Age, cumulative psychosocial risk factors, and health disparities in U.S. adult cigarette smoking: 2002–2019 Authors: Fang Fang Chen, Elias M Klemperer, Tyler G Erath, Michael DeSarno, Adam M Leventhal, Stephen T Higgins Published: 2025 ABSTRACT AbstractBackgroundUnderstanding psychosocial factors that underpin smoking disparities is important for developing targeted interventions to reduce tobacco use and its associated adverse health consequences. This study investigated associations between cumulative psychosocial risks, age, and the interplay of these two factors in U.S. smoking disparities over time.MethodsParticipants included adults aged ≥ 18 years in the 2002–2019 US National Survey on Drug Use and Health. Psychosocial factors were combined into a cumulative-risk index with four levels: No-, Low-, Moderate-, and High-risk. Logistic regression analyses tested whether smoking prevalence varied by risk level and age over time.ResultsGreater cumulative risk was associated with higher smoking prevalence. Using No-risk group prevalence (4.2%) as the comparator, smoking prevalence was greatest in the High-risk group (50.8%; OR = 20.9 [95% CI, 19.8–22.0]), followed by the Mod-risk group (29.7%; OR = 5.8 [5.6–5.9]), and low-risk group (15.2%; OR = 2.4 [95% CI, 2.4–2.5]). Importantly, these disparities persisted over time.Smoking prevalence declined across time in all age groups under age 65 and the rate of decline decreased with age: Age 18–25 showed the greatest decline (34.4% to 13.7%; OR = .88 [95% CI, .88–.88]), followed by age 35–49 (29.4% to 20.8%; OR = .95 [.94–.95]), age 26–34 (30.2% to 22.4%; OR = .95 [.94–.95]), and Age 50–64 showed the slowest decline trend (21.3% to 18.8%; OR = .96 [95% CI, .96–.97]). No significant decline over time for age 65 + years was observed (10.4% to 8.7%; OR = 1.00 [.98–1.01]).There was a significant association of risk level and age in changes of smoking prevalence over time. For ages 18–25 and 35–49, smoking prevalence declined over time across all risk levels. For Ages 26–34, smoking prevalence decreased between 2012 and 2019 across all risk levels but not during 2002–2011. However, for ages 50–64 and 65 + , smoking prevalence was stable for all risk levels between 2002 and 2019.ConclusionsDespite overall declines in smoking prevalence, disparities by cumulative risk levels and age remained steady and substantial over nearly two decades. Including consideration of smoking disparities by cumulative risk levels and age may enhance the development of more targeted and effective interventions to reduce smoking prevalence.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12916-025-04421-3.Keywords: Cumulative psychosocial risks, Age, Health disparities, Health inequities, Smoking prevalence Background Smoking continues to pose a major U.S. public health concern, despite significant declines in smoking prevalence over the years [1–3]. Cigarette smoking disproportionately impacts those facing psychosocial disadvantages such as poverty or mental health conditions. These disadvantages are not merely individual attributes but reflect systemic and often unjust social conditions, including unequal access to education, housing instability, discrimination, and limited behavioral health services. As a result, smoking contributes to inequities in morbidity and mortality, particularly among socially and economically marginalized populations [4, 5]. Understanding the behavioral trajectory and psychosocial underpinnings of disparities in smoking risk is crucial for developing more effective and targeted interventions to reduce tobacco use and its associated health consequences [6]. Such disparities are closely linked to various risk factors, including socioeconomic status [7–9], psychological problems/distress [7, 9, 10] and alcohol/drug problems [11–13]. Previous research has primarily focused on individual risk factors [3] in the context of tobacco-related health disparities. However, it is important to understand the combined influence of various risk factors in tobacco research. Various forms of psychosocial challenges often occur together, and their cumulative presence may contribute to increasingly elevated probabilities of smoking. Indeed, smoking risk has been shown to vary as an orderly function of the cumulative number of risk factors that an individual has experienced [11–13]. Adult smoking prevalence also tends to vary across age groups. For example, a recent study examining age-specific trends in U.S. adult smoking between 2011 and 2022 demonstrated that while all age groups experienced declines, the pace of decline differed markedly in the data from the National Health Interview Survey (NHIS) [3]. Historically, U.S. adults under age 65 exhibited high smoking prevalence, reaching approximately 28.5% for young adults (ages 18–24), 25.6% for adults (ages 25–44), and 22.7% for middle-aged adults (ages 45–64) in the early 2000s [7]. However, the rate of decline in smoking prevalence over time has varied across age groups: it decreased sharply among young adults (ages 18–24) to 4.8%, but more gradually among adults (aged 25–44) reaching 12.5%, and even more slowly among middle-aged adults (ages 45-64) falling to 15.1% in 2022 [2]. As a result of these differing rates of decline, middle-aged adults now have the highest smoking rate among all age groups. Older adults (ages 65 +) consistently showed the lowest smoking prevalence, remaining relatively stable from 9.3% in 2002 to 8.7% in 2022 [2, 7] in sharp contrast to the substantial declines among younger age groups. The recent study [3] further demonstrated how these age trends differed by key sociodemographic risk factors such as income and education, underscoring the importance of examining the intersection between age and individual risk factors when tracking changes in smoking prevalence over time. However, far less is known about how cumulative psychosocial risk might shape smoking disparities within and across age groups. Age may interact with cumulative risk factors in multiple ways. One hypothesis is that accumulated risks may exert a consistent and uniform influence across the lifespan, producing elevated smoking prevalence at all ages. Another possibility is that age may moderate the impact of cumulative risks in more nuanced, developmental patterns—for instance, in younger individuals, a greater number of risk factors may accelerate smoking initiation and uptake, whereas in older adults, multiple risk factors may contribute to the persistence and reinforcement of smoking behavior. By incorporating age-focused surveillance research within a cumulative risk framework, this study has the potential to make a substantive contribution. Cumulative risk models integrate multiple psychosocial determinants, such as socioeconomic disadvantage, mental health challenges, and adverse social environments, into a single index, enabling a more holistic assessment of vulnerability. Applying this approach to smoking prevalence allows us to assess not only whether disparities exist, but also whether the magnitude and nature of these disparities shift across age groups. To our knowledge, no prior work has systematically examined the interplay between cumulative risk and age in smoking prevalence, which is critically important for tailoring interventions to different life stages. Despite an overall decline in smoking prevalence, substantial disparities in smoking persist [1–3]. This study aims to address existing knowledge gaps regarding smoking disparities over time by systematically examining how smoking prevalence trend (i.e., lifetime smoking of > 100 cigarettes and past month cigarette use) varies by (1) cumulative psychosocial risk levels; (2) age; and (3) the interplay between cumulative smoking risk and age—a relationship has not been previously examined and could be important to developing more targeted and effective interventions to reduce smoking. Methods Data source Data are from individuals aged 18 years or older who participated in the 2002–2019 National Survey on Drug Use and Health (NSDUH) [14]. The NSDUH data set includes nationally representative cross-sectional data on measures of tobacco, alcohol, and illicit drug use, and of other health-related issues in the U.S. population among civilian, noninstitutionalized individuals 12 years of age and above. We used the eighteen years between 2002 and 2019 as changes in survey methods and administration in 2002 raise validity concerns about including earlier years, and COVID-related method changes in 2020 and later years preclude comparisons with prior years. The NSDUH data collection procedure received approval from the institutional review board at RTI International. Trained interviewers collected data through personal visits to households and noninstitutional group quarters. Verbal informed consent was obtained from each participant. The use of audio computer-assisted self-administered interviewing ensured a private and confidential method for respondents to input their answers [15]. Main outcomes and measures Current smoking Current cigarette smoking status is defined as lifetime smoking of > 100 cigarettes and past month cigarette use. Quit ratios Quit ratio is defined as the number of former smokers (reports lifetime cigarettes > 100 and not having smoked a cigarette in the past year) over lifetime smokers (reports lifetime cigarettes > 100). Findings related to quit ratios are presented in Additional file 1_Quit Ratios. Risk factors Eleven risk factors were assessed. Variables of interest were determined based on previously identified demographic and psychosocial predictors of smoking status, including less than a 4-year college education, low annual income (i.e., < $50,000), alcohol binge drinking, illicit drug use/dependence, and psychological distress [11–13, 16]. Additional risk factors included being unmarried [17], having moved in the past year as an indication of housing instability [18], no seat belt use as an indication of high risk taking [19], lack of religious beliefs [20], and perceived risk in smoking [21]. Detailed coding information for these risk factors is provided in the Additional file 2: e-Method. Associations of individual risk factors with cumulative risk index are presented in Additional file 2: e-Table 1. Associations of individual risk factors with smoking prevalence are presented in Additional file 2: e-Table 2. Age Age is defined at the individual level for each survey year (i.e., all 18-year-olds in 2002 were born in the same year); thus, the observed trends reflect age-specific population patterns across time, not within-person changes. Time A two-year time bins were formed between 2002 and 2019 to reduce variability due to sampling variation. We focused on the years between 2002 and 2019 because this is the longest period during which common survey methods were used in NSDUH. Cumulative psychosocial risk levels Each risk factor was first re-categorized as a binary variable (risk or no-risk) and then summed in a cumulative risk index for each survey year. This approach aligns with the standard cumulative risk methodology [22], where multiple risk indicators are aggregated into a composite index to capture co-occurrence at a point in time. Different cut-off points were identified based on the distribution of the risk factors in relation to the outcome variables. Sensitivity analyses were performed based on different cut-off points. The final thresholds were based on model fit and model parsimony. Participants were classified into the following four risk levels: those with 0–1 risk factor were categorized as No-risk, those with 2–3 risk factors as Low-risk, those with 4–5 risk factors as Moderate-risk, and 6 or more risk factors as High-risk. Race and ethnicity The determination of race and ethnicity was based on respondents’ self-classification, using classifications established by the US Census Bureau. More detailed description is provided in Additional file 2: e-Method. Statistical analysis Descriptives of smoking prevalence by sociodemographic characteristics and cumulative risk levels were analyzed using the Proc SurveyMeans procedure in SAS 9.4 software [23]. Logistic regression analyses were conducted to test whether smoking prevalence varied by sociodemographic characteristics and cumulative-risk levels across time, with the Proc SurveyLogistic procedure in SAS [23]. The above analyses were conducted to examine age differences over time and for cumulative-risk level differences over time within each age group. E-cigarettes began to gain significant visibility in the United States in 2011, which marked the beginning of a rapid increase in e-cigarette awareness and use [24]. The rate of declining for cigarette smoking appeared to be faster after 2012. A dichotomous variable of Piece was created to separate time into two pieces, one for 2002–2011 (Piece1) and the other for 2012–2019 (Piece2). Data were weighted by using population weight during analysis to adjust for the differential probability of selection and response. Covariates included gender, race/ethnicity, and region in which the respondent resided (rural vs. non-rural). To examine the decline in smoking prevalence for each age group by cumulative-risk levels, the interaction effect of Risk Level × 2-Year Bin was tested within each age group. All analyses were performed at the individual level. For each analysis, p < 0.05 (2-tailed) was considered statistically significant. Results Sociodemographic characteristics Participant baseline characteristics are presented in Table 1. Briefly, this cross-sectional sample included 706,891 adults age 18 years and older across survey years 2002–2019. About half were Male (46.6% unweighted), Non-Hispanic White (63.2%), had less than a 4-year college degree (76.9%), had an annual family income of less than $50,000 (58.1%), and most resided in an urban area (79.2%). Annual weighted interview response rates ranged from 64.9% to 79.0% during 2002–2019 (Mean = 72.7%, SD = 0.04). Table 1Sample descriptive statistics and association with smoking prevalenceCigarette smokingTotal N = 706,891Unweighted sampleWeighted smoking prevalence Test of differencesN%Smoking %95% CIOdds ratio95% CISurvey year 2002–2003733969.225.4%24.9%25.9%1.661.601.72 2004–005745359.324.8%24.3%25.3%1.611.551.67 2006–007742289.324.4%23.9%24.9%1.571.521.63 2008–009752119.423.4%22.9%23.9%1.481.431.54 2010–011780529.722.2%21.7%22.7%1.391.341.44 2012–2013752939.421.5%21.0%21.9%1.331.281.38 2014–0158523210.620.1%19.7%20.4%1.221.181.26 2016–0178517910.618.4%18.0%18.8%1.101.061.14 2018–019 (Ref)8576510.717.1%16.7%17.4%Age year Age 18–530473943.125.2%25.0%25.4%3.313.173.44 Age 26–411956716.928.5%28.2%28.9%3.913.744.08 Age 35–916262923.025.1%24.8%25.3%3.273.143.41 Age 50–47176810.220.8%20.4%21.2%2.572.462.69 Age 65+ (Ref)481886.89.3%8.9%9.6%Gender Female37754353.419.3%19.1%19.4%0.740.720.75 Male (Ref)32934846.624.5%24.2%24.7%Race/ethnicity Hispanic11113915.716.2%15.9%16.6%0.630.610.65 Black8697012.322.1%21.6%22.5%0.920.900.95 American Indian/ Alaska Native101321.434.1%32.3%36.0%1.691.551.83 NonHisp Native HI/Other Pac Isl35140.522.0%19.5%24.6%0.920.791.07 Asian285354.09.7%9.2%10.2%0.350.330.37 Multi-Race196642.830.6%29.2%31.9%1.431.351.53 White (Ref)44693763.223.5%23.3%23.7%Education < High School11097915.726.6%29.6%30.5%3.593.483.70 High School21822030.922.9%27.2%27.8%3.173.093.26 Some College21416330.317.5%22.7%23.3%2.492.422.56 >= College (Ref)16352923.17.1%10.5%10.9%Family income Less than $20,00016871523.930.4%30.1%30.8%2.622.562.69 $20,000–$49,99924148734.225.2%24.9%25.5%1.741.691.78 $50,000–$74,99911283016.020.1%19.8%20.5%1.301.271.33 $75,000+ (Ref)18385926.014.3%14.1%14.5%Rurality Rural14678120.826.9%26.5%27.3% Urban (Ref)56011079.220.8%20.6%21.0%1.401.371.43 Sample descriptive statistics and association with smoking prevalence Cumulative-risk and smoking outcomes The weighted proportion of respondents categorized into the No-, Low-, Mod-, and High-cumulative risk categories and relationships between risk levels and smoking outcomes when collapsing data across survey years are presented in Table 1. Using smoking prevalence in the No-risk group (4.7%) as the comparator, smoking prevalence was greatest in the High-risk group (50.8%, odds ratio (OR) = 20.9 [95% Confidence Interval (CI), 19.8–22.0]), followed by the Mod-risk (29.7%, OR = 5.8 [5.6–5.9]), and Low-risk (15.2%, OR = 2.4 [95% CI: 2.4–2.5]) groups. This pattern was stable over time (Additional file 2: e-Fig. 1). Table 2 shows that risk factors remained stable over time and were associated with smoking consistently. Table 2Risk levels and association with smoking prevalence over timeCigarette smokingTotal N = 706,891 Test of differencesSample N (%)Weighted sample %Weighted smoking %95% CIOdds ratio95% CIAge 18+ (overall sample collapsed over time) No Risk (Ref)74,59217.54.7%4.5%4.9% Low Risk228,70941.515.2%14.9%15.4%2.42.392.50 Moderate Risk249,90729.229.7%29.4%30.0%5.85.645.92 High Risk153,68311.850.8%50.4%51.2%20.919.8321.95 Overall21.8%21.6%21.9%2002–2003 No Risk (Ref)687416.25.7%4.9%6.4% Low Risk22,20341.217.8%17.1%18.5%2.52.322.65 Moderate Risk25,87429.733.6%32.6%34.5%5.85.386.21 High Risk18,44513.055.6%54.3%56.9%20.918.0224.23 Overall25.4%24.9%25.9%2004–2005 No Risk (Ref)665416.96.1%5.3%6.8% Low Risk22,59541.317.8%17.0%18.5%5.65.195.99 Moderate Risk26,30729.132.7%31.7%33.6%5.65.195.99 High Risk18,97912.754.6%53.4%55.9%18.716.1221.68 Overall24.8%24.3%25.3%2006–2007 No Risk (Ref)718017.55.6%4.9%6.2% Low Risk22,55340.817.6%16.9%18.4%2.52.332.67 Moderate Risk26,03228.932.1%31.2%33.1%5.55.145.94 High Risk18,46312.954.2%52.9%55.4%20.117.4423.13 Overall24.4%23.9%24.9%2008–2009 No Risk (Ref)736418.65.7%4.9%6.5% Low Risk23,39542.016.5%15.7%17.2%2.52.342.70 Moderate Risk27,21928.432.9%31.9%33.9%6.25.796.74 High Risk17,23311.055.2%53.9%56.5%20.317.4223.68 Overall23.4%22.9%23.9%2010–2011 No Risk (Ref)782317.94.5%3.8%5.1% Low Risk24,40541.816.3%15.5%17.0%2.42.272.60 Moderate Risk28,08728.930.3%29.3%31.2%5.45.055.85 High Risk17,73711.451.3%50.0%52.6%22.519.1026.54 Overall22.2%21.7%22.7%2012–2013 No Risk (Ref)727317.74.7%4.0%5.4% Low Risk23,78742.114.5%13.9%15.2%2.72.482.86 Moderate Risk27,64629.029.6%28.7%30.5%6.66.107.12 High Risk16,58711.352.8%51.5%54.2%22.719.3126.66 Overall21.5%21.0%21.9%2014–2015 No Risk (Ref)10,58617.64.0%3.5%4.5% Low Risk29,99941.513.5%13.0%14.0%2.42.222.51 Moderate Risk29,36229.528.1%27.3%28.8%5.95.546.31 High Risk15,28511.448.0%46.8%49.1%22.119.1825.58 Overall20.1%19.7%20.4%2016–2017 No Risk (Ref)10,35517.83.5%3.0%3.9% Low Risk29,81541.612.5%12.0%13.0%2.32.172.45 Moderate Risk29,63829.625.9%25.1%26.6%5.65.266.02 High Risk15,37111.144.6%43.4%45.7%22.319.3025.67 Overall18.4%18.0%18.8%2018–2019 No Risk (Ref)10,48317.63.4%2.9%3.8% Low Risk29,95741.511.2%10.7%11.7%2.32.142.43 Moderate Risk29,74229.623.9%23.2%24.6%5.75.296.08 High Risk15,58311.441.7%40.6%42.9%20.617.6923.98 Overall17.1%16.7%17.4% Risk levels and association with smoking prevalence over time Age and smoking outcomes The relationship between age and smoking prevalence when collapsing data across survey years is complex (Table 1). The group with the highest smoking prevalence is age 26–34 years (28.5%), followed by age 18–25 (25.2%), age 35–49 (25.1%), and age 50–64 years (20.8%). The age 65 + group had the lowest smoking prevalence (9.3%). Smoking prevalence over time Cumulative risk Associations between smoking prevalence and cumulative risk are presented in Table 2 and Fig. 1. Smoking prevalence decreased across all risk levels between 2002 and 2019: for No-risk group (5.7% to 3.4%), OR = 0.93 [95% CI, 0.91–0.94]; Low-risk group (17.8% to 11.2%), OR = 0.94 [95% CI, 0.93–0.95]; High-risk group (55.6% to 41.7%), OR = 0.94 [95% CI, 0.93–0.95]; and Mod-risk group (33.6% to 23.9%), OR = 0.95 [95% CI, 0.95–0.96]. Risk level differences in trends of cigarette smoking prevalence among US adults Across all but the No-risk group, the rate of decline was slower in 2002–2011 than in 2012–2019: for the Low-risk group, 2002–2011, b = − 0.03, OR = 0.98 [95% CI, 0.96–0.99] vs. in 2012–2019, b = − 0.09, OR = 0.91 [95% CI, 0.89–0.94], and the difference was significant, OR = 0.94 [95% CI, 0.91–0.96]; for the Mod-risk group, 2002–2011, b = − 0.02, OR = 0.98 [95% CI, 0.96–0.99] vs. in 2012–2019, b = − 0.10, OR = 0.91 [95% CI, 0.89–0.93], and the difference was significant, OR = 1.06 [95% CI, 1.03–1.08]; for the High-risk group in 2002–2011, b = − 0.03, OR = 0.97 [95% CI, 0.96–0.99] vs. in 2012–2019, b = − 0.15, OR = 0.86 [95% CI, 0.84–0.88], and the difference was significant, OR = 0.90 [95% CI, 0.87–0.92]. There was no significant difference in decline between the two time periods for the No-risk group: 2002–2011, b = − 0.05, OR = 0.95 [95% CI, 0.91–0.99] vs. 2012–2019, b = − 0.12, OR = 0.89 [95% CI, 0.83–0.95]. These findings indicate that the cumulative influence of these psychosocial disadvantages on smoking is persistent across the 18-year period. Age Associations between smoking prevalence and age are presented in Fig. 2. Smoking prevalence declined across all age groups from 2002 to 2019. The three younger age groups showed larger declines: for Age 18–25 group (34.4% to 13.7%), b = –0.13, OR = 0.88 [95% CI, 0.88–0.88], with a linear decreasing trend in odds per two-year bin; Age 26–34 group (30.2% to 22.4%), b = –0.06, OR = 0.95 [95% CI, 0.94–0.95]; Age 35–49 group (29.4% to 20.8%), b = –0.06, OR = 0.95 [95% CI, 0.94–0.95]. The two oldest groups exhibited small-to-no declines: Age 50–64 (21.3% to 18.8%), b = –0.06, OR = 0.96 [95% CI, 0.96–0.97]; Age 65 + (10.4% to 8.7%), b = 0.00, OR = 1.00 [95% CI, 0.98–1.01].Across all but the Age 65 + group, the rate of decline was slower in 2002–2011 than in 2012–2019. For Age 18–25 group, there was a decline in 2002–2011, b = –0.09, OR = 0.92 [95% CI, 0.91–0.93] vs. a faster decline in 2012–2019, b = –0.24, OR = 0.79 [95% CI, 0.77–0.80], and the difference was significant, OR = 0.86 [95% CI, 0.84–0.88]. For Age 26–34 group, there was no decline in 2002–2011 (30.2% to 30.3%), b = 0.01, OR = 1.01 [95% CI, 0.99–1.03] vs. a decline in 2012–2019 (30.6% to 22.4%), b = –0.17, OR = 0.84 [95% CI, 0.82–0.87], and the difference was significant, OR = 0.84 [95% CI, 0.81–0.86]. For Age 35–49 group, there was a small decline in 2002–2011, b = –0.04, OR = 0.96 [95% CI, 0.95–0.98] vs. a faster decline in 2012–2019, b = –0.08, OR = 0.92 [95% CI, 0.90–0.94], and the difference was significant, OR = 0.96 [95% CI, 0.93–0.98]. For Age 50–64 group, there was no decline in 2002–2011 (21.3% to 20.8%), b = 0.00, OR = 1.00 [95% CI, 0.98–1.02] vs. a slight decline in 2012–2019 (20.0% to 18.8%), b = –0.05, OR = 0.95 [95% CI, 0.92–0.98], and the difference was significant, OR = 0.95 [95% CI, 0.91–0.99]. For Age 65 + group, there was no decline in 2002–2011 (10.4% to 9.5%), b = 0.00, OR = 1.00 [95% CI, 0.96–1.05] vs. no decline in 2012–2019 (9.6% to 8.7%), b = –0.02, OR = 0.98 [95% CI, 0.93–1.03], and the difference was not significant, OR = 0.98 [95% CI, 0.91–1.04]. Age differences in trends of cigarette smoking prevalence among US adults Interaction of cumulative-risk level × age Interactions of cumulative risk and age are presented in Fig. 3A–E, which reveals distinct age-specific trends in smoking prevalence across risk levels: For Age 18–25 years: smoking prevalence declined sharply from 2002 to 2019 across all risk levels, aligning with national data showing dramatic reductions among young adults, whose smoking rates dropped from ~ 35% in the early 2000s to ~ 12% by 2022 [2, 3, 7]. A–E Age × risk-level differences in trends of cigarette smoking prevalence among US adults For Age 26–34 and Age 35–49: both age groups experienced general declines across risk tiers, though the decline was more gradual compared to younger adults, consistent with earlier findings that individuals under 40 saw steeper declines in smoking than those aged 40–64 in recent years [2, 3, 7]. For Age 50–64: Smoking prevalence decreased among low-risk individuals, declined slightly among those with moderate risk, and remained stable among the high-risk group. These patterns reflect slower smoking reduction in older adults, particularly among high-risk individuals with entrenched smoking habits and greater dependence. For Age 65 + : Only low-risk adults showed slight prevalence declines; moderate-risk remained stable; high-risk showed a slight increase in smoking. This mirrors national data showing no decline—and even a modest rise—in smoking prevalence among U.S. adults 65 and older between 2011 and 2022 [2, 3, 7]. These trends highlight the interplay between age and cumulative psychosocial risk in shaping smoking behavior over time. Discussion The overarching aim of this study was to understand how disparities in smoking have changed over the last two decades, with a specific emphasis on cumulative risk factors, chronological age, and the potential interplay between those two characteristics. There was a consistent association between higher cumulative-risk levels and greater smoking risk, which aligns with earlier research demonstrating the influence of cumulative psychosocial disadvantages on smoking prevalence [11–13]. Although all risk levels exhibited significant declines in smoking prevalence, the rate of change varied across risk levels, further underscoring the critical role that cumulative risk plays in shaping smoking disparities. The relationship between age and smoking is multifaceted, with both declining and stable patterns across groups, generally consistent with prior research [3]. More specifically, smoking prevalence declined steadily among adults aged 18–25 and 35–49, mirroring national survey trends showing decreases among younger adults [3]. In contrast, adults aged 50–64 experienced only modest declines, while prevalence among those 65 and older remained largely unchanged [3]. The lack of decline in the 65 + group may reflect floor effects given the already low prevalence, underscoring the importance of attention to this often overlooked population [25]. Adults aged 26–34 exhibited mixed trends, falling between the clearer patterns observed in younger and older groups. Overall, these findings emphasize the importance of considering age when examining smoking patterns over time. Across all five age groups, increases in cumulative risk were associated with greater smoking risk. That is, those in the High-risk group had the highest smoking prevalence, followed by the Mod-risk and Low-risk groups. Importantly, risk level and age interacted with each other. Although there was a declining trend in smoking patterns over time, the rate and direction of change varied by cumulative-risk levels and by age groups. Our findings demonstrate distinct age-related trajectories in smoking prevalence when examined through the lens of cumulative risk. Young adults (18–25) experienced pronounced declines across all risk levels, paralleling documented steep reductions in smoking in this demographic [2, 3, 7]. In contrast, older adults (50 +) displayed attenuated declines, especially among high-risk individuals, suggesting behavioral entrenchment and higher dependence [2, 3, 7]. For adults 65 + , we observed no decline, and even a slight increase among high-risk individuals, consistent with national trends showing stable or rising smoking prevalence in this age group from 2011–2022 [2, 3, 7]. These divergent trajectories underscore the interaction between age, cumulative risk, and public health progress, highlighting the need for tailored interventions across life stages. The study’s findings shed light on the intricate relationship between cumulative risk, age, and smoking patterns over the 18-year period examined. This study diverges from prior research by highlighting the differential effects of cumulative risk and smoking risk across various age groups. While previous studies have acknowledged age-specific smoking patterns, they often did not account for the compounded effects of multiple risk factors in chronic disease epidemiology [26]. The age by cumulative psychosocial risk findings also have policy implications that cessation strategies may need to be age- and risk-tailored. For example, younger adults (18–34) with elevated risk, who showed steeper declines but remained vulnerable to psychosocial triggers, may be best supported by interactive digital modalities, like mobile or web platforms, which readily meet them in their preferred environments and foster personalized engagement. Conversely, perhaps midlife and older adults (50 +) with high risk, who exhibited persistent smoking, are likely to benefit most from integrated support, combining pharmacotherapy and counseling delivered through healthcare settings. Aligning intervention intensity and format with both age and cumulative-risk burden may have the potential to help concentrate resources on the populations most resistant to generalized campaigns, ultimately reducing smoking disparities. In summary, this study’s findings both corroborate and extend upon existing literature, emphasizing the importance of considering cumulative risk factors alongside age when analyzing smoking patterns. This approach also has the potential to provide a more robust framework for developing effective smoking cessation strategies tailored to different risk and age groups. Limitations While we believe this study has the potential to contribute significantly to our understanding of smoking patterns and their relationship with cumulative risk factors and age, there are limitations to consider. First, the study relies on self-reported data with the risk of underreporting. Although self-reported data on smoking are generally valid [27], it tends to underestimate smoking prevalence [28]. Second, the study considers a broad range of cumulative risk factors, but the selection of these factors does not encompass all potentially relevant variables that could influence smoking risk. Third, we acknowledge that our analysis does not fully account for the nuanced interplay of age, period, and cohort effects. Key federal and state tobacco control policy, such as the 2009 Family Smoking Prevention and Tobacco Control Act, state-level tax increases, and smoke-free air laws, likely contributed to declines in smoking during the study period. In addition, the absence of state identifiers in the public-use NSDUH file prevents direct modeling of policy exposure. Nevertheless, the primary aim of this study was not to estimate policy effects but to examine how age and cumulative psychosocial risk jointly influence smoking prevalence over time. Future research incorporating state-level policy data, regional variation, or longitudinal designs could more directly assess the contribution of tobacco control policies to observed trends in smoking disparities. Cohort effects present another limitation, as smoking behaviors can vary significantly across cohorts due to differences in socialization, policy exposure, and smoking norms. While this study examined patterns in smoking prevalence by age group across survey years, we caution that these results reflect age-by-period patterns and should not be interpreted as isolating cohort effects. Specifically, age reflects life stage or developmental period (e.g., emerging adulthood, midlife); period refers to calendar-year-specific influences that affect all age groups (e.g., national tobacco policy changes); cohort captures shared experiences among people born around the same time (e.g., those growing up during a period of strong tobacco control). Because the NSDUH public-use data include only age in broad brackets and do not contain exact age or birth year, we could not conduct formal age-period-cohort modeling. Future research should focus on integrating longitudinal datasets and employing advanced age-period-cohort modeling techniques to capture the nuanced interaction of age, period, and cohort effects. Incorporating policy-related variables, such as state-level regulations and cessation interventions, would provide a clearer picture of how federal and state policies shape smoking behaviors. Conclusions This study has the potential to inform national public health policies on reducing smoking-related health disparities by shedding light on the intricate relationships between cumulative risk, age, and smoking prevalence. The study underscores the need for targeted public health interventions and policies that consider the unique challenges faced by different age groups and those with varying cumulative risk levels. Despite the overall substantial declines in smoking, important disparities persist, highlighting the importance of considering these differences when developing U.S. tobacco control and regulatory policies. Supplementary Information Additional file 1: e-Results-Quit Ratios. Additional file 2: e-Methods – Description of Risk Factors and Race/Ethnicity. e-Table 1 Individual Risk Factors and Association with Cumulative Risk Index. e-Table 2 Individual Risk Factors and Association with Smoking Behaviors. e-Fig. 1 Smoking Disparities—Odds Ratio Over Time for Cigarette Smoking Prevalence. Acknowledgements We thank Substance Abuse and Mental Health Services Administration (SAMHSA) for providing access to the NSDUH 2002–2019 datasets. Abbreviations Confidence Interval National Health Interview Survey National Survey on Drug Use and Health Odds ratio Research Triangle Institute Substance Abuse and Mental Health Services Administration Tobacco Centers of Regulatory Science Authors’ contributions F.F.C. contributed significantly to the conceptualization and design of the study, as well as to the analysis and interpretation of data, and the writing of the manuscript. E.K. T.H., M.D., A.L. and S.H. played key roles in the study design, data analysis, interpretation of the findings, and revisions of the manuscript. Funding This research was supported by National Institute of General Medical Sciences (NIGMS) Center of Biomedical Research Excellence Award P30GM149331 and Tobacco Centers of Regulatory Science (TCORS) award U54DA036114 from the National Institute on Drug Abuse (NIDA) and Food and Drug Administration (FDA). The views in this report represents the views of the authors and may not represent the views of NIGMS, NIDA, or FDA. Data availability This study uses publicly available data from the National Survey on Drug Use and Health (NSDUH), 2002–2019, which can be accessed through the Substance Abuse & Mental Health Data Archive (SAMHDA) at [https://www.datafiles.samhsa.gov/](https://www.datafiles.samhsa.gov). Declarations Ethics approval and consent to participate This study is based on secondary analysis of the NSDUH 2002–2019 public-use datasets, which are de-identified and publicly available. Therefore, ethics approval and informed consent were not required. Consent for publication All authors approved the final manuscript and consent to publication. Competing interests The authors declare no competing interests. Footnotes Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials Additional file 1: e-Results-Quit Ratios. Additional file 2: e-Methods – Description of Risk Factors and Race/Ethnicity. e-Table 1 Individual Risk Factors and Association with Cumulative Risk Index. e-Table 2 Individual Risk Factors and Association with Smoking Behaviors. e-Fig. 1 Smoking Disparities—Odds Ratio Over Time for Cigarette Smoking Prevalence. Data Availability Statement This study uses publicly available data from the National Survey on Drug Use and Health (NSDUH), 2002–2019, which can be accessed through the Substance Abuse & Mental Health Data Archive (SAMHDA) at [https://www.datafiles.samhsa.gov/](https://www.datafiles.samhsa.gov).