Discussion

Mortality is higher in men than women across Europe, but there is considerable variation in the magnitude of this gap (from an ‘excess’ of 188 deaths per 100 000 per year in Iceland to 942 per 100 000 per year in Ukraine). Smoking-related deaths accounted for around 40% to 60% of the gender gap, while alcohol-related mortality typically accounted for around 20% of the gender gap. The range in the contribution of smoking-related deaths reflects gender differences in the uptake of smoking by gender in earlier decades.7

The strengths of this analysis are the use of best available data from the WHO (who endeavour to quality assure the data) and the inclusion of most large European countries (with the exception of Turkey). There are, however, also some limitations to consider. First, alcohol and tobacco use contribute to some shared causes of mortality and morbidity and so any division into alcohol-related and smoking-related causes will underestimate the scope of influence of each on mortality. For example, liver cancer has been linked with excessive alcohol consumption for some time, but more recently smoking has been also shown to be associated with this malignancy. Similarly, tobacco and alcohol consumption contribute to the development of cancers of the aerodigestive tract.8 This is the rationale behind ‘Peto's method’ of benchmarking smoking-related deaths within each country against lung cancer deaths,9 as one method of estimating smoking related deaths. The WHO's definitions acknowledge that smoking and alcohol contribute to some causes of death (oesophageal and throat cancer) but not others. Unfortunately, the cause-specific data required for applying any alternative definitions were not available from the WHO-HFA database and so we were constrained by the WHO's definitions. A second limitation is the potential for differential coding practices to bias the cause-specific death rates between countries despite the use of the International Classification of Diseases system. Third, the WHO-HFA database only provides these data by gender for all ages together so we were unable to examine relationships in specific age groups.

It is no surprise that two of the most important health behaviours, smoking and hazardous drinking continue to account for substantial proportions of the gender gap in mortality because health behaviours have long been a powerful way of portraying gendered identities.10–15 For example, it has been suggested that cultural portrayals of drinking keep shifting to maintain a gendered distinction in drinking behaviours, so that as men and women both modify their drinking behaviours, considerable effort is devoted to constructing men's drinking in different ways to women's drinking.16 The balance of the contribution of smoking-related and alcohol-related deaths and the degree of consistency of the pattern across Europe is perhaps more of a surprise given the complexity of associations between gender, smoking and drinking over the preceding decades, the long lag time between exposure to smoking and disease (estimated to be between 20 and 35 years for lung cancer,17 and 30 to 40 years for other cancers such as colorectal cancer18), the considerable variation in the magnitude of the gender gap in all-cause mortality and the different stages in the smoking epidemic that countries of Northern, Western, Southern and Eastern Europe have reached. What is clear is that smoking accounts for a substantial part of the gender difference in mortality in contemporary Europe. The importance of health behaviours (and in particular smoking) in accounting for a large proportion of the gender gap in mortality is often lost in discussion of gender and health. For example, it is often suggested that other factors (such as differences in consultation for illness) account for much of this difference (see for example White and Witty19) despite a paucity of robust evidence.20

The continuing uptake of smoking among a significant minority of young people, and increases in detrimental patterns of alcohol consumption, point to the ongoing need for public health measures to reduce health-damaging behaviours. However, the continuing links between smoking and drinking and cultural constructions of gender demonstrate that action to reduce smoking and harmful drinking cannot be tackled at an individualistic level alone. These behaviours are culturally bound and these cultural constructions of behaviours are partially shaped by and exploited by the alcohol and tobacco industries (see for example Amos and Haglund, and Toll and Ling11 21), in addition to people's structural opportunities and constraints.22

Profound changes in the population level of smoking and in the magnitude of the gender gap in smoking should contribute to smaller gender differences in mortality in coming decades, beginning first in countries in Western and Northern Europe that reached the fourth stage of the tobacco epidemic first.7 However, the extent to which this is realised will depend on the ways in which other health risk behaviours are patterned by gender.