healthscotland.com

Passive smoking and associated causes of death in adults in Scotland

Contents:	Study Aim Executive Summary Study Aim Methods Analysis Methods Analysis Discussion Key references Discussion Key references Appendix Executive Summary Appendix

Analysis

Measurement of second-hand smoke exposure

It is extremely difficult to measure the amount of second-hand smoke experienced by individuals and impossible to produce a direct measure of an individual's cumulative exposure, the more relevant aetiological factor. Early studies of the impact of passive smoking used as a pragmatic indicator, marriage to a smoker, and most of these studies concentrated on lifelong non-smoking wives as the index group. This turned out to be a sensible choice as when studies of the levels of urinary and serum cotinine were conducted among passive smokers (non-smoking women married to smoking men) and controls (non-smoking women married to non-smoking men), there was a substantial difference in mean levels. This meant that passive smokers as a group experienced higher cotinine levels than controls which was the important issue, even though a small number of controls had higher individual levels than a small number of passive smokers.

The use of such a definition was supported by questionnaire evidence that passive smokers 'at home' were also more likely to accept passive exposure at work or in other social circumstances. Later studies concentrated on exposures both at home, at work and in other circumstances and consistently observed increased risks in relation to exposures in the different settings. More recently, the British Regional Heart Study has reported risks of coronary heart disease and stroke in relation to serum cotinine measures directly among non-smokers and has observed markedly increased risks for the former disease. Thus, we can conclude that the use of definitions such as 'married to a smoker' is valid as an indicator of increased exposure. Indeed, they may well be conservative in trying to estimate the 'true' risk of passive exposure.

Lung cancer

It is widely accepted that passive exposure to cigarette smoke can cause lung cancer among those who have never smoked. ETS is classified as a Class 1 (strong epidemiological and biological human and animal evidence) carcinogen by IARC, the International Agency for Research on Cancer.

How risky is passive smoking in relation to lung cancer?

The interpretation of the evidence linking exposure to second-hand smoke with lung cancer is constrained by the imprecision of risk estimates. This is especially true for Scotland as there are few studies locally which have attempted to make such an estimate and all have been based on very small numbers of passively exposed cases. The most recent international analysis which is based on large numbers of cases, includes adjustment for potential confounders and exposure misclassification, has been produced by the International Agency for Research on Cancer (IARC) in the International Journal of Cancer in 2004 (Brennan, 2004). This concludes that there is a clear dose-response relationship consistent with a causal association between exposure to second-hand smoke and the development of lung cancer among never smokers. The risk estimates are set at 16 to 18% for average exposure and up to 32% for long-term exposure from combined spousal, workplace and social sources.

How do these estimates relate to the Scottish situation?

Two relevant factors pertain. First, the degree of passive exposure will almost certainly be related to the extent of active smoking. As Scotland has been known historically as the lung cancer capital of the world, it would seem reasonable to infer that the level of active smoking has been markedly higher than other places in the world (particularly those taking part in the above-mentioned international study). Therefore, the level of passive exposure locally will have been higher and it would seem reasonable to increase the magnitude of the estimates quoted. Secondly, all studies into the effects of passive smoking tend to use as a control group 'never smokers' married to 'never smokers' (or some equivalent for workplace exposure). However, studies of urinary cotinine levels in this type of control group have found that over 60% have detectable levels - in other words the control groups themselves are substantially contaminated. Again, this figure of 60% may be a conservative estimate as far as Scotland is concerned.

It is impossible to quantify with any precision by how much the Scottish estimate should be increased and in this document we are arbitrarily setting the average exposure risk at 25%.

What proportion of the non-smoking population is exposed to second-hand smoke?

The objective in this part of the report is to try and estimate how much past exposure impacts on current lung cancer levels in Scotland. As there is a substantial lead-time between exposure and development of lung cancer - a minimum of 20 years and up to 40 years -we are consequently looking for smoking data from some 30 years ago. It is extremely fortuitous that data from the Renfrew/Paisley survey conducted by Professor Victor Hawthorne is available for that time period. As this was a population-based study and co-habitees (mostly husbands and wives) took part in large numbers it also provides valuable data on spousal exposure. This suggested that only about 19% of men were lifelong non-smokers and about one third of these (6%) were regularly exposed to second-hand smoke at home. Some 46% of women were lifelong non-smokers and about three-quarters of these (32%) were passively exposed at home. The majority of women were not working at that time and their total passive exposure derived mainly from the home situation. Men however derived substantial exposures at work although those exposed at home were more likely to be in contact with second-hand smoke at work. The revised figures for regular exposure from either source have been taken as 12% for men and 33% for women.

The total number of lung cancer deaths occurring each year in Scotland is approximately 2,300 among men and 1,700 in women. These deaths can be distributed among the different smoking categories as shown in Table 1.

Table 1 Total number of lung cancer deaths in Scotland by sex and smoking category

Smoking category	Relative risk*	Men		Women
		% in each smoking category*	Number of lung cancer deaths	% in each smoking category*	Number of lung cancer deaths
Never	1	19	82	46	204
1-14 cigs/day	5	12	260	19	422
15-24 cigs/day	8	29	1005	24	852
25+ cigs/day	10	16	693	3	133
Ex-smokers	2.5	24	260	8	89
Total deaths**			2300		1700

* Relative risk estimates and smoking prevalence figures derived from the Renfrew/Paisley survey

**This includes 1867 male and 1256 female lung cancer deaths attributable to active smoking

Thus, we would estimate that the annual number of lung cancer deaths occurring among lifelong non-smokers in Scotland totals 82 for men and 204 for women. The number of lung cancer deaths in Scotland annually caused by 'active' smoking is 81% (male) and 74% (female). This amounts to 1867 in men and 1256 among women, a total of over 3,100 lung cancer deaths.

The distribution of the 82 (male) and 204 (female) lung cancer deaths in lifelong non-smokers among those who and have and have not been passively exposed is shown in Table 2.

Table 2 Total number of lung cancer deaths in lifelong non-smokers by ETS exposure category

ETS exposure	Relative risk of lung cancer	Men		Women
		% of never smokers in each group	Number of lung cancer deaths	% of never smokers in each group	Number of lung cancer deaths
No	1	33	23	25	43
Yes	1.25	67	59	75	161

This implies that 220 (59+161) lung cancer deaths occur among lifelong non-smokers who are passively exposed. However, not all of these deaths can be directly attributed to passive exposure. The attributable risk figures for passive smoking among lifelong non-smokers and lung cancer are 14% (12 men) and 16% (32 women). This amounts to 44 lung cancer deaths per year for both sexes combined.

This figure may seem very low. Paradoxically, this reflects the fact that the 'active' smoking prevalence for Scotland has historically been so high - up to 80% ever-smoked among adult men and 55% among adult women. Thus the contribution of passive as opposed to active exposure to cigarette smoke is minimal in that there have been few men and a minority of women who experience passive exposure only. There is some evidence that ex-smokers have an increased risk of lung cancer due to subsequent passive exposure but no evidence that active smokers have increased risks due to additional passive exposure.

Thus the bottom line for Scotland at present is that of the 4,000 lung cancer deaths occurring annually, over 3,100 are caused by active smoking and about 50 are caused by passive exposure.

Ischaemic heart disease

Numerous studies have shown that individuals who smoke are at an increased risk of developing and dying from IHD. What is less understood is the mechanism by which this occurs. Whether smoking directly causes IHD or exacerbates other mechanisms is not clear. In contrast to lung cancer and respiratory disease, the dose-response relationship between amount smoked among active smokers and IHD risk is weak. This would seem to suggest that cigarette smoking per se is the critical risk factor and not the amount smoked. This has led to considerable controversy about the magnitude of IHD risk associated with passive exposure. Some argue that if the ratio of dose of passive:active smoking seen in lung cancer risk (between 1 in 50 to 1 in 100) is transferred to IHD risk, then this latter risk associated with passive exposure is negligible. Alternatively, as smoking or not smoking is more critical than amount smoked for active smokers, then passive exposure would represent a more substantial risk than simply applying a direct calibration of dose received. The epidemiological evidence would seem to support this latter argument. Even so, the estimates derived to date for IHD risk associated with passive smoking do seem to be on the high side and it is possible that some, but not all, of this excess risk is due to misclassification of passive smoking status or unexplained confounding, a problem which led to some of the early estimates for lung cancer being on the high side.

Taking the estimates suggested by the review by Law, and using the same method as above for lung cancer, we estimate there are 758 male IHD deaths annually in Scotland and 1814 IHD female deaths among lifelong non-smokers in Scotland (Table 3).

Table 3 Total number of ischaemic heart disease deaths in Scotland by sex and smoking category

Smoking category	Relative risk of IHD	Men		Women
		% in each smoking category	Number of IHD deaths	% in each smoking category	Number of IHD deaths
Never	1	19	758	46	1814
1-14 cigs/day	1.62	12	776	19	1214
15-24 cigs/day	1.95	29	2256	24	1845
25+ cigs/day	1.81	16	1156	3	214
Ex-smokers	1.31	24	1254	8	413
Total deaths*			6200		5500

* This includes 2210 male and 1557 female IHD deaths attributable to active smoking

Table 4 below shows the distribution of the 758 male IHD deaths and 1814 female IHD deaths in lifelong non-smokers between those who have and those who have not been passively exposed to ETS.

From the table it can be see that 544 IHD deaths occur amongst male lifelong non-smokers and 1432 IHD deaths amongst female lifelong non-smokers who have been exposed to ETS. Of these 109 male IHD cases and 286 IHD cases can be attributed to passive smoking.

Table 4 Total number of ihd deaths in lifelong non-smokers by ets exposure category

ETS exposure	Relative risk of IHD	Men		Women
		% of never smokers in each group	Number of IHD deaths	% of never smokers in each group	Number of IHD deaths
No	1	33	214	25	382
Yes	1.25	67	544	75	1432

Stroke

Very little information exists about the risk of stroke in relation to ETS. Studies in New Zealand and the USA have produced positive associations, although the magnitudes have varied substantially. The former estimated the excess risk at 82% and the latter at 29%. When compared with the risk observed in studies of active smokers generally, the former figure appears high. I have therefore taken the lower figure as the 'true' estimate as it appears more plausible in relation to active smoking and stroke and because it is of the same order of magnitude as that observed for IHD, as IHD and stroke share similar levels of risk for active smoking.

Table 5 Total number of stroke deaths in Scotland by sex and smoking category

Smoking category	Relative risk of Stroke	Men		Women
		% in each smoking category	Number of Stroke deaths	% in each smoking category	Number of Stroke deaths
Never	1	19	348	46	1554
1-14 cigs/day	1.41	12	310	19	905
15-24 cigs/day	1.65	29	876	24	1338
25+ cigs/day	1.78	16	522	3	180
Ex-smokers	1.01	24	444	8	273
Total deaths*			2500		4250

*This including 668 male and 872 female stroke deaths attributable to active smoking.

Table 6 shows the distribution of the 348 male stroke deaths and 1554 female stroke deaths in lifelong non-smokers between those who have and those who have not been passively exposed to ETS. In total there are 252 deaths amongst male and 1235 deaths amongst female lifelong non-smokers exposed to ETS due to stroke. Of these 57 male stroke deaths and 278 female stroke deaths can be attributed to passive smoking.

However, it should be noted that these estimates are based on a limited number of studies and further epidemiological studies are required to refine them.

Table 6 Total number of stroke deaths in lifelong non-smokers by ets exposure category

ETS Exposure	Relative risk of stroke	Men		Women
		% of never smokers in each group	Number of stroke deaths	% of never smokers in each group	Number of stroke deaths
No	1	33	96	25	319
Yes	1.29	67	252	75	1235

Respiratory disease (excluding lung cancer)

Given that the risk of respiratory disease is increased among active smokers, and that there is good evidence that respiratory function, particularly FEV 1, is impaired among lifelong non-smokers who are exposed to second-hand smoke, it seems entirely plausible that respiratory disease should be included among the diseases affected by passive smoking. Most studies have tended to concentrate on the effects of passive smoking on the respiratory development of children, rather than investigate the consequences among adults. There are few estimates of the likely risk available. As an estimate I have chosen to assume that the shape of the exposure-response equation is more like that observed for lung cancer than that for the cardiovascular system. As the risks for respiratory disease among active smokers are lower than the corresponding risks for lung cancer by a factor of about 2.5, I have assumed a risk of 1.10 for respiratory disease among passive smokers (c.f. 1.25 for lung cancer among passive smokers).

Table 7 Total number of respiratory disease deaths in Scotland by sex and smoking category

Smoking category	Relative risk	Men		Women
		% in each smoking category	Number of respiratory	% in each smoking category	Number of respiratory deaths
Never	1	19	297	46	1030
1-14 cigs/day	1.85	12	347	19	787
15-24 cigs/day	2.28	29	1032	24	1225
25+ cigs/day	3.37	16	842	3	226
Ex-smokers	1.29	24	483	8	231
Total deaths*			3000		3500

*This includes 1439 male and 1261 female respiratory deaths attributable to active smoking

Table 8 shows the distribution of the 297 male respiratory disease deaths and the 1030 female respiratory disease deaths between lifelong non-smokers who have and who have not been exposed to ETS. In total there are 205 deaths amongst male and 791 deaths amongst female lifelong non-smokers exposed to ETS due to respiratory disease. Of these, 19 male respiratory disease deaths and 72 female respiratory disease deaths can be attributed to passive smoking.

Table 8 Total number of respiratory disease deaths in lifelong non-smokers by ets exposure category

Passive exposure	Relative risk	Men		Women
		% of never smokers in each group	Number of respiratory deaths	% of never smokers in each group	Number of respiratory deaths
No	1	33	92	25	240
Yes	1.10	67	205	75	791

Table 9 summarises the estimated deaths from all four diseases attributable to active and passive smoking. In total, it is estimated that there are about 865 deaths per annum attributable to passive smoking

Table 9 Summary of number of deaths (four causes) attributable to active and passive smoking

Cause of death	Deaths
	Total	Attributable to active smoking	Attributable to passive smoking*
Lung cancer	4000 (2300/1700)**	3123 (1867/1256)	44 (12/32)
IHD	11700 (6200/5500)	3767 (2210/1557)	395 (109/286)
Stroke	6750 (2500/4250)	1540 (668/872)	335 (57/278)
Respiratory	6500 (3000/3500)	2700 (1439/1261)	91 (19/72)
4 causes combined	28950 (14,000/14,950)	11130 (6184/4946)	865 (197/668)

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