Ecological momentary assessment and mediation analyses reveal relationships between affect, craving, and substance use in daily life

Background and AimsCraving, the desire to use substances, is a key factor and diagnostic criterion for substance use disorders (SUDs) that predicts substance use. Experiencing affective states, both positive (PA) and negative (NA), are also implicated in SUDs. Yet, the interrelationships among affect, craving, and substance use remain unclear, particularly at the within-person level of analysis. This study aimed to use ecological momentary assessment (EMA) to test whether craving mediates the association between PA/NA and subsequent substance use. DesignObservational study using EMA over a 14-day period and multilevel mediation models with bootstrapping. SettingOutpatient addiction treatment centers in which participants were enrolled during the data collection period. ParticipantsAdults with SUDs enrolled in a pilot randomized controlled trial, evaluating an app-delivered mindfulness-based intervention for SUDs, that completed the baseline EMA period and were also randomized to a treatment condition (N=36). MeasurementsParticipants completed EMA surveys three times daily [midday, afternoon, evening] for 14 days, reporting on current PA, NA, craving, and substance use. We used multilevel mediation modeling to test within-person associations and examine whether craving mediated the link between affect (PA and NA) and subsequent substance use. FindingsMediation analysis showed an indirect pathway via craving on the link between affect and substance use. Higher PA was associated with lower craving at the same EMA survey, which in turn led to reduced likelihood of substance use at the next survey ({beta}=-0.04, p=0.02). Conversely, higher NA was associated with higher craving at the same survey, leading to greater likelihood of substance use at the next survey ({beta}=0.08, p=0.03). ConclusionAt the within-person level, craving mediates the association between affective states and subsequent substance use in individuals with SUD. PA predicted lower craving, which in turn predicted lower likelihood of substance use, while NA predicted higher craving, which in turn predicted greater likelihood of substance use. Our findings suggest that a promising strategy may be targeting momentary affect in daily life, such as through ecological momentary interventions.