Aim: To compare predictors of smoking initiation in two longitudinal studies in California conducted during periods when adolescent smoking prevalence was increasing (1993-96) and decreasing (1996-99).
Design, setting and participants: Cohorts of 12-15-year-old never smokers were identified from the cross-sectional 1993 and 1996 California Tobacco Surveys (large population-based telephone surveys) and followed-up 3 years later (1993-96, n = 1764; 1996-99, n = 2119).
Measures: We compared cohort transition rates to any smoking by follow-up in risk groups defined by known predictors of smoking initiation at baseline. Besides examining predictors individually, risk groups were defined using a multivariate analysis.
Findings: Overall, transition to any smoking by follow-up occurred in 38.3 +/- 4.0% (% +/- 95% confidence interval) of never smokers in the 1993-96 cohort and 31.1 +/- 2.6% in the 1996-99 cohort. For most predictors, the transition rate for adolescents with the characteristic was the same or only slightly lower in the 1996-99 cohort compared to the 1993-96 cohort, but the transition rate in those without the characteristic was generally much lower, thus increasing the power of the predictor. The multivariate analysis confirmed that compared to the 1993-96 cohort, transition occurred much less often in the 1996-99 cohort for adolescents at low rather than at medium or high risk of future smoking.
Conclusions: The turnaround in California adolescent smoking in the mid-1990s, when smoking began to decline, appears to come primarily from adolescents already at low risk of future smoking (as defined by a variety of predictors), who transitioned to smoking at much lower rates than previously.