Comparison of the intervention and control groups at the 4-month follow-up.

The means and their 95% confidence intervals were estimated from maximum likelihood mixed effects models adjusting for stratification variables, age and gender. The effect size was calculated from the difference in the means at 4 months divided by their pooled SD. P-values compare the two groups and are from 1) a mixed model (using all randomized individuals, including those with missing follow-up data) which adjusted for stratification variables, age, and gender, 2) a t-test, and 3) a permutation test done within the 4 strata defined by baseline BDI and absenteeism. Permutation test is based on Monte-Carlo simulation with 1,000,000 samples (used to get exact distribution of t-statistic). †: Because this variable did not have the baseline measurement, the means at month 4 and their difference were examined by regression models adjusting for stratification variables, age and gender. BDI-II: Beck Depression Inventory-II. HPQ: World Health Organization Health and Work Performance Questionnaire.