Classification and course of alcohol problems among adolescents in addictions treatment programs.

BACKGROUND

This study tested whether adolescents in addictions treatment programs could be reliably classified into milder and more severe alcohol problem categories using latent class analysis, and examined the relation of latent class structure to DSM-IV alcohol diagnoses. Transition probabilities among the latent classes at 1-year follow-up and their predictors were used to characterize the short-term course of alcohol problems.

METHODS

Adolescents recruited from addictions treatment were administered a modified section of the Structured Clinical Interview for the DSM (SCID) alcohol use disorders (AUD) during or shortly after addictions treatment and at 1-year follow-up. DSM-IV lifetime conduct disorder diagnosis was assessed at baseline using a modified version of the Schedule for Affective Disorders and Schizophrenia for School-Age Children (K-SADS).

RESULTS

A 3-class solution that represented increasing severity of alcohol problems (i.e., asymptomatic, mild, and severe) provided the best fit to the alcohol symptom data, and was consistent across gender, ethnicity, and presence of conduct disorder. Cross-classification of latent class membership with DSM-IV alcohol diagnosis indicated only a fair level of agreement. There was an overall decrease in alcohol problem severity from baseline to follow-up. Conditional transition probabilities indicated that males and those with conduct disorder, particularly for those in the severe class at baseline, had a lower likelihood of transitioning to the asymptomatic class at 1-year follow-up.

CONCLUSIONS

The latent class structure of alcohol problems in an adolescent clinical sample was best represented by a 3-class severity-based model. Cross-classification of latent class membership with DSM-IV alcohol diagnosis suggests the potential utility of re-organizing diagnostic criteria to better distinguish the relative severity of abuse and dependence diagnoses. Transitions between latent classes were fairly common at 1-year follow-up, and characterized by a general reduction in alcohol symptoms.