Background:
Previous studies in the U.S. show higher smoking rates among those with lower levels of education attainment. Less is known about how these smoking patterns vary by birth cohort or how they may be driven by different demographic profiles across education groups. Furthermore, limited attention has been given to differences in smoking behaviors between those with less than a high school degree and those with 8th grade education or less.

Methods:
Data from the National Health Interview Survey 1966-2015 were utilized to obtain smoking-related information for U.S. adults aged 25 years or older. We developed age-period-cohort models with constrained natural splines to estimate smoking prevalence in groups categorized with five education levels: ≤8th grade, 9-11th grade, high school graduate or GED, some college, and at least a college degree. Annual probabilities of smoking initiation, cessation and intensity by age, birth cohort (1890-1990), gender, and education level were also estimated by the models. Additional regression analyses were conducted to identify sociodemographic factors that may explain smoking disparities across education subgroups.

Results:
The probability of smoking initiation was highest among individuals with 9-11^th grade education and lowest among those with a college degree or more. The initiation probability among those with ≤8^th grade education decreased by birth cohort, resulting in this group having the second lowest smoking prevalence after those with a college degree or more in more recent birth cohorts. The smoking cessation probability was highest among those with a college degree or more. Additional analyses suggest that the low smoking rates among those with ≤8^th grade education may be driven by the increasing proportion of non-US born Hispanics in this group.

Conclusions:
This study identifies population characteristics that may be driving smoking disparities between levels of educational attainment, providing detailed insights into change in smoking patterns by education for different U.S. birth cohorts.