What is the association of cigar, pipe, and cigarette use with mortality risk in the US population?
In this nationally representative longitudinal cohort study of 357 420 participants comparing exclusive cigar, pipe, and cigarette use with mortality risk, a statistically significantly elevated risk of tobacco-related cancer mortality among exclusive current and former cigarette smokers, current cigar smokers, and current pipe smokers, as well as a statistically significantly elevated risk of mortality from most examined causes of death among both current daily and nondaily cigarette smokers, were demonstrated.
Use of cigar, pipe, and cigarettes each confers mortality risks.
Tobacco products have changed in recent years. Contemporary mortality risk estimates of combustible tobacco product use are needed.
To investigate the mortality risks associated with current and former use of cigars, pipes, and cigarettes.
Design, Setting, and Participants
The National Longitudinal Mortality Study is a longitudinal population-based, nationally representative health survey with mortality follow-up that includes demographic and other information from the Current Population Survey, tobacco product use information from the Tobacco Use Supplement, and mortality data from the National Death Index. In this study, participants provided tobacco use information at baseline in surveys starting from 1985 and were followed for mortality through the end of 2011. The study includes 357 420 participants who reported exclusively using cigar, pipes, or cigarettes or reported never using any type of tobacco product.
Current or former exclusive use of any cigar (little cigar, cigarillos, large cigar), traditional pipe, or cigarette and never tobacco use. Information on current daily and nondaily use was also collected. Estimates adjusted for age, sex, race/ethnicity, education, and survey year.
Main Outcomes and Measures
All-cause and cause-specific mortality as identified as the primary cause of death from death certificate information.
Of the 357 420 persons included in the analysis, the majority of current and former cigar and pipe smokers were male (79.3%-98.0%), and smokers were more evenly divided by sex (46% of current daily smokers were male). There were 51 150 recorded deaths during follow-up. Exclusive current cigarette smokers (hazard ratio [HR], 1.98; 95% CI, 1.93-2.02) and exclusive current cigar smokers (HR, 1.20; 95% CI, 1.03-1.38) had higher all-cause mortality risks than never tobacco users. Exclusive current cigarette smokers (HR, 4.06; 95% CI, 3.84-4.29), exclusive current cigar smokers (HR, 1.61; 95% CI, 1.11-2.32), and exclusive current pipe smokers (HR, 1.58; 95% CI, 1.05-2.38) had an elevated risk of dying from a tobacco-related cancer (including bladder, esophagus, larynx, lung, oral cavity, and pancreas). Among current nondaily cigarette users, statistically significant associations were observed with deaths from lung cancer (HR, 6.24; 95% CI, 5.17-7.54), oral cancer (HR, 4.62; 95% CI, 1.84-11.58), circulatory death (HR, 1.43; 95% CI, 1.30-1.57), cardiovascular death (HR, 1.24; 95% CI, 1.11-1.39), cerebrovascular death (stroke) (HR, 1.39; 95% CI, 1.12-1.74), and chronic obstructive pulmonary disease (HR, 7.66; 95% CI, 6.09-9.64) as well as for daily smokers.
Conclusions and Relevance
This study provides further evidence that exclusive use of cigar, pipes, and cigarettes each confers significant mortality risks.
Combustible tobacco products, including cigars, pipes, and cigarettes, continue to represent the leading cause of preventable mortality in the United States. Mortality risks for cigarette smokers in the United States have been extensively studied and analyzed,1,2 but comparable estimates for users of other combustible tobacco products, such as cigars and pipes, are more limited. The 2014 Surgeon General’s Report on the health effects of cigarette smoking3 found that cigarettes are responsible for approximately 480 000 premature deaths annually. Although less common than cigarette use, US pipe and cigar use is notable. In 2015, an estimated 12.5 million people in the United States aged 12 years or older were current cigar smokers (past 30 days).4 In a recent analysis of National Adult Tobacco Survey 2013-2014 data, the authors report that 5.4% of US adults aged 18 years or older met the threshold for ever use (smoked ≥50 cigars over the lifetime) and currently used cigars daily, some days, or rarely.5 Similarly, 0.8% of US adults met the lifetime use threshold for pipes (smoked ≥50 pipes over the lifetime) and used a pipe daily, some days, or rarely. Quiz Ref IDBetween 2000 and 2015, the purchase and consumption of noncigarette tobacco products increased by 117.1% (83.8% per capita), including an increase in cigar consumption by 85.2% (56.8% per capita) and an increase in pipe tobacco consumption by 556.4% (455.7% per capita), while total cigarette consumption decreased by 38.7% (48.1% per capita).6
In 1998, a National Cancer Institute report concluded that cigar smoking causes cancer of the oral cavity, larynx, esophagus, lung, and possibly pancreas.7 In addition, cigar smokers are at an increased risk of heart disease and other lung disease, and their risk of oral cavity cancer and esophageal cancer is comparable to that of cigarette smokers.7 A 2010 analysis indicated that cigar smoking is responsible for approximately 9000 premature deaths and more than 140 000 years of potential life lost among US adults aged 35 years or older.8 However, few morbidity and mortality studies of cigar smoking have been published over the past 2 decades.9 Even less information is available on traditional pipes, and few studies have examined large numbers of exclusive pipe smokers using population-level data.10,11 These products have evolved substantially over the years and studies are needed to understand the contemporary disease risks of cigar and pipes. In this study, we provide updated estimates on the mortality risks of exclusive use of cigars and pipes utilizing data from the Tobacco Use Supplement to the Current Population Survey (TUS-CPS) linked to the National Longitudinal Mortality Study (NLMS).
The NLMS is a nationally representative sample of the civilian, noninstitutionalized US population and currently consists of the Annual Social and Economic Supplements (March 1973 to March 2011); the TUS-CPS for February 1978, April 1980, August 1980, December 1980, and September 1985; and a subset of the 1980 Census that is linked to mortality data from the National Death Index (through 2011).12 The present analysis includes a subset of the NLMS, namely participants who completed Tobacco Use Questionnaires that were included over the years as supplements to the US Census Bureau’s data collection activities. Our analysis includes participants who completed the 1992-1993, 1995-1996, 1998-1999, 2000, 2001-2002, 2003, 2006-2007, and 2010-2011 TUS-CPS questionnaires as well as participants who completed the similar tobacco questionnaire on the September 1985 CPS. A total of 640 726 NLMS participants from these surveys provided tobacco use information. Individuals with missing or incomplete tobacco use information or other covariates were excluded from the analysis. Because tobacco-related deaths are most likely to occur between ages 35 and 80 years, we excluded participants younger than 35 years or older than 80 years at the time of surveys (n = 229 648). We further excluded current or former exclusive smokeless tobacco users (n = 4709) and current or former users of multiple tobacco products (n = 48 949), because our analysis sought to estimate mortality risks for exclusive combustible tobacco users. After these exclusions, a total of 357 420 participants were included in the analysis and 51 150 death events were recorded. The National Institutes of Health determined that this study was exempt from human subject institutional review board review because data were not identifiable.
Tobacco use was classified as exclusive current or former use of cigarettes, any cigar (little cigar, cigarillos, large cigar), or traditional pipe, and never tobacco use. We defined individuals who reported having smoked 100 cigarettes or more in their lives as ever cigarette smokers. Current smokers reported smoking 100 lifetime cigarettes or more and reported smoking daily or some days at the time of the survey, and former smokers reported that they had smoked 100 cigarettes or more in their lifetime but currently do not smoke. Questions assessing tobacco product use evolved over the course of the September 1985 Supplement to the CPS and TUS-CPS surveys. In 1985, 1992-1993, 1995-1996, and 1999, cigar and pipe smokers reported having ever used these products on a regular basis, and current users reported that they used them at the time of survey. In 1998 and 2000, cigar and pipe users reported having ever used these products, with current users reporting daily or someday use at the time of survey and former users reporting no current use. Beginning with the 2001-2002 TUS-CPS cycle, cigar and pipe users reported having used these products even once, with current users reporting daily or someday use and former users reporting no current use. We defined never tobacco users as survey participants who indicated that they had never used cigars, pipes, and cigarettes. To examine the effect of smoking frequency, current users of cigars, pipes, and cigarettes were further divided into daily and nondaily users based on their reported daily or some day use.
Mortality was categorized by the underlying cause of death based on International Statistical Classification of Diseases, 10th Revision (ICD-10) coding using a standardized list of 113 causes developed by the National Center for Health Statistics. Mortality risks were calculated for all causes, tobacco-related cancers (lip, oral cavity, and pharynx [ICD-10: C00-C14]; esophagus, stomach, colon and rectum, liver, and pancreas [ICD-10: C15, C16, C18-C21, C22, C25]; larynx, trachea, bronchus, and lung [ICD-10: C32, C33-C34]; cervix, kidney, and bladder [C53, C64-C65, C67]), circulatory diseases (ICD-10: I00-I09, I20-I25, I26-I28, I29-I51, I60-I69, I70, I71, I72-I78), cardiovascular diseases (ICD-10: I00-I09, I20-I25, I26-I28, I29-I51, I70, I71, I72-I78), cerebrovascular disease (ICD-10: I60-I69), influenza and pneumonia (ICD-10: J10-J11, J12-J18, J40-J42, J43, J44), chronic obstructive pulmonary disease (COPD) (ICD-10: J40-J42, J43, J44), and diabetes (ICD-10: E10-E14).3 We also examined mortality from 2 major groups of tobacco-related cancers: lip, oral cavity, and pharynx cancer (ICD-10: C00-C14), and trachea, bronchus, and lung cancer (ICD-10: C33-C34) separately. Given the small number of deaths among tobacco user groups within each TUS-CPS cycle, we combined all TUS-CPS cycles into 1 analytic data set.
Mortality risks were estimated using Cox proportional hazards models with age as the underlying time variable. For analyses of both overall and cause-specific mortality, follow-up was from age at survey response through death (from any cause) or end of follow-up (December 31, 2011). For analyses of cause-specific mortality, we censored at the age of death from another cause. Multivariable models were used that adjusted for sex, race/ethnicity (non-Hispanic white, non-Hispanic African American, non-Hispanic other race, or Hispanic), educational attainment (less than high school graduate or equivalent, high school graduate or equivalent, some college education, or a bachelor’s degree or more education), and CPS survey year (1985, 1992-1993, 1996, 1999, 2000, 2001, 2002, 2003-2004, 2006, 2007, 2008, and 2010-2011). Multivariable models were developed based on review of previously published research studies on tobacco-related mortality risks.1 Hazard ratios (HRs) with a 2-tailed P value <.05 were considered statistically significant. Sampling weights were incorporated in all models and reweighted to account for differences in cohort size. This method produced results that account for and are more representative of the US population distribution. The proportional hazards assumption was assessed through visual inspection of –log-log plots of survivor function by time (age). Analyses were conducted with Unix SAS, version 9.4 (SAS Institute) using the relevant NLMS survey weights set to the noninstitutional US population size, appropriately accounting for the cluster sampling design of the original surveys.12
Table 1 presents participant demographic characteristics by tobacco use status. The majority (79.3%-98.0%) of current and former cigar and pipe smokers were male, whereas current cigarette smokers were more evenly divided by sex (46% of current daily smokers were male). Tobacco users for each product were more likely to be non-Hispanic whites. Most cigar and pipe users had at least some college education, but cigarette smoking was more concentrated among those with less than a college education. Specifically, 41.2% (95% CI, 38.6%-43.8%) of daily pipe smokers and 30.5% (95% CI, 27.8%-33.3%) of daily cigar smokers had at least a bachelor’s degree, compared with 11.4% (95% CI, 11.1%-11.7%) of current daily cigarette smokers. Proportional hazards assumptions were met in this data set.
Table 2 presents HRs for each mortality outcome by tobacco use status. Quiz Ref IDBoth current (HR, 1.98; 95% CI, 1.93-2.02) and former (HR, 1.32; 95% CI, 1.29-1.35) cigarette smokers had higher all-cause mortality risks than never tobacco users and were at a higher risk of dying from tobacco-related cancers, including bladder, esophagus, larynx, lung, oral cavity, and pancreas (current: HR, 4.06; 95% CI, 3.84-4.29; former: HR, 1.97; 95% CI, 1.86-2.09). Lung cancer mortality risks among current (HR, 11.82; 95% CI, 10.73-13.03) and former (HR, 4.15; 95% CI, 3.75-4.59) cigarette smokers were especially high. Compared with never combustible tobacco users, current and former cigarette smokers also had an increased risk of dying from tobacco-related diseases, such as oral cancer (current: HR, 9.02; 95% CI, 5.78-14.1; former: HR, 2.70; 95% CI, 1.66-4.39), cardiovascular disease (current: HR, 1.44; 95% CI, 1.38-1.51; former: HR, 1.11; 95% CI, 1.07-1.16), cerebrovascular disease (current: HR, 1.24; 95% CI, 1.12-1.36), COPD (current: HR, 11.01; 95% CI, 9.72-12.5; former: HR, 4.00; 95% CI, 3.54-4.53), and diabetes (current: HR, 1.18; 95% CI, 1.03-1.35; former: HR, 1.18; 95% CI, 1.05-1.33).
Quiz Ref IDWe also observed an elevated risk of death among exclusive current cigar and pipe users in relation to never tobacco users, although the HR for pipe users was statistically significant only in the age-adjusted model given fewer mortality events among pipe users (Table 2). In multivariable models, current exclusive cigar smokers (HR, 1.20; 95% CI, 1.03-1.38) had higher all-cause mortality risks than never tobacco users. Analyses of specific causes of death among cigar and pipe smokers were limited by modest sample sizes. However, both current cigar smokers (HR, 1.61; 95% CI, 1.11-2.32) and current pipe smokers (HR, 1.58; 95% CI, 1.05-2.38) had an increased risk of dying from tobacco-related cancers. Quiz Ref IDSimilarly, both current cigar smokers (HR, 3.26; 95% CI, 1.86-5.71) and current pipe smokers (HR, 1.51; 95% CI, 0.61-3.74) exhibited a higher risk of dying from lung cancer, although the risk estimate for lung cancer mortality among pipe smokers was not statistically significant and was based on 5 or fewer deaths in this category. We observed a nonstatistically significant association between current exclusive cigar smoking and COPD (HR, 2.44; 95% CI, 0.98-6.05); however, this estimate is based on 5 or fewer deaths.
Table 3 presents mortality risks among current tobacco users stratified by daily and nondaily use. Most cigarette and pipe smokers reported daily use; however, cigar smokers somewhat more commonly reported nondaily use. In general, daily users had higher mortality risks than nondaily users of each tobacco product, as expected. We observed associations for nondaily cigarette smokers with lung cancer (HR, 6.24; 95% CI, 5.17-7.54), oral cancer (HR, 4.62; 95% CI, 1.84-11.58), circulatory death (HR, 1.43; 95% CI, 1.30-1.57), cardiovascular death (HR, 1.24; 95% CI, 1.11-1.39), cerebrovascular death (stroke) (HR, 1.39; 95% CI, 1.12-1.74), and COPD (HR 7.66; 95% CI, 6.09-9.64), compared with never tobacco users. Among daily cigar users, mortality risks from tobacco-related cancer (HR, 1.80; 95% CI, 1.20-2.69), lung cancer (HR, 4.18; 95% CI, 2.34-7.46), and COPD (HR, 3.29; 95% CI, 1.33-8.17) were elevated and statistically significant. Among daily pipe users, we observed an elevated risk of mortality from all tobacco-related cancers (HR, 1.75; 95% CI, 1.16-2.64), and a nonstatistically significantly elevated risk of all-cause mortality (HR, 1.13; 95% CI, 0.96-1.34), and diabetes mortality (HR, 1.98; 95% CI, 0.98-4.01). For other mortality end points, however, there were generally too few deaths among daily and nondaily users to generate robust risk estimates.
Our analyses of NLMS data revealed that exclusive current cigarette smokers and exclusive current cigar smokers have higher all-cause mortality risks than never tobacco users. The risk of dying from tobacco-related cancers is higher for current and former exclusive cigarette smokers, current exclusive pipe smokers, and current exclusive cigar smokers than for never tobacco users. Elevated mortality risks also were found for former and current nondaily exclusive cigarette smokers for most causes of mortality assessed.
Specifically, we found an increased risk among cigar smokers for lung cancer and tobacco-related cancers overall compared with never tobacco users, consistent with previous results.13 We also found that current exclusive cigar smokers have increased all-cause mortality risk (HR, 1.20; 95% CI, 1.03-1.38), which is an outcome that, to our knowledge, has not been estimated for the US population since the Cancer Prevention Study-I (HR, 1.08; 95% CI, 1.05-1.12), using data collected more than 40 years ago.7 Among current daily users, we found increased COPD mortality risk for cigar smokers compared with never tobacco users (HR, 3.29; 95% CI, 1.33-8.17), a result that confirms a previous elevated, nonstatistically significant result among regular cigar users (HR, 1.42; 95% CI, 0.96-2.03).7 Our estimate of increased tobacco-related cancer mortality risk for current daily pipe smokers (HR, 1.75; 95% CI, 1.16- 2.64) is also consistent with previous results that found increased risks for cancers such as oropharyngeal (relative risk [RR], 3.90; 95% CI, 2.15-7.08), laryngeal (RR, 13.1; 95% CI, 5.20-33.1), esophageal (RR, 2.44; 95% CI, 1.51-3.95), and lung (RR, 5.00; 95% CI, 4.16-6.01).10 We observed lower risks for daily pipe and cigar smoking than for daily cigarette smoking. It is likely that daily cigar and pipe smokers use these products less frequently per day than cigarette users. Indeed, ever cigarette smoking on these questionnaires was defined as using at least 100 cigarettes in the lifetime; however, definitions for ever use of cigars and pipes did not use the same criteria. A 2015 review by Chang et al9 indicates that few additional studies examining cigar mortality risks have been published in recent years, strengthening the importance of our findings.
The HR for all-cause mortality among current smokers relative to never smokers in the American Cancer Society’s Cancer Prevention Study II (baseline year 1982), who were surveyed around the same time as the first participants in our study, was 2.33 (95% CI, 2.26- 2.40) for men and 2.08 (95% CI, 2.02-2.14) for women,1 compared with our overall all-cause mortality HR of 1.98 (95% CI, 1.93-2.02). The HRs for former smokers were 1.42 (95% CI, 1.38-1.45) for men and 1.33 (95% CI, 1.29-1.37) for women, compared with an HR of 1.32 (95% CI, 1.29-1.35) in this study. The comparability of our smoking-related mortality risk estimates with similar cohorts underscores the validity of our approach and highlights the addition of cigar- and pipe-related mortality risk estimates to the literature. It is informative to have contemporary mortality risk estimates for other combustible tobacco products.
This study has several strengths. The NLMS combined data set is a nationally representative prospective study. Tobacco use information is collected prior to death and sampling weights are used to provide representative RRs. In addition, the detailed TUS-CPS questionnaire allowed us to analyze current daily and nondaily tobacco users separately as well as former tobacco users. The size of this study enabled us to examine exclusive users of specific tobacco products and generate estimates of overall mortality, cancer mortality, and other major causes of death. Also, our study included follow-up from 1985 to 2011, spanning a more recent time period than previous studies. Our study is unique in that we harmonized tobacco use data, including information on frequency of use, from the national surveys over many years creating a large sample size to enable examination of associations of multiple combustible tobacco products and tobacco-related mortality outcomes.
There are also limitations to consider. Tobacco use questions in the TUS-CPS have changed over time; however, these slight changes in questionnaire wording likely did not substantially alter our findings, although classifying participants reporting pipe or cigar use at least once as ever-users in 2001-2002 may have attenuated results for nondaily and former users. Also, TUS-CPS data lacked detailed frequency information for cigar and pipe use over the years, which restricted our ability to examine frequency of tobacco use in greater detail than daily or nondaily. We were also unable to examine the influence of multiple tobacco product use in these data. Because the TUS-CPS spans several decades, mortality follow-up time across surveys varies; however, we examined results by survey year and the results were comparable. Quiz Ref IDIn addition, because tobacco use was only assessed at baseline, these estimates may not represent lifetime use patterns (eg, tobacco use cessation during follow-up would underestimate risks). Tobacco use information was collected primarily by self-report, with approximately 20% by proxy questionnaire, and therefore is potentially subject to errors in recall. However, studies have illustrated that self-report tobacco use information is reliable and valid.14 Likewise, given exclusion of teens and young adults, proxy data on tobacco use status is reasonably consistent with self-report for middle-aged and older adults.15 Our study lacked data on some potentially confounding variables, including alcohol, body weight, and family history of disease, although prior US studies have not indicated that these factors strongly confound associations between tobacco product use and disease.16 We also lacked information on use of water pipes and e-cigarettes; however, these products were uncommonly used during the study period. Finally, even given the large size of included surveys, there were limited numbers of deaths among some tobacco use strata, such as nondaily pipe smokers.
Our study provides further evidence that cigar, pipe, and cigarette use confers mortality risks, even among nondaily current cigarette users, with lower risks observed among former users than current users. These data underscore the importance of cessation to reduce mortality and morbidity from combustible tobacco use. These risk estimates are consistent with published estimates from other studies, thus demonstrating the reliability of NLMS risk estimates. Given rapid changes in tobacco products available in the marketplace, our results provide information on the contemporary mortality risks of 2 noncigarette combustible tobacco products.
Accepted for Publication: December 20, 2017.
Corresponding Author: Carol H. Christensen, PhD, MPH, Office of Science, Center for Tobacco Products, Food and Drug Administration, 10903 New Hampshire Ave, Bldg 71, Room G335, Silver Spring, MD 20993 (firstname.lastname@example.org).
Published Online: February 19, 2018. doi:10.1001/jamainternmed.2017.8625
Author Contributions: Drs Christensen and Ms Cosgrove had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Christensen, Rostron, Altekruse, Apelberg, Inoue-Choi, Freedman.
Acquisition, analysis, or interpretation of data: Christensen, Rostron, Cosgrove, Altekruse, Hartman, Gibson, Inoue-Choi, Freedman.
Drafting of the manuscript: Christensen, Rostron, Cosgrove, Altekruse, Gibson, Inoue-Choi, Freedman.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Christensen, Rostron, Cosgrove, Gibson, Inoue-Choi, Freedman.
Obtained funding: Altekruse.
Administrative, technical, or material support: Christensen, Cosgrove, Altekruse, Inoue-Choi, Freedman.
Study supervision: Apelberg, Freedman.
Conflict of Interest Disclosures: None reported.
Census Bureau Disclaimer: This article is released to inform interested parties of research and to encourage discussion. Any views expressed on statistical, methodologic, technical, or operational issues are those of the authors and not necessarily those of the US Census Bureau, Food and Drug Administration (FDA), and National Institutes of Health (NIH).
Disclaimer: This publication represents the views of the authors and does not represent FDA/Center for Tobacco Products nor NIH/National Cancer Institute position or policy.
et al. 50-Year trends in smoking-related mortality in the United States. N Engl J Med
. 2013;368(4):351-364.PubMedGoogle ScholarCrossref
et al. 21st-century hazards of smoking and benefits of cessation in the United States. N Engl J Med
. 2013;368(4):341-350.PubMedGoogle ScholarCrossref
US Department of Health and Human Services. The Health Consequences of Smoking—50 Years of Progress: A Report of the Surgeon General, 2014. Washington, DC: US Dept of Health and Human Services; 2014.
Center for Behavioral Health Statistics and Quality Substance Abuse and Mental Health Services Administration. 2015 National Survey on Drug Use and Health: Detailed Tables. Rockville, MD: Substance Abuse and Mental Health Services Administration; 2015.
et al. Tobacco product use among adults—United States, 2013-2014. MMWR Morb Mortal Wkly Rep
. 2016;65(27):685-691.PubMedGoogle ScholarCrossref
B. Consumption of combustible and smokeless tobacco - United States, 2000-2015. MMWR Morb Mortal Wkly Rep
. 2016;65(48):1357-1363.PubMedGoogle ScholarCrossref
National Cancer Institute, ed. Cigars: Health Effects and Trends: Smoking and Tobacco Control Monograph. Vol 9. Bethesda, MD: National Cancer Institute; 1998.
B. Mortality and economic costs from regular cigar use in the United States, 2010. Am J Public Health
. 2014;104(9):e86-e91.PubMedGoogle ScholarCrossref
BJ. Systematic review of cigar smoking and all cause and smoking related mortality. BMC Public Health
. 2015;15:390.PubMedGoogle ScholarCrossref
EE. Association between exclusive pipe smoking and mortality from cancer and other diseases. J Natl Cancer Inst
. 2004;96(11):853-861.PubMedGoogle ScholarCrossref
GA. Pipe smoking in the United States, 1965-1991: prevalence and attributable mortality. Prev Med
. 1996;25(2):91-99.PubMedGoogle ScholarCrossref
J. National Longitudinal Mortality Survey Files and Technical Systems Reference Manual. Suitland, MD: Census Department; 2014.
MJ. Cigar smoking in men and risk of death from tobacco-related cancers. J Natl Cancer Inst
. 2000;92(4):333-337.PubMedGoogle ScholarCrossref
PD. Factors associated with discrepancies between self-reports on cigarette smoking and measured serum cotinine levels among persons aged 17 years or older: Third National Health and Nutrition Examination Survey, 1988-1994. Am J Epidemiol
. 2001;153(8):807-814.PubMedGoogle ScholarCrossref
LJ. On consistency of self- and proxy-reported regular smoking initiation age. J Subst Abus Alcohol
. 2013;1(1):1001.PubMedGoogle Scholar
SJ. Smoking vs other risk factors as the cause of smoking-attributable deaths: confounding in the courtroom. JAMA
. 2000;284(6):706-712.PubMedGoogle ScholarCrossref