Time-varying effects of a text-based smoking cessation intervention for urban adolescents

Introduction: Craving to smoke is understood as an important mechanism for continued smoking behavior. Identifying how smoking interventions operate on craving with particular populations is critical for advancing intervention science. This study's objective was to investigate the time-varying effect of a text-delivered smoking cessation intervention. Methods: Toward this end, we used ecological momentary assessment (EMA) data collected from a five-day, automated text-messaging smoking cessation randomized clinical trial with 200 urban adolescents. We employed a time-varying effect model (TVEM) to estimate the effects of stress (time-varying covariate) and baseline nicotine dependence level (time-invariant covariate) on craving over six months by treatment condition. The TVEM approach models behavioral change and associations of coefficients expressed dynamically and graphically represented as smooth functions of time. Results: Controlling for gender, age, and current smoking, differences in trajectories of craving between intervention and control conditions were apparent over the course of the study. During months 2 to 3, the association between stress and craving was significantly stronger among the control group, suggesting treatment dampens this association during this time period. The intervention also reduced the salience of baseline dependence among treatment adolescents, with craving being reduced steadily over time, while the control group increased craving over time. Conclusions: These results provide insight into the time-varying nature of treatment effects for adolescents receiving a text-based smoking cessation intervention. The ability to specify when in the course of an intervention the effect is strongest is important in developing targeted and adaptive interventions that can adjust strategically with time.