Electric bed heater use is considered a good surrogate for personal nighttime magnetic field exposure, because these devices are used throughout the night over several months, are in close contact with the body, and experimentally have been shown to emit stronger fields than other residential sources (5–15 milligauss for waterbeds and 10–20 milligauss for electric blankets). 1–3 As a result, there is considerable public health concern regarding the potential risk of magnetic fields from electric bed heaters on reproductive health. 3–13
A few laboratory studies have reported alterations in the development of chick embryos exposed to magnetic fields. 14–17 Olcese et al18 demonstrated the involvement of the visual system in mediating magnetic field effects on pineal melatonin synthesis in rats. Stevens 19 hypothesized that the adverse health effects associated with magnetic field exposures may be due to changes in the nighttime production of the regulatory hormone melatonin, which in turn may increase circulating gonadal hormones and hence increase the risk of adverse reproductive outcomes.
The epidemiologic studies that have assessed the association between electric bed heater use and adverse reproductive outcomes show inconsistent results. Three studies that examined birth defects found little association with electric blanket use. 7–9 One of these three studies, 9 however, found a higher rate of urinary tract birth defects among the pregnancies during which the mothers used electric blankets and had a history of infertility (odds ratio = 4.4; 95% confidence interval = 0.9–22.7, 5 exposed cases). An increased risk of brain cancer has been seen among the children of mothers who used electric blankets while they were pregnant. 10 Bracken et al11 found little association between the use of electric bed heaters and low birth weight and intrauterine growth retardation. Waterbeds and electric blankets were combined, and season of use and heater setting were not considered, however. Wertheimer and Leeper 3 found that among electric bed heater users, fetal loss and fetal growth delay occurred more frequently in months of increasingly cold nights, when increasing exposure from electric bed heater use would be expected. In a recent study, Belanger et al12 found that women who reported electric blanket use at the time of conception had a slightly higher rate of spontaneous abortion than those of non-users. The rates were not increased when the blankets were used daily on a high setting. Waterbed users had slightly lower rates than did non-users.
Our current study is part of a prospective study designed to examine the associations between spontaneous abortion and first trimester environmental exposures. The purpose of this study is to assess whether reported use of electric bed heaters is positively associated with clinical spontaneous abortion.
There are five a priori hypotheses: (1) electric bed heater users have higher spontaneous abortion rates than non-users, with electric blanket users having the highest rates; (2) the association increases with increasing duration of use and temperature setting; (3) the association is more pronounced among users who did not use bedroom heat during the night (assuming bed heaters would continue to cycle on and off in rooms with cooler ambient temperatures); (4) the association is more pronounced for those women who conceived in months with increasingly cold nights; and (5) the association is more pronounced among women who interviewed earlier in pregnancy, because they have a greater probability of having an earlier pregnancy loss, and earlier gestations may be more sensitive to magnetic field exposures. This last assumption is partially supported by the work of Juutilainen et al, 13 who found an increased risk of preclinical spontaneous abortion and front-door magnetic field measures.
We conducted a small, post hoc, experimental measurement survey of four conventional electric blankets to determine (1) whether the overnight magnetic field levels corresponding to the study’s reported setting-duration combinations would rank similarly as the observed relative risk estimates of spontaneous abortion associated with these combinations and (2) whether any of the setting-duration magnetic field estimates of exposures received by the uterus, by the middle body surface, or by the retina were substantially higher than background levels. Even though the uterus is the most feasible target site for spontaneous abortion, the retina and middle body surface were also evaluated, because the retina has previously been proposed as a feasible site 18 and the middle body surface could serve as a site representing the maximum exposure to the body.
Subjects and Methods
Details of the study population recruitment and ascertainment of pregnancy outcome are described elsewhere. 20
Pregnant women were recruited between January 1990 and September 1991 when they called for their first prenatal appointment at one of three California Kaiser Permanente Medical Care Program facilities (Santa Clara and Walnut Creek in northern California and Fontana in southern California). Eligible women had to be at least 18 years old, with 13 weeks’ gestation or less, and Spanish or English speaking. A total of 7,881 women were evaluated by the prenatal appointment clerks, of which 6,179 (78.4%) were eligible and willing to be interviewed, 424 were ineligible, and 1,278 refused participation. Of the 6,179 women who agreed to be interviewed, 5,342 (86.5%) completed the interview. The overall eligible participation rate was 72% (5,342 of 7,457).
Pregnancy outcomes for the 5,342 interviewed women were ascertained by searching computerized hospital records (73%), abstracting medical records for women without computer matches (18%), and following up with telephone interviews or mailed questionnaires completed by the women, or by matches to California vital records (8%). Fewer than 1% (N = 35) of the outcomes could not be determined.
Spontaneous abortions (N = 499) were defined as pregnancies that ended at 20 weeks’ gestation or less. Multiple births (N = 55) were considered as a single pregnancy outcome in the analyses. Elective abortions (N = 128) and ectopic (N = 13) and molar pregnancies (N = 4) were excluded from analyses.
Women were interviewed by telephone between 4 and 13 weeks of gestation. Twenty-two per cent of the interviews were conducted by the 6th week of gestation, 50% by the 8th week, 87% by the 10th week, and only 1% in the 13th week of gestation. Telephone interviews were conducted using a computer-assisted system to collect information about demographics, reproductive and medical history, potential confounders, and environmental exposures. Users of electric bed heaters and bedroom heat at night reported whether they used the heater any time between their last menstrual period date and the interview date. Women who used the heaters only to warm the bed were considered non-users. Bed heater temperature setting and number of hours of use per night were assessed for the usual use pattern over the 7 days before the interview. All women used blankets purchased or received as gifts before 1989, when the new low-magnetic field blankets were placed on the market.
Electric Bed Heater Measurement Assessment
EMDEX-C meters, set at a 5-second sampling rate, were used to measure overnight magnetic field intensities of four different pre-1989 conventional blankets. The manufacturers of the blankets were not known. Measurements were done for the off, low, medium, and high settings for the magnetic field time-weighted average and for a rate of change metric assessing the time-weighted average of the change in magnetic field strength levels between measurement samples.
Three meter positions were used: (1) the middle of the mattress covered by the blanket (estimating the middle body surface position), (2) the middle of the mattress covered by the blanket but separated 10 cm away from the blanket by a foam block (estimating the uterus position), and (3) the “pillow” position about 10 cm away from the edge of the blanket (estimating the retina position).
Four meters, one at each meter position, measured magnetic fields simultaneously one setting at a time. Measurements were taken on four different nights, during which the blanket was off all night the first night, on for 1 hour the second night, on for 5 hours the third night, and on for 8 hours the fourth night.
Estimates of the overnight magnetic field levels for each metric were calculated for (1) off all night (non-users), (2) on for 1 hour or less, (3) on for 2–5 hours, and (4) on for 6–8 hours. We used the midpoint hour in the hour intervals that were used to categorize the setting-duration use combinations of the study. For example, for the combination 2–5 hours “on” and 3–6 hours “off,” we assumed 3.5 hours “on” and 4.5 hours “off.”
The measurements were done without anyone sleeping under the blankets. The room temperature at the time of measurements was 61°F.
Only women who responded to the question about use of electric bed heaters any time between their last menstrual period date and their date of interview (N = 5,144) were considered. The denominator for calculating rates of spontaneous abortions included all livebirths (N = 4,614), spontaneous abortions (N = 498), and stillbirths (N = 32). Spontaneous abortion rates for electric bed heater users were compared with those of non-users (1) for four types of electric bed heaters and (2) by bed heater temperature setting and hours of use during the night. Possible modification and confounding of the estimated effect measures were examined in detail for (1) use of bedroom heat during the night, (2) gestational age at interview, and (3) month of last menstrual period. Use of bedroom heat at night while sleeping was grouped into three categories (heat off, heat on some of the night, and heat on most of the night). Gestational age at interview was divided into three categories (up to 6 weeks, between 7 and 9 weeks, and 10 weeks or more).
Potential confounders and modifiers of the effect estimates were assessed first by stratified analysis. For this, the outcome was dichotomized into spontaneous abortion and no spontaneous abortion, and the exposure was dichotomized into bed heater use and non-use. Each bed heater type was assessed separately. For ease of comparing adjusted rates with crude rates, odds ratios are presented, because crude relative risks and odds ratios and their corresponding 95% confidence intervals were similar. For those variables of which the odds ratios were homogeneous across strata, Mantel-Haenszel 21 analysis was used to calculate adjusted odds ratios. Multiple logistic regression was used to assess the independent association between electric bed heater use and spontaneous abortion. Covariates initially considered were prior fetal loss, gestation at interview, nausea, body mass index, maternal age, race, income, education, smoking, alcohol, caffeine, water (tap and bottle) consumption, perception of health, season of electric bed heater use, recruitment location (Kaiser facility), and bedroom heat on at night. Covariates were categorized as indicated in Table 1. Variables included in the final logistic regression models were those that showed a 10% change in the crude effect estimate after the Mantel-Haenszel adjustment. In the final model, covariates that were important for at least one of the bed heater types were used. The final model used spontaneous abortion as the dependent variable and electric bed heater use as the independent variable, with each heater type included as a dummy variable and maternal age, race, income, bedroom heat on at night, perceived health status, tapwater, alcohol, cigarette and caffeine consumption, gestational age at interview, and recruitment location as covariates. For two subtypes of electric blanket use, low setting for 6 hours or more and high setting for an hour or less, the final logistic models used maternal age, alcohol and cigarette consumption, perceived health, heat on at night, and gestation at interview.
Multiple logistic regression analyses were also used to evaluate effect modification between electric blanket use and waterbed use and maternal age, season at last menstrual period, heat on at night, race, income, and gestational week of interview.
Electric Bed Heater Use Patterns
Among the 5,144 pregnant women, 10% (524) used an electric blanket, 15% (796) used a waterbed, 5% (259) used a heating pad, and only 1% (55) used an electrically heated mattress pad any time between their last menstrual period month date and their interview date. Electric blanket use was prominent from September through March, with the highest prevalence of use (18.3%) found in December and the lowest prevalence of use found in July (1.5%). Waterbed use was constant throughout the year. Electric blanket use was more prevalent among women living in northern California (Santa Clara and Walnut Creek), and waterbed use was more prevalent among women living in southern California (Fontana) (Table 1). Of those who used electric blankets within a week of the interview, the majority used low settings for the duration of the night. Those who used the blanket on high settings generally did so for an hour or less (see Table 4). Those who used waterbeds within a week of the interview mainly used medium and high settings for the duration of the night (see Table 3).
Electric Bed Heater Users vs Non-Users
Table 1 summarizes the characteristics of women by use of the two most prevalent bed heater types, electric blankets and waterbeds. Compared with non-users, electric blanket users and waterbed users were more likely to be white; to consume cigarettes, alcohol, and caffeine; to have a higher household income; to perceive their health as good; and to use bedroom heat at night. Electric blanket users were slightly older than non-users, whereas waterbed users were slightly younger.
As shown in Table 2, the risk of spontaneous abortion for women who did not use any type of electric bed heater was 10.2%, a risk comparable with what is reported in the literature (10–20%). 22 The risk of spontaneous abortion for women who used electric bed heaters was slightly lower, with electric blanket users having the lowest risk (7.4%).
The crude odds ratios for each bed heater type were all below 1. After controlling for potential confounders, these estimates increased slightly, but the odds ratios were still below 1 for electric blankets and heating pads (Table 2).
Users vs Non-Users by Bed Heater Temperature Setting and Hours of Use
We examined the influence of the bed heater temperature setting and hours of use during the night on the association of spontaneous abortions only for electric blanket and waterbed users, because these electric bed heater types were the most prevalent (Table 3). Only 71% of the users responded to the questions regarding setting and hours of use for the usual use pattern over the 7 days before the interview. It was predicted that spontaneous abortion rates would increase with increasing temperature setting and duration of use. When compared with non-users, risks of spontaneous abortions were lower for those women (N = 225) using the blankets at low settings (odds ratio = 0.5, 95% confidence interval = 0.3–0.9), similar for those (N = 113) using the blankets at medium settings (odds ratio = 1.0, 95% confidence interval = 0.5–1.8), and only slightly higher for the 33 women using the blankets at high settings (odds ratio = 1.6, 95% confidence interval = 0.6–3.3). For hours of nighttime electric blanket use, risks of spontaneous abortion decreased as hours of use increased (chi-square test for trend, p-value = 0.03). This decrease was pronounced for women who used the electric blanket for 6 hours or more (odds ratio = 0.6, 95% confidence interval = 3.0–1.0). No pattern was found for waterbed users with respect to temperature setting and hours of use.
The combined association of hours of electric blanket use and setting is shown in Table 4. Compared with non-users, the 171 women who used the blanket at a low setting for 6 or more hours had the lower spontaneous abortion risks (adjusted odds ratio = 0.6, 95% confidence interval = 0.3–1.1). The 20 women who used the blanket at a high setting for an hour or less had higher risks than non-users (adjusted odds ratio = 2.9, 95% confidence interval = 1.1–9.0). There was no spontaneous abortion among the 13 women who used the electric blankets at high settings for more than 1 hour.
Users vs Non-Users by Nighttime Bedroom Heat Use
Electric bed heater users who reported use during the week before the interview, compared with non-users, were also more likely to use bedroom heat at night (see Table 1). Room heat on at night while using an electric bed heater at the same time may increase the ambient room temperature, switching off the duty cycle, resulting in no current and hence little or no magnetic field exposure. Hence, it was predicted that an increased risk of spontaneous abortion would be pronounced for electric bed heater users who slept with the bedroom heat turned off. This was not the case. The odds ratios for spontaneous abortions were similar across the room heat categories (Table 5).
Users vs Non-Users by Gestational Week at Interview
Interview week at gestation was considered as a possible modifier of the odds ratio. We hypothesized that spontaneous abortion risks would be progressively higher for women who were interviewed earlier in gestation and at risk of earlier spontaneous abortions. The data do not support this hypothesis for clinical spontaneous abortions (Table 6). For electric blanket users, we found a U-shaped pattern. Women who were interviewed at 4–6 weeks and at 10–13 weeks of gestation had higher spontaneous abortion risks than non-users, whereas those who were interviewed at 7–9 weeks of gestation had similar risks. We also found that electric blanket users had lower risks than non-users regardless of gestational age of clinical spontaneous abortion. No such pattern was found for waterbed users.
Users vs Non-Users by Season of the Last Menstrual Period Month
The odds ratios of spontaneous abortion for electric blanket users compared with non-users were not increased in the cool to colder months (September through January) around conception as predicted (Table 7). Electric blanket users had a decreased risk of spontaneous abortion after adjusting for last menstrual period month (adjusted odds ratio = 0.7, 95% confidence interval = 0.7–0.9). The risks of spontaneous abortion for waterbed users compared with non-users showed a random pattern across last menstrual period month with a slight decrease rate compared with non-users when adjusted for last menstrual period month (adjusted odds ratio = 0.8, 95% confidence interval = 0.6–1.1).
Experimental Blanket Measurement Survey
Overnight Exposure Estimates by Setting-Duration Combinations
Table 8 shows the estimated overnight magnetic field levels, by meter position, and metric for each of the setting-duration combinations reported in Table 4. For the time-weighted average levels, as predicted in the literature, a substantial (tenfold) increase in exposure over background levels would be delivered only to the middle body surface for those who used their blankets on medium or high settings for 2 or more hours (22% of the electric blanket users in the study).
For the rate of change metric, a substantial increase in exposure over background levels would be delivered to the middle body surface and to the uterus for users of medium and high settings, as well for those who used their blankets on low settings for most of the night. The retina would receive a greater exposure over background levels for users of high settings for 6–8 hours (representing only 3% of the blanket users of the study among the parent cohort).
Ranking of Setting-Duration Exposures with Spontaneous Abortion Relative Risks
Table 8 also ranks the estimated setting-duration exposures from low to high and lists the corresponding odds ratios for each setting-duration combination. There was no increasing order of odds ratios with increasing setting-duration exposure estimate for either metric. In fact, using a high setting for 1 hour or less, a combination corresponding to high odds ratio, did not rank high as to exposure.
The cycling pattern was similar for the middle body surface and the uterus meter positions, with the middle body surface having higher peaks. The two major cycling patterns are shown in Figure 1, a and b. Low temperature settings usually did not cycle (Figure 1a). When they did cycle, the off periods were considerably longer than the on periods (Figure 1b). The medium settings showed an initial long on period (a plateau) followed by a consistent on and off cycling. Fields of blankets set on high peaked initially at their plateau for a longer period of time than those set on medium and then cycled on and off consistently thereafter. No obvious cycling pattern was observed for the magnetic field tracings estimating the retina exposures.
Our study results do not support the main hypothesis that reported use of electric bed heaters during the first trimester of pregnancy increases the risk of clinical spontaneous abortions. Overall, electric bed heater users had a slightly lower rate of spontaneous abortions than found for non-users. This finding applied for electric blanket users despite the fact that they were slightly more likely to have risk factors for spontaneous abortions reported in the literature and in our study population, such as older maternal age and consumption of cigarettes, alcohol, and caffeine.
Consistent with our findings, Belanger et al12 found a slightly lower risk of spontaneous abortion for users of electrically heated waterbeds compared with non-users. Nevertheless, unlike our findings, they found a slightly higher risk of spontaneous abortion for electric blanket users. They found an adjusted odds ratio of 1.7 and a 95% confidence interval of 1.0–3.2 for their 153 electric blanket users at conception, whereas in our study we found an adjusted odds ratio of 0.8 and a 95% confidence interval of 0.6–1.2 for our 524 electric blanket users between their last menstrual period date and interview date. Our results differ from those of Belanger et al12 (Breslow-Day homogeneity P-value = 0.003 21), but we can think of no reason to explain these different results.
Possible Explanations of Lower Spontaneous Abortion Rates among Electric Blanket Users
The lower rate of spontaneous abortion among electric blanket users, especially users who keep blankets at low settings for most of the night, remains lower for most analyses and after adjustment for potential confounders. There are four main explanations for the lower-than-average spontaneous abortion risks we observed among electric blanket users.
First, there is no association; the decreased risks of spontaneous abortions observed for users compared with non-user may be a chance event.
The second, and perhaps most feasible, explanation is that the association is confounded. It could be that women who use electric blankets are a unique group of women with unspecified factors for being reproductively healthy. For example, the type of woman who feels cold all the time and needs to warm her house and her bed might be physiologically different and have a better pregnancy prognosis. Women who used electric blankets were more likely than non-users to use bedroom heat at night. We examined body mass index as a possible flag for “feeling cold” (assuming that thin women may be more likely to feel cold) and also all the recognized risk factors for clinical spontaneous abortions as possible confounders. We also explored the possibility that those who used the blanket at low settings and at high settings were different populations. None of these analyses shed light on our results.
Third, the association is the result of a recruiting bias, in which women who use electric blankets and spontaneously abort before the clinical recognition of their pregnancies are less likely to be recruited than users who carry their pregnancies to the point that a clinical spontaneous abortion occurs. We cannot rule out that possibility; however, recruitment of the study participants did not seem to be biased, because known risk factors for clinical spontaneous abortions were assessed and were consistent with what would have been expected.
Finally, electric blankets might indeed prevent clinical spontaneous abortion if used at a low setting. One could hypothesize that low-level magnetic fields or perhaps heat has a biologically beneficial effect. Litovitz et al23 have reported that 40-milligauss fields (an extremely high level not found from residential sources) protect hens’ eggs from anoxia, but other than this there is no support in theory or in previous data.
There are a number of limitations to this study. First, only clinical spontaneous abortions are assessed. Spontaneous abortions are a heterogeneous group representing fetal death by all causes. This study was not designed to assess the association of electric bed heater use on the entire spectrum of spontaneous abortion. The evaluation of the association on a variety of homogenous spontaneous abortion subgroups, such as preclinical, chromosomally abnormal, or morphologically abnormal losses, may identify a group more sensitive to magnetic field exposures than the broader, more heterogeneous clinical spontaneous abortion group assessed in this study.
Second, other residential nighttime sources of magnetic field exposures could not be controlled for in this study. Electric blanket users are different from non-users with respect to several characteristics (see Table 1), and hence, indirectly, their overall nighttime magnetic field exposures may be different. For example, assume that lower-income residential environments are more likely to have magnetic field sources that emit higher fields than higher-income home environments. Given this, a non-user may live in a higher-magnetic field home than a user, given that non-users have lower income.
Third, this study was not designed to examine the association of nighttime residential magnetic field exposures on spontaneous abortions but rather to assess the “exposure” indirectly as self-reported electric bed heater use. It may be that electric blankets and waterbeds, as used by this study population, may not account for a significant source of nighttime magnetic field exposures. Wilson et al24 tested standard, conventional electric blankets on high settings in a Washington State laboratory and found that levels measured 10 cm over the blanket surface were lower than assumed. The laboratory readings were 7.5 milligauss for electric blankets and 2.2 milligauss for electrically heated waterbeds, whereas the residential bed spot readings were 4.5 milligauss for electric blankets, 2.0 milligauss for electrically heated waterbeds, 2.9 milligauss for mattress pads, and 2.4 milligauss for heating pads. In our experimental bed survey, like that of Wilson et al, 24 the time-weighted average magnetic fields measured by the meters 10 cm from the surface of the electric blankets were modest, even at a high setting. At low settings (used by 61% of the study participants), at which blankets often did not cycle, the magnetic field estimates were close to background levels.
Finally, the use of electric blankets at high settings could not be adequately evaluated owing to the small number of women using the blanket at this setting. On the basis of our experimental bed survey, electric blankets at this setting may be an important nighttime magnetic field source.
This study has several strengths. First, the assessment of electric bed heater use was prospective, because women were interviewed within their first trimester of pregnancy.
Second, the study size was large for a prospective cohort study, resulting in a large number of spontaneous abortions to examine (nearly 500) and allowing us to assess confounding adequately and to assess effect modification for at least waterbed users and electric blanket users at low settings (the majority of users).
Third, electric blankets (N = 524 with 52 expected spontaneous abortions) and waterbeds (N = 796 with 80 expected spontaneous abortions) and the pattern of their use were examined in more detail than previous studies. A priori we knew that waterbeds had a different pattern of magnetic field exposure than electric blankets. The waterbed heating element is localized in the bed sometimes at one side or the other, sometimes at the head or the foot or at the center. This variation would induce more random misclassification of exposure as compared with electric blankets, which deliver exposure more uniformly across the bed.
Fourth, half of the interviews occurred before 8 weeks of gestation, allowing for our study to capture a fair number of early clinical spontaneous abortions.
Finally, a supplementary electric blanket survey was done post hoc to help us understand the assumption behind using electric blankets as a surrogate for a nighttime magnetic field source of exposure and the spontaneous abortion-electric blanket relative risk pattern observed for the various setting-duration combinations.
These findings provide little support for the hypothesis that electric bed heaters increase the risk spontaneous abortions. The decreased association of electric blankets used at low settings is most likely due to uncontrolled confounding. The initial assumption that electric bed heaters would deliver strong time-weighted average magnetic field levels to the uterus may be wrong (especially for low settings), so that although we can examine the association of electric blankets themselves, electric blankets may not be a strong enough magnetic field source, as generally used by our study population, to examine the relation between time-weighted average magnetic fields and spontaneous abortion.
We thank our scientific advisory panel (George Hutchison, Martin Misakian, Lowell Sever, and Nancy Wertheimer) for critical comments, the staffs of the Kaiser Permanente Medical Care facilities in the study areas for their help, and Cecily Wilder for manuscript preparation.
1. Florig HK, Hoburg JF. Power-frequency magnetic fields from electric blankets. Health Phys 1990; 58:493–502.
2. Preston-Martin S, Peters JM, Yu MC, Garabrant DH, Bowman JD. Myelogenous leukemia and electric blanket use. Bioelectromagnetics 1988; 9:207–213.
3. Wertheimer N, Leeper E. Possible effects of electric blankets and heated waterbeds on fetal development. Bioelectromagnetics 1986; 7:13–22.
4. Meyer RE, Aldrich TE, Easterly CE. Effects of noise and electromagnetic fields
on reproductive outcomes. Environ Health Perspect 1989; 81:193–200.
5. Brent RL. Reproductive and teratologic effects of electromagnetic fields
. In: Committee on Interagency Radiation Research and Policy Coordination: Health Effect of Low Frequency Electric and Magnetic Fields: Oak Ridge Associated Universities. Washington DC: Office of Science Technology and Policy, 1992; V1–V70.
6. Wilson BW, Wright CW, Morris JE, Buschbom RL, Brown DP, Miller DL, Sommers-Flannigan R, Anderson LE. Evidence for an effect of ELF electromagnetic fields
on human pineal gland function. J Pineal Res 1990; 9:259–269.
7. Dugos L, Vena J, Byers T, Sever L, Bracken MB, Marshall E. Congenital defects and electric bed heating in New York State: a register-based case-control study. Am J Epidemiol 1992; 135:1000–1011.
8. Milunsky A, Ulcickas M, Rothman KJ, Willett W, Jick SS, Jick H. Maternal heat exposure and neural tube defects. JAMA 1992; 268:882–885.
9. Li D, Checkoway H, Mueller BA. Electric blanket use during pregnancy in relation to the risk of congenital urinary tract anomalies among women with a history of subfertility. Epidemiology 1995; 6:485–489.
10. Savitz DA, John EM, Klecker RC. Magnetic field exposure from electric appliances and childhood cancer. Am J Epidemiol 1990; 131:763–773.
11. Bracken MB, Belanger K, Hellenbrand K, Dlugosz L, Holford TR, McSharry J-E, Addesso K, Leaderer B. Exposure to electromagnetic fields
during pregnancy with emphasis on electrically heated beds: association with birthweight and intrauterine growth retardation. Epidemiology 1995; 6:263–270.
12. Belanger K, Leaderer B, Hellenbrand K, Holford TR, McSharry J, Power ME, Bracken MB. Spontaneous abortion and exposure to electric blankets and heated water beds. Epidemiology 1998; 9:36–42.
13. Juutilainen J, Matilainen P, Saarikoski S, Läära E, Suonio S. Early pregnancy loss and exposure to 50-Hz magnetic fields. Bioelectromagnetics 1993; 14:229–236.
14. Delgado JMR, Leal J, Monteagudo JL, Garcia MG. Embryological changes induced by weak, extremely low frequency electromagnetic fields
. J Anat 1982; 134:533–551.
15. Juutilaiinen J, Saali K. Development of chick embryos in 1 Hz to 100 kHz magnetic fields. Radiat Environ Biophys 1986; 25:135–140.
16. Pafkova H, Jerabek J. Interaction of MF 50 Hz, 20 mT with high dose of X-rays: evaluation of embryotoxicity in chick embryos. Rev Environ Health 1994; 10:235–241.
17. Farrell JM, Litovitz TL, Penafiel M, Montrose CJ, Doinov P, Barber M, Brown KM, Litovitz TA. The effect of pulsed and sinusoidal magnetic fields on the morphology of developing chick embryos. Bioelectromagnetics 1997; 18:431–438.
18. Olcese J, Reuss S, Vollrath L. Evidence for the involvement of the visual system in mediating magnetic field effects on pineal melatonin synthesis in the rat. Brain Res 1985; 333:382–384.
19. Stevens RG. Electric power use and breast cancer: a hypothesis. Am J Epidemiol 1987; 125:556–561.
20. Swan SH, Waller K, Hopkins B, Windham GC, Fenster L, Schaefer C, Neutra RR. A prospective study of spontaneous abortion: relation to amount and source of drinking water consumed in early pregnancy. Epidemiology 1998; 9:126–133.
21. Breslow NE, Day NE. Statistical Methods in Cancer Research. vol. 1. The Analysis of Case-Control Studies. IARC Scientific Pub. No. 32. Lyon: International Agency for Research on Cancer, 1980.
22. Kline J, Stein Z. Spontaneous abortion. In: Bracken M, ed. Perinatal Epidemiology. London: Oxford University Press, 1984; 23–51.
23. Litovitz T, DiCarlo A, Farrell JM. Why is it possible for ELF magnetic fields to induce bioeffects? (Abstract). The Bioelectromagnetics Society (BEMS) Twentieth Annual Meeting, June 1998.
24. Wilson BW, Lee GM, Yost MG, Davis KC, Heimbigner T, Buschbom RL. Magnetic field characteristics of electric bed-heating devices. Bioelectromagnetics 1996; 17:174–179.