Association of mortality with high temperatures in a temperate climate: England and Wales.

BACKGROUND: It is well known that high ambient temperatures are associated with increased mortality, even in temperate climates, but some important details are unclear. In particular, how heat-mortality associations (for example, slopes and thresholds) vary by climate has previously been considered only qualitatively. METHODS: An ecological time-series regression analysis of daily counts of all-cause mortality and ambient temperature in summers between 1993 and 2006 in the 10 government regions was carried out, focusing on all-cause mortality and 2-day mean temperature (lags 0 and 1). RESULTS: All regions showed evidence of increased risk on the hottest days, but the specifics, in particular the threshold temperature at which adverse effects started, varied. Thresholds were at about the same centile temperatures (the 93rd, year-round) in all regions-hotter climates had higher threshold temperatures. Mean supra-threshold slope was 2.1%/°C (95% CI 1.6 to 2.6), but regions with higher summer temperatures showed greater slopes, a pattern well characterised by a linear model with mean summer temperature. These climate-based linear-threshold models capture most, but not all, the association; there was evidence for some non-linearity above thresholds, with slope increasing at highest temperatures. CONCLUSION: Effects of high daily summer temperatures on mortality in English regions are quite well approximated by threshold-linear models that can be predicted from the region's climate (93rd centile and mean summer temperature). It remains to be seen whether similar relationships fit other countries and climates or change over time, such as with climate change.

Exposure-response relationship between lung cancer and polycyclic aromatic hydrocarbons (PAHs).

OBJECTIVES: To estimate the exposure-response function associating polycyclic aromatic hydrocarbon (PAH) exposure and lung cancer, with consideration of smoking. METHODS: Mortality, occupational exposure and smoking histories were ascertained for a cohort of 16,431 persons (15,703 men and 728 women) who had worked in one of four aluminium smelters in Quebec from 1950 to 1999. A variety of exposure-response functions were fitted to the cohort data using generalised relative risk models. RESULTS: In 677 lung cancer cases there was a clear trend of increasing risk with increasing cumulative exposure to PAH measured as benzo(a)pyrene (BaP). A linear model predicted a relative risk of 1.35 (95% CI 1.22 to 1.51) at 100 microg/m(-3) BaP years, but there was a significant departure from linearity in the direction of decreasing slope with increasing exposures. Among the models tried, the best fitting were a two-knot cubic spline and a power curve (RR = (1+bx)(p)), the latter predicting a relative risk of 2.68 at 100 microg/m(-3) BaP years. Additive models and multiplicative models for combining risks from occupational PAH and smoking fitted almost equally well, with a slight advantage to the additive. CONCLUSION: Despite the large cohort with long follow-up, the shape of the exposure-response function and the mode of combination of risks due to occupational PAH and smoking remains uncertain. If a linear exposure-response function is assumed, the estimated slope is broadly in line with the estimate from a previous follow-up of the same cohort, and somewhat higher than the average found in a recent meta-analysis of lung cancer studies.

Estimating reduction in occupational disease burden following reduction in exposure.

BACKGROUND/OBJECTIVE: Many occupational exposures causing disease cannot feasibly be eliminated entirely, but policies that reduce the exposures may be under consideration. This paper sets out to clarify how to estimate the reduction in occupational disease following a reduction in exposure, and shows a real-data illustration for doing this. METHODS: Modest extensions of standard expressions for attributable fractions permit estimation of fractions by which cases would be reduced by policies that do not eliminate exposure but change exposure distributions. However, this requires information on the exposure-response relation and on distribution of exposures. RESULTS: From hypothetical scenarios and a real example this paper explores how attributable cases are distributed by exposure level and, in particular, the proportion by which attributable cancers are reduced by eliminating exposures above a limit (the classic occupational limit regulation). It shows how this depends on the shape of the exposure-response relation and to some extent the shape of the exposure distribution, as well as on the proportion exposed above the limit. For linear no-threshold relations and left-skewed exposure distributions, the majority of the burden may be in a large number of people experiencing small relative risks, and thus may not be tackled by a strategy to reduce exposures above a certain limit. CONCLUSION: With appropriate data, estimating the disease burden in terms of the distribution of exposure is straightforward and can help to clarify the likely outcome of an intervention.

The effect of temperature on food poisoning: a time-series analysis of salmonellosis in ten European countries.

We investigated the relationship between environmental temperature and reported Salmonella infections in 10 European populations. Poisson regression adapted for time-series data was used to estimate the percentage change in the number of cases associated with a 1 degree C increase in average temperature above an identified threshold value. We found, on average, a linear association between temperature and the number of reported cases of salmonellosis above a threshold of 6 degrees C. The relationships were very similar in The Netherlands, England and Wales, Switzerland, Spain and the Czech Republic. The greatest effect was apparent for temperature 1 week before the onset of illness. The strongest associations were observed in adults in the 15-64 years age group and infection with Salmonella Enteritidis (a serotype of Salmonella). Our findings indicate that higher temperatures around the time of consumption are important and reinforce the need for further education on food-handling behaviour.

Mortality and temperature in Sofia and London.

STUDY OBJECTIVE: Heat and cold have been associated with increased mortality, independently of seasonal trends, but details are little known. This study explores associations between mortality and temperature in two European capitals-Sofia and London-using four years of daily deaths, air pollution, and weather data. DESIGN: Generalised additive models were used to permit non-linear modelling of confounders such as season and humidity, and to show the shape of mortality-temperature relations-using both two day and two week average temperatures separately. Models with linear terms for heat and cold were used to estimate lags of effect, linear effects, and attributable fractions. PARTICIPANTS: 44701 all age all cause deaths in Sofia (1996-1999) and 256464 in London (1993-1996). MAIN RESULTS: In London, for each degree of extreme cold (below the 10th centile of the two week mean temperature), mortality increased by 4.2% (95% CI 3.4 to 5.1), and in Sofia by 1.8% (0.6 to 3.9). For each degree rise above the 95th centile of the two day mean, mortality increased by 1.9% (1.4 to 2.4) in London, and 3.5% (2.2 to 4.8) in Sofia. Cold effects appeared after lags of around three days and lasted-particularly in London-at least two weeks. Main heat effects occurred more promptly. There were inverse associations at later lags for heat and cold in Sofia. CONCLUSIONS: Average temperatures over short periods do not adequately model cold, and may be inadequate for heat if they ignore harvesting effects. Cold temperatures in London, particularly, seem to harm the general population and the effects are not concentrated among persons close to death.

The combination of effects on lung cancer of cigarette smoking and exposure in quebec chrysotile miners and millers.

Although it is well known that both cigarette smoke and microscopic airborne asbestos fibres can cause lung cancer, evidence as to how these two agents combine is nebulous. Many workers have believed in the multiplicative theory, whereby asbestos increases the risk in proportion to the risk from other causes. However, evidence against this theory is mounting: a recent review concluded that the multiplicative hypothesis was untenable, and that the relative risk of lung cancer from asbestos exposure was about twice as high in non-smokers as in smokers, a finding largely independent of type of asbestos fibre. The criteria for entry to the current study were met by 7279 men in the 1891-1920 birth cohort of Quebec chrysotile miners and millers. The data consisted of date of birth, place of employment, smoking habit, asbestos exposure accumulated to age 55 and, for those 5527 who died between 1950 and June 1992, date and cause of death; 533 of the deaths were from lung cancer. For the principal analyses, ex-smokers were excluded from the study cohort, which comprised 5888 men, of whom 473 died of lung cancer. The conventional form of analysis is simply of the double dichotomy: non-smokers of cigarettes, 'unexposed' and exposed; all others, 'unexposed' and exposed. The respective standardized lung cancer mortality ratios (SMRs) were 0.29 and 0.62; and 1.37 and 1.72. Thus, the differences in relative risk, due to exposure, were closely similar, 0.33 and 0.35. On the other hand, the effects of asbestos measured by the corresponding ratios, 2.12 and 1.25, did differ, being 1.7 times as high in non-smokers as in others. The principal analysis was much more penetrating: the method was to fit models to a 'disaggregated' 6 x 10 array, by smoking habit (excluding ex-smokers) and asbestos exposure, of lung cancer SMRs. Both linear and log-linear models were fitted: the former included the additive and linear-multiplicative; the latter embraced the more conventional multiplicative form. The additive model fitted much the best. The fit of each multiplicative model was improved by the introduction of an interaction term that implied a less than multiplicative relationship. Thus smoking and exposure to chrysotile appear to have acted independently in causing lung cancer, with 10 cigarettes a day having an effect roughly equivalent to exposure amounting to 700 million particles per cubic foot x years. The refutation of the multiplicative hypothesis in these data reinforces its inapplicability in general; but the additive hypothesis is not generally applicable either. Indeed, there seems to be no good reason to believe that interactions conform to any simple theory. The implications are important.

Case-referent survey of young adults with mesothelioma: I. Lung fibre analyses.

OBJECTIVES: Our study aimed to determine the lung tissue concentration of asbestos and other mineral fibres by type and length in persons with mesothelioma aged 50 yr or less at time of diagnosis, compared to controls of similar age and geographical region. In this age group it was thought that most, but not all, work-related exposures would have been since 1970, when the importation of crocidolite, but not amosite, was virtually eliminated. METHODS: Eligible cases were sought from recent reports by chest physicians to the SWORD occupational disease surveillance scheme. Lung tissue samples were obtained at autopsy from 69 male and four female cases, and mineral fibres identified, sized and counted by electron microscopy. Fibre concentrations per microg dry tissue were compared with similar estimates from a control series of autopsies of sudden or accidental deaths. Unadjusted, and adjusted odds ratios calculated by logistic regression, assessed relative risk in relation to fibre type, length and concentration. RESULTS: Unadjusted and adjusted odds ratios increased steadily with concentration of crocidolite, amosite, tremolite and all amphiboles combined. There was also some increase with chrysotile, but well short of statistical significance. Incremental risk examined in a linear model was as highly significant for all amphiboles together as individually. Short, medium and long amphibole fibres were all associated with increased risk in relation to length. Mullite and iron fibres were significant predictors of mesothelioma when considered without adjustment for confounding by amphiboles, but, after adjustment, were weak and far from statistically significant. CONCLUSIONS: In this young age group, amosite and crocidolite fibres could account for about 80% of cases of mesothelioma, and tremolite for some 7%. The contribution of chrysotile, because of low biopersistence, cannot be reliably assessed at autopsy, but to the extent that tremolite is a valid marker, our results suggest that it was small. The steep linear trend in odds ratio shown by amphiboles combined indicates that their effects may be additive, with increased risk from the lowest detectable fibre level. Non-asbestos mineral fibres probably made no contribution to this disease. Contrary to expectation, however, some 90% of cases were in men who had started work before 1970; this was so whether or not amosite or crocidolite was found in lung tissue.

The determinants of Canadian children's personal exposures to magnetic fields.

Study of the health effects of magnetic fields often depends on identifying determinants and hence indicators of personal exposure. This study identified determinants of children's exposure to magnetic fields and constructed a prediction model for them. For 632 children participating in a case-control study of childhood leukemia, we made direct measures of exposure over 48 h using a portable device, together with observations on candidate determinants. A child's age and sex, the proportion of time spent in the home, and their parents' education or income were very weak predictors of (logged) mean 48 h magnetic field (R(2) < 1%). More important were province (R(2) = 8.0%) and type of residence (R(2) = 11.3%). Low temperatures at the time of measurement were associated with high fields (about 20% increase for each 10 degrees C below 14, R(2) = 4.9%). Several visible attributes of wiring around residences predicted exposure, mostly captured in the Wertheimer-Leeper wire code (R(2) = 13.5%). Stationary 24 h measurement in the bedroom (R(2) = 63.3%) and spot measurements outside the house (R(2) = 40.7%) predicted personal exposures best. Adding other minor predictors increased only slightly variance explained by 24 h stationary (R(2) = 66.2%) and spot (R(2) = 46.8%) measurements. Without spot or stationary measurements, the best model was much less powerful (R(2) = 29.0%). We conclude that spot measurements outside the residence provide a moderately effective basis for estimating exposure for children living there, but do not perform as well as 24 h stationary measurements in the child's bedroom. Although several other easily-observed variables were associated with personal exposure, they were weak determinants, either individually or in combination.

Methodological approaches to the analysis of hierarchical studies of air pollution and respiratory health--examples from the CESAR study. Central European Study on Air pollution and Respiratory Health.

OBJECTIVES: Many studies of air pollution and health are carried out over several geographical areas, and sometimes over several countries. This paper explores three approaches to analysis in such studies: a non hierarchical model, a two-stage analysis, and multilevel modelling. Illustrations are given using a preliminary subset of data from the CESAR study. DESIGN: The Central European Study on Air pollution and Respiratory Health (CESAR) was conducted in 25 areas within six Central European countries, enrolling 20,271 schoolchildren. Pollution averages were calculated for each area. Associations between pollution and health outcomes were estimated under different models. MAIN RESULTS: A regression analysis of log FVC (forced vital capacity) on PM10, ignoring the geographical hierarchy, estimated a significant mean drop in FVC (adjusted for confounders) of 2.2% (95% CI 0.5% to 1.3%), p=0.007, from the area with the lowest PM10 to that with the highest. A multilevel model (mlm), using data for all children, but with random effects at area and country level, estimated a drop of 2.8% (-0.6% to 6.1%), p=0.110. A two-stage analysis (mean log FVC, adjusted for confounders, was estimated for each area using regression, and these means then regressed on PM10) estimated a drop of 2.6% (-0.5% to 5.5%), p=0.101. Simulation exercises showed the non hierarchical method to be very inadequate in the context of the CESAR study, with only half of all 95% confidence intervals for the estimated PM10 slope containing the true value (i.e., that used to create the simulated data). The two-stage and multilevel modelling methods gave results which were substantially better, though both underperformed slightly. All three methods appeared to give unbiased slope estimates. CONCLUSIONS: Acknowledgement of hierarchical structures is essential in statistical inference--standard errors can be substantially incorrect when they are ignored. Multilevel, random-effects models correctly address hierarchical structures, though having few units at higher levels can cause problems in convergence, especially where complex modelling is required. Two-stage analyses, acknowledging hierarchy, provide simple alternatives to random-effects models.

Exposures of children in Canada to 60-Hz magnetic and electric fields.

OBJECTIVES: This study sought to characterize personal exposures of Canadian children to 60-Hz magnetic and electric fields and explain the variability. METHODS: Altogether 382 Canadian children up to 15 years of age wore meters recording 60-Hz electric and magnetic fields over 2 days. Meter location was noted. Thereafter, meters measured fields in the center of the children's bedrooms for 24 hours. Personal exposures were calculated for home, school or day care, outside the home, bedroom at night, and all categories combined (total). RESULTS: The arithmetic mean (AM) was 0.121 microT [geometric mean (GM): 0.085 microT), range 0.01-0.8 microT] for total magnetic fields. Fifteen percent of the total exposures exceeded 0.2 microT. The AM of the total electric fields was 14.4 (GM 12.3, range 0.82-64.7) V/m. By location category, the highest and lowest magnetic fields occurred at home during the day (0.142 microT) and during the night (0.112 microT), respectively. Measurements during sleep provided the highest correlation with total magnetic field exposure. Province of measurement explained 14.7% of the variation in the logarithms of total magnetic fields, and season accounted for an additional 1.5%. CONCLUSIONS: This study has identified differences in children's magnetic field exposures between provinces. Measurements at night provided the best surrogate for predicting total magnetic field exposure, followed by at-home exposure and 24-hour bedroom measurements. Electrical heating and air conditioning, wiring type, and type of housing appear to be promising indicators of magnetic field levels.

Power-frequency electric and magnetic fields and risk of childhood leukemia in Canada.

In a case-control study of childhood leukemia in relation to exposure to power-frequency electric and magnetic fields (EMF), 399 children resident in five Canadian provinces who were diagnosed at ages 0-14 years between 1990 and 1994 (June 1995 in British Columbia and Quebec) were enrolled, along with 399 controls. Exposure assessment included 48-hour personal EMF measurement, wire coding and magnetic field measurements for subjects' residences from conception to diagnosis/reference date, and a 24-hour magnetic field bedroom measurement. Personal magnetic fields were not related to risk of leukemia (adjusted odds ratio (OR) = 0.95, p for trend = 0.73) or acute lymphatic leukemia (OR = 0.93, p for trend = 0.64). There were no clear associations with predicted magnetic field exposure 2 years before the diagnosis/reference date or over the subject's lifetime or with personal electric field exposure. A statistically nonsignificant elevated risk of acute lymphatic leukemia was observed with very high wiring configurations among residences of subjects 2 years before the diagnosis/reference date (OR = 1.72 compared with underground wiring, 95% confidence interval 0.54-5.45). These results provide little support for a relation between power-frequency EMF exposure and risk of childhood leukemia.

Geographical variation in anophthalmia and microphthalmia in England, 1988-94.

OBJECTIVE: To investigate the geographical variation and clustering of congenital anophthalmia and microphthalmia in England, in response to media reports of clusters. DESIGN: Comparison of pattern of residence at birth of cases of anophthalmia and microphthalmia in England in 1988-94, notified to a special register, with pattern of residence of all births. Three groups studied included all cases, all severe cases, and all severe cases of unknown aetiology. OUTCOME MEASURES: Prevalence rates of anophthalmia and microphthalmia by region and district, and by ward population density and socioeconomic deprivation index of enumeration district grouped into fifths. Clustering expressed as the tendency for the three nearest neighbours of a case to be more likely to be cases than expected by chance, or for there to be more cases within circles of fixed radius of a case than expected by chance. RESULTS: The overall prevalence of anophthalmia and microphthalmia was 1.0 per 10 000 births. Regional and district variation in prevalence did not reach statistical significance. Prevalence was higher in rural than urban areas: the relative risk in the group of wards of lowest population density compared with the most densely populated group was 1.79 (95% confidence interval 1.15 to 2.81) for all cases and 2.37 (1.38 to 4. 08) for severe cases. There was no evidence of a trend in risk with socioeconomic deprivation. There was very little evidence of localised clustering. CONCLUSIONS: There is very little evidence to support the presence of strongly localised environmental exposures causing clusters of children to be born with anophthalmia or microphthalmia. The excess risk in rural areas requires further investigation.

Effect of measurement error on epidemiological studies of environmental and occupational exposures.

Random error (misclassification) in exposure measurements usually biases a relative risk, regression coefficient, or other effect measure towards the null value (no association). The most important exception is Berkson type error, which causes little or no bias. Berkson type error arises, in particular, due to use of group average exposure in place of individual values. Random error in exposure measurements, Berkson or otherwise, reduces the power of a study, making it more likely that real associations are not detected. Random error in confounding variables compromises the control of their effect, leaving residual confounding. Random error in a variable that modifies the effect of exposure on health--for example, an indicator of susceptibility--tends to diminish the observed modification of effect, but error in the exposure can create a supurious appearance of modification. Methods are available to correct for bias (but not generally power loss) due to measurement error, if information on the magnitude and type of error is available. These methods can be complicated to use, however, and should be used cautiously as "correction" can magnify confounding if it is present.

Task-based estimation of past exposures to 60-hertz magnetic and electric fields at an electrical utility.

OBJECTIVES: Past exposures of electric utility workers to extremely low-frequency (ELF) magnetic (B) and electric fields (E) in Quebec were estimated. METHODS: The current intensities were measured and durations of exposures determined for tasks or work locations in 14 job categories. Past task or location intensities were extrapolated from the present on the basis of interviews with long-term workers and utility personnel. Past task or location durations were estimated for the long-term workers. Time-weighted average (TWA) exposures for past periods were reconstructed for jobs from the intensity and duration estimates. RESULTS: Magnetic fields were estimated to have increased the most over time for substation and distribution-line jobs. Magnetic field exposures for jobs in the generation and transmission of electricity were estimated to have increased very little. For substation jobs, the ratios of magnetic fields in 1945 to those in 1990 ranged from 0.42 to 0.69; the corresponding figures for distribution-line jobs ranged from 0.36 to 0.94. For electric fields in substations, the estimated increase over time was less than for magnetic fields, the 1945:1990 ratios ranging from 0.59 to 0.88. For the distribution line jobs, the 1945:1990 ratios for electric fields were less than 1.0 in 4 cases (0.6 to 0.89), more than 1.0 in 3 others (1.13 to 2.01) and unchanged in 1. CONCLUSIONS: Reconstruction of TWA exposures allowed changes in the intensity and the duration of exposures to be considered separately. Documentation of the intensity and duration of exposures for different tasks allows exposure reconstruction for jobs that have ceased to exist. The method is applicable elsewhere if exposure-monitoring records allow the level and duration of exposures for tasks or locations to be calculated and if estimates of past durations and intensities of exposures can be reliably obtained.

Exposure to 60-Hz magnetic and electric fields at a Canadian electric utility.

OBJECTIVES: The purpose of this study was to estimate exposure to extremely low frequency (ELF) magnetic and electric fields in the Québec electrical utility Hydro-Quebec. METHODS: Personal exposures to ELF magnetic and electric fields were measured for workers randomly selected from 32 job categories at Hydro-Québec. Weekly arithmetic and geometric means, and other indices of exposure were estimated from 465 worker-weeks of data. RESULTS: By job category, the arithmetic means of the ELF magnetic field ranged from 0.09 to 2.36 microT. Those of the ELF electric field exposures ranged from 2.5 to 400 V.m-1. Within each field, correlations of either the arithmetic or geometric means with alternative indices, including an index of the time rate of change, were generally high (r > or = 0.8). Exceptions were the 20th percentile of the electric means and the proportion of time above 12.4 and 100 microT. The day-to-day variation of exposure was greater than the variation between workers. The median between-day and between-worker components of variance (as geometric standard deviations) by job category were 2.13 and 1.71 for magnetic fields (2.24 and 1.81 for electric fields). CONCLUSIONS: Substation workers, hydroelectric generating station operators, and cable splicers showed the highest arithmetic means for 60-Hz magnetic fields above 1 microT. For 60-Hz electric fields, forestry workers, equipment electricians in 735 kV substations, and distribution linemen (contact method) had arithmetic mean exposures greater than 100 W.m-1. Of the total variance in the logarithms of the weekly magnetic and electric field means, job category explained 49.6% and 59.5%, respectively.

Optimizing power in allocating resources to exposure assessment in an epidemiologic study.

We consider an epidemiologic study with a fixed budget, in which resources may be put into increasing sample size or into improving accuracy of exposure assessments. To maximize study power (efficiency), improving accuracy is preferable if and only if the proportional increase in the square of the validity coefficient is more than the proportional increase in total study costs per subject that is required to achieve it. (The validity coefficient is the correlation between the true exposure and the approximate assessment in the study base.) This is most likely to be so if the cost of exposure measurement remains a small proportion of the overall costs per subject. The design with maximum power will not generally have minimum bias in measure of effect, so that alternative optimality criteria are required if this bias is important.

Exposure to magnetic fields estimated from last job held in an electrical utility in Québec, Canada: a validation study.

OBJECTIVES: To investigate how closely the variables of exposures to magnetic fields based on the last job held in an electrical utility in Québec, Canada, compared with those based on the workers' entire employment history with the company. METHODS: In large cohort studies, the last job held is often used to assign exposure to the study subjects. Exposure was assigned in this way for a mortality study of a cohort of electrical utility workers in Québec. For the present study, a sample of the cohort was used to compare the exposure estimates obtained from the last job with those obtained from full work histories. RESULTS: The correlations between indices based on the last job and on all jobs varied between 0.75 and 0.78. The mean was slightly lower when only the last job was used. The last job was particularly good in identifying the most highly exposed people (for the exposure cut off point of 90th percentile for the last job and for all jobs, sensitivity = 0.69, specificity = 0.97, kappa = 0.66). The results suggest that although not all workers starting in highly exposed jobs stayed in them, it seemed that the workers who ended their working life in highly exposed jobs had stayed in these jobs throughout their working life. CONCLUSION: The results indicated some (but not catastrophic) loss of information when estimates of exposure were based on the last job only.

A case cohort study of suicide in relation to exposure to electric and magnetic fields among electrical utility workers.

OBJECTIVES: This case cohort study examines whether there is an association between exposure to electric and magnetic fields and suicide in a population of 21,744 male electrical utility workers from the Canadian Province of Québec. METHODS: 49 deaths from suicide were identified between 1970 and 1988 and a subcohort was selected comprising a 1% random sample from this cohort as a basis for risk estimation. Cumulative and current exposures to electric fields, magnetic fields, and pulsed electromagnetic fields (as recorded by the POSITRON meter) were estimated for the subcohort and cases through a job exposure matrix. Two versions of each of these six indices were calculated, one based on the arithmetic mean (AM), and one on the geometric mean (GM) of field strengths. RESULTS: For cumulative exposure, rate ratios (RR) for all three fields showed mostly small non-significant increases in the medium and high exposure groups. The most increased risk was found in the medium exposure group for the GM of the electric field (RR = 2.76, 95% CI 1.15-6.62). The results did not differ after adjustment for socioeconomic state, alcohol use, marital state, and mental disorders. There was a little evidence for an association of risk with exposure immediately before the suicide. CONCLUSION: Some evidence for an association between suicide and cumulative exposure to the GM of the electric fields was found. This specific index was not initially identified as the most relevant index, but rather emerged afterwards as showing the most positive association with suicide among the 10 indices studied. Thus the evidence from this study for a causal association between exposure to electric fields and suicide is weak. Small sample size (deaths from suicide) and inability to control for all potential confounding factors were the main limitations of this study.

A mortality study of electrical utility workers in Québec.

OBJECTIVES: The objective of this study was to investigate the mortality of electrical utility workers exposed to electric and magnetic fields. METHODS: A historical cohort mortality study was carried out on 21,744 workers who were employed in an electrical company in the province of Québec between 1970 and 1988. The last job held by each study subject was coded. A job exposure matrix (JEM) was used to estimate the exposure to 60 Hz electric and magnetic fields, and pulsed electromagnetic fields (as recorded by the PEMF (POSITRON) meter) in this job. Standardised mortality ratios (SMRs) were calculated relative to Québec men. Because the exposure was exclusively among blue collar workers, the remainder of the analyses by exposure were restricted to them. Rate ratios (RRs) in the exposed groups relative to the background groups were estimated with Poisson regression. There were 1582 deaths by the end of follow up. RESULTS: SMRs were almost all below one and never substantially increased, although there were a few increased rate ratios (RRs). There was a significant RR of 2.00 (95% confidence interval (95% CI) 1.37-2.93) for deaths caused by accidents and violence in workers exposed to magnetic fields, RR of 1.82 (95% CI 1.25-2.65) for electric fields, and RR of 1.62 (95% CI 1.13-2.32) for pulsed electromagnetic fields. Occupational accidents related to power lines explain some of the excess of deaths from accidents and violence. Some association was found between magnetic fields and leukaemia, brain cancer, and suicide, between electric fields and brain cancer and suicide, and between pulsed electromagnetic fields and lung cancer, but these were not significant. CONCLUSION: These results are broadly reassuring that major causes of death are not strongly associated with exposure to electric and magnetic fields, but small numbers and approximate exposure assessments preclude the denial of any risk, in particular if it were to affect a rare cause of death.

Comparing standardized mortality ratios.

It is well known that the ratio of two standardized mortality ratios (SMRs) is not in general an unbiased estimate of the underlying within-stratum ratio of rates of one subcohort relative to another. It is also established, although less well known, that a sufficient condition for unbiasedness is that the underlying stratum-specific rates in each of the two subcohorts be proportional to the reference population. Further, the ratio of SMRs is more precise than the wholly internal (Poisson regression) estimate of rate ratio. In data that are compatible with the proportionality assumption, use of the ratio of SMRs thus buys precision at the cost of increased vulnerability to bias. To further elucidate choice between methods, we derive expressions for the asymptotic precision of each. These show that improved precision of ratio of SMRs depends on the extent to which the distribution of expected deaths over strata is different in the two cohorts, or equivalently, on the variance over strata of the proportion of expected deaths falling in the first cohort. The results are illustrated by hypothetic examples.