METHODS AND RESULTS

Daily deaths 1998–2000 from the Office of National Statistics, for men and women in England and Wales aged 65–74 years and 50–59 years, were extracted by class (when recorded), as 1 (professional), 2 (managerial and technical), 3N (non-manual skilled), 3M (manual skilled), 4 (partly skilled), or 5 (unskilled), with between 896 and 66 477 deaths in each age, sex and class group. Married women with only domestic occupations at home, assigned their husband’s class, are classified as housewives. “Central England” temperatures, from the Meteorological Office, were daily means of Squire’s Gate Lancashire, Manchester Airport, Malvern, Rothamstead.

Daily mortality at 18°C was estimated by generalised linear modelling with Poisson distribution, of mortality on temperature, over the linear range 0°C to 18°C. The regression coefficient of mortality on temperature and its standard error, expressed as percentage of deaths at 18°C, gave cold related mortality and standard error for each group. A second explanatory variable, mean influenza deaths 10 days before to 10 days after, allowed for influenza. Temperature was lagged three days to give the closest relation to all cause mortality.⁴ Statistical comparisons were assessed by the t test. To minimise problems from multiple comparisons, these were made only between and within classes 5 and 1.

Cold related mortality in the retired (65–74) age group was generally higher in men of class 5 (unskilled) than class 1 (professional), or other classes, with little difference between men, and women or housewives, of any class (table 1). In the working age group (50–59), women in class 5 had significantly higher cold related mortality than those in class 1, but in men in class 5 cold related mortality was on average lower than in men of any other class. It was also significantly lower in class 5 among men than women, or housewives who represented 62% of these women (table 1), both in direct comparison and in relation to comparisons of men and women in class 1.

CONCLUSION

In men of working age (50–59) cold related mortality was low in class 5 compared with any other class, though it was high in class 5 men of the retired age group (65–74). In the class 5, but not class 1, men of working age cold related mortality was also low compared with women or housewives of the same class and age group. This implies a beneficial effect of work related factors in men of class 5 but not class 1, independent of home environment and income. The simplest explanation is that internal heat production from manual work protected class 5 men against daytime cold stress.

International surveys also point to an important role of out of home factors. Cold related mortality in eight regions of Europe varied inversely with the effectiveness of measures taken by the people in the different regions to avoid both outdoor and indoor cold.⁵ Home heating and insulation has received much attention in Britain, but less attention has been paid to out of home factors. We suggest increased emphasis on reducing exposure to outdoor cold stress, in campaigns to reduce winter mortality.