Exposure to Extremely Low Frequency Magnetic Fields Among Working Women and Homemakers

Transcription

1 PII: S (01) Ann. occup. Hyg., Vol. 45, No. 8, pp , British Occupational Hygiene Society Published by Elsevier Science Ltd. All rights reserved Printed in Great Britain /01/$20.00 Exposure to Extremely Low Frequency Magnetic Fields Among Working Women and Homemakers ALLISON L. MCCURDY, LOUIS WIJNBERG, DANA LOOMIS, DAVID SAVITZ and LEENA A. NYLANDER-FRENCH * Department of Environmental Sciences and Engineering, School of Public Health, The University of North Carolina at Chapel Hill, Chapel Hill, NC 27599, USA; Department of Epidemiology, School of Public Health, The University of North Carolina at Chapel Hill, Chapel Hill, NC 27599, USA Given concerns with potential health effects of exposure to magnetic fields, the goal of this study was to examine the magnitude and sources of occupational and residential exposure to extremely low frequency (primarily 60 Hz) magnetic fields among women. Exposure to 60 Hz magnetic fields was surveyed among cases and controls recruited for a study of breast cancer in 25 counties in North Carolina. The 273 women who participated wore an integrating personal magnetic-field exposure meter (AMEX 3-D) that measured their time-weighted average (TWA) exposure. A questionnaire was administered to determine the duration and frequency of electric appliance and machinery use. The geometric mean (GM) of the TWA exposure for employed women was mt (range mt) and for homemakers mt (range mt). Women working in manufacturing and industrial facilities had the highest exposure (GM mt, range mt), while nurses and health technicians (GM mt, range mt) and teachers and school administrators (GM mt, range mt) had the lowest exposures. Job titles, unless very limited in scope and/or environment, self-reported information about equipment use, potential exposure sources, time, and distance were not good predictors of magnetic-field exposure. Furthermore, the results show that occupations previously observed to have increased risk of breast cancer, such as teachers, nurses, administrative support, and housewives, did not have elevated average magnetic field exposures. Therefore, it is questionable whether exposure to power frequency magnetic fields is the cause of the increased breast cancer risk seen in these occupations British Occupational Hygiene Society. Published by Elsevier Science Ltd. All rights reserved Keywords: magnetic field; exposure variability; mixed-effects multiple linear regression model; women; breast cancer INTRODUCTION The association between exposure to extremely low frequency electric and magnetic fields (ELF EMF) and cancer has been investigated in many epidemiological studies. A majority of the studies has focused on brain cancer and leukemia (Floderus et al., 1993; Kheifets et al., 1997; Sahl et al., 1993; Savitz et al., 1988; Theriault et al., 1994). However, breast cancer has been of increasing interest in relation to magneticfield exposure (Coogan et al., 1996a,b; Loomis, 1991; Received 21 August 2000; in final form 27 February *Author to whom correspondence should be addressed. Tel.: ; Fax: ; leena french@unc.edu Loomis et al., 1994; Stenlund and Floderus, 1997; Stevens, 1987; Stevens and Davis, 1996). Savitz (1993) noted that the most pressing need in this research area is to measure patterns of magneticfield exposures in a variety of worker populations and then to perform traditional epidemiological evaluations of cancer occurrence. Many of the past magnetic-field exposure assessments have focused on men and electrical occupations. Recently, several investigators have examined whether workers in particular occupations have a greater risk of developing breast cancer (Calle et al., 1998; Coogan et al., 1996a; Habel et al., 1995; Morton, 1995; Petralia et al., 1998; Pollán and Gustavsson, 1999). However, accurate exposure measurements for different occupations held by women are needed to assess the evidence for 643

2 644 A. L. McCurdy et al. a causal relationship between breast cancer and occupational exposure to magnetic fields and to help plan future studies. The goal of this study was to monitor women in a variety of jobs to gain knowledge about how and to what extent women are exposed to magnetic fields at work and at home. MATERIALS AND METHODS Study population Study participants were identified from a sample of women who were participating in the first phase of the Carolina Breast Cancer Study (CBCS) (Newman et al., 1995). Briefly, the CBCS is an ongoing population based case-control study of risk factors in the development of breast cancer. Only CBCS subjects who had participated in one sub-study of the project were asked to participate in this investigation of magnetic-field exposures. Participants were between the ages of 27 and 79 yr and included both breast cancer cases and controls who lived in 25 counties in central and eastern North Carolina. The cases were identified from the North Carolina Cancer Registry and the controls were identified from the Division of Motor Vehicles (under age 65) and Health Care Financing Administration (Medicare) (age 65 79). The CBCS investigators did not reveal the cancer status of the participants to the investigators in this study. Eight hundred participants were initially identified. However, because inclusion of retired persons resulted in an excessive proportion of homemakers, all persons over the age of 65 were removed from the contact list limiting the number of potential participants to 610. A total of 273 women (45%) agreed to participate in the study. The reasons for non-participation were unwillingness to participate (103 women, 17%) or the participant was unreachable (222 women, 36%) due to disconnected phone, wrong number, change of address, or subject s unavailability during at least four contact attempts. Of the non-participants, 71% were over the age of 45 while 29% were under 45, and 54% were white, 43% black, and 3% of another race. Twelve women (2%) had died since their initial participation in the CBCS. After initial agreement to participate, 71 women (12%) were ultimately excluded from the study due to meter malfunctions, improper use of the meter, failure to wear the meter within the study period, family emergencies, illness, or death. A total of 202 women (33%) provided data appropriate for further analyses. This study was reviewed and accepted by the University of North Carolina School of Medicine Institutional Review Board. Interview and questionnaire Prior to measurement, a brief telephone interview was conducted concerning potential magnetic-field exposure sources at work (employed women only) or at home (homemakers only). Questions were targeted to collect information about the types of electrical equipment or appliances commonly used, the typical duration of use, and the approximate distance of the user from the equipment. Employed women were asked about their use of small (i.e., handheld), medium (i.e., larger than handheld, up to the approximate size of a copy machine), and large (e.g., elevators, large industrial equipment, electrical equipment rooms, generators, air conditioning/heating units) equipment that were potential magnetic-field sources. Homemakers were asked about the proximity of power lines to their home, the type of residence (i.e., apartment, duplex, single-family home, townhouse, or other), the size of the home, and the type of water pipes (i.e., metal or plastic) in the home. The interview data were used to help interpret exposure measurement information. Exposure measurements Exposure measurements were collected using an AMEX 3-D meter, which is a small integrating personal magnetic-field exposure meter that measures in the Hz range but is calibrated to a unit response with 60-Hz magnetic fields (Enertech Consultants, Campbell, CA). This meter has an accuracy of ±20% over the range of 0.02 to 15 µt and a linear response from 0.05 to 9.7 µt (Enertech Consultants, 1990). The AMEX 3-D meter only produces a timeweighted average (TWA) and is not able to demonstrate exposure peaks like other meters such as the EMDEX. However, the TWA has been indicated to be a useful exposure metric (Armstrong et al., 1990; Bracken and Patterson, 1996; Savitz et al., 1994). Furthermore, the AMEX 3-D meter has the advantage of being transportable and easy to use by the participant. Calibration of the AMEX-3D meters was conducted by Enertech Consultants at the beginning and at the conclusion of the study. Women were instructed to wear the AMEX 3-D meter inside a pouch attached to the belt or in a pocket. A TWA was calculated by dividing the cumulative exposure intensity value (in µt-h, read from the meter) by the time the instrument was worn (in hours, reported by the participant). Whenever possible, two measurements were collected for each participant. Potential participants were contacted in the order in which they appeared on the alphabetical list of contacts. Upon agreement to participate, the telephone interview was completed, and a meter, consent form, and blank diary form were sent to the participant. The participant decided the day on which she would wear the meter. Employed women were instructed to turn the exposure meter on immediately upon arriving at work and to turn it off just prior to leaving work. Homemakers were instructed to turn the meter on immediately upon waking up in the morning and to turn it off just prior to going to bed at night. All parti-

3 Extremely low frequency magnetic field exposure 645 cipants were instructed to wear the meter continuously during the monitoring period regardless of whether they left their workplace or home. Upon the return of a meter that had been worn by a participant, the meter was electrically discharged to yield the TWA measurement for the day worn. A diary of tasks and/or jobs completed by the participant accompanied each measurement. The diary elicited information about specific tasks and the use of electrical equipment, duration of equipment use, and distance from equipment on the day the meter was worn. From the diary, exposure time was estimated by calculating the percentage of time the participant reported being within three feet of an electrical appliance or piece of equipment that was on during the period the meter was worn. Although every attempt was made to collect a second full-day measurement between one and two months after the first measurement, this was not always possible (e.g., participant was not reached immediately, participant s personal situation precluded another measurement) and led to variation of times between the measurements. A questionnaire asking about any changes at work or in routine (e.g., new job, different position, no longer working) was sent to each participant at the time of the second measurement. Statistical analysis All data were coded and entered into a computer database. After invalid measurement data were excluded, descriptive statistics and multiple regression analyses were generated with the SAS system software PC version 6.12 (SAS Institute, Cary, NC) and all differences were evaluated at a two-tailed significance level of Exposure measurements were log-transformed for statistical analyses. Normality of the distribution of exposure by group and sample was tested using PROC Univariate. Several studies have indicated that magnetic-field exposures are approximately normally distributed when log-transformed (Bowman et al., 1992; Bracken and Patterson, 1996; Guenel et al., 1993; Sahl et al., 1994). A repeated measurement strategy similar to that of Kromhout et al. (1995) was used to investigate exposures and to elucidate the relative influence of worker, group (job), and environment (work or home) on magnetic-field exposure. Employed women were placed in occupational groups (selected a priori) based on their job title and their description of work tasks. The job groups of teachers/school administrators and nurses/medical technicians were selected, because women are a majority in these occupations. A large number of women also work in offices completing tasks that are difficult to classify more specifically than as administrative tasks. Therefore, women working in an office environment who did not perform a very specialized task were consolidated into one group, office workers. Computer programmers/technicians and industrial workers were chosen as job groups, because women in these occupational areas have potential to be exposed to high magnetic fields. The miscellaneous group was used to accommodate women who did not fall into the other job groupings, but worked outside the home. The within-person and between-person variance components were estimated by applying mixed-model multiple linear regression analysis (PROC Mixed) with a RANDOM statement (Littell et al., 1996b). Mixed-model multiple linear regression analysis (PROC Mixed) with a REPEATED statement (Littell et al., 1996a) was used to investigate the relationship between magnetic-field exposure and fixed effects representing occupation (job group) and measurement (i.e., 1 or 2) with exposure time, as calculated from the diary (percentage of time participant reported being within 3 ft of a potential magnetic-field source while wearing the meter) as a covariate. A general model [Model (1)] was constructed using standard regression techniques to determine significant fixed effects. Y ijk ln(x ijk ) m Y a i g ij c ijk R ijk e ijk, (1) for i 1, 2,, G job groups, j 1, 2, j i workers in group i, and k 1, 2 measurements per subject, where X ijk the magnetic-field exposure of the kth measurement from the jth worker in the ith job group, Y ijk the natural log-transformed value of X ijk, the overall mean, m Y a i g ij the fixed effect for the ith job group, the random effect for the jth worker in the ith group, c ijk the fixed effect for the exposure time expressed in the diary for the kth measurement on the jth worker in the ith job group R ijk e ijk the exposure time during the kth measurement on the jth worker in the ith job group, the random effect for the kth measurement on the jth worker in the ith job group. The effect of reported use of medium-size equipment (the most commonly reported type of equipment) on magnetic-field exposure was assessed for employed women. The effect of power line distance, type of home, size of home, and water pipe material on magnetic-field exposure was assessed for homemakers. Both assessments were made using mixed-effects linear models (PROC Mixed) with the log-transformed magnetic-field exposure as the response variable. Two models were employed to assess these effects, one for employed women [Model (2)] and one for homemakers [Model (3)]. Model (2)

4 646 A. L. McCurdy et al. for employed women is given by the following expression: Y ijkl ln(x ijkl ) m Y a i b j g ijk ijkl, (2) for i 1, 2 distance categories from medium-size equipment (generally 3 ft from medium-size equipment or generally >3 ft from medium-size equipment), j 1, 2,, C categories of medium-size equipment, and k 1, 2, k ij workers at distance i using j categories of medium-size equipment category, and l 1, 2 measurements per subject, where X ijkl the magnetic-field exposure of the lth measurement from the kth worker exposed at ith distance to the jth categories of medium-size equipment, Y ijkl the natural log-transformed value of X ijkl, m Y the overall mean, a i the fixed effect for the ith distance, b j the fixed effect for the jth category of medium-size equipment, γ ijk the random effect for the kth worker at ith distance using the jth category of mediumsize equipment, e ijkl the random effect for the lth measurement on the kth worker exposed to the jth category of medium-size equipment in the ith job group. Model (3) for homemakers is given by the following expression: Y ijklmn ln(x ijklmn ) m Y a i b j c k (3) d l g ijklm ijklmn, for i 1, 2 distance categories of the home from a power station or power lines ( 30 ft or >30 ft), j 1, 2,, T types of homes, k 1, 2, 3 size categories of home ( 1000 ft 2, ft 2, >2000 ft 2 ), l 1, 2 types of water pipes (metal or plastic), m 1, 2,, m ijkl homemakers living ith distance from powerlines in jth type of home of kth size having lth type of water pipes, n 1, 2 measurements per subject, where X ijklmn the magnetic-field exposure of the nth measurement from the mth homemaker living ith distance from powerlines in the jth type of home of the kth size having lth type of water pipes, Y ijklmn the natural log-transformed value of X ijklmn, m Y the overall mean, a i the fixed effect for the ith distance, the fixed effect for the jth category of home, b j c k d l the fixed effect for the kth size of home, the fixed effect for the lth type of water pipe, g ijklm e ijklmn the random effect for the mth homemaker living at ith distance from powerlines in a jth type of home of kth size having lth type of water pipes, the random effect for the nth measurement on the mth homemaker living at ith distance from powerlines in a jth type of home of kth size having lth type of water pipes. The underlying distribution of magnetic-field exposures was assumed to be lognormal. In each of the three models, it was further assumed that the subject random effect (g) and the measurement random effect (e) were mutually independent and normally distributed with means of zero and variances s 2 B and s 2 W, representing the between- and within-subject variance components, respectively. For Model (1), it follows that E(Y ijk ) m Y a i c ijk R ijk, for Model (2) E(Y ijk ) m Y a i b j, and, for Model (3) E(Y ijkl ) m Y a i b j c k d l. RESULTS Study population At the completion of the study, a total of 333 measurements were obtained from 202 women. Of the participants, 133 were employed and 69 were homemakers. Thirteen percent of the participants were between the ages of 30 and 40, 39% between the ages of 41 and 50, 33% between the ages of 51 and 60, and 15% over the age of 60. The study population consisted of 81% Caucasians and 19% African- Americans; the proportions were similar for employed women and homemakers. A wide range of occupations typically held by women, including nurses and teachers, was obtained. For the employed women, the following six groups were evaluated: (1) teachers and school administrators (14% of the study participants), (2) nurses and health technicians (7%), (3) office workers (27%), (4) computer programmers and technicians (4%), (5) industrial workers (3%), and (6) miscellaneous workers (11%). Women who were not employed but worked as volunteers outside the home were classified as employed women and were included in the groups according to their job tasks. The homemaker group (34%) consisted of women who did not report working outside the home (homemakers and retired persons) and they were treated as a control group for the employed group. Exposure assessment Approximately 2/3 of the participants (136 women) provided two magnetic-field measurements, while the remaining 1/3 (66 women) provided only one measurement. The proportion of employed women and homemakers providing two measurements did not differ (68 and 65%, respectively).

5 Extremely low frequency magnetic field exposure 647 The geometric means and 95% confidence intervals for the TWA magnetic-field exposures for all groups are presented in Table 1. The geometric mean of the TWA exposure for employed women was µt (range µt) and for homemakers µt (range µt). Group 1 (teachers and school administrators) had the lowest average exposure of all the occupational groups (geometric mean µt, range µt) while Group 5 (industrial workers) had the highest exposure (geometric mean µt, range µt). Comparisons of the within- and between-person variance components for each occupational group are shown in Table 1. The within-person variation was less than the between-person variation in every group. Group 2 (nurses and health technicians) and Group 4 (computer programmers and technicians) had the smallest estimated within- and between-worker variance components while Group 5 (industrial workers) had the greatest estimated within- and betweenworker variance components. Mixed-effects regression models were applied to investigate the association between the log-transformed magnetic-field exposure and the fixed effects (job group, distance from equipment, exposure time listed in the diary, proximity of home to power lines, type of home, size of home, and type of water pipe). This analysis included all participants regardless of the number of measurements collected. For Model (1), the occupation was strongly associated with magnetic-field exposure (the P-value for the global test was ). The log-transformed mean magnetic-field exposure for the employed women was statistically different from the homemakers (the P-value for the global test was ). As shown in Table 2, the occupational groups that were significantly different from the homemakers were Group 5 (industrial workers, P 0.015) and Group 6 (miscellaneous workers, P 0.004). When Groups 5 and 6 (14% of the employed women) were removed, the difference between the mean magneticfield exposure of the employed women and the homemakers was no longer statistically significant (the P- value for the global test was 0.159). The exposure time as calculated from the diary (percentage of time participant reported being within 3 ft of a potential magnetic-field source while wearing the meter) was not a significant predictor of magneticfield exposure (P 0.539) (Table 2). Since only a few participants used small- or largesize equipment, the variable medium-size equipment was investigated in more detail as an explanatory variable for exposure. Regarding Model (2), the use of a typewriter was significantly associated with elevated magnetic-field exposure for employed women (P 0.006) while the use of a copy machine was associated with decreased magnetic-field exposure (P 0.046) (Table 3). The distance variable and other categories of medium-size equipment were not predictive of magnetic-field exposure. Regarding homemakers [Model (3)], the proximity to above-ground power lines (P 0.293), the type of home (P 0.239), or the size of home (P 0.863) were not predictive of magnetic-field exposure. The type of water pipe was a significant predictor of magnetic-field exposure (P 0.033). However, the magnetic-field exposures of women reporting metal water pipes in the home was significantly lower than those of women reporting plastic water pipes (P 0.01). Since metal pipes can be a significant source of Table 1. Geometric mean (µt) and 95% confidence intervals for the TWA magnetic-field exposure and within- and between-person variance components of the log-transformed magnetic-field exposure for each group Group No. of No. of Geometric Geometric Between-person Within-person workers samples mean 95% CI variance variance s 2 BW BWS g s 2 WW WWS g 1. Teachers and school administrators 2. Nurses and health technicians 3. Office Workers Computer programmers and technicians 5. Industrial workers Miscellaneous All employed women Homemakers CI Confidence interval for the geometric mean. s 2 BW Variance component due to person. BWS g Estimated geometric standard deviation of between-person distribution of exposure. s 2 WW Variance component due to days. WWS g Estimated geometric standard deviation of within-person distribution of exposure.

6 648 A. L. McCurdy et al. Table 2. Point estimates and standard errors for fixed effects related to log-transformed magnetic-field exposure [Model (1)] Fixed effect Estimate Standard error P-value Intercept (homemakers) Teachers and school administrators Nurses and health technicians Office workers Computer programmers and technicians Industrial workers Miscellaneous workers Exposure time Table 3. Point estimates and standard errors for fixed effects related to log-transformed magnetic-field exposure of the employed women (occupational groups 1 6) [Model (2)] Fixed effect Estimate Standard error P-value % Subjects a Intercept b Distance 3 ft c Computer Copy machine Communication equipment Typewriter Audiovisual equipment Medical equipment Lawn care equipment Cleaning equipment Other medium-size equipment a Percent of subjects who reported the use of medium-size equipment. b Variables (1) distance >3 ft and (2) no use of each of the medium-size equipment were used as references. c Participant generally worked within 3 ft of medium-size equipment. ground currents (General Electric Company, 1989; Zaffanella, 1992), which are presumed to contribute to residential magnetic field exposure, it is surprising that women who reported metal pipes had lower magnetic field exposure. DISCUSSION The lack of exposure measurements continues to be a significant obstacle to drawing conclusions about the association between magnetic-field exposure and potential health risks (e.g., breast cancer). Bracken (1993) observed that magnetic fields in the workplace appear to be comparable with residential levels, unless a clearly defined high-current source is present. We observed a substantial difference between employed women and homemakers using the mixedeffects Model (1). However, when Group 5 (industrial workers) and Group 6 (miscellaneous workers) were removed from the mixed-effects analysis, there was no longer a statistically significant difference between the employed women and the homemakers, suggesting that exposure of other employed women were similar to those of homemakers. Conflicting results have been reported about the variability of magnetic-field exposures in different occupational groups (Deadman et al., 1988; Guenel et al., 1993; Kromhout et al., 1995; van der Woord et al., 1999). Regardless of group, the within-worker variation was less than the between-worker variation (Table 1). This suggests that magnetic-field exposure for each participant was more stable over time than it was between other participants in the same occupational group. This observation may partially be due to the fact that the participants were able to decide themselves which day to wear the meter and, therefore, they could have chosen similar days for both measurements. However, this seems unlikely, since at least one month was allowed between the measurements. Group 1 (teachers and school administrators), Group 2 (nurses and health technicians) and homemakers were fairly homogenous with minimal exposure variability. Therefore, the group means could be used as exposure estimates in epidemiological studies with some confidence. Group 4 (computer programmers and technicians) also had minimal variability, but the sample size was small (N 6). Group 3 (office worker) included a large number of job titles and a fair amount of variability diluting the usefulness of the obtained exposure estimate in epidemiological studies. It was not possible to separate the office workers into smaller groups because of the relatively small sample size and the large number of

7 Extremely low frequency magnetic field exposure 649 job titles given by the participants. However, a majority of the office workers (98%) did claim to routinely use a computer during their workday. It is unlikely that partitioning the office workers into more specific and smaller groups by job title would have produced significantly different results because of the underlying amount of variability within the group. Because of the limited number of participants working in industrial jobs (Group 5) and the amount of variation observed in this occupational group, this data should be used with caution when considering the use of the obtained exposure estimate for other industrial workers in epidemiological studies. The source(s) that could have contributed to the high magnetic-field exposure of several of the jobs in Group 6 (miscellaneous workers) were obvious. For example, the Sheriff s Office dispatcher (mean TWA µt) was surrounded by electronic maps, radios, and computers, the hospital cook (mean TWA µt) and school cafeteria manager (mean TWA µt) worked with industrial cooking equipment including ovens, steam kettles, and walkin freezers, and the public library assistant (mean TWA µt) used computers and electronic scanners. However, several participants with above average exposure did not have obvious sources of magnetic-field exposure in their work environment. For example, an artist, who reportedly did not use any electrical equipment, had a mean TWA of µt. However, this is consistent with previously reported exposure of sculptors, painters, and other related artists (mean 0.25 µt, N 4) (Floderus et al., 1996). This demonstrates the difficulty and associated uncertainty of using jobs and job titles to classify magneticfield exposure in epidemiological studies. Several items [e.g., bonnet-type hairdryers used in beauty parlors and carpentry equipment (electric saws and sanders)], which could help explain increased magnetic-field exposure (TWA>0.2 µt) in the homemaker group, were reported several times. Unfortunately, the AMEX 3-D meter is unable to measure exposure peaks, which could confirm that these magnetic-field sources significantly contributed to the overall TWA measurement. The low magnetic-field exposures measured in homemakers, teachers and school administrators, nurses and health technicians, computer programmers and technicians, and office workers are consistent with results reported in other previously published occupational studies (Bracken and Patterson, 1996; Enertech Consultants, 1998b; Floderus et al., 1996; Schiffman et al., 1998). Twenty-eight percent of the employed women and 13% of the homemakers had mean TWA magnetic-field exposures above 0.2 µt, which has been used as an arbitrary boundary to define the exposed group in some epidemiological studies (NIEHS, 1995). Our results are consistent with the findings of a study of 1000 persons, which encompassed both occupational and residential magnetic-field exposures (Enertech Consultants, 1998a). In that study, about 21% of workers had greater than 0.2 µt TWA magnetic-field exposure during the atwork period and about 14% had greater than 0.2 µt TWA magnetic-field exposure while at home. Coogan et al. (1996b) tried to categorize magneticfield exposure for different occupations using information about time spent near magnetic-field sources. They observed a 43% increased risk of breast cancer in women who reported an occupation with potential for high exposure to magnetic-fields. However, they suggested that their study was limited by the potential for exposure misclassification. Our study demonstrated that self-reporting of proximity to the source was not very helpful in determining magnetic-field exposure. Our results are consistent with another study in which magnetic-filed exposures were not associated with the numbers and types of electrical office components (Schiffman et al., 1998). The fact that time and distance from electrical equipment did not correlate with magnetic-field exposure supports the hypothesis that electrical components in a home or office environment do not significantly contribute to overall magnetic-field exposure. This is logical, because magnetic-field strength for such point sources drops off as a function of the inverse cube of the distance. In general, information about equipment, potential exposure sources, time, and distance, which are important variables in the industrial hygiene assessment of some chemical exposures, do not appear to be good predictors of magnetic-field exposure. Our data provide information about magnetic-field exposures for women in several occupations, which have received little attention thus far. With the possible exception for industrial workers, women are not exposed to high levels of magnetic fields in commonly held occupations. However, more precise exposure measurements are needed to determine if this low-level exposure is important to the health of women. Based on the results of this study (i.e., nonsignificant difference in exposure between different occupational groups when the small industrial worker and miscellaneous groups were removed), it is questionable whether the increased risk of breast cancer observed in occupations such as teachers, nurses, administrative support, and housewives (Calle et al., 1998; Habel et al., 1995; Morton, 1995) is likely to be the result of exposure to magnetic fields. Furthermore, our study demonstrates that the usual factors for classifying exposure to magnetic fields for epidemiological studies, based upon time and proximity to various types of electrical equipment, were not good predictors of magnetic-field exposure. Acknowledgements We are grateful to all the volunteers who participated in this investigation. We also thank Susan Jackson and Beth Newman for their assistance and support. This project was supported by the National Cancer Institute Specialized Pro-

8 650 A. L. McCurdy et al. gram of Research in Excellence in Breast Cancer (P50- CA58223). REFERENCES Armstrong BG, Deadman JE, Theriault G. Comparison of indices of ambient exposure to 60-Hertz electric and magnetic fields. Bioelectromagnetics 1990;11: Bowman JD, Sobell E, London SJ, Thomas DC, Garabrant DH, Pearce N, Peters JM. Electric and magnetic field exposure, chemical exposure, and leukemia risk in electrical occupations. Report TR , Project , Electric Power Research Institute, Palo Alto (CA), Bracken TD. Exposure assessment for power frequency electric and magnetic fields. Am Ind Hyg Assoc J 1993;54: Bracken TD, Patterson RM. Variability and consistency of electric and magnetic field occupational exposure measurements. J Expos Anal Envir Epidemiol 1996;6: Calle EE, Murphy TK, Rodriguez C, Thun MJ, Heath SWJ. Occupation and breast cancer mortality in a prospective cohort of U.S. women. Am J Epidemiol 1998;148: Coogan PF, Clapp RW, Newcomb PA, Mittendorf R, Bogden G, Baron JA, Longnecker MP. Variation in female breast cancer risk by occupation. Am. J Ind Med 1996;30: Coogan PF, Clapp RW, Newcomb PA, Wenzl TB, Bogdan G, Mittendorf R, Baron JA, Longnecker MP. Occupational exposure to 60-Hertz magnetic fields and risk of breast cancer in women. Epidemiology 1996;7: Deadman JE, Camus M, Armstrong BG, Heroux P, Cyr D, Plante M, Theriault G. Occupational and residential 60-Hz electromagnetic fields and high-frequency electric transients: exposure assessment using a new dosimeter [published erratum appears in American Industrial Hygiene Association Journal 1996 Jun;57(6):580 3]. Am Ind Hyg Assoc J 1988;49: Enertech Consultants. AMEX-3D magnetic field meter (threeaxis model) operating instructions, version 1.00, Enertech Consultants, Campbell, CA, Enertech Consultants. The 1000 people study. EMF RAPID Program, Engineering Project #6, Department of Energy, Campbell, CA, 1998a. Enertech Consultants. Summary for EMF RAPID project #3. summary.html, Enertech Consultants, Campbell, CA, 1998b. Floderus B, Persson T, Stenlund C. Magnetic-field exposures in the workplace: reference distribution and exposures in occupational groups. Int J Occup Envir Hlth 1996;2: Floderus B, Persson T, Stenlund C, Linder G, Johansson C, Kiviranta J, Parsman H, Lindblom M, Wennberg A, Ost A, Knave B. Occupational exposure to electromagnetic fields in relation to leukemia and brain tumors: a case-control study in Sweden. Cancer Causes Contr 1993;4: General Electric Company. Pilot study of residential power frequency magnetic fields. EPRI EL-6509, Final report, Electric Power Research Institute, Palo Alto, CA, Guenel P, Nicolau J, Imbernon E, Warret G, Goldberg M. Design of a job exposure matrix on electric and magnetic fields: selection of an efficient job classification for workers in thermoelectric power production plants. Int J Epidemiol 1993;22:S Habel LA, Stanford JL, Vaughan TL, Rossing MA, Voigt LF, Weiss NS, Daling JR. Occupation and breast cancer risk in middle-aged women. J Occup Envir Med 1995;37: Kheifets LI, Kavet R, Sussman SS. Wire codes, magnetic fields, and childhood cancer. Bioelectromagnetics 1997;18: Kromhout H, Loomis DP, Mihlan GJ, Peipens LA, Kleckner RC, Iriye R, Savitz DA. Assessment and grouping of occupational magnetic field exposure in five electric utility companies. Scand J Work Envir Hlth 1995;21: Littell RC, Milliken GA, Stroup WW, Wolfinger RD. Analysis of repeated measures data, In: SAS system for mixed models. Cary, NC: SAS Institute; p Littell RC, Milliken GA, Stroup WW, Wolfinger RD. Random effects models, In: SAS system for mixed models. Cary, NC: SAS Institute; p Loomis DP. Cancer of breast among men in electrical occupations. Lancet 1991;339: Loomis DP, Savitz DA, Ananth CV. Breast cancer mortality among female electrical workers in the United States. J Nat Cancer Inst 1994;86: Morton WE. Major differences in breast cancer risks among occupations. J Occup Envir Med 1995;37: Newman B, Moorman PG, Millikan R, Qaqish BF, Geradts J, Aldrich TE, Liu ET. The Carolina Breast Cancer Study: integrating population-based epidemiology and molecular biology. Breast Cancer Res Treat 1995;35: NIEHS. Questions and answers about EMF: electric and magnetic fields associated with the use of electric power. DOE/EE-0040, National Institute of Environmental Health Sciences and U.S. Department of Energy, Research Triangle Park, NC, Petralia SA, Vena JE, Freudenheim JL, Marshall JR, Michalek A, Brasure J, Swanson M, Graham S. Breast cancer risk and lifetime occupational history: employment in professional and managerial occupations. Occup Envir Med 1998;55:43 8. Pollán M, Gustavsson P. High risk occupations for breast cancer in the Swedish female working population. Am J Pub Hlth 1999;89: Sahl JD, Kelsh MA, Greenland S. Cohort and nested case-control studies of hematopoietic cancers and brain cancer among electric utility workers. Epidemiology 1993;4: Sahl JD, Kelsh MA, Smith RW, Aseltine DA. Exposure to 60- Hz magnetic fields in the electric utility work environment. Bioelectromagnetics 1994;15: Savitz DA. Epidemiologic studies of electric and magnetic fields and cancer: strategies for extending knowledge. Envir Hlth Perspect 1993;101: Savitz DA, Ohya T, Loomis DP, Senior RS, Bracken TD, Howard RL. Correlations among indices of electric and magnetic field exposure in electric utility workers. Bioelectromagnetics 1994;15: Savitz DA, Wachtel H, Barnes FA, John EM, Tvrdik JG. Casecontrol study of childhood cancer and exposure to 60-Hz magnetic fields. Am J Epidemiol 1988;128: Schiffman A, Breyesse PN, Kanchanaraska S, Cutler T, Fan V. Characterization of extremely low frequency magnetic field exposures of office workers. Appl Occup Envir Hyg 1998;13: Stenlund S, Floderus B. Occupational exposure to magnetic fields in relation to male breast cancer and testicular cancer: a Swedish case-control study. Cancer Causes Contr 1997;8: Stevens RG. Electric power use and breast cancer: a hypothesis. Am J Epidemiol 1987;125: Stevens RG, Davis S. The melatonin hypothesis: electric power and breast cancer. Envir Hlth Perspect 1996;104: Theriault G, Goldberg M, Miller AB, Armstrong B, Guenel P, Deadman J, Imbernon E, To T, Chevalier A, Cyr D, Wall C. Cancer risks associated with occupational exposure to magnetic fields among electric utility workers in Ontario and Quebec, Canada, and France: Am J Epidemiol 1994;139: van der Woord MP, Kromhout H, Barregard L, Jonsson P. Within-day variability of magnetic fields among electric utility workers: consequences for measurement strategies. Am Ind Hyg Assoc J 1999;60: Zaffanella LE. Survey of residential magnetic field sources. EPRI TR , Interim report, Electric Power Research Institute, Palo Alto, CA, 1992.