Below are abstracts presented at several recent scientific meetings and a comment on the important Linet study published in the New England Journal of Medicine in 1997.
APPLICATIONS OF DECISION ANALYSIS TO MANAGING MAGNETIC FIELDS IN CALIFORNIA SCHOOLS. H.K. Florig1, M. Henrion2, A.R. Sheppard3, B. Bernstein4 and K. Soohoo3.1Carnegie Mellon Univ., Pittsburgh, Pennsylvania 15213, USA,2Lumina Decision Systems, Los Altos, California 94022, USA, 3Asher Sheppard Consulting, Redlands, California 92373, USA, 4EcoAnalysis, Inc., Ojai, California 93023, USA.
Some 10% of schools in the U.S. are situated close enough to transmission lines that magnetic field levels on school property are comparable to levels associated with the high risk cohorts in epidemiologic studies of residential EMF exposure and childhood cancer. This has created a tough policy problem for local and state governments as officials search for solutions that strike a balance between the costs of reducing exposure levels and the uncertain health and other benefits of such action. Although transmission lines are not the only source of magnetic field exposure to school children, transmission lines are the most frequent focus of parents' concerns about magnetic field exposure at school. We are in the midst of a three year effort to apply techniques of decision analysis to systematically evaluate policy options for this problem. By being explicit about objectives, values, and expectations about uncertain events, decision analysis makes it easier to identify areas of disagreement among stakeholders and to identify those decision components that most limit the ability to discriminate between policy alternatives. Opinions vary widely about what the goals of EMF policy for schools ought to be. Some subscribe to a narrow model that includes mitigation cost and health risk as the only outcomes that matter. Others consider a much broader range of objectives including reducing parental worry, maintaining confidence in public institutions, minimizing administrative burdens, and distributing risk more equally. Our research shows that policy preferences are highly dependent on which subset of these objectives is embraced. Within the narrow framework of risk cost tradeoffs, we find that the factors that contribute most to uncertainty in policy choices for individual schools are willingness to pay for risk reduction, the probability that milligauss levels of EMF exposure are really hazardous, and the efficacy of mitigation that reduces average field levels. For policies at the state level which affect many schools, additional factors that contribute substantially to uncertainty in policy choice are statewide mitigation costs and the population exposure reduction that would result from any given policy. The value of information for these two factors exceeds the costs of gathering that information by a wide margin. Future work will examine the policy implications of considering distribution lines and in building sources of magnetic field exposures to school children. This work was supported by the Electric and Magnetic Fields Program of the California Department of Health Services, Vincent DelPizzo, Director of EMF Extramural Research.
Top of Page
Annual Review of Research on Biological Effects of Electric and Magnetic Fields from the Generation, Delivery and Use of Electricity, San Diego, California, USA, November 9-16, 1997. (four abstracts)
A STRUCTURE FOR EVALUATING ALTERNATIVE STATEWIDE POLICIES ON POTENTIAL RISKS ASSOCIATED WITH EXPOSURE TO 60 HZ MAGNETIC FIELDS IN CALIFORNIA PUBLIC SCHOOLS AND DAYCARE CENTERS. L.G. Geissinger1, H.K. Florig2, B. Bernstein3 and J. Jostes4. 1Paradigm Planning and Research, Vashon, Washington 98070, USA, 2Carnegie Mellon University, Pittsburgh, Pennsylvania 15213, USA, 3EcoAnalysis, Inc., Ojai, California 93023, USA, 4Dudek Associates, Santa Barbara, California 93101, USA.
Studies of the possible human health effects of 60 Hz magnetic fields (EMF) have yielded mixed results, leaving varying perceptions about risk and the need for protective action. State agencies and organizations have been given limited support for taking a statewide perspective on concerns about EMF in schools, yet share responsibility for shaping public policy as it applies to assuring the safety of public schools, both now and in the future. The objectives of this research are to identify policy options at the statewide level for coping with the EMF problem in schools and to develop a framework for ongoing evaluation of these policy alternatives. A set of sample policies was developed based on a literature review and discussions with local school officials, statewide decision-makers, and a statewide steering committee. The policies reflect a range of approaches to risk and focus on actions that directly affect EMF exposure (e.g. field strength limits) or that indirectly reduce exposure by, for instance, providing information. Significant flexibility exists within each type of policy, based upon its level of implementation, financing mechanism, approach to public involvement, and relationship to ethical principles. Options in the sample policy set were qualitatively compared based on social, ethical, political, risk, economic, and procedural considerations. Distinctions are drawn between competing concepts of social and environmental justice, and between equity and fairness. Where appropriate, quantitative models of mitigation cost, exposure reduction, and risk reduction are used to analyze a subset of the total set of policy alternatives (see related abstract, H.K. Florig and M. Henrion). Decision analysis is a useful way for structuring parts of complicated policy problems, but must be recognized for both its strengths and limitations. The main strength of decision analysis is that it forces the discipline of systematic analysis on the policy debate, thus helping stakeholders identify more precisely their areas of agreement and disagreement. Limitations of decision analysis stem from difficulties in finding edifying representations of important qualitative attributes, and from simplifications that are needed to render a quantitative analysis analytically tractable. From a process perspective, decision analysis can be beneficial by helping stakeholders identify issues of most importance, but can also be alienating to those who are wary of placing too much command of policy analysis in the hands of technical experts. The focus of this project is not to find the "right" answer to the debate over what policy course should be adopted for EMF and schools; rather it is to put forward a structure for evaluating the problem that is more representative of underlying values and concerns than is possible using decision analysis alone. This work was supported by the Public Health Institute and the California Department of Health Services.
Top of Page
EXPLORING COST-BENEFIT TRADE-OFFS AND THE EFFECT OF WAITING IN DECISIONS INVOLVING EMF LITIGATION IN SCHOOLS. H.K. Florig1 and M. Henrion2. 1Carnegie Mellon University, Pittsburgh, Pennsylvania 15213, USA, 2Lumina Decision Systems, Inc., Los Gatos, California 95030, USA.
Environmental policy decisions involve choices of both the extent and timing of health-saving intervention. Pollution standards can be set at different levels, depending on the extent of protection desired and the costs of achieving that protection. Both the risk-saving potential and the costs of environmental interventions are usually quite uncertain, however, so it can be difficult to tell whether any given policy option is a good buy or not. For cases in which ongoing research can reduce uncertainties in risk and cost, delaying implementation of an intervention might be preferred to immediate implementation. We explore these questions in the context of 60 Hz magnetic field (EMF) mitigation in California schools. A cost-benefit model is used to examine the effects of extent and timing on the costs and benefits of a hypothetical decision to implement exposure standards for EMF exposure from net current loops. (Recent measurements of EMF exposure in California schools  show that net current loops are the most common source of exposure to above-average EMF levels in these facilities.) Future uncertainty in scientific knowledge about EMF risk is represented as a degree of expectation that scientists will reach consensus one way or the other on the EMF hazard question by any given year. The model also explicitly incorporates uncertainty in EMF hazard, mitigation costs, discount rates for mitigation expenditures and health benefits, the lifetime of mitigation investments, and many other factors. For a given exposure standard and timing strategy, the model computes probability distributions for the present value of mitigation costs (including the costs of survey and diagnosis), the amount of population exposure reduction, and the present equivalent population risk reduction. Given a user-specified willingness-to-pay to avoid morbidity and mortality risk, the model also computes the net benefits of the policy action. As one would expect, acting now is the dominant strategy when expectations are high that the science will soon show that EMF exposure is hazardous or when mitigation policies can be designed to address only the very most cost-effective cases. Estimates of the present value of the total costs of surveying, diagnosis, and mitigation for exposure standards addressing net current loops range from $10-50 million, depending on the level of the exposure standard and assumptions about unit costs. Estimates of the present equivalent childhood cancer risk avoided by exposure standards for net current loops range from zero to a few hundred cases, depending on the probability that current EMF epidemiologic evidence reflects a real EMF hazard, assumptions about dose-response, the level of the exposure standard, and other factors.  Zaffanella, L., "Progress Report on Field Reduction Cost Assessment," Enertech Consultants, Inc., May 17, 1997. This work was supported by the Public Health Institute and the California Department of Health Services.
Top of Page
CALIFORNIA SCHOOL EMF SURVEY AND FIELD REDUCTION COST ASSESSMENT-INTERIM RESULTS. V. DelPizzo1, L.E.Zaffanella2 and C. Hooper3. 1EMF Program, California Department of Health Services, Emeryville, California 94608, USA, Enertech Consultants, 2Lee, Massachusetts 01238, and 3Campbell, California 95008, USA.
Objective: An important part of the California EMF Program is a "School Exposure Assessment" project initiated in 1995 and scheduled for completion by the end of 1998. The purpose of the this project is to determine the level of EMF exposure in California public schools, identify field sources, evaluate field reduction techniques, and assess the cost of field reduction. The ultimate goal is to provide data for a quantitative analysis of policy options related to public schools. The objective of this paper is to present and discuss interim results obtained after completing two-thirds of the survey. Method: A sample of 90 schools was chosen to achieve the statistical goals of the study. The measurement protocol consists of determining the spatial field distribution of the rms (root mean squared) magnetic field in all areas of the schools, identifying the principal sources in each area, and recording the source parameters that may affect the cost of field reduction. The emphasis is on source identification and characterization. Power line characteristics are recorded and their lateral profile is measured. So are the characteristics of other "area sources": electrical panels, net currents, fluorescent lights, etc. The survey data are entered in a database which can be used in conjunction with exposure and cost algorithms to calculate exposure and mitigation costs. The data obtained from EMF measurements at 56 of the 90 schools were available for analysis. Some of the most important results obtained from the analysis of the data are reported and discussed in this paper. Discussion: The results obtained in 56 schools are evaluated in terms of the goals of the study. The structure of the California School EMF database is described. Examples of results obtained from this database are reported: (1) average magnetic fields measured in each of the approximately 3300 school areas surveyed, (2) magnetic field exceeded in a given percentage (e.g. 50%, 10%, 5%) of the space of each of the school areas, (3) breakdown of the previous results by type of area (classrooms, other indoor student occupied areas, staff occupied areas, outdoor areas), (4) percentage of areas in each school that exceeds specified magnetic field levels, (5) a separate evaluation for each source type (transmission line, distribution line, net current, electrical panel, etc.) of the average magnetic field produced by the that source type in each school area. The paper discusses results of several other types of measurements: electric fields, DC fields, magnetic field transients, and temporal variations of the magnetic field during the school day. The process of calculating the cost of reducing the magnetic field is described. The surveyed school's distribution of magnetic field in each school area is used to determine the number of school areas that meet a specified target (e.g. average field less than 1.7 mG, and 5 mG not to be exceeded in more than 5% of the space). The sources responsible for the field in the areas that do not meet the target are identified. For each of these areas, all known mitigation options are assessed to determine whether they are capable of reducing the field below the target value. The cost of implementing the various options is calculated and the option corresponding to the lowest cost is determined. The cost of meeting a given target is then calculated for each school by adding the costs for each area. This work is performed as a contract with the Public Health Institute and is managed by the California Department of Health Services.
Top of Page
ELECTRIC BLANKET MAGNETIC FIELDS BY SETTING, DURATION OF USE, AND POSITION OF THE METER FROM THE BLANKET. G. Lee, R. Neutra and L. Hristova. California Department of Health Services, Berkeley, California 94704, USA.
Introduction: Electric blankets are considered a major source of night-time residential magnetic field exposure because they are generally used throughout the night over several months, are in close contact with the body, and are believed to emit stronger (10-20 mG) fields than other residential sources. Hence, electric blankets were hypothesized to be a potential risk factor for adverse health effects through the effects of magnetic fields and suppression of the night-time melatonin peak. The results of our prospective study of the risk of self-reported use of electric blankets on spontaneous abortions did not support this hypothesis; electric blanket users had a lower rate of spontaneous abortions than non-users. Objective: A small exposure assessment study of conventional electric blankets was conducted to determine: 1) how the magnetic field measurements were influenced by setting, and by position of the measurement meter on the mattress from the blanket (representing exposure to the skin, to the ovary, and to the retina); 2) if an estimate of dose using setting and duration of use for various metrics was high enough, relative to background exposures, to substantially influence the magnetic fields delivered to the retina, to the ovary or to the skin; and 3) if there was an exposure metric which would convert the "messy" setting-duration dose response curve findings of the prospective electric blanket study. Conclusions: For the main hypothesized receptor for magnetic fields, the retina, users of electric blankets would not receive an excess of exposures compared to non-users. If the uterus was a receptor for magnetic fields, only women who use the blanket at medium or high settings for at least half of the night would receive an exposure substantially higher than non-users. This small group of women did not have elevated miscarriage rates in our prospective study. Only for the skin, which has not been hypothesized as a receptor for magnetic fields, would electric blanket users of high to medium settings for most of the night receive substantially higher magnetic field exposures than non-users.
Top of Page
Bioelectromagnetic Society Twentieth Annual Meeting, St. Petersburg, Florida, USA, June 8-11, 1998 (two abstracts)
INCONSISTENCIES BETWEEN PROSPECTIVE AND RETROSPECTIVE RESULTS OF THE CALIFORNIA STUDY OF PREGNANCY OUTCOMES. G Lee, V DelPizzo, M Yost, L Hristova, RR Neutra. California Dept of Health Services, 5900 Hollis St, Emeryville 94608, California USA.
A prospective pilot study was conducted as an exposure validation for a retrospective nested case- control study on pregnancy outcomes and EMF exposure in the San Francisco Bay Area. In this study, 220 women in the early stage of pregnancy (the mean time since last menstrual period [LMP] was about 10 weeks) were requested to wear an EMDEX meter for a period of at least 24 h, during which these women were also asked to keep an activity diary. Measurements were averaged over the following periods: total time of measurement; total time spent in the home; total time spent at home, but not in bed (to exclude the possible contribution of electric bed heating); total time spent in bed; total time spent at work, total time spent away from home but not at work. Beside the TWA, a standardized rate-of-change metric (RCM) was used to measure the consistency or "spikiness" of the field1. The magnetic field exposure so recorded was compared to that determined by wire coding and spot measurements about 20 weeks later. The latter data were used to validate exposure in the nested case-control study2, which was gathered retrospectively, at around 30 weeks of gestation. We noted that the subject availability for the retrospective measurements was somewhat lower than anticipated and apparently differential. When personal EMDEX measurements were regarded as the "gold standard," the predictive value positive of the retrospective measurements was low, although the prospective and retrospective TWA data were significantly and moderately strongly correlated3. These considerations prompted us to examine the relationship between prospectively assessed exposure (free from non-participation bias and presumably closer to the "true" exposure) and spontaneous abortion (SAB), although this was not the intended purpose of the prospective study. The incidence of spontaneous abortion was found to be higher among women with elevated TWA magnetic field. This possible association was weaker for exposure measured only during the bedtime hours or for the whole 24-h period, while it was strongest and statistically significant for exposure measured over the whole time the subject spent at the home as well as for the time spent away from both home and work. The sample did not include enough subjects to explore possible association due to occupational exposure. In the Bay Area, wire coding classification is only a weak surrogate of the TWA magnetic field and did not appear to be associated to increased SAB rates. No clear association existed between the RCM and SAB rates. Adjustment for confounding by known or suspected SAB risk factors did not substantially change these findings. These observations are based on a relatively small number of records and can only be regarded as hypothesis generating. The California EMF Program has funded a much larger replication study, expected to be completed in late 1999.
REFERENCES 1. Lee et al, Abstract J3, First World Congress for Electricity and Magnetism in Biology and Medicine, 1992 2. Lee et al, Case-Control study assessing personal magnetic field measures, powerline wire configuration and miscarriage, Bioelectromagnetic Society 20th Annual Meeting, 1998. 3. DelPizzo et al. Short Term Variation in Personal Exposure to Residential Magnetic Fields, The Annual Review of Research Effects of Electric and Magnetic Fields from the Generation, Delivery and Use of Electricity, November, 1995.
Top of Page
THE RELATIONSHIP BETWEEN RESIDENTIAL MAGNETIC FIELD MEASURES AND MISCARRIAGE G Lee, RR Neutra, L Hristova, M Yost, S Swan, V DelPizzo, California Department of Health Services and Public Health Institute, Berkeley, CA
A nested case-control study was done to assess the relationship between miscarriage and exposures to residential magnetic fields. An EMDEX meter was used to measure personal magnetic fields around 30 weeks after the participants' last menstrual period. Exposures were divided into those received at work, at home, at night while sleeping (bed) and elsewhere (other than the work and home environments). Each participants' residence was characterized into four wire-code categories: very high current configuration (VHCC, estimating high magnetic field exposure); ordinary high current configuration (OHCC); ordinary low current configuration (OLCC); and underground power-lines (Buried, estimating low magnetic field exposure) based on the types of and proximity of power-lines near the home. Analyses were done only for those women who did not move since their first trimester of pregnancy. Personal time-weighted average (TWA) magnetic fields comparisons were made for exposures greater than or less than 2.0 milligauss (mG) between cases (N= 155) and controls (N=507). The odd ratios and 95% confidence intervals for each of the exposure components were: 0.77, 0.34-1.74 for bed; 1.0, 0.48-2.07 for home; 0.58, 0.26-1.32; for total home (bed +home); 0.80, 0.42-1.54 for work, 0.88, 0.43-1.81 for other; and 1.03, 0.54-1.97 for the total 24-hour period. For wire-codes, each of the 3 above ground categories were compared to the buried category between cases and controls. The odd ratios and 95% confidence intervals were: 1.22, 0.71-2.15 for VHCC; 0.94, 0.58-1.52 for OHCC; and 0.96, 0.60-1.52 for OLCC. These results indicate that there is no association between late pregnancy (30 week gestation) personal TWA measurements, wire-code categories and miscarriage. The results did not change even after confounders were considered. We assumed that retrospective measures would represent prospective measures. This assumption may not be valid. We discuss this possibility in another paper presented at this meeting.
Top of Page
Unfortunately, the study itself does not support this conclusion. The study states that the results "provide little support of the hypothesis that living in homes with high time-weighted-average magnetic field levels or in homes close to electrical transmission lines is related to the risk of Childhood ALL (acute lymphocytic leukemia)." However, the authors conclude, "We cannot exclude the possibility of a small increase in risk among children in homes with very high magnetic fields, as suggested in studies using historical estimates of residential magnetic field exposure (the Scandinavian studies)."
Q: What were the study's results?
A: The authors found absolutely no association between ALL and living in close proximity to powerlines. They used a method known as "wire-coding" (developed by Wertheimer and Leeper for their 1982 study of EMF and adult cancers) to classify houses with respect to proximity to powerlines. However, it was not possible to wire-code all the houses in the study, and therefore the results are less precise than one would expect given the size of the study. In other words, the number of houses wire-coded was too small to rule out the 50% increase in risk of childhood leukemia presented in the National Academy of Sciences review.
On the other hand, the authors found a higher than expected cancer rate among children living in houses with uncommonly high time-weighted-average magnetic field levels. Specifically, it found that (1) children exposed to magnetic field levels of 2 mG or more had a cancer rate 24% higher compared to children living in homes with magnetic field levels less than 0.65 mG, and (2) children exposed to magnetic field levels of 3 mG or more had a cancer rate 72% higher compared to children living in homes with magnetic field levels less than 0.65 mG. The first results were not statistically significant; that is, it could not be ruled out by chance. The second results, however, were statistically significant. The authors had intended to use 2 mG as a cut-point for assessing their conclusions when they planned the study. The data in the article, however, indicate that taken as a group cancer patients were always exposed to magnetic fields higher than the control group.
Q: Was Linet's study better planned and executed than the others?
A: In general, it was a well-conducted study. There were some hard-to- overcome weaknesses, for example, the reliance on random telephone dialing to recruit controls (a method that excludes persons without a telephone or who are hard to reach by phone), and a less-than-perfect participation rate (78% of cancer cases and 63% of eligible controls).
The study used significantly improved techniques. In this study, houses were measured within two years of diagnosis. In others, this type of measurement was taken several or many years later. Since residential fields may change significantly over time, a long delay may make it difficult to detect an association.
Q: Was the Linet study very large?
A: The study was undoubtedly larger than previous studies. However, in order to be able to show an association reliably, a study must also include a sufficient number of subjects exposed to the agent being studied. Since high magnetic fields in residences are uncommon, even a study as large as this may have been "too small." To be 90% sure that a study would not miss a 50% increase in cancer risk in children exposed to more than 2 mG, one would have to measure magnetic fields in the homes of 1075 cases and as many controls. Similarly, to detect the same risk in children living in very high wirecode homes, one would have to code 1720 cancer homes and as many controls. Therefore, to be able to dismiss the hypothesis, the Linet study should have been two to four times larger than it was. Large studies like this one encounter new difficulties such as different wiring practices across regions in the US. The NAS report noted that "wire codes might not be useful in multi city epidemiological studies."
Q: How can this hypothesis be confirmed or rejected?
A: The most promising approach appears to be a systematic evaluation of all research results, epidemiological and experimental. For epidemiological studies this can be done by meta-analysis, that is, pooling data from different studies. A meta-analysis conducted for the NAS report has determined that a statistically significant association between wirecodes (i.e., proximity to powerlines) and childhood leukemia does exist. However, it found no association with measured fields. Preliminary inclusion of the Linet data in this meta-analysis does not significantly change these findings.
It should be noted that two German studies, two Canadian studies and one British study will be reported in the next few years, all dealing with childhood leukemia. These should be added to the meta-analysis. Several epidemiological studies of powerlines, magnetic fields and female breast cancer and miscarriage are also underway, as well as a variety of experimental studies from the federal research program. The California Department of Health Services' Electrical and Magnetic Fields Program intends to consider all this evidence in its final risk evaluation when most of this information will be available in 1999.
1515 Clay Street, Suite 1700 Oakland, CA 94612