healthscotland.com

Passive smoking and associated causes of death in adults in Scotland

Appendix

Sensitivity analyses

A number of assumptions have been made in order to estimate the impact of exposure to ETS on the number of associated causes of death currently occurring in Scotland among adults. This appendix examines the impact of using alternative assumptions on the total numbers of deaths which can be considered to be associated with passive smoking. Three alternatives are considered.

• Using smoking prevalence data derived from the Scottish Heart Health Study conducted between 1984 and 1986.
• Using relative risks for active smoking and lung cancer which are an average of those derived from the Renfrew/Paisley study and the UK Doctors’ Study.
• Allowing the vascular causes of death to have a shorter latency in terms of disease progression to that assumed for the respiratory causes.

Different smoking prevalence data

The distribution of active cigarette smoking has been taken from the Renfrew/Paisley survey. Whilst this is appropriate in that it represents a time period which is relevant to the development of the chronic diseases studied and it is a general population survey, it may not represent the pattern of active smoking in Scotland at that time. Judging by the lung cancer SMR for Renfrew local government district for the second half of the 1980s, the mortality rate for lung cancer was about 12% higher than that for Scotland as a whole. Similar excesses are apparent for IHD and stroke.

The calculations for the four causes of death have been redone using the distribution of cigarette smoking as measured by the Scottish Heart Health Survey. The main differences as compared to the Renfrew/Paisley survey were:

• a greater percentage of lifelong non-smokers among men (27% vs 19%)
• a lower percentage of lifelong non-smokers among women (42% vs 46%)
• a higher percentage of ex-smokers among men (34% vs 24%)
• a higher percentage of ex-smokers among women (20% vs 8%)

Table 1 Estimated deaths using alternative smoking prevalence data

*Among lifelong non-smokers; **(male deaths/female deaths)

This calculation produced an estimate of 945 deaths attributed to passive smoking among lifelong non-smokers (compared with 865 deaths calculated from Renfrew/Paisley estimates, an increase of about 9%).

Different relative risks for active smoking and lung cancer

The relative risks for lung cancer in the West of Scotland are low compared to most other studies of lung cancer and smoking. This is because the rate of lung cancer in lifelong non-smokers is higher than in most other studies. It is not clear exactly why this is so but asbestos exposure among shipyard and other workers is a strong factor in the West of Scotland and atmospheric pollution appears to influence lung cancer rates particularly in the Glasgow area. As an alternative analysis, the relative risks are doubled for lung cancer (i.e. the rate of lung cancer among lifelong non-smokers for Scotland as a whole is assumed to be about half that seen in the Renfrew/Paisley survey). A 50% increase is also incorporated into the calculation for respiratory disease, but no alteration is made to the risks for IHD and stroke. Cigarette smoking data from the SHHS is used.

Table 2 Estimated deaths using alternative relative risk for active smoking and lung cancer

*Among lifelong non-smokers; **(male deaths/female deaths)

This produces an estimate of 903 deaths, compared with 945 deaths under the earlier assumptions and 865 deaths in the original calculation.

Allowing different latencies for vascular and respiratory causes of death

A third possibility is to assume that the latencies for lung cancer and respiratory disease are different from deaths due to vascular causes (IHD, stroke). Using direct evidence on length of latencies and indirect evidence on the length of time it takes for a the risk for a ‘quitter’ to return to that of a lifelong non-smoker, it can be inferred that the latency for lung cancer and respiratory disease is of the order of 20–40 years, but that for the vascular causes is much shorter (5–10 years). Applying these assumptions to the Scottish situation leads to an estimate of 925 deaths associated with exposure to second-hand smoke (compared to previous estimates of 865, 945 and 903 deaths).

Table 3 Estimated deaths assuming different latencies for vascular and respiratory deaths

*Among lifelong non-smokers; **(male deaths/female deaths)