Long-term follow-up of participants with mental ill health in the Danish return-to-work program: A secondary analysis of a controlled trial.

Districts took part in four different ways and how they were assigned is not fully documented; in three districts the programme and control offices were not assigned at random, so the authors analysed the data as an observational study. Districts where individuals were not randomised may be open to selection bias. Implementation varied between municipalities. Half of the first consultations with a return-to-work coordinator happened after the ninth week, referral rates to the interdisciplinary teams differed widely between municipalities, and the process evaluation found large differences in quality. What participants actually received therefore varied depending on where they lived. In districts where the same staff ran both the programme and ordinary case management, the training given to the programme teams may also have improved ordinary case management, which would hide any real difference between the groups. More money followed each person assigned to the programme, which may have given job centres an incentive to classify people as eligible for the trial. Six comparisons were made across two outcomes and three time points, and the authors set each interval at 99% so that the confidence that all six intervals contain the true value is about 94%. The significant 5-year result is one of those six. The 5-year result held when the three location-separated districts were removed, but in those districts alone it was smaller and not significant, and it was not significant when the analysis adjusted for which jobs and benefits office people attended. Reasons for sickness absence were self-reported and carried no diagnostic codes, and only people in one of the three return-to-work categories were eligible, so the sample is narrower than everyone off sick with mental health problems. This is a secondary analysis of an existing trial. Whether the programme improved the quality or speed of workability assessments was never measured.

The intervention was designed and implemented by the Danish National Research Centre for the Working Environment. The Danish National Prevention Fund paid for implementing it in the participating municipalities, and the Danish Ministry of Employment paid for its design and evaluation. The authors declared no potential conflicts of interest with respect to the research, authorship and publication of the article.

Do not read the 5-year result as proof the programme gets people back to work. It was one of six comparisons, it was smaller and not significant in the three districts where the programme ran from separate offices, and it disappeared when the analysis took account of which job centre people attended. The outcome covering death, sickness absence benefits and disability pension showed no effect at any follow-up.