Cigar Smoking and All Cause and Smoking Related Mortality
Cigar Smoking and All Cause and Smoking Related Mortality
We conducted a systematic review of epidemiological studies published prior to June 2014 that examined the association between cigar smoking and all-cause and cause-specific mortality. In doing so, we followed guidelines for systematic reviews from the Institute of Medicine (IOM), the Meta-analysis Of Observational Studies in Epidemiology (MOOSE) group and the Cochrane Collaboration. For cause-specific mortality, we selected causes and conditions identified as smoking-related in the 2004 Surgeon General's Report on the health effects of smoking and IARC's 2012 summary monograph "A Review of Human Carcinogens". Based on these criteria, studies of cigar smoking and the following causes of death were identified in the research literature and included in this review: oral cancer, nasopharyngeal carcinoma, esophageal cancer, stomach cancer, colon and rectal cancer, liver cancer, pancreatic cancer, laryngeal cancer, lung cancer, bladder cancer, kidney cancer, atherosclerosis, coronary heart disease, stroke, aortic aneurysm, and chronic obstructive pulmonary disease (COPD).
We conducted a search in PubMed using the following terms: (cigar[tw] OR cigars[tw] OR cigarillo[tw] OR cigarillos[tw] OR cheroot[tw] OR cheroots[tw] OR stogies[tw]) AND (death[tw] OR mortality[tw]), yielding 100 potentially pertinent references. We also conducted a search in EMBASE using the following terms: (cigar OR cigars OR cigarillo OR cigarillos OR cheroot OR cheroots OR stogies) AND ('death'/exp OR death OR 'mortality'/exp OR mortality), yielding 207 potential references. Finally, we searched ISI Web of Science using the terms: Topic = (cigar OR cigars OR cigarillo OR cigarillos OR cheroot OR cheroots OR stogies) AND Topic = (death OR mortality), yielding 92 potential references. After removing duplicate records among the three search results, there were a total of 246 references.
At each stage of the study screening, two reviewers (CMC and CGC) independently reviewed the studies and made selections for inclusion (Figure 1). Final selections were made after discussion and consensus was reached on discrepant results. All selected studies were screened by title and abstract, and the full texts of the subset of relevant papers were then reviewed. A total of 227 references were excluded from review (see Additional file 1 http://www.biomedcentral.com/1471-2458/15/390/additional).
(Enlarge Image)
Figure 1.
Flowchart of study selection of prospective studies of cigar and all-cause and cause-specific mortality.
Because the focus of this review is on studies that examine current cigar smoking at baseline, we excluded an earlier cohort study conducted by the American Cancer Society that looked at mortality risks for lifetime ever cigar smokers (w1-w2). A total of 108 references did not report mortality risk associated with cigar use (w3-w110). An additional 13 non-English studies were also excluded (w111-w123). We excluded 90 non-epidemiologic or non-human studies, including reviews and commentaries (w124-w213). Two studies were excluded because they examined causes of death (prostate cancer and multiple myeloma) that were not classified as smoking-related in the 2004 Surgeon General's Report or 2012 IARC monograph (w214-w215). Because the aim of this review was to compare mortality risks relative to never tobacco users or never smokers, two studies with former or current cigarette smokers in their reference groups were excluded (w216-w217). We excluded 7 references because they only contained results from cohort studies that were also published in studies that were selected for review (w218-w224). Three references were excluded because they were unpublished meeting abstracts (w225-w227). Finally, three additional references were identified by handsearching the references of relevant studies. In total, 22 articles on cigar smoking and all-cause and cause-specific mortality were included in this review (Figure 1 and Table 1).
Two different sets of data were extracted. The first dataset consisted of study-level characteristics including cohort name, country where study was conducted, publication year of study, number of cigar users, age range of cohort participants, year of enrollment, length of follow-up, outcome type (i.e. all-cause or cause-specific mortality), and relative mortality risk measure. The second dataset consisted of association-level characteristics including relative risk estimates as well as characteristics used to stratify estimates such as age, type and level of cigar exposure, and duration of cigar exposure. Thus, multiple entries could occur for a particular study. Additionally, reported International Classification of Diseases (ICD) codes from the relevant revision were extracted for cause-specific mortality estimates. Although one included study, Shanks and Burns (1998), did not report ICD codes in their analysis of Cancer Prevention Study I (CPS-I) data, it can be assumed that their cause of death coding system was similar to ICD 7, given that similar studies of CPS-I have used ICD 7 codes. Data extraction was independently conducted by two reviewers (CMC and CGC). Similar to the selection of studies, data entries were compared and any differences in data extraction were resolved through discussion.
A number of studies did not report 95% confidence intervals and/or relative mortality risks (see "Effect measure reported" column in Table 1). For these studies, we calculated these estimates using the available data. For standardized mortality ratios (SMRs), we used the relevant formula from Breslow et al. (equation 2.15) to calculate 95% confidence intervals. For incidence rate ratios (IRRs), we calculated 95% confidence intervals using Wald Limits.
The main exposure of interest was either current exclusive cigar smoking or current exclusive cigar/pipe smoking. Current cigarette smokers are excluded from this analysis to better isolate the effects of cigar smoking on disease risk. Past cigarette smoking patterns are likely to affect current cigar smoking patterns. CPS-I results suggest that secondary cigar smokers (current, exclusive cigar smoking with a history of previous cigarette or pipe smoking) are about twice as likely to report inhaling cigar smoke as primary cigar smokers (current, exclusive cigar smoking with no history of previous cigarette or pipe smoking) (42.0% vs. 21.6%). Therefore, secondary cigar smokers may have different disease risks compared with primary cigar smokers not only due to past cigarette use, but also due to differences in cigar use, particularly with inhalation. Thus, for studies that assessed this information, we present results for primary and secondary cigar smokers separately as well as separate studies that combine cigar and pipe use from studies of cigar use only.
Methods
We conducted a systematic review of epidemiological studies published prior to June 2014 that examined the association between cigar smoking and all-cause and cause-specific mortality. In doing so, we followed guidelines for systematic reviews from the Institute of Medicine (IOM), the Meta-analysis Of Observational Studies in Epidemiology (MOOSE) group and the Cochrane Collaboration. For cause-specific mortality, we selected causes and conditions identified as smoking-related in the 2004 Surgeon General's Report on the health effects of smoking and IARC's 2012 summary monograph "A Review of Human Carcinogens". Based on these criteria, studies of cigar smoking and the following causes of death were identified in the research literature and included in this review: oral cancer, nasopharyngeal carcinoma, esophageal cancer, stomach cancer, colon and rectal cancer, liver cancer, pancreatic cancer, laryngeal cancer, lung cancer, bladder cancer, kidney cancer, atherosclerosis, coronary heart disease, stroke, aortic aneurysm, and chronic obstructive pulmonary disease (COPD).
We conducted a search in PubMed using the following terms: (cigar[tw] OR cigars[tw] OR cigarillo[tw] OR cigarillos[tw] OR cheroot[tw] OR cheroots[tw] OR stogies[tw]) AND (death[tw] OR mortality[tw]), yielding 100 potentially pertinent references. We also conducted a search in EMBASE using the following terms: (cigar OR cigars OR cigarillo OR cigarillos OR cheroot OR cheroots OR stogies) AND ('death'/exp OR death OR 'mortality'/exp OR mortality), yielding 207 potential references. Finally, we searched ISI Web of Science using the terms: Topic = (cigar OR cigars OR cigarillo OR cigarillos OR cheroot OR cheroots OR stogies) AND Topic = (death OR mortality), yielding 92 potential references. After removing duplicate records among the three search results, there were a total of 246 references.
Study Selection
At each stage of the study screening, two reviewers (CMC and CGC) independently reviewed the studies and made selections for inclusion (Figure 1). Final selections were made after discussion and consensus was reached on discrepant results. All selected studies were screened by title and abstract, and the full texts of the subset of relevant papers were then reviewed. A total of 227 references were excluded from review (see Additional file 1 http://www.biomedcentral.com/1471-2458/15/390/additional).
(Enlarge Image)
Figure 1.
Flowchart of study selection of prospective studies of cigar and all-cause and cause-specific mortality.
Because the focus of this review is on studies that examine current cigar smoking at baseline, we excluded an earlier cohort study conducted by the American Cancer Society that looked at mortality risks for lifetime ever cigar smokers (w1-w2). A total of 108 references did not report mortality risk associated with cigar use (w3-w110). An additional 13 non-English studies were also excluded (w111-w123). We excluded 90 non-epidemiologic or non-human studies, including reviews and commentaries (w124-w213). Two studies were excluded because they examined causes of death (prostate cancer and multiple myeloma) that were not classified as smoking-related in the 2004 Surgeon General's Report or 2012 IARC monograph (w214-w215). Because the aim of this review was to compare mortality risks relative to never tobacco users or never smokers, two studies with former or current cigarette smokers in their reference groups were excluded (w216-w217). We excluded 7 references because they only contained results from cohort studies that were also published in studies that were selected for review (w218-w224). Three references were excluded because they were unpublished meeting abstracts (w225-w227). Finally, three additional references were identified by handsearching the references of relevant studies. In total, 22 articles on cigar smoking and all-cause and cause-specific mortality were included in this review (Figure 1 and Table 1).
Data Extraction
Two different sets of data were extracted. The first dataset consisted of study-level characteristics including cohort name, country where study was conducted, publication year of study, number of cigar users, age range of cohort participants, year of enrollment, length of follow-up, outcome type (i.e. all-cause or cause-specific mortality), and relative mortality risk measure. The second dataset consisted of association-level characteristics including relative risk estimates as well as characteristics used to stratify estimates such as age, type and level of cigar exposure, and duration of cigar exposure. Thus, multiple entries could occur for a particular study. Additionally, reported International Classification of Diseases (ICD) codes from the relevant revision were extracted for cause-specific mortality estimates. Although one included study, Shanks and Burns (1998), did not report ICD codes in their analysis of Cancer Prevention Study I (CPS-I) data, it can be assumed that their cause of death coding system was similar to ICD 7, given that similar studies of CPS-I have used ICD 7 codes. Data extraction was independently conducted by two reviewers (CMC and CGC). Similar to the selection of studies, data entries were compared and any differences in data extraction were resolved through discussion.
A number of studies did not report 95% confidence intervals and/or relative mortality risks (see "Effect measure reported" column in Table 1). For these studies, we calculated these estimates using the available data. For standardized mortality ratios (SMRs), we used the relevant formula from Breslow et al. (equation 2.15) to calculate 95% confidence intervals. For incidence rate ratios (IRRs), we calculated 95% confidence intervals using Wald Limits.
The main exposure of interest was either current exclusive cigar smoking or current exclusive cigar/pipe smoking. Current cigarette smokers are excluded from this analysis to better isolate the effects of cigar smoking on disease risk. Past cigarette smoking patterns are likely to affect current cigar smoking patterns. CPS-I results suggest that secondary cigar smokers (current, exclusive cigar smoking with a history of previous cigarette or pipe smoking) are about twice as likely to report inhaling cigar smoke as primary cigar smokers (current, exclusive cigar smoking with no history of previous cigarette or pipe smoking) (42.0% vs. 21.6%). Therefore, secondary cigar smokers may have different disease risks compared with primary cigar smokers not only due to past cigarette use, but also due to differences in cigar use, particularly with inhalation. Thus, for studies that assessed this information, we present results for primary and secondary cigar smokers separately as well as separate studies that combine cigar and pipe use from studies of cigar use only.