Implementation of the individual placement and support approach – facilitators and barriers

The individual placement and support approach (IPS) has become a widely recognized evidence-based practice to provide work for more people with severe mental illness. The aim of this literature review was to identify and evaluate research on implementation of IPS, focusing on facilitators and barriers. Contextual, local organizational, cooperation/team and individual factors influence the implementation process. Key facilitators are the use of a fidelity scale to measure and develop quality and the employment of skilled local leaders and IPS specialists. Barriers are located at the contextual level, when the national employment policy and regulation contradict the IPS scheme, and at the local level, where mental health professionals’ negative attitudes towards the IPS scheme and a culture based on a medical approach challenge the implementation of IPS. The evaluation of research in IPS implementation show that most studies are empirically driven, using different understandings of implementation and have a poor theoretical underpinning of the studies. The need for further studies based on comparative methods and more developed theoretical framework is discussed. ARTICLE HISTORY Received 22 September 2015 Accepted 22 July 2016


Introduction
According to OECD (2012), the unemployment rates are 3-6 times higher for people with severe mental disorder than for people with no mental disorder. However, studies indicate that with adequate support, people with mental health problems are able to get a job. The Individual Placement and Support -IPSapproach has become a widely recognized evidence-based practice (EBP) to increase work frequency for persons with severe mental illness (Kinoshita et al. 2013). IPS is based on a core set of principles, namely 1. Focus on Competitive Employment, 2. Eligibility Based on Client Choice, 3. Integration of Rehabilitation and Mental Health Services, 4. Attention to Client Preferences, 5. Personalized Benefits Counselling, 6. Rapid Job Search, 7. Systematic Job Development and 8. Time-Unlimited and Individualized Support (Drake, Bond, and Becker 2012).
The IPS model developed in the 1980s, inspired by the supported employment model (Drake, Bond, and Becker 2012). A systematic review from 2013 (Kinoshita et al. 2013) compares supported employment and IPS with other approaches for finding employment. Drawing from a total of 2259 people with mental health problems in 14 studies, the review has two main findings: (1) Supported employment increases the length and time of a person's employment; (2) Persons on supported employment find jobs faster. Supported employment and IPS are better than other approaches in these two respects. Due to the good results in the USA, research on the effect of IPS has been carried out in many countries, including several European countries with similar positive results ( van Erp et al. 2007;Koletsi et al. 2009;Rinaldi, Miller, and Perkins 2010;Hasson, Andersson, and Bejerholm 2011;Knaeps, DeSmet, and Van Audenhove 2012). Therefore, the interest in the IPS-supported employment approach is growing among professionals as well as among governments and policymakers.
Implementation is about the efforts to incorporate a programme or practice at the community, agency or practitioner levels. Implementation research has shown that the implementation of an innovative and evidence-based method, like the IPS-programme, does not occur in a vacuum (Fixsen et al. 2005;Meyers, Durlak, and Wandersman 2012). Existing contextual factors influence the implementation. Potential barriers may be rooted in structural factors such as a welfare model where vocational rehabilitation services are based on a prevocational training approach. The acceptance of new knowledge based on evidence-based methods can be challenged by professionals (Fixsen et al. 2005). Organizational structure can create barriers for the integration and cooperation between mental health and employment services (Boardman 2013). Knowledge about which factors affect implementation of IPS is important to support national and local processes, as is the development of an effective supported employment approach that can support people with mental health problems in finding employment.
Previous reviews have focused on the literature on IPS implementation in a specific country or on producing a general summary of the literature. Rinaldi, Miller, and Perkins (2010) review the evidence of implementing IPS in England. They identify four themes that emerge as challenges to implementation: fear on the part of professionals, individuals and their families that work will have a negative impact on the client, a culture of low expectations, failure to provide the support that is proven to work, and the global recession. Contreras et al. (2012) review the research from Australia and New Zealand and maintain that the central challenge for implementing IPS is the incorporation of the vocational assistance into the public mental health services. Boardman (2013) summarizes the research on IPS and possible barriers for implementation and points to clients' motivation, self-efficacy and previous employment history. Other barriers are attitudinal such as low expectations from clinical staff, and employers' prejudice. Additional factors that influence implementation are structural such as the welfare and employment policies and regulations, or are embedded in the organizational context and the way the context affects the professionals' attitudes and work performance.
While the previous reviews present important information on factors influencing IPS implementation, this review is based on a systematic review method. The primary aim is to perform an analysis of potential facilitators and barriers when implementing IPS. The secondary aim is to evaluate the research on IPS implementation to generate an overview of the methods and theories used, and discuss these findings in relation to future studies.

Reviewing the implementation of IPSconcepts and methods
The research field of studying implementation is broad and can be located in different traditions and disciplines. Nilsen et al. (2013) distinguish between the traditions that developed out of (a) policy implementation research and (b) implementation science, with the former researching the implementation of 'Big P', understood as policies in the form of formal laws, rules and regulations, and the latter studying the 'small p', understood as the methods to promote the systematic uptake of research findings and other EBPs into routine practice in health services and care (Nilsen et al. 2013;Nilsen 2015). Common to the two traditions is the investigation of the process of translating intention into desired change, examining the output and outcome of the intended policy, programme or guideline. In the implementation science tradition, implementation is defined as a step in the process from the phase where someone explore and decide to proceed with the incorporation of a programme until the programme reaches sustainability. Fixsen et al. (2005) distinguish between six stages of the implementation process: exploration and adoption, programme installation, initial implementation, full operation, innovation and sustainability. Meyers, Durlak, and Wandersman (2012) have identified the critical steps in quality implementation, and show how most of these steps should be addressed before implementation begins, that is, developing an assessment strategy, considering and deciding on adaption, obtaining explicit buy-in from critical stakeholders, fostering a supportive community/organizational climate, and creating a structure for implementation. The implementation concept is used in a broad sense, as a concept that defines all phases in the process and in a more narrow sense, as the phase where the programme actually begins/the initial implementation phase (Meyers, Durlak, and Wandersman 2012;Fixsen et al. 2005).
Furthermore, the studies of factors that function as facilitators or barriers for implementation output and outcome are common topics and several frameworks have been developed to describe these factors, place them into levels and develop theoretical frameworks or models. The strength of these frameworks is that they describe empirical phenomena by fitting them into a set of categories and thereby provide an overview of ideas and practices that shape the complex implementation process and can help researchers and practitioners use the ideas of others who have implemented similar programmes (Meyers, Durlak, and Wandersman 2012). The frameworks imply a systems approach to implementation because they point to multiple levels of influence and acknowledge that there are relationships within and across levels (Nilsen 2015). Nevertheless, also recognizing that different barriers and enablers may interact in various settings and the interaction between factors can be difficult to explain and predict (Proctor et al. 2009;Winter 2012;Nilsen 2015). According to Nilsen (2015), it is furthermore possible to distinguish between different categories of theories, models and frameworks. Nilsen develops a taxonomy of five categories of theoretical approaches to implementation studies. One approach is based on a process model that focuses on steps in the process of translating research into practice. The second approach is determinant frameworks that focus on determinants, acting as barriers and enablers to influence implementation. The third approach draws on classic theories that originate from other disciplines like sociology and organizational theory. The forth approach is implementation theories developed to provide understanding and/or explanation of aspects of implementation, and the fifth approach draws on evaluation frameworks (Nilsen 2015, 13).
IPS is known as an EBP based on the core principles mentioned in the introduction. The implementation of IPS can be supported by fidelity reviews that guide the local community or agency in reaching high-quality implementation. The aim of this review is to examine potential factors that facilitate or function as barriers when implementing IPS, and to evaluate the research on IPS implementation. The literature covering IPS can be identified using the concept in the search strategy. However, the research on factors influencing the implementation of IPS can be reported using other terms and from other approaches, which makes it difficult to perform a comprehensive review of the implementation topic. A pragmatic solution to this challenge is to review the studies, which are reported as implementation studies or studies that report factors influencing implementation. The limitation of such a strategy is that potentially relevant studies may not be included; for instance, studies that examine the IPS specialists' skills, roles and discretionary decisions when delivering the IPS programme to clients and organizations. Such studies may use other labels and terms than the concept of implementation and are therefore not included in this review. Given the state of the field, we have chosen to have a rather open conceptual understanding of the implementation topic with the aim to get an insight into how the included papers understand implementation. We will therefore look into how the studies included view implementation with a reference to Nilsen's taxonomy, and how future implementation studies may learn from these findings.
This literature review was conducted as part of a research project on implementing IPS in Denmark (Bonfils 2015). The focus on the Danish context has guided our search strategy, as we have been interested in studies and previous reviews from other western countries, particularly other Nordic countries, as these countries have similar welfare systems to that of Denmark.

Search strategy
The search strategy involved two steps. First, we made a search for relevant articles published in English in the databases SocIndex, Cinahl, PsycArticles, PsycInfo, Academic Search Elite, and PubMed, using the following keywords: (Psychiatric rehabilitation OR mental illness OR psychiatric disabilities OR mental health care OR recovery) AND (individual placement and support OR IPS) AND (implementation OR implementing OR from research to practice).
Second, we made a search in SwePub.se, Libris.kb.se, su.diva-portal.org/smash/search.jsf, bibsys.no, NORA, Bibliotek.dk, Forskningsbasen.deff.dk and ucviden.dk for Nordic studies using the keyword 'individual placement and support OR IPS'. The search was conducted in October 2013. A second search was made in April 2015 to include studies, which had been published since.

Selection of studies
We included: . Studies that explore the implementation of IPS from a theoretical or empirical approach, using either qualitative methods, quantitative methods or mixed methods. Fidelity studies were included. . Original articles and review articles, including systematic reviews. . Articles based on knowledge from the implementation of IPS in relation to people with severe mental illness. . Studies conducted in Western countries. . Studies published since and including 1996.
We excluded: . Articles not published in English or one of the following Nordic languages: Swedish, Norwegian or Danish.

Screening process
The screening process was conducted in two stages: 1. screening title and abstract; 2. full texts. Stage 1 was conducted by two of the four team members (Inge Storgaard Bonfils (ISB) and Henrik Hansen (HH)), and started with a pilot study of 20 abstracts, aimed at ensuring consensus among the reviewers in applying the eligibility criteria. References were categorized as 'relevant' or 'not relevant' according to the eligibility criteria described. All discrepancies were discussed. After the pilot study, both reviewers read and screened all articles. Stage 2 was conducted by two of the four team members (ISB and HH). We screened all full-text articles applying the eligibility criteria and disagreements were discussed. We then checked the reference list of the included articles and included articles that seemed relevant. Screening in the second search April 2015 was conducted by ISB.

Criteria for judging paper quality
The selected papers based on qualitative methods were assessed for methodological validity using standardized Critical Appraised Skill Programme for qualitative studies (Critical Appraisal Skills Programme (CASP) Qualitative Research Checklist 31 May 2013). Studies based on surveys or mixed methods were judged looking at the transparency, trustworthiness and ethical consideration of the research design. The judging of paper quality was conducted as part of the data extraction process and we used the judging to make a 'sensitivity analysis' to assess the possible impact of the study quality on the review's findings (Thomas and Harden 2008;Barnett-Page and Thomas 2009).

Data analysis
A conceptual framework was developed to identify factors that facilitate or function as barriers in the implementation of IPS. Insight from both implementation science, represented by Proctor et al.'s (2009) overview of implementation research in mental health service, Nilsen's (2015) overview of implementation theories, models and frameworks and policy implementation research represented by Winter and Nielsen (2008) and the Integrated Implementation Model, influenced the development of the framework. The framework is based on a systems approach and distinguishes between factors that influence implementation on different levels; the larger system/policy environment, which we define as contextual factors, the organizational level, the group or team level and the individual level concerning the frontline staff . A pilot study was conducted using this framework to generate key themes and develop an overarching conceptual framework dividing the factors into categories and subcategories on the four analytical levels; (1) contextual factors at the system and society level (welfare system, labour market/economic situation, attitudes and tradition/culture), (2) factors at the local organizational level (finance, fidelity, leadership and supervision), (3) factors concerning cooperation at the group level (the relation between IPS specialist and clinical team) and (4) factors at the individual level (IPS specialist, clinical team members). Three reviewers (ISB, HH and Helle Stentoft Dalum) participated in the pilot study, the judging of paper quality and data extraction. Disagreements were resolved by discussion until consensus was reached. ISB made the data extraction of papers included in the second search April 2015 to identify if or how these papers added to the analysis of factors that facilitate or function as barriers for implementing IPS.

Results
In total, 261 articles were identified in the databases and reference list checking in the first search in 2013. Eighteen papers were included, based on primary studies, reports/evaluations and reviews. In the second search in April 2015, 85 papers were identified, and 3 were included, based on primary studies and evaluations reports. Altogether 21 papers were included. Most often, deletions occurred when (1) the studies were not applicable/did not explore IPS or (2) implementation of IPS was mentioned in the article, but was not evaluated in any way in the study presented.

Study design, methods and theoretical approach
The 21 articles report studies from the USA, Australia, UK, Canada, Sweden, the Netherlands and Belgium.  (Rinaldi, Miller, and Perkins 2010). A third study was based on a framework, which outlines an implementation theory of six core elements for successful change (Schneider and Akhta 2012). The fourth study draws on organizational theory and a conceptual framework that aim to shed light on organizational and contextual factors influencing implementation (Menear et al. 2011). The fifth study (Markström et al. 2015) developed a model for analysing data based on several implementation theories.
The remaining articles framed the study within an empirical discussion of mental health services and supported employment, however, without a clear definition or reference to a specific implementation theory, framework or model. However, some of the papers are based on fidelity studies or include fidelity studies as part of the study design (van Erp et al. 2007;Waghorn et al. 2007;Killackey and Waghorn 2008;Boyce et al. 2008;Cocks and Boaden 2009;Knaeps, DeSmet, and Van Audenhove 2012;Marshall et al. 2008;Shepherd et al. 2012;Markström et al. 2015). These studies imply an understanding of successful implementation as equivalent to high fidelity. Other studies focus on facilitators and barriers and draw on a kind of determinant framework approach. Some of these studies combine fidelity studies and a determinant framework approach (Bejerholm, Larsson, andHofgren 2011, Boardman 2013;Killackey and Waghorn 2008;Knaeps, DeSmet, and Van Audenhove 2012;Kostick, Whitley, and Bush 2010;Marshall et al. 2008;Quimby, Drake, and Becker 2001; van Erp et al. 2007;Waghorn et al. 2007). A third category of studies take the point of departure from a process model, describing the process of implementing IPS. Some of these studies combine a process evaluation with fidelity studies, determinant models and/or formative evaluations (Oldman et al. 2005;Becker et al. 2007;Folkesson 2014;Socialstyrelsen 2014;Markström et al. 2015). To summarize, the studies of IPS implementation are characterized by the use of a variety of theoretical approaches and in some cases, the absence of conceptual models. The studies use different methods; however, these are primarily qualitative methodology and case studies. The papers are rather diverse when it comes to methodological quality and study design. However, in our judging of the papers' quality, we found that the studies all employed a design appropriate to the research questions addressed. Some studies can, however, be criticized for being insufficient in their theoretical approach, and this could be addressed in further investigations (Table 1).

Findings based on the analysis
The factors that influence implementation are reported as both facilitators and barriers. Some articles highlight barriers. Some focus on facilitators and some emphasize both. Figure 1 gives an overview over the main findings. In the following section, we will present the facilitators and barriers in general terms and in the next section, we will discuss the strengths and limitations of the analysis (Figure 1).

Contextual factors
Not all studies report contextual factors, as the studies did not directly address the issue of how factors on system and society level influence implementation. Fifteen of the studies reveal potential barriers through the existing literature and/or examine the influence of the system level through case studies and interviews with leaders and frontline staff. Despite the variations in data provided to highlight contextual factors, several key factors were pointed out. National employment policy and regulations based on a 'train then place' model are reported as barriers for implementing IPS (van Erp et al. 2007;Rinaldi, Miller, and Perkins 2010;Hasson, Andersson, and Bejerholm 2011;Boardman 2013;Socialstyrelsen 2014;Markström et al. 2015). An example: when national rules for assessment of work capacity contradict the zero exclusion policy of IPS, and job seekers are required to take part in an assessment before entering an employment programme (Waghorn et al. 2007). Another factor is the 'benefit trap', as people risk losing social benefits if they become employed (van Erp et al. 2007;Boardman 2013). The lack of financial incentives to work acts as a structural barrier that demotivates clients, as working may not improve their financial situation. Structural hindrances to inter-sectoral and/or inter-governmental collaboration between mental health services and the employment system constitute a third factor. The structural division of the services makes collaboration and integration difficult (Waghorn et al. 2007;Knaeps, DeSmet, and Van Audenhove 2012;Markström et al. 2015). The national labour market and the unemployment rate affect the opportunities to find work in general. Some of the studies (van Erp et al. 2007;Schneider and Akhta 2012;Socialstyrelsen 2014) emphasize that a labour market that offers few opportunities for low-skilled workers, challenges the work situation for people with mental health problems who lack skills and previous work experience. Another hindering factor is the employer's willingness to employ people with mental disabilities as they fear the employee cannot live up to their expectations (Knaeps, DeSmet, and Van Audenhove 2012). The general attitudes and culture among professionals and employers are also mentioned as factors that influence implementation (Rinaldi, Miller, and Perkins 2010;Bejerholm, Larsson, and Hofgren 2011;Hasson, Andersson, and Bejerholm 2011;Boardman 2013). Mental health professionals' negative attitudes towards the IPS scheme challenge IPS implementation because they often emphasize that symptoms must disappear before a person is ready for vocational rehabilitation. The IPS approach also challenges the attitudes and culture among vocational workers and other professionals with a 'cure, care and graduate rehabilitation' approach who wish to protect their clients and have low expectations of their ability to work (Rinaldi, Miller, and Perkins 2010;Hasson, Andersson, and Bejerholm 2011;Shepherd et al. 2012).

Local organization and cooperation
All the included studies report on factors on the local organizational and cooperation level. A general experience is that the local implementation of IPS is a process that takes time (Rinaldi, Miller, and Perkins 2010;Schneider and Akhta 2012;Shepherd et al. 2012;Folkesson 2014;Socialstyrelsen 2014;Markström et al. 2015). A number of factors influence the process, and the IPS/SE fidelity scale is a common method to measure the degree to which the IPS model is implemented. The use of fidelity studies is highlighted as a method to develop the quality of the service and as a quality indicator for management information. Low fidelity scores are reported in several studies. Examples of this are a low degree of integration of vocational rehabilitation with mental health treatment, eligibility criteria for access to the scheme, high caseloads, limitations to the extent of follow-up and outreach, the use of key account employers rather than respecting individual job preferences  (Boyce et al. 2008;Cocks and Boaden 2009;Schneider and Akhta 2012;Boardman 2013;Knaeps, DeSmet, and Van Audenhove 2012).
The ability to secure funding for the scheme in the long run is reported as a factor that supports the implementation of IPS (Oldman et al. 2005;Rinaldi, Miller, and Perkins 2010;Socialstyrelsen 2014). However, the funding model may also function as a barrier if it is based on short-term outcomes and other criteria that constrain the use of the IPS scheme (Boyce et al. 2008;Cocks and Boaden 2009).
The role of leadership is highlighted as a factor for successful implementation. Successful leadership is described as having a personal commitment to the programme, taking on a persuasive and assertive role, taking on the role of a change agent, having the authority to transfer or discharge IPS specialists or programme leaders who refuse to follow the IPS approach, having administrative skills and setting performance standards and clear job expectations (van Erp et al. 2007;Marshall et al. 2008;Rinaldi, Miller, and Perkins 2010;Shepherd et al. 2012). Supervision and the continuous education of IPS specialists are also highlighted as factors that facilitate the implementation of IPS. The supervisor must have first-hand experience as an IPS specialist and provide team support. Rinaldi, Miller, and Perkins (2010) and Boardman (2013) highlight that it required training and consultation from experienced IPS coordinators or purveyors to implement IPS. Studies (Marshall et al. 2008;Rinaldi, Miller, and Perkins 2010) show how the use of outcome data and success stories and user stories support the staff in having positive attitudes towards the IPS scheme. However, factors such as staff and programme leaders' turnover, data collection and other types of administrative work that takes time, challenge the managers' time for supporting the implementation of the programme (Oldman et al. 2005; van Erp et al. 2007).
The IPS approach is based on the integration of IPS specialists in mental health treatment teams and support from the clinical team members facilitates the process. However, some studies (van Erp et al. 2007;Boyce et al. 2008;Hasson, Andersson, and Bejerholm 2011;Menear et al. 2011;Knaeps, DeSmet, and Van Audenhove 2012) show how inadequate cooperation challenges the implementation. Different values, attitudes and beliefs in the client's ability to work create conflicts between the clinical team and the IPS specialists. There are also examples of conflicts developing due to a lack of coordination of activities between the clinical team members and the IPS specialist (Quimby, Drake, and Becker 2001). Initiatives such as consensus building, training before launching the IPS programme and regular team meetings support the development of a positive cooperation (Oldman et al. 2005;van Erp et al. 2007;Folkesson 2014).

Individual level
Half of the papers report on how factors related to the individual level influence implementation. The main focus is on the IPS specialist and the development of a person-centred approach. However, only few papers address this issue in details. The development of a person-centred approach interacts with the welfare system's demands on the client and the support and demands on the IPS specialist, which make it difficult to separate these factors from the contextual and organizational level. As an example, Bejerholm, Larsson, and Hofgren (2011) mention how the IPS specialists find it difficult to explain the rules and regulations of the welfare system to clients, leading to the clients becoming concerned and frightened of losing financial security if they follow their goal of obtaining employment. This example points to how the IPS specialists' knowledge of the national welfare policy, rules and regulations plays a role in creating a positive partnership with the client. Kostick, Whitley, and Bush (2010) examine the notion of client-centredness from the perspective of the IPS specialist and supervisors, identifying barriers and facilitators to implementation in accordance with the IPS model. They interpret client-centredness as a product of negotiations between client and IPS specialist, and employers, agencies and other para-professionals. They identify four principal factors influencing implementation of client-centredness, the first two related to the client's anxieties about their interest and abilities, and their ability to have realistic work preferences. IPS specialists address these issues through communicating their commitment and flexibility and negotiate unrealistic job expectations, according to the IPS model. IPS specialists experience this as a fragile balance between meeting the client where he/she is at, and encouraging clients to transcend self-doubt. In this process, IPS specialists must be skilled in envisioning how clients will respond to particular work environments and work with employment-seeking efforts accordingly. The two other principal factors are related to what we call the organizational level. These are factors that influence the IPS specialist's ability to maintain motivation to be client-centred through support from supervisors with first-hand experience, thus facilitating the implementation process, as mentioned in the previous section. Pressure from agency to meet specific outcome numbers functions as barriers, as client preferences may be overlooked to increase the performance standards. Institutional factors influence the ability of IPS specialists to match clients' preferences, as the specialist needs to collaborate with other institutions. This inter-agency collaboration can function as a barrier if it compromises client-centredness and the IPS model; however, it also functions as a facilitating component if they expand the scope of resources available to clients and hereby promote positive employment outcomes. These two examples of studies that address the individual level point out how the IPS specialist's skills, knowledge and experience are essential to a successful implementation of the scheme. The IPS specialist is in the frontline of delivering the service to the client and transform the programme into everyday practice (Cocks and Boaden 2009;Kostick, Whitley, and Bush 2010). The studies, however, interpret factors concerning the individual level in relation to the contextual and organizations level, and thereby point out the relationships within and across the levels. The development of skilled IPS specialists is an important issue in the implementation of the IPS programme and as such integrated into the development of high fidelity programmes on the organizational level.

Discussion
Although we screened 346 titles and abstracts, studies that evaluate factors that influence implementation may be reported using other terms and approaches than the ones used in this search strategy and selection of studies. The limitation of this review is that it is difficult to perform a comprehensive review of the implementation topic, and given the state of the field, the review has become more loose to capture and summarize the facilitators and barriers revealed in the included papers. As mentioned, we made a 'sensitivity analysis' to assess the possible impact of the study on the review's findings. We found that the poorer quality studies contributed comparatively little to the analysis. The facilitators and barriers documented in these studies were also revealed in higher quality studies. The better studies had a more developed analysis and contributed most.
The findings of this review show that factors influencing IPS implementation are multiple and differentiated. An important learning point is that there is an inherent interdependency between the various factors and it can therefore be difficult to differentiate between factors on different levels. Contextual factors rooted in the welfare systems' traditions are institutionalized in the local organizational context of the mental health services and the vocational rehabilitation services. The attitudes, values, management and administrative traditions and cultures also influence the implementation on the local level. However, the different barriers and enablers may interact differently in various settings. We find that the interpretation of factors influencing IPS implementation is challenged by a lack of clarity about how contextual settings influence the implementation process. The first point we want to address is that the studies of IPS implementation have been carried out in different western countries (USA, Australia, Canada, UK, Sweden, The Netherlands and Belgium) and the contextual differences between countries regarding health, employment, social care and welfare systems are not fully clarified in the papers and in this review. The differences in context and welfare systems may have influenced findings in various ways as the implementation of IPS is organized differently across countries. An example from Australia shows how a successful implementation of the IPS model was developed through bypassing the federal disability employment system (Killackey and Waghorn 2008). Studies from Sweden show how the main barriers for implementing IPS are found in the national welfare system, law and regulation (Bejerholm, Larsson, and Hofgren 2011;Hasson, Andersson, and Bejerholm 2011;Socialstyrelsen 2014;Markström et al. 2015). Further studies based on comparative methods are therefore needed to clarify the influence different welfare systems and -models have on IPS implementation.
Secondly, we want to point out, that contextual setting also implies an understanding of the way the political environment and government support the implementation process, and how political and administrative settings may have influenced the results of the studies included in the review. In some cases, IPS has been implemented as part of a national strategy and supported by government funding (examples of this are: Socialstyrelsen 2014; Markström et al. 2015). In other cases, IPS have been implemented as part of an RCT study and supported financially through the research project. Bejerholm, Larsson, and Hofgren (2011) point out how the implementation of IPS in a Swedish city was supported by the mental healthcare services and not the public employment services and social insurance agencies. This may have influenced the process and the result of the study as the mental health care regarded the IPS as part of their service and had positive attitudes towards the model, whereas the other agencies may have been reluctant and not fully committed to the process.
Finally, contextual setting can also imply an understanding of how the implementation of IPS today is supported through international collaboration and experts. Menear et al. (2011) show how experts from the USA have had direct involvement in supporting IPS implementation in three provinces in Canada, facilitating the process. However, the influence of international collaboration on national or local implementation has so far not been examined.
Studies of IPS implementation are an emerging research field and so there is a need to advance the development of both the methodological and conceptual framework to guide the research further (Proctor et al. 2009;Nilsen et al. 2013). However, the development of a grand implementation theory may not be the solution. As stated by Nilsen (2015), implementation is too multifaceted and complex a phenomenon to allow for universal explanations. The use of theories in designing the study can, however, open for a more explicit questioning of the assumptions regarding, for instance, the interaction between factors on different levels that facilitate or act as barriers for implementation. Knowledge from social and political science can support the understanding of how the outer context influences the political and cultural milieu on the organizational and individual level in which IPS is carried out. Future studies could address this question.

Conclusion
The article contributes to the knowledge of factors influencing IPS implementation through a review of the literature which included 21 papers on factors influencing IPS implementation. Facilitators and barriers to implementation were located on the contextual level, on the local organizational level, and on the individual level, and a conceptual framework was developed to organize and analyse the findings. Key findings for a successful implementation are the use of a fidelity scale to measure and develop the quality of the scheme at a local level, and the employment and development of skilled key implementers such as the local leadership and the IPS specialist. Barriers are located at the contextual level when the national employment policy and regulations contradict the IPS scheme, and the local level, where different values, attitudes and beliefs in the client's ability to work create conflicts between the clinical team and the IPS specialist. Knowledge about factors influencing implementation is important to guide the research of IPS implementation further and with regard to supporting national and local processes. Further studies based on comparative methods are needed to clarify the influence different welfare systems and models have on IPS implementation.

Disclosure statement
No potential conflict of interest was reported by the authors.

Notes on contributors
Inge Storgaard Bonfils, Ph.D., Cand.Scient.Pol., is associate lecturer at the Department of Social Work, Metropolitan University College, Denmark.
Henrik Hansen is associate lecturer, and holds a Master of Science (MSc) degree in Health (Occupational Therapy).
Lene Falgaard Eplov, Ph.D., is Senior consultant and Head of program for research in rehabilitation, recovery & shared care at Mental Health Centre Copenhagen, Denmark.