Table 1 Statistical methods for outcomes and ancillary analyses
Outcome | Hypothesis | Outcome measure | Methods of analysis |
|---|---|---|---|
1) Primary | |||
 a) Drinks per drinking day at 6 months | BMT reduced outcome from baseline to 6 months | Drinks per drinking day during the last 90 days in the Timeline Follow Back [continuous] | T test |
2) Secondary | |||
 a) Alcohol use pattern at 6 months | Back to a low-risk use or abstinence after the treatment | Presence of a low-risk pattern: less than 100 g of ethanol a week and no binge drinking occasions (i.e., more than three SD in women and 4 in men) during the last 90 days in the Timeline Follow Back [binary] | Chi-squared test |
 b) Frequency of heavy drinking at 6 months | Reduction | Number of heavy drinking occasions (i.e., more than three SD in women and more than four in men) during the last 90 days in the Timeline Follow Back [continuous] | T test |
 c) Most extended period of abstinence during the last 3 months | Augmentation | Number of days of abstinence within the last 90 days in the Timeline Follow Back [continuous] | T test |
 d) Severity of dependency at 6 months | Reduction | Score in the Alcohol DAYS, SEV, and WORST SEV score in the Substance Dependence Severity Scale (last 30 days) [continuous] | T test |
 e) Alcohol related negative consequences | Reduction | Total consequences score in the Drinker Inventory of Consequences [continuous] | T test |
3) Subgroup analyses | |||
 a) High v/s low severity | Greater effect in low severity. |  |  |
 b) Motivational level | Higher motivation intensifies the treatment effect. |  |  |
 c) Educational level | Higher education intensifies the treatment effect. |  |  |
 d) Male v/s female | Sex interacts with treatment effect. |  |  |
4) Sensitivity analyses | |||
 a) Per-protocol analysis |  |  | a) T test/chi-squared |
 b) Adjustment for baseline variables |  |  | b) Linear model (multivariate regression) |
 c) Missing data imputation |  |  | c) Multiple imputation (missing-at-random assumption) |