Multivariable random intercept logistic regression models with psychological and social work factors at T1 as predictors of mental distress above the cut-off for "caseness" at T2^a.

aSeparate regressions were run for each factor and subjects reporting mental distress at baseline were excluded. Age, sex, skill level, and mental distress at T1 as a continuous variable were included in all regressions. bResponse categories were yes/no. **p<.01. ***p<.0005, which was the bonferroni-adjusted threshold of statistical significance.