The outcome of interest was relapse during the COVID-19 pandemic, a binary variable (any vs. none). Participants were asked if they had been abstinent but drank alcohol again since January 2020 or if they had previously controlled their drinking but increased their alcohol consumption since January 2020. An endorsement of either option was coded as a relapse event; however, as the study’s eligibility criteria excluded current heavy drinkers, we refer to these events as mild relapse (i.e., resolved at the time of data collection). We identified a set of candidate predictor variables a priori to examine in multivariate models, which fell into the following groups: lifetime AUD problem characteristics; recovery threats; recovery supports; and demographic characteristics.
