Evidence-informed decision about (de-)implementing return-to-work coordination – a case study

Christina Tikka (  christina.tikka@ttl. ) University of Amsterdam: Universiteit van Amsterdam https://orcid.org/0000-0003-2078-8715 Jos Verbeek Amsterdam UMC Location AMC Coronel Institute for Work and Health: Amsterdam UMC Locatie AMC Coronel Instituut voor Arbeid en Gezondheid Jan Hoving Amsterdam UMC Location AMC Coronel Institute for Work and Health: Amsterdam UMC Locatie AMC Coronel Instituut voor Arbeid en Gezondheid Regina Kunz University of Basel: Universitat Basel

It has been shown that, especially when studies do not nd bene cial effects of an intervention, practice and setting are perceived different from what is evaluated in studies and results are not translated to practice (10). In the case of Finland, the government considerably invested in the development and implementation of RtW coordination (11,12), while the Cochrane review could not nd a bene cial effect (6), and evidence from Finland was missing in the review.
The disparity between the scienti c evidence and occupational health practice in Finland calls for a more detailed analysis, to aid those making evidence-informed recommendations and decisions regarding (de-)implementing RtW coordination interventions.

Methods
The aim is to assess the available evidence and draw conclusions about (de-)implementing RtW coordination in Finland using the evidence-to-decision framework (9). We analyse a) how comparable the interventions in the Cochrane review (6) are to RtW coordination practice in Finland and b) collect and assess the available evidence on RtW coordination intervention in Finland using the criteria listed in the evidence-to-decision framework (9).
We conducted a systematic literature search in PubMed, webpages of Finnish research and government institutes, and reference lists of included studies (Supplemental Table 5). One researcher (CT) screened title and abstract and included any type of publication describing or evaluating interventions to coordinate RtW in Finland and the Cochrane review.
We invited experts on RtW coordination in Finland to participate in an online survey in March and April 2019 using Survey Monkey. We de ned experts as a) researchers who participated in the development of a training course on RtW coordination in Finland (13) and b) participants of the training course. Both expert groups were invited via email and reminders were sent after one and two weeks. Participants of the training course were invited one day after the training course by the course coordinator. Participants of the research project were invited by CT. In the survey both expert groups were asked about the difference between RtW coordination and usual care for workers on sick leave in Finland (Supplemental Table 7).
We used content analysis methods to extract and summarize the literature and survey data. Prior to dataextraction, we de ned categories (such as content of RtW coordination) and corresponding themes (meetings, workers' needs assessment, RtW plan, implementation management) based on the description of interventions included in the review (Supplemental Table 4). One author (CT) collected data for each category (names, year, setting, participants, content) for RtW coordination and usual care interventions in the review and Finland using Excel.
Data analysis was done independently by two authors (CT and JV). Within each category, authors summarized similarities and differences between RtW coordination in Finland and the review and judged each category as either similar, mostly similar, different, or unclear if data for a comparison was missing.
Judgements were compared using Excel and disagreements resolved via discussion.
We applied all criteria from the EtD framework for health system and public health recommendations (9) to draw conclusions about (de-)implementing RtW coordination in Finland.
We compared the costs of sick leave for all workers on long-term sick-leave in Finland with RtW coordination to those without RtW coordination. We used 254 Euro per day for the costs of sick leave in Finland based on the average lost production, including salaries (14,15) plus personnel costs (16,17). We calculated the average total sick leave days per worker who has already been on 4 weeks sick-leave as 106 days using statistical data provided by the Finish Social Insurance Institution (KELA) (18).
We calculated the costs of the intervention based on the content of RtW coordination in Finland described by our ndings and assumed that a) a basic workers needs' assessment is part of usual care, b) a RtW plan is developed during a RtW coordination meeting and can be directly implemented, c) the RtW coordination meeting is additional to care as usual, d) changes to the plan and evaluation of the implementation are dealt with in follow-up meetings. Therefore, we calculated the costs of RtW coordination as the costs of the meetings without adding the costs for the workers needs assessment or the implementation of the RtW plan. We used an average of 1,5 RtW coordination meetings per worker on sick leave in Finland based on 3years registry data of a Finnish occupational health service provider (19). We estimated the costs of the intervention at 300 Euro, considering salary costs for one hour for the employer, employee, and physician (20).
For the effect of RtW coordination on sick leave we based our calculation on the effect estimate found by the Cochrane review (Analysis 1.3) (6). The meta-analysis found a statistical non-signi cant bene cial effect of RtW coordination after 12 months follow up but showed substantial heterogeneity in studies. In our model we used an effect of RtW coordination that was not excluded by the analysis and assumed a reduction of 5 days of annual sick leave per worker on long term sick leave. We used the assumed effect of RtW coordination to calculate the effect that this would have on the costs of sick leave. We updated the literature search from the Cochrane review (6) until 07 March 2019 and included randomised controlled trials that ful l the inclusion criteria of the review to judge if new studies are available that are likely to change the results of the review.

Results
We included the Cochrane review (6) and six publications that described the content of RtW coordination interventions in Finland. Three publications from Finland were recommendations and part of information and training material (12,21,22). Three publications reported empirical data of the content and process of actual RtW coordination interventions in Finland evaluated between 2014 and 2018 (19,23,24). The review included 14 randomised controlled trials, conducted in six European countries, Canada, and USA (Supplemental Table 6). For our survey, we invited 39 of 42 eligible participants due to missing contact information. Two participants declined participation. We received responses from 24% (10 of 42).

RtW coordination -Cochrane review vs Finland
We judged RtW coordination and usual care interventions in Finland and those evaluated in the Cochrane review (6) to be mostly similar ( Table 1). Coordination of RtW included at least one face-to-face meeting between the physician and the worker, which was often joined by the employer. In these meetings participants discussed the progress of RtW and temporary work accommodations. The workers' needs assessment consisted of an evaluation of the workers' disability and functioning as well as considering factors from the type of work and the workplace. The RtW plan contained goals and multiple actions, such as temporary work accommodations. Goals could be a full or partial RtW or being available for the labour market in other ways. The plan was jointly developed by the health care professionals and the worker, but also other participants could join the development process, such as the employer or the worker's support person. Mostly, the employer or the occupational physician was responsible for implementing the RtW plan and contacting the worker to ensure goals were achieved. The plan could be changed if this were deemed appropriate. In the Cochrane review, interventions always included a workers' needs assessment that focuses on workability and barriers for RtW and an individually tailored RtW plan. According to Finnish recommendations this is very similar to interventions in Finland. However, in practice not all components of the RtW coordination intervention are delivered as intended. Low adherence to the study intervention was described for 2 out of the 14 included trials in the review. Similarly, the survey results showed that in Finland the content of the workers' needs assessment and the RtW plan doesn't always comply with the recommendations and might only include factors that are either related to the individual worker or his workplace (Supplemental Table 7). Table 1 Similarities and differences between RtW coordination interventions in the Cochrane review and in Finland

Categories
Summary Judgement

Names
Mostly use of keywords that suggest coordination, only some studies in the review used keywords that did not suggest coordination (such as case management and consultation)

Setting
Interventions are mostly situated in European welfare states and can start after a long-term sick leave of the worker.
In the Cochrane review all workers were at least 4 weeks on sick leave but almost half of the Finnish workers had less than 4 weeks accumulated sick-leave and may not have been on sick leave at the time of the RtW meeting.

Mostly similar
Year(s) studied Data from different but overlapping timespans 1 , most studies were recent and conducted after the year 2000 Mostly similar Participants No differences: The worker, the employer or a workplace representative, and a physician (most often occupational physician) participate in the intervention. Possibility for other health care providers (such as occupational health nurse or physiotherapist) and other stakeholders (such as occupational safety representative, social worker) to participate in the intervention.

Content
No differences: Interventions include: -at least one face-to-face meeting between worker and coordinator, which is often but not always joined by the employer.
-a workers' needs assessment that includes a focus on employee's work ability.
-a collaboratively developed RtW plan which consists of dates, goals, and actions for RtW -one person responsible for the implementation of the RtW plan (evaluating the progress and making changes to the RtW plan if appropriate).
In practice the RtW coordination intervention might not always be fully implemented as recommended.

Similar
Intervention duration 2 In the review, interventions lasted 3 months until more than 1 year. Information about the duration of the intervention in Finland was missing.
Unclear usual care the worker may receive general advice to return to work and communication between health care providers is possible.

Effect of the intervention
The review found no bene cial effect on four outcomes across all time points, although some of the con dence intervals around the effects did not exclude a clinically relevant bene t (6). For example, there were no statistically signi cant effects after 12 months on time to RtW (low-quality evidence), cumulative sickness absence (low quality evidence), the proportion of participants at work at end of the follow-up (low quality evidence), nor on the proportion of participants who had ever returned to work (moderate quality evidence) (Fig. 1). The included evidence does not provide results on the importance of the outcome for participants. We judged that a small decrease in the duration of sick leave might not be relevant to the individual worker but for the employer, especially of small companies and blue-collar workers.
The update of the search strategy from the review identi ed 2858 references including duplicates. Three studies were eligible for inclusion in the review and included between 98 and 180 participants. One study found a decrease in sick leave by 10 days (25) while two reported no statistically signi cant effects on RtW (26,27). Evaluation of the effects of RtW coordination in Finland is available from one uncontrolled beforeafter study (19) and from our survey of expert opinions. Results from both studies indicated that RtW coordination in Finland increase the number of workers returning to work and decrease the duration of sick leave compared to usual care.

Costs
In Finland there were on average 1.6 meetings per worker receiving RtW coordination (19), which cost on average 480 € per person. The cost analysis shows that the cost for RtW coordination equals the costs of about two days of sick leave (Table 2). A ve days reduction of cumulative sick leave would result in 790 € savings, when coordination of RtW is used. Other criteria To improve the RtW process, the Ministry of Health and Social Affairs in Finland invested in projects that aim to improve the coordination between the many stakeholders involved in the RtW process (11). Further, the Finnish Institute of Occupational Health (FIOH) promotes and provides training courses on how to coordinate the RtW process (12). The intervention is a common intervention by occupational health service providers in Finland (24).

Drawing conclusions based on the EtD framework criteria
Based on the evidence we made a judgement for each of the EtD framework criteria (Table 3).
We judged RtW to be a priority for Finland due to the support of the Ministry and FIOH for RtW coordination.
Joined meetings between the employer, employee and occupational health service providers have become a common RtW coordination intervention in Finland but might not be implemented well.
We found the interventions evaluated in the Cochrane review (6) to be similar to those currently implemented in Finland. The systematic review could not show a desirable or undesirable effect of RtW coordination on sick leave compared to usual care. The quality of the evidence of the effect of RtW coordination is of moderate to very low-quality, due to the imprecision of the results and the risk of bias in primary studies. We judged that the results of the review do not exclude a small bene cial effect of coordination of RtW compared to usual care.
A small reduction in the amount of sick leave days per year may not be relevant to workers from the individual perspective. From the population perspective, however, already a small reduction of 5 days of sick leave would reduce the total costs of sick leave.
We judged that the resource requirements of RtW coordination were little in comparison to the costs of sick leave, but that the evidence on bene ts and harms was not in favour of the intervention. Long-term sick-leave has an important impact on the worker and the society in terms of productivity. RtW is a recognised priority by policy makers in Finland.

Bene ts and harms
RtW coordination compared to usual care does not increase nor decrease the length of sick leave and neither increases the number of workers returning to work.
Certainty of the evidence* Moderate quality evidence for the outcomes: -Cumulative sickness absence in workdays for follow-up of 6 months and more than 12 months -Proportion who had ever returned to work -long-term follow-up: 12 months Low quality evidence for the outcomes: -Time to return to work: for follow-up of 6 months, 12 months and more than 12 months -Cumulative sickness absence in workdays -long-term follow-up: 12 months -Proportion who had ever returned to work -very long-term follow-up: more than 12 months -Proportion at work at end of the follow-up -follow-up: 6 months, 12 months, more than 12 months Very low-quality evidence for the outcomes: -Proportion who had ever returned to work -short-term follow-up: 6 months Outcome importance The included evidence does not provide information on stakeholders' values of a possibly small decrease in the duration of sick leave. Duration of long-term sick leave in Finland lasts on average 106 days. A small decrease by 5 days of sick leave might not be that relevant to the individual worker but for the employer, especially of small companies and blue-collar workers, might be relevant.
Balance between desirable and undesirable effects Coordination of RtW did not have a desirable or undesirable effect on RtW.
Resource use

Resource requirements
The analysis of the costs showed that the saving from the reduction of 5 days of sick leave outweigh 1.6 times the cost of RtW coordination.

Certainty of the evidence
The analysis of the costs was done as a brief calculation that may not include all important items of the costs and bene ts of RtW coordination, such as the costs of implementing the plan.

Criteria Judgement
Impact on health equity Interventions that increase RtW improve the access to the labour market and decrease inequity between healthy and disabled workers. The effect of usual care and RtW coordination on sick leave might be similar.

Acceptability
The intervention is already a common intervention by occupational health service providers in Finland (24). We did not evaluate the attitudes of workers, employers and health service providers towards the intervention.

Feasibility
Our survey showed that the intervention might not be implemented according to the recommendations. We did not evaluate important barriers that would prevent the implementation of RtW coordination in Finland.
* The quality of evidence re ects the extent to which the review authors (6) are con dent that an estimate of the effect is correct. (28)

Discussion
Comparison with other studies According to some Finnish experts in our survey and the ndings from a single before-after study, RtW coordination in Finland is considered effective in increasing the number of workers returning to work and decreasing the duration of sick leave compared to no coordination of care. In contrast to these opinions, the results from the Cochrane review do not show a considerable effect. The review included studies up to 2016 and provided evidence of moderate to very low-quality showing no bene cial effect of RtW coordination on sick leave. New high-quality studies with large sample sizes may change the results. We found three new eligible studies, that could be included in an update of the review. However, based on the study ndings and methods used, studies were unlikely to alter the results of the review or the quality of the evidence base.
Our ndings of no bene cial effect of RtW coordination is consistent with the ndings of multiple other Cochrane reviews that could not nd considerable effects of additional clinical interventions on RtW for workers on sick leave compared to usual care (29-33).

Strength and limitations
Routinely collected data on the quality of RtW coordination is missing and publications on RtW interventions usually describe the ideal intervention. Empirical data on what really happens in practice are scarce. We used data from a systematic literature search including grey literature and interviewed experts in a survey to describe the (intended) content and process of the RtW coordination practice in Finland. We compared in detail the current practice in Finland to what has been evaluated in studies in the Cochrane review. We don't think that additional empirical data about Finnish practice would considerably change our conclusions about the similarities and differences between the interventions.
We combined data from different sources and study designs but data on stakeholders' values, attitudes, barriers to implementation of RtW coordination, and costs is either missing or very limited. Future studies that show little support from stakeholders for the intervention could alter our ndings and support de-implementation strategies of RtW coordination. On the other hand, large support from stakeholders could support better implementation of RtW coordination.
Even so we do not know barriers to RtW coordination in Finland, our survey results show that coordination might not be implemented as recommended. Although our survey included a small number of participants and results from bigger studies could alter our ndings, it is questionable whether better implementation would achieve a larger decline in sick leave days. Other interventions might be better suited alternatives, such as changes in sickness certi cation policies or incentives for employers to improve RtW rates of their workers (34). We don't think that additional data about implementation barriers to RtW coordination would considerably change our conclusions.
Our cost analysis indicated cost bene ts from RtW coordination if two days of sick leave were averted. A possible small decrease of sick leave might result in meaningful economic consequences important for employers and society.

Conclusion
Our study provides an example on how to make transparent evidence-informed decisions that consider the wider social and political environment. RtW coordination practice in Finland and the interventions in the Cochrane review are similar and the review ndings apply to Finland. Considering all EtD framework criteria, including costs and certainty of the evidence, investment in de-implementation strategies or better implementation of RtW coordination interventions in Finland is currently not required.
New studies evaluating the effect and the costs of the intervention based on better quality data would help improve the evidence base. Both would empower decision makers to implement interventions that are clinically and economically worthwhile. We recommend that changes in RtW practices should be implemented as part of a controlled evaluation study, including detailed descriptions of the content of interventions and usual care. New studies need to be su ciently powerful to detect small but clinically relevant effect sizes, such as two days of reduced sick leave. Given the popularity of RtW coordination, a randomized controlled trial of RtW coordination in Finland would be di cult to realize.
Decision makers can use the EtD framework and its' criteria as a tool to make transparent evidence-based decisions in occupational health and safety. We advise to call for a comprehensive cost-bene t analysis, an assessment of stakeholders' values, and better-quality evidence on the effectiveness of coordination on time to RtW for Finland.

Availability of data and materials
The datasets supporting the conclusions of this article are included within the article and its additional les.

Supplemental Tables
Supplemental Table 4 Categories and themes for data-extraction and analysis of RtW coordination What is the difference between the process of occupational health negotiations most commonly conducted in Finland during the last three years (below referred to as usual practice) and the process now described on the e-learning platform "Työterveysneuvottelu-ratkaisuja-työhön" (below referred to as best practice)?
Best practice does include an assessment of the worker's needs, usual negotiation practice does not.

2 50%
Best practice does lead to an individually tailored return-to-work plan directed at the worker, the workplace, and the employer that includes more than one possible action (e.g. treatment and work accommodation), usual negotiation practice does not.

2 60%
With best practice the affected worker has at least one joint face-toface meeting with the occupational physician and the employer/supervisor (and maybe others such as occupational health nurse), in usual negotiation practice the worker has no joint face-to-face meetings with the occupational physician and the employer/supervisor.

20%
There is no difference between best practice and usual practice for occupational health negotiations in Finland. What is/are the main difference/s between occupational health negotiations and other return-to-work interventions (usual care) in Finland?
Compared to occupational health negotiations, in most cases, other interventions only focus on treatment and health or work capacity evaluations and an assessment of possible work accommodations is missing.

3 50%
In most cases, with other interventions the worker receives no guidance on how to return-to-work, but with occupational health negotiations he does.

2 50%
In most cases, with other interventions the advice on how to returnto-work is only general but with occupational health negotiations 4 1 50% the worker receives an individually tailored return-to-work plan directed at the worker, the workplace, and the employer.
In most cases, other interventions do not include joined face-to-face meeting(s) with occupational health services and workplace representative, but occupational health negotiations do.
All interventions are similar effective in the length of sick leave and the number of workers returning to work Occupational health negotiations are more effective than other interventions (worker take shorter sick-leaves and/or more worker return-to-work) Occupational health negotiations that ful l all criteria advocated on the e-learning platform are more effective than occupational health negotiations that don't ful l all criteria (worker take shorter sickleaves and/or more people return-to-work) Other outcome or statement (please specify) (comments) 2 0 20% Comment 1: depending on the circumstances (original: "valitaan tilanteen mukaan") Comment 2: Effectiveness of the interventions need to be assessed in their own contexts (based on the case in question). It is impossible to compare effectiveness of the different interventions directly. The model provided for ideal intervention on the e-learning platform does not result to shorter sick-leaves and/or more people return-to-work but suggest a smoothier negotiation process in the occupational health negotiation and joint decision making concerning RtW.
Total number of respondents: 6 4 a researchers b training course participants Supplemental