Onset and Cessation of Smoking: Temporal Dynamics and Racial Difference in Educational Smoking Disparities among Women

Despite extensive research on the impact of smoking on health, there remains an inadequate understanding of the temporal relationship between smoking, education levels, and racial/ethnic groups. This study aims to clarify the influence of smoking initiation and cessation on the changing trends of educational disparities in smoking prevalence among women from diverse racial/ethnic backgrounds across varying birth cohorts. Using a sample of 336,732 women from the U.S. National Health Interview Survey (1997–2018), we employ Cox proportional hazards models to analyze the relative risks of smoking initiation and cessation across birth cohorts by race/ethnicity and education. Our investigation reveals that the education gap in smoking prevalence has experienced the most notable increase across cohorts among White women, followed by Black and Hispanic women. Particularly pronounced is the widening of the education gap in smoking initiation rates among White women in successive cohorts. This is primarily driven by a decline in smoking initiation among college-educated women and a marked surge in smoking initiation among those without a college degree. Conversely, among Black and Hispanic women, the rate of smoking initiation has declined in recent cohorts, with a more substantial decrease among those with a college degree. In contrast, smoking cessation rates have consistently risen across cohorts within each racial group, regardless of their education level. Hence, the expanding education gap in smoking prevalence across birth cohorts predominantly arises from the increasing disparity in smoking initiation rates. Given that smoking initiation typically occurs before college education is completed, this may underscore a growing trend of smokers being selected into cohorts without a college degree. We discuss the implications of this phenomenon within the context of selection based on family socioeconomic status and individual traits.