Initiation, cessation and relapse of tobacco smoking over a 3-year period among participants aged ≥15 years in a large longitudinal cohort in rural South Africa

Tobacco smoking is increasing in many low-and-middle-income countries, but data about initiation and cessation patterns are sparse, particularly in rural areas. We investigated changes in smoking status and their determinants in rural South Africa. Participants enrolled in the Vukuzazi population cohort in rural KwaZulu-Natal, South Africa completed a baseline tobacco behavioural survey during 1 May 2018 to 31 March 2020. A follow-up survey was conducted during 4 May 2021 to 18 November 2022 among all participants aged ≥15 years who reported current and former smoking at baseline (to detect cessation and relapse) and in a random selection of participants aged 15–29 years who reported never smoking at baseline (to detect initiation). We fit regression models to estimate smoking initiation (from never to current or former smoking), cessation (from current to former smoking) and relapse (from former to current smoking) between baseline and follow-up, and to investigate the sociodemographic and behavioural variables associated with each outcome. Of those recruited, 52% (754/1448) participated in the follow-up survey, which occurred a median of 3.0 years (IQR: 2.6–3.2) from baseline. Initiation, cessation and relapse occurrence was 12.0% (95% CI: 8.4–16.8), 12.9% (95% CI: 10.0–16.5) and 10.9% (95% CI: 4.4–24.2), respectively. Males had significantly higher odds of initiation than females (adjusted odds ratio [AOR] 12.81, 95% confidence interval [CI]: 3.54–46.36). Moderate/heavy smoking (≥10 products per day; AOR 0.27, 95% CI: 0.08–0.93 relative to light smoking