Results
In total, there were 22 selected studies that examined cigar smoking and mortality from 16 cohorts (Table 1). All of the studies were from prospective cohorts for which vital status was usually determined through either active follow-up or linkage to a death registry (Strachan conducted a case–control study nested within a prospective cohort). Four studies were published from the Dorn study cohort which approximately 250,000 government life-insurance policy holders (mostly US World War I veterans) responded to questionnaires on tobacco use mailed to them in the 1950s. Two studies were published from the Whitehall study cohort in which over 18,000 men aged 40–69 years from the British Civil Service underwent an examination between 1967 and 1969 which included reporting health history and lifetime smoking habits. Finally, three studies were published from the American Cancer Society's Cancer Prevention Study II (CPS-II) cohort in which over a half million male volunteers from all 50 US states completed a self-administered questionnaire in 1982. Women were not included in CPS-II studies of cigar use because they were not asked if they smoked cigars. One of the 16 cohorts began in the 1940s, five cohorts began in the 1950s, five cohorts began in the 1960s, four cohorts began in the 1970s, and one cohort began in the 1980s. The studies were conducted primarily in North America and Europe with 12 in the US, five in the United Kingdom (UK), one in Canada, and four in Nordic countries (Denmark, Sweden, and Finland).
We found a substantial amount of variation across studies in terms of study characteristics including the definition of tobacco exposure (e.g. assessment of past cigarette smoking, inclusion of current pipe smoking), dose categories, and adjustment for possible confounding risk factors. Furthermore, CPS-I and II were the only cohorts large enough to yield results for primary cigar smokers and secondary cigar smokers separately for most outcomes. Thus, we present the results descriptively, rather than giving pooled relative risk estimates. In the subsequent results, we refer to effect estimates as "mortality ratios" or MRs, regardless of whether they are age-standardized mortality ratios, hazards ratios (HR), incidence rate ratios (IRR) or odds ratios (OR). We present MRs for current cigar smoking only (rather than ever or former cigar smoking). The referent group is never smokers or never tobacco users. Results for all-cause mortality and cause-specific mortality are presented. Estimates for groups of causes, such as cancers of the digestive system or genitourinary system, were not included in our results. For each outcome, we present results in the following order of exposure: primary cigar smoking (current, exclusive cigar smoking with no previous history of cigarette or pipe smoking), primary cigar and/or pipe smoking (current, exclusive cigar and/or pipe smoking with no previous history of cigarette smoking), secondary cigar smoking (current, exclusive cigar smoking with previous history of cigarette or pipe smoking), secondary cigar and/or pipe smoking (current, exclusive cigar and/or pipe smoking with previous history of cigarette smoking), any current cigar smoking (regardless of cigarette/pipe smoking history), and any current cigar/pipe smoking (regardless of cigarette smoking history). Following overall cigar exposure, we then present results (when available) by level of cigar exposure (cigars per day), inhalation (any or the reported level), and duration of cigar smoking.
All-cause Mortality
All-cause mortality results are presented in Table 2 and Table 3. Two studies reported all-cause mortality risk among primary cigar smokers. Shanks and Burns found a significant positive association (MR = 1.08, 95% CI = 1.05–1.12) based on 3,754 cigar smoker deaths in CPS-I, whereas Ben-Shlomo reported an inverse association between cigar smoking and all-cause mortality based on a much smaller cohort with only 9 cigar smoker deaths (Table 2). In three cigar studies that also included pipe smoking, primary pipe/cigar smoking was associated with slight increases in all-cause mortality risks (MRs 1.09 to 1.37), and all of these associations just missed conventional thresholds for statistical significance with the lower bounds of the 95% confidence intervals being at least 0.98. All-cause mortality risks were increased in two studies reporting risks for secondary cigar smoking (MRs 1.12 to 1.20), and one association was statistically significant while the other just missed statistical significance with 0.99 as the lower bound of the 95% confidence interval. All three studies on secondary cigar/pipe smoking found significant increases in all-cause mortality (MRs 1.13 to 1.46). Overall, MRs for all-cause mortality among current cigar smokers ranged from 0.48 to 1.60 and 0.70 to 1.68 among current cigar/pipe smokers. Among primary cigar smokers in CPS-1, smoking 3 or more cigars per day (MRs 1.08–1.17) and higher levels of cigar smoke inhalation (MRs 1.19–1.60) were both associated with significantly increased risk of all-cause mortality (Table 3).
Oral Cancer
Results for mortality from oral cancer are presented in Table 4 and Table 5. Shanks and Burns and Shapiro, Jacobs, and Thun both found significantlyelevated risk of oral cancer mortality for primary cigar smokers in CPS-I and II (Table 4). Secondary cigar smoking was also significantly associated with death from oral cancer in CPS-I. Overall, current cigar smoking MRs ranged from 4 to 7.9. Strong dose–response trends were observed for cigars per day, level of inhalation, and duration in both studies of primary cigar smoking in CPS-I and II (Table 5). Primary cigar smokers reporting no inhalation also had significantly elevated risk of oral cancer mortality in CPS-I (MR = 6.98, 95% CI = 4.13–11.03). Risk of oral cancer was elevated but not statistically significant (MR = 2.12, 95% CI = 0.43–6.18) among primary cigar smokers who smoked 1–2 cigars per day in CPS-I, presumably due to small sample size, based on the wide confidence interval. Risk could not be estimated in CPS-II due to 0 exposed cases smoking 1–2 cigars per day.
Esophageal Cancer
Results for deaths from esophageal cancer are presented in Table 6 and Table 7. Shanks and Burns observed a significant positive association between primary cigar smoking and esophageal cancer mortality in CPS-I (MR = 3.60, 95% CI = 2.17–5.62) based on 19 exposed cases (Table 6). Shapiro, Jacobs, and Thun observed an elevated, but non-significant association in CPS-II (MR = 1.80, 95% CI = 0.90–3.70) based on 9 exposed cases, and was thus underpowered to detect this association. Shanks and Burns also observed a significant association for secondary cigar smoking in CPS-I. Current cigar smoking MRs ranged from 1.8 to 6.5. In CPS-I, Shanks and Burns found strong dose–response trends for cigars per day and depth of inhalation among primary cigar smokers (Table 7), even though an elevated risk was still observed among those reporting no inhalation (MR = 3.40, 95% CI = 1.90, 5.61). Risk of esophageal cancer was elevated but not statistically significant among primary cigar smokers who smoked 1–2 cigars per day in both CPS-I (MR = 2.28, 95% CI = 0.74–5.33) and CPS-II (MR = 1.80, 95% CI = 0.60–5.00, based on 4 exposed cases). Based on the wide confidence intervals, both cohorts appear to be underpowered to detect an association at this level of exposure due to small sample size.
Stomach Cancer
Mortality risk estimates for stomach cancer are presented in Table 8 and Table 9. Overall, current cigar smoking MRs were 1.2 in the Dorn study cohort and 2.29 from CPS-II, with the estimate from CPS-II being statistically significant (Table 8). Stronger associations were observed for the highest levels of both cigars per day and inhalation in CPS-II (Table 9). Cigar smoking was found to be associated with stomach cancer mortality in CPS-II for both less than 40 years of cigar smoking and 40 years or more of cigar smoking.
Liver Cancer
Results for liver cancer are presented in Table 10. Studies of liver cancer mortality risk for cigar smoking are limited. Both studies of current cigar smoking showed significant associations with deaths from liver cancer; however, neither study reported cigarette smoking history.
Pancreatic Cancer
Results for deaths from pancreatic cancer are presented in Table 11 and Table 12. Shanks and Burns found a significant association between primary cigar smoking and pancreatic cancer mortality in CPS-I, while Shapiro, Jacobs, and Thun did not find a significant association in CPS-II (Table 11). Shanks and Burns also found a significant association with pancreatic cancer mortality risk for secondary cigar smokers. Overall, current cigar smoking MRs ranged from 1 to 1.8. A weak dose–response trend was seen for cigars per day in CPS-I and II (Table 12). Inhalation was associated with pancreatic cancer mortality in CPS-II, and some evidence of a dose–response trend for inhalation was observed in CPS-I.
Laryngeal Cancer
The results for deaths from laryngeal cancer are presented in Table 13 and Table 14. Shanks and Burns and Shapiro, Jacobs, and Thun found significant positive associations between primary cigar smoking and laryngeal cancer mortality risk in CPS-I and CPS-II that were of similar magnitude (MRs = 10) (Table 13). Kahn also found a significant association of the same magnitude among current cigar smokers. Shanks and Burns and Shapiro, Jacobs, and Thun observed a strong dose–response relationship for cigars per day and level of inhalation, and Shapiro, Jacobs, and Thun observed similar trends for duration of cigar smoking (Table 14). Shanks and Burns also observed that primary cigar smokers reporting no inhalation in CPS-I had significantly elevated risk of mortality from laryngeal cancer (MR = 10.6, 95% CI = 3.87–23.07). Risk of laryngeal cancer was highly elevated but not statistically significant among primary cigar smokers who smoked 1–2 cigars per day in both CPS-I (MR = 6.45, 95% CI = 0.72–23.27) and CPS-II (MR = 6.00, 95% CI-0.70–53.50, based on 1 exposed case). Based on the wide confidence intervals, both cohorts appear to be underpowered to detect an association at this level of exposure due to small sample size.
Lung Cancer
Lung cancer mortality risk estimates are presented in Table 15 and Table 16. Shanks and Burns and Shapiro, Jacobs, and Thun found significant positive associations with lung cancer mortality risk in CPS-I and II, although Ben-Shlomo did not observe a significant association in a much smaller cohort (Table 15). Wald found a significant association between primary cigar/pipe smoking and lung cancer mortality risk. Shanks and Burns and Ben-Shlomo found a significant positive association between lung cancer mortality and secondary cigar smoking as did Wald for secondary cigar/pipe smoking. Overall, current cigar smoking MRs ranged from 1.59 to 7.64, and current cigar/pipe smoking MRs ranged from 3.19 to 8.64. Dose–response trends were observed for cigars per day, level of inhalation, and duration in CPS-I and II data (Table 16).
Bladder Cancer
The results for deaths from bladder cancer are presented in Table 17 and Table 18. Neither Shanks and Burns nor Shapiro, Jacob, and Thun found an association between primary cigar smoking and bladder cancer mortality risk in CPS-I and CPS-II, respectively, and Shanks and Burns did not find an association with secondary cigar smoking (Table 17). Overall, current cigar smoking MRs ranged from 0.94 to 1.9. There was no dose–response trend for cigars per day among primary and secondary cigar smokers in both CPS-I and CPS-II. Shanks and Burns did not find increased mortality risk with depth of inhalation in CPS-I, but Shapiro, Jacobs, and Thun did find a significant association with inhalation among primary cigar smokers (Table 18).
Coronary Heart Disease
Mortality results due to coronary heart disease (CHD) are presented in Table 19 and Table 20. Primary cigar smoking was significantly associated with CHD mortality in CPS-I (MR = 1.05, 95% CI = 1.00–1.11). Primary cigar smoking was also associated with CHD mortality in CPS-II men aged 30–74 years (MR = 1.30, 95% CI = 1.05–1.62), but not in men 75 years and older. Primary cigar smoking was also not associated with CHD mortality in the much smaller Whitehall study cohort. The one study on primary cigar/pipe smoking from Wald did not observe an increased risk of death from CHD. CPS-I also found increased CHD mortality risk to be associated with secondary cigar smoking (MR = 1.09, 95% CI = 1.01–1.18), although the small Whitehall study cohort did not. Wald did not find an association between secondary cigar/pipe smoking and CHD mortality risk. Overall, the current cigar smoking MRs ranged from 0.45 to 1.30, and the current cigar/pipe MRs ranged from 0.98 to 1.67. There was a weak dose–response relationship with cigars per day and level of inhalation among primary but not secondary cigar smokers in the CPS-I study (Table 20). Primary cigar smoking for 25 years or more was also associated with increased CHD mortality risk in CPS-II men aged 30–74 years.
Stroke
Results for stroke are presented in Table 21 and Table 22. Shanks and Burns did not find a significant association with stroke mortality risk for either primary or secondary cigar smoking in CPS-I (Table 21). Overall, current cigar smoking MRs ranged from 0.92 to 1.08, and Haheim found a significantly increased risk of death from stroke for cigar/pipe smoking (MR = 3.6, 95% CI = 1.05–12.3). However, there was no dose–response trend for cigars per day and level of inhalation among primary and secondary cigar smokers in CPS-I (Table 22).
Aortic Aneurysm
Mortality risk estimates for aortic aneurysm are presented in Table 23 and Table 24. Shanks and Burns found significant positive associations for aortic aneurysm mortality for both primary and secondary cigar smoking in CPS-I (Table 23). Overall, current cigar smoking MRs ranged from 1.76 to 5.10, while Strachan found that current cigar/pipe smoking was associated with a five-fold increased risk of death from aortic aneurysm (MR = 5.40, 95% CI = 1.90–15.30). There were no clear dose–response trends for cigars per day and levels of inhalation with in CPS-I (Table 24). Notably, risk of aortic aneurysm was significantly elevated among primary cigar smokers who smoked 1–2 cigars per day in CPS-I (MR = 1.82, 95% CI = 1.11–2.81).
COPD
Results for deaths from chronic obstructive pulmonary disease (COPD) are presented in Table 25 and Table 26. Shanks and Burns found that COPD mortality was not associated with primary cigar smoking but was significantly associated with secondary cigar smoking (MR = 4.39, 95% CI = 3.02–6.16) in CPS-I (Table 25). Wald found that COPD mortality was not associated with primary cigar/pipe smoking but was significantly associated with secondary cigar/pipe smoking. Overall, current cigar smoking MRs ranged from 0.79 to 4.39, while current cigar/pipe smoking MRs ranged from 1.11 to 1.68. For cigars per day, there were no dose–response trends for primary cigar smoking, but there were strong dose–response trends for secondary cigar smoking in CPS-I (Table 26). For level of inhalation, there were suggestive dose–response trends for primary and secondary cigar smoking.
Other Causes of Death
For a number of causes of death (atherosclerosis, cancers of the kidney, nasopharynx, colon and rectum), there were only single estimates of associations with cigar smoking (Table 27). All estimates come from the Dorn study cohort. Although estimates for colon and rectal cancer were also reported by Kahn, Heineman reported more up-to-date associations for these cancer sites. Only colon and rectal cancer were significantly associated with primary current cigar/pipe smoking after adjusting for age, calendar time, year of questionnaire response, SES, sedentary job. For colon cancer only, there was a significant dose-trend for the number of cigars smoked a day after adjusting for the same factors (p-trend = 0.004, data not shown).