Dynamic Associations between Negative Affect and Craving within- and across-Day

Introduction: The annual US cost of opioid abuse now exceeds $78 billion. Treatment of opioid abuse, a typically relapsing disorder, may especially benefit from greater knowledge of intra-personal states that may precede relapse. Primary among such intra-personal states are negative affect and craving for illicit substances, with the former being recognized as a proximate influence on relapse risk. To investigate how these two intra-personal states are linked to each other, this study explores the within- and across-day associations of negative affect and craving among opioid addicts in residential treatment.

Methods: Ecological momentary assessment data (n = 73) were collected from participants 4x daily for 12 days. To provide a more nuanced view of the relationship between negative affect and craving we employed time-varying effect models (TVEM). TVEM provides advantages over multilevel models for these data by smoothly capturing the dynamic relationship between variables both within- and across-days.

Results: 2002 observations with full information were analyzed. These were drawn from 4x daily measurements for 73 individuals across 12 days. Within-day negative affect predicted increased craving at all points. This effect grew until afternoon, (βmax = .033, 95% CIs [.022, .043]) then decreased until evening. Across-day negative affect interacted more dynamically with craving. Across the first week the relationship between negative affect and craving weakened. At day 8, this trend reversed and negative affect increasingly predicted craving through day 12 (βmax = 0.036, 95% CIs [.022, .050]). The within-day effect of age on craving was slightly significant (βmax = .026, 95% CIs [.013, .040]), with greater influence during early and late day. The across-day effect of age on craving exhibited minimal and diminishing significance (βmax = .025, 95% CIs [.007, .042]). Males experienced greater within-day craving (βmax = 0.60, 95% CIs [.22, .99]), peaking around midday. Across-day effects of gender were significant between days 1 – 5 (βmax = 0.58, 95% CIs [.12, 1.03]) and 9 – 11.5 (βmax = 0.48, 95% CIs [.004, .96]).

Conclusions: To our knowledge, this is the first study to note a latent rebound response in the association of negative affect and craving. This may be due to participant anticipation of stressors upon discharge from treatment. Future studies should examine differences in this association between individuals discharging to natural environments vs. extended treatment placements. Results demonstrate TVEM can be used to obtain unique insights into time-varying processes underlying critical periods.