Residential Distance to High-voltage Power Lines and Risk of Neurodegenerative Diseases: a Danish Population-based Case-Control Study

Transcription

1 American Journal of Epidemiology The Author Published by Oxford University Press on behalf of the Johns Hopkins Bloomberg School of Public Health. All rights reserved. For permissions, please Vol. 177, No. 9 DOI: /aje/kws334 Advance Access publication: April 9, 2013 Original Contribution Residential Distance to High-voltage Power Lines and Risk of Neurodegenerative Diseases: a Danish Population-based Case-Control Study Patrizia Frei, Aslak Harbo Poulsen*, Gabor Mezei, Camilla Pedersen, Lise Cronberg Salem, Christoffer Johansen, Martin Röösli, and Joachim Schüz * Correspondence to Aslak Harbo Poulsen, Danish Cancer Society Research Center, Danish Cancer Society, Strandboulevarden, 2100 Copenhagen Ø, Denmark ( aslak@cancer.dk). Initially submitted March 22, 2012; accepted for publication July 23, The aim of this study was to investigate the possible association between residential distance to high-voltage power lines and neurodegenerative diseases, especially Alzheimer s disease. A Swiss study previously found increased risk of Alzheimer s disease for people living within 50 m of a power line. A register-based case-control study including all patients diagnosed with neurodegenerative diseases during the years was conducted among the entire adult population of Denmark. Using conditional logistic regression models, hazard ratios for ever living close to a power line in the time period 5 20 years before diagnosis were computed. The risks for developing dementia, Parkinson s disease, multiple sclerosis, and motor neuron disease were not increased in persons living within close vicinity of a power line. The risk of Alzheimer s disease was not increased for ever living within 50 m of a power line (hazard ratio = 1.04, 95 confidence interval: 0.69, 1.56). No dose-response according to number of years of living within 50 m of a power line was observed, but there were weak indications of an increased risk for persons diagnosed by the age of 75 years. Overall, there was little support for an association between neurodegenerative disease and living close to power lines. dementia; environmental exposure; magnetic fields; neurodegenerative diseases Abbreviations: CI, confidence interval; CPR, Central Population Register; ELF-MF, extremely low frequency magnetic field(s); HR, hazard ratio; ICD-10, International Classification of Diseases, Tenth Revision. Transmission of electric power is a prerequisite of modern life and gives rise to extremely low frequency (ELF) electric and magnetic fields (MF) in the vicinity of power lines. ELF-MF have been classified as possibly carcinogenic by the International Agency for Research on Cancer because of limited evidence in humans in relation to childhood leukemia and inadequate evidence in experimental animal studies, as well as a lack of a known biological mechanism (1, 2). For noncancer endpoints, a 3- to 4-fold risk increase for Alzheimer s disease related to occupational exposure to ELF-MF was suggested in an initial report by Sobel et al. (3). Subsequent studies summarized in reviews and metaanalyses on occupational ELF-MF exposure (4 6) found less consistent results, but indications of an increased risk of Alzheimer s disease remained, with reported risks on the order of for exposure levels of 0.1 µt to 0.5 µt (4). A Swedish study suggested that the increased risk might be restricted to cases with disease onset by age 75 years (7), which was confirmed in 2 subsequent studies (8, 9), and the risk was found to be higher in men than in women (8). For other neurodegenerative diseases, amyotrophic lateral sclerosis, the most frequent motor neuron disease, has been quite consistently linked with electrical occupations (5, 6). Even if an association were found in workers, extrapolation to the general public would remain difficult for example, due to occupational coexposure to other potential risk factors, particularly electric shocks (6). Recently, the first study analyzing a possible association between residential ELF-MF exposure and neurodegenerative diseases was conducted in Switzerland (10). An increased risk of Alzheimer s disease was observed for people living within 50 m of a high-voltage power line compared with people living at a distance of 600 m, and risks were highest for people who had lived within 50 m of a 970

2 Power Lines and Neurodegenerative Diseases 971 power line for at least 15 years (hazard ratio (HR) = 2.00, 95 confidence interval (CI): 1.21, 3.33). However, the study had to rely on information from death certificates, with likely underreporting of neurodegenerative diseases. Furthermore, it was not possible to take into account full residential histories, due to reliance on census data that were collected only every 10 years. Because of the intriguing Swiss findings for Alzheimer s disease, independent replication became a key research priority, as recommended by the World Health Organization (2) and the Scientific Committee on Emerging and Newly Identified Health Risks (11). Our aim in this study was to investigate the association between residential distance to power lines and risk of neurodegenerative diseases, particularly Alzheimer s disease, in Denmark, using data that allowed an improved study design in comparison with the original Swiss study. MATERIALS AND METHODS We conducted a register-based case-control study among the entire adult population of Denmark (approximately 5.5 million inhabitants). Since 1968, the Central Population Register (CPR) has assigned each Danish resident a personal identification number (CPR number) at birth or on the date of immigration, making it possible to track every resident in and across all Danish registers. Identification of cases and controls and controls were selected among all adult (aged 20 years) residents of Denmark (excluding Greenland and the Faroe Islands) with a valid CPR number. were identified from the Danish Hospital Discharge Registry, which was established by the Danish National Board of Health in 1977 and in which more than 99 of all hospitalizations for somatic diseases are registered (12), including outpatients from 1994 onwards. The dates of admission and discharge and up to 20 diagnoses per discharge, according to a modified Danish version of the International Classification of Diseases, Tenth Revision (ICD-10) (from 1994 onwards), are available for each patient. All persons hospitalized for the first time with a neurodegenerative disease from January 1, 1994, to December 31, 2010, were defined as cases. The following groups of neurodegenerative diseases were defined: Alzheimer s disease (ICD-10 codes F00 and G30), vascular dementia (ICD-10 code F01), other dementia (ICD-10 codes F02, F03, A81.0, B22.0, G10, and G31), Parkinson s disease (ICD-10 code G20), multiple sclerosis (ICD-10 code G35), and motor neuron disease (ICD-10 code G12). Any person with 2 or more of the above diagnoses was considered a case for all registered diagnoses. with dementia (Alzheimer s disease, vascular dementia, and other dementia) were included only if their first diagnosis was made at the age of 65 years, because of differences in etiology between very-early-onset ( 65 years) dementia cases and later-onset dementia cases and because dementia diagnoses in the registers have been validated primarily in the elderly population (ages 65 years) so far (13). For each case, 6 controls were randomly selected from the CPR, individually matched by gender and date of birth (±30 days). The controls had to be free of the disease and alive at the date of diagnosis of the matched case. Information on marital and vital status for cases and controls was extracted from the CPR. Residential history and geocoding of addresses The residential history of all cases and controls from 1968 onwards was obtained from the CPR. The data set included addresses and all dates of moving in and out. Because of poor address quality before 1974 (mainly due to missing house numbers), we chose to consider address history only up to 20 years before diagnosis. Information on geographical coordinates, placed within the house or, where applicable, the apartment (estimated deviation not more than 1 2 m), and number of floors in the building was obtained from the Danish Building and Dwelling Register, which is maintained by the Danish enterprise and construction authority for 95.2 of the 2,044,842 relevant addresses. Persons with a continuous geocoded address history dating back to 20 years before diagnosis were included in the analysis (94.2 of eligible cases and 89.1 of eligible controls). Socioeconomic variables On the basis of a m grid cell net of Denmark maintained by Statistics Denmark, we obtained information on socioeconomic indicators for all addresses. Income was provided as average disposable income per household in the grid cell, and information on education was based on the person with the highest level of education in the household. Location of power lines For all current and historical overhead power lines operated with alternating current at a voltage level of 132 kv, we obtained the geographical location, the date of entering service and, where appropriate, the date of termination of operation from all 7 Danish transmission companies responsible for these voltage levels. The majority of the power lines were digitalized from orthophotos, and the accuracy of these is estimated to be 3 5 m. For some historical lines, the geographical coordinates were identified through land surveying. The locations of the power lines were validated against current maps of power lines in Denmark and, for a sample of lines, historical maps of power lines. The mapped grid totaled 4,336 km of current and historical power lines ( kv, kv, kv, and kv). For lines for which only the year of entering and exiting service was available (77.5), we set the date of entering service to December 31 and the date of termination to January 1 to avoid erroneously classifying people as exposed. For each address, we determined the shortest distance to any of the power lines. Geographical data were processed in ArcGIS 9.3 (ESRI, Redlands, California). Statistical analysis Using conditional logistic regression, we calculated hazard ratios and associated 95 confidence intervals for

3 972 Frei et al. Table 1. Characteristics of (at Diagnosis) With Neurodegenerative Disease and in a Study of Residential Distance to Highvoltage Power Lines and Risk of Neurodegenerative Diseases, Denmark, persons who had, during the time period of 5 20 years before diagnosis, ever lived within a certain distance (in categories: <50 m, 50 <200 m, or 200 <600 m) of a power line, compared with those who never did (considering the closest category for each person). We estimated crude hazard ratios as well as hazard ratios adjusted for relevant confounders, which were collected for the address at which the participant lived at diagnosis (listed in Table 1). To investigate a potential dose-response relationship, we further categorized exposure according to time of residency within 50 m of a power line. For better comparability with the Swiss study (10), we also performed all analyses for only 220- to 400-kV lines using the same reference category as for the main analyses ( 600 m from all 132-kV lines). We also repeated the analyses taking into account exposure during the whole time period of 20 years before diagnosis. For Alzheimer s disease, we performed secondary analyses by age at diagnosis (65 75 years vs. >75 years) and by Diagnosis Alzheimer s Disease Vascular Dementia Other Dementia gender. Because of remarkable changes in clinical dementia practice in the past 2 decades owing to major advances in the understanding of the pathophysiology of the disease and in diagnostics and treatment (14), we investigated whether the risk was different for more recently diagnosed Alzheimer s disease cases. The validity of registered dementia diagnoses in 2003 was shown to be high (14) and thereby suitable for register-based epidemiologic studies about dementia. On the basis of this knowledge, we used the year 2003 as a cutoff. Data were analyzed using SAS 9.2 (SAS Institute Inc., Cary, North Carolina). RESULTS The numbers and characteristics of cases and controls at the date of diagnosis are shown by disease in Table 1. More cases than controls were married at the time of diagnosis. More cases with dementia or Parkinson diagnoses tended to Total no. 20, ,217 10,207 55,969 68, ,897 Female gender 13, , , , , , Mean age, years (SD) 81.0 (6.5) 81.1 (6.5) 80.6 (6.7) 80.7 (6.7) 82.5 (7.0) 82.6 (6.9) Marital status a Missing data b , , , , Never married 4, , , , , , Ever married c 9, , , , , , Married at diagnosis 5, , , , , , Disposable income First quartile 6, , , , , , Second and third quartiles (plus missing data) d 9, , , , , , Fourth quartile 4, , , , , , in highest educational level e <23.5 6, , , , , , <36.0 (plus 6, , , , , , missing data) f , , , , , , floors in residential building 1 floor (plus missing 12, , , , , , data) g >1 floor 8, , , , , , Urbanity of residence Urban 17, , , , , , Rural 3, , , , , , Table continues

4 Power Lines and Neurodegenerative Diseases 973 Table 1. Continued be in the lowest income group compared with their controls, which was not observed for multiple sclerosis and motor neuron disease. The distribution of educational levels was similar for cases and controls, except for Alzheimer s disease, where the proportion of cases in the highest education group was higher than for the controls. more often lived in multistory apartment buildings and in urban environments. The risks of developing a neurodegenerative disease according to distance from a 132- to 400-kV power line are shown in Table 2. Overall, risks were not elevated for people living closest (<50 m) to a power line. The risk of Diagnosis Parkinson s Disease Multiple Sclerosis Motor Neuron Disease Alzheimer s disease for people living within 50 m of a power line was 1.04 (95 CI: 0.69, 1.56). Results of analyses considering only 220-kV and 400-kV power lines were based on smaller numbers but were similar. The adjusted hazard ratio for Alzheimer s disease for people living within 50 m of a 220- or 400-kV power line was 1.31 (95 CI: 0.50, 3.46; n = 5). There was no dose-response according to the number of years spent living within 50 m of a power line for any of the diseases (Table 3). For Alzheimer s disease, the highest risk estimate was observed in the middle group (between 5 and 9 years), and the risk for those living within 50 m of a power line for 10 or more years was slightly Total no. 16,925 90,060 8,234 34,535 2,990 14,996 Female gender 7, , , , , , Mean age, years (SD) 74.4 (10.1) 75.0 (9.6) 44.7 (13.2) 46.5 (13.5) 64.9 (13.0) 66.2 (12.3) Marital status a Missing data b , , Never married 2, , , , , Ever married c 7, , , , , , Married at diagnosis 6, , , , , , Disposable income First quartile 4, , , , , Second and third quartiles (plus missing data) d 8, , , , , , Fourth quartile 4, , , , , in highest educational level e <23.5 5, , , , , <36.0 (plus 5, , , , , , missing data) f , , , , , , floors in residential building 1 floor (plus missing 11, , , , , , data) g >1 floor 5, , , , , Urbanity of residence Urban 12, , , , , , Rural 4, , , , , Abbreviation: SD, standard deviation. a Being in a registered partnership is considered equivalent to being married. b 4.2 of data were missing. c But not married at the time of diagnosis. d 1 of data were missing. e Percentage of persons with more than 12 years of education, grouped into 3 equal-sized groups. f 1 of data were missing. g 5.1 of data were missing.

5 974 Frei et al. decreased. Results remained virtually unchanged when we used the exposure window of 20 years before diagnosis (data not shown). Subgroup analyses on Alzheimer s disease With regard to Alzheimer s disease, there was no difference in risk estimates between men and women; adjusted hazard ratios for living within 50 m of a power line were 0.99 (95 CI: 0.53, 1.85) for men and 1.07 (95 CI: 0.63, 1.83) for women. Table 4 presents secondary analyses by age at diagnosis and calendar year of diagnosis. The risk for persons diagnosed at ages years was increased for those living within 50 m of a power line (adjusted HR = 1.92, 95 CI: 0.95, 3.87), while among those diagnosed at a later age, the hazard ratio was slightly decreased (HR = 0.81, 95 CI: 0.48, 1.34). In addition, the risk was slightly increased for cases living within 50 m of a power line diagnosed in 2003 or later, but those diagnosed earlier had a decreased hazard ratio (n = 2). Restricting analyses of age at diagnosis to persons diagnosed in 2003 or later yielded a significantly increased risk for those diagnosed by the age of 75 years (HR = 2.59, 95 CI: 1.17, 5.76), based on 9 cases, and a risk close to 1 for those diagnosed after the age of 75 years (Table 4). DISCUSSION The results of our population-based study do not suggest an increased risk of developing neurodegenerative diseases from living within close vicinity of a high-voltage power line. Overall, the risk of Alzheimer s disease was not increased. There were some weak indications from secondary analyses of an increased risk for persons diagnosed by the age of 75 years among people living within 50 m of a power line, which became statistically significant when data were restricted to cases diagnosed in 2003 or later, but with the overall finding of no association, other subgroups also yielded somewhat decreased risk estimates. Strengths and limitations The current study was register-based, covering the entire Danish population, and therefore eliminated the problem of selection bias. Although slightly more cases than controls had a full address history up to 20 years before diagnosis (94.2 vs. 89.1), the exposure distribution for the cases and controls for which we had partial address information was similar to the distribution for cases and controls with complete address histories (cases: 0.14 vs had ever lived within 50 m of a power line; controls: 0.24 vs. 0.17). We had individual information on relevant confounders such as socioeconomic variables and very accurate address information for determination of the geographical coordinates. Our study provides improvements over the Swiss study (10) in several respects. Our results were based on hospitalization data rather than information from death certificates, where underreporting of neurodegenerative diseases can be expected, especially for dementia diagnoses (15, 16). In Denmark, it is legally possible to obtain reimbursement for medication expenses only if the drug has been prescribed by a neurologist; therefore, we would expect that only a few patients with these diagnoses were not registered in the Danish Hospital Discharge Registry. Additionally, we had continuous information on the residential history of all study participants, while census data collected every 10 years were used for the Swiss study. This allowed us to calculate cumulative exposure more precisely. We were able to take into account a large time window for exposure (up to 20 years before diagnosis). By doing this, we allowed for the fact that the pathophysiological processes, especially for dementia diseases, begin years before diagnosis. It has been suggested that there is a temporal lag of about a decade between the deposition of amyloid β and the clinical manifestation of Alzheimer s disease (17). Usually, cognitive symptoms exist for a few years in dementia patients before admission to a hospital (13); therefore, we excluded the time period of 5 years before diagnosis. Conducting the analyses using the time frame 20 years before diagnosis yielded virtually the same results. We put a lot of effort into identifying all current and historical power lines with voltage levels of 132 kv. While the validity of Parkinson s disease diagnoses (18) and dementia diagnoses (in the elderly population of Denmark (evaluated in the year 2003 (13))) was found to be quite high, the validity of dementia subtypes was shown to be less reliable (13): 33 of Alzheimer s disease cases in 2003 were misclassified as dementia without specification. Nevertheless, for 81 of all patients recorded as having Alzheimer s disease, the recorded diagnosis was correct. Thus, the majority of our identified Alzheimer s disease cases would have been true cases of Alzheimer s disease, which is reassuring for the results. For vascular dementia, however, only 18.5 of 27 cases recorded as vascular dementia were confirmed as vascular dementia, with the rest being mainly dementia without specification (48.1) and Alzheimer s disease (22.2); therefore, the results for vascular dementia must be interpreted with caution. A limitation of our study was the potential for exposure misclassification. Distance to a power line is only a crude proxy for exposure to magnetic fields (19), as the magnetic field generated by power lines is determined by other factors, including load characteristics and placement of phases. However, a study in Denmark using personal measurement devices found that the magnetic fields in residences near power lines (defined as <100 m from 400-kV lines, <50 m from 132-/150-kV lines, and <25 m from 50-/ 60-kV lines) were markedly higher than those in houses farther away (geometric mean: 0.29 µt vs µt) (20). Other potential sources of ELF-MF from power lines not considered in this study were underground transmission cables, overhead power lines with lower voltage ( 132 kv), and transformer stations. In Denmark, most of the urban Copenhagen area is served by underground cables. In most situations, the field generated by an underground cable only exceeds background levels within a few meters of the cable, limiting the number of potentially exposed residences (21). Therefore, although there will be some highly exposed residences among persons classified as living beyond 600 m

6 Power Lines and Neurodegenerative Diseases 975 Table 2. Hazard Ratios for Neurodegenerative Disease Among Persons Who Had Ever Lived Within a Certain Distance of 132- to 400-kV Power Lines in the Time Span 5 20 Years Before Diagnosis, Denmark, Diagnosis and Distance From 132- to 400-kV Power Line, m a Crude Adjusted b HR 95 CI HR 95 CI Alzheimer s disease 0 < , , < , , , < , , , , ,693 1 Referent 1 Referent Vascular dementia 0 < , , < , , < , , , ,684 52,733 1 Referent 1 Referent Other dementia 0 < , , < , , , <600 2,635 16, , , , ,307 1 Referent 1 Referent Parkinson s disease 0 < , , < , , , < , , , ,864 84,164 1 Referent 1 Referent Multiple sclerosis 0 < , , < , , < , , , ,220 30,519 1 Referent 1 Referent Motor neuron disease 0 < , , < , , < , , ,754 13,834 1 Referent 1 Referent Abbreviation: CI, confidence interval; HR, hazard ratio. a Persons diagnosed with a neurodegenerative disease in Denmark during b Adjusted for disposable income, education, urbanization category, number of floors in the residential building, and marital status. from a power line, they are likely to constitute a very small percentage of the total reference population. We were not able to investigate occupational exposures, which could have constituted an additional source of ELF-MF exposure for some of the study participants. Comparison with previous studies and interpretation Except for the Swiss study (10), most studies that have investigated the relationship between exposure to ELF-MF and neurodegenerative diseases have been conducted in occupational settings. The direct comparison of residential studies with occupational studies is generally limited because of potential occupational coexposures and different exposure patterns. Exposures in occupational settings are generally greater than those in the general population (4, 6). In addition, while exposure levels in residential settings can be expected to be fairly constant, they might be more variable in occupational settings. Since no established mechanism by which magnetic fields could produce neurodegenerative diseases has been identified so far (2, 22), it is unclear which aspect of exposure

7 976 Frei et al. Table 3. Hazard Ratios for Neurodegenerative Disease According to Cumulative Duration of Residency Within 50 m of 132- to 400-kV Power Lines in the Time Span 5 20 Years Before Diagnosis, Denmark, Diagnosis and Cumulative Duration of Residency Within 50 m of a 132- to 400-kV Power Line, years a is relevant for health, if any, and peak exposures might (for example) be more influential than average exposures. There was little evidence of an association between magnetic field exposure and Parkinson s disease in occupational settings and in the Swiss residential study (5, 6, 10), which corresponds to our findings. Regarding motor neuron disease, in the Swiss study (10) and a very recent Brazilian study (23), no increased risks were observed, but the statistical power of these studies was very limited because of few cases. The occupational studies, however, found a more consistent link between electrical occupations and motor neuron disease (5, 6). Confounding by electric shock has been discussed as a Crude Adjusted b HR 95 CI HR 95 CI Alzheimer s disease , , , , 3.68 < , , 2.45 Always lived 600 m away 19, ,693 1 Referent 1 Referent Vascular dementia , , , , 4.29 < , , 1.63 Always lived 600 m away 9,684 52,733 1 Referent 1 Referent Other dementia , , , , 1.43 < , , 1.56 Always lived 600 m away 65, ,307 1 Referent 1 Referent Parkinson s disease , , , , 3.12 < , , 3.08 Always lived 600 m away 15,864 84,164 1 Referent 1 Referent Multiple sclerosis , , , , 1.51 < , , 2.56 Always lived 600 m away 7,220 30,519 1 Referent 1 Referent Motor neuron disease , , , , < , , 4.30 Always lived 600 m away 2,754 13,834 1 Referent 1 Referent Abbreviations: CI, confidence interval; HR, hazard ratio. a Persons diagnosed with a neurodegenerative disease in Denmark during b Adjusted for disposable income, education, urbanization category, number of floors in the residential building, and marital status. potential explanation for the findings due to weaker evidence of an association when magnetic field levels were measured (6), but the evidence is controversial (9). A recent Danish study did not find an increased risk of amyotrophic lateral sclerosis for persons who had experienced an electric shock, but the statistical power was low in that study as well (24). With regard to Alzheimer s disease or dementia, the results of previous occupational studies are inconsistent but generally indicate increased risks for stronger magnetic fields (4 6). Associations for Alzheimer s disease were found at levels of 0.1 µt to 0.5 µt (4). Residential exposures within 50 m of a power line can be on the same order of magnitude: The

8 Power Lines and Neurodegenerative Diseases 977 Table 4. Hazard Ratios a for Diagnosis of Alzheimer s Disease Among Persons Living Within 50 m of a 132- to 400-kV Power Line in the Time Span 5 20 Years Before Diagnosis, According to Age at Diagnosis and Year of Diagnosis, Denmark, Year of Diagnosis < All Years Age at Diagnosis, years b HR 95 CI HR 95 CI HR 95 CI Crude estimate , , , 3.35 > c , , 1.18 All ages , , , 1.37 Adjusted estimate d , , , 3.87 > c , , 1.34 All ages , , , 1.56 Abbreviations: CI, confidence interval; HR, hazard ratio. a Reference category: always having lived 600 m from a 132- to 400-kV power line. b Persons diagnosed with Alzheimer s disease in Denmark during c No risk calculation was possible because of missing cases. d Adjusted for disposable income, education, urbanization category, number of floors in the residential building, and marital status. Danish company energinet.dk (Fredericia, Denmark), the owner of the electricity infrastructure in Denmark, has calculated magnetic field levels for the most common pylons with average load currents in Denmark: For a 2-system 400-kV line, calculated magnetic fields at a distance of 50 m from the line were between 0.3 µt and 0.9 µt for optimal phase configurations and up to 1.1 µt for other configurations, and for a 132-kV line, they were approximately 0.1 µt with optimal phase configurations and 0.3 µt with other configurations (21). Potential work-related coexposures, such as exposure to solvents, pesticides, and lead, were frequently discussed as potential confounders for dementia (4, 5). In addition, Garcia et al. (4) found indications of publication bias, with larger studies on Alzheimer s disease showing a smaller degree of association than smaller studies. Unlike suggestions in some studies (8, 25), we have not observed risk differences between men and women for Alzheimer s disease. Regarding latency of the effects, the results of some occupational studies indicate a rather late-acting influence in the disease process, while other studies have found effects for accumulated exposure throughout life (4, 7, 9, 25). The Swiss study found increased risk estimates for persons living within 50 m of a 220- to 380-kV power line for at least 15 years (HR = 2.00, 95 CI: 1.21, 3.33), indicating a cumulative effect (10). In the present study, we did not observe such a risk increase for people living close to power lines for at least 10 years. As suggested by occupational studies in Sweden (7 9), we found indications that age of onset might play a role. The proportion of patients with relatively early Alzheimer s disease onset might have been increased in the Swiss study, since underreporting of dementia diagnoses on death certificates increases with the age of the deceased (15, 16), which could be a potential explanation for the risk increase found in the Swiss study. Conclusions Overall, we did not observe an increased risk of neurodegenerative diseases among persons living close to power lines. Our study therefore confirms the findings of a previous study (10) of no association for Parkinson s disease, multiple sclerosis, motor neuron disease, vascular dementia, and other types of dementia. Furthermore, it does not confirm the finding of an increased risk of Alzheimer s disease. Given the high quality and representativeness of our study, this is reassuring, since together the studies provide little evidence that living close to a power line increases the risk of any neurodegenerative disease. We found some weak suggestions that ELF-MF exposure might increase the risk of Alzheimer s disease when diagnosed by the age of 75 years, although those subgroup analyses must be interpreted with care given the overall finding of no association. If the latter observed association were causal, ELF-MF exposure would explain 5 (0.02) of the Alzheimer s disease cases in our study population. ACKNOWLEDGMENTS Author affiliations: Danish Cancer Society Research Center, Danish Cancer Society, Copenhagen, Denmark (Patrizia Frei, Aslak Harbo Poulsen, Camilla Pedersen, Christoffer Johansen); Environment Department, Electric Power Research Institute, Palo Alto, California (Gabor Mezei); Memory Disorders Research Group, Department of Neurology, Copenhagen University Hospital, Rigshospitalet, Denmark (Lise Cronberg Salem); Department of Epidemiology and Public Health, Swiss Tropical and Public Health Institute, University of Basel, Basel, Switzerland (Patrizia Frei, Martin Röösli); and

9 978 Frei et al. Section of Environment and Radiation, International Agency for Research on Cancer, Lyon, France (Joachim Schüz). This work was supported by a fellowship for prospective researchers awarded by the Swiss National Science Foundation (stipend PBBSP ) to P.F. A.H.P. was supported by a stipend for doctoral students from the Danish Graduate School in Public Health Science. All other authors contributed to this work on the basis of their respective core budget positions. Additional funding was provided by the Electric Power Research Institute to the Danish Cancer Society Research Center (contract EP-P38793/C17252). We thank Rikke Baastrup for support with the address data, Nick Martinussen for help in the preparation of the data, the power-line companies for providing data on the locations of their power lines, Kasper Grue Understrup of the Danish National Board of Health for providing data on cases and controls, and Henrik Hobel of DM Partner A/S (Copenhagen, Denmark) for providing geocodes. Special thanks go to Martin Hertach for supporting our analyses involving ArcGIS with his excellent expertise. Conflict of interest: none declared. REFERENCES 1. International Agency for Research on Cancer. Non-Ionizing Radiation, Part 1: Static and Extremely Low-Frequency (ELF) Electric and Magnetic Fields. (IARC Monographs on the Evaluation of Carcinogenic Risks to Humans, vol 80). Lyon, France: International Agency for Research on Cancer; World Health Organization. Extremely Low Frequency Fields. (Environmental Health Criteria, vol 238). Geneva, Switzerland: World Health Organization; Sobel E, Davanipour Z, Sulkava R, et al. Occupations with exposure to electromagnetic fields: a possible risk factor for Alzheimer s disease. Am J Epidemiol. 1995;142(5): Garcia AM, Sisternas A, Hoyos SP. Occupational exposure to extremely low frequency electric and magnetic fields and Alzheimer disease: a meta-analysis. Int J Epidemiol. 2008; 37(2): Hug K, Röösli M, Rapp R. Magnetic field exposure and neurodegenerative diseases recent epidemiological studies. Soz Präventivmed. 2006;51(4): Kheifets L, Bowman JD, Checkoway H, et al. Future needs of occupational epidemiology of extremely low frequency electric and magnetic fields: review and recommendations. Occup Environ Med. 2009;66(2): Feychting M, Pedersen NL, Svedberg P, et al. Dementia and occupational exposure to magnetic fields. Scand J Work Environ Health. 1998;24(1): Andel R, Crowe M, Feychting M, et al. Work-related exposure to extremely low-frequency magnetic fields and dementia: results from the population-based study of dementia in Swedish twins. J Gerontol A Biol Sci Med Sci. 2010; 65(11): Feychting M, Jonsson F, Pedersen NL, et al. Occupational magnetic field exposure and neurodegenerative disease. Epidemiology. 2003;14(4): Huss A, Spoerri A, Egger M, et al. Residence near power lines and mortality from neurodegenerative diseases: longitudinal study of the Swiss population. Am J Epidemiol. 2009; 169(2): Scientific Committee on Emerging and Newly Identified Health Risks, European Commission. Research Needs and Methodology to Address the Remaining Knowledge Gaps on the Potential Health Effects of EMF. Brussels, Belgium: European Commission; ( ph_risk/committees/04_scenihr/docs/scenihr_o_024.pdf). (Accessed March 12, 2010). 12. Danish National Board of Health. The Activity in the Hospital Care System [in Danish]. Copenhagen, Denmark: Danish National Board of Health; Phung TK, Andersen BB, Hogh P, et al. Validity of dementia diagnoses in the Danish hospital registers. Dement Geriatr Cogn Disord. 2007;24(3): Phung TK, Waltoft BL, Kessing LV, et al. Time trend in diagnosing dementia in secondary care. Dement Geriatr Cogn Disord. 2010;29(2): Ganguli M, Rodriguez EG. Reporting of dementia on death certificates: a community study. J Am Geriatr Soc. 1999; 47(7): Østbye T, Hill G, Steenhuis R. Mortality in elderly Canadians with and without dementia: a 5-year follow-up. Neurology. 1999;53(3): Sperling RA, Aisen PS, Beckett LA, et al. Toward defining the preclinical stages of Alzheimer s disease: recommendations from the National Institute on Aging-Alzheimer s Association workgroups on diagnostic guidelines for Alzheimer s disease. Alzheimers Dement. 2011;7(3): Olsen JH, Tangerud K, Wermuth L, et al. Treatment with levodopa and risk for malignant melanoma. Mov Disord. 2007;22(9): Maslanyj M, Simpson J, Roman E, et al. Power frequency magnetic fields and risk of childhood leukaemia: misclassification of exposure from the use of the distance from power line exposure surrogate. Bioelectromagnetics. 2009;30(3): Skotte JH. Exposure to power-frequency electromagnetic fields in Denmark. Scand J Work Environ Health. 1994; 20(2): Energinet.dk. Catalog: Magnetic Field Strengths for Different Types of Power Lines [in Danish]. ( SiteCollectionDocuments/Danske20dokumenter/ Klimaogmiljo/01_10_362520Magnetfelter_ Magnetfelternes20størrelse20-20Katalog.pdf). (Accessed April 23, 2011). Fredericia, Denmark: Energinet. dk; Swanson J, Kheifets L. Biophysical mechanisms: a component in the weight of evidence for health effects of power-frequency electric and magnetic fields. Radiat Res. 2006;165(4): Marcilio I, Gouveia N, Pereira Filho ML, et al. Adult mortality from leukemia, brain cancer, amyotrophic lateral sclerosis and magnetic fields from power lines: a case-control study in Brazil. Rev Bras Epidemiol. 2011;14(4): Grell K, Meersohn A, Schüz J, et al. Risk of neurological diseases among survivors of electric shocks: a nationwide cohort study, Denmark, Bioelectromagnetics. 2012;33(6): Qiu C, Fratiglioni L, Karp A, et al. Occupational exposure to electromagnetic fields and risk of Alzheimer s disease. Epidemiology. 2011;15(6):