The Effectiveness of Cognitive-Behavioral Therapy in Helping People on Sick Leave to Return to Work: A Systematic Review and Meta-analysis.

All 34 comparison groups were unblinded, and every included trial scored between 4 and 7 out of 10 on the PEDro quality scale, so none was high quality. Trials measured the same symptoms on different scales and at different time points, which produced very high heterogeneity (93% of the variation in the sick leave result was between studies). Control groups were not comparable: some received usual care, some nothing, one Qigong, one graded activity, so the word control means different things in different rows of the same analysis. Sixteen of the 34 comparison groups combined CBT with other treatments such as graded activity, workplace interventions or motivational interviewing, so the pooled effect cannot be attributed to CBT alone. The average effect on sick leave was 3.649 days, which the authors themselves say needs weighing against the time and cost of delivering CBT. Working ability did not improve even though sick leave fell, which raises the possibility that people returned to work while still symptomatic. The review states that heterogeneity above 50% 'meant a significant effect of CBT'. That is not what this statistic measures, and few of the subgroup findings could be reproduced by another dataset. The subgroup findings rest on between 1 and 11 trials each, so they are exploratory rather than confirmed.

The only funding statement in the text names Griffith University as the funder. No competing interests are declared in the material provided.

Do not read this as evidence that CBT on its own gets people back to work. Sixteen of the 34 treatment groups combined CBT with other treatments such as graded activity, workplace interventions or motivational interviewing, control groups ranged from usual care to Qigong, every included trial was unblinded and scored only 4 to 7 out of 10 for quality, and 93% of the variation in the sick leave result was disagreement between studies rather than a consistent effect. The claims about session length, group delivery and homework come from small groups of trials and should be treated as leads, not conclusions.