Collaborator or competitor: assessing the evidence supporting the role of social enterprise in health and social care

ABSTRACT In many countries, social enterprise has been introduced into a competitive market-oriented environment as a substitute for publicly owned services, particularly in healthcare. In the United Kingdom, evidence for this move seems to derive from case studies where social enterprise operates in collaboration – as opposed to competition – with publicly owned services. Our systematic review demonstrates that there is no evidence to support the role of social enterprise as a substitute for publicly owned services. However, there is evidence to show that where social enterprise operates in a collaborative environment, enhanced outcomes can be achieved, such as connectedness, well-being and self-confidence.


Introduction
Since the late 1990s, social enterprisesbroadly understood to mean businesses that trade for a social purposehave been increasingly promoted by governments as a tool to deliver a wide range of policy outcomes.Perhaps, nowhere has this emphasis been more widely felt than the health and social care sector in the United Kingdom (Roy et al. 2013).The former Secretary of State for Health, Andrew Lansley MP, famously declared that he wanted the National Health Service (NHS) to become the 'largest social enterprise sector in the world' when launching his 2010 White Paper Liberating the NHS.This statement was received with consternation by those working in the NHS fearing privatization; from many working in social enterprises, who saw their role as working alongside the public sector rather than replacing it; and from within academia, who saw this as the potential culmination of the New Public Management paradigm with the third sector becoming 'cost effective service providers' of public services (Bovaird 2014).
It is widely noted that there is little evidence to support the trajectory towards social enterprises as a substitute for public services in health (Heins et al. 2010), with policy seemingly driven by ideology, and an implicit acceptance of the New Public Management paradigm (Nicholls and Teasdale 2016).The various policies (discussed later in this paper) are notable in that the 'evidence' they draw upon is limited to descriptive case study examples of 'successful' social enterprises operating in the health sector.It is notable, however, that these illustrative examples have tended to be of social enterprises operating as a complement to the existing publicly funded services, deriving from instances where social enterprise was but one player in a collaborative, pluralistic welfare system (Buckingham 2009).It is therefore somewhat ironic that this 'evidence' is then used to justify an alternative trajectory, whereby social enterprise becomes a substitute to the existing (public) providers, including the 'spinning out' of provision from the National Health Service (NHS) (Hazenberg and Hall 2016;Miller, Millar, and Hall 2012;Sepulveda 2015), particularly given that it would seem that third sector organizations operating in a competitive environment behave much like for-profit providers (Heins et al. 2010;Heins and Bennett 2016).The portrayal of evidence therefore requires an analysis of the contexts in which that evidence was gathered.
In this paper, we present the results of a systematic review addressing the research question: Do social enterprises provide improved outcomes in comparison with usual care in health and social care systems?We separate studies which examined social enterprise as a complement to the existing providers from those that examined social enterprise as a substitute to the existing services.Our contribution is therefore to provide a more nuanced account of the potential role that social enterprise could play in health and social care systems.Systematic reviews are considered a valuable method for critically integrating primary studies, increasing the generalizability and assessing the consistency of evidence (Mulrow 1994).They collect together what is known using accountable methods, and have long been widely recognized as a robust form of research in health and social science studies (Chandler and Hopewell 2013;Petrosino et al. 2001).
Our paper proceeds as follows: Firstly, we offer an account of the role of social enterprise in health and social care, focusing on policies developed in the UK and the descriptive case studies used to justify these policies.Secondly, an overview of previous academic studies that explore the contribution of social enterprises in health and social care sector is provided.Third, we describe the methods employed for our systematic review.Ours is the first study to systematically analyse differences in outcomes depending on the collaborative or competitive contexts within which social enterprise operates and so, in our findings, we distinguish between studies treating social enterprise as a substitute for public provision, and those treating social enterprise as complementary.Our findings suggest that there is no evidence to support the role of social enterprise as a substitute for publicly owned services.However, there is evidence to show that where social enterprise operates in a collaborative environment, enhanced outcomes can be achieved, such as connectedness, well-being and self-confidence.Finally, we turn to the wider public management literature to help interpret our findings.
From complement to substitute: the evolving role of social enterprise in UK health policy The concept of social enterprise is highly contested, variously understood in different geographies, cultures and at different points in time (Teasdale 2012a).Scholars have traced two broad competing traditions of social enterprise: the approach most often seen in the United States, which sees social enterprise as commercially sustainable organizations competing against other providers to sell goods and/or deliver services (Dees 1998;Kerlin 2006); and a continental European approach, which sees social enterprises as 'hybrid' organizations deriving revenue from a range of sources and working collaboratively with governments to deliver goods and services within a mixed economy of welfare (Defourny and Nyssens 2010;Evers and Svetlik 1993).
The United Kingdom has borrowed from both of these conceptualizations in developing what is widely seen as the most developed institutional support structure for social enterprise in the world (Nicholls 2010;Teasdale 2012a).The definition of social enterprise originally developed by the UK Government's Department of Trade and Industry is sufficiently broad to capture both of these approaches: A social enterprise is a business with primarily social objectives, whose surpluses are principally reinvested for that purpose in the business or in the community, rather than being driven by the need to maximise profit for shareholders and owners.(Department of Trade and Industry 2002, 8) This loose definition has been flexibly interpreted in a variety of ways (Teasdale, Lyon, and Baldock 2013).For the purposes of the review conducted in this paper we have added the criteria that social enterprises should not distribute surpluses to external shareholders.This equation of social enterprise with the third sector is in line with the way in which social enterprise is usually understood in the United Kingdom and much of mainland Europe (Defourny, Hulgård, and Pestoff 2014), and excludes newer forms of social business which seek to marry private profit with social benefit (Teasdale, Lyon, and Baldock 2013).
Initially, policymakers saw social enterprise as a vehicle to regenerate deprived communities as just one player in a collaborative, pluralistic welfare system (Buckingham 2009;Teasdale 2012a).Over time, however, social enterprise has become the preferred mechanism for delivering a range of public services, most notably in health and social care. 1 The Social Enterprise Investment Fund was designed to increase the supply of social enterprises in the healthcare sector (Hall, Miller, and Millar 2012).The 'Right to Request' and 'Right to Provide' policies within the National Health Service (NHS) in England aimed to encourage staff in the publicly owned NHS to 'spin out' the existing services (Miller, Millar, and Hall 2012), and also facilitate the sub-contracting of healthcare provision to social enterprises.Social enterprises were increasingly promoted as alternative providers to publicly owned provision, ostensibly because they were seen as offering higher levels of innovation, cost-efficiency and responsiveness to societal needs (Allen 2009;Buckingham 2009;Hall, Miller, and Millar 2016).
As has been widely noted, the various social enterprise policies of the UK Government were not predicated on research evidence (Nicholls and Teasdale 2016).However, it has been common for policy documents to use 'successful' examples of social enterprises delivering public services to illustrate the rationale behind the policy trajectory as highlighted above.A full analysis of the wide range of social enterprises presented in various policy documents is beyond the scope of this paper, and so here we offer just three examples.The first social enterprise strategy Social Enterprise: A Strategy for Success (Department of Trade and Industry 2002, 23) identified the Bromley-by-Bow Centre as an exemplar of social enterprise which offers a 'variety of integrated projects linking health with education, training and family support'.Similarly, the Big Life Group, identified as an exemplar of social enterprise within the ambitiously titled subsequent strategy Social Enterprise: Scaling New Heights (Office of the Third Sector 2006), played an important role linking groups to healthcare services in South Manchester in the 2000s.The Little Angels example provided in the 2009 publication Enterprising Services, meanwhile, was described as providing breastfeeding services in deprived areas through supporting new mothers to access new and existing services, where the commissioning hospital 'created seamless links between NHS and Children's Services and those of Little Angels to create holistic integrated support' (Office of the Third Sector 2009, 27).We would draw attention to the fact that in each of these cases the emphasis is on the role that social enterprise can play in connecting the existing services within a collaborative health system, but which have then been used to justify social enterprise as a substitute to health and social care providers.Similarly, the focus on Social Return on Investment as a tool to measure the impact of social enterprise in Department of Health-funded programmes such as the Social Enterprise Investment Fund has tended to focus on the additional benefits that social enterprise can bring to an otherwise unchanged health system (Millar and Hall 2013), rather than attempting to account for services which would be lost if they were replaced by social enterprises (Alcock et al. 2012).In essence, the illustrative case studies used in the policy literature bear little relation to the types of organization that the current policy environment favours.Perhaps, however, the academic evidence can better help to justify these policy choices?
The evidence behind social enterprise as a deliverer of health services The limited academic literature on social enterprise and health has tended to focus on the (often indirect) well-being aspects of engagement with/employment in (nonhealthcare related) social enterprises (Roy et al. 2014;Roy, Baker, and Kerr 2017).Some research has focused on the role of social enterprises as a viable means of delivering 'co-produced' services in rural and remote communities (Farmer, Hill, and Munoz 2012), while other work has explored the processes involved in 'spinning off' parts of publicly funded and delivered healthcare systems into social enterprises (Addicott 2011;Miller, Millar, and Hall 2012), or evaluating the marketization of healthcare in the UK via social enterprise (Hall, Miller, and Millar 2012).
Only one systematic review has explored the role of social enterprise in a health context, and here the focus was on the health and well-being generated by social enterprises operating outside of the formal healthcare system.Roy et al. (2014) analysed the mechanisms and pathways through which social enterprise might act as a public health intervention.Their results suggested that social enterprises may have a positive impact on health and well-being, although only five studies were included in the final analysis, the majority of which focused on WISEs ('work integration social enterprises') which aim to provide employment for people disengaged or significantly disadvantaged from the mainstream labour market, such as people with intellectual disabilities.However, that paper did not aim to compare the outcomes of social enterprises with those of usual care providers; rather, it sought to understand the impact of social enterprise activity on health outcomes through acting on social determinants.Roy et al. (2014) concluded that social enterprise could enhance skills and employability, increasing self-reliance, reducing stigmatization, building social capital, and improving health behaviours.
Other systematic reviews have studied the effects of ownership on general health care providers (Berendes et al. 2011;Heins et al. 2010), dialysis centres (Devereaux et al. 2002) and nursing homes (Comondore et al. 2009;Hillmer et al. 2005;Xu, Kane, and Shamliyan 2013), and have included the wider Third Sector as one of the comparators.Some studies have concluded that non-profit facilities provide a significantly higher quality of staffing, and improvements in physical health in areas such as reduced infection rates and pain (e.g.pressure ulcers) (Comondore et al. 2009;Hillmer et al. 2005).However, other studies have found no 'consistent evidence that non-profits perform better than the private sector in either non-universal or universal health systems and [. ..] that in a competitive environment non-profit providers behave much like for-profit providers' (Heins et al. 2010, 523), in both developed and developing countries (Berendes et al. 2011).This conclusion has also been partially confirmed by more recent research based on case studies in health care (Heins and Bennett 2016) and is supportive of work highlighting the risk of marketization of non-profit organizations (Dart 2004;Eikenberry and Kluver 2004) particularly in health and social care systems (Aiken and Bode 2009;Eikenberry and Kluver 2004).
This brief discussion of previous studies reveals that no clear positive results can be detected, and thus it is not possible to support the policy focus on social enterprise as a preferred mechanism for delivering services.In part, this might derive from the lack of shared understanding as to what social enterprise is, and the wide variety of contexts within which the individual studies have taken place.It is notable for the purposes of this paper that no existing studies have analysed if there are differences in outcomes when social enterprise operates in a more collaborative environment alongside the existing public providers, as opposed to in a more competitive environment as a substitute for public providers.

Methods
Conducting a systematic literature review has been considered a robust form of research for addressing this issue because of the possibility of increasing generalizability of results and assessing consistency of evidence (Mulrow 1994).In health care research, this method is a useful vehicle for exploring the contribution of an intervention in terms of its effectiveness and relative costs (Donaldson, Mugford, and Vale 2002).Systematic reviews are widely used to inform policy-makers, practitioners and civil society (Chalmers, Hedges, and Cooper 2002).
The research question used for informing our study was: Do social enterprises provide improved outcomes in comparison with usual care in health and social care systems?
Due to the need to provide a clear definition for the purposes of a systematic review, we defined social enterprises broadly as non-profit distributing organizations that achieved (some of) their income through trading (in public or private-sector markets) in order to achieve their social purpose.To ensure the validity of the chosen methods and comprehensiveness of the study's search strategy, a protocol for the review was developed in collaboration with the research team and shared with two other experts in systematic reviews.PRISMA guidelines were used for the design of the research (Moher et al. 2009) and the protocol was registered on the International Prospective Register of Systematic Reviews (PROSPERO) under reference CRD42014009524. 2 To be included, studies had to meet the following criteria: • Primary research (any type of methods) evaluating and explaining the contribution in terms of efficacy and efficiency (i.e.quality of life, well-being and health) of social enterprise in comparison (either as a substitute or as a complement) to usual care providers in any publicly-funded health and social care.Cohort studies (one group pre-and post-evaluation) were also included as their baseline results can mimic usual care.However, conscious of this assumption, the weight given to them in the synthesis of the studies was lower than that of other quantitative methodology with better counterfactual evidence (i.e.quasi-experimental, RCT or case control studies).
• Only papers analysing social enterprises in publicly-funded health care systems or in sectors managed by the public part of non-publicly-funded health care systems were included.However, as we are conscious of the differences among the settings, the importance of contexts is outlined in the discussion session; • We also included articles that examine third sector organizations that meet our definition of social enterprise (above).Third sector organizations (and synonyms) were also included as keywords in the search strategy.However, where there was no evidence of the third sector organization 'trading' (either in the papers or subsequent follow up searches), then the papers in question were excluded. 3• Foundation Trusts (as well as dialysis centres and nursing homes) were excluded in line with the assumption that they form part of the system of 'usual care'. 4 Additional information about inclusion and exclusion criteria of studies is reported in Appendix 1.The following databases were searched: Sociological Abstracts, Web of Knowledge, ASSIA, International Bibliography of social sciences, Cochrane, DARE, NHS EED and HTA, Campbell, PUBMED, Ethos, OpenGrey, PLANEX and CINHAL. 5In addition, specific journals relating to social enterprise and non-profits 6 were screened.All the materials identified in the research were evaluated using a two-stage screening process with the inclusion criteria screened against titles and abstracts at the first stage.Of the 24,717 papers initially identified, 20% of the abstracts randomly selected after duplicates were removed were double screened to decrease possible biases.In all cases, disagreements regarding the inclusion or exclusion of a paper were resolved through group discussion.Relevant studies were retrieved and their full text screened by the lead researcher.However, any doubts about inclusion and exclusion based on the full text were discussed among the research team.An Excel database was created to record the reasons for exclusions at each stage.Figure 1 describes the flow of studies through the processes of, and screening for, inclusion.
The quality assessment of quantitative papers was undertaken using the EPHPP framework (Effective Public Health Practice Project), which has been recognized as a valid, reliable and easy to understand tool (Thomas et al. 2004).For qualitative papers, the framework by Mays and Pope (2000) was applied.This framework has been adopted by several prior systematic reviews in the health and social care field (Humphreys et al. 2007;Robinson and Spilsbury 2008), and so is considered a relevant and recognized tool.For mixed methods papers the tool developed by Pluye et al. (2009) was used.This tool is one of the most often used scoring systems for evaluating mixed methods research papers (Heyvaert et al. 2013;Tashakkori and Teddlie 2010).
Specific frameworks were used for extracting the data.The dimensions of these frameworks reflect the research methods used in the included studies, although the common categories used were study design, population, outcome and findings.Data were extracted by the lead researcher and discussed with the team in several meetings.Social enterprise is an emergent field with no standardized approach to measuring impact (Millar and Hall 2013) and so there was a significant degree of heterogeneity in the results both within and between different methods.Thematic analysis was brought to bear to identify common themes in the data (Barnett-Page and Thomas 2009).Although not initially an intention of the study, when analysing the papers we began to notice differences between results which appeared to correlate to the contextual setting.Therefore, we subsequently differentiated results according to whether the social enterprise activity occurred in a competitive or collaborative healthcare system.Our evaluation of context took place at the micro level of the study, rather than the country (or time-period) within which the study took place.We classified papers treating social enterprise as a substitute when the social enterprise is compared to the same service provided by an organization in the public sector.For example, Luo and Yu (1994) compared a Chinese work integration social enterprise directly with a similar service provided by the public sector.We classified papers treating social enterprise as complementary when the service they provided was additional to 'usual care'.Here, the impact of the intervention was measured through comparison with people receiving only 'usual care' (or pre-and post-intervention).For example, Higgins, McKevitt, and Wolfe (2005) compared patients in a hospital setting receiving reading services from a social enterprise with patients not receiving the service.

Findings
As Figure 1 shows, a total of 24,717 papers were initially identified, of which 771 were screened in full text.After excluding papers on the grounds of them not involving an analysis of social enterprise; not assessing any type of health outcomes; and/or not comparing intervention with usual care, 25 studies were eventually included.10 of the studies were conducted on social enterprise in the UK, four in Australia, four in Canada, three in the US, and one each in Bangladesh, China, Norway and Sweden.Seven papers studied the outcomes provided by social enterprise as a substitute to usual care, acting in a more competitive healthcare system, while 18 papers analysed the contribution of social enterprise as a complement to usual care providers, in a more collaborative setting.The heterogeneity of the different studies precluded the use of aggregative synthesis tools such as statistical meta-analysis.Our findings and discussion sections utilize a configurative approach (see Gough, Thomas, and Oliver 2012)in this case narrative synthesisto enable exploration of how the effects on health and well-being of beneficiaries might change in different (policy and/or health) contexts.Our discussion section also highlights some of the difficulties in conducting systematic reviews in relatively broad policy areas such as this, and what we learned from the process.

Social enterprise in a competitive setting
The seven studies in which social enterprise was a substitute to mainstream providers focused on a range of different interventions, and utilized a wide variety of outcome measures.Three of the seven identified studies compared outcomes among providers in different healthcare settings (Allen et al. 2012;Knapp et al. 1999;Shepherd et al. 1996).Of these, just one explicitly analysed differences between social enterprise and publicly owned provision, finding that social enterprises do not provide more innovative or more effective care than usual care providers (Allen et al. 2012).However, they can more readily involve stakeholders in decision making.One study focused on complex interventions that support healthcare delivery through social and physical activities (Edwards et al. 2009).One paper analysed a parenting programme which aimed to reduce behavioural problems in children (Gardner, Burton, and Klimes 2006); one paper analysed the role of work integration social enterprise in reducing mental illness (Luo and Yu 1994) and one paper compared nutrition intervention programmes in developing countries (Khan and Ahmed 2003).Sample sizes differed across papers, partly due to study design, and are reported in Table 1.
Seven themes were identified through the thematic analysis of papers where social enterprises operated in a competitive setting: physical health; mental health; wellbeing; connectedness within the community; confidence and empowerment; cost differences; and medication and health service utilization.
Improved physical health was highlighted as one of the possible contributions made by social enterprises in two of the studies (Edwards et al. 2009;Knapp et al. 1999).However both studies failed to identify a clear and long-term improvement trend.Knapp et al. (1999) showed a significant improvement in mobility and incontinence along with a slight significant deterioration in the total disability score in comparison to public and private sector providers; while Edwards et al. (2009) found a non-statistically-significant difference in healing ulcers at 24 weeks, a significant decrease in ulcer area and a significant decrease in pain experienced by people participating in the organization.
Mental health outcomes were explored in three papers.Specific measures related to depression (Edwards et al. 2009;Gardner, Burton, and Klimes 2006) and mental health clinical outcomes (Luo and Yu 1994) were analysed.Mental health clinical outcomes showed a significant improvement (Luo and Yu 1994) while depression scales did not show any significant differences (Edwards et al. 2009;Gardner, Burton, and Klimes 2006).
Three papers focused on social determinants of health and well-being, including social connectedness and self-confidence and a clearer trend was detected with those.Specific measures used related to behavioural change (Gardner, Burton, and Klimes 2006), and quality of life and well-being (Edwards et al. 2009;Shepherd et al. 1996) were explored.Two of the papers presented positive significant results in terms of quality of life, activities daily living score and changing behaviour (Edwards et al.  (Shepherd et al. 1996).In the two studies where two follow-up analyses were provided, the long-term effect of the intervention was confirmed (Edwards et al. 2009;Gardner, Burton, and Klimes 2006).One paper explored involvement with, and connectedness in, the community, in terms of interactions between people (Edwards et al. 2009).However, no significant impact was found.Two papers explored outcomes in terms of self-confidence and empowerment (Edwards et al. 2009;Gardner, Burton, and Klimes 2006).Edwards et al. (2009) found that the social enterprise mode of delivery had advantages as measured on the Rosenberg Self-Esteem scores in comparison with usual care providers, while Gardner, Burton, and Klimes (2006) highlighted a significant increase in the confidence of to play independently.Three of the papers considered differences in costs (Khan and Ahmed 2003;Knapp et al. 1999;Shepherd et al. 1996).Knapp et al. (1999) found that social enterprise had a higher cost than for-profit organizations, but a lower cost than the public sector.Khan and Ahmed (2003) affirmed that social enterprise did not appear to be more efficient than the public sector after evaluating the cost per participant.However, in the same paper, the authors claimed that the lower costs of the public sector had been achieved by not enrolling all the people who were in need of the programme.Shepherd et al. (1996) highlighted the difficulties in collecting data related to costs, and they did not present any results.
Finally, one paper explored the reduction in the use of medications and health service utilization (Luo and Yu 1994), highlighting the presence of a lower range use of medication use among the beneficiaries of social enterprise, and a significant lower relapse rate, or rate of readmission to hospitals.However, the authors recognized that it was not possible based on the data they had collected to relate the lower use of medication directly to the intervention.

Social enterprise in a collaborative setting
Studies which focused on social enterprises in a more collaborative setting also analysed a wide range of interventions utilizing a wide range of outcome measures.12 of the 18 studies identified focused on complex interventions supporting healthcare delivery through physical activities, lunch clubs, arts promotion, reading services and information and advice for patients with chronic illnesses (Cheifetz et al. 2014;Ewles et al. 2001;Finn, Bishop, and Sparrow 2007;Greenspan and Handy 2008;Higgins, McKevitt, and Wolfe 2005;Jenkins, Brigden, and King 2013;Jones et al. 2013;Keller et al. 2004;Kirrmann 2010;Leroux 2005;Stathi and Sebire 2011;Zabriskie, Lundberg, and Groff 2005).Five papers analysed the impact of work integration social enterprises on the mental health of beneficiaries (Ferguson 2012;Jackson et al. 2009;Lysaght, Jakobsen, and Granhaug 2012;Norman 2006;Williams, Fossey, and Harvey 2012).One paper analysed the outcomes of a housing and information programme for homeless beneficiaries (Hawk and Davis 2012).The sample sizes of these studies varied according to the study design and they are reported alongside interventions and study design in Table 2.This time, six themes emerged from the thematic analysis of the work of social enterprises operating in a collaborative setting: physical health; mental health; well- being; connectedness within the community; confidence and empowerment and medication and health service utilization.Three papers highlighted significant improvements in physical activity and motor performance between baseline and first follow-up (Cheifetz et al. 2014;Jones et al. 2013;Leroux 2005), while two studies detected a contribution of the social enterprise in improving beneficiaries' attitudes towards physical activity (Ewles et al. 2001;Stathi and Sebire 2011).However, the low retention rate of the participants in the long-term in one of the papers (Leroux 2005) and a lack of improvement in some of the physical activity scales in another (Cheifetz et al. 2014) occurred at follow-up.One paper drew attention to speech improvement in people with strokes (Jenkins, Brigden, and King 2013) and one paper identified a possible impact in terms of changed clinical outcomes or the severity of illness (Hawk and Davis 2012).Jenkins, Brigden, and King (2013) stated that a percentage (the exact number was not declared) of patients who had received the services of the organization identified a speech improvement.However, the lack of clear data prevented the drawing of any conclusions about physical health outcomes.Hawk and Davis (2012) found that participation with the organization had a possibly significant relation with a decreased HIV viral load.However, the lack of an external control group means that their results cannot be directly attributed to the social enterprise.
Three papers explored a possible impact on mental health.In Ferguson (2012), a (non-significant) decrease in depressive symptoms compared with no change in the control group was observed.There was a significant improvement in depression in the intervention group between the baseline and the follow-up.A significant increase in psychological well-being was detected by Finn, Bishop, and Sparrow (2007), and a positive reduction of mental ill health was found by Jones et al. (2013).
Several studies focused on social determinants of health and well-being, such as connectedness and self-confidence, through measures such as the effect on total life satisfaction (Ferguson 2012), emotional well-being (Williams, Fossey, and Harvey 2012), and general well-being (Cheifetz et al. 2014;Finn, Bishop, and Sparrow 2007;Jones et al. 2013;Keller et al. 2004).Ferguson (2012) reported a significantly higher improvement in total life satisfaction in the intervention group from the baseline to the end of the follow-up.Finn, Bishop, and Sparrow (2007) found a significant improvement in well-being factors while Williams, Fossey, and Harvey (2012) and Keller et al. (2004) reported improvements in emotional and general well-being from qualitative interviews with beneficiaries.However, the lack of significance with regard to the Global Well-Being scores and Social Well-Being Subscale (Cheifetz et al. 2014) and the lack of a comparison group in almost all of the studies (Cheifetz et al. 2014;Finn, Bishop, and Sparrow 2007;Jones et al. 2013;Keller et al. 2004;Williams, Fossey, and Harvey 2012) meant that it was not possible to attribute the results solely to the social enterprise interventions.Qualitative papers identified the ability of social enterprises to create mechanisms respectively addressing patients' well-being needs, recovery and quality of life (Greenspan and Handy 2008;Higgins, McKevitt, and Wolfe 2005;Kirrmann 2010); increasing their motivation and adaptation to specific situations (Higgins, McKevitt, and Wolfe 2005) and supporting a sense of vitality and relaxation (Ewles et al. 2001;Stathi and Sebire 2011).
13 papers analysed the contribution of social enterprise to increased connectedness in the community, showing the possibilities of supporting reintegration (Jenkins, Brigden, and King 2013;Zabriskie, Lundberg, and Groff 2005), increasing family and peer support (Ferguson 2012) and social engagement (Jones et al. 2013).Ferguson (2012) observed a significant increase in family support in the intervention group versus a decrease among her control group.Moreover, the paper identified a borderline-significant difference in peer support.Jenkins, Brigden, and King (2013) reported an improvement in social contact, although the lack of a control group means no causal relationship can be identified.No significant impact was found by Jones et al. (2013) in terms of social engagement and by Zabriskie, Lundberg, and Groff (2005) in terms of community involvement.Reintegration into the community was explored in qualitative papers, identifying social enterprise as leading to increasing interactions with other people (Finn, Bishop, and Sparrow 2007;Higgins, McKevitt, and Wolfe 2005;Stathi and Sebire 2011), fostering social and support networks (Ewles et al. 2001), reconnecting people with life expectations (Kirrmann 2010), and feelings of inclusion and engagement within the community (Greenspan and Handy 2008;Keller et al. 2004;Lysaght, Jakobsen, and Granhaug 2012;Norman 2006).
Four papers considered the effect of social enterprise on the self-confidence and self-esteem of the beneficiaries (Ferguson 2012;Finn, Bishop, and Sparrow 2007;Jenkins, Brigden, and King 2013;Keller et al. 2004).Ferguson (2012) explored this topic both in the qualitative and quantitative parts of her paper.In the quantitative analysis, the effect on self-esteem was not found to be statistically significant.However, in her qualitative analysis, social enterprise appeared to increase beneficiaries' overall sense of self-worth.This finding was confirmed by other studies that explored sense of self-worth, confidence and feelings of empowerment (Ewles et al. 2001;Finn, Bishop, and Sparrow 2007;Jenkins, Brigden, and King 2013;Keller et al. 2004;Kirrmann 2010;Lysaght, Jakobsen, and Granhaug 2012;Stathi and Sebire 2011).
Finally, three papers considered the effect of social enterprise-led activity on medication and hospitalization (Finn, Bishop, and Sparrow 2007;Jackson et al. 2009;Kirrmann 2010).A moderate association between a reduction in the use of mainstream mental health services/medication and the extent of involvement in the organization was explored by Finn, Bishop, and Sparrow (2007) while Kirrmann (2010) identified the perception of decreased medication as a possible outcome.Jackson et al. (2009) showed that social enterprise beneficiaries had significantly fewer emergency department, ambulatory and hospitals visits.However, no significant difference was found in relation to the average duration of their hospital stays.

Discussion
Before discussing the results, it is important to highlight some of the limitations of a systematic review of this type.Systematic reviews are probably best suited to relatively homogenous forms of public policy intervention, ideally with clearly defined outcome measurements.As highlighted in the introductory sections of this paper, social enterprise is a fluid concept capturing a wide range of organizational forms.Understandings of social enterprise vary across countries where the term is constructed and conceptualized differently (Teasdale 2012a).We attempted to address this to some extent through imposing specific social enterprise criteria for studies to be included (essentially non-profit distributing organizations engaged in trading).
Even so, our definition would be contested by some commentators as too narrow (as it excludes some profit distributing forms of social enterprise), and by others as too broad, as it does not insist on criteria such as democratic ownership and/or governance (see Defourny and Nyssens 2010).Additionally, the health impacts of the different forms of social enterprise are wide ranging, thus making direct comparison problematic.No two papers explored the same research questions.Studies utilized different methodological approaches.Our review examined evidence across very different social and health care systems situated within very different public policy contexts.Studies were undertaken at different points in time.Healthcare policy and funding frames have undergone rapid transformation, particularly in the UK and US, in recent years.Thus, even if our findings had demonstrated conclusively the health benefits of social enterprise (whether in a collaborative or competitive setting), it would be problematic to generalize to all social enterprises, health or policy contexts.
Eighteen of the studies analysed social enterprises acting within a collaborative health system.Only seven studies treated social enterprise as a substitute to usual care.Thus, our paper offers support to those who argue that policy support for social enterprise in health and social care is based on limited evidence, particularly regarding the replacement of publicly owned services.However, we were more concerned in this paper with exploring whether we can identify differences when social enterprise operates in a collaborative rather than competitive environment.A synthesis of the outcomes and findings is provided in Table 3.
Regarding clinical outcomes (physical and mental health), there is no clear evidence showing that delivery by a social enterprise has demonstrable long-term benefits (whether in a more collaborative or competitive context).Both improvements and deteriorations in physical and mental health outcomes were explored in papers where social enterprise was used as a substitute to usual care.Slightly more positive results were presented in papers that analysed social enterprises acting as complement to usual care, particularly regarding improvements in physical activity and decreases in depressive symptoms.While the lack of long-term evidence and comparison groups in some of the papers further downplay the small number of studies reporting positive results, it would certainly seem worth conducting further research into how social enterprises might be better utilized within collaborative health care systems to enhance physical activity, and reduce depression.
Only a very small number of studies explored differences in costs (or savings) between social enterprise and usual care providers.Understandably, all these studies treated social enterprise as a substitute to public provision, and results were unclear.However, given that one of the claims behind social enterprise is that it is more efficient than public provision (Bovaird 2014), it would certainly seem worth conducting more economic analyses.Similarly, given that treating social enterprise as a complement to usual care would necessarily have cost implications, it is important to conduct research which begins to unpack the cost-benefit implications of introducing additional social enterprise provision to the welfare mix.
It is notable that the clearest positive outcomes occurred among measures such as well-being, connectedness, confidence and empowerment, and that these outcomes were more obvious when social enterprise operated in a more collaborative setting.Here, several papers showed that social enterprise produces positive results in terms of interactions with communities and families, the inclusion of beneficiaries, feelings of engagement, perceptions of social support and increased sense of self-worth.Despite a lack of statistically significant findings in quantitative papers and the lack of comparison groups in some studies, a clear positive trend in some papers, particularly qualitative, with some longer-term results, can be assumed.When social enterprise is treated as a substitute to usual care, only two papers showed positive results.Only one paper analysed outcomes related to connectedness, without showing positive results.While the number of papers is small, it might be worth exploring whether social enterprises operating in a more collaborative environment are better able to promote connectedness among beneficiaries as a consequence of not competing for funds with organizations and groups they are connecting them to.The benefits of social enterprise (whether in a collaborative or competitive environment) would therefore appear to centre most clearly on well-being, connectedness, confidence and empowerment, rather than on those more clinical measures which would be the primary justification for utilizing social enterprise as a health provider.Where positive clinical outcomes were observed, these were more obvious in a collaborative setting, and focused on measures that clearly relate to well-being, such as a reduction in depression, or enhanced physical activity.The wider academic literature helps us interpret these findings.

Conclusion
Our original research question sought to ascertain whether social enterprises provide improved outcomes in comparison with usual care in health and social care systems.
Additionally we sought to understand the importance of context through distinguishing between studies according to whether the social enterprise activity occurred in a competitive or collaborative healthcare system.With reference to our original research question there is insufficient evidence to provide conclusive answers.In part this is because of the heterogeneity of organizational types included within the label social enterprise, the wide variety of health impacts that different studies have focused upon, and the variety of contexts in which these social enterprises operate.
Our study does suggest, however, that social enterprises may lead to improved health outcomes in some circumstances, particularly as regards well-being and mental health.A particular ethos within certain types of social enterprise may help them act as a positive mechanism for increasing self-reliance, building social capital and improving health behaviours (viz.Roy et al. 2014).Social enterprises, when integrated into the existing service provision for disadvantaged groups, and relatively free to focus on social benefits rather than intense competition for funding, can act as a space for improvement of well-being and quality of life (Farmer et al. 2016;), hence reducing social isolation and social exclusion (Teasdale 2010).It is more difficult to conceptualize how organizational type would have a bearing on some other types of clinical activity captured within this review, for example in healing ulcers.Institutional theory would suggest that, over time, organizations tend to mimic the behaviour of competitors (or indeed collaborators) in the same field (Nicholls and Teasdale 2016).Professional codes, procedural rules, legitimacy, and income-based reward systems may outweigh organizational form, particularly for those (larger) contracts where local innovation or discretion may be low.Hence, organizational form may have less of an impact on outcomes than size (Taylor, Langan, and Hoggett 1994).Together, the above would help explain why there is no clear pattern in clinical health improvements for social enterprises operating in a competitive environment.
Our findings do offer some support to those highlighting the importance of context, and particularly the collaborative or competitive environments within which social enterprises operate.Well over a decade ago Osborne and McLaughlin (2004) identified that treating third sector organizations as service agents risked the distinctive competencies of the sector, and their distinctive contribution to public services.While competitive tendering can improve service quality among third sector organizations, our findings suggest that service quality, at least in relation to health outcomes, cannot be demonstrated to be measurably better than when delivered by a social enterprise than by public provision.The increased pressures of operating in a resource-constrained environment where financial considerations must override a commitment to clients can lead social enterprise providers to focus only on those services (and clients) that are profitable (Dart 2004) to the extent that the services they provide are no different than those offered by other providers.As Buckingham (2009) shows, marketization leads social enterprises to behave differently: not simply through reducing trust in other organizations, but also through wanting to keep all the 'income' from their client, or pass on as much of the costs of treatment to other organizations (Teasdale 2012b).
Where provision via a social enterprise does have positive health outcomes, it would seem that a collaborative environment facilitates (or even enables) these benefits to wellbeing and mental health.While conducted primarily to inform UK policy, our paper has relevance to all countries engaging social enterprise in public service delivery, as it suggests that this is best undertaken in more collaborative public policy environments.Future research studies might usefully attempt to control for environmental factors in assessing the impact of social enterprise compared with other organizational types on well-being on mental health, perhaps through 'natural experiments'.Indeed, the enhanced social connectedness offered by social enterprises was observed as occurring only in a collaborative setting.Collaboration between providers within healthcare systems on the ground means that professionals can direct beneficiaries/clients/patients to the most appropriate service.As Bovaird (2014) highlighted, collaborations between third sector organizations and between networks of third sector organizations and providers in the public sector can theoretically lead to 'economies of scope', particularly where strong trust-based relations between third sector organizations and communities can harness self-help and the co-production of services.As our study suggests, evaluating the benefits of these economies of scope is difficult, and any benefits are usually ignored by evaluative (and commissioning) frameworks anyway (Smith 2013).However, research suggests that it is through these economies of scope that the potential well-being benefits of social enterprise may be realized.More exploratory work is necessary to conceptualize how wider economies of scope are generated and to develop tools to evaluate their benefits.However this, of course, assumes a rational approach to evidence-based policymaking that is not apparent in the UK, where moves to privatization since the 2012 Health and Social Care Act and subsequent Sustainability and Transformation Plans (Alderwick and Ham 2017) arguably insert financial and ideological imperatives over and above such outcomes.Notes 1. See Nicholls and Teasdale (2016) and Sepulveda (2015) for a more comprehensive overview of social enterprise policies of the UK Government. 2. See https://www.crd.york.ac.uk/prospero/display_record.php?RecordID=9524 .3. The papers did not always highlight the presence of trading in the organizations analysed.
Therefore, an analysis of each organization's website was also undertaken to explore if the organization was involved in trading.Papers were excluded when neither the paper nor the website outlined any trading activities within the organization.4.Although there is no standard definition of 'usual care', for the purposes of this Review we define this as the routine care usually received by patients for prevention or treatment of diseases in a given context.5.The full search strategy is available from the corresponding author on request.6. Searches were conducted in: Voluntas, Nonprofit and Voluntary Sector Quarterly, Social Enterprise Journal and Journal of Social Entrepreneurship.

Disclosure statement
No potential conflict of interest was reported by the authors. (Continued).
Criteria Inclusion Exclusion Study and Design Comparative studies of any design (qualitative and quantitative, experimental, semi-experimental) examining any type of performance (efficacy, efficiency or both) in at least two population groups which include social enterprises or third sector organizations are included.Moreover, all systematic reviews related to the topic will be evaluated in order to obtain their reference lists and identify possible papers.
Studies which do not compare at least two categories will be excluded.Opinion papers will not be included.Studies developed in languages other than English will not be included.

Figure 1 .
Figure 1.Results of the search and study selection process.

Table 1 .
Social enterprise as a substitute to usual care: papers included.Burton, and Klimes 2006) while one paper did not show any significant evidence in terms of the Global Well-Being scale

Table 2 .
Social enterprise as a complement to usual care: papers included.

Table 3 .
Social enterprise results.