JAMA Health Forum – Health Policy, Health Care Reform, Health Affairs | JAMA Health Forum | JAMA Network
[Skip to Navigation]
Sign In
Figure 1.  Estimated Annual Prevalence of Smoking in the US Using the Baseline and Graphic Health Warning Scenarios, 2022 to 2100
Estimated Annual Prevalence of Smoking in the US Using the Baseline and Graphic Health Warning Scenarios, 2022 to 2100

Current smokers include individuals who had smoked within the past 2 years and recently quit. Shading represents the ranges of the scenarios. NHIS indicates National Health Interview Surveys.

Figure 2.  Cumulative Smoking-Attributable Deaths Averted in the US Using a Graphic Health Warning Scenario, 2022 to 2100
Cumulative Smoking-Attributable Deaths Averted in the US Using a Graphic Health Warning Scenario, 2022 to 2100

The solid line represents the main policy scenario (initiation reduced by 10% and cessation increased by 50% in 2022). Shading represents the most optimistic (initiation reduced by 15% and cessation increased by 75% in 2022) and conservative (initiation reduced by 5% and cessation increased by 25% in 2022) scenarios.

Figure 3.  Cumulative Life-Years Gained in the US Using a US Graphic Health Warning Scenario, 2022 to 2100
Cumulative Life-Years Gained in the US Using a US Graphic Health Warning Scenario, 2022 to 2100

The solid line represents the main policy scenario (initiation reduced by 10% and cessation increased by 50% in 2022). Shading indicates the most optimistic (initiation reduced by 15% and cessation increased by 75% in 2022) and conservative (initiation reduced by 5% and cessation increased by 25% in 2022) scenarios.

Table.  Estimated Cumulative Smoking-Attributable Deaths Averted and Life-Years Gained From 2012 to 2100 in the US
Estimated Cumulative Smoking-Attributable Deaths Averted and Life-Years Gained From 2012 to 2100 in the US
1.
US Food & Drug Administration. Cigarette health warnings. 2020. Accessed January 6, 2021. https://www.fda.gov/tobacco-products/labeling-and-warning-statements-tobacco-products/cigarette-health-warnings
2.
Public Health and Tobacco Policy Center. FDA graphic warnings. 2020. Accessed July 23, 2021. https://tobaccopolicycenter.org/tobacco-control/tobacco-litigation/fda-graphic-warnings/
3.
Levy  DT, Mays  D, Yuan  Z, Hammond  D, Thrasher  JF.  Public health benefits from pictorial health warnings on US cigarette packs: a SimSmoke simulation.   Tob Control. 2017;26(6):649-655. doi:10.1136/tobaccocontrol-2016-053087 PubMedGoogle ScholarCrossref
4.
Levy  DT, Warner  KE, Cummings  KM,  et al.  Examining the relationship of vaping to smoking initiation among US youth and young adults: a reality check.   Tob Control. 2019;28(6):629-635. doi:10.1136/tobaccocontrol-2018-054446 PubMedGoogle ScholarCrossref
5.
Meza  R, Jimenez-Mendoza  E, Levy  DT.  Trends in tobacco use among adolescents by grade, sex, and race, 1991-2019.   JAMA Netw Open. 2020;3(12):e2027465-e2027465. doi:10.1001/jamanetworkopen.2020.27465 PubMedGoogle ScholarCrossref
6.
Jeon  J, Meza  R, Krapcho  M, Clarke  LD, Byrne  J, Levy  DT. Actual and counterfactual smoking prevalence rates in the U.S. population via microsimulation. Risk Anal. 2012;32(suppl 1):S51-S68. doi:10.1111/j.1539-6924.2011.01775.x
7.
Moolgavkar  SH, Holford  TR, Levy  DT,  et al.  Impact of reduced tobacco smoking on lung cancer mortality in the United States during 1975-2000.   J Natl Cancer Inst. 2012;104(7):541-548. doi:10.1093/jnci/djs136 PubMedGoogle ScholarCrossref
8.
Jeon  J, Holford  TR, Levy  DT,  et al.  Smoking and lung cancer mortality in the United States from 2015 to 2065: a comparative modeling approach.   Ann Intern Med. 2018;169(10):684-693. doi:10.7326/M18-1250 PubMedGoogle ScholarCrossref
9.
Tam  J, Levy  DT, Jeon  J,  et al.  Projecting the effects of tobacco control policies in the USA through microsimulation: a study protocol.   BMJ Open. 2018;8(3):e019169. doi:10.1136/bmjopen-2017-019169 PubMedGoogle Scholar
10.
Holford  TR, Levy  DT, McKay  LA,  et al.  Patterns of birth cohort-specific smoking histories, 1965-2009.   Am J Prev Med. 2014;46(2):e31-e37. doi:10.1016/j.amepre.2013.10.022 PubMedGoogle ScholarCrossref
11.
Creamer  MR, Wang  TW, Babb  S,  et al.  Tobacco product use and cessation indicators among adults—United States, 2018.   MMWR Morb Mortal Wkly Rep. 2019;68(45):1013-1019. doi:10.15585/mmwr.mm6845a2 PubMedGoogle ScholarCrossref
12.
Rosenberg  MA, Feuer  EJ, Yu  B,  et al Chapter 3: cohort life tables by smoking status, removing lung cancer as a cause of death. Risk Anal. 2012;32(suppl 1):S25-S38. doi:10.1111/j.1539-6924.2011.01662.x
13.
Institute of Medicine.  Health Implications of Raising the Minimum Age for Purchasing Tobacco Products. The National Academies Press; 2015.
14.
LCFIT. LCFIT: The Lee-Carter Fitter. 2009. Accessed July 17, 2021. http://lcfit.demog.berkeley.edu/
15.
Thun  MJ, Carter  BD, Feskanich  D,  et al.  50-year trends in smoking-related mortality in the United States.   N Engl J Med. 2013;368(4):351-364. doi:10.1056/NEJMsa1211127 PubMedGoogle ScholarCrossref
16.
Hammond  D, Reid  JL, Driezen  P,  et al.  Are the same health warnings effective across different countries? an experimental study in seven countries.   Nicotine Tob Res. 2019;21(7):887-895. doi:10.1093/ntr/nty248 PubMedGoogle ScholarCrossref
17.
Drovandi  A, Teague  P-A, Glass  B, Malau-Aduli  B.  A systematic review of the perceptions of adolescents on graphic health warnings and plain packaging of cigarettes.   Syst Rev. 2019;8(1):25. doi:10.1186/s13643-018-0933-0 PubMedGoogle ScholarCrossref
18.
Brewer  NT, Hall  MG, Noar  SM,  et al.  Effect of pictorial cigarette pack warnings on changes in smoking behavior: a randomized clinical trial.   JAMA Intern Med. 2016;176(7):905-912. doi:10.1001/jamainternmed.2016.2621 PubMedGoogle ScholarCrossref
19.
Noar  SM, Francis  DB, Bridges  C, Sontag  JM, Ribisl  KM, Brewer  NT.  The impact of strengthening cigarette pack warnings: systematic review of longitudinal observational studies.   Soc Sci Med. 2016;164(164):118-129. doi:10.1016/j.socscimed.2016.06.011 PubMedGoogle Scholar
20.
Warner  KE, Mendez  D. How much of the future mortality toll of smoking can be avoided? Tob Control. Published online June 16, 2020. doi:10.1136/tobaccocontrol-2019-055530
21.
Caro  JJ, Briggs  AH, Siebert  U, Kuntz  KM; ISPOR-SMDM Modeling Good Research Practices Task Force.  Modeling good research practices—overview: a report of the ISPOR-SMDM Modeling Good Research Practices Task Force-1.   Med Decis Making. 2012;32(5):667-677. doi:10.1177/0272989X12454577 PubMedGoogle ScholarCrossref
22.
Siebert  U, Alagoz  O, Bayoumi  AM,  et al.  State-transition modeling: a report of the ISPOR-SMDM Modeling Good Research Practices Task Force–3.   Med Decis Making. 2012;32(5):690-700. doi:10.1177/0272989X12455463 PubMedGoogle ScholarCrossref
23.
Feirman  SP, Donaldson  E, Glasser  AM,  et al.  Mathematical modeling in tobacco control research: initial results from a systematic review.   Nicotine Tob Res. 2016;18(3):229-242. doi:10.1093/ntr/ntv104 PubMedGoogle ScholarCrossref
24.
Feirman  SP, Glasser  AM, Rose  S,  et al.  Computational models used to assess US tobacco control policies.   Nicotine Tob Res. 2017;19(11):1257-1267. doi:10.1093/ntr/ntx017 PubMedGoogle ScholarCrossref
25.
National Cancer Institute Cancer Intervention and Surveillance Modeling Network. CISNET publication support and modeling resources. 2018. Accessed July 17, 2021. https://resources.cisnet.cancer.gov/projects/.
26.
Alcaraz  A, Hernández-Vásquez  A, Palacios  A,  et al.  Health and economic impact of health warnings and plain tobacco packaging in seven Latin American countries: results of a simulation model.   Nicotine Tob Res. 2020;22(11):2032-2040. doi:10.1093/ntr/ntaa104 PubMedGoogle ScholarCrossref
27.
US Food & Drug Administration. FDA commits to evidence-based actions aimed at saving lives and preventing future generations of smokers. April 29, 2021. Accessed July 1, 2021. https://www.fda.gov/news-events/press-announcements/fda-commits-evidence-based-actions-aimed-saving-lives-and-preventing-future-generations-smokers
28.
Apelberg  BJ, Feirman  SP, Salazar  E,  et al.  Potential public health effects of reducing nicotine levels in cigarettes in the United States.   N Engl J Med. 2018;378(18):1725-1733. doi:10.1056/NEJMsr1714617 PubMedGoogle ScholarCrossref
29.
Noar  SM, Barker  J, Bell  T, Yzer  M.  Does perceived message effectiveness predict the actual effectiveness of tobacco education messages? a systematic review and meta-analysis.   Health Commun. 2020;35(2):148-157. doi:10.1080/10410236.2018.1547675 PubMedGoogle ScholarCrossref
30.
Noar  SM, Hall  MG, Francis  DB, Ribisl  KM, Pepper  JK, Brewer  NT.  Pictorial cigarette pack warnings: a meta-analysis of experimental studies.   Tob Control. 2016;25(3):341-354. doi:10.1136/tobaccocontrol-2014-051978 PubMedGoogle ScholarCrossref
31.
Hammond  D.  Health warning messages on tobacco products: a review.   Tob Control. 2011;20(5):327-337. doi:10.1136/tc.2010.037630 PubMedGoogle ScholarCrossref
32.
Yong  HH, Borland  R, Thrasher  JF,  et al Mediational pathways of the impact of cigarette warning labels on quit attempts. Health Psychol. 2014;33(11):1410-1420. doi:10.1037/hea0000056
33.
Noar  SM, Francis  DB, Bridges  C, Sontag  JM, Brewer  NT, Ribisl  KM.  Effects of strengthening cigarette pack warnings on attention and message processing: a systematic review.   Journal Mass Commun Q. 2017;94(2):416-442. doi:10.1177/1077699016674188 PubMedGoogle ScholarCrossref
34.
Thrasher  JF, Brewer  NT, Niederdeppe  J,  et al.  Advancing tobacco product warning labels research methods and theory: a summary of a grantee meeting held by the US National Cancer Institute.   Nicotine Tob Res. 2019;21(7):855-862. doi:10.1093/ntr/nty017 PubMedGoogle ScholarCrossref
35.
Huang  J, Chaloupka  FJ, Fong  GT.  Cigarette graphic warning labels and smoking prevalence in Canada: a critical examination and reformulation of the FDA regulatory impact analysis.   Tob Control. 2014;23(suppl 1):i7-i12. doi:10.1136/tobaccocontrol-2013-051170 PubMedGoogle ScholarCrossref
36.
Azagba  S, Sharaf  MF.  The effect of graphic cigarette warning labels on smoking behavior: evidence from the Canadian experience.   Nicotine Tob Res. 2013;15(3):708-717. doi:10.1093/ntr/nts194 PubMedGoogle ScholarCrossref
Limit 200 characters
Limit 25 characters
Conflicts of Interest Disclosure

Identify all potential conflicts of interest that might be relevant to your comment.

Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.

Err on the side of full disclosure.

If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.

Not all submitted comments are published. Please see our commenting policy for details.

Limit 140 characters
Limit 3600 characters or approximately 600 words
    Views 3,382
    Original Investigation
    September 24, 2021

    Estimated Prevalence of Smoking and Smoking-Attributable Mortality Associated With Graphic Health Warnings on Cigarette Packages in the US From 2022 to 2100

    Author Affiliations
    • 1Department of Health Policy and Management, Yale University School of Public Health, New Haven, Connecticut
    • 2Department of Epidemiology, University of Michigan School of Public Health, Ann Arbor
    • 3Department of Health Promotion, Education, and Behavior, Arnold School of Public Health, University of South Carolina, Columbia
    • 4School of Public Health and Health Systems, University of Waterloo, Waterloo, Ontario, Canada
    • 5Department of Biostatistics, Yale University School of Public Health, New Haven, Connecticut
    • 6Lombardi Comprehensive Cancer Center, Georgetown University Medical Center, Washington, DC
    JAMA Health Forum. 2021;2(9):e212852. doi:10.1001/jamahealthforum.2021.2852
    Key Points

    Question  What are the estimated population health outcomes associated with implementation of graphic health warnings on cigarette packages in 2022 and with delayed implementation of the warnings since 2012, as originally planned, in the US?

    Findings  In this decision analytical model using simulation modeling of smoking prevalence and smoking-attributable mortality in the US from 2012 to 2100, graphic health warnings, if implemented from 2022 to 2100, would be associated with 539 000 smoking-attributable deaths averted and 7.9 million life-years gained, and if implemented in 2012, with 718 000 deaths averted and 11.2 million life-years gained through 2100.

    Meaning  These findings suggest that graphic health warnings on cigarette packages may be associated with important public health benefits, and industry litigation and procedural delays to implementation may have been harmful for public health.

    Abstract

    Importance  Starting in 2022, the US Food and Drug Administration (FDA) plans to require all cigarette packages in the US to display graphic health warnings depicting health harms associated with smoking. The FDA originally planned to implement such warnings in 2012, but tobacco industry litigation delayed the effort.

    Objective  To assess the estimated population health outcomes associated with a policy requiring graphic health warnings on cigarette packages in the US and with a 10-year delay in implementation.

    Design, Setting, and Participants  This decision analytical model used simulation modeling of smoking prevalence and smoking-attributable mortality in the US from 2012 to 2100, using the Cancer Intervention and Surveillance Modeling Network smoking population model. The study was conducted from October 2020 to July 2021.

    Main Outcomes and Measures  The primary outcomes were annual adult smoking prevalence, smoking-attributable deaths averted, and life-years gained vs the baseline scenario. A baseline scenario assuming no graphic health warnings was compared with expected outcomes of implementing graphic health warnings in 2022 vs 2012. Policy effects were considered under varying assumptions of the association of the policy with smoking initiation and cessation, ranging from most conservative to most optimistic. A maximum smoking reduction scenario in which all smoking would stop by the end of 2022 was evaluated.

    Results  In the baseline scenario, an estimated 13.2 million smoking-attributable deaths would have occurred from 2012 to 2100. Under a maximum smoking-reduction scenario, 5.5 million of these deaths would be averted. Implementation of graphic health warnings from 2022 to 2100 would be associated with 539 000 (range, 275 000-794 000) smoking-attributable deaths averted and 7.9 million (range, 4.0-11.6 million) life-years gained, representing less than 10% of the 5.5 million estimated smoking-attributable deaths averted and 81.8 million life-years gained in the maximum smoking-reduction scenario. Implementation from 2012 to 2100 would be associated with 33.2% (range, 32.9%-33.5%) more deaths averted (718 000; range, 365 000 to 1.1 million) and 42.7% (range, 42.3%-43.1%) more life-years gained (11.2 million; range, 5.7-16.6 million) compared with implementation in 2022.

    Conclusions and Relevance  This decision analytical model estimated that FDA cigarette graphic health warnings, if implemented in 2022, would be associated with public health benefits. The model also estimated that more smoking-attributable deaths would have been averted if the policy had been implemented in 2012. Industry litigation and delays to implementing tobacco regulations may have been harmful for public health.

    Introduction

    Cigarette health warnings in the US have remained unchanged for the past 35 years, but starting in 2022, pending legal challenges, the US Food and Drug Administration (FDA) will require cigarette manufacturers to display graphic health warnings covering 50% of the front and back of cigarette packages. These warnings will also be required in cigarette advertisements. Cigarette health warnings will feature 1 of 13 graphic images that depict some of the negative health consequences of smoking and brief warning statements (eg, “Smoking causes COPD, a lung disease that can be fatal”).1 The FDA originally planned to implement graphic health warnings in 2012, but tobacco industry litigation and resulting court decisions struck down the initial warnings, delaying implementation.1 Public health organizations and physicians filed a lawsuit against the FDA for these delays, and a court-ordered deadline forced the FDA to finalize new graphic health warnings nearly a decade later.2 The tobacco industry has also issued repeated legal challenges to the revised warning designs, and the implementation date is contingent on further court rulings.

    Previous estimates of the potential population health benefits of graphic health warnings in the US were based on smoking prevalence trends through 2016.3 Since then, the prevalence of smoking—especially among youths and young adults—has decreased more than expected.4,5 Recent shifts in smoking trends may affect the outcomes associated with new regulations, including graphic health warnings. Past estimates also have not separately evaluated the outcomes for specific birth cohorts. This study used a simulation model to estimate smoking prevalence and mortality outcomes associated with graphic health warnings in the US under the planned policy for 2022 and compared these with estimated outcomes if the policy had been implemented in 2012.

    Methods
    Smoking History Generator Population Model

    This decision analytical model simulated smoking and mortality outcomes associated with graphic health warnings in the US using the Cancer Intervention and Surveillance Modeling Network (CISNET) Smoking History Generator Population Model, a discrete deterministic compartmental model version of the CISNET Smoking History Generator microsimulation model.6-9 The model simulated the annual number of never smokers, current smokers, and former smokers in the US by age (0-99 years), gender (male, female), birth cohort (1864-2100), and calendar year (1964-2100). Former smokers were further categorized by the number of years since quitting smoking (from 1 to 40 or more). Age, birth cohort, and calendar year were modeled in 1-year increments. A schematic representation of the model is shown in the eFigure 1 in the Supplement. In the model, simulated never smokers became current smokers and current smokers became former smokers based on annual smoking initiation and cessation probabilities. Further details are provided in the eAppendix in the Supplement. Data were analyzed from October 2020 to July 2021. This study did not involve human participants or private information and was therefore exempt from institutional review board review.

    Smoking Initiation and Cessation

    Annual probabilities of smoking initiation and cessation by age, gender, calendar year, and birth cohort were derived from the CISNET Lung Working Group using National Health Interview Surveys from 1965 to 2018.8,10 Current smoking was defined as having smoked at least 100 cigarettes in a lifetime and having smoked at any time in the past 2 years. This stringent definition of current smoking avoided the need to model relapse among individuals who had recently quit smoking but produced smoking prevalence estimates that were slightly greater than standard survey estimates.11 Annual probabilities for the birth cohorts beginning with the cohort born in 1864 were derived using age-period-cohort models of National Health Interview Surveys data on smoking initiation and cessation and on the prevalence of never, current, and former smoking status.8-10 In the baseline scenario, probabilities of initiating smoking were projected into the future by holding them fixed at the estimated level for the 2000 birth cohort (ie, individuals aged 18 years at the last survey in 2018). Analogously, probabilities of smoking cessation were projected into the future by holding these fixed at the value estimated for the 1990 birth cohort (ie, individuals aged 28 years at the last survey in 2018).8,9

    Mortality

    Population dynamics were based on US Census birth estimates and on mortality probabilities generated by the CISNET Lung Working Group that varied by age, gender, birth cohort, and smoking status (never, current, or former smoker by years since quitting).8,12,13 Future mortality rates by cohort, age, and gender were projected using the Lee-Carter method and then partitioned into mortality rates for never, current, and former smokers using the approach of Rosenberg et al.12-14

    For mortality among former smokers, we used probabilities as a function of age (a), year (y), and years since quitting (ysq) based on the CISNET probabilities of mortality among never and current smokers. Specifically, the probability of mortality for a former smoker (ufs) was calculated by taking the maximum between the probability of mortality for a never smoker (uns) of the same age and year and the probability of mortality for a current smoker (ucs) of the same age and year multiplied by the relative risk (RR) of mortality for a former smoker compared with mortality for current smoker by years since quitting (RRfs): ufs(a,y,ysq) = max (ucs[a,y] × RRfs[ysq], uns[a,y]).

    Estimates of the RR of all-cause mortality among former smokers compared with current smokers by the years-since-quitting categories were derived from Thun et al.15 We transformed these into single-year estimates by taking the midpoint of each years-since-quitting category and interpolating it using an exponential model, constraining the (RRfs) to be between 0 and 1 (eTable 1 and eFigures 2 and 3 in the Supplement). This approach ensured that the probability of mortality for a former smoker (ufs) was between those for a never smoker (uns) and current smoker (ucs). The relative risk of mortality among former smokers who quit more than 40 years ago was held constant at the relative risk among former smokers who quit 40 years ago.

    Policy Parameters

    We modeled (1) a baseline scenario without graphic health warnings, (2) graphic health warning policy effects that modified underlying initiation and cessation probabilities based on a warning implementation date of January 1, 2022, and (3) graphic health warning policy effects if the warnings had been implemented on January 1, 2012. The policy effect sizes were based on estimates developed by previous researchers3,9 and were consistent with those of other studies.16-19 However, because policies in the Smoking History Generator act through changes in smoking initiation and cessation and Levy et al3,9 reported the outcomes for smoking initiation and prevalence, we further calibrated the model to translate the prevalence effects from Levy et al into changes in cessation probabilities. Our main policy scenario decreased initiation by 10% and increased cessation by 50%. Whereas the association of graphic health warnings with smoking initiation was treated as constant over time, we used a decay rate for cessation that reflected most smokers quitting in response to the policy within the first few years after its implementation. Specifically, the cessation effect was applied with 2012 or 2022 as year 0 (y0), with a decay rate of 20% in subsequent years (eg, 50% × [1 − 0.2]yeary0). The decay rate of 20% was based on prior calibration of the CISNET Smoking History Generator microsimulation model in the evaluation of the effects of cigarette taxes on smoking prevalence and was similarly used by the validated SimSmoke model.9

    We explored a range of policy scenarios in which graphic health warnings reduced smoking initiation across all ages by 5% to 15% and increased cessation across all ages by 25% to 75% (9 scenarios in 2 policy implementation years). In addition, we evaluated a maximum smoking-reduction scenario in which all smoking stopped by the end of 2022, with no new smoking initiation and complete cessation among all current smokers. This estimated the largest possible public health gains that could be achieved through intervention, because smoking-attributable deaths among existing former smokers cannot be avoided with further intervention.20

    Modeling Practices

    When possible, we adhered to recommendations for best research practices in simulation modeling.21,22 We took a societal perspective to simulate the US population over a long time with annual cycle lengths and used a model structure consistent with existing smoking models.23,24 Our state-transition model simulated US cohorts using transition probabilities for smoking that were specific to the historical patterns of each birth cohort. These smoking initiation and cessation transition probabilities were the most representative data sources for our purposes and are among the most widely used in the field of tobacco simulation modeling. Although the model maintained the Markov property, former smokers were distinguished by time since quitting, which allowed the model to simulate changes in their mortality risk depending on how recently they stopped smoking. For intervention effects, existing literature was challenging to translate into model-ready inputs. Therefore, we relied on initiation and cessation effect size estimates determined by experts on cigarette graphic health warnings, which were used in a previous analysis.3 Because a policy on graphic health warnings for cigarettes has never been implemented in the US, we were unable to perform external validation for our policy scenarios; however, the model successfully reproduced observed smoking patterns using the baseline scenario. To facilitate transparency, the CISNET initiation, cessation, and mortality inputs are publicly available online.25 Model equations are provided in the eAppendix in the Supplement.

    Statistical Analysis

    For each scenario, we used methods similar to those of Tam et al9 to estimate smoking prevalence, smoking-attributable deaths averted, and life-years gained from policy implementation to 2100 in the US population. In brief, smoking-attributable deaths (SAD) were calculated using the Equation:

    Image description not available.

    where for a given gender (g), age (a), and year (y), the prevalence of current smoking (prevcs,g(a,y)) was multiplied by the difference in mortality rate between a current smoker (ucs,g(a,y)) and never smoker (uns,g(a,y)) and then applied to the size of the population (Pg(a,y)) at that gender, age, and year to produce the estimated number of smoking-attributable deaths associated with current smoking. This was summed with the number of corresponding smoking-attributable deaths associated with former smoking to produce the total estimated number of smoking-attributable deaths.

    The estimated number of smoking-attributable deaths averted in each year was calculated by subtracting the difference in deaths between policy and baseline scenarios across all ages. To calculate the number of life-years gained, we calculated the difference in the total number of life-years lived in the baseline scenario and each policy scenario. All calculations were performed using R, version 4.0.4 (R Project for Statistical Computing).

    Results

    The model reproduced observed US smoking trends from 2010 to 2018, projecting that smoking prevalence would reach 5% before 2060 in the baseline scenario (Figure 1). In the baseline scenario, the model estimated that smoking prevalence would decrease from 20.5% in 2012 to 13.6% in 2022 and 4.6% in 2100. Simulated implementation of graphic health warnings in both 2022 and 2012 was associated with modest reductions in estimated smoking prevalence compared with the baseline scenario. In the scenario with graphic health warnings implemented in 2022, the model estimated that smoking prevalence would decrease from 13.6% in 2022 to 4.2% (range across policy scenarios, 4.0%-4.4%) in 2100. If the warnings had been implemented in 2012, the model estimated that smoking prevalence would have decreased to 12.3% (range, 11.7%-12.9%) in 2022 and 4.2% (range, 4.0%-4.4%) in 2100.

    In the baseline scenario, the model estimated 13.2 million smoking-attributable deaths from 2012 to 2100, with 8.8 million among men and 4.4 million among women. In the maximum smoking-reduction scenario, in which all smoking ceased by 2022, the model estimated that 5.5 million smoking-attributable deaths would be avoided (men, 3.3 million; women, 2.1 million) through 2100. The total estimated number of life-years gained would be 81.8 million (men, 52.1 million; women, 29.8 million). The full range of estimates for cumulative smoking-attributable deaths averted and life-years gained by 2100 based on the scenarios of policy implementation in 2012 and 2022 are presented in the Table (cumulative results by 2032, 2042, and 2052 are shown in eTables 2-4 in the Supplement).

    In the main policy scenario, implementation in 2022 would be associated with an estimated 539 000 (range, 275 000-794 000) smoking-attributable deaths averted (Figure 2) and 7.9 million (range, 4.0-11.6 million) life-years gained (Figure 3) from 2022 to 2100 and would avert 4.1% (range, 2.1%-6.0%) of the estimated number of smoking-attributable deaths in the baseline scenario. In the scenario with a graphic health warnings policy implemented in 2022, 9.9% (range, 5.0%-14.6%) of the total number of smoking-attributable deaths would be averted and 9.6% (range, 4.9%-14.2%) of the life-years would be gained compared with the maximum smoking reduction scenario.

    If graphic health warnings had been implemented in 2012, the model estimated that policy implementation would have been associated with 33.2% (range, 32.9%-33.5%) more smoking-attributable deaths averted (718 000; range, 365 000 to 1.1 million) and 42.7% (range, 42.3%-43.1%) more life-years gained (11.2 million; range, 5.7-16.6 million) compared with estimates based on implementation in 2022. A total of 5.4% (range, 2.8%-8.0%) of the estimated number of smoking-attributable deaths in the baseline scenario from 2012 to 2100 would have been averted.

    Discussion

    In this decision analytical model, implementation of graphic health warnings was associated with reductions in estimated smoking-related mortality. This was consistent with earlier estimates in the US3 and other countries.26 The reductions in overall estimated smoking prevalence were similar to those reported in Levy et al3: 10% (range, 4%-19%). However, this study’s model estimated fewer smoking-attributable deaths averted during a longer period owing to the model’s projections of lower smoking prevalence in the baseline scenario compared with the projections of Levy et al.3 Recent studies have found that smoking rates among US youths and young adults have decreased at especially fast rates4,5; thus, the outcomes of future policy changes will be attenuated by these underlying trends at baseline.

    This model found that implementing graphic health warnings in 2012, as originally planned based on the FDA’s 2011 rule, may have been associated with 33.2% more smoking-attributable deaths averted and 42.7% more life-years gained, suggesting potential health consequences associated with delayed implementation and tobacco industry litigation. Future legal challenges to FDA tobacco regulations may be associated with substantial harms to population health and should also be quantified.

    We acknowledge that the counterfactual of having had no delays to the implementation of graphic health warnings in 2012 is unlikely to reflect real-world policy making. Past policy delays cannot be undone, and protracted litigation in the present tobacco regulatory environment may be inevitable. Other factors beyond legal challenges, such as limited agency capacity (eg, resources and personnel) and changing leadership, can delay the implementation of regulations. In some instances, such as a lawsuit filed against the FDA by public health organizations, a court decision may expedite policy implementation. However, by quantifying the potential losses associated with delayed implementation, the findings of this study suggest a need to act quickly while minimizing the chances of future procedural delays.

    Only 4% of all estimated smoking-attributable deaths were averted in this study’s scenario of implementing a graphic health warnings policy in 2022. Furthermore, the model estimated that graphic health warnings implemented from 2022 to 2100 would be associated with 9.9% of the total number of smoking-attributable deaths avoided compared with the maximum smoking-reduction scenario. This finding suggests the need for progressive policy action that goes beyond graphic health warnings. The Biden Administration’s recent commitment to implement a federal ban on menthol flavoring in combustible cigarettes may have positive implications for public health.27 Even larger health gains would likely be achieved by reducing the level of nicotine in combustible cigarettes to minimally addictive levels.28

    We believe that rigorous evaluations of graphic health warnings using state-of-the-art methods are needed to accurately quantify their outcomes. Future research should carefully evaluate the effects of the planned graphic health warnings after implementation and pay particular attention to changes to smoking behaviors in response to the policy, including smoking cessation, reduced smoking intensity, switching to noncigarette tobacco products, smoking initiation, and youth experimentation with smoking. Furthermore, researchers should consider how multiple factors, including e-cigarette use and other tobacco control policies, could confound estimates of policy outcomes.

    Limitations

    This study has limitations. Although the literature on graphic health warnings reports an associated public health benefit,16,17,19,29-33 uncertainty remains about the true magnitude of their effect on smoking behavior, especially with regard to smoking initiation.34 To our knowledge, the most rigorous analysis conducted to date evaluated the graphic health warnings policy in Canada using the US as a control.35 The current study estimated changes to overall smoking prevalence and not smoking behaviors per se, similar to other analyses.19,35 We modeled smoking through initiation and cessation behaviors, which required inputs that were challenging to produce using existing empirical methods. Even in studies with appropriate controls, causality cannot be evaluated because graphic health warnings can be implemented at the same time as other policies. For these reasons, we structured this study as an uncertainty analysis by evaluating policy effects under both optimistic and conservative policy conditions using previously reported upper and lower threshold values based on experts’ opinion of the likely association of graphic health warnings with smoking initiation and cessation. Although initiation effects are understudied,34 other research on cessation effects was consistent with our estimates.18,19,36

    We also did not consider potential associations between graphic health warnings and the quantity of cigarettes consumed by individual smokers. We did not find substantial evidence in the literature to justify such associations. Although the quantity of cigarettes smoked may decrease for those who continue to smoke after the implementation of such a policy, the mean overall quantity of cigarettes smoked may increase if those who quit as a result of a graphic health warning policy smoked less intensely than did the remaining smokers.

    We did not present uncertainty analyses in our baseline projections. Although this could influence our baseline scenario results in terms of absolute numbers, this study focused on the health gains associated with policies. Any uncertainty in the underlying baseline model inputs (all-cause mortality rates and CISNET initiation and cessation estimates) would carry through to the policy scenario, and relative changes would remain qualitatively the same.

    The effects estimated by the model are based on the CISNET projections of smoking rates using data through 2018. We did not account for changes in smoking or changes in underlying mortality as a result of the COVID-19 pandemic. The health benefits associated with graphic health warnings will depend on a range of factors not considered in this model, including tobacco control policies, industry activities, and the use of other tobacco and nicotine products such as e-cigarettes. In addition, we did not explicitly consider e-cigarettes in the model. However, the underlying CISNET smoking parameters captured recent changes in smoking initiation and cessation through 2018 and cohort effects that may reflect recent e-cigarette use. Graphic health warnings proposed by the FDA in 2011 differ from those planned for 2022, but this study assumed that the association of the warnings with smoking behaviors would be the same.

    Although this model’s results were consistent with earlier estimates using the SimSmoke model,3 they differed in absolute terms partly because the current model did not explicitly incorporate the effects of past policies. Furthermore, the projections accounted for more recent decreases in the prevalence of smoking by birth cohort, which resulted in a lower projected baseline smoking prevalence in the absence of graphic health warnings.

    Conclusions

    This decision analytical model estimated that FDA cigarette graphic health warnings, if implemented in 2022, would be associated with public health benefits. The model also estimated that more smoking-attributable deaths would have been averted if the policy had been implemented in 2012. This study’s findings suggest that preventing delays to policy implementation should be a public health priority.

    Back to top
    Article Information

    Accepted for Publication: August 4, 2021.

    Published: September 24, 2021. doi:10.1001/jamahealthforum.2021.2852

    Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2021 Tam J et al. JAMA Health Forum.

    Corresponding Author: Rafael Meza, PhD, Department of Epidemiology, University of Michigan School of Public Health, 1415 Washington Heights, Ann Arbor, MI 48109 (rmeza@umich.edu).

    Author Contributions: Dr Meza had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

    Concept and design: Tam, Thrasher, Levy, Meza.

    Acquisition, analysis, or interpretation of data: Tam, Jeon, Hammond, Holford, Levy, Meza.

    Drafting of the manuscript: Tam, Thrasher, Levy, Meza.

    Critical revision of the manuscript for important intellectual content: All authors.

    Statistical analysis: Holford, Levy, Meza.

    Obtained funding: Tam, Meza.

    Administrative, technical, or material support: Hammond.

    Supervision: Tam, Meza.

    Conflict of Interest Disclosures: Dr Tam reported receiving grants from the National Cancer Institute (NCI), National Institutes of Health (NIH) during the conduct of the study and being a tobacco regulatory science fellow at the US Food and Drug Administration’s Center for Tobacco Products through the National Academy of Medicine. Dr Hammond has testified on behalf of public health authorities in response to industry legal challenges, including challenges to health warning regulations. Dr Holford reported receiving grants from the NCI, NIH during the conduct of the study. Dr Meza reported receiving grants from the NCI, NIH during the conduct of the study. No other disclosures were reported.

    Funding/Support: This study was supported by awards U01CA253858 (Drs Meza, Levy, and Holford), U01CA199284 (Drs Meza, Levy, and Holford), and U54CA229974 (Drs Meza and Levy) from the NCI, NIH.

    Role of the Funder/Sponsor: The funder had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

    Disclaimer: The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.

    References
    1.
    US Food & Drug Administration. Cigarette health warnings. 2020. Accessed January 6, 2021. https://www.fda.gov/tobacco-products/labeling-and-warning-statements-tobacco-products/cigarette-health-warnings
    2.
    Public Health and Tobacco Policy Center. FDA graphic warnings. 2020. Accessed July 23, 2021. https://tobaccopolicycenter.org/tobacco-control/tobacco-litigation/fda-graphic-warnings/
    3.
    Levy  DT, Mays  D, Yuan  Z, Hammond  D, Thrasher  JF.  Public health benefits from pictorial health warnings on US cigarette packs: a SimSmoke simulation.   Tob Control. 2017;26(6):649-655. doi:10.1136/tobaccocontrol-2016-053087 PubMedGoogle ScholarCrossref
    4.
    Levy  DT, Warner  KE, Cummings  KM,  et al.  Examining the relationship of vaping to smoking initiation among US youth and young adults: a reality check.   Tob Control. 2019;28(6):629-635. doi:10.1136/tobaccocontrol-2018-054446 PubMedGoogle ScholarCrossref
    5.
    Meza  R, Jimenez-Mendoza  E, Levy  DT.  Trends in tobacco use among adolescents by grade, sex, and race, 1991-2019.   JAMA Netw Open. 2020;3(12):e2027465-e2027465. doi:10.1001/jamanetworkopen.2020.27465 PubMedGoogle ScholarCrossref
    6.
    Jeon  J, Meza  R, Krapcho  M, Clarke  LD, Byrne  J, Levy  DT. Actual and counterfactual smoking prevalence rates in the U.S. population via microsimulation. Risk Anal. 2012;32(suppl 1):S51-S68. doi:10.1111/j.1539-6924.2011.01775.x
    7.
    Moolgavkar  SH, Holford  TR, Levy  DT,  et al.  Impact of reduced tobacco smoking on lung cancer mortality in the United States during 1975-2000.   J Natl Cancer Inst. 2012;104(7):541-548. doi:10.1093/jnci/djs136 PubMedGoogle ScholarCrossref
    8.
    Jeon  J, Holford  TR, Levy  DT,  et al.  Smoking and lung cancer mortality in the United States from 2015 to 2065: a comparative modeling approach.   Ann Intern Med. 2018;169(10):684-693. doi:10.7326/M18-1250 PubMedGoogle ScholarCrossref
    9.
    Tam  J, Levy  DT, Jeon  J,  et al.  Projecting the effects of tobacco control policies in the USA through microsimulation: a study protocol.   BMJ Open. 2018;8(3):e019169. doi:10.1136/bmjopen-2017-019169 PubMedGoogle Scholar
    10.
    Holford  TR, Levy  DT, McKay  LA,  et al.  Patterns of birth cohort-specific smoking histories, 1965-2009.   Am J Prev Med. 2014;46(2):e31-e37. doi:10.1016/j.amepre.2013.10.022 PubMedGoogle ScholarCrossref
    11.
    Creamer  MR, Wang  TW, Babb  S,  et al.  Tobacco product use and cessation indicators among adults—United States, 2018.   MMWR Morb Mortal Wkly Rep. 2019;68(45):1013-1019. doi:10.15585/mmwr.mm6845a2 PubMedGoogle ScholarCrossref
    12.
    Rosenberg  MA, Feuer  EJ, Yu  B,  et al Chapter 3: cohort life tables by smoking status, removing lung cancer as a cause of death. Risk Anal. 2012;32(suppl 1):S25-S38. doi:10.1111/j.1539-6924.2011.01662.x
    13.
    Institute of Medicine.  Health Implications of Raising the Minimum Age for Purchasing Tobacco Products. The National Academies Press; 2015.
    14.
    LCFIT. LCFIT: The Lee-Carter Fitter. 2009. Accessed July 17, 2021. http://lcfit.demog.berkeley.edu/
    15.
    Thun  MJ, Carter  BD, Feskanich  D,  et al.  50-year trends in smoking-related mortality in the United States.   N Engl J Med. 2013;368(4):351-364. doi:10.1056/NEJMsa1211127 PubMedGoogle ScholarCrossref
    16.
    Hammond  D, Reid  JL, Driezen  P,  et al.  Are the same health warnings effective across different countries? an experimental study in seven countries.   Nicotine Tob Res. 2019;21(7):887-895. doi:10.1093/ntr/nty248 PubMedGoogle ScholarCrossref
    17.
    Drovandi  A, Teague  P-A, Glass  B, Malau-Aduli  B.  A systematic review of the perceptions of adolescents on graphic health warnings and plain packaging of cigarettes.   Syst Rev. 2019;8(1):25. doi:10.1186/s13643-018-0933-0 PubMedGoogle ScholarCrossref
    18.
    Brewer  NT, Hall  MG, Noar  SM,  et al.  Effect of pictorial cigarette pack warnings on changes in smoking behavior: a randomized clinical trial.   JAMA Intern Med. 2016;176(7):905-912. doi:10.1001/jamainternmed.2016.2621 PubMedGoogle ScholarCrossref
    19.
    Noar  SM, Francis  DB, Bridges  C, Sontag  JM, Ribisl  KM, Brewer  NT.  The impact of strengthening cigarette pack warnings: systematic review of longitudinal observational studies.   Soc Sci Med. 2016;164(164):118-129. doi:10.1016/j.socscimed.2016.06.011 PubMedGoogle Scholar
    20.
    Warner  KE, Mendez  D. How much of the future mortality toll of smoking can be avoided? Tob Control. Published online June 16, 2020. doi:10.1136/tobaccocontrol-2019-055530
    21.
    Caro  JJ, Briggs  AH, Siebert  U, Kuntz  KM; ISPOR-SMDM Modeling Good Research Practices Task Force.  Modeling good research practices—overview: a report of the ISPOR-SMDM Modeling Good Research Practices Task Force-1.   Med Decis Making. 2012;32(5):667-677. doi:10.1177/0272989X12454577 PubMedGoogle ScholarCrossref
    22.
    Siebert  U, Alagoz  O, Bayoumi  AM,  et al.  State-transition modeling: a report of the ISPOR-SMDM Modeling Good Research Practices Task Force–3.   Med Decis Making. 2012;32(5):690-700. doi:10.1177/0272989X12455463 PubMedGoogle ScholarCrossref
    23.
    Feirman  SP, Donaldson  E, Glasser  AM,  et al.  Mathematical modeling in tobacco control research: initial results from a systematic review.   Nicotine Tob Res. 2016;18(3):229-242. doi:10.1093/ntr/ntv104 PubMedGoogle ScholarCrossref
    24.
    Feirman  SP, Glasser  AM, Rose  S,  et al.  Computational models used to assess US tobacco control policies.   Nicotine Tob Res. 2017;19(11):1257-1267. doi:10.1093/ntr/ntx017 PubMedGoogle ScholarCrossref
    25.
    National Cancer Institute Cancer Intervention and Surveillance Modeling Network. CISNET publication support and modeling resources. 2018. Accessed July 17, 2021. https://resources.cisnet.cancer.gov/projects/.
    26.
    Alcaraz  A, Hernández-Vásquez  A, Palacios  A,  et al.  Health and economic impact of health warnings and plain tobacco packaging in seven Latin American countries: results of a simulation model.   Nicotine Tob Res. 2020;22(11):2032-2040. doi:10.1093/ntr/ntaa104 PubMedGoogle ScholarCrossref
    27.
    US Food & Drug Administration. FDA commits to evidence-based actions aimed at saving lives and preventing future generations of smokers. April 29, 2021. Accessed July 1, 2021. https://www.fda.gov/news-events/press-announcements/fda-commits-evidence-based-actions-aimed-saving-lives-and-preventing-future-generations-smokers
    28.
    Apelberg  BJ, Feirman  SP, Salazar  E,  et al.  Potential public health effects of reducing nicotine levels in cigarettes in the United States.   N Engl J Med. 2018;378(18):1725-1733. doi:10.1056/NEJMsr1714617 PubMedGoogle ScholarCrossref
    29.
    Noar  SM, Barker  J, Bell  T, Yzer  M.  Does perceived message effectiveness predict the actual effectiveness of tobacco education messages? a systematic review and meta-analysis.   Health Commun. 2020;35(2):148-157. doi:10.1080/10410236.2018.1547675 PubMedGoogle ScholarCrossref
    30.
    Noar  SM, Hall  MG, Francis  DB, Ribisl  KM, Pepper  JK, Brewer  NT.  Pictorial cigarette pack warnings: a meta-analysis of experimental studies.   Tob Control. 2016;25(3):341-354. doi:10.1136/tobaccocontrol-2014-051978 PubMedGoogle ScholarCrossref
    31.
    Hammond  D.  Health warning messages on tobacco products: a review.   Tob Control. 2011;20(5):327-337. doi:10.1136/tc.2010.037630 PubMedGoogle ScholarCrossref
    32.
    Yong  HH, Borland  R, Thrasher  JF,  et al Mediational pathways of the impact of cigarette warning labels on quit attempts. Health Psychol. 2014;33(11):1410-1420. doi:10.1037/hea0000056
    33.
    Noar  SM, Francis  DB, Bridges  C, Sontag  JM, Brewer  NT, Ribisl  KM.  Effects of strengthening cigarette pack warnings on attention and message processing: a systematic review.   Journal Mass Commun Q. 2017;94(2):416-442. doi:10.1177/1077699016674188 PubMedGoogle ScholarCrossref
    34.
    Thrasher  JF, Brewer  NT, Niederdeppe  J,  et al.  Advancing tobacco product warning labels research methods and theory: a summary of a grantee meeting held by the US National Cancer Institute.   Nicotine Tob Res. 2019;21(7):855-862. doi:10.1093/ntr/nty017 PubMedGoogle ScholarCrossref
    35.
    Huang  J, Chaloupka  FJ, Fong  GT.  Cigarette graphic warning labels and smoking prevalence in Canada: a critical examination and reformulation of the FDA regulatory impact analysis.   Tob Control. 2014;23(suppl 1):i7-i12. doi:10.1136/tobaccocontrol-2013-051170 PubMedGoogle ScholarCrossref
    36.
    Azagba  S, Sharaf  MF.  The effect of graphic cigarette warning labels on smoking behavior: evidence from the Canadian experience.   Nicotine Tob Res. 2013;15(3):708-717. doi:10.1093/ntr/nts194 PubMedGoogle ScholarCrossref
    ×