Abstract

Heart rate variability (HRV) results from the action of neuronal and cardiovascular reflexes, including those involved in the control of temperature, blood pressure and respiration. Quantitative spectral analyses of alterations in HRV using the digital Fourier transform technique provide useful in vivo indicators of beat-to-beat variations in sympathetic and parasympathetic nerve activity. Recently, decreases in HRV have been shown to have clinical value in the prediction of cardiovascular morbidity and mortality. While previous studies have shown that exposure to power-frequency electric and magnetic fields alters mean heart rate, the studies reported here are the first to examine effects of exposure on HRV. This report describes three double-blind studies involving a total of 77 human volunteers. In the first two studies, nocturnal exposure to an intermittent, circularly polarized magnetic field at 200 mG significantly reduced HRV in the spectral band associated with temperature and blood pressure control mechanisms (P Å 0.035 and P Å 0.02), and increased variability in the spectral band associated with respiration (P Å 0.06 and P Å 0.008). In the third study the field was presented continuously rather than intermittently, and no significant effects on HRV were found. The changes seen as a function of intermittent magnetic field exposure are similar, but not identical, to those reported as predictive of cardiovascular morbidity and mortality. Furthermore, the changes resemble those reported during stage II sleep. Further research will be required to determine whether exposure to magnetic fields alters stage II sleep and to define further the anatomical structures where fieldrelated interactions between magnetic fields and human physiology should be sought. Bioelectromagnetics 19:98 -106, 1998. ᭧ 1998 Wiley-Liss, Inc. Key words: power-frequency; heart rate variability; EKG; HRV; EMF; Fourier transform INTRODUCTION tion has been paid to whether field exposure alters the natural variability in the human cardiac rhythm. A Since the late 1960s many research studies have metronome beats at a constant rate; a healthy heart does examined the biological consequences of exposure to not. Even in quiescent conditions, heart rate typically power frequency electric and magnetic fields. In the exhibits significant beat-to-beat variability. This type first human exposure study performed in our labora-of variability is not consciously perceived by a person; tory, a large number of physiological and performance and it should not be confused with heart rate reactivity, variables were screened with generally negative results the slowing or speeding of heart rate in direct response [Graham et al., 1987]. The most relevant exception for to perceived situational or personal stimuli (eg., exerpurposes of this report was a biologically small but cise, anxiety, relaxation, etc.). Heart rate variability statistically significant slowing of mean heart rate (lengthening of the cardiac interbeat interval) resulting from exposure to combined 60 Hz electric and mag-Contract grant sponsors: Electric Power Research Institute, The Departnetic fields. Field-related slowing of heart rate has con-ment of Energy, The National Institute of Environmental Health Sciences, Midwest Research Institute, and A.S. Consulting and Research, tinued to be observed in subsequent studies performed Inc. with healthy young men in our laboratory were screened to assure they met the specific criteria set Spectral analysis techniques have long been ap-for participation (male, age 18 to 35 years, no chronic plied in the field of cardiovascular physiology [e.g., disease or disability, no recent serious acute illness, no see monographs by McDonald, 1974 and by Milnor, medications, regular sleep habits, no night work). All 1982], and their use in performing quantitative analyses volunteers provided written informed consent. of HRV has become an area of emerging importance in clinical medicine. Certain alterations in HRV have Procedures newly-recognized prognostic value in a number of imProcedures common to all studies are described portant conditions [coronary artery disease, Hayano et first, followed by those specific to each of the three al., 1990; post-infarction risk, His vital signs were recorded, and the indwelling cathetinuously followed cohort study for cardiovascular disease in the world -have reported that quantitative ter for the collection of multiple blood samples was inserted in an arm vein. IBI was recorded using silverspectral assessment of HRV offers prognostic information for mortality risk beyond that provided by the silver chloride biopotential electrodes attached with double-stick adhesive tape to prepared skin sites on the evaluation of traditional risk factors The quantitative methods used by basic physiolo-right clavicle and the seventh intercostal space under the left ancillary midline, corresponding to the standard gists and clinicians are essentially identical to those we report here in our evaluation of alterations in HRV electrocardiographic Lead II configuration. Beckman electrode paste served as the contact medium. resulting from magnetic field exposure. This report describes three recent human studies in which volunteers Each man slept overnight in the Human Exposure Test Facility at MRI from 2300 h to 0700 h under slept through the night in the laboratory while being exposed to intermittent or continuous power-frequency double blind test conditions. After the volunteer got into bed in the exposure room, a blood sample was magnetic fields. Sufficient cardiac data were collected to make a detailed examination of HRV. The main obtained. The double-blind/field control system was activated at 2300 h. On sham control nights, the control purpose of these studies was not to evaluate HRV, but rather to determine if nocturnal exposure to magnetic system did not energize the field generating equipment. On exposure test nights, the field generating equipment fields suppressed plasma levels of the pineal hormone melatonin volunteers, and each man was sham exposed in one session and exposed to the magnetic field in another It should be noted that collection of cardiac data was not the major purpose of the studies reported here. test session. This study differed from the previous ones in that it evaluated the effects of continuous exposure If for any reason problems were encountered with the cardiac recording system (e.g., an electrode became to the magnetic field. The 60 Hz, circularly-polarized magnetic field was activated at 2300 h at a field strength loose), the experimenters were instructed to turn the system off and to do nothing that might interfere with of 200 mG, and it remained on continuously until 0700 h. Complete cardiac data were collected for 26 volunthe major thrust of the study, the collection of blood samples for melatonin assays. Therefore complete car-teers (52 all-night test sessions). Data were missing from proportionately more subjects in Study 3 than in diac data was available for only a subset of the subjects in each of the studies. Except for a requirement for Study 1, because complete data for both sessions was required for a subject to be included. completeness of the data, there were no other exclusions in the subset of subjects included in the cardiac Exposure Facility analyses reported here. Characteristics and control systems of the MRI Study 1 Study 1 used an independent groups design. Human Exposure Test Facility have been documented Thirty-three healthy young men were randomly as-and are described in Cohen et al. [1992] and in Dietrich signed to three groups matched on age, body size and et al. [1995]. A systematic protocol using test instrusensitivity of melatonin to light exposure: a sham con-ments traceable to the National Institute of Standards trol group; a group exposed to a 10 mG magnetic field, and Technology was followed to verify the exposure and a group exposed to a 200 mG magnetic field. Expo-characteristics of the facility and to calibrate the resure to the circularly-polarized field during the night cording equipment. was intermittent: 1 h off/1 h on. For example, midnight Subjects were exposed to a uniform (4% -7%) to 0100 h was ''on,'' 0100 to 0200 h was ''off'' and circularly-polarized 60 Hz magnetic field generated in so on. During the hours designated as ''on,'' the field the facility by six Helmholtz coils surrounding each of switched on and off every 15 s throughout the hour. the exposure rooms in both the vertical and horizontal For example, from 12:00:00 to 12:00:15 the field was axes. The horizontal field axis is oriented from the on, from 12:00:15 to 12:00:30 the field was off, and doorway to the rear of the exposure room, and the so on. IBI was recorded continuously through the night. vertical axis, from floor to ceiling. Each field axis is Complete cardiac data were collected on 29 of the 33 independently energized from an adjustable autotransvolunteers. former. The horizontal field current is shifted from the vertical field current by a phase angle of 90 degrees. Study 2 Study 2 used a repeated measures design in Subjects slept on a cot in the facility with their bodies which each subject served as his own control. Forty oriented in line with the horizontal field component healthy, young men participated in two all-night test (North to South). The facility was composed of two sessions. In one test session, all of the men were sham test rooms, A and B. In room A, the geomagnetic field exposed from 2300 h to 0700 h. In the other test seswas 44.1 mT; the horizontal component was 10.1 mT sion, all of the men were exposed to the identical and the vertical component was 42.9 mT. In room B, 200 mG intermittent exposure condition used in Study the geomagnetic field was 53.0 mT; the horizontal com-1. Since the 10 mG exposure condition used in Study ponent was 22.3 mT and the vertical component was 1 gave negative results on all measures examined, this 48.1 mT. Illumination in the facility is provided by condition was not used in Study 2 or Study 3. The incandescent lamps located above translucent ceiling volunteers were assigned randomly to two counterbalpanels and maintained at less than 10 lux during test anced orders of testing (sham/exposed, exposed/sham), sessions. Illumination levels were measured using a and all sessions were conducted double-blind. ComDigital Photometer (Model J16, Tektronix, Beaverton, plete cardiac data were collected for 22 volunteers (44 Oregon). all-night test sessions). Data were missing from proportionately more subjects in Study 2 than in Study 1, Measures because complete data for both sessions were required Cardiac Interbeat Interval, or IBI, is a measure for a subject to be included. of the duration in milliseconds between successive R waves in the cardiac cycle. As such, it provides a direct Study 3 Study 3 followed the design of Study 2. An additional 40 healthy, young men served as their own and sensitive measure of heart rate. recording system consisted of six special-purpose iso-focus of spectral analysis was based on the results of numerous psychophysical and medical research studies lation amplifiers with high common mode rejection and high input impedance, operating as the input to [see Porges and Bohrer, 1990; Data from Study 1 were also used to determine ply, and optical couplers were used to further isolate the amplifiers from the field generator. Physiological if exposure influenced the chaotic, aperiodic components of the human cardiac rhythm. For chaos-theory signals were selectively filtered for residual 60 Hz interference. The cardiac signals were conditioned analyses, we used an optomized FORTRAN implementation of Pincus' Approximate Entropy (ApEn) statisthrough the amplifiers in the measurement system, passed to a hardware detector (Beckman Instruments, tic, which is an approximation of the KolmogorovSinai Entropy statistic 1991; Pincus and Viscarello, 1992). Signal Processing Statistics Each volunteer generated 25,000 to 30,000 heartThe traditional approach consisted of performing beats over the course of a night. The series of continu-analysis of variance for mixed designs (ANOVA; Sysously recorded interbeat time intervals was converted tat for Windows, V. 5.0 for Study 1 and BMDP4V for to instantaneous heart rate, which provided a regularly-Studies 2 and 3) on both heart rate mean and heart rate spaced time series with a 1 s resolution. The time pe-standard deviation, as well as spectral data. In Study riod selected for analysis was midnight to 0600 h. Since 1, Group (sham, 10 mG, 200 mG) was the ''between blood was drawn every hour, a 10-min segment subjects'' variable. Hour (1 through 6) and Period (1 (5 min before and after the hour) was deleted to remove through 3) were the ''within subjects'' variables. In possible artifacts associated with obtaining blood sam-Studies 2 and 3, Order of exposure (sham, real vs. ples. The six 50-min ''hours'' were each further di-real, sham) was the ''between subjects'' variable. Field vided into three ''Periods'' corresponding to the first, (sham, 200 mG), Hour and Period were the ''within middle and last thirds of the hour. subject'' variables. Significant effects (P õ 0.05) and Spectral analyses using the digital Fourier trans-trends (P õ 0.10) were followed by simple effects form were performed on the cardiac time series to analyses. determine if exposure altered specific, periodic components of heart rate variability known to be mediated by the activity of the sympathetic and parasympathetic RESULTS branches of the autonomic nervous system on the heart. Study 1 For each Period, a time series of 1024 points was analyzed as follows: (1) any linear trend (including the The data from Study 1 were used to compare and contrast the usefulness of traditional, spectral and mean value) was first removed; (2) a Hamming window was applied; (3) a digital Fourier transform (DFT) was chaos-theory based approaches to analysis. Traditional measures of variability, such as the standard deviation performed; and (4) the resulting spectrum was smoothed using a 7-point moving average [Black-(SD), provided a way to compare the present results with the results of previous HRV studies [ Kamath and man & Tukey, 1958; Bloomfield, 1976; Press et al., 1986; Marple, 1987]. The results of the DFT are ex- Chaos-theory based analyses allowed the quantification series had a time resolution of 1 s, the Nyquist limit for this spectral analysis is 0.5 Hz. of aperiodic phenomena previously noted in medical studies of cardiac variability in patient populations. Spectral analyses provided measures of total power, absolute and relative power in specific frePerformance of traditional analyses using AN-OVA revealed no significant differences between the quency bands (e.g. 0 to 0.1 Hz and 0.15 to 0.40 Hz), as well as the ratios of power between specific bands. test groups in mean heart rate or heart rate standard deviation data. This was not unexpected, given the Selection of these particular frequency bands as the not affected by exposure group or by field-on versus field-off hours. Similar negative results were obtained when the ratio between power in the entire high band and total power was examined. As an example, the three panels in netic field significantly reduced percent total power in Performance of the chaos-theory based analyses the low spectral band (F Å 5.99; df Å 1,20; P Å 0.02). provided negative results. The healthy, young men who This result replicates our previous observation in Study participated in Study 1 exhibited chaotic heart rate variations within the normal range reported by others using the ApEn statistic In contrast to traditional and chaos-theory based techniques, spectral analyses performed on the cardiac time series revealed significant differences between the test groups. As shown in band in subjects exposed to 200 mG magnetic fields most likely results from a reduction in low-frequency Study 3 periodic components in the heart rate spectrum. Thus, intermittent exposure to 200 mG magnetic fields apThis study was undertaken to determine if continuous magnetic field exposure produces differential ef-pears to result in an alteration of the normal nocturnal variability that is inherent in human cardiac rhythm. fects on HRV in human volunteers. Twenty-six additional volunteers provided complete cardiac interbeat In addition, we also observed an increase in power at a spectral index of the normal respiratory arrhythmia interval data in both their sham control and magnetic field exposure test sessions. These data were submitted in volunteers exposed to intermittent 200 mG magnetic fields in the first study. to spectral analysis using the procedures described above. Unlike our previous results with intermittent The results obtained in Study 2 confirmed and extended our previous results. Spectral analysis of heart magnetic field exposure, we observed no significant decrease in the low band spectral power upon continu-rate time series again revealed a statistically significant reduction in the power of the low-frequency band ous field exposure (Field by Hour interaction F Å 0.71; df Å 5,120; P Å 0.61). Again in contrast to our previous (0.0 -0.1 Hz) in subjects exposed to 200 mG intermittent magnetic fields. The difference observed between results with intermittent exposure, we also failed to see significant increases in high band spectral power (Field sham control and magnetic field conditions was significant, as was the interaction between field conditions by Hour interaction F Å 1.33; df Å 5,120; P Å 0.26). HRV during exposure to the continuous magnetic field and time of night. Since systemic thermoregulation and blood pressure control are active at all times and are was not significantly different from HRV during sham control conditions. known to vary during the night, these results again suggest a field-induced alteration in an underlying normal physiological process. DISCUSSION Spectral analysis of heart rate time series also revealed a statistically significant increase in high band The magnetic field-induced alterations in heart rate variability on healthy human volunteers that we (0.15 Hz to 0.40 Hz) power in volunteers exposed to 200 mG intermittent magnetic fields. The power in this have observed have proven reproducible in two independent studies of strong design and high statistical frequency band is associated with the natural respiration-induced alteration of heart rate. The interaction power conducted under strict double-blind conditions. Exposure of human volunteers to intermittent 60 Hz between field conditions and time of night was significant. Since respiration-induced sinus arrhythmia is acmagnetic fields resulted in alterations in HRV. Spectral analyses of heart rate time series revealed statistically tive at all times, and is known to vary during the night, our results again suggest a field-induced alteration in significant reductions in the power of the low-frequency band (0 -0.1 Hz) in volunteers exposed to inter-an underlying normal physiological process. Unlike previous results with intermittent magmittent 200 mG magnetic fields, and a similar reduction in low band power when comparing sham vs. field netic field exposure, exposure to the continuous magnetic field did not significantly decrease low band exposure nights in the same individual. Power in this frequency band is associated with neural control of power or increase high band power. This is important because it indicates that intermittency of magnetic field thermoregulation and blood pressure control. A reduction in power in the low-frequency band exposure may be an important parameter in the human cardiac responses that result from those exposures. Incould be due to one of two distinct processes: (1) a reduction in low-frequency periodic components in the termittent exposure resulted in statistically significant alterations in HRV, and these effects were reproducible heart rate spectrum (e.g., Traub -Hering -Meyer wave), or (2) a reduction in aperiodic components in in two separate studies with non-overlapping human volunteers conducted in two different years. In contrast, the heart rate time series (e.g., chaotic variability). Chaos-theory based analyses of the heart rate time se-continuous field exposure had no observable effect on HRV. ries provided a direct examination of chaotic variability. When chaotic variability is present, it appears in Our analyses of the effects of exposure to intermittent fields (each study individually and combined) spectral analyses as very low frequency power. Analyses of heart rate time series using the Approximate have revealed two significant effects in opposite directions, a decrease in power in the low spectral band and Entropy (ApEn) statistic did not reveal effects of field exposure on this parameter. The combination of spec-an increase in power in the high spectral band. important to note that virtually all clinical studies of raised by these studies of magnetic field exposure on human heart rate variability. HRV as a predictor of cardiovascular risk report that higher risk is associated with a reduction in low band power, as we have observed in this study. The data are ACKNOWLEDGMENTS