DSM-5 latent classes of alcohol users in a population-based sample: Results from the Sao Paulo Megacity Mental Health Survey, Brazil

Abstract

Background: We aimed to identify different categorical phenotypes based upon the DSM-V criteria of alcohol use disorders (AUD) among alcohol users who had at least one drink per week in the past year (n = 948). Methods: Data are from the Sao Paulo Megacity Mental Health Survey collected in 2005-2007, as part of the World Mental Health Survey Initiative. A latent class analysis of the 11 DSM-5-AUD criteria was performed using Mplus, taking into account complex survey design features. Weighted logistic regression models were used to examine demographic correlates of the DSM-5-AUD latent classes. Results: The best latent-class model was a three-class model. We found a ""non-symptomatic class"" (69.7%), a ""use in larger amounts class"" (23.2%), defined by high probability (>70%) of the ""use in larger amounts"" criterion only, and a ""high-moderate symptomatic class"" (7.1%), defined by high-moderate probability of all the 11 AUD criteria. Compared to those in the non-symptomatic class, individuals in the ""high-moderate symptomatic class"" were more likely to have been married, have lower educational attainment and to be unemployed or in non-regular/informal employment. Those on the ""use in larger amounts class"" were more likely to have been married or never married. Conclusion: The two symptomatic classes clearly represented the dimensionality of the new proposed AUD criteria, and could be more specifically targeted by different prevention or treatment strategies. DSM-5-AUD has the advantage of shedding light on risky drinkers included in the ""use in larger amounts class"", allowing for preventive interventions, which will reach a large number of individuals.