of risk for GWEN using data (basically negative data) from similar broadcast installations that had been in use for many years.
First, despite a large growth in radio and television broadcasting over the last several decades, concurrent time-series analyses reflect no parallel increase in overall cancer morbidity and mortality above the “noise” level of the data. Because the time-averaged electric fields induced in a body standing near the outer fence of a GWEN RN facility are comparable to or smaller than those induced by exposures to radio and television broadcasts in urban areas, GWEN risks are unlikely to be significantly greater than those associated with commercial broadcasting.
Second, systems for public-health surveillance regularly review cancer incidence and mortality data at the resolution of the census tract. This surveillance activity would be expected to detect any excess cancer risks around broadcast facilities that are larger than the detection threshold of the surveillance apparatus. That no convincing evidence has so far been found suggests that, whatever the public health risks related to broadcast exposure might be, they are smaller than the detection threshold. A recent epidemiological study of cancer risk in a population exposed to AM-broadcast fields as high as 16 V/m during work hours found no excess cancer risk. However, the statistical power to detect cancers other than short-latency ones, such as leukemia, was low. Only two cases of leukemia and no leukemia deaths were identified in a work population of 8,000. Both cases were among indoor workers who had the lowest exposure. Among outdoor workers, 1.11 cases of leukemia would have been expected. Despite the negative findings in this study, upper bound estimates from these data are close to the background rate, indicating the variability encountered in this type of analysis.
On the basis of the negative findings of epidemiological surveys related to AM broadcast stations, the committee estimates that the total excess number of cancer deaths associated with operation of the entire GWEN communication system over a 70-yr lifetime is less than one. This upper-bound estimate is well below the level that could be detected by an epidemiological survey system. The committee recognizes the inherent limitations in any effort to establish an upper bound estimate of cancer risk using negative data. However, such a bounding argument is likely to produce large numbers relative to the actual (unknown) risk. The upper bound cancer risk estimated here from negative epidemiological
findings on health effects of RF broadcast systems clearly indicates the absence of a significant cancer risk associated with GWEN emissions.
A number of actions might be taken to reduce population exposures to GWEN LF and UHF emissions. They include choosing from among candidate sites to avoid those with homes nearby (this is currently an important factor in site selection), placing the UHF-transmitting antenna in each site so as to minimize UHF exposures of those living nearby, placing the UHF-transmitting antennae on higher poles, increasing the size of the GWEN RN sites to reduce field strengths at the site boundaries, and reducing the number of GWEN RNs in the final GWEN system. Each of those options carries certain costs. The primary risk-management challenge is to decide which, if any, of the costs are justified.
The results of a critical evaluation of published literature on the biological and health effects of EMFs and of an analysis of the relevance of this information to the potential health effects of GWEN fields indicate that these fields should have only a minimal, and probably undetectable, impact on public health. In 1987, an environmental-impact statement on the effects of the GWEN system on humans and natural biota was published. The EIS focused on thermal mechanisms of interaction of GWEN fields with living systems and on the risks to humans posed by shock and burn phenomena associated with the charging of electrically conductive objects in the vicinity of GWEN towers.
Since the publication of the EIS, a number of new issues have arisen concerning the possible biological and health effects of EMFs, particularly those that might be associated with nonthermal events. The present report evaluates information from studies conducted at frequencies varying from ELF to microwaves in the context of public exposure to GWEN fields. Because there is little published information about the effects of LF signals, the committee used information from AM broadcasting to establish an upper-bound estimate of cancer risk to people living in the vicinity of GWEN stations. On the basis of its review of published literature on the biological and health effects of electromagnetic fields, the committee concludes that the excess risk of cancer death (that is, the risk that cannot be ruled out from health surveillance data) associated with exposure to GWEN fields is less than one additional death over a 70-yr period for persons living within 10 km of the entire system of GWEN sites.
The conclusions of this report reinforce those of the EIS, in that no evidence of adverse effects of GWEN fields on public health was found. The National Research Council committee recommends that this report be used in conjunction with the original EIS as a comprehensive assessment of the potential public health impact of the GWEN system.