- N J Wald, professor,
- C Watt H, statistician
- a BUPA Epidemiological Research Group Department of Environmental and Preventive Medicine Wolfson Institute of Preventive Medicine St
- bc Bartholomew's and the Royal London School of Medicine and Dentistry London EC1M 6BQ
- Correspondence to: Professor Wald
- Accepted 10 April 1997
Objective: To estimate the extent to which cigarette smokers who switch to cigars or pipes alter their risk of dying of three smoking related diseases–lung cancer, ischaemic heart disease, and chronic obstructive lung disease.
Design: A prospective study of 21 520 men aged 35-64 years when recruited in 1975-82 with detailed history of smoking and measurement of carboxyhaemoglobin.
Main outcome measures: Notification of deaths (to 1993) classified by cause.
Results: Pipe and cigar smokers who had switched from cigarettes over 20 years before entry to the study smoked less tobacco than cigarette smokers (8.1 g/day v 20 g/day), but they had the same consumption as pipe and cigar smokers who had never smoked cigarettes (8.1 g) and had higher carboxyhaemoglobin saturations (1.2% v 1.0%, P<0.001), indicating that they inhaled tobacco smoke to a greater extent. They had a 51% higher risk of dying of the three smoking related diseases than pipe or cigar smokers who had never smoked cigarettes (relative risk 1.51; 95% confidence interval 0.96 to 2.38), a 68% higher risk than lifelong non-smokers (1.68; 1.16 to 2.45), a 57% higher risk than former cigarette smokers who gave up smoking over 20 years before entry (1.57; 1.04 to 2.38), and a 46% lower risk than continuing cigarette smokers (0.54; 0.38 to 0.77).
Conclusion: Cigarette smokers who have difficulty in giving up smoking altogether are better off changing to cigars or pipes than continuing to smoke cigarettes. Much of the effect is due to the reduction in the quantity of tobacco smoked, and some is due to inhaling less. Men who switch do not, however, achieve the lower risk of pipe and cigar smokers who have never smoked cigarettes. All pipe and cigar smokers have a greater risk of lung cancer than lifelong non-smokers or former smokers.
The health risks from smoking pipes or cigars are less than those from smoking cigarettes, but there is little direct evidence on these risks in cigarette smokers who switch to pipes or cigars
This prospective study shows that smokers who switch from cigarettes to pipes or cigars halve their combined risk of dying of lung cancer, ischaemic heart disease, or chronic obstructive lung disease compared with continuing smokers, but their risk was still about 50% higher than that of lifelong non-smokers
Some of this reduction in risk was due to reduced inhaling, but most of it was due to a reduction in the amount of tobacco smoked
The best option is either not to smoke or to give up altogether; failing that, switching to pipes or cigars is better than continuing to smoke cigarettes
It is recognised that cigarette smokers inhale tobacco smoke while men who smoke only pipes or cigars tend not to. Pipe and cigar smokers are at lower risk of the main smoking related diseases, probably in part because of this difference in inhaling habit1 and because they may smoke less tobacco. There is evidence that smokers who switch from cigarettes to pipes or cigars tend to maintain their acquired inhaling habits.2 3 4 5 6 7 The extent to which this occurs and the extent to which it negates the potential health benefits associated with smoking cigars or pipes rather than cigarettes is uncertain. We used data from the British United Provident Association (BUPA) study to examine these points in greater detail in relation to three diseases caused by smoking–namely, lung cancer, ischaemic heart disease, and chronic obstructive lung disease. In particular, we assessed the merits of switching from smoking cigarettes to smoking pipes and cigars.
Subjects and methods
The BUPA study, a prospective study of 21 520 professional and business men aged 35-64 years who attended the BUPA Medical Centre in London for a routine health examination between 1975 and 1982, has been described previously.8 At the time of each examination, which included a blood pressure measurement, a detailed history of smoking was obtained, including self reported inhaling habits classified into four categories (nil, slight, moderate, deep). A blood sample was collected, carboxyhaemoglobin saturation and various other factors (including serum cholesterol) measured, and serum samples stored at -40°C. The study was restricted to men with NHS numbers so that their NHS records could be flagged and the Office of Population Censuses and Surveys could inform us of all deaths and their certified causes. Further information was then sought from the doctor who certified death. Eight hundred and twenty nine men were lost to follow up because they emigrated. The average follow up time was 14 years and 4 months. This report is based on the deaths that occurred in the study up to October 1993. The codes of ICD-9 (international classification of diseases, ninth revision) used to classify the three specified smoking related diseases were 162 for lung cancer, 410-414 for ischaemic heart disease, and 416, 491, 492, 496, and 519 for chronic obstructive lung disease.
To assess the benefits of switching from cigarettes to cigars or pipes, or both, we compared mortality from the three smoking related diseases in current pipe or cigar smokers who had switched (switchers) from smoking cigarettes at least 20 years before entry to the study (that is, 20 years before 1975-82) with that in pipe or cigar smokers who were current smokers at entry and had never smoked cigarettes (non-switchers). The 20 year interval between switching and the date of the examination was selected to avoid the excess health hazards attributable to past cigarette smoking. We verified this by showing that men who had stopped smoking cigarettes for at least 20 years before entry to the study and had not switched to pipes or cigars had death rates from the three specified diseases that were similar to those in lifelong non-smokers. We also compared death rates in the switchers with those in cigarette smokers who did not smoke cigars or pipes, with those in former smokers, and with those in lifelong non-smokers.
Consumption of tobacco was estimated on the basis of one cigarette containing 1 g of tobacco, one small cigar (cheroot) containing 2 g, and one large cigar containing 5 g. The weight of tobacco used by pipe smokers was recorded directly on the questionnaire. Carboxyhaemoglobin saturations were measured (taking the mean of two measurements) with an IL182 co-oximeter, as described in detail previously, for all men at the time they attended the medical centre.9 Risks of mortality from the three specified diseases were compared by using Cox's proportional hazards survival analysis adjusted for age at entry to the study. We used survival analysis to take account of the differing lengths of follow up, which were mainly attributable to the eight year recruitment period. The results were also analysed by logistic regression (results not shown), which gave virtually identical results. The association between carboxyhaemoglobin measurements and self described inhaling category was examined by analysis of variance.
Table 1) shows the data collected in 1975-82 on median tobacco consumption and carboxyhaemoglobin saturations according to smoking group: men who currently smoked cigars or pipes and had never smoked cigarettes, men who currently smoked cigars and pipes but smoked cigarettes at least 20 years previously, and men who smoked only cigarettes. Tobacco consumption was similar in the two groups of pipe and cigar smokers (switchers and non-switchers) (table 1), but the median carboxyhaemoglobin saturation was higher in the switchers (1.2% v 1.0%, P<0.0001 by the Wilcoxon rank sum test). The mean carboxyhaemoglobin saturations in both groups of pipe and cigar smokers were much lower than in current cigarette smokers (table 1). In lifelong non-smokers and in former smokers the median carboxyhaemoglobin saturation (reflecting endogenous production and exposure to atmospheric carbon monoxide) was 0.7% (10th-90th centile 0.4-1.1).
Table 2) shows self described inhaling category according to smoking group. One third of the switchers, and about half of the non-switchers said that they inhaled, and 95% of current cigarette smokers said that they inhaled; this indicates clear differences between the three groups.
Table 2) also shows an increasing trend in carboxyhaemoglobin saturation with deeper self described inhaling; cigarette smokers had higher saturations than other smokers except for men in the lightest category of smoking. This was also the case after we allowed for type of product smoked and weight of tobacco smoked (P<0.0001 by analysis of variance). In the pipe and cigar smokers there was no material difference in carboxyhaemoglobin saturations within each inhaling category between switchers and non-switchers; self described inhaling therefore accounted for the difference in carboxyhaemoglobin saturations shown in table 1).
Table 3) shows the mortality and the number of deaths by 1993 from the three smoking related diseases and from all causes for the three groups of current smokers and for former smokers relative to lifelong non-smokers. The combined risk of the three diseases in current cigarette smokers was 3.18 times higher (95% confidence interval 2.55 to 3.84) than in lifelong non-smokers; individual relative risk estimates were 2.27 for ischaemic heart disease, 16.4 for lung cancer, and 29.5 for chronic obstructive lung disease. Having given up smoking cigarettes 20 or more years before entry to the study reduced the risk to about that of a lifelong non-smoker (rows 1 and 2 in table 3). The risk in switchers was much less than among continuing cigarette smokers (rows 4 and 5, P<0.001) but somewhat higher than the risk in lifelong pipe and cigar smokers (rows 3 and 4, P=0.07). Mortality from all causes was also graded as expected. The rates were lower in non-smokers; successively higher in former smokers, pipe and cigar smokers who had never smoked cigarettes, and pipe and cigar smokers who had previously smoked cigarettes; and substantially higher in cigarette smokers. The relative risks of ischaemic heart disease were not materially changed after we allowed for serum cholesterol concentration and blood pressure. They were 0.97 (95% confidence interval 0.71 to 1.33) for former smokers, 1.00 (0.68 to 1.47) for pipe and cigar smokers who never smoked cigarettes, 1.31 (0.85 to 2.03) for those who had switched from cigarettes, and 2.18 (1.74 to 2.73) for cigarette smokers.
Table 4) shows the risk of the three smoking related diseases in the switchers compared with other groups. The risk of all three diseases combined was 46% lower in switchers than in continuing cigarette smokers (relative risk 0.54) and 68% higher than in lifelong non-smokers. It was 51% higher than in pipe or cigar smokers who had not switched from cigarettes.
Interpretation of results
Our results indicate that current pipe or cigar smokers who switched from cigarettes smoke about the same amount of tobacco as pipe and cigar smokers who never smoked cigarettes but that they tend to inhale more. Most of the reduction in the risk of dying from ischaemic heart disease, lung cancer, and chronic obstructive lung disease combined compared with that in continuing cigarette smokers is attributable to the fact that pipe and cigar smokers smoke less tobacco than the cigarette smokers (median 8.1 g a day compared with 20 g a day). The risk of dying from the three diseases combined in these switchers was less (24% less; 95% confidence interval 55% less to 27% more) than in light cigarette smokers who smoked the same amount of tobacco–that is, those who smoked between one and 14 cigarettes a day with a median of 8 g tobacco a day–so the reduction may also be explained in part by reduced inhaling. There may be other differences in lifestyle between the switchers and non-switchers, though this is probably unlikely because adjustment for serum cholesterol concentration and systolic blood pressure made no difference to the risk estimates.
Our results indicate a specific adverse effect among pipe smokers who switched from cigarettes that is not due to former cigarette smoking. The carboxyhaemoglobin saturations shown in table 1) suggest this is due to increased inhaling. For example, the mean carboxyhaemoglobin saturation in pipe smokers who never smoked cigarettes was about twice the background level (1.4 v 0.7); it was nearly three times the background level in those who switched from cigarettes (1.9 v 0.7). An interesting result, though one that is not directly relevant to the question we sought to answer, was that carboxyhaemoglobin saturation was related to the risk of the three specified smoking related diseases independently of smoking category or amount smoked. Indeed, after adjustment for carboxyhaemoglobin saturations, smoking was no longer significantly related to risk. We estimated that regardless of smoking category, the risk of dying of the three specified diseases increased by 22% per 1% increase in carboxyhaemoglobin saturation.
Two previous studies have compared mortality in pipe and cigar smokers who switched from smoking cigarettes with those who never smoked cigarettes, but neither yielded conclusive results. The paper by Kaufman et al was a case-control study of cigar and pipe smoking in relation to myocardial infarction in young men (aged 40-54 years).10 The group of men who stopped smoking cigarettes and switched to pipes or cigars included men who could have switched as recently as two years previously, so that any excess risk could have been associated with the residual effect of cigarette smoking rather than the effect of pipe or cigar smoking. The Whitehall study showed that pipe or cigar smokers who previously smoked cigarettes had a higher mortality than pipe or cigar smokers who did not previously smoke cigarettes,11 though again this result could have been due to the residual effects of cigarette smoking in the former cigarette smokers. The study did show that cigarette smokers who switched to smoking pipes (but not to smoking cigars) had a higher mortality from all causes than former cigarette smokers (relative risk 1.17; 1.03 to 1.34), with a significant increase in risk of ischaemic heart disease. The difference was explained by the pipe smokers' previous cigarette smoking habits.
Public health implications
Our results have public health implications. Cigarette smokers may be able to stop smoking when they receive advice and support and by using aids such as nicotine chewing gum.12 Some, however, may still have difficulty in giving up smoking altogether, and these smokers would be better off changing to cigars or pipes instead of continuing to smoke cigarettes. Much of the effect is due to the reduction in the quantity of tobacco smoked; the rest may be attributable to reduced inhalation. Cigarette smokers may find it easier to reduce consumption by changing to smoking cigars or pipes rather than by smoking fewer cigarettes. The risk of the three specified smoking related diseases, none the less, is still higher than in men who only ever smoked pipes or cigars, higher than in those who gave up smoking altogether–and, of course, higher than in men who never smoked at all.
We thank Sir Richard Doll, Leo Kinlen, Jim Haddow, and Malcolm Law for their helpful comments.
Funding: BUPA Medical Research and Development Ltd gave financial support to the BUPA Epidemiological Research Group within the Wolfson Institute of Preventive Medicine.
Conflict of interest: None.