Main findings
The aim of the study was to evaluate the effect of PE in individuals at risk of having a mental disorder. Participating in the PE sessions had no influence on the chance of full RTW during the first 6 months, but during the first 3 months, participants in the intervention group had a significantly higher risk of not having fully returned to work. The same pattern was seen for the outcome first RTW; however, no significant difference was observed during the first 3 months. The risk of not returning to work during the first 3 months was highest for individuals who had participated in four to six sessions compared to the control group.
The intervention did not decrease the level of symptoms of depression or anxiety or any other of the psychological symptoms. It did not improve mental health related QoL; however, individuals in the intervention group improved their scores on internal LoC at both 3 and 6 months.
Interpretation of outcomes
The significantly higher risk in the intervention group of not returning to work during the first 3 months might be due to an ambition to complete the PE programme before they went back to work. It is plausible since individuals who participated four to six times had an even higher risk of not going back to work compared to the risk of all participants in the intervention group. For all individuals allocated to the intervention group, the chance for first RTW was not significantly lower than in the control group. This may be because they had been able to attend the course while working part time. As part of usual care, individuals from both groups participated in other courses arranged by the job centres, e.g. psychology sessions and mindfulness therapy. It was not examined whether participating in those courses resulted in a higher risk of not returning to work. However, it might not be the participation in PE or other courses in itself that delayed RTW, but the fact that they participated in a research project and, therefore, wanted to finish the intervention even though they were ready to RTW. It has been presumed that participating in an intervention programme for several weeks may obstruct the natural RTW and, hence, introduce a negative effect [
36]. Another explanation for delayed RTW could be that the course made them more aware of their mental health symptoms, and therefore, they felt worse and postponed RTW. However, participants in the intervention group did not score higher on mental health symptoms after the intervention compared to the control group.
If PE or course participation may, in general, result in a higher risk of not returning to work, it is important to be aware of when implementing interventions. Maybe the risk is more pronounced when the intervention is offered close to the start of the sickness absence period. Most workers will return to work rapidly within the first months after reporting sick [
37,
38]. Participating in interventions at an early stage could therefore prolong RTW. In individuals on sick leave due to low back pain, the optimum time window for the start of an effective structured intervention has been suggested to be approximately 8 to 12 weeks after start of the sickness absence [
36]. Our intervention was, on average, provided 7–8 weeks after the start of sickness absence. However, it could be questioned whether the intervention started too early because participants in the control group returned to work significantly earlier than did the intervention group during the first 3 months after the intervention was initiated.
Psychoeducation
This specific type of PE was not effective in facilitating RTW and improving mental health. This could be due to the intervention not being specific and tailored to the participants’ individual needs. PE is usually applied to a group of patients with one specific diagnosis [
17]. In this study, the participants could suffer from sub-clinical as well as clinical depression, anxiety, and somatoform disorder besides feeling distressed. Broad inclusion criteria were applied because we believed that the topics that were taught and discussed in the psychoeducation sessions would be relevant for sick-listed individuals with different mental health problems. Another reason for the broad inclusion criteria was to test an intervention that could be implemented by the social workers in the job centres without asking medical doctors for specific diagnostic information.
Another reason for not finding an effect could be the open groups, which were used in order to offer the intervention as rapidly as possible, as it has been shown to be important for RTW outcome [
11]. This, however, resulted in a lack of continuity in the PE because participants had not all attended the same previous sessions. Furthermore, the participants were not well connected socially since they only took part in a few sessions together. This also limited their opportunity to exchange experiences with other participants.
Another reason for not finding an effect could be that the sessions might have been based too much on lectures and too little on discussions. Thus, the content of the sessions might not have been sufficiently aimed at the participants’ own challenges. It is possible that homework would have helped the participants to work with the topics and make them part of their daily lives. We did not measured how well they used what they had been taught.
Furthermore, the course may have focused too much on mental health and not enough on RTW. The nurses were not accustomed to working with individuals on sick leave or giving advice on RTW issues; however the physiotherapist, the social worker, and the psychologist were. Finally, PE was given in addition to the standard offers to individuals on sick leave in Denmark. Thus about 40 % of the individuals participated in activities offered by the job centres, and about 65 % received treatment for their mental health, mostly from a GP or a psychologist. Moreover, the social workers encouraged the participants to resume to part time work partially, which may facilitate RTW [
39]; however, the effect in individuals with mental disorders is inconsistent [
38,
40].
In the analysis of the effect of the intervention in this study, the content of usual care must be considered. The effectiveness of the intervention, in this case PE, is a relative measure and depended on the effect in the usual care group, which may have been effective in itself.
Strength and limitations
The major strength of this study was the randomised design and the large group of participants. Register data were used to measure RTW, which is preferable compared to self-reported data in regard to receiving more accurate information on the sick leave period [
41].
The social workers were not sufficiently blinded for the allocation and were able to correctly identify two-thirds of the individuals in the control group, which could introduce confounding. It is possible that they could have let participants in the control group return to work earlier than those in the intervention group.
The intervention was offered at an early stage in the sickness absence period. As a result, participants were randomised before they had given written consent. This could introduce possible risk of bias, but it did not seem to have affected the final results.
Thus the participants knew their allocation before they provided written consent; however, this did not seem to influence the relative participation rates because the same number of individuals from each group dropped out of the study after randomisation. The internal validity of the study does not seem to have been threatened because no differences were found between the dropouts in the two groups. Reasons for dropping out of the study and reasons for not attending the PE session as intended were not collected.
Some participants completed the baseline questionnaire after they had started the intervention. Analyses were not adjusted for baseline score as this could introduce information bias. However, scores on symptoms of depression and anxiety (SCL-90-R) at baseline and the score on SCL-8 AD were similar for the two groups. The SCL-8 AD consists of items on symptoms of depression and anxiety and was completed by participants before they knew about their allocation. Moreover, the scores on the remaining baseline questions seemed to be similar between the two groups. However, the significantly higher score on internal LoC in the intervention group at both 3 and 6 months might be explained by the difference that was already present at baseline.
Generalisation
Effectiveness of RCTs depends on the context in which they are conducted. Effectiveness in RCTs in the field of RTW will differ due to heterogeneity in populations, characteristics of the workers and workplaces, and differences in the social system [
42]. The study was performed in individuals on sick leave in a Danish setting, and all participants received the standard care from the job centres and health care system.
The present study was conducted in collaboration with the job centres, because the goal was to assess the effect of the intervention as it would work in a realistic setting. The participants were included based on a simple screening instrument (SCL-8 AD). Thus, considerable variation in reasons for sickness absence, symptoms, and diagnoses was allowed.
One-third of the eligible individuals participated in the study. The study population consisted of more women than men and of individuals who were intermediate to highly educated, on sick leave due to mental health problems, and had low recovery expectations, which is similar to another Danish study [
43]. It is possible that those accepting to participate were more eager to return to work compared to those not accepting to participate. If the last two-thirds had participated, it is likely that the results would have been different from those in the present study.
PE was taught by different health professionals, which circumvents ascribing the effect to have been due to the influence of a single person.