What proportion of cancer deaths are attributable to cigarette smoking in each US state?
In this study of population-attributed fraction of cancer deaths due to cigarette smoking, cigarette smoking explained a high proportion of cancer deaths in all states, but this proportion was highest in several Southern states, notably Kentucky, Arkansas, Tennessee, West Virginia, and Louisiana.
Strengthened tobacco control is needed to reduce the burden of cancer death in all states.
State-specific information about the health burden of smoking is valuable because state-level initiatives are at the forefront of tobacco control. Smoking-attributable cancer mortality estimates are currently available nationally and by cancer, but not by state.
To calculate the proportion of cancer deaths among adults 35 years and older that were attributable to cigarette smoking in 2014 in each state and the District of Columbia.
Design, Setting, and Participants
The population-attributable fraction (PAF) of cancer deaths due to cigarette smoking was computed using relative risks for 12 smoking-related cancers (acute myeloid leukemia and cancers of the oral cavity and pharynx; esophagus; stomach; colorectum; liver; pancreas; larynx; trachea, lung, and bronchus; cervix uteri; kidney and renal pelvis; and urinary bladder) from large US prospective studies and state-specific smoking prevalence data from the Behavioral Risk Factor Surveillance System.
Main Outcomes and Measures
The PAF of cancer deaths due to cigarette smoking in each US state and the District of Columbia.
We estimate that at least 167 133 cancer deaths in the United States in 2014 (28.6% of all cancer deaths; 95% CI, 28.2%-28.8%) were attributable to cigarette smoking. Among men, the proportion of cancer deaths attributable to smoking ranged from a low of 21.8% in Utah (95% CI, 19.9%-23.5%) to a high of 39.5% in Arkansas (95% CI, 36.9%-41.7%), but was at least 30% in every state except Utah. Among women, the proportion ranged from 11.1% in Utah (95% CI, 9.6%-12.3%) to 29.0% in Kentucky (95% CI, 27.2%-30.7%) and was at least 20% in all states except Utah, California, and Hawaii. Nine of the top 10 ranked states for men and 6 of the top 10 ranked states for women were located in the South. In men, smoking explained nearly 40% of cancer deaths in the top 5 ranked states (Arkansas, Louisiana, Tennessee, West Virginia, and Kentucky). In women, smoking explained more than 26% of all cancer deaths in the top 5 ranked states, which included 3 Southern states (Kentucky, Arkansas, and Tennessee), and 2 Western states (Alaska and Nevada).
Conclusions and Relevance
The proportion of cancer deaths attributable to cigarette smoking varies substantially across states and is highest in the South, where up to 40% of cancer deaths in men are caused by smoking. Increasing tobacco control funding, implementing innovative new strategies, and strengthening tobacco control policies and programs, federally and in all states and localities, might further increase smoking cessation, decrease initiation, and reduce the future burden of morbidity and mortality associated with smoking-related cancers.
Smoking prevalence in the United States has been more than halved since the release of the first Surgeon General’s Report on the health hazards of cigarette smoking in 1964, as a result of increased awareness and implementation of public health policies against smoking.1(p33) Nevertheless, there are still 40 million current adult cigarette smokers, and smoking remains the largest preventable cause of death from cancer and other diseases.2 Cigarette smoking accounted for an estimated 28.7% of all cancer deaths in US adults 35 years and older in 2010.3 However, there are no such estimates by state, despite substantial geographic variation in smoking prevalence.4 State-specific smoking-attributable mortality is particularly valuable for public health advocates and policy makers because state-level initiatives are at the forefront of tobacco control efforts. Herein, we estimate the proportion of all cancer deaths explained by cigarette smoking in adults older than 35 years in each of the 50 states and the District of Columbia (DC). For convenience, we refer to DC as a state hereafter.
This analysis used deidentified publicly available data and thus is not considered human subjects research; no institutional review board approval was necessary. We estimated the state-specific proportion of cigarette smoking–attributable cancer mortality (SACM) using methods similar to those of the 2014 Surgeon General’s Report,1 based on 12 cancers caused by cigarette smoking (acute myeloid leukemia and cancers of the oral cavity and pharynx; esophagus; stomach; colorectum; liver; pancreas; larynx; trachea, lung, and bronchus; cervix uteri; kidney and renal pelvis; and urinary bladder). To avoid potential bias, we calculated the overall population-attributable fraction (PAF) for cancer deaths in each state using the weighted sums method.5 Specifically, we first calculated the PAF for each sex and age group (35-49, 50-54, 55-59, 60-64, 65-69, 70-74, 75-79, 80-84, ≥85 years) in each state, using the standard formula for multicategory exposure6:
where s represents age; p0, p1, and p2, the proportion of never, former, and current smokers, respectively; and RR1 and RR2, the relative risk for former and current smokers, respectively, compared with never smokers.
Age-, sex-, and state-specific smoking prevalence (never, former, or current) were calculated on the basis of data from the 2014 Behavioral Risk Factor Surveillance System survey (BRFSS), which is the only national survey designed to provide reliable state-level estimates of health behaviors.4 Smoking prevalence estimates were based on 372 759 survey participants 35 years and older who provided information on smoking status. These smoking prevalence estimates were generated from the weighted public data provided by the Centers for Disease Control and Prevention (CDC). Weighting was based on characteristics such as sex, age, race and ethnicity, education, and marital status to adjust for nonresponse bias and ensure that the sample was representative.7 Age- and sex-specific (but not state-specific) relative risks for death for current and former smoking status were those for a composite outcome of any of the 12 smoking-related cancers as reported from analyses of the Cancer Prevention Study-II (442 960 participants) and Pooled Contemporary Cohort (954 029 participants).3
For each state, the number of smoking-attributable cancer deaths in each age and sex group was calculated by multiplying the age- and sex-specific PAFs by the corresponding observed 2014 cancer death counts obtained from the National Center for Health Statistics.8 The total number of smoking-attributable cancer deaths in each state was then calculated by summing across all age and sex groups. Finally, the overall SACM in each state was calculated by dividing the number of estimated smoking-attributable cancer deaths by the total number of cancer deaths among persons 35 years and older in each state. The 95% confidence intervals on the SACM were estimated via a bootstrap method,9 with 5000 simulations.
To illustrate the geographic variation in SACM, we mapped the results grouping states by number rank (1 being the highest SACM).
Differences in SACM between states may be partly due to differences between states in racial and ethnic composition because smoking prevalence substantially varies by race/ethnicity.2 To compare a measure of SACM between states that was not influenced by racial and ethnic composition, we calculated SACM by state for non-Hispanic white (NHW) men. We then assessed whether variation in SACM across states in NHW men was similar to that for all races/ethnicities combined using the Spearman correlation. Sparse data precluded similar comparison for other racial/ethnic and sex groups. However, we also calculated national SACM estimates for NHWs, non-Hispanic blacks, and Hispanics using smoking prevalence from the National Health Interview Survey (NHIS) (51 637 participants 35 years and older) during 2013 to 201410 and relative risks of cancer death as described herein.
Finally, as a sensitivity analysis, we compared the SACM in 4 regions (South, Midwest, West, and Northeast according to the Bureau of Census classification)11 using smoking prevalence from the NHIS with that estimated using smoking prevalence from the BRFSS. We used Stata, version 13.1, and SAS 9.4 to perform the analyses. A 2-sided P value of .05 was used to determine statistical significance.
In 2014, at least 167 133 cancer deaths (28.6% of all cancer deaths) in persons older than 35 years in the United States were attributable to cigarette smoking, with 103 609 of these deaths occurring in men (62.0%) and 63 524 in women (38.0%) (Table). The proportion of SACM ranged from 21.8% in Utah to 39.5% in Arkansas among men, and from 11.1% in Utah to 29.0% in Kentucky among women. Many of the states with the highest proportions of SACM were located in the South, including 9 of the top 10 states for men (Arkansas, Louisiana, Tennessee, West Virginia, Kentucky, Alabama, Mississippi, North Carolina, and Oklahoma) (Figure 1). Notably, smoking explained nearly 40% of adult male cancer deaths in 5 of these states. Southern states dominated the top 10 SACM states among women as well, but the second- and third-ranked states were Alaska (27.5%) and Nevada (27.1%)—which ranked 18th and 20th, respectively, among men. For both sexes combined, 7 of the top 10 states were located in the South, 2 in the West (Alaska and Nevada), and 1 in the Midwest (Missouri). While California had the lowest SACM after Utah, it had the highest number of deaths explained by smoking, because of its large population.
State-specific rankings for NHW men were generally similar to those for all races/ethnicities combined (Spearman correlation coefficient = 0.91, P < .001), with the notable exception of DC (SACM of 18.5% in NHW men vs 33.3% in men overall) (eFigure 1 in the Supplement). Nationally, non-Hispanic blacks had the highest SACM (27.2%), followed by NHWs (26.0%) and Hispanics (19.8%) (Figure 2). Finally, estimates of SACM at the regional level using smoking prevalence from the NHIS were equivalent to those estimated using smoking prevalence from the BRFSS (eTable 1 in the Supplement).
In most states, approximately one-third of cancer deaths in men and one-quarter in women were explained by cigarette smoking. However, consistent with smoking-attributable all-cause mortality,12 cancer deaths were associated with cigarette smoking less often in Western states and more often in the South, particularly among men. For example, smoking accounted for nearly 40% of cancer deaths among men in 5 Southern states. The larger burden of SACM in men than in women likely reflects a lower prevalence of smoking among women than men in the older birth cohorts.13,14 However, sex differences in SACM may diminish in the future because smoking histories and risk of mortality from smoking-related diseases are comparable for men and women in more recent birth cohorts.15 In fact, female smoking prevalence recently surpassed male smoking prevalence in South Dakota, Montana, and Arkansas.16
Higher SACM in the South is driven by higher historic smoking prevalence, which has prevailed in large part due to weaker tobacco control policies and programs. Policy initiatives are heavily influenced by the tobacco industry in all states,17,18 especially those in the South,19 where 95% of the US tobacco crop is grown.20 Although spending on tobacco control is inversely associated with smoking prevalence,21,22 only 5 states spent at least 50% of the amount recommended by the CDC in 2016.23 In particular, 8 of the 21 states that spend less than 10% of the CDC-recommended amount are located in the South (Alabama, Georgia, Kentucky, North Carolina, South Carolina, Tennessee, Texas, and Virginia). Tobacco control spending by all states combined was less than $500 million in 2016, far less than the $10 billion spent annually by the tobacco industry on marketing.24
Public smoking restrictions and high cigarette prices (through excise taxes, price promotion restrictions, and minimum price laws)25 are among the most effective tobacco control policies,26,27 and both are primarily legislated by states. Again, the least restrictive public smoking policies and most affordable cigarettes are found in the South. Nine of 14 states with the least comprehensive smoke-free indoor air laws are in this region28 and the mean cigarette excise tax is $0.49 in major tobacco states, compared with $1.80 in other states (and as high as $4.35 in New York).29 However, there are signs that the tobacco industry’s influence has waned somewhat in Southern tobacco-growing states in recent years, facilitating improvement in tobacco control policies30 and highlighting the opportunity for more rapid progress in the future.
The higher SACM in Southern states may also be due in part to disproportionately high levels of low socioeconomic status, which is associated with higher smoking prevalence2 and lower smoking cessation rates.31 Smoking prevalence among adults with a high school education or less are 2 to 4 times those among college graduates,2 and people with a lower educational attainment are less aware of the health hazards of smoking.32 Only half (50%) of adults in Kentucky have more than a high school education, compared with 68% in Colorado.31 In addition, racial differences in smoking prevalence and population distribution may account for some variation in the SACM by state. For example, black men have a higher SACM and a higher proportion of smoking-attributable all-cause mortality,33 reflecting historically higher smoking prevalence compared with white men.34 In some Southern states (eg, Louisiana, Mississippi), blacks account for more than 30% of the population compared with less than 5% in many Western and Northern states (eg, Utah, Connecticut).35 Conversely, some states, such as California and Texas, are disproportionately populated by Hispanics,35 among whom SACM is lower. Nevertheless, the SACM by state for NHW men is generally similar to that of all men, indicating that variation in racial composition is unlikely to be the driving factor for state differences in SACM. Of note, DC showed the lowest SACM for NHW men, reflecting the large proportion of highly educated men (85% with a bachelor’s degree or more)31 in whom awareness about the health hazards of smoking is highest.32 The comparatively low SACM in Utah reflects the religious prohibition of smoking among Mormons.36
In addition to Southern states, Alaska and Nevada had particularly high SACM, especially among women. In Alaska, which had the second highest SACM in women, smoking prevalence was the same in men and women in 2009, in contrast to most states where it was 10% to 60% higher in men.16 Nevada is one of a handful of non-Southern states that still allows smoking in bars and casinos.37 A previous study of smoking-attributable all-cause mortality found that Nevada had the highest fraction of deaths explained by smoking of any state.38 Missouri is another non-Southern state with high SACM, ranking seventh for both sexes combined. It has the lowest cigarette excise tax ($0.17) of any state, 90% lower than the national mean of $1.65.29
Tobacco control has been credited with preventing approximately 8 million premature deaths in the United States over the past 5 decades, equivalent to 157 million years of life saved.39 Our data show that there remains the potential to avert many more premature deaths in light of suboptimal funding for tobacco control programs, not only in the South, but in all states. As of 2016, two-thirds of states lack 100% smoke-free laws in public places to protect the general public from second-hand smoke40; no state41 has taxes on cigarettes that account for at least 75% of the retail price, as recommended by the World Health Organization42; and only 1 state (North Dakota) funds its tobacco control programs at the level recommended by the CDC.23 The Affordable Care Act includes coverage of cessation treatments without cost-sharing for the privately and Medicare insured. However, for Medicaid enrollees—who are twice as likely to smoke2—coverage is state governed, and only 7 states provide comprehensive coverage (Connecticut, Indiana, Massachusetts, Minnesota, Nevada, Pennsylvania, and Vermont).43 Although there has generally been a stagnation in the adoption of traditional comprehensive tobacco control,44 some states and localities have implemented innovative approaches to fight the tobacco epidemic. For instance, California, Hawaii, and 145 smaller localities have increased the tobacco sales age to 21 years45—a measure supported by the Institute of Medicine.46 Likewise, communities across the United States have passed laws that limit or prohibit smoking in multifamily housing.47 The federal government can do more to accelerate cessation and discourage initiation, including requiring manufacturers of tobacco products to reduce nicotine content to nonaddictive levels,48 increasing federal tobacco taxes, and maintaining funding of antismoking campaigns.49 With fully one-third of tobacco-related cancer deaths in men and one-quarter in women preventable with current knowledge, tobacco control should spearhead the Cancer Moonshot initiative to accelerate progress against cancer. However, it is important to realize that given the lag time between tobacco use and cancer diagnosis,50 the impact of today’s policies will be most evident on the future cancer burden.
Our study likely underestimated deaths caused by tobacco use for several reasons. First, only 12 cancers were included, for consistency with the Surgeon General’s report1; however, cigarette smoking is associated with excess mortality for additional cancers.3,51 Second, self-reported data are known to underestimate smoking prevalence.52 Third, deaths caused by tobacco exposures other than active cigarette smoking, including second-hand smoke, pipes, hookahs, cigars, smokeless tobacco, and electronic nicotine delivery systems, were not included in our analysis. Due to changing patterns of tobacco use,53 products other than cigarettes may account for a greater proportion of all tobacco-related cancer deaths in the future. Finally, confidence intervals for SACM in some states were relatively wide due to limited precision of smoking prevalence estimates available from the BRFSS in some age groups. However, the BRFSS is the only national survey designed to provide estimates of state-level smoking status. Although the response rate for the BRFSS is lower (47%) than that for the NHIS (61%),10 the surveys report generally comparable smoking prevalence estimates,54 which generate similar SACM when compared at the regional level. Notably, higher SACM was less apparent in the Census Bureau–defined Southern region because it includes states such as Maryland, which has exceptionally low smoking prevalence (16.4% in 2013),55 and Texas, which has a large lower-smoking Hispanic population.56 This illustrates the high variability of smoking-attributable disease within regions and supports the value of state-specific analyses.
The proportion of cancer deaths attributable to cigarette smoking varies substantially across states and is highest in the South, where up to 40% of cancer deaths in men are caused by smoking. However, the human costs of cigarette smoking are high in all states, regardless of ranking. Increasing tobacco control funding, implementing innovative new strategies, and strengthening tobacco control policies and programs, federally and in all states and localities, might further increase smoking cessation, decrease initiation, and reduce the future burden of smoking-related cancers.
Accepted for Publication: August 24, 2016.
Corresponding Author: Joannie Lortet-Tieulent, MSc, Surveillance and Health Services Research, American Cancer Society, 250 Williams St, NW, Atlanta, GA 30303 (email@example.com).
Published Online: October 24, 2016. doi:10.1001/jamainternmed.2016.6530
Author Contributions: Mss Lortet-Tieulent and Goding Sauer 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. Both senior authors, Drs Jacobs and Jemal, contributed equally.
Study concept and design: Jemal.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Lortet-Tieulent, Goding Sauer, Jemal.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Lortet-Tieulent, Goding Sauer, Miller, Fedewa, Jacobs.
Study supervision: Jemal.
Conflict of Interest Disclosures: None reported.
Funding/Support: This work was supported by the Intramural Research Department of the American Cancer Society. All authors work in the Intramural Research Department.
Role of the Funder/Sponsor: All authors work for the American Cancer Society, but the management of the American Cancer Society played no part in the decision to perform the analysis and to submit the manuscript for publication.
Additional Contributions: We thank W. Dana Flanders, MD, DSc, MPH, Emory University, for his statistical support and Michal Stoklosa, MA, American Cancer Society, for advice on tobacco control. They did not receive compensation for these contributions.
US Department of Health and Human Services. The Health Consequences of Smoking—50 Years of Progress: A Report of the Surgeon General. Atlanta, GA: US Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health; 2014.
et al. Current cigarette smoking among adults—United States, 2005-2014. MMWR Morb Mortal Wkly Rep
. 2015;64(44):1233-1240.PubMedGoogle ScholarCrossref
et al. What proportion of cancer deaths in the contemporary United States is attributable to cigarette smoking? Ann Epidemiol
. 2015;25(3):179-182.e1, e171.PubMedGoogle ScholarCrossref
Centers for Disease Control and Prevention. Behavioral Risk Factor Surveillance System. Atlanta, GA: US Department of Health and Human Services, Centers for Disease Control and Prevention. 2014. http://www.cdc.gov/brfss/
. Accessed September 19, 2016.
S. Interval estimation by simulation as an alternative to and extension of confidence intervals. Int J Epidemiol
. 2004;33(6):1389-1397.PubMedGoogle ScholarCrossref
Centers for Disease Control and Prevention (CDC). State-specific smoking-attributable mortality and years of potential life lost—United States, 2000-2004. MMWR Morb Mortal Wkly Rep
. 2009;58(2):29-33.PubMedGoogle Scholar
JE. Cigarette smoking among successive birth cohorts of men and women in the United States during 1900-80. J Natl Cancer Inst
. 1983;71(3):473-479.PubMedGoogle Scholar
et al. 50-year trends in smoking-related mortality in the United States. N Engl J Med
. 2013;368(4):351-364.PubMedGoogle ScholarCrossref
Centers for Disease Control and Prevention (CDC). State-specific prevalence of cigarette smoking and smokeless tobacco use among adults—United States, 2009. MMWR Morb Mortal Wkly Rep
. 2010;59(43):1400-1406.PubMedGoogle Scholar
SA. Tobacco lobby political influence on US state legislatures in the 1990s. Tob Control
. 2001;10(2):124-134.PubMedGoogle ScholarCrossref
US v Philip Morris. Civil action No. 99-2496 (GK) (DC Cir 2006).
SA. Multiple streams approach to tobacco control policymaking in a tobacco-growing state. J Community Health
. 2014;39(4):633-645.PubMedGoogle ScholarCrossref
D. The impact of tobacco control programs on adult smoking. Am J Public Health
. 2008;98(2):304-309.PubMedGoogle ScholarCrossref
et al. State tobacco control spending and youth smoking. Am J Public Health
. 2005;95(2):338-344.PubMedGoogle ScholarCrossref
Robert Wood Johnson Foundation, Campaign for Tobacco-Free Kids, American Cancer Society Cancer Action Network, American Heart Association, American Lung Association, Americans for Nonsmokers’ Rights. Broken promises to our children: a state-by-state look at the 1998 state tobacco settlement 17 years later. Washington, DC; 2015. http://www.tobaccofreekids.org/microsites/statereport2016/
. Accessed September 19, 2016.
KM. Beyond excise taxes: a systematic review of literature on non-tax policy approaches to raising tobacco product prices. Tob Control
. 2016;25(4):377-385.PubMedGoogle ScholarCrossref
Centers for Disease Control and Prevention (CDC). State smoke-free laws for worksites, restaurants, and bars—United States, 2000-2010. MMWR Morb Mortal Wkly Rep
. 2011;60(15):472-475.PubMedGoogle Scholar
SA. Tobacco-control policies in tobacco-growing states: where tobacco was king. Milbank Q
. 2015;93(2):319-358.PubMedGoogle ScholarCrossref
BW. Smoking knowledge and behavior in the United States: sociodemographic, smoking status, and geographic patterns. Nicotine Tob Res
. 2008;10(10):1559-1570.PubMedGoogle ScholarCrossref
TF; Centers for Disease Control and Prevention (CDC). Cigarette smoking—United States, 1965-2008. MMWR Suppl
. 2011;60(1):109-113.PubMedGoogle Scholar
C. Religion and reduced cancer risk: what is the explanation? a review. Eur J Cancer
. 2008;44(17):2573-2579.PubMedGoogle ScholarCrossref
et al. Surveillance for smoking-attributable mortality and years of potential life lost, by state—United States, 1990. MMWR CDC Surveill Summ
. 1994;43(1):1-8.PubMedGoogle Scholar
et al. Tobacco control and the reduction in smoking-related premature deaths in the United States, 1964-2012. JAMA
. 2014;311(2):164-171.PubMedGoogle ScholarCrossref
et al. State Medicaid Coverage for tobacco cessation treatments and barriers to coverage—United States, 2014-2015. MMWR Morb Mortal Wkly Rep
. 2015;64(42):1194-1199.PubMedGoogle ScholarCrossref
SD. Stuck in neutral: stalled progress in statewide comprehensive smoke-free laws and cigarette excise taxes, United States, 2000-2014. Prev Chronic Dis
. 2016;13:E80.PubMedGoogle ScholarCrossref
Institute of Medicine. Public Health Implications of Raising the Minimum Age of Legal Access to Tobacco Products. Washington, DC: Institute of Medicine; 2015.
HK. It shouldn’t take another 50 years: accelerating momentum to end the tobacco epidemic. Clin Cancer Res
. 2014;20(7):1719-1720.PubMedGoogle ScholarCrossref
AD. Stages of the cigarette epidemic on entering its second century. Tob Control
. 2012;21(2):96-101.PubMedGoogle ScholarCrossref
et al. Smoking and mortality—beyond established causes. N Engl J Med
. 2015;372(7):631-640.PubMedGoogle ScholarCrossref
M. The accuracy of self-reported smoking: a systematic review of the relationship between self-reported and cotinine-assessed smoking status. Nicotine Tob Res
. 2009;11(1):12-24.PubMedGoogle ScholarCrossref
et al. Tobacco product use among adults—United States, 2013-2014. MMWR Morb Mortal Wkly Rep
. 2016;65(27):685-691.PubMedGoogle ScholarCrossref
P. Tracking chronic disease and risk behavior prevalence as survey participation declines: statistics from the behavioral risk factor surveillance system and other national surveys. Prev Chronic Dis
. 2008;5(3):A80.PubMedGoogle Scholar
L; Centers for Disease Control and Prevention (CDC). State-specific prevalence of current cigarette smoking and smokeless tobacco use among adults aged ≥18 years—United States, 2011-2013. MMWR Morb Mortal Wkly Rep
. 2015;64(19):532-536.PubMedGoogle Scholar