Purpose: Employer involvement in primary health care rehabilitation has been suggested as a promising avenue to curb increasing sick-leave rates due to common mental disorders. The aim of this study is to examine organizational prerequisites for employer involvement in the primary mental health care setting. Methods: In 2017, an intervention was launched at eleven primary health care units to assist rehabilitation coordinators and general practitioners with involving employers early on in the return-to-work process for patients with stress-related disorders. Eleven pre-intervention and ten post-intervention interviews with general practitioners, managers and rehabilitation coordinators were thematically analysed using the Consolidated Framework for Implementation Research (CFIR). The centres that succeeded with implementation were compared to the centres that failed implementation. Results: Employer contact was more easily established at centres where general practitioners were genuinely dedicated to issues related to insurance medicine, and where the coordinator had a clear and distinctive role in relation to other staff members. At the successful centres, the cooperation between general practitioners and rehabilitation coordinators was structured and the intervention’s core component – that is, employer involvement – could be added to pre-existing workflows. Conclusion: Primary care stakeholders aiming to involve employers should establish solid routines and structures for insurance medicine with clear roles for the involved professional groups and should stabilize the staff situation before this step is launched.

• Primary healthcare • Mental health rehabilitation • Return-to-work • Job stress • Evaluation

Björk L, Glise K, Pousette A, Holmgren K, Bertilsson M (2021) Implementing Employer Involvement for Patients with Stress-Related Disorders in Swedish Primary Care – A Process Evaluation. J Family Med Community Health 8(1): 1183.

RC: Rehabilitation Coordinator; GP: General Practitioner; RTW: Return to Work; CFIR: Consolidated Framework for Implementation Research

Health care and social insurance systems across Europe are struggling to combat high levels of sickness absence due to burnout [1-3]. In Sweden, sick leaves due to adjustment disorders and reactions to severe stress – a spectrum of diagnoses that includes burnout – are increasing and the average sick-leave due to burnout lasts about six months [4]. These disorders justify compensation from the social security system1 and the vast majority of sick-leave notes are dispensed by general In Sweden, the first fourteen days of a sick-leave period are covered by the employer (except for the first qualifying day); thereafter, the Social Insurance Agency takes over the financial responsibility. practitioners (GPs) within the primary healthcare system [5,6]. Assessing a patient’s work capacity in relation to functionality and diagnosis can be a difficult task for a GP, particularly when it comes to psychiatric diagnoses [7-9]. In addition to a number of important factors at the individual level, such as symptom severity and expectation to return to work [10], workplace-level factors such as good social support from supervisors and colleagues are important determinants for a successful return-to-work (RTW) process in patients with common mental disorders [11,12]. Meta-analyses of effective interventions have shown that employer involvement is a crucial component in increasing patients’ RTW, particularly in patients with stress-related disorders [13,14]. Some attempts at employer involvement in stress rehabilitation have been studied in Sweden, and have shown both beneficial effects [15-17] and no effects [18]. These attempts indicate that more outcome research is needed to evaluate workfocused interventions in a primary care setting, and that further qualitative research could assist primary health care givers in configuring and providing work-focused services. In a scoping review of non-pharmacological interventions to support patients with mental disorders in obtaining, keeping or returning to work, Reed and Kalaga [19] identify a number of obstacles that must be overcome in order for work to be a focus in a primary care setting. Systemic barriers include poor coordination and communication between GPs and occupational GPs, a lack of access to rehabilitation services, and time constraints. Another barrier mentioned in the review relates to the difficulties GPs face when trying to determine the effects of being away from – or being engaged in – work on a patient’s symptoms, well-being and recovery. These findings indicate that further qualitative research could assist primary health care givers in configuring and providing work-focused services.

Setting and Aim Swedish health care is managed at a regional level. Rehabilitation coordinators (RCs) – have been introduced in all regions. Their role is to provide individual support to sick-listed patients, and to coordinate health care services with other public services and employers. RCs are concentrated in primary care but are also present in inpatient care settings where sickness certification is common. To evaluate whether a systematic procedure involving the patient’s employer can reduce the time for RTW in patients with stress-related disorders, the PRIMA trial was conducted in the Västra Götaland region in 2017 (see [20] for details). Public (n=15) and private (n=7) primary health care centres were randomly distributed to intervention and control groups. The target group for the trial comprised the GPs and RCs at the eleven intervention centres (Table 1),

and by extension their patients, who had been sick-listed from work due to stress-related disorders [?]. Through lectures and workshops with experienced clinicians and researchers, the GPs and RCs in the intervention group were trained in understanding both why and how the employer can be involved at an early stage in the rehabilitation of patients with stress-related disorders. Back at their practices, the participants then followed a standardized work procedure that involved an examination of stressors in the patient’s work and private life situation, as well as early contact with the patient’s employer. The prime hypothesis of PRIMA is that early contact between the caregiver, patient and employer will stimulate the employer to take measures at the workplace and motivate the patient to advance his or her work reintegration. Including the patient’s workplace and employer in rehabilitation is not a typical practice in primary health care, and it is likely that several contextual factors affected the centres’ degree of success in this endeavour. Such factors are the focus of the present study. Through a systematic and theoretically guided comparison between the primary health care centres that succeeded in implementing employer contact and the centres that failed to do so, the aim of the present study is to gain knowledge about the prerequisites for employer involvement in a primary mental health care setting. By ‘implementation’, we refer to the process of putting the intervention to use in a given organization. Theoretical framework Damschroder and colleagues [21] have presented the Consolidated Framework for Implementation Research (CFIR), which is an overarching taxonomy of the various factors that have been found in earlier research to be essential for the effective implementation of interventions in health care services. The framework consists of five domains (in italics below), each of which contains several interacting constructs that researchers can select from when designing a strategy for implementation,and when evaluating its progress and effectiveness. The first domain relates to the core and peripheral intervention characteristics. While the core components of an intervention are indispensable for all participants, the peripheral components can be adapted to fit with local conditions. The core component of the PRIMA intervention was its standardized work procedure that included early employer contact. The second and third domains in the CFIR framework are the outer setting and inner setting of an intervention. While the outer setting includes ‘the economic, political, and social context within which an organization resides’ (30, p.5), the inner setting includes ‘features of structural, political, and cultural contexts through which the implementation process will proceed’ (ibid). It is primarily the centres’ inner settings that are in focus within the present study. The fourth domain relates to the characteristics of the individuals involved in the intervention and/or its implementation process. The fifth and final domain involves the implementation process. In PRIMA, several measures were taken by the research team to facilitate an active change process. Each of the five domains includes several specific constructs. For example, the first domain, intervention characteristics, includes the constructs of adaptability (i.e. the degree to which an intervention can be adapted to meet local needs) and complexity (i.e. the perceived difficulty of implementation). CFIR terminology will be used to discuss the results of the present study.

Categorization of successful and unsuccessful implementation As part of the standardized work procedure, a signed informed consent to participate form and a questionnaire used by the RC to interview the patient’s employer were collected for each included patient. At the end of the 14-month implementation period, these forms were collected from the intervention centres. GPs and RCs at five centres had a successful implementation as they did well in (1) recruiting patients and (2) contacting the patients’ employers, while six centres failed implementation (see the last two columns in Table 1).

Data collection Before the intervention, group interviews were conducted with representatives from the eleven intervention centres. All centres were requested to invite one GP, the RC and the centre manager to the interview. At some centres, the manager also practiced as a GP or as a RC and thus represented both functions in the interview (Table 2).

The interviews took place at the centres during ordinary working hours and were sometimes disturbed by the ‘everyday life’ at the centre. For example, one GP had to interrupt the interview to see to an urgent patient matter, and one participant manager arrived late due to a staffing errand. Thus, it was necessary for the interviewer to take a pragmatic approach to the interview situation and to be flexible in relation to daily events. All but two interviews were conducted by the same interviewer for coherence. The interviews were semistructured and included questions about how often patients with stress-related disorders were treated at the centre, and about the therapies and procedures used to treat and rehabilitate these patients back to work. Examples of questions are: ‘What kind of treatment and other services are you able to offer patients with stress-related disorders?’ and ‘Please describe how the RC and other staff members collaborate at this centre’ [Table 2]. After the intervention, a follow-up interview was conducted at each centre. One centre declined participation due to a high workload. The post-intervention interview took place about one year after the first interview, and the centre manager, the RC and a GP who had been involved during implementation were invited. However, their participation could not always be fulfilled due to staff turnover, parental leaves and retirements (Table 2). The post-intervention interviews included questions about how the intervention had been received the staff members; and about factors that had facilitated or hindered implementation. All pre- and post-intervention interviews were audio-recorded and transcribed verbatim.

Analysis Transcripts were transferred to the NVivo 11 Software and coded. The codes were derived from the research team’s ideas about how the intervention could be assumed to affect the participating GPs and RCs and, by extension, the included patients and their employers. For example, we theorized that the standardized work procedure would only be implemented if the RC had enough time and commitment to follow the study protocol (generating the codes RC workload and RC commitment),and that the intervention would only work if the employers were willing to share information with the caregiver (generating the code employer engagement). Other codes emerged from the data. These codes mainly concerned hindrances that were related to the intervention itself – such as issues related to the criteria of patient inclusion – and that had not been anticipated beforehand by the research team. The codes were arranged under the following four themes: intervention procedure, intervention content, intervention-environment fit and staff situation and commitment. These themes are presented in the results section and will be related to the broader CFIR domains in the discussion. By using the framework matrix tool in NVivo, the coded data was organized in matrices. Ten matrices were created: two per theme – one matrix for the successful centres, and one for the unsuccessful centres – with cases (i.e. primary health care centres) in rows, codes in columns and the coded data in cells. In the next step, the information in each cell was condensed, and illustrative quotations were identified. Finally, each pair of matrices (successful centres vs. unsuccessful centres) was used for qualitative content analysis, per theme. The analysis was performed by the first author, and the second and last authors carried out independent readings of the matrices.

Intervention procedure The study protocol included criteria for patient inclusion. One criterion excluded patients with past sick-leave periods, and it caused GPs and RCs to hastily dismiss patients that were not entirely new to them, rather than having to go back into the patient’s record and check for previous periods of sickleave. Other sources of confusion had to do with the diagnosis. Some participants thought that they were only supposed to include patients with burnout. Overall, the participants in the unsuccessful group thought that the intervention was narrowly targeted, and that misconception obstructed the implementation: […] one should have broader inclusion criteria from the beginning so that the patients are admitted immediately. One should not have to search for them. (RC, Centre II, postintervention interview) There were no reports from the participants in either group that indicated dissatisfaction with the support they received from the research team during the implementation.

Intervention content As a first step after inclusion, the coordinator had the patient fill out a questionnaire about, for example, work and private life stressors and health. The patient’s answers were used as a basis for further dialogue. The RCs – and even some GPs in the successful group – reported that the questionnaire was helpful in obtaining a deepened understanding of the patient’s situation. However, some coordinators noticed that the patients were sometimes exhausted after having answered the questionnaire. After consulting with the GP, the next step for the RC was to contact the employer by phone, following a second short questionnaire. This step was the core component of the intervention. The coordinators reported that it was often difficultto reach the employer, and that this procedure was the most time-consuming component of the intervention. However, all of the RCs and many of the GPs in the successful group viewed this step as a very positive experience: I feel much more comfortable contacting the employer, and I have done so much more after the project [...] It does not feel like such a great thing to contact employers, I see that it is great. So that has been the biggest gain I think, to understand the importance of getting employers on board early in the process. (RC, Centre C, post-intervention interview) Some of the patients were even able to return to work directly after this contact. According to the standardized work procedure, the next step was for the RC, the GP and the patient to agree on a time to meet the employer and create a plan for the patient’s RTW. The project provided a short guideline for the meeting that included suggestions regarding work accommodations. The GP’s often full-time schedule presented an obstacle to this step, as it made it difficult to find an occasion when everyone could meet. One RC in the successful group was particularly frustrated over this last step, and thought of it as something that delayed rather than facilitated the patient’s RTW. Since the RCs and GPs in the unsuccessful group failed to follow the protocol in terms of patient recruitment, their views on the standardized work procedure could not be assessed. Intervention-environment fit There was a clear difference between the successful and unsuccessful centres in terms of the organizational structures and routines that were in place for working with sickness certification. At the successful centres, the coordinator had a clear role in this process; furthermore, formalized routines of cooperation between GPs and RCs, as well as initiatives to include employers, were already in place before the intervention: My function is to coordinate the care process for the individual [...] so that there will not be waiting times […] And also, of course, to try to involve employers as early as possible and with the right timing. (RC, Centre C, pre-intervention interview) Managers and RCs at the successful centres stated that they joined the project to enhance these structures and routines. However, each centre in this group had its own routines and there was no standard workflow identified across centres. At one of the centres, the coordinator and the centre manager – who was also a senior GP – held regular meetings at which they continuously evaluated and developed a specific routine for sickness certification that every GP at the centre was required to follow. As part of this routine, all new GPs who started to work at the centre received information on how to use the coordinator’s services. At the successful centres, the RC’s work tasks often included early employer contact. In both groups, the RCs reported that the inclination to use the coordinator’s service varied among GPs: some GPs were reluctant to involve the coordinator, while others routinely consulted with the RC. At the successful centres, many GPs not only explicitly appreciated the coordinators’ support, but also regarded the function as essential for modern primary health care:I would say the rehab coordinator is absolutely necessary for the mission we have now [...] If it were only up to us GPs ... then, then it would not work! The employer contact… (GP, Centre A, post-intervention interview) At centres with well-established routines for GP-RC collaboration, and with experience in involving the employer early on in rehabilitation, the standardized work procedure enhanced already existing routines: I think it has linked in quite well with what we had before. [...] The rehab coordinator [...] is an appreciated resource for the GPs [...] so that [the GP] can concentrate on the patient meeting and on medical issues. (Manager/GP, Centre E, post-intervention interview) In contrast, participants from centres in the unsuccessful group claimed that they joined the project in order to establish structures and routines for insurance medicine. These centres were characterized by a lack of established routines for organizing GP-RC cooperation: A nurse who starts working walks straight into a structure: ‘This is your schedule, you’ll be answering the phone at this time, you will do this, and you will do that!’ But when a coordinator walks into a health care centre, it’s just: ‘Well?!’ (RC, Centre I, post-intervention interview) The same informant described how this lack of structure affected the RC’s role: I am quite concerned about how it looks when it comes to routines, structures, staffing problems… so there are many variables that do not work. And it makes it…it does not matter how damn good I am, to speak plainly. How big a difference do I make? (RC, Centre I, post-intervention interview) It was also uncommon for the RC in this group to have contact with employers, and if they did have such contact, it was late in the rehabilitation process. Many of the RCs struggled to find their role in relation to the GPs and to get involved early on in the rehabilitation process: I can’t go to a GP and say: ‘Have you done a-b-c? […] Have you asked what they [the patients] do at work? Why, have you asked why the patient has anxiety?’ [...] I get no structures from the top to keep to […] I have people that I cannot tell what to do because everything is ‘right’. (RC, Centre I, post-intervention interview) Staff situation and commitment During the 14 months of implementation, centres in both groups went through periods of significant changes: senior staff members retired, centres expanded or changed facilities, and many managers struggled to recruit and retain specialist GPs. Participants from most centres reported a high general workload as an important hindering factor in implementation. What clearly distinguished the centres with successful implementation, despite these conditions, was the genuine commitment to the project that was shown by a few or several GPs, at least initially. Among the successful centres, two centres (A and C) had a single enthusiastic GP who recruited patients and monitored them according to the standardized work procedure in close cooperation with the RC; two centres (B and E) had several GPs who initially recruited afew patients each, but whose commitment faded over time; and one centre (D) had GPs who were all committed from beginning to end: That [the GPs’ engagement] is certainly a contributing factor to the fact that we have found so many patients. For everyone has been on track! (RC, Centre D, post-intervention interview) The RCs in the successful group received the opportunity to test the standardized work procedure, including employer contact, together with their GP colleagues; this experience strengthened them in forming future employer contacts. In this group, the RCs became advocates for the intervention and were able to recruit patients and make employer contact even when individual GPs were temporarily absent or had left the centre. In three out of five centres in this group, the coordinators were also sometimes replaced or temporarily absent. Yet these centres managed to maintain a certain level of continuity in implementation. A strained work situation and high staff turnover particularly characterized all centres at which the implementation failed. At these centres, the coordinators had irregular work schedules, and often had to leave their RC duties to cover for absent colleagues. The intervention was never prioritized: It was perhaps unfortunate that we participated [in the project], because it was a high, a giant, load on the health care staff. One might not have had the space to pass on information when new staff arrived, nor to implement the working method. (RC, Centre III, post-intervention interview) Also, the GPs’ commitment to the implementation was weak at the unsuccessful centres. The RCs of this group never received the opportunity to test the core features of the intervention in cooperation with their GP colleagues, and never became advocates for implementation.

Employers are key stakeholders in the question of when patients with stress-related disorders will return to work. Research on effective approaches for employer involvement in a primary care setting is sparse and inconclusive, and the field is open for new models and methods. The PRIMA intervention supports the RC - a new function in Swedish primary care - in order to involve the employer early on in the rehabilitation process. By comparing the implementation process at five centres that succeeded with the implementation with the process at six unsuccessful centres, this study examines the general prerequisites for primary mental health care to involve employers in the work reintegration process. Considering how contextual phenomena interact to both facilitate and hinder implementation is important in understanding how – and at best why – interventions work [22- 24]. In the following section, the results are organized and discussed in relation to the domains and relevant constructs of the CFIR framework [21] in order to enhance comparability to studies with other populations and settings. Intervention characteristics The results provide some answers on what the large-scaleimplementation of employer involvement should look like in a primary care setting. For example, the results indicate that the targeted population needs to be broadly defined. The strict inclusion criteria in PRIMA caused uncertainty among many staff members, and it is plausible that looser criteria would have made it possible for the participant GPs and RCs to include more patients. It is also likely that if GP attendance at the meeting between the caregiver, patient and employer would have formulated as a recommendation, rather than as a requirement, this would have lowered the threshold for the RC to follow through with the standardized procedure. In CFIR terminology, the perceived difficulty of implementation – that is, the intervention’s complexity – prevented the participants from getting started with, and following through with, the implementation. It is clearly difficult to balance the need for consistent implementation across sites with the need for a design that is adaptive to local conditions, and to meet the participants’ need for simplicity (e.g. simple tools and a broad scope) while maintaining the required accuracy of scientific rigor (e.g. measurements with satisfactory psychometric properties and a well-defined population). Keeping the intervention ‘simple’ is particularly challenging when patients with stressrelated disorders are the target, as recent findings indicate that interventions with several components are more efficient for RTW than ‘simple’ interventions [13-14]. The substantial efforts that were made by the research team to adapt the intervention according to the input from experienced primary care clinicians in the design phase were insufficient. Difficulties have also been reported elsewhere in implementing RTW interventions in real care settings – despite an elaborate design and enthusiastic advocates for change [e.g. 25]. Outer setting Theoretically, it could be assumed to be more difficult to establish employer contact in districts with many unemployed or self-employed patients. However, successful centres were present in both rural and urban areas, and in both rich and poor neighbourhoods. This finding implies that employer involvement can be implemented regardless of the care centre’s socioeconomic and geographical context. Inner setting There was a clear difference in the level of preparation at the successful versus unsuccessful centres in terms of preexisting organizational structures and routines for working with sickness certification and insurance medicine. At the successful centres, formalized routines of cooperation were in place between the GPs and RCs. However, these structures and routines differed considerably between centres, and there was no indication of any generally established model or procedure. Nevertheless, at centres where such structures and routines existed, the standardized work procedure was introduced into an environment that aligned with the core component of employer involvement; this implementation also strengthened the already existing routines. In CFIR terminology, the compatibility between the intervention and the already existing values, workflows and systems at the successful centres was high, which facilitated the implementation. These results indicate that initiatives aiming toinvolve employers early on in the RTW process should ensure that routines for insurance medicine and solid structures for working with sickness certification are established before this step is launched. Implementation did not work at the centres that joined the project with the aim of establishing structures and routines for insurance medicine. The results also show that the primary health care centres were struggling with a high workload and staff discontinuity, and that this situation was particularly salient at the centres where the implementation failed. Time constraints among GPs have been reported to hamper RTW interventions in earlier studies [e.g. 26]. In CFIR terms, such structural characteristics can hinder change initiatives; ‘change enthusiasts’ quit, newcomers do not receive an adequate introduction to the initiative, and the time and interest required to adopt new procedures and methods are lacking. In this situation, well-established and formal work roles, solid structures and work routines can make the different functions less dependent on the specific individuals who perform them. RCs at both the successful and unsuccessful centres perceived that the preparedness to use the coordinator’s service varied among the GPs. At the successful centres, however, not only did the GPs value the coordinator’s services, but the coordinator also had a clear and well-established role in relation to the GPs. Furthermore, at the successful centres, one or more GPs were committed to the implementation and collaborated with the RC to secure the standardized work procedure. With time, the coordinators became increasingly confident with involving employers, and their belief in their own ability to ‘execute courses of action to achieve implementation goals’ [21, p. 9] – that is, their self-efficacy – increased. High-quality communication and collaboration between GPs and RCs were thus important prerequisites for implementation; in addition, at least one GP at the centre needed to be convinced that the intervention would bring about a positive change. In CFIR vocabulary, the results indicate that employer involvement in primary health care is enabled when GPs are committed and act as ‘opinion leaders’ or even ‘champions’ for this endeavour. The importance of having physicians ‘on board’ is a widely recognized prerequisite for successful change implementation in a health care setting [27-28]. In Reed and Kalaga’s [19] review of work-focused interventions in primary care, GPs’ knowledge and confidence in understanding the impact of work on mental health also played a crucial role. For RCs with an unclear status and role at their workplace, and with no peer among the GPs, implementation was not feasible. The intervention did not serve to clarify the role of coordinators in settings where this role was not already satisfactorily developed. Managers that are considering expanding their units’ services to include employer contact should therefore ensure that the occupational roles of the involved staff members in the unit are clear. Implications Taken together, the results of this study lead us to make the following recommendations for primary care stakeholders aiming to involve employers in the rehabilitation of stressrelated disorders: (1) make sure to establish solid routines and structures for insurance medicine and sickness certificationin which the professional roles and mandates of the involved professional groups are explicit and clear; (2) make sure that the work situation at the centre in terms of staff stability and workload permits the GPs to engage in the change process; and (3) pursue a high degree of adaptation to local workflows and routines, in terms of both how employers are involved and how patients are referred to the programme. A ‘good fit’ between intervention and context has been emphasized as critical in both generic [21] and primary-care-specific [29] implementation frameworks, and this study adds empirical support to this claim.

The categorization of participating centres into a successful group and an unsuccessful group can be disputed. An average number of employer contacts per successful centre of 10.6 patients (as calculated from Table 1) is quite low. As a matter of fact, only one centre came close to the target of twenty patients [20], with sixteen employer contacts. This result can only be considered ‘successful’ in comparison with the failure to implement the intervention that was seen in the unsuccessful group. However, the intervention’s standardized work procedure was followed for 53 patients in total, and this was considered to motivate a comparison between the centres that at least tried to implement and the centres that never moved on from the decision to commit to implementation to implementation itself. A second limitation relates to the inherent difficulties of implementation research. Taking time from busy health care staff members and managers to collect data can be problematic, and involves maintaining a continuous balance between the staff’s work situation and the criteria for scientific accuracy. For example, we intended to interview one GP, the RC and the manager at each centre both pre- and post-intervention, but this could not always be accomplished due to high workloads and unforeseen events in the daily practice. The lack of coherence in the data collection procedure – with some interviews being conducted with all professions present, and others with only one or two – might have contributed to a certain social desirability bias. For example, the RCs might have been less outspoken when managers and GPs were participating in the interview session. Since the analytical strategy was to analyse all data from the successful centres as a group, and then compare it with the data from the unsuccessful group, this eventual bias in the data from single centres was reduced.

In the present study, GPs and RCs at eleven Swedish primary care centres were followed in their attempts to introduce early employer contact in order to induce RTW in patients with stress-related disorders. It was concluded that primary care stakeholders aiming to involve employers should establish solid routines and structures for insurance medicine with clear roles for the involved professional groups and should stabilize the staff situation before this step is launched.

This study was funded by the Swedish Research Council for Health, Working Life and Welfare, Grant No. 2015-00703. The authors would like to acknowledge Marie Hammer,Sandra Pettersson, Anja Beno and Karin Pokorny for their contributions to this study. The study is part of the New Ways research programme on mental health at work, in the Section for Epidemiology and Social Medicine, Institute of Medicine, Sahlgrenska Academy, University of Gothenburg.

1. Hassard J, Teoh KRH, Visockaite G, Dewe P, Cox T. The cost of workrelated stress to society: A systematic review. J Occup Health Psychol. 2018; 23: 1-17.

2. Schaufeli WB. Burnout in Europe: Relations with national economy, governance, and culture. Research Unit Occupational and Organizational Psychology and Professional Learning. KU Leuven, Belgium; 2018.

3. Eurofound. Burnout in the workplace: A review of data and policy responses in the EU. Publications Office of the European Union, Luxembourg: Publications Office of the European Union, Luxembourg. 2018.

4. The Swedish social insurance agency. Social. försäkringsrapport. 2020: 8.

5. Löfgren A, Hagberg J, Arrelöv B, Ponzer S, Alexanderson K. Frequency and nature of problems associated with sickness certification tasks: a cross-sectional questionnaire study of 5455 physicians. Scandinavian journal of primary health care. 2007; 25: 178-185.

6. Söderberg E, Alexanderson K. Sickness certificates as a basis for decisions regarding entitlement to sickness insurance benefits. Scandinavian Journal of Public Health. 2005; 33: 314-320. 

7. Swedish NAO. Bedömning av arbetsförmåga vid psykisk ohälsa- en process med stora utmaningar [Assessment of work ability in mental health – a process with major challenges]. 2018.

8. Wynne-Jones G, Mallen CD, Main CJ, Dunn KM. What do GPs feel about sickness certification? A systematic search and narrative review. Scandinavian journal of primary health care. 2010; 28: 67-75.

9. Bertilsson M, Maeland S, Löve J, Ahlborg G, Jr, Werner EL, Hensing G. The capacity to work puzzle: a qualitative study of physicians’ assessments for patients with common mental disorders. BMC Fam Pract. 2018; 19: 133.

10.Nigatu YT, Liu Y, Uppal M, McKinney S, Gillis K, Rao S, et al. Prognostic factors for return to work of employees with common mental disorders: a meta-analysis of cohort studies. Soc Psychiatry Psychiatr Epidemiol. 2017; 52: 1205-1215.

11.de Vries H, Fishta A, Weikert B, Rodriguez Sanchez A, Wegewitz U. Determinants of Sickness Absence and Return to Work Among Employees with Common Mental Disorders: A Scoping Review. Journal of Occupational Rehabilitaion. 2017; 1-25.

12.Gragnano A, Negrini A, Miglioretti M, Corbiere M. Common Psychosocial Factors Predicting Return to Work After Common Mental Disorders, Cardiovascular Diseases, and Cancers: A Review of Reviews Supporting a Cross-Disease Approach. J Occup Rehabil. 2017.

13.Perski O, Grossi G, Perski A, Niemi M. A systematic review and metaanalysis of tertiary interventions in clinical burnout. Scand J Psychol. 2017; 58: 551-561.

14.Mikkelsen MB, Rosholm M. Systematic review and meta-analysis of interventions aimed at enhancing return to work for sick-listed workers with common mental disorders, stress-related disorders, somatoform disorders and personality disorders. Occupational and Environmental Medicine. 2018.

15.Eklund M, Erlandsson LK. Return to work outcomes of the Redesigning Central Björk L, et al. (2021) J Family Med Community Health 8(1): 1183 (2021) 8/8 Björk L, Glise K, Pousette A, Holmgren K, Bertilsson M (2021) Implementing Employer Involvement for Patients with Stress-Related Disorders in Swedish Primary Care – A Process Evaluation.  8(1): 1183. 

16.Karlson B, Jonsson P, Palsson B, Abjornsson G, Malmberg B, Larsson B, et al. Return to work after a workplace-oriented intervention for patients on sick-leave for burnout--a prospective controlled study. BMC Public Health. 2010; 10: 301.

17.Karlson B, Jonsson P, Osterberg K. Long-term stability of return to work after a workplace-oriented intervention for patients on sick leave for burnout. BMC Public Health. 2014; 14: 821.

18.Salomonsson S, Santoft F, Lindsäter E, Ejeby K, Ljótsson B, Öst L-G, et al. Cognitive–behavioural therapy and return-to-work intervention for patients on sick leave due to common mental disorders: a randomised controlled trial. Occup Environ Med. 2017; 74: 905-912.

19.Reed K, Kalaga H. Focusing on employment in primary mental health care: A scoping review. Work. 2018; 59: 3-13.

20.Björk L, Glise K, Pousette A, Bertilsson M, Holmgren K. Involving the employer to enhance return to work among patients with stressrelated mental disorders – study protocol of a cluster randomized controlled trial in Swedish primary health care. BMC Public Health. 2018; 18: 838.

21.Damschroder LJ, Aron DC, Keith RE, Kirsh SR, Alexander JA, Lowery JC. Fostering implementation of health services research findings into practice: a consolidated framework for advancing implementation science. Implementation Science. 2009; 4: 50.

22.Pawson R, Tilley N. Realistic Evaluation. London: SAGE; 1997.

23.Nilsen P, Bernhardsson S. Context matters in implementation science: a scoping review of determinant frameworks that describe contextual determinants for implementation outcomes. BMC Health Services Research. 2019; 19: 189.

24.Watson DP, Adams EL, Shue S, Coates H, McGuire A, Chesher J, et al. Defining the external implementation context: an integrative systematic literature review. BMC Health Services Research. 2018; 18: 209.

25.van Oostrom SH, van Mechelen W, Terluin B, de Vet HC, Knol DL, Anema JR. A workplace intervention for sick-listed employees with distress: results of a randomised controlled trial. Occup Environ Med. 2010; 67: 596-602.

26.Bakker IM, Terluin B, van Marwijk HWJ, van der Windt DAWM, Rijmen F, van Mechelen W, et al. A Cluster-Randomised Trial Evaluating an Intervention for Patients with Stress-Related Mental Disorders and Sick Leave in Primary Care. PLOS Clinical Trials. 2007; 2.

27.Greening J. How can we improve the effective engagement of doctors in clinical leadership? Leadership in Health Services. 2012; 25: 20-26.

28.Bååthe F, Norbäck LE. Engaging physicians in organisational improvement work. Journal of Health Organization and Management. 2013; 27: 479-497.

29.Lau R, Stevenson F, Ong BN, Dziedzic K, Treweek S, Eldridge S, et al. Achieving change in primary care—causes of the evidence to practice gap: systematic reviews of reviews. Implementation Science. 2016; 11: 40.