Our search identified 16,422 papers, after duplicates were removed. Thirty-six reviews were included in our review (Fig. 1) [4, 23,24,25,26,27,28,29,30,31, 34, 61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85]. Studies that were reviewed in full but did not meet our inclusion criteria were focused on describing collaboration models or theories [86,87,88,89,90,91,92,93,94,95,96,97,98,99,100], evidence related to service-level interventions [101,102,103,104,105,106,107,108,109,110,111,112,113,114,115,116,117,118,119,120,121,122] or inter-professional collaboration [123,124,125,126,127] without a clear focus on organizational collaboration, collaboration within the health care system [128,129,130,131,132], or collaboration primarily at national or international levels [133, 134]. Other studies were excluded because they were not a review of empirical evidence or the phenomenon reviewed was unclear [135,136,137,138,139,140,141,142,143,144,145,146,147,148,149,150,151], or they repeated or were superseded by another study from the same authors [152,153,154,155,156,157]. One study could not be obtained [158].
Collaboration type and context
Table S1 (see Additional File, Table S1) describes the context and type of collaborations in the included studies. The studies reviewed collaborations from a range of contexts. Some studies focused on collaborations in a single country, such as the UK [4, 23, 61, 63,64,65, 80] or US [28, 72, 79]. The majority of studies included evidence on collaborations from multiple countries and contexts (or did not define the country contexts of the studies reviewed). The definitions of collaboration used in the studies varied widely, as did the types of organizations involved and the aims of the collaborations.
Some studies reviewed collaborations with broad population health goals, such as preventing disease and reducing health inequalities [4, 26,27,28,29,30,31, 34, 61, 66, 76, 79]. These collaborations often involved health care and social services organizations, public health agencies, and other sectors including housing and education, and more. Other studies reviewed evidence on collaborations with a narrower scope or focus, such as integration between health care and social services [23,24,25, 62, 63, 68, 80, 84], or care for people with mental health needs [67, 73, 78].
Even within single studies, multiple kinds of collaboration were typically studied together, and interventions were often weakly described. Many reviews combined evidence on collaboration at an organizational-level (such as joint planning or funding of services) with more targeted strategies or interventions that resulted from organizational collaboration (such as care coordination programs for target populations). Evidence related to the impact and functioning of organizational partnerships was therefore hard to disentangle from evidence on related policies and interventions—for example, evidence on specific service-level changes.
Quality of evidence
Overall, the quality of evidence reviewed was weak (see Additional File, Table S2). The methods of the studies varied, including umbrella reviews, systematic reviews and meta-analyses, scoping reviews, and narrative reviews. Of the reviews that reported data on collaboration impacts, most were deemed to be critically low quality and only three reviews were deemed to be high quality. Of the reviews that only reported data on factors influencing collaboration functioning, most had multiple sources of potential bias—including weak search strategies and limited approaches to assessing and reporting risk of bias in the studies they reviewed. We have not excluded studies from our narrative synthesis based on the quality of the reviews, but we do note limitations or uncertainty in the evidence presented.
Impacts of collaboration
Twenty-one studies reported on collaboration impacts at a mix of individual and population levels [4, 23,24,25,26, 28, 30, 31, 62, 63, 67, 69, 71,72,73,74, 77, 78, 81, 82, 84]. They included evidence on health outcomes and health-related behaviours, service access and quality, resource use and spending, and organization or system-level processes related to collaboration. Evidence from the studies on collaboration impacts is summarized in Table S3 (see Additional File, Table S3).
Health outcomes
Most studies assessing the impact of collaboration on health outcomes, such as quality of life, mortality or health equity, found no, mixed, or limited evidence of impact. A review and meta-analysis of collaboration between local health and non-health agencies for health improvement found little or no evidence of health benefits [26]. Meta-analysis of effects on mortality, for example, found no effect (relative risk = 1.04 in favour of control, 95% CI 0.92 to 1.17) (see Additional File, Table S3 for effects on morbidity) [26]. A review of public sector collaborations to improve health in targeted communities in England found no evidence of population health improvements [4]. A review of community-level interventions to improve health in the US found insufficient evidence related to population health impacts—though it did find that these interventions could contribute to positive changes in health-related behaviours, such as smoking [28]. Ndumbe-Eyoh and Moffat found mixed impacts on health outcomes and limited evidence on equity impacts of collaboration to improve health for disadvantaged groups [31].
Mason et al. reviewed integrated funding initiatives between health care and social services agencies and found that, in the studies assessing health effects (such as quality of life and mortality), most reported no significant difference compared with usual care [25]. Cameron’s et al’s review of collaboration between health care and social services agencies found no or marginal improvements in health outcomes in studies with comparative designs (some studies with weaker designs reported improvements) [23]. Winters et al. found that most studies did not report positive outcomes (though did not define these outcomes clearly) [24]. And Liljas et al’s review of collaboration to provide more integrated care for older people with multimorbidity found that no studies examining mortality effects reported significant changes in mortality rates [82]. Five reviews found that evidence on health outcomes was limited [63, 67, 69, 74, 84].
Evidence of impact from some kinds of collaborations was more promising. Anderson et al. reviewed evidence on community coalitions to reduce health inequalities among minority groups and concluded that community coalition-driven interventions could benefit minority populations [30]. Community-level system changes—for example, focused on improving housing or green spaces—had little or no impact on measures of health status or health behaviour. But interventions targeting changes in the health and social care system—for example, to improve quality of care—led to small improvements on measures of health status or behaviour in large samples of community residents (though the evidence was rated as very low certainty). More positive impacts were also reported from lay community health worker and group-based health education interventions in large samples of community residents (though, again, the evidence was rated low or very low certainty). Bagnall et al’s review of systems approaches to reducing obesity found that most studies reported some positive effects, including on health-related behaviours and body mass index [81]. One review of collaboration between health and a range of non-health sectors (such education and housing) to prevent and control vector-borne disease also found positive effects in the majority of studies that measured outcomes, including incidence and prevalence of disease [77].
Lopez-Carmen et al’s review of collaboration to improve mental health among indigenous children found few quality evaluations to draw on, but identified some studies reporting positive outcomes among children and their families receiving particular interventions [73]. Similarly, a review of primary care and public health collaboration described weaknesses in the evidence but reported some positive outcomes at an individual and population level, related to chronic disease management, disease control, and maternal child health [71].
Service use and quality
Several reviews reported evidence that collaboration in a mix of contexts could improve access to services [31, 62, 71, 73, 77], including for disadvantaged groups [31, 73]. Cooper et al’s review of collaboration in children and young people’s mental health services, however, found more mixed evidence—with some studies suggesting more equitable access and others reporting reductions in access [67]. Some models of financial integration between health care and social service agencies may also have the unintended effect of reducing access for some groups (for example, by creating financial incentives to deny access to more costly patients) [25].
A review of integrated care interventions (including a mix of organizational and service level changes to improve coordination of services) found inconsistent evidence on overall health care utilization and activity, as well as on a range of specific utilization measures (such as clinician contacts and length of stay) [62]. Cameron et al’s review of health and social care collaboration found some evidence that intermediate care could reduce inappropriate admissions to institutional care [23]. Liljas et al’s review of health and social care collaboration found mixed evidence on hospital admissions, readmissions, and length of stay [82].
Evidence of impact on quality of services was mixed. Five reviews reported mixed impacts on quality of care [25, 67, 84] and patient satisfaction [82]. Dowling et al’s review of health and social care partnerships in England found no clear or consistent evidence of improvements in services [63]. Baxter et al. found evidence of improvements in patient satisfaction and perceived quality of care related to integrated care interventions [62]. A review of primary care and public health collaboration also reported some improvements in quality of care [71].
Reviews of some targeted collaboration interventions reported positive impacts. For example, a review and meta-analysis of collaboration to improve child welfare outcomes in the US found that family drug treatment courts were positively associated with entry (odds ratio = 2.94, 95% CI 1.50 to 5.75) and completion (odds ratio = 2.07, 95% CI 1.26 to 3.41) of substance use services, and that family drug treatment courts (odds ratio = 2.40, 95% CI 1.75 to 3.29) and recovery coaches (odds ratio = 1.52, 95% CI 1.17 to 1.99) were associated with increased likelihood of children being reunited with families [72]. Whiteford et al. found that collaboration between mental health care agencies and non-medical supports could increase housing stability, reduce recidivism rates, and improve employment-related outcomes, though also noted that positive outcomes had not been reported in all programs reviewed [78]. The reviews of these interventions, however, were of critically low quality (see Additional File, Table S2).
Resource use and spending
Evidence on the resource use and spending impacts of collaboration was limited and mixed. Hayes et al’s review of collaborations between health and non-health organizations found some evidence of increased costs, and—even though economic data were not provided for all studies—‘it was clear that in most studies the interventions required additional resources’ [26].
Two reviews focused on collaborations to deliver more integrated health and care services found inconsistent evidence on costs [62], or weak evidence that did not generally report cost reductions [23]. A review of collaboration between mental health care and non-medical supports found that while some studies reported ‘improved cost efficiency across sectors’, there were also examples of initiatives where costs fell for some partners but increased for others [78].
A review of integrated funding initiatives between health and social care organizations [25] reported impacts on hospital costs and utilization together. It found that most schemes reported mixed (14 schemes) or unclear evidence (5 schemes), around a third of studies found no significant effect on hospital costs or utilization (11 schemes), three schemes reported a reduction in hospital costs or utilization, and one scheme reported increased utilization. Other studies sought evidence on the cost impacts of collaboration but found little data [30, 77, 81, 84].
Process impacts
Some reviews reported qualitative evidence on changes in organizational or system-level processes as impacts of collaboration. These included improved collaboration processes and organizational capacity [73, 78], stakeholder buy-in [74], implementation of policies and programs related to health promotion [28], and an increased focus on health inequalities in local plans [4]. Potential negative impacts included increased time commitment and challenges to professional identities [67]. There were also some unintended consequences of financial incentives related collaboration, such as ‘upcoding’ (of ‘nursing home certifiable’ patients in a US demonstration program) [25], cost shifting [78], and skewing local priorities [71]. More broadly, qualitative evidence on factors shaping collaboration functioning—outlined in the following section—describe various processes that can support or constrain joint working.
Factors influencing collaboration functioning
Twenty-nine studies reported on factors shaping the success and functioning of organizational collaboration (see Additional File, Table S4). This included factors related to collaboration aims and motivation, resources and capabilities, cultures and relationships, governance and leadership, and external context (Fig. 2). The factors overlap and interrelate, and sometimes come into conflict.
Factors influencing collaboration functioning and example interactions between them
Notes: The interactions between factors are examples identified in the studies reviewed. They are not an exhaustive list of all interactions between the factors identified. The relationships may move in both directions (eg involving staff may help create a shared vision, while having a shared vision may help with the task of engaging other partners), and may support or constrain collaboration in different contexts (eg national policies can help or hinder)
Motivation and purpose
Organizations collaborated for different reasons, which shaped how they worked together. A shared vision between local organizations and clear aims for collaboration were commonly identified as factors contributing to partnership success [23, 27,28,29, 34, 63, 64, 66, 69, 74, 76,77,78, 80, 81, 83, 85]. Involving organizational staff was identified as one route to creating a shared vision [23, 28, 66, 76]; and a shared vision, in turn, may help with the task of engaging other partners [28]. On the flipside, unclear or unrealistic aims, competing agendas, and uncertain benefits were all identified as factors that can hold back organizational collaboration [23, 61, 64, 67,68,69, 71, 76, 77].
National policies supported local partnerships to emerge in various contexts studied (see section on external context) [27, 71, 76, 80, 83, 85]—though some studies also noted that national government policies mandating local collaboration may reflect an underlying lack of motivation for joint working among local agencies, and could create conditions for future conflict [65, 85]. Commitment to collaboration from local leaders and staff was commonly thought to be needed for partnerships to work effectively [28, 29, 34, 66,67,68,69,70, 77, 79].
Relationships and cultures
Multiple studies described how collaboration was more likely to be successful if partners trust each other [23, 63, 67, 68, 70, 71, 81, 84, 85] and have positive relationships [23, 25, 29, 64, 67, 73, 74, 77, 79, 81, 83]. For example, Davies et al. identified lack of trust between health care staff and care homes as a barrier to integrated working [84]. Historic relationships between agencies—present or absent; good or bad—shaped how local partnerships developed and functioned [23, 24, 28, 34, 66, 68, 69, 74, 77, 85].
Relationships were also affected by cultural and professional differences between agencies and staff within them—often identified as barriers to collaboration [23, 25, 65, 67,68,69, 76,77,78, 85]. In some cases, shared values could bring local agencies together—for example, united by a commitment to good governance or reducing health inequalities [71, 81]. But differences in values could also fundamentally undermine collaboration efforts. Williams, for example, found that philosophical differences between health and social care and criminal justice agencies—between ‘care and control’ sectors—contributed to various structural and procedural challenges experienced among crime prevention and reduction partnerships in England [65].
Clarity on roles and responsibilities of different agencies was thought to help collaborations make decisions, implement programs, and function effectively [23, 24, 66, 67, 69, 71, 74,75,76, 78, 85]. Lack of clarity could lead to protectionism, concerns about loss of power, and underuse of particular skills or services within the partnership [23, 64, 75]. For example, Green et al. described how lack of understanding of aboriginal health workers among public service agencies contributed to their underutilization within partnerships to improve care for indigenous children [75]. Developing clear frameworks and processes for collaboration [23, 29, 64] and joint training for staff between agencies [69] (see section on resources and capabilities) were both identified as mechanisms that could help improve clarity on organizational roles. Yet role clarity may not be needed for all kinds of partnerships, or at all levels within them. Corbin et al. noted that flexibility on roles may help partnerships be more inclusive and garner increased resources [34]. And, at a service level—for example, for staff delivering programs within the partnership—flexibility may be needed to support multidisciplinary teams to function [23].
How and when partners communicate was widely thought to affect how collaborations work [23, 24, 27, 29, 34, 61, 65,66,67,68,69,70,71, 73,74,75,76,77,78,79, 83, 85]. The simple interpretation from the literature is that good communication helps, while poor communication makes things harder. Good communication was thought to be open and frequent (though partners may disagree about how and when communication should happen [34]) [24, 66, 71, 74, 79, 85], and involve sharing of information and best practice [27, 61, 73, 83]. Various mechanisms were identified to help agencies do this (see section on resources and capabilities), such as regularly scheduled meetings and protocols for information sharing [24, 74]. But communication issues within the partnerships studied were widespread, exacerbated by lacking or incompatible information systems, conflicting procedures, lack of trust between organizations and professions, and more [23, 61, 65, 67,68,69, 78]. The quality of communication was thought to affect various other factors shaping collaboration success, such as trust and understanding between partners [24, 66, 70, 71, 75].
Resources and capabilities
Organizational collaborations depended on having sufficient resources to fund and deliver interventions, such as new service models or programs [28, 29, 34, 66, 67, 69,70,71, 73, 74, 76, 80, 81, 83, 85]. Lack of resources for joint working—funding, staff, equipment—was identified as a common barrier to collaboration [23, 34, 61, 67,68,69, 71, 73, 76,77,78, 85], and could result in increased staff workload [75, 80]. Short-term or uncertain funding also held back some collaborations [61, 64, 69, 71].
Sharing resources between agencies—for example, through pooled budgets—was identified as one mechanism that may facilitate joint working [23, 61, 69, 71, 83], and, in some cases, as a route to accessing additional resources [83]. But studies also found that sharing resources could lead to challenges in ensuring equitable funding between agencies [24, 25, 69, 71, 85], and could create fears of cost-shifting among some partners [68, 85]. Ultimately, having resources is not enough: resources also needed to be used effectively by local agencies to generate positive impact [74, 79].
The ability of organizations to collaborate was shaped, in part, by the infrastructure in place between them. Shared processes and systems—such as agreements for sharing information, joint meetings, and planning processes (see section on governance and leadership)—were thought to support organizations to communicate and work together [24, 29, 34, 64, 67, 70, 71, 73, 74, 80, 83]. Several studies suggested that co-location or close proximity of teams may support joint working [34, 67, 71, 78, 80, 83]—though this appears to relate largely to teams delivering services. Health impact assessments—a mix of methods and tools to help identify the health and equity impacts of a particular policy or program—were also identified as a key mechanism for local governments and other partners seeking to promote intersectoral action [27, 76]. On the flipside, fragmented or conflicting processes between agencies could hold back local partnerships [24, 25, 65, 67, 68, 70, 71, 77, 85]. Collaborations also needed skills to plan and implement their chosen interventions [28, 29, 69, 74, 79, 85], and monitor and evaluate the results [27, 28, 34, 63, 69, 74, 76,77,78, 81].
The skills and capabilities of staff also shaped collaboration functioning. Staff able to work across organizational and professional boundaries—sometimes referred to as ‘boundary spanners’—were thought to contribute to partnership success [24, 34, 64, 67, 71, 75, 85]. Staff training on collaboration processes and joint training between agencies were thought to help improve collaboration and understanding between sectors [23, 24, 28, 64, 67, 69,70,71, 74,75,76, 79, 83, 84], while high staff turnover or lack of continuity of key staff could hold back collaboration [24, 34, 69, 70, 73, 84].
Governance and leadership
Differences in decision-making processes, competition for power and resources, and lack of accountability between agencies were identified as barriers to collaboration [25, 66, 68, 70, 71, 76, 77, 85]. As a result, clear decision-making and accountability arrangements were thought to contribute to collaboration success [34, 63, 64, 66, 78, 85]. The literature is not particularly clear what this means in practice. Example mechanisms included conflict-resolution processes [78, 79, 83], such as a neutral convener [83], and formalized rules or contracts between agencies [71, 79].
The literature is more clear, however, that good governance means involving all relevant agencies and stakeholders [24, 27, 28, 34, 68, 70, 71, 74, 77,78,79, 81, 85]. This includes front-line staff [24, 64, 69, 71]. But broad membership could also bring challenges for decision-making, such as limiting the chance of consensus [66]. Multiple studies identified the importance of community involvement—including direct involvement of community members and community-based organizations—for collaborations to be successful [24, 28, 63, 66, 71, 81]. This may help ensure that organizations understand community needs and design appropriate interventions [66, 71, 73, 81].
Senior leaders played a key role in shaping how local collaborations and their governance functioned—for better or worse [24, 28, 29, 34, 63, 66,67,68,69,70, 74, 76,77,78,79,80,81, 85]. Leadership commitment was thought to be needed for collaborations to work (see section on motivation and purpose). Leaders used their power to free up resources for joint working [34, 70, 76], help resolve conflicts [29, 78], promote openness and information sharing between agencies [34, 74], and more. But leaders could also block partnership working by defending territorial or organizational interests [68, 70]. The collective involvement of organizational leaders—beyond their individual impact—was thought to contribute to partnership effectiveness [24, 28, 34, 66].
External factors
Collaborations do not exist in a vacuum. While the internal characteristics of partnerships—their leadership, governance, composition, and so on—mattered, the broader context in which local agencies operated shaped how they worked together and the impact they could achieve.
National policy context influenced the local partnerships reviewed. On the one hand, national policies promoted or incentivized joint working in several contexts studied [27, 65, 71, 75, 76, 80, 83, 85]. Government policies on tackling health inequalities, for example, facilitated local partnerships to develop in Europe and elsewhere [27, 76]. In some contexts, such as the UK, national policymakers also mandated partnership working between agencies—though some studies suggested that doing so risks lowering the chances of partnership success, including by undermining the time needed to develop local relationships [66, 85]. On the other hand, some studies suggested that national policy priorities could dampen or conflict with local priorities [76, 85]. And constantly shifting national policies may confuse or undermine local partnerships—as was experienced by those involved in area-based partnerships between health care, social services, and other sectors in England in the late 1990s and early 2000s [61]. Policies not directly focused on collaboration—for example, payment systems in the health care sector—could also create barriers to collaboration between health and social care [80].
The institutional and organizational context of health care, social services, and other sectors also shaped local collaborations [25, 63, 68, 75]. For example, Mason et al. identified differences in national performance systems, pension schemes, and employment arrangements as barriers to collaboration between health and social services in the UK [25]. Several studies also noted that organizational restructuring could hold back local collaboration [23, 61, 69, 71]—for example, by creating uncertainty among agencies [71] and requiring leaders to renegotiate relationships [61].
Other external factors identified in the literature included political context [27, 34, 77]—for instance, ‘political will’—and the geographical location and boundaries of collaborating agenices [61, 69, 77, 85]. Finally, the social and economic context within which partnerships operate was recognized as a factor shaping collaboration functioning potential and impact [28, 34, 76].
link
More Stories
Supply Chain Collaboration for Resilience
Balancing the tightrope of Competition and Collaboration
Digital marketer’s checklist for strategic data collaboration