Advertisement

Smoking Disparities by Level of Educational Attainment and Birth Cohort in the U.S.

Open AccessPublished:January 15, 2023DOI:https://doi.org/10.1016/j.amepre.2022.06.021

      Introduction

      Little is known about how U.S. smoking patterns of initiation, cessation, and intensity vary by birth cohort across education levels or how these patterns may be driven by other demographic characteristics.

      Methods

      Smoking data for adults aged ≥25 years was obtained from the National Health Interview Surveys 1966–2018. Age-period-cohort models were developed to estimate the probabilities of smoking initiation, cessation, intensity, and prevalence by age, cohort, calendar year, and gender for education levels: ≤8th grade, 9th–11th grade, high school graduate or GED, some college, and college degree or above. Further analyses were conducted to identify the demographic factors (race/ethnicity and birthplace) that may explain the smoking patterns by education. Analyses were conducted in 2020–2021.

      Results

      Smoking disparities by education have increased by birth cohort. In recent cohorts, initiation probabilities were highest among individuals with 9th–11th-grade education and lowest among individuals with at least a college degree. Cessation probabilities were higher among those with higher education. Current smoking prevalence decreased over time across all education groups, with important differences by gender. However, it decreased more rapidly among individuals with ≤8th grade education, resulting in this group having the second lowest prevalence in recent cohorts. This may be driven by the increasing proportion of non-U.S. born Hispanics in this group.

      Conclusions

      Although smoking is decreasing by cohort across all education groups, disparities in smoking behaviors by education have widened in recent cohorts. Demographic changes for the ≤8th-grade education group need special consideration in analyses of tobacco use by education.

      INTRODUCTION

      Smoking is the leading cause of morbidity and mortality in the U.S.
      National Center for Chronic Disease Prevention and Health Promotion. (U.S.)
      Office on Smoking and Health.
      • Lariscy JT.
      Smoking-attributable mortality by cause of death in the United States: an indirect approach.

      Smoking& tobacco use: fast facts and fact sheets. Centers for Disease Control and Prevention. https://www.cdc.gov/tobacco/data_statistics/fact_sheets/fast_facts/index.htm. Updated October 6, 2021. Accessed May 28, 2022.

      More than 400,000 smoking-attributable deaths occur each year, with millions of individuals also living with smoking-related diseases or disabilities.
      National Center for Chronic Disease Prevention and Health Promotion. (U.S.)
      Office on Smoking and Health.
      Even though smoking prevalence is steadily decreasing in the general population, the decline is not uniform, and some groups continue to have a higher prevalence than others. For instance, in 2018, current smoking prevalence was 21.8% (95% CI=19.9, 23.8) among adults with less than a high school education, compared with 7.1% (95% CI=6.2, 7.9) among adults with an undergraduate college degree.

      Current cigarette smoking among adults in the United States. Centers for Disease Control and Prevention.https://www.cdc.gov/tobacco/data_statistics/fact_sheets/adult_data/cig_smoking/index.htm. Updated December 15, 2020. Accessed December 8, 2021.

      The current inverse relationship between educational attainment and smoking prevalence is consistent with studies conducted in the U.S. and other high-income countries.
      • Zhu BP
      • Giovino GA
      • Mowery PD
      • Eriksen MP.
      The relationship between cigarette smoking and education revisited: implications for categorizing persons’ educational status.
      • Pampel FC.
      The persistence of educational disparities in smoking.
      • Huisman M
      • Kunst AE
      • Mackenbach JP.
      Inequalities in the prevalence of smoking in the European Union: comparing education and income.
      • Curtin F
      • Morabia A
      • Bernstein M.
      Smoking behavior in a Swiss urban population: the role of gender and education.
      • Giskes K
      • Kunst AE
      • Benach J
      • et al.
      Trends in smoking behaviour between 1985 and 2000 in nine European countries by education.
      • Tomioka K
      • Kurumatani N
      • Saeki K.
      The association between education and smoking prevalence, independent of occupation: a nationally representative survey in Japan.
      In addition, individuals with higher educational attainment report more interest in quitting smoking and are more likely to quit.
      National Center for Chronic Disease Prevention and Health Promotion. (U.S.)
      Office on Smoking and Health.
      ,
      U.S. National Cancer Institute
      A socioecological approach to addressing tobacco-related health disparities.
      However, little is known about how these patterns have changed over time or by generation.
      Although much of the literature on smoking prevalence by education relies on cross-sectional data, research generally does not consider how patterns change across generations. In particular, variations in smoking prevalence by education level likely reflect the patterns of initiation and cessation over time and across birth cohorts. Age-specific smoking initiation and cessation probabilities; smoking intensity; and prevalence of current, former, and never smoking have been shown to vary widely by birth cohort for the entire U.S. population and for different racial groups.
      • Harris JE.
      Cigarette smoking among successive birth cohorts of men and women in the United States during 1900-80.
      • Rodriquez EJ
      • Oh SS
      • Pérez-Stable EJ
      • Schroeder SA.
      Changes in smoking intensity over time by birth cohort and by Latino National background, 1997-2014.
      • Freedman DM
      • Tarone RE
      • Doody MM
      • et al.
      Trends in reproductive, smoking, and other chronic disease risk factors by birth cohort and race in a large occupational study population.

      Meza R, Cao P, Jeon J, et al. Patterns of birth cohort-specific smoking histories by race and ethnicity in the U.S. Am J Prev Med. In press.

      However, little is known about how these smoking patterns may vary by cohort for different education groups in the U.S. and other countries.
      • Pampel F
      • Legleye S
      • Goffette C
      • Piontek D
      • Kraus L
      • Khlat M.
      Cohort changes in educational disparities in smoking: France, Germany and the United States.
      • Escobedo LG
      • Peddicord JP.
      Smoking prevalence in U.S. birth cohorts: the influence of gender and education.
      • Manuel DG
      • Wilton AS
      • Bennett C
      • Rohit Dass A
      • Laporte A
      • Holford TR
      Smoking patterns based on birth-cohort-specific histories from 1965 to 2013, with projections to 2041.
      • Midlöv P
      • Calling S
      • Sundquist J
      • Sundquist K
      • Johansson SE.
      The longitudinal age and birth cohort trends of smoking in Sweden: a 24-year follow-up study.
      • Christopoulou R
      • Lillard DR
      • Balmori de la Miyar JR.
      Smoking behavior of Mexicans: patterns by birth-cohort, gender, and education.
      The failure to distinguish by cohort may be owing to the lack of long-term, nationally representative longitudinal studies as well as methodologic challenges to estimating cohort patterns from cross-sectional survey data.
      The theory of epidemiologic transitions of social behaviors indicates that individuals from higher SES backgrounds initially adopt behaviors (e.g., cigarette smoking, fast-food consumption) before their counterparts from lower SES backgrounds but then quit these behaviors earlier because they recognize their underlying harms.
      • Fleischer NL
      • McKeown RE.
      The second epidemiologic transition from an epidemiologist's perspective.
      ,
      • Pearson TA.
      Education and income: double-edged swords in the epidemiologic transition of cardiovascular disease.
      In contrast, individuals from low SES backgrounds are often late adopters, because of economic or social costs and lack of access, but then tend to maintain risky behaviors at higher rates than individuals from high SES backgrounds because of a variety of reasons, including less support for quitting and targeted marketing from industries.
      • Fleischer NL
      • McKeown RE.
      The second epidemiologic transition from an epidemiologist's perspective.
      • Pearson TA.
      Education and income: double-edged swords in the epidemiologic transition of cardiovascular disease.

      Liber AC, Sánchez-Romero LM, Cadham CJ, et al. Tobacco couponing: a systematic review of exposures and effects on tobacco initiation and cessation. Nicotine Tob Res. In press. Online February 10, 2022. https://doi.org/10.1093/ntr/ntac037.

      Investigating cohort trends in U.S. smoking patterns across education levels with long-term data may help to provide further insights into the dynamics of the social transition of smoking behaviors.
      In this study, an extension of previously developed methodology based on age-period-cohort statistical models and data from the National Health Surveys 1966–2018 was conducted to estimate trends in smoking initiation and cessation probabilities, smoking prevalence, and smoking intensity by birth cohort and gender for 5 education groups: ≤8th grade, 9th–11th grade, high school graduate or GED, some college, and at least a college degree. Because variations in demographic composition across education groups is known to influence differences in smoking patterns by education,
      • Nguyen-Grozavu FT
      • Pierce JP
      • Sakuma KK
      • et al.
      Widening disparities in cigarette smoking by race/ethnicity across education level in the United States.
      ,
      • Pampel F
      • Khlat M
      • Bricard D
      • Legleye S.
      Smoking among immigrant groups in the United States: prevalence, education gradients, and male-to-female ratios.
      trends in the racial/ethnic and global region of birth composition within education groups have been examined to provide additional context to smoking patterns and disparities by educational attainment.

      METHODS

      Study Sample

      Publicly available deidentified data from the National Health Interview Surveys (NHIS) from 1966 to 2018 were utilized for this study. NHIS started to collect smoking-related information since 1965, but the 1965 survey had different education categories from those of subsequent years, so the 1965 data were excluded. Data from 2019 onward were excluded because of survey design and education assessment changes beginning in 2019.
      Questionnaire redesign.
      Survey data and questions utilized to construct smoking variables are presented in Appendix Text 1 (available online).

      Measures

      Educational attainment level was classified into 5 categories―(1) ≤8th grade, (2) 9th–11th grade, (3) high school graduate or GED, (4) some college, and (5) at least a college degree―to reflect major education groups across birth cohorts. Consistent with other analyses, all analyses were restricted to individuals aged ≥25 years because most people reach their highest education attainment by age 25 years.
      U.S. Census Bureau
      Educational attainment in the United States; 2020.

      Statistical Analysis

      This was done by adapting the Holford et al.
      • Holford TR
      • Levy DT
      • McKay LA
      • et al.
      Patterns of birth cohort-specific smoking histories, 1965-2009.
      methodology to estimate historical smoking patterns for each education group. For each group, modes were applied to estimate the following age-specific smoking parameters by single-year birth cohort (1900–2000), calendar year (1966–2018), and gender using NHIS sample adult weights, which ensure that the data are nationally representative: smoking initiation and cessation probabilities; prevalence of ever, current, and former smoking; and smoking intensity among adults who were currently smoking. Trends in smoking intensity were estimated using the reported number of cigarettes per day (CPD), which is a proxy measure of smoking exposure and to calculate pack-years; these measures are used in risk prediction models
      • Tammemägi MC
      • Katki HA
      • Hocking WG
      • et al.
      Selection criteria for lung-cancer screening.
      and to determine lung cancer screening eligibility in the U.S.
      • Krist AH
      • Davidson KW
      • et al.
      U.S. Preventive Services Task Force
      Screening for lung cancer: U.S. Preventive Services Task Force recommendation statement.
      A detailed description of the model and model parameters are presented in Appendix Text 2 (available online).
      Because variations in demographic composition across education groups are known to affect smoking disparities by education,
      • Nguyen-Grozavu FT
      • Pierce JP
      • Sakuma KK
      • et al.
      Widening disparities in cigarette smoking by race/ethnicity across education level in the United States.
      ,
      • Pampel F
      • Khlat M
      • Bricard D
      • Legleye S.
      Smoking among immigrant groups in the United States: prevalence, education gradients, and male-to-female ratios.
      demographic characteristics of the 5 education groups considered were examined. Specifically, this was done by examining the percentage of American Indian/Alaska Native, Asian and Pacific Islander, Hispanic, non-Hispanic Black, and non-Hispanic White individuals in each education group from 1978 to 2018 and the percentage of individuals born in the U.S. and in other regions of the globe in each education group from 2000 to 2018.

      RESULTS

      Age-specific smoking patterns by level of educational attainment and gender for selected birth cohorts 1910, 1930, 1950, 1970, and 1990 are shown in Figure 1, Figure 2, Figure 3, Figure 4. Additional cohorts by calendar year are shown in Figure 1, Figure 2, Figure 3, Figure 4 (available online). Interactive figures by user-specified birth cohort, education, smoking parameter, and plot perspective could be found at https://apps.cisnetsmokingparameters.org/education/.
      Figure 1
      Figure 1Age-specific smoking initiation probabilities (percentage) for selected birth cohorts by education attainment and gender (females—top panels, males—bottom panels).
      HSG, high school graduate.
      Note: Lines represent the initiation probabilities for ≤8th grade (red), 9th–11th grade (orange), HSG or GED (sky blue), some college (blue), and college degree or above (black). An interactive version of this figure's data can be found at https://apps.cisnetsmokingparameters.org/education/.
      Figure 2
      Figure 2Age-specific smoking-cessation probabilities (percentage) for selected birth cohorts by education attainment and gender (females—top panels, males—bottom panels).
      HSG, high school graduate.
      Note: Lines represent the initiation probabilities for ≤8th grade (red), 9th–11th grade (orange), HSG or GED (sky blue), some college (blue), and college degree or above (black). An interactive version of this figure's data can be found at https://apps.cisnetsmokingparameters.org/education/.
      Figure 3
      Figure 3Age-specific current smoker prevalence for selected birth cohorts by education attainment and gender (females—top panels, males—bottom panels).
      HSG, high school graduate.
      Note: Lines represent the initiation probabilities for ≤8th grade (red), 9th–11th grade (orange), HSG or GED (sky blue), some college (blue), and college degree or above (black). An interactive version of this figure's data can be found at https://apps.cisnetsmokingparameters.org/education/.
      Figure 4
      Figure 4Age-specific mean cigarettes per day for selected birth cohorts by education attainment and gender (females—top panels, males—bottom panels).
      HSG, high school graduate.
      Notes: Lines represent the initiation probabilities for ≤8th grade (red), 9–11th grade (orange), HSG or GED (sky blue), some college (blue), and college degree or above (black). An interactive version of this figure's data can be found at https://apps.cisnetsmokingparameters.org/education/.
      Overall, initiation probabilities increased by age until around ages 15–20 years and then decreased (Figure 1). The peak usually occurred at younger ages among lower education groups. Females in general had lower initiation probabilities than males. Among females, individuals with at least a college degree had the highest initiation probabilities in earlier birth cohorts (at or before 1920) than other education groups, which decreased over cohorts and become the lowest in recent cohorts (at or after 1990). Among females born after 1930, those with a 9th–11th-grade education had the highest initiation probabilities than other education groups. Females in the 8th-grade or less educational group had the lowest smoking initiation probabilities in cohorts born before 1980 and the second lowest in cohorts born after 1990. Differences in initiation by education level were relatively minor among females before the 1930 birth cohort but have widened and persisted after the 1950 birth cohort. Among males, initiation probabilities across all education groups increased by cohort until the 1920–1930 cohorts and then decreased thereafter. Before the 1920 cohort, the initiation probabilities among males were highest among individuals with 9th–11th-grade education, followed by males with ≤8th-grade education, high school graduate or GED, some college, and college or above. However, in cohorts born after 1950, males with ≤8th-grade education had the second lowest initiation probabilities. Relative differences in male initiation probabilities by education have also increased by cohort.
      Birth cohort‒specific cessation probabilities increased by age in general. However, for cohorts born after 1950, cessation probabilities increased rapidly at young ages, peaked around age 40 years, decreased until age 50 years, and then increased in older ages (Figure 2). The decrease in older ages may be an artifact of the analytic approach, which relies on data from older cohorts to project cessation rates at older ages for recent cohorts; probabilities shown beyond 2018 for more recent cohorts are based on estimates from earlier cohorts at the same age. For example, individuals born in the 1970 cohort were aged 48 years in 2018, which is near the point at which this peak is observed. At younger ages, cessation rates increased rapidly in cohorts born after 1990. This pattern was more apparent for individuals with some college or a college degree or above, reflecting larger increases in cessation among young adults in this group. Cessation probabilities were highest among individuals with a college degree or above, followed by some college, whereas the lowest 3 education groups had similar levels of cessation probabilities. Differences in cessation probabilities between education groups have been widening after the 1950 birth cohort. Cessation patterns were similar between females and males, with males having slightly higher cessation rates than females in earlier birth cohorts (at or before 1910).
      Smoking initiation and cessation probabilities were used to estimate current smoking prevalence by age, birth cohort, gender, and education level (Figure 3). With their generally lowest initiation probabilities and highest cessation probabilities, the college or above education group had the lowest current smoker prevalence across all ages and birth cohorts for both genders, except females born before the 1910 birth cohort. Among individuals with >8th-grade education, starting with the 1950 cohort, smoking prevalence was the highest overall among individuals with 9th–11th-grade education, followed by individuals with a high school diploma or GED and then those with some college education. Smoking prevalence among individuals with ≤8th-grade education was highest or second highest for cohorts born before 1930, except for females born before 1910, but decreased at a fast rate and become second or third lowest after the 1970 birth cohort. Increasing differences in both smoking initiation and cessation across birth cohorts by education level resulted in correspondingly greater differences in current smoking by education level across cohorts.
      Before cohort 1950, the average CPD increased with age until ages 40–50 years and then decreased (Figure 4); however, in recent cohorts, average CPD remained relatively constant until age 50 years and then decreased. In general, males had slightly higher CPD consumption than females.
      Average CPD did not differ much by education group before the 1930 birth cohort but showed a clear gradient, with differences widening by education level after the 1950 birth cohort. The highest education group had the highest average CPD among males born before 1930 and the second highest among females born before 1910 but the lowest average CPD after the 1950 birth cohort. Males with 9th–11th-grade education and born after 1950 birth cohort had the highest CPD, followed by those with high school degree or GED, some college, ≤8th-grade education (becoming third highest after 1990 birth cohort), or at least a college degree. Among females born after 1970, the average CPD was highest among those with ≤8th-grade education, followed by those with 9th–11th-grade education, high school degree or GED, some college, or at least a college degree.
      Because of their higher initiation and lower cessation probabilities, lower education groups in general had longer mean smoking duration across birth cohorts (Appendix Figure 5, available online). Furthermore, consistent with observed CPD patterns, lower education groups had higher pack-years in recent birth cohorts, except for individuals with ≤8th-grade education who had pack-years similar to those seen among those with some college education (Appendix Figure 6, available online).
      Figure 5
      Figure 5Distribution of race/ethnicity across years by education attainment using the National Health Interview Surveys 1978–2018 data.
      AIAN, American Indian/Alaska Native; API, Asian and Pacific Islander; HISP, Hispanic; NHB, non-Hispanic Black; NHW, non-Hispanic White.
      Figure 6
      Figure 6Distribution of global region of birth across years by education attainment using the National Health Interview Surveys 2000–2018 data.
      Note: Top panel: whole U.S.; bottom panel: immigrants only.
      Race/ethnicity and global region of birth distributions vary considerably over time across education groups. In particular, the proportion of Hispanic adults in the ≤8th-grade education group has increased considerably since the mid-1990s, reaching ∼50% of the group in recent years (Appendix Table 1, available online, and Figure 5). Furthermore, the proportion of non-U.S. born individuals in the ≤8th-grade education group also increased considerably, reaching >72% in 2018, with most foreign-born individuals coming from Mexico, Central America, and the Caribbean (Figure 6).

      DISCUSSION

      Nationally representative survey data from 1966 to 2018 were used to examine smoking behaviors by age, birth cohort, calendar year, and gender for 5 educational attainment groups. Current smoking prevalence decreased in all education groups; however, differences between groups widened over time. Among males, smoking prevalence was higher among individuals with lower educational attainment across all cohorts. Among females, smoking prevalence was higher among those who were more highly educated in earlier cohorts but was higher among those with less education in more recent cohorts. One exception to these patterns is that in recent birth cohorts, individuals with ≤8th-grade education had low smoking rates similar to those seen among individuals with some college education. This is largely explained by the increasing proportion of Hispanic and foreign-born individuals from Mexico and Central America in that group, who have relatively low smoking rates.
      The estimated widening of smoking disparities by education is consistent with previous findings.
      • Pampel FC.
      The persistence of educational disparities in smoking.
      ,
      U.S. National Cancer Institute
      A socioecological approach to addressing tobacco-related health disparities.
      ,
      • Pierce JP
      • Fiore MC
      • Novotny TE
      • Hatziandreu EJ
      • Davis RM.
      Trends in cigarette smoking in the United States. Educational differences are increasing.
      This study further shows how these disparities developed over generations for different smoking behaviors that occur throughout the life course (initiation, cessation, and intensity). Smoking initiation usually occurs early in life before the highest level of education attainment is acquired. Hence, disparities in initiation reflect the influence of factors such as lower family income or lower parental education, which are also associated with higher rates of smoking experimentation and lower rates of educational achievement.
      • Maralani V.
      Understanding the links between education and smoking.
      • Assari S
      • Mistry R
      • Caldwell CH
      • Bazargan M.
      Protective effects of parental education against youth cigarette smoking: diminished returns of blacks and Hispanics.
      American Psychological Association
      Education and socioeconomic status factsheet.
      • Chassin L
      • Presson CC
      • Sherman SJ
      • Edwards DA.
      Parent educational attainment and adolescent cigarette smoking.
      By contrast, smoking cessation occurs later in life. Individuals with lower educational attainment are known to face more social and financial stressors during adulthood, with less access to resources and information that can support the adoption of healthier behaviors.
      • Businelle MS
      • Kendzor DE
      • Reitzel LR
      • et al.
      Mechanisms linking socioeconomic status to smoking cessation: a structural equation modeling approach.
      Moreover, tobacco companies are known to target individuals from low SES and education backgrounds with price discounts and other marketing strategies, which also affect smoking rates.

      Liber AC, Sánchez-Romero LM, Cadham CJ, et al. Tobacco couponing: a systematic review of exposures and effects on tobacco initiation and cessation. Nicotine Tob Res. In press. Online February 10, 2022. https://doi.org/10.1093/ntr/ntac037.

      Similar factors influence the intensity of smoking among people who currently smoke, measured in this study as CPD and pack-years, which also showed increasing relative differences between low and high education groups by cohort.
      The analysis estimates provide a view of the history of smoking behaviors across different education and gender groups during the past century and the social transition of these behaviors. Because the smoking epidemic for females started later than for males, a full social transition of smoking from higher education to lower education could be observed for females across birth cohorts for each smoking behavior analyzed: initiation, cessation, intensity, and prevalence.
      • Fleischer NL
      • McKeown RE.
      The second epidemiologic transition from an epidemiologist's perspective.
      In contrast, among males, individuals with lower education had higher smoking prevalence and initiation probabilities and lower cessation probabilities across cohorts. However, patterns of smoking intensity (mean CPD) by cohort among males did show a transition from higher levels among highly educated groups in earlier cohorts to higher levels among less educated groups in recent cohorts. These findings are consistent with those of the few studies that examined the social transition theory empirically in the U.S.
      • Pampel F
      • Legleye S
      • Goffette C
      • Piontek D
      • Kraus L
      • Khlat M.
      Cohort changes in educational disparities in smoking: France, Germany and the United States.
      ,
      • Escobedo LG
      • Peddicord JP.
      Smoking prevalence in U.S. birth cohorts: the influence of gender and education.
      For example, Escobedo et al. examined cohort trends of current smoking prevalence by age, gender, education level, and race/ethnicity using U.S. survey data from 1978 to 1988 and found decreases by cohort in smoking prevalence among individuals with at least a high school education but no decreases among individuals with less than a high school education.
      • Escobedo LG
      • Peddicord JP.
      Smoking prevalence in U.S. birth cohorts: the influence of gender and education.
      The study also found differences in these patterns by gender, with smoking prevalence in earlier birth cohorts higher among females with at least high school education than among females with lower education but lower among more educated females in more recent cohorts.
      • Escobedo LG
      • Peddicord JP.
      Smoking prevalence in U.S. birth cohorts: the influence of gender and education.
      The finding of relatively low smoking rates among individuals with ≤8th-grade education indicate that this group deserves special consideration. This could be potentially explained by increasing proportions of Hispanic adults and immigrants from Mexico and Latin America in the ≤8th-grade education group. Hispanic adults have a lower smoking prevalence than other U.S. racial‒ethnic groups, particularly in recent birth cohorts.

      Meza R, Cao P, Jeon J, et al. Patterns of birth cohort-specific smoking histories by race and ethnicity in the U.S. Am J Prev Med. In press.

      Smoking is also less prevalent among individuals with less education in Mexico, with Mexican immigrants in the U.S. having lower smoking rates than U.S.-born Mexican Americans.
      • Christopoulou R
      • Lillard DR
      • Balmori de la Miyar JR.
      Smoking behavior of Mexicans: patterns by birth-cohort, gender, and education.
      ,
      • Fleischer NL
      • Ro A
      • Bostean G.
      Smoking selectivity among Mexican immigrants to the United States using binational data, 1999-2012.
      ,
      • Bostean G
      • Ro A
      • Fleischer NL.
      Smoking trends among U.S. Latinos, 1998-2013: the Impact of Immigrant Arrival Cohort.
      Interestingly, among U.S.-born individuals, smoking prevalence among adults with ≤8th-grade education was also lower than among adults with 9th–11th-grade education.
      • Zhu BP
      • Giovino GA
      • Mowery PD
      • Eriksen MP.
      The relationship between cigarette smoking and education revisited: implications for categorizing persons’ educational status.
      Individuals from this group might not experience socialization into smoking that mostly occurs during high school, resulting in lower smoking rates.
      • Zhu BP
      • Giovino GA
      • Mowery PD
      • Eriksen MP.
      The relationship between cigarette smoking and education revisited: implications for categorizing persons’ educational status.
      The results strongly suggest that studies of tobacco use by education need to disaggregate the education groups with less than a high school education into finer subcategories to properly capture the behaviors of individuals with the lowest levels of education.
      The results for 5 different education groups could guide interventions that facilitate greater reductions in smoking among groups with high smoking rates. For example, adults in the 9th–11th-grade education group have the highest smoking prevalence because of having the highest smoking initiation rates and lowest cessation rates. Hence, tobacco control interventions that prevent initiation, such as school-based smoking prevention curricula, should start early—ideally before 9th grade—to reach adolescents who may be at potentially higher risk of initiating smoking during high school. Although the ≤8th-grade group generally has low smoking initiation probabilities, they also have low smoking-cessation probabilities. It is therefore important to prioritize resources for effective smoking-cessation interventions for people with less education.
      The estimates can also facilitate the development of simulation models to assess the impact of tobacco control interventions by education group.
      • Tam J
      • Jeon J
      • Thrasher JF
      • et al.
      Estimated prevalence of smoking and smoking-attributable mortality associated with graphic health warnings on cigarette packages in the U.S. from 2022 to 2100.
      • Tam J
      • Levy DT
      • Jeon J
      • et al.
      Projecting the effects of tobacco control policies in the USA through microsimulation: a study protocol.
      • Le TT
      • Mendez D.
      An estimation of the harm of menthol cigarettes in the United States from 1980 to 2018.
      • Levy DT
      • Tam J
      • Sanchez-Romero LM
      • et al.
      Public health implications of vaping in the USA: the smoking and vaping simulation model.
      Previous estimates for the overall U.S. population by gender
      • Holford TR
      • Levy DT
      • McKay LA
      • et al.
      Patterns of birth cohort-specific smoking histories, 1965-2009.
      ,
      • 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.
      and race
      • Holford TR
      • Levy DT
      • Meza R.
      Comparison of smoking history patterns among African American and white cohorts in the United States born 1890 to 1990.
      have already been used for simulation models of lung cancer outcomes
      • 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.
      ,
      • Meza R
      • Jeon J
      • Toumazis I
      • et al.
      Evaluation of the benefits and harms of lung cancer screening with low-dose computed tomography: modeling study for the U.S. Preventive Services Task Force.
      • Cao P
      • Jeon J
      • Levy DT
      • et al.
      Potential impact of cessation interventions at the point of lung cancer screening on lung cancer and overall mortality in the United States.
      • de Koning HJ
      • Meza R
      • Plevritis SK
      • et al.
      Benefits and harms of computed tomography lung cancer screening strategies. A comparative modeling study for the U.S. Preventive Services Task Force.
      • Meza R
      • Cao P
      • Jeon J
      • et al.
      Impact of joint lung cancer screening and cessation interventions under the new recommendations of the U.S. Preventive Services Task Force.
      • Moolgavkar SH
      • Holford TR
      • Levy DT
      • et al.
      Impact of reduced tobacco smoking on lung cancer mortality in the United States during 1975–2000.
      • Toumazis I
      • de Nijs K
      • Cao P
      • et al.
      Cost-effectiveness evaluation of the 2021 U.S. Preventive Services Task Force recommendation for lung cancer screening.
      as well as simulation models of tobacco control interventions.
      • Tam J
      • Jeon J
      • Thrasher JF
      • et al.
      Estimated prevalence of smoking and smoking-attributable mortality associated with graphic health warnings on cigarette packages in the U.S. from 2022 to 2100.
      • Tam J
      • Levy DT
      • Jeon J
      • et al.
      Projecting the effects of tobacco control policies in the USA through microsimulation: a study protocol.
      • Le TT
      • Mendez D.
      An estimation of the harm of menthol cigarettes in the United States from 1980 to 2018.
      • Levy DT
      • Tam J
      • Sanchez-Romero LM
      • et al.
      Public health implications of vaping in the USA: the smoking and vaping simulation model.
      ,
      • Holford TR
      • Meza R
      • Warner KE
      • et al.
      Tobacco control and the reduction in smoking-related premature deaths in the United States, 1964–2012.
      ,
      • Apelberg BJ
      • Feirman SP
      • Salazar E
      • et al.
      Potential public health effects of reducing nicotine levels in cigarettes in the United States.
      Future analyses can use these estimates to evaluate the potential impact of different interventions or policies on smoking prevalence and disparities across education groups.

      Limitations

      This study has limitations. Education was classified into 5 groups, and the proportion of ≤8th-grade education dropped from 30% in 1966 to 3.4% in 2018. The reduction in sample size in recent years may result in unstable estimation for recent birth cohorts. For females with ≤8th-grade education, a simplified version of the corresponding age-period-cohort models (reducing the df) was used to obtain stable results (Appendix Text 2, available online). Second, the high school graduate group included those with a GED degree. Individuals with GED degrees have higher rates of smoking than high school graduates.
      • Martinez SA
      • Beebe LA
      • Terrell DR
      • Thompson DM
      • Campbell JE.
      Tobacco use patterns among GED recipients.
      However, GED information was not collected in NHIS until 1997. Future analyses will attempt to estimate separate smoking parameters for those with GED using data from NHIS 1997 and onward. Third, the survey data on which this analysis is based may be subject to bias for several reasons: smoking prevalence may be underestimated when based on self-report of a socially undesirable health behavior,
      • Klesges RC
      • Debon M
      • Ray JW.
      Are self-reports of smoking rate biased? Evidence from the second National Health and nutrition Examination Survey.
      declining survey response rates may increase the potential for nonresponse bias,
      • Czajka JL
      • Beyler A.
      Declining Response Rates in Federal Surveys: Trends and Implications (background paper).
      and reliance on self-reported retrospective data on smoking initiation and cessation may be subject to recall bias.
      • Volk RJ
      • Mendoza TR
      • Hoover DS
      • Nishi SPE
      • Choi NJ
      • Bevers TB.
      Reliability of self-reported smoking history and its implications for lung cancer screening.
      Furthermore, the NHIS surveys the household non-institutionalized population, thereby excluding homeless, psychiatric inpatient, or imprisoned populations who are disproportionately of low SES. In addition, self-reported CPD may also be biased toward round numbers (data heaping), particularly because cigarette packs are standardized to 20 cigarettes.
      • Klesges RC
      • Debon M
      • Ray JW.
      Are self-reports of smoking rate biased? Evidence from the second National Health and nutrition Examination Survey.
      ,
      • Wang H
      • Heitjan DF.
      Modeling heaping in self-reported cigarette counts.
      This issue was addressed using CPD categories that have frequently reported CPD values as mean values.
      Finally, smoking patterns by education level were characterized in this study, but the results do not consider underlying mechanisms nor can they determine whether an actual causal relationship exists between education and smoking. For example, smoking initiation probabilities for individuals with a college degree or above were the lowest among recent birth cohorts. However, the average age of college graduates is around 22 years, which is past the peak age of smoking initiation (around 18 years). Therefore, obtaining a college degree or above is unlikely to directly reduce smoking initiation rates but instead reflects the influence of other individual and family and social characteristics that affect initiation.
      • Maralani V.
      Understanding the links between education and smoking.

      CONCLUSIONS

      Historical birth cohort patterns of smoking behaviors were estimated for different education groups. Although current smoking prevalence decreased by birth cohort across all education groups, disparities in smoking behaviors by educational attainment have widened in recent birth cohorts. These disparities justify targeted interventions for people with less education. Finally, relatively low smoking prevalence and changes in demographic composition among adults with ≤8th-grade education indicate that this group merits special consideration in future studies of tobacco use by education.

      CRediT AUTHOR STATEMENT

      Pianpian Cao: Conceptualization, Data curation, Formal analysis, Methodology, Visualization, Software, Writing–original draft. Jihyoun Jeon: Conceptualization, Formal analysis, Visualization, Writing–review and editing. Jamie Tam: Conceptualization, Formal analysis, Writing–review and editing. Nancy L. Fleischer: Conceptualization, writing–review and editing. David T. Levy: Conceptualization, Funding acquisition, Writing–review and editing. Theodore R. Holford: Conceptualization, Funding acquisition, Methodology, Writing–review and editing. Rafael Meza: Conceptualization, Formal analysis, Funding acquisition, Methodology, Software, Supervision, Visualization, Writing–review and editing.

      ACKNOWLEDGMENTS

      The study sponsor had no role in the study design; collection, analysis, and interpretation of data; writing of the report; or the decision to submit the report for publication.
      This project was funded through National Cancer Institute Grants U01CA199284 and U01CA253858.
      No financial disclosures were reported by the authors of this paper.

      Appendix. SUPPLEMENTAL MATERIAL

      REFERENCES

        • National Center for Chronic Disease Prevention and Health Promotion. (U.S.)
        Office on Smoking and Health.
        Health Consequences Smok-50 Years of Progress: A Report of the Surgeon General. U.S.: Centers for Disease Control and Prevention, 2014 (Accessed December 11, 2019)
        • Lariscy JT.
        Smoking-attributable mortality by cause of death in the United States: an indirect approach.
        SSM Popul Health. 2019; 7100349https://doi.org/10.1016/j.ssmph.2019.100349
      1. Smoking& tobacco use: fast facts and fact sheets. Centers for Disease Control and Prevention. https://www.cdc.gov/tobacco/data_statistics/fact_sheets/fast_facts/index.htm. Updated October 6, 2021. Accessed May 28, 2022.

      2. Current cigarette smoking among adults in the United States. Centers for Disease Control and Prevention.https://www.cdc.gov/tobacco/data_statistics/fact_sheets/adult_data/cig_smoking/index.htm. Updated December 15, 2020. Accessed December 8, 2021.

        • Zhu BP
        • Giovino GA
        • Mowery PD
        • Eriksen MP.
        The relationship between cigarette smoking and education revisited: implications for categorizing persons’ educational status.
        Am J Public Health. 1996; 86: 1582-1589https://doi.org/10.2105/AJPH.86.11.1582
        • Pampel FC.
        The persistence of educational disparities in smoking.
        Soc Probl. 2009; 56: 526-542https://doi.org/10.1525/sp.2009.56.3.526
        • Huisman M
        • Kunst AE
        • Mackenbach JP.
        Inequalities in the prevalence of smoking in the European Union: comparing education and income.
        Prev Med. 2005; 40: 756-764https://doi.org/10.1016/j.ypmed.2004.09.022
        • Curtin F
        • Morabia A
        • Bernstein M.
        Smoking behavior in a Swiss urban population: the role of gender and education.
        Prev Med. 1997; 26: 658-663https://doi.org/10.1006/pmed.1997.0187
        • Giskes K
        • Kunst AE
        • Benach J
        • et al.
        Trends in smoking behaviour between 1985 and 2000 in nine European countries by education.
        J Epidemiol Community Health. 2005; 59: 395-401https://doi.org/10.1136/jech.2004.025684
        • Tomioka K
        • Kurumatani N
        • Saeki K.
        The association between education and smoking prevalence, independent of occupation: a nationally representative survey in Japan.
        J Epidemiol. 2020; 30: 136-142https://doi.org/10.2188/jea.JE20180195
        • U.S. National Cancer Institute
        A socioecological approach to addressing tobacco-related health disparities.
        HHS, National Institutes of Health, National Cancer Institute, Bethesda, MDPublished 2017 (Accessed December 14, 2021)
        • Harris JE.
        Cigarette smoking among successive birth cohorts of men and women in the United States during 1900-80.
        J Natl Cancer Inst. 1983; 71: 473-479https://doi.org/10.1093/jnci/71.3.473
        • Rodriquez EJ
        • Oh SS
        • Pérez-Stable EJ
        • Schroeder SA.
        Changes in smoking intensity over time by birth cohort and by Latino National background, 1997-2014.
        Nicotine Tob Res. 2016; 18: 2225-2233https://doi.org/10.1093/ntr/ntw203
        • Freedman DM
        • Tarone RE
        • Doody MM
        • et al.
        Trends in reproductive, smoking, and other chronic disease risk factors by birth cohort and race in a large occupational study population.
        Ann Epidemiol. 2002; 12: 363-369https://doi.org/10.1016/S1047-2797(01)00295-2
      3. Meza R, Cao P, Jeon J, et al. Patterns of birth cohort-specific smoking histories by race and ethnicity in the U.S. Am J Prev Med. In press.

        • Pampel F
        • Legleye S
        • Goffette C
        • Piontek D
        • Kraus L
        • Khlat M.
        Cohort changes in educational disparities in smoking: France, Germany and the United States.
        Soc Sci Med. 2015; 127: 41-50https://doi.org/10.1016/j.socscimed.2014.06.033
        • Escobedo LG
        • Peddicord JP.
        Smoking prevalence in U.S. birth cohorts: the influence of gender and education.
        Am J Public Health. 1996; 86: 231-236https://doi.org/10.2105/AJPH.86.2.231
        • Manuel DG
        • Wilton AS
        • Bennett C
        • Rohit Dass A
        • Laporte A
        • Holford TR
        Smoking patterns based on birth-cohort-specific histories from 1965 to 2013, with projections to 2041.
        Health Rep. 2020; 31: 16-31https://doi.org/10.25318/82-003-x202001100002-eng
        • Midlöv P
        • Calling S
        • Sundquist J
        • Sundquist K
        • Johansson SE.
        The longitudinal age and birth cohort trends of smoking in Sweden: a 24-year follow-up study.
        Int J Public Health. 2014; 59: 243-250https://doi.org/10.1007/s00038-013-0535-5
        • Christopoulou R
        • Lillard DR
        • Balmori de la Miyar JR.
        Smoking behavior of Mexicans: patterns by birth-cohort, gender, and education.
        Int J Public Health. 2013; 58: 335-343https://doi.org/10.1007/s00038-012-0376-7
        • Fleischer NL
        • McKeown RE.
        The second epidemiologic transition from an epidemiologist's perspective.
        in: Zuckerman MK Modern Environments and Human Health. John Wiley & Sons, Inc, Hoboken, NJ2014: 353-368https://doi.org/10.1002/9781118504338.ch19
        • Pearson TA.
        Education and income: double-edged swords in the epidemiologic transition of cardiovascular disease.
        Ethn Dis. 2003; 13 (Accessed January 29, 2022): S158-S163
      4. Liber AC, Sánchez-Romero LM, Cadham CJ, et al. Tobacco couponing: a systematic review of exposures and effects on tobacco initiation and cessation. Nicotine Tob Res. In press. Online February 10, 2022. https://doi.org/10.1093/ntr/ntac037.

        • Nguyen-Grozavu FT
        • Pierce JP
        • Sakuma KK
        • et al.
        Widening disparities in cigarette smoking by race/ethnicity across education level in the United States.
        Prev Med. 2020; 139106220https://doi.org/10.1016/j.ypmed.2020.106220
        • Pampel F
        • Khlat M
        • Bricard D
        • Legleye S.
        Smoking among immigrant groups in the United States: prevalence, education gradients, and male-to-female ratios.
        Nicotine Tob Res. 2020; 22: 532-538https://doi.org/10.1093/ntr/ntz022
      5. Questionnaire redesign.
        NHIS, Centers for Disease Control and Prevention, 2019 (Updated November 9, 2021Accessed December 29, 2021)
        • U.S. Census Bureau
        Educational attainment in the United States; 2020.
        U.S. Census Bureau, Suitland, MDPublished April 21, 2021 (Accessed January 9, 2022)
        • Holford TR
        • Levy DT
        • McKay LA
        • et al.
        Patterns of birth cohort-specific smoking histories, 1965-2009.
        Am J Prev Med. 2014; 46: e31-e37https://doi.org/10.1016/j.amepre.2013.10.022
        • Tammemägi MC
        • Katki HA
        • Hocking WG
        • et al.
        Selection criteria for lung-cancer screening.
        N Engl J Med. 2013; 368: 728-736https://doi.org/10.1056/NEJMoa1211776
        • Krist AH
        • Davidson KW
        • et al.
        • U.S. Preventive Services Task Force
        Screening for lung cancer: U.S. Preventive Services Task Force recommendation statement.
        JAMA. 2021; 325: 962-970https://doi.org/10.1001/jama.2021.1117
        • Pierce JP
        • Fiore MC
        • Novotny TE
        • Hatziandreu EJ
        • Davis RM.
        Trends in cigarette smoking in the United States. Educational differences are increasing.
        JAMA. 1989; 261: 56-60https://doi.org/10.1001/jama.261.1.56
        • Maralani V.
        Understanding the links between education and smoking.
        Soc Sci Res. 2014; 48: 20-34https://doi.org/10.1016/j.ssresearch.2014.05.007
        • Assari S
        • Mistry R
        • Caldwell CH
        • Bazargan M.
        Protective effects of parental education against youth cigarette smoking: diminished returns of blacks and Hispanics.
        Adolesc Health Med Ther. 2020; 11: 63-71https://doi.org/10.2147/AHMT.S238441
        • American Psychological Association
        Education and socioeconomic status factsheet.
        American Psychological Association, Washington, DCPublished July 2017 (Accessed February 25, 2022)
        • Chassin L
        • Presson CC
        • Sherman SJ
        • Edwards DA.
        Parent educational attainment and adolescent cigarette smoking.
        J Subst Abuse. 1992; 4: 219-234https://doi.org/10.1016/0899-3289(92)90031-R
        • Businelle MS
        • Kendzor DE
        • Reitzel LR
        • et al.
        Mechanisms linking socioeconomic status to smoking cessation: a structural equation modeling approach.
        Health Psychol. 2010; 29: 262-273https://doi.org/10.1037/a0019285
        • Fleischer NL
        • Ro A
        • Bostean G.
        Smoking selectivity among Mexican immigrants to the United States using binational data, 1999-2012.
        Prev Med. 2017; 97: 26-32https://doi.org/10.1016/j.ypmed.2017.01.004
        • Bostean G
        • Ro A
        • Fleischer NL.
        Smoking trends among U.S. Latinos, 1998-2013: the Impact of Immigrant Arrival Cohort.
        Int J Environ Res Public Health. 2017; 14: 255https://doi.org/10.3390/ijerph14030255
        • Tam J
        • Jeon J
        • Thrasher JF
        • et al.
        Estimated prevalence of smoking and smoking-attributable mortality associated with graphic health warnings on cigarette packages in the U.S. from 2022 to 2100.
        JAMA Health Forum. 2021; 2e212852https://doi.org/10.1001/jamahealthforum.2021.2852
        • 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; 8e019169https://doi.org/10.1136/bmjopen-2017-019169
        • Le TT
        • Mendez D.
        An estimation of the harm of menthol cigarettes in the United States from 1980 to 2018.
        Tob Control. 2022; 31: 564-568https://doi.org/10.1136/tobaccocontrol-2020-056256
        • Levy DT
        • Tam J
        • Sanchez-Romero LM
        • et al.
        Public health implications of vaping in the USA: the smoking and vaping simulation model.
        Popul Health Metr. 2021; 19: 19https://doi.org/10.1186/s12963-021-00250-7
        • 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: 684-693https://doi.org/10.7326/M18-1250
        • Holford TR
        • Levy DT
        • Meza R.
        Comparison of smoking history patterns among African American and white cohorts in the United States born 1890 to 1990.
        Nicotine Tob Res. 2016; 18: S16-S29https://doi.org/10.1093/ntr/ntv274
        • Meza R
        • Jeon J
        • Toumazis I
        • et al.
        Evaluation of the benefits and harms of lung cancer screening with low-dose computed tomography: modeling study for the U.S. Preventive Services Task Force.
        JAMA. 2021; 325: 988-997https://doi.org/10.1001/jama.2021.1077
        • Cao P
        • Jeon J
        • Levy DT
        • et al.
        Potential impact of cessation interventions at the point of lung cancer screening on lung cancer and overall mortality in the United States.
        J Thorac Oncol. 2020; 15: 1160-1169https://doi.org/10.1016/j.jtho.2020.02.008
        • de Koning HJ
        • Meza R
        • Plevritis SK
        • et al.
        Benefits and harms of computed tomography lung cancer screening strategies. A comparative modeling study for the U.S. Preventive Services Task Force.
        Ann Intern Med. 2014; 160: 311-320https://doi.org/10.7326/M13-2316
        • Meza R
        • Cao P
        • Jeon J
        • et al.
        Impact of joint lung cancer screening and cessation interventions under the new recommendations of the U.S. Preventive Services Task Force.
        J Thorac Oncol. 2022; 17: 160-166https://doi.org/10.1016/j.jtho.2021.09.011
        • 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: 541-548https://doi.org/10.1093/jnci/djs136
        • Toumazis I
        • de Nijs K
        • Cao P
        • et al.
        Cost-effectiveness evaluation of the 2021 U.S. Preventive Services Task Force recommendation for lung cancer screening.
        JAMA Oncol. 2021; 7: 1833-1842https://doi.org/10.1001/jamaoncol.2021.4942
        • Holford TR
        • Meza R
        • Warner KE
        • et al.
        Tobacco control and the reduction in smoking-related premature deaths in the United States, 1964–2012.
        JAMA. 2014; 311: 164-171https://doi.org/10.1001/jama.2013.285112
        • 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: 1725-1733https://doi.org/10.1056/NEJMsr1714617
        • Martinez SA
        • Beebe LA
        • Terrell DR
        • Thompson DM
        • Campbell JE.
        Tobacco use patterns among GED recipients.
        J Health Care Poor Underserved. 2018; 29: 1488-1508https://doi.org/10.1353/hpu.2018.0108
        • Klesges RC
        • Debon M
        • Ray JW.
        Are self-reports of smoking rate biased? Evidence from the second National Health and nutrition Examination Survey.
        J Clin Epidemiol. 1995; 48: 1225-1233https://doi.org/10.1016/0895-4356(95)00020-5
        • Czajka JL
        • Beyler A.
        Declining Response Rates in Federal Surveys: Trends and Implications (background paper).
        Published June 15, 2016 (Washington, DCMathematica Policy Research Report) (Accessed September 7, 2022)
        • Volk RJ
        • Mendoza TR
        • Hoover DS
        • Nishi SPE
        • Choi NJ
        • Bevers TB.
        Reliability of self-reported smoking history and its implications for lung cancer screening.
        Prev Med Rep. 2020; 17101037https://doi.org/10.1016/j.pmedr.2019.101037
        • Wang H
        • Heitjan DF.
        Modeling heaping in self-reported cigarette counts.
        Stat Med. 2008; 27: 3789-3804https://doi.org/10.1002/sim.3281