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 significant dose-response relation with number of cigarettes smoked per day (table 3). Furthermore male current smokers displayed significantly increased risks of cancer of the upper aerodigestive sites, liver and biliary passages, and bladder (table 2). Male smokers also displayed a significantly positive dose-response relation with number of cigarettes smoked for cancer of the upper aerodigestive sites, cancer of the liver and cirrhosis of the liver (table 3). Notably no statistically significantly increased relative risk of all cerebrovascular disease or any specific cerebrovascular diagnosis was observed at any level of exposure for either men or women.

No major differences were found between male and female smokers in the relative risks of any of the studied diseases other than tumour diseases, where the male current smokers displayed consistently higher relative risk estimates than the female current smokers did (table 2). There were, however, no significant differences in relative death rates between male and female smokers at different levels of exposure (table3).

When analysing the follow up period in two parts there was a tendency towards lower relative mortality from ischaemic heart disease and all cause mortality in the later part of the follow up period, 1982–96, than the earlier, 1969–82, for male smokers (table 4). There were no significant differences in relative death rates between male and female smokers, at any level of exposure, in either of the two periods of follow up.

After dividing the cohort according to age at smoking debut it was shown that younger age at debut seems to increase the risks of all cause mortality among both male and female smokers. The relative death rates of ischaemic heart disease and all cause mortality were somewhat higher in female smokers. However, the confidence intervals are wide and the differences were not significant (table 5).

Excluding the non-inhalers did not change the results in any substantial way (table 6). Though women and men display a different pattern of inhalation, self described degree of inhalation proved to have little impact on the relative risk estimates of the studied diseases and an even smaller effect on the male/female risk ratios.

Discussion

Swedish men began smoking to a larger extent around the second world war and women 20 years later. The proportion of daily smokers among men in Sweden peaked in the 1960s, when nearly half of the male population were daily smokers, and the rates then decreased to remain stable for the past 15–20 years.11 18 At the same time the rates for women have been steadily increasing, particularly in the younger age groups. This has resulted in a larger proportion of daily smokers among women than men in the younger age categories. Today women and men in Sweden, overall, smoke to the same extent and the proportion of daily smokers is between 20 and 25%.19

It should be noted that smoking is not the sole risk factor of the diseases in this study. Potential confounders apart from those explicitly assessed and accounted for (place of residence and age) are, for example, diet, stress, physical activity, alcohol consumption, socioeconomic status and occupational exposures. The socioeconomic gradient in smoking does not differ between men and women in this cohort.11

Alcohol consumption has been associated with increased risks of, for example, cancer of the upper digestive tract,20cerebrovascular disease and ischaemic heart disease.21Differences in alcohol consumption between different groups may, in Sweden, to some extent be indicated by the incidence of the largely alcohol related diseases cancer of the liver20 and cirrhosis of the liver.22 The relative risks of accidents, suicide and violence was significantly increased among both men and women and also these causes of death may be associated with certain lifestyles. The relative risks of both cirrhosis and cancer of the liver is clearly higher among the men than the women in this study and this could partly be a result of a higher alcohol consumption than expected among the men in the sample. It is therefore possible that the effects of smoking are somewhat less pronounced for certain causes than indicated in this study, especially for men.

It should be pointed out that we have not been able to adjust for any changes in smoking habits after 1963. Any gender differences in smoking cessation are probably small23 and have only limited effects on the results. During the 1960s and 70s the proportion of female smokers increased, especially women in the younger age categories started smoking. It is therefore possible that a higher proportion of the younger women than the younger men took up smoking after the smoking habits were registered. However, in the study only 1.3% of the deaths occurred among women between 18 and 29 years old in 1963 and the potential effects on results are probably small. It is also rare to take up smoking at an older age and the possible bias is therefore negligible.24 In 1963 smoking was socially accepted and bias attributable to underreporting is probably small. Notably only few women smoked more than 15 cigarettes per day in 1963. This gives rise to somewhat more unstable relative risk estimates in these categories and the statistical power to detect differences in dose-response gradient between men and women is therefore limited.

Consistently with previous studies clear associations between smoking and the risk of ischaemic heart disease, aortic aneurysm, peptic ulcer, bronchitis and emphysema, were observed for both men and women.3-5 14-17 In this study men and women seem to be at equal risks for most of these smoking related hazards as similar associations were found for both genders at different levels of tobacco exposure from cigarettes. However, the female smokers consistently displayed higher relative death rates from ischaemic heart disease than the male smokers did, though not significantly. This gender difference has also previously been reported.10 In this study no association was found between smoking and cerebrovascular disease at any level of tobacco exposure for either men or women. In some previous research cerebrovascular disease has been considered to be partly caused by smoking.16 17 21 There are however some other studies that have failed to show any association between smoking and some forms of cerebrovascular diseases25 26 and the relation is still a matter of debate.27 It should be noted that cerebrovascular disease is a broad category of diseases with a complex aetiology.5

Statistically significant relations between cigarette smoking and cancers of the lung, upper aerodigestive sites, bladder and pancreas have consistently been observed among both men and women.3-6 10 15 17 18 These findings are in line with the results from this study.

When all current smokers were compared with never smokers, not adjusted for amount, the male smokers generally displayed higher relative risks of cancer than the female smokers did. This gender differential was reduced when the current smokers were divided into groups by amount smoked. The fact that female current smokers are at a lower relative risk than male current smokers seems to be a result largely of the lack of female smokers in the high consuming categories. It seems as if different levels of exposure largely can explain the difference between the genders in mortality rates from tumour diseases. This is not unexpected as men took up smoking earlier than women, and men have also generally been smoking more heavily than women.21

It should be noted that a similar relative risk in women and men related to smoking does not mean that the habit has the same public health consequences. The baseline mortality rates are lower in women than in men, for, for example, ischaemic heart disease, which leads to lower absolute risk differences if the relative risks are equal.

In conclusion, this representative study of Swedish men and women shows that women are generally as susceptible to the negative consequences of smoking as men are. It should be noted that although the cohort is large the number of women smoking more than 16 cigarettes per day is limited. The relative risk estimates for the female smoker in the high consuming categories should therefore be interpreted with caution. As men are smoking less at the same time as there has been an increase in the number of female smokers it can be expected that male smoking related mortality will decrease at the same time as the negative effects of smoking among women will largely appear in the future.