Reciprocal Associations Between Cannabis Initiation and Recurrent Binge Eating Onset in US Adults

Background: Cannabis use and binge eating often co-occur; however, the order and timing of their onset are not well described in population-based samples, limiting our understanding of potential risk pathways between the two behaviors. Method: We examined the order and timing of onset of these behaviors among 12,314 adults from the Collaborative Psychiatric Epidemiology Surveys. Cannabis initiation and recurrent binge eating (i.e., binge eating at least twice weekly for 3 months) were dated using ages of onset reported at a single interview. Survey-weighted survival and discrete-time models estimated cumulative probabilities and adjusted odds ratios (ORs) for onset of each behavior after the other. Results: Among 251 adults reporting both behaviors, cannabis initiation came first for 62.3% (95% CI [52.9, 71.7]) and binge eating came first for 33.2% (95% CI [23.8, 42.5]). The median interval between behaviors was 4 to 5 years. Within 10 years of cannabis initiation, 2.6% (95% CI [1.7, 3.6]) had reported binge eating onset; in contrast, within 10 years of binge eating onset, 35.6% (95% CI [24.7, 45.0]) had reported cannabis initiation. Prior cannabis initiation was associated with higher annual odds of binge eating onset (OR = 2.23, 95% CI [1.46, 3.39], p < .001), and prior binge eating was associated with higher annual odds of cannabis initiation (OR = 1.61, 95% CI [1.07, 2.42], p = .024). In secondary analyses, the association of prior cannabis initiation with binge eating onset was significant among women (OR = 2.63, p < .001) but not men (OR = 1.81, p = .113). The sex interaction was not significant after correction (q = .063). Conclusions: Each behavior was associated with later onset of the other, suggesting a two-way association rather than a single direction of influence. Thus, clinicians treating either behavior should consider asking about the other. However, because onset ages were recalled, prospective research is critical to test causal direction.