Modeling complexity of EMA data: Time-varying lagged effects of negative affect on smoking urges for subgroups of nicotine addiction

Introduction: Ecological momentary assessments (EMA) are increasingly used in studies of smoking behavior. Through EMA, examination of lagged relationships is particularly useful for establishing a temporal order of events and for identifying types and timing of risk factors. The time-varying effect model (TVEM) handles EMA data challenges and addresses unique questions about the time-varying effects. Methods: Generalized TVEM was applied to EMA data from a smoking cessation study to investigate a "time-varying lagged" effect of negative affect on high smoking urges. Participants included 224 smokers with a smoking history of 23.1 years (SD = 9.8) smoking 27.3 cigarettes per day (SD = 10.7), which provided 11,394 EMAs following a quit attempt and prior to a smoking lapse. Results: The effect of negative affect was found to vary as a function of a time lag, with stronger immediate effects: estimated odds ratio (OR) of 2.7 for the lower nicotine-dependence group (time to first morning cigarette > 5 min, 57.6%) and OR of 2.4 for the higher nicotine-dependence group (≤ 5 min). The magnitude of the effect persisted up to 7 hr while decreasing over time. Conclusions: This analysis confirmed the importance of negative affect as a precursor of smoking urges while showing that the magnitude of the effect varies over time. An assumption of a constant lagged effect may bias estimates of the relationships and fail to provide a comprehensive outlook of the relational dynamics.