Table 3 presents a full multivariate model which extends the information in Table 2 by estimating all covariates simultaneously to determine each individual covariate effect on AUD remission after adjusting for all other covariates. For the outcome of initial remission, three predictors were significant at p<.05 (two others were marginally significant at p<.10): Higher typical drinking frequency and alcohol dependence (versus abuse) at T10 were negatively related to initial remission such that drinking less frequently and having alcohol abuse (instead of dependence) at T10 were associated with initial remission (similar to the univariate effects); for these probands, having a higher level of education at T10 predicted initial AUD remission. For the marginal effects (trends), a lower LR and having had formal AUD treatment at T10 tended to be associated with a lower rate of initial AUD remission.
