LCA is a statistical tool to extract respondent profiles that are distinguished by probabilities of responses to the set of variables included in the analysis (i.e. the latent class indicators). We used LCA to extract distinct and mutually exclusive respondent profiles from patterns of 13 psychiatric diagnoses (213 = 8,192 possible response patterns) among those with AUD. LCA was estimated with the maximum likelihood robust (MLR) estimator with 200 random sets of starting values and 20 optimizations to ensure the models converged to an appropriate solution (Nylund et al. 2007).
