Measurement and choice of drinking outcome variables in the COMBINE Study.

OBJECTIVE: The COMBINE Study, a federally funded multisite clinical trial, endeavored to choose drinking and related outcome variables that were scientifically sound and had both convergent validity with previously published studies and face validity for clinical meaning. This article describes these variables and the methods used to collect them.
METHOD: In choosing the primary outcome drinking variables, the mechanisms of action of naltrexone and acamprosate were considered along with previously published results for them and the psychosocial therapies utilized (Project MATCH). In addition, enough previous data were required to abstract meaningful power calculations for sample size estimates. Attention was paid to methodological detail in collection of drinking data, and confirmatory biological variables (carbohydrate deficient transferrin and gamma glutamyl transpeptidase) were incorporated into the study design.
RESULTS: Daily standard drinks were collected by calendar-based methods with a stated goal of 90% within-treatment drinking data to be collected. "Percentage of days abstinent" and "time to first heavy drinking day" were chosen as primary outcome variables. Standardized daily alcohol consumption data can be applied to various statistical approaches, including hierarchical linear modeling and multiple relapse event analyses, which can evaluate a progression of improvement or worsening over time.
CONCLUSIONS: Trained individuals using calendar-based methods attempting to collect all daily drinking data, independent of treatment dropout, should enhance interpretive validity of treatment differences. Convergence of drinking data with biological marker changes, quality of life, craving and health services utilization will enhance the overall validity of both the within-treatment and the posttreatment results for the COMBINE Study.