Deaths caused by secondhand smoke: estimates are consistent

In 2001 Woodward and Laugesen estimated the number of deaths caused by secondhand cigarette smoke in New Zealand, using an indirect method based on studies of disease specific mortality risks.¹ Most of the relative risks used in this estimation were taken from studies conducted in other countries. We now have an opportunity to check the accuracy of this estimate using a more direct method based on all cause mortality risks taken from a recent New Zealand study.²

Hill et al compared mortality among New Zealand never smokers living with cigarette smokers with that of never smokers in non-smoking households.² They report adjusted mortality rate ratios for 45–74 year olds from two periods: 1981–4 and 1996–9. For men the ratios were 1.17 (95% confidence interval (CI) 1.05 to 1.30) and 1.16 (95% CI 1.04 to 1.30) respectively; for women 1.06 (95% CI 0.97 to 1.16) and 1.28 (95% CI 1.16 to 1.42). Assuming a rate ratio of 1.15 constant over age and sex, and applying this to 1996 census counts of never smokers living in households with at least one smoker (approximately 55 340 adults), we estimate that passive smoking accounts for 73.5 deaths per year in the 45–74 year age group.

We have repeated the calculations conducted by Woodward and Laugesen, restricting the analysis to deaths caused by exposures in the home, and including only the age group 45–74. The base is again the 1996 New Zealand census population. The results are 2.7 lung cancer, 57.9 heart disease, and 46.3 stroke deaths per year (106.9 in total). This estimate includes never-smokers and ex-smokers (compared with the study by Hill et al, which was restricted to lifetime never smokers²). In their 2001 paper, Woodward and Laugesen undertook sensitivity analysis showing that the overall number of deaths was reduced by 45% if ex-smokers were excluded.¹ In this instance, 106.9 would come down to 58.8 deaths per year. Note that this does not include deaths that may be caused by other passive smoking related conditions (such as chronic lung disease or other cancers). Thus, 58.8 deaths per year is in close agreement with the estimated 73.5 deaths based on the study by Hill et al.²

Both estimates of the number of deaths caused by passive smoking have their weaknesses—for example, Hill et al had to assume that living with a smoker was a reliable measure of exposure to second hand smoke.² As a result, these calculations should be viewed as a guide to, not a precise measure of, the burden of disease. But it is encouraging that two different methods of estimating attributable deaths in the same population produce broadly consistent answers. It should add to the confidence with which policymakers, health educators, and others use estimates of the passive smoking burden, while conscious of the significant uncertainties that accompany all calculations of this kind.