Extremely low-frequency magnetic fields are designated as possibly carcinogenic in humans based on an epidemiologic association with childhood leukemia, though no biologic mechanism has been identified.1 In a national case-control study of childhood cancers in England and Wales, both a significant trend (P < 0.01) in risk of leukemia with distance from power lines and a significant dose-response relationship comparing risk of leukemia for those living <200 m and 200–600 m from a power line with those living >600 m were observed.2 These associations were, however, not fully explained by exposure to magnetic fields because no statistically significant excess risk was observed with increasing magnetic fields.3
Some reports have also implicated exposures to extremely low-frequency magnetic fields in the incidence of adult cancers, including leukemia,1,4–6 brain and central nervous system cancers,7,8 malignant melanoma,9 and female breast cancer.10 However, evidence for an association between magnetic fields and cancer risks in adults is inconsistent and weaker than for childhood leukemia.5 Several studies in adults have included only small number of cases and based their exposure estimates solely on distance from power line, which is a relatively poor proxy for magnetic field strength.11 To address these shortcomings, we carried out a large national case-control study of cancer incidence among adults living near high-voltage overhead power lines in England and Wales. We included the estimation of magnetic fields and distance from power lines. We focused on the high-voltage power lines where residential exposures to magnetic fields from the power lines tend to dominate over other sources.5,12
We chose a case-control design over other approaches such as Poisson regression of small area data because it enabled analysis of individual-level information, including residential addresses of cases and controls, which is available within the National Cancer Registry to a ~0.1 m accuracy. This high level of geographic resolution is important because magnetic fields fall off rapidly with distance from a power line. Furthermore, population data for England and Wales are collected from census only every 10 years, leading to possible inaccuracies in denominator information in the intercensual years.13 A similar case-control design was used for the national study of childhood cancers and power lines across England and Wales.2,3
We identified eligible cases and controls from the National Cancer Registry held by the Small Area Health Statistics Unit at Imperial College London.14 Case cancers in England and Wales were 15–74 years of age with a diagnosis of leukemia, brain or central nervous system cancers, malignant melanoma, or female breast cancer diagnosed 1974–2008, and living within 1000 m of a high-voltage overhead power line. Diagnostic codes of the case cancers according to the eighth, ninth, and tenth revisions of the International Classification of Disease, together with controls selected from a range of cancers not considered to be associated with electric and magnetic fields, are given in eTable 1 (http://links.lww.com/EDE/A641). Exclusions from the controls were malignant neoplasms of lymphatic and hematopoietic tissues other than leukemias; nonmelanoma skin cancer; cancers of the lip, oral cavity, pharynx, and respiratory cancers, including larynx and lung (because it has been hypothesized that electric fields emitted by high-voltage power lines, and the charged ions they produce, might be associated with these cancers)15; and cancers of ill-defined, secondary, and unspecified sites.
We obtained grid references and construction dates (by year) of the ~21,800 pylons for all the high-voltage (400 and 275 kV) overhead power lines in England and Wales from National Grid, together with the few 132 kV lines (0.1% of the total at this voltage, by length) that form part of the National Grid rather than regional distribution networks. We constructed zones delineating 1000 m of the power lines for each year 1969–2008 and linked these to the cancer database addresses within a Geographical Information System.
We located residential addresses (0·1 m accuracy) of case and control cancers recorded at the time of diagnosis using ADDRESS-POINT16; depending on the year, we successfully located 89% to 96% of the cancer database addresses. On the basis of diagnosis address, we first calculated distances from all operating power lines for the year before diagnosis. This was done to ensure that cases and controls were living near an operational power line on the date of diagnosis because information on the power lines network was available only annually and new power lines could be added to the network at any point during a year. We included all eligible cases of leukemia, brain/central nervous system cancers, and malignant melanoma within 1000 m of a power line. For female breast cancer, we selected a 50% random sample because the large numbers precluded calculation of magnetic fields for the complete set of cases, though we used both the 50% sample and all eligible cases for distance-based analyses. We identified a pool of 164,095 possible control cancers within 1000 m of a power line in the year before diagnosis. We randomly selected control cancers from the pool, with replacement, for the various case cancers; of the total of 79,507 control cancers thus selected, 18,767 (23.6%) were used for more than one case cancer type (see eTable 1, http://links.lww.com/EDE/A641). We included one control per case for female breast cancer and three per case for the other cancers, frequency-matched to cases by year of diagnosis and region (there are 10 regions in England and Wales; regions are administrative areas, the highest tier in the subnational division used by central Government). We then obtained distances of diagnosis address from operating power lines for year of diagnosis and 5 years before diagnosis for the analyses.
National Grid provided estimates (in 2011) of magnetic fields (blind of case/control status) using their EM2D program,12 which was also used to estimate magnetic fields in the national study of overhead power lines and childhood cancers.3 Magnetic field estimates were provided for year of diagnosis and for 5 years before diagnosis, for all diagnosis addresses (cases/controls) lying within 200 m of an operating high-voltage power line, except those with transposed phasing for which magnetic fields within 100 m of a line were provided.12 Magnetic fields for people living further from a power line were assumed to be <100 nanotesla (nT) in categorical analyses and zero in analyses of magnetic fields considered as a continuous variable. To maintain confidentiality, the list of addresses provided to National Grid was supplemented by an additional 10% addresses within 200 m.
EM2D takes into account the distance from the power line, tower design, clearance above ground, phase arrangement, and historical load data—estimated for the winter peak and scaled to give an annual average.12 For consistency, modeled estimates of magnetic field exposures were obtained at an elevation of 1 m above ground level.12 On the basis of the configuration of power lines for the year of diagnosis, there were 9586 case and control cancer locations within 200 m of a power line; a directly calculated magnetic field estimate of ≥100 nT was obtained using EM2D for 1845 (19.2%) of these, with an estimate of <100 nT (zero in continuous analyses) assigned for 7,576 (79.0%). Data were insufficient to provide an estimate for the remaining 165 (1.7%) case/control addresses (eg, because there were no available load data or the load data were illegible for the relevant power line); this number was slightly higher (254, 2.7%) for the power lines configuration 5 years before diagnosis. Data were insufficient to provide a magnetic field estimate for similar proportions of cases and controls for both year of diagnosis (1.9%  cases, 1.6%  controls) and 5 years before diagnosis (2.6% , 2.7% ).
Information on Potential Confounders
In addition to stratifying by year of diagnosis/region, we included age, sex (except female breast cancer), and small area (enumeration district) measures of rurality and deprivation (because individual-level information on deprivation and smoking are unavailable in the National Cancer Registry) as potential confounders. We used CORINE 1990/2000 land-cover data (codes 1–11, urban; codes >11, rural)17 as a measure of rurality and Carstairs score to denote deprivation18 because this was available for all 35 years of the study. We used Carstairs 1981 and 1991 at the enumeration district level for cancers diagnosed 1974–1985 and 1986–1995, respectively, and thereafter Carstairs 2001, based on census output areas—on average, there are approximately 12,000 enumeration districts or output areas per region. Negative Carstairs scores indicate areas that are more affluent and positive scores, more deprived, than the average across England and Wales.18
Information on migratory patterns and social class of individuals is unavailable in the National Cancer Registry. We examined migration patterns for women, with breast cancer/control cancers, who lived within 1000 m of a power lines and were included in the Office for National Statistics’ Longitudinal Study—a 1% sample of the UK population with data on individuals linked across decennial censuses.19 We also examined individual measures of social class by distance. We used data for 1991 (n = 14,044 women 15–74 years of age) and 2001 (n = 14,451) because residential postcodes are unavailable in the Longitudinal Study for earlier periods.
We used logistic regression to estimate cancer risks by categories of distance and magnetic field. Category breaks were based on those used in the national childhood study,2,3 with an additional magnetic field break between 200 and 400 nT and for the highest exposures (≥1000 nT); this was possible in our study because there are many more cancers in adults than in children. Models were adjusted for age, sex (except female breast cancer), deprivation, and rurality. Because deprivation and rurality are area-level (enumeration district) measures, this leads to a multilevel data structure; we acknowledge the possibility of within-area dependence by including area-level random effects in our models. P values for tests for linear trend across categories were based on median distance/magnetic field in each category. Additional analyses treated distance and magnetic field as continuous measures. For distance, we used (1) inverse rank and (2) reciprocal of distance. For magnetic fields, we tested for nonlinearity by including a quadratic term. We carried out a sensitivity analysis and reran all the models excluding the rarest cancers from the control pool. We excluded the rarest 5% (3,870) of control cancers comprising cancers of other respiratory organs, pituitary, adrenal and other endocrine glands, male genital, other digestive, eye, bone, gallbladder, small intestine, heart, and thyroid. All analyses were carried out in the statistical package R.20 The study received ethical approval from the London MREC Committee.
Within 1000 m of a power line, 7823 leukemia cases, 6781 brain/central nervous system cancers, 9153 malignant melanoma cases, and 58,404 breast cancer cases were included (Table). Case cancers tended to be younger, slightly more affluent (based on Carstairs score) and, for malignant melanoma and breast cancer, less urban than control cancers. There was a higher proportion of women among malignant melanoma cases than among controls. Mean magnetic field estimates among the controls ranged from 7.4 to 10.6 nT (Table), but much higher values, ranging up to 8592 nT, were calculated very close to a power line (Fig. 1). Cases of female breast cancer lived on average closer to a power line than controls (Table).
Results by distance for year of diagnosis are shown in Figure 2 and eTable 2 (http://links.lww.com/EDE/A641). With adjustment for confounders, odds ratios (ORs) for leukemia, brain/central nervous system cancer, and malignant melanoma closest to the power lines (0–49 m) compared with distances 600–1000 m ranged from 0.82 (95% confidence interval = 0.61–1.11, 66 cases) for malignant melanoma to 1.22 (0.88–1.69) for brain/central nervous system cancer. For female breast cancer, after adjustment for confounders, ORs at 0–49 m compared with 600–100 m were 1.01 (0.85–1.21) and 1.07 (0.93–1.24) for the 50% and 100% sample, respectively (Fig. 2 and eTable 2, http://links.lww.com/EDE/A641).
Results of the magnetic field analyses for year of diagnosis are shown in Figure 3 and eTable 3 (http://links.lww.com/EDE/A641). In comparison with the reference group (<100 nT), we observed no meaningful excess risks or trends of risk with magnetic field strength for any of the four cancers examined. In adjusted analyses at the highest estimated field strength (≥1000 nT) compared with <100 nT, the OR ranged from 0.68 (0.39–1.17) for malignant melanoma to 1.08 (0.77–1.51) for female breast cancer. Quadratic terms included in the regression analyses did not improve the fit of the model.
Pearson correlation coefficients between distance/magnetic field from power lines in year of diagnosis and 5 years before diagnosis were r = 1.00 and r = 0.85, respectively. Trends in risk by distance and magnetic field at diagnosis address 5years before diagnosis were similar to those for year of diagnosis (see eTables 2 and 3, http://links.lww.com/EDE/A641). Excluding rare cancers from the control group as a sensitivity analysis had minimal effect on the results for year of diagnosis and 5 years before diagnosis (data not shown). We also examined data for female breast cancer cases younger than and older than 50 years of age. With adjustment for confounders, there were no convincing trends in risk at either age, and there were no differences in trends between the two age groups (data not shown).
In the Longitudinal Study, 68% of women moved into or out of areas within 100 m of power lines from 1991 to 2001. There were no meaningful differences in migration patterns between those with breast cancer and other cancers, and no difference in social class (based on individual occupation) for women living ≤100 m and from >100 to ≤1000 m of a power line (P = 0·66; see eTable 4, http://links.lww.com/EDE/A641).
This national study, with 35 years of observation, is by far the largest to date to investigate the risk of adult cancers near high-voltage overhead power lines.4,6,10,21,22 Because of the size of the population and the focus on the high-voltage overhead power lines (400 and 275 kV), we included scores of people with estimated exposures ≥1000 nT, substantially increasing the exposure range above that reported in previous studies. At these levels, overhead power lines are the dominant source of exposure to magnetic fields.12 After adjustment for deprivation and other confounders, we found no excess risks or trends for leukemia, brain/central nervous system cancers, malignant melanoma, or female breast cancer in relation to distance or magnetic fields from high-voltage overhead power lines in England and Wales.
In addition to its size, important strengths of our study include the use of a geographical approach to maximize exposure variation23 and estimation of magnetic field exposures at the home address of individual cases and controls using a validated modeling procedure,12 carried out blind to case/control status; we thus identified homes near power lines with high and, in some cases, extremely high, magnetic field exposures. The use of magnetic field estimates in our study is a major strength, given that distance from power lines has been shown to be a relatively poor proxy of residential magnetic field exposure—indeed, studies using distance alone have been considered “uninterpretable.”11 Because magnetic field exposures fall off rapidly with distance, a particular strength was the availability in the cancer registry of individual address locations with high accuracy (~0.1 m), which we used in the estimation of magnetic fields. This provided an important advantage of our case-control design over a small area approach. Because areas near power lines are more affluent than the average for England and Wales, and the four cancers included here are all associated with higher social class,24–27 we needed to pay specific attention to possible confounding by deprivation. This affected, in particular, the distance-based analyses; adjustment for deprivation and other confounders resulted in lower ORs for these cancers with distance from a power line. Adjustment for deprivation should also address possible local variations in the uptake of screening programs for breast cancer (among cases) and cervical and colon cancers (among controls) because evidence to date suggests variation in uptake largely reflecting socioeconomic status.28–30
Our study has limitations. In the absence of population address data, we used as controls a range of cancers not considered to be associated with electric and magnetic fields, as was done in a Taiwanese study.4 The use of cancer controls has the advantage that, as with the case cancers, ascertainment levels are high,31 they are known to have been living near a power line at the time of diagnosis, and magnetic field exposures can be estimated to the same degree of accuracy as for cases; however, the use of cancer controls might introduce bias (toward the null) if magnetic fields are positively associated with the control cancers. We had no direct information on the migration of cases or controls, and so were unable to allow for cumulative exposures or latency, except for providing estimates at diagnosis address 5 years before diagnosis, which were highly correlated with estimates for year of diagnosis. Data from the Longitudinal Study indicate that over a 10-year period, there was substantial in- and out-migration among the population living within 100 m of power lines. To the extent that migration affected case and control cancers similarly, this would tend to dilute any risk, if exposed people moved out of the area and unexposed moved in. Furthermore, calculation of magnetic fields was based on historical load data with inherent inaccuracies12; these were unlikely to be differential with respect to case or control location, and so this may have led to underestimation of any risk. We were only able to take into account the magnetic fields from overhead power lines and had no measurements of actual exposures within the home. However, long-term average background fields in homes32 are at least an order of magnitude below our highest category of magnetic field (≥1000 nT).
Comparison With Other Studies
The International Agency for Research on Cancer (IARC) designated extremely low-frequency magnetic fields as possibly carcinogenic in humans, based on an epidemiologic association with childhood leukemia.1 Results of the national study of childhood cancers and overhead power lines in England and Wales were consistent with the IARC evaluation, showing a statistically significant dose-response association between childhood leukemia and proximity of home address at birth to high-voltage power lines: relative risk = 1.69 (95% confidence interval = 1.13–2.53) and 1.23 (1.02–1.49) for those living <200 m and 200–600 m from a power line compared with those who lived >600 m from a line, respectively.2 In the same national study, the risk of childhood leukemia was 2.00 (0.18–22.04) in the highest field category (≥400 nT) compared with the lowest (<100 nT) and 1.14 (0.57–2.32) with each 200 nT higher magnetic field. No statistically significant associations were thus observed with magnetic field exposures. However, annual average magnetic fields were fairly low in the study; only 2.5% of cases and controls lived at addresses where the field was estimated to be ≥100 nT giving wide confidence intervals around the risk estimates.3 The study authors concluded that exposure to magnetic fields in year of birth was unlikely to fully explain the observed association between childhood leukemia and distance from power lines.3
The national childhood cancer study used a case-control approach similar to that adopted here for adults, with use of the National Grid EM2D program to estimate magnetic fields based on power line characteristics and residential addresses of cases and controls. Evidence in adults, including for leukemia and brain cancer, was considered by the IARC review to be inconsistent and weaker than for childhood leukemia, whereas evidence for breast cancer, in particular, was inadequate to make causal judgments.1 Subsequently, the World Health Organization5 concluded that the evidence did not support an epidemiologic association between extremely low-frequency magnetic fields and female breast cancer, and a recent meta-analysis reached a similar conclusion.33 National, population-based studies in Finland,22 Sweden,21 and Norway10,34 reported conflicting results for risks of adult leukemia and breast cancer associated with residential exposures to magnetic fields from high-voltage power lines, though compared with our study, numbers were small, particularly for those with the highest exposures (categorized as >200 nT in those studies).
In summary, our results do not support an epidemiologic association of adult cancers with proximity to residential magnetic fields from high-voltage overhead power lines. Unless new biologic hypotheses emerge, our findings should help to settle a long-standing debate on the safety of residential exposures to extremely low-frequency magnetic fields from high-voltage overhead power lines and adult cancers.
We thank Sara Morris, Norman Cobley, Catherine Keshishian, Nina Iszatt, and Peter Hambly (Imperial College London) for their help with data acquisition and analysis early on in the study. We are grateful to John Swanson (National Grid) who provided data on the locations of power lines and carried out the magnetic fields analyses in EM2D, blinded to case-control status. We thank the cancer registries in England and Wales and the Office for National Statistics for providing data on cancer incidence. The permission of the Office for National Statistics to use the Longitudinal Study is gratefully acknowledged (clearance no. 30104), as is the help provided by staff of the Centre for Longitudinal Study Information and User Support (CeLSIUS). CeLSIUS is supported by the ESRC Census of Population Programme (Award Ref: RES-348-25-0004). Census output is Crown copyright and is reproduced with the permission of the Controller of HMSO and the Queen’s Printer for Scotland. The Small Area Health Statistics Unit is funded by the Health Protection Agency as part of the MRC-HPA Centre for Environment and Health at Imperial College London. P.E. acknowledges support from the National Institute for Health Research (NIHR) Biomedical Research Centre based at Imperial College Healthcare NHS Trust and Imperial College London. P.E. is a NIHR Senior Investigator.
1. IARC Working Group on the Evaluation of Carcinogenic Risks to Humans. Non-ionizing Radiation, Part 1: Static and Extremely Low-frequency (ELF) Electric and MAGNETIC FIELDS. 2002 Lyon IARC (Monographs on the Evaluation of Carcinogenic Risks to Humans, 80)
2. Draper G, Vincent T, Kroll ME, Swanson J. Childhood cancer in relation to distance from high voltage power lines in England and Wales: a case-control study. BMJ. 2005;330:1290
3. Kroll ME, Swanson J, Vincent TJ, Draper GJ. Childhood cancer and magnetic fields from high-voltage power lines in England and Wales: a case-control study. Br J Cancer. 2010;103:1122–1127
4. Li CY, Thériault G, Lin RS. Residential exposure to 60-Hertz magnetic fields and adult cancers in Taiwan. Epidemiology. 1997;8:25–30
5. World Health Organization. Extremely Low Frequency Fields. Environmental Health Criteria. 2007;Vol. 238 Geneva, Switzerland World Health Organization
6. Feychting M, Forssén U, Floderus B. Occupational and residential magnetic field exposure and leukemia and central nervous system tumors. Epidemiology. 1997;8:384–389
7. Savitz DA, Loomis DP. Magnetic field exposure in relation to leukemia and brain cancer mortality among electric utility workers. Am J Epidemiol. 1995;141:123–134
8. Wertheimer N, Leeper E. Magnetic field exposure related to cancer subtypes. Ann N Y Acad Sci. 1987;502:43–54
9. Floderus B, Stenlund C, Persson T. Occupational magnetic field exposure and site-specific cancer incidence: a Swedish cohort study. Cancer Causes Control. 1999;10:323–332
10. Kliukiene J, Tynes T, Andersen A. Residential and occupational exposures to 50-Hz magnetic fields and breast cancer in women: a population-based study. Am J Epidemiol. 2004;159:852–861
11. Maslanyj M, Simpson J, Roman E, Schuz J. Childhood leukaemia: misclassification of exposure from the use of the ‘distance from power line’ exposure surrogate. Bioelectromagnetics. 2009;30:183–188
12. Swanson J. Methods used to calculate exposures in two epidemiological studies of power lines in the UK. J Radiol Prot. 2008;28:45–59
13. Arnold RA, Diamond ID, Wakefield JCElliott P, Wakefield JC, Best NG, Briggs DJ. The use of population data in spatial epidemiology. In: Spatial Epidemiology: Methods and Applications. 2000 Oxford OUP:30–50
14. Elliott P, Westlake AJ, Hills M, et al. The Small Area Health Statistics Unit: a national facility for investigating health around point sources of environmental pollution in the United Kingdom. J Epidemiol Community Health. 1992;46:345–349
15. Henshaw DL, Ross AN, Fews AP, Preece AW. Enhanced deposition of radon daughter nuclei in the vicinity of power frequency electromagnetic fields. Int J Radiat Biol. 1996;69:25–38
17. European Environment Agency. CORINE Land Cover 2000 (CLC2000) 100 m - Version 8/2005 Version 2.. 2005 Copenhagen, Denmark European Environment Agency
18. Carstairs V, Morris R. Deprivation: explaining differences in mortality between Scotland and England and Wales. BMJ. 1989;299:886–889
19. Office for National Statistics. . What is the ONS Longitudinal Study? Office for National Statistics Longitudinal Survey. 2010 Available at: http://www.celsius.lshtm.ac.uk/what.html
. Accessed 6 January 2011
20. R Development Core Team.. R: A language and environment for statistical computing. 2009 Available at: http://www.r-project.org
. Accessed 6 January 2011
21. Feychting M, Forssén U, Rutqvist LE, Ahlbom A. Magnetic fields and breast cancer in Swedish adults residing near high-voltage power lines. Epidemiology. 1998;9:392–397
22. Verkasalo PK, Pukkala E, Kaprio J, Heikkilä KV, Koskenvuo M. Magnetic fields of high voltage power lines and risk of cancer in Finnish adults: nationwide cohort study. BMJ. 1996;313:1047–1051
23. Elliott P, Savitz DA. Design issues in small-area studies of environment and health. Environ Health Perspect. 2008;116:1098–1104
24. Shack L, Jordan C, Thomson CS, Mak V, Møller HUK Association of Cancer Registries. . Variation in incidence of breast, lung and cervical cancer and malignant melanoma of skin by socioeconomic group in England. BMC Cancer. 2008;8:271
25. Klassen AC, Smith KC. The enduring and evolving relationship between social class and breast cancer burden: a review of the literature. Cancer Epidemiol. 2011;35:217–234
26. Faggiano F, Partanen T, Kogevinas M, Boffetta P. Socioeconomic differences in cancer incidence and mortality. IARC Sci Publ. 1997;138:65–176
27. Kroll MEC, Stiller A, Murphy MFG, Carpenter LM. Childhood leukaemia and socioeconomic status in England and Wales 1976–2005: evidence of higher incidence in relatively affluent communities persists over time. Br J Cancer. 2011;105:1783–1787
28. Baker D, Middleton E. Cervical screening and health inequality in England in the 1990s. J Epidemiol Community Health. 2003;57:417–423
29. Maheswaran R, Pearson T, Jordan H, Black D. Socioeconomic deprivation, travel distance, location of service, and uptake of breast cancer screening in North Derbyshire, UK. J Epidemiol Community Health. 2006;60:208–212
30. Henley NC, Hole DJ, Kesson E, Burns HJ, George WD, Cooke TG. Does deprivation affect breast cancer management? Br J Cancer. 2005;92:631–633
31. Møller H, Richards S, Hanchett N, et al. Completeness of case ascertainment and survival time error in English cancer registries: impact on 1-year survival estimates. Br J Cancer. 2011;105:170–176
32. Swanson J, Kaune WT. Comparison of residential power-frequency magnetic fields away from appliances in different countries. Bioelectromagnetics. 1999;20:244–254
33. Chen C, Ma X, Zhong M, Yu Z. Extremely low-frequency electromagnetic fields exposure and female breast cancer risk: a meta-analysis based on 24,338 cases and 60,628 controls. Breast Cancer Res Treat. 2010;123:569–576
34. Tynes T, Haldorsen T. Residential and occupational exposure to 50 Hz magnetic fields and hematological cancers in Norway. Cancer Causes Control. 2003;14:715–720
Supplemental Digital Content
Copyright © 2013 Wolters Kluwer Health, Inc. All rights reserved.