Mortality among male and female smokers in Sweden: a 33 year follow up.(Statistical Data Included)

Abstract

Study objective-It is still unclear if men and women are equally susceptible to the hazards of tobacco smoking. The objective of this study was to examine smoking related mortality among men and women. Design-In 1963 a questionnaire concerning tobacco smoking habits was sent out to a random sample from the 1960 Swedish census population. Date and cause of death have been collected for the deceased in the cohort through 1996.

Setting-Sweden.

Participants-The survey included a total of 27 841 men and 28 089 women, aged 18-69 years. The response rate was 93.1% among the men and 95.4% among the women.

Main results-After adjustment for age and place of residence positive associations were found between cigarette smoking and mortality from ischaemic heart disease, aortic aneurysm, bronchitis and emphysema, cancer of the lung, upper aerodigestive sites, bladder, pancreas in both men and women, but not from cerebrovascular disease. When the effect of amount of the cigarette consumption was considered, female smokers displayed, for example, slightly higher relative death rates from ischaemic heart disease. However, no statistically significant gender differential in relative mortality rates was observed for any of the studied diseases.

Conclusions-Women and men in this Swedish cohort seem equally susceptible to the hazards of smoking, when the gender differential in smoking characteristics is accounted for. Although the cohort under study is large, there were few female smokers in the high consuming categories and the relative risk estimates are therefore accompanied by wide confidence intervals in these categories.

(J Epidemiol Community Health 2001 ;55:825-830)

Historically female smokers have been at a lower mortality risk than male smokers. [1-5] It has been argued that these results are reflections of the fact that women have been smoking less heavily and for a shorter period of time than men. [5,6] Over the past couple of decades the number of women who smoke have increased dramatically. Accordingly, more recent studies with longer follow up have presented mortality rates of women more similar to those of men and some studies have even suggested that women may be more sensitive to some of the adverse health effects of smoking. [7-10]

Studies with a representative selection of national populations are scarce. Furthermore, additional studies of the long term effects of smoking among women have been recommended to further examine possible differences in susceptibility and aetiology between men and women. [10] The objective of this study was to contribute to this knowledge by comparing the relations between smoking and selected smoking related diseases among men and women in a large prospective cohort created from a random sample of the 1960 Swedish census population.

Methods

This study is based on the Swedish 1963 smoking habit survey, earlier described in detail. [2 11] Briefly, the smoking habit survey is based on a age stratified sample of 55 930 people (27 841 men and 28 089 women), aged 18-69 years, selected randomly from a register of the 1960 Swedish census population. Through three questionnaires and additional follow up telephone interviews a total response rate of 94.2% (95.4% for women and 93.1% for men) was achieved. The survey consisted of, for example, questions about present and former (previous nine years) consumption of different types of tobacco, duration of smoking, age when taking up the habit, manner of smoking, and quantity. The questions about quantity smoked were answered by fixed response alternatives, 1-3, 4-7, 8-15, 16-25 and more than 25 cigarettes smoked per day, respectively. In the analysis the categories 1-3 and 4-7 were merged because of because of small numbers.

Information on mortality was obtained from the Cause of Death Registry of the Swedish Central Bureau of Statistics, which collects and codes all death certificates in Sweden. By using the 10 digit civic registration number, unique to every person living in Sweden, record linkages have been possible between the smoking habit survey and the Cause of Death Registry. All deaths through 1996 have been collected. In the analysis the first five years of follow up (up to 1968) were excluded to reduce the risk of inverted causality--that is, that the individual changes the smoking habit because of the lethal disease. Since 1987 underlying and contributing causes of death in the Cause of Death Registry have been coded according to the 9th revision of the international Classification of Diseases (ICD) and between 1969 and 1987 according to the 8th revision. In the analysis all classifications have been translated to the 8th revision according to a translation list constructed by the Central Bureau of Statistics. [12] T he analysis of the cause specific mortality has been restricted to the underlying cause of death.

Cox proportional hazards regression model [13] was used to calculate cause specific relative mortality rates for different groups defined by smoking habits (that is, amount smoked). Using a stratified Cox model the rates were standardised both by age in 1963 in seven age groups (18-29, 30-39, 40-49, 50-54, 55-59, 60-64, 65-69 years), and place of residence in three categories: major towns (Stockholm, Gothenburg, Malmo), other towns, and rural districts. This stratified model is less restricted as it permits a different (non-proportional) baseline hazards function in each of the subgroups defined. [13] Dose-response tests were calculated for trends in the rates where never smokers, smokers of 1-7, 8-15, 16-25, and more than 25 cigarettes per day were coded as 1, 2, 3, 4, and 5 respectively (former smokers are not included). Smokers of cigarettes only are included in this study because less than 1% of the women smoked other tobacco products compared with 36% of the men. The male smokers of other tobacco produc ts, mostly pipe smokers, have previously been analysed as a separate group. [14] The reference group consists of those who reported that they had never smoked any tobacco products regularly. Finally, when noncodable answers were excluded, a total of 16 458 men and 25 086 women were included in the analysis and a total of 877 635 person years were accumulated (547 889 among the women and 329 736 among the men). The women started smoking at a later age than the men in the cohort, with an average age of debut of 22.9 and 18.6 years respectively. Furthermore, 10% of the female and 6% of the male smokers were non-inhalers. A further description of the cohort under study is given in table 1.

As the female smokers have started smoking at an older age than the male smokers have, we divided the cohort according to age at smoking debut. We divided the cohort into those who started smoking before 25 years of age and those who were 25 years or older and analysed the groups separately.

We have chosen to present the relative death rates for selected causes of death considered to be closely related to smoking previously [3-5] [15-17] or otherwise viewed as interesting, for example, alcohol related.

Results

Increased relative risks, with 95% confidence intervals (CI) not covering unity, among all current smokers (with never smokers as reference) (table 2) as well as significant dose-response trends with amount smoked (table 3) were seen for ischaemic heart disease, aortic aneurysm, bronchitis and emphysema, peptic ulcer, cancer of the trachea, bronchus and lung, accidents, suicide and violence, and all causes of death for both male and female smokers. In addition both male and female current smokers displayed significantly increased risks of cancer of the pancreas (table 2), and the male smokers also showed a …