How do interorganisational collaborations in health care work, why and whom do they benefit?

A realist synthesis

The purpose of this chapter is to present findings from the second and third stage of this realist synthesis (see Figure 1). Drawing on reviews, MRTs and case studies and organisational evaluation literature, we aim to glean initial CMOCs and test these CMOCs against case study literature. In doing so, the chapter seeks to better understand how CMOC configurations of IOC are chained together causally and how differences between contexts affect their implementation. The chapter will present a refined programme theory, explaining how, why and for whom IOCs in health care work.

Excerpts of the chapter have been reproduced with permission from two sources. First, Aunger et al.135 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: https://creativecommons.org/licenses/by/4.0/. The text below includes minor additions and formatting changes to the original text. Second, Aunger et al.138 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: https://creativecommons.org/licenses/by/4.0/. The text below includes minor additions and formatting changes to the original text.

Paper selection

Theory gleaning searches

From the systematic search, a total of 2769 titles and abstracts were screened, which were filtered down to 117 full texts. The Google Scholar searches conducted on 11 March 2020 produced 426,000 results on this specific day, and the first 40 pages of results were screened, resulting in four further papers. At this stage, 52 papers were included (see Figure 10). The 52 papers were then screened for relevance (i.e. whether the papers included sufficient descriptive depth regarding contextual factors, mechanisms and outcomes underlying IOC), which resulted in 35 included papers. Reference-scanning and citation-tracking resulted in a further four papers, giving a total of 39 papers included in the theory gleaning analysis (see Figure 10).

FIGURE 10 A PRISMA (Preferred Reporting Items for Systematic reviews and Meta-Analyses) diagram reflecting all papers included in this realist synthesis. NT, Nuffield Trust; THF, The Health Foundation. (Figure not reproduced here; see the original article.)

Agreement between independent reviewers was 100%. Fourteen purposively identified papers were also drawn on, which outlined the MRTs used to elucidate the workings of mechanisms, bringing the total number of papers included in the gleaning phase to 53.

Theory refinement searches

For the new systematic search, the Social Policy and Practice database search identified 2144 papers. After deduplication against our existing literature library for this project (please see additional file 1 for the full details of this search strategy) there were 1092 papers. Abstracts were then screened for relevance and 104 papers remained. At this point, papers were most frequently excluded because of not being related to IOCs or for not being case studies. After full-text screening, 48 papers were considered eligible for the review; however, after screening for relevance and rigour, only 13 papers were included. Papers were removed for being insufficiently descriptive (i.e. lacking relevance). The searches on the websites of The King’s Fund, the NICE, the Nuffield Trust, The Health Foundation and NHS Employers resulted in an additional 15 papers. Thirteen of these papers were included, but two were removed for lack of relevance. Two further studies were identified through citation-tracking and a final paper was identified in a department newsletter after the search was completed. Four other papers were also brought over on rescreening of studies from the gleaning phase. As a result, 33 total unique papers were included in this theory refinement portion of this realist synthesis.

Literature characteristics

There was a total of 86 studies when n the theory gleaning and refinement stages were combined (see Figure 10). Papers in the literature covered a wide range of IOC types (see Table 6), including 55 case studies, evaluative reports or case–control studies, 16 reviews and 1 briefing.15,17,21,22,32,37,44,47,60,84,85,94,100,102,113,119,145–200

In addition, 14 theoretical papers were included. The 14 papers comprised one paper on partnership synergy,115 one paper on trust,14 two papers on conflict,201,202 one paper on power,203 one paper on coordination,125 one paper on leadership,204 two papers on organisational flexibility,203,205 one paper on task complexity,206 two papers on confidence and formalisation207,208 and two papers on proximity theory.209,210

Theory gleaning

Partnership synergy as a middle range theory

Frequently mentioned in seven of the systematically reviewed studies15,47,60,113,158,167,168,175 was the concept of partnership synergy, which was first coined by Lasker et al.115 as a means for explaining how partnerships achieve advantage over independent, competitive working. Therefore, this theory was adopted as a MRT, which explains how there are ‘partnership functioning’ (hereby collaborative functioning) mechanisms essential to explaining the processes of working together, as well as ‘partnership performance’ (hereby collaborative performance) mechanisms that underpin the improvements that collaborating seek to attain. Lasker et al.115 put forward partnership synergy as an intermediate outcome that comes after the functioning of the partnership, but precedes the effectiveness of it (see Figure 11).115 This means that, when working well together, a combination of resources and skills of the partners is what enables achievement above and beyond what would have been possible individually.

FIGURE 11 Simplified depiction of our MRT and the essential roles of trust, confidence and faith. This figure has been adapted with permission from two sources. First, Aunger et al.138 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: https://creativecommons.org/licenses/by/4.0/. The figure includes minor additions and formatting changes to the original text. Second, Lasker et al.115 (Figure not reproduced here; see the original article.)

Partnership synergy can be considered a mechanism whereby a context of high collaborative functioning leads to greater collaborative synergy and, therefore, improved collaborative performance. Improved collaborative performance is likely to be an outcome in itself, which results from mechanisms involved with an improved ability to achieve health care-related outcomes, such as reduced duplication of effort, economies of scale and competitive advantage.115 However, these performance-related mechanisms will depend on the aims and structure of each individual collaboration. We also add to the MRT the concept of collaborative inertia, which was put forward by Huxham,113 one of the systematically identified studies.

Collaborative inertia occurs when organisations and actors get ‘bogged down’ in the day-to-day functioning of the partnership.113 While trying to optimise the daily functioning, achievement of the actual aims of the collaboration fall by the wayside, as significant resource and time is devoted to collaborative functioning rather than accomplishment of outcomes. It is possible that a collaboration will engage in a period of inertia in its earlier stages of formation, before synergy is later achieved. This concept of inertia was also put forward by a number of the included studies60,113,119,158,184 and is defined as when ‘the output from collaborative arrangements often appears to be negligible or the rate of output to be extremely slow’,13 and this is also depicted in Figure 12.

FIGURE 12 Depiction of key aspects of refined programme theory. Adapted with permission from Aunger et al.138 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: https://creativecommons.org/licenses/by/4.0/. The figure includes minor additions and formatting changes to the original text. (Figure not reproduced here; see the original article.)

This MRT theory, taken together, proposes that the mechanisms comprising ‘partnership functioning’ need to have their context configured very favourably before ‘synergy’ and, therefore, enhanced performance can be achieved. However, we have reframed the concept of partnership synergy as ‘collaborative behaviour’ to make it a more well-defined and testable concept. As partnership functioning relies on many other contextual factors and the mechanisms that enable collaboration, these will be explored in the following section as elements that enable synergy (or collaborative behaviour) and, therefore, partnership effectiveness and accomplishment of its aims.

Gleaning initial context–mechanism–outcome configurations

Our gleaning phase identified that several mechanisms are essential to how collaborations function. The mechanisms included trust, risk tolerance, faith, confidence, conflict, power, interpersonal communication, leadership, cultural integration, perception of progress and perception of task complexity. Our MRT incorporated the trust-building loop by Vangen and Huxham14 and its focus on risk-taking as a driver of collaborative, rather than competitive, attitudes and behaviours. Within the trust-building loop, trust acts as an enabler for organisations to enhance their risk tolerance. Furthermore, we identified a role for ‘faith’,138 where, both individually and collectively, actors have belief that the collaborative endeavour (i.e. the intervention) is a virtuous and beneficial undertaking, thereby worthy of working on. Faith is, therefore, likely to drive actors to dedicate time and effort to engaging in collaborative behaviour, but will change over time in response to other mechanisms and contextual factors.

Therefore, we posited that trust and faith serve as dual drivers for actors to begin behaving collaboratively. In addition, trust and faith also serve as mechanisms in realist terms. In our theory, at the gleaning phase, building trust and faith were key processes that constitute the ‘collaborative functioning’ stage of collaborations. Achievement of a high level of trust and faith across the organisation allows for a synergistic state in which partners achieve maximal collaborative behaviour. Our prior work also indicated that in more integrative types of collaboration (e.g. a merger) or collaborations that are mandated, trust may be progressively replaced by ‘confidence’ in contractual mechanisms as a means for driving collaborative behaviour.138 This is because much of the risk of engaging in collaborative behaviour is enshrined in contractual obligation rather than the building of trusting, robust interpersonal relationships.

Our theory identified that, at least initially, collaboration requires daily efforts to maintain interpersonal ties and build relationships for ‘collaborative functioning’ to occur. Mechanisms found to be key to collaborative functioning identified in the review were trust and faith, with conflict, interpersonal communication, leadership and cultural integration being within different CMOCs (see Figure 11).138 As is evident in this applied example, these CMOCs operate through one another, forming chains that are situated in temporal stages as the collaboration develops over time (see Figure 12). Our theory posited that once trust and faith reach a certain threshold, then a novel state is entered, termed ‘collaborative synergy’, in which the benefits of collaboration can be attained. Since our refined version of the theory, this has been reframed as ‘collaborative behaviour’ to make ‘synergy’ a more tangible concept.138 Driven by the integration of skills, knowledge and resources of partners, performance benefits may include innovations brought about by sharing of expertise, cost savings from better bargaining power and reduced duplication of effort across health systems.

For this section of the chapter, we adopt our theory from the gleaning phase as the MRT for understanding in this refinement phase. A full description of the CMOCs identified in this literature can be viewed in Box 1. These CMOCs were taken as a foundation for the refinement phase of the realist synthesis and the refined theory will be explored in more detail in the following section. Further details of the theory gleaning stage can be found in Aunger et al.138

Theory refinement

Refining the context–mechanism–outcome configurations of collaborative functioning

This refinement of our realist synthesis gave greater insight into how CMOCs are situated in the timeline of development of collaborations, which is how we have structured this section of the chapter. In addition, refinement of our realist synthesis has given further insight into atemporal mechanisms (i.e. mechanisms that can activate at any time) and how these affect the process of collaborating. This section will explain the various mechanisms of the refined theory, when they activate and in response to which contextual factors.

Context–mechanism–outcome configuration coding and establishment of context–mechanism–outcome configuration chains: refinement of mechanisms

Extraction of CMOCs from included studies resulted in 338 CMOCs being identified, many of which were functionally identical and analogous to demiregularities. The majority of the CMOCs were explored in the prior section.138 As previously mentioned, coding of CMOCs was performed according to which mechanism a context activates. The below ‘initial mechanisms’ were used as preliminary deductive codes; however, these were updated as CMOCs when other mechanisms were identified in the literature (see Table 7). By the end of the process, novel CMOCs were not being identified, as they were all analogous to demiregularities that had been already found. In terms of frequency, the CMOCs most identified were CMOCs with trust or confidence as the mechanism, then ‘perception of progress’, faith, interpersonal communication and information-sharing, task complexity, cultural assimilation, conflict, and clarity and sharedness of vision. For the sake of brevity and because of the complexity of IOC, every CMOC and their contextual factors will not be explored here; however, further details of the CMOCs are provided in Report Supplementary Material 1, along with the full table of CMOCs identified in each respective paper. In the refinement stage, some mechanisms from the gleaning phase were redesignated as parts of the contextual environment, including leadership and power, which now form aspects of the contextual environment, rather than being mechanisms in themselves.

Mechanisms specific to early stages of collaboration

Our initial CMOCs138 identified that essential to establishing ‘initial faith’ (i.e. whether or not engaging in collaboration is feasible and worth the risk and effort) are factors such as financial constraints, the regulatory environment and its favourability to collaboration and organisational size (which may affect the perceived difficulty of the task). A further contextual factor identified in this refinement stage was the reputation of the specific form of collaboration being considered. For example, some papers176 referred to negative perceptions due to collaborations being associated with privatisation of the NHS, and this lowered actors’ desire to engage with this form of collaboration in the first place (i.e. their initial faith in the endeavour). Also tied into faith as a precursor mechanism is the perceived legitimacy of collaboration, which often affects a collaboration from the outset. Impacting this mechanism is stakeholder involvement, which can serve to increase its legitimacy in the eyes of staff, whether or not a partnership is voluntary and whether or not staff perceive the collaboration as a threat professionally.173

Related to the level of initial faith is that of initial trust. Initial trust was put forward during the theory gleaning phase of our realist synthesis as likely to come into play in the ‘connecting’ phase of collaboration, during which organisations seek partners and establish initial relationships.213 Contextual elements identified as essential to determining this initial level of trust are the history of collaborating or competing between the organisations,37,176,181,191 organisational reputations44 and, at a later stage, the strength of legal agreements.186 These factors have been found to enhance or undermine trust. For example, legal agreements can act as an initial reassurance when relying on a partner, but can also undermine collaborations if the legal agreements do not allow for attribution of collaborative behaviours to altruistic intent.22,44 Furthermore, this refinement stage also identified that a context of historical health system failures can lower initial trust,187 as evidenced by NHS Providers, which put forward that ‘a legacy of challenges, can lead to a break-down in trust and dialogue and an entrenchment of organisational ‘fortress mentalities’.187 As trust and risk are intricately linked, then this level of initial trust is essential to setting the degree of risk an organisation is willing to take on with its partner, which can affect the aims and outcomes a partnership seeks to accomplish.44,100

Middle stage of collaboration

During the ‘mid-life’ of a collaboration, a multitude of factors come into play that can help rapidly increase the level of trust, buffering against potential conflicts that may occur. Chief among these factors are ‘quick wins’ with a partner, which also help to increase faith.44,166 These small successes serve to rapidly bolster trust and can be increased further through factors such as seconding staff188,189 and having open, honest, stable and empathetic leadership.187,197 Alongside these quick wins are longer-term battles, such as the need to ensure effective interpersonal communication between key organisational actors, managing conflict100,163,167,183 and either creating a new culture or helping build bridges between existing ones.37,145,153

With respect to ensuring appropriate communication between collaborators, a number of contextual elements are key. Geographical proximity is one element that is difficult to mitigate, as a greater geographical distance between collaborators increases time spent building relationships significantly by allowing for ease of arranging meetings and enabling informal interaction.165 Greater geographical proximity was most often cited as improving communication, but can also be unhelpful if conflict is already occurring.172,192 In addition, having a larger size and/or quantity of organisations involved can make communicating more difficult because of the increased number of involved actors and moving parts.85 Compatibility of IT systems,37,160 joint staff appointments183 and having regular collaboration-wide meetings can also work to increase trust as an outcome through the mechanism of interpersonal communication.102,158 When cultures are mismatched or not mutually understood, conflict can occur, which thereby reduces trust.188 Improving cultural assimilation by configuring the context is also possible, which can also go on to enhance trust. A mutual cultural understanding can be fostered by ensuring that a shared vision of the collaboration is in place153 by having a cross-organisational ‘inspirational leader’ who also engages in role-modelling behaviours158 and by supporting staff through the transition.173 It mutual cultural understanding also be improved by having joint teams of staff to work on shared goals, which can improve a sense of collegiality.150,154 It was evident in the literature that certain passive elements, such as the pre-existing degree of cultural distance and whether or not the collaboration is perceived as forced on staff, can also significantly change the difficulty of cultural assimilation.197

Atemporal elements impacting collaboration

Conflicts between organisations can occur as a result of deteriorations in trust, as a result of ‘acute events’, such as failures on specific tasks, or from accumulating tensions caused by cultural distance.100,138 Conflicts directly cause a loss of faith in the collaboration and trust between partners.138 Therefore, there is a reciprocal relationship between conflict and trust, and conflict and faith (see Figure 2), and this is supported by excerpts such as ‘Conflict, for example due to competition between partners, increases the difficulty in predicting the partner’s behaviour and increases the uncertainty in the decision to trust’.44 Reductions in faith can also lead to intraorganisational conflict.138 We also found that conflict can be modulated when it occurs by the approach to conflict resolution. Conflict resolution is now a mechanism in itself in this refined theory, which can dampen the impact of conflict on trust or faith.183

This refinement stage further identifies that use of external, impartial deal brokers and committees can lessen the impact of conflict on trust by moving the locus of that trust to the third party rather than the partner,167 along with having robust governance structures that are not imbalanced in either direction in terms of power.183 Likewise, conflict itself can be mitigated by leaders bringing a constructive approach to conflicts, by proactively attempting to reduce power imbalances and by avoiding or managing any senses of takeovers in the case of mergers or other more integrative collaboration types.44 Although conflict interacts with both trust and faith, so too does the degree to which the collaborative vision is shared and its clarity. The clarity of vision is more keenly interlinked with faith, which is supported by quotes such as ‘most sources concur that a clear vision and/or mission statement should include attainable goals and that lack of clarity about vision can be a serious barrier to engagement’.167 Here, engagement can be considered similar to our concept of faith. However, the sharedness of the vision works through the mechanism of trust, as sharedness relates directly to the interorganisational perception of each organisation. The clarity and sharedness of vision are affected by patient and public engagement, which helps keep the focus on improving care quality (rather than secondary objectives), having inclusive decision-making processes and stable leadership. In many cases, significant leadership turnover meant starting over with trust-building exercises due to large changes in vision occurring.155,171

As a final look into the role of faith, a key mechanism that links into faith as an outcome is ‘the perception of progress’. The perception of progress is interwoven with faith, but is not entirely the same concept (as one can have faith without much perception of progress). The perception of progress is essential for ensuring that momentum is maintained and that there is no stall into what is termed ‘collaborative inertia’, a situation in which there is insufficient faith to maximise work on the collaboration.158 Having an increased perception of progress increases faith and a lesser perception of forward momentum reduces it, and this is supported by quotes from health-care leaders, such as:

So it is harder and less dynamic at the start, until you get a drumbeat going. Then it becomes easier because the peer group start doing it for you.197

Affecting this perception of progress are contextual factors, such as appropriate degrees of ambition (as overambition can lead to disappointment),190,193 implementation of ‘quick wins’, having effective planning (which ensures that staff are working on the most appropriate projects at the right time156) and, importantly, having effective evaluation and dissemination processes (which ensure that staff are aware of the progress being made25,193).

These various review findings suggest a web of contextual elements, operating through many mechanisms to produce many outcomes and forming causal chains (see Figure 13). Some of these elements can be altered to be more beneficial to implementers (e.g. keeping ambitions realistic) and some cannot (e.g. geographical proximity). Although these elements discussed are those that underlie the functioning of the partnership, the review also identifies mechanisms that underlie material improvements to organisational performance.

FIGURE 13 Example of how contextual elements and mechanisms may operate over time to drive a voluntary-type partnership into collaborative synergy. The lines represent how levels of trust and faith may change independently over time in response to the various numbered events. Adapted with permission from Aunger et al.138 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: https://creativecommons.org/licenses/by/4.0/. The figure includes minor additions and formatting changes to the original text. (Figure not reproduced here; see the original article.)

Refining context–mechanism–outcome configurations driving partnership performance

As trust and faith are maximised, work on managing conflicts, building collaborative structures and establishing relationships decreases. As a result, time and effort are freed up to create the conditions for collaborative ‘synergy’, which in our refined theory refers to a willingness to engage in collaborative, rather than competitive, behaviour (see Figure 3). Once risk tolerance and faith are maximised, then partners can strike out to work on risky innovations together, share best practices, pool resources, bid for contracts together, increase influence in the local health economy, reduce duplication of effort and better focus on inequalities in the health system, and these may be the mechanisms through which ‘ultimate outcomes’ of collaboration (i.e. performance improvements) are realised, occurring because of an accumulation of skill, knowledge and resources being brought to bear against problems encountered by the partnership. However, these benefits seldom come to fruition because of the sheer amount of work and goodwill that is required to rise above the quagmire of the daily functioning of the partnership, explaining why many collaborations are not successful. Studies capture contexts where ‘too much was being expected too soon . . . as they were still working out how to function operationally before they could accomplish goals such as decreasing workload and improving care’.37

Achievement of this synergistic state is extremely difficult, as many collaborative endeavours are ‘set up for failure’ because of initial contextual conditions being configured in such an unfavourable manner that it becomes insurmountably difficult to build the relationships required.214 These difficult contextual conditions can include a regulatory environment that still revolves around competition rather than collaboration,183 mandated partnerships and integrated care initiatives that do not allow for building of relationships, brought into place in conditions of pre-existing histories of competition and ‘bad blood’187 and a lack of financial support or consideration provided for collaborative endeavours to be implemented properly.176 As outlined by our theory, these common contextual factors serve to undermine initial trust in partners and initial faith in the process and increase task complexity to a degree that makes them very difficult to overcome. One quote from an included study of buddying arrangements in the UK reflected this clearly:

Interviewees repeatedly said how difficult it would have been if their buddying arrangement had been imposed, and indeed those arrangements seen to be imposed by regulatory bodies appear to be have been the least successful.165

This information provides implications for cross-sector initiatives, such as ICS or accountable care organisations taking place in the UK, as it demonstrates that context can be configured very unfavourably in these mandated arrangements, making it difficult to build trust and faith. In addition to many of these arrangements being mandated, cross-sector working brings additional challenges by requiring working between workforces of differing professional backgrounds. These professional differences manifest in a greater degree of cultural divide (a barrier that was referenced by many of the included case studies44,100). Likewise, in cross-sector arrangements, the number of partners and size of the involved organisations is likely to be greater, which further increases the difficulty of communicating effectively and clearly, and results in much higher task complexity. These are all concerns that require great tact to mitigate. Collaborative performance will be explored further in a later chapter by drawing on our empirical findings.

Refinements by collaboration type and collaborative functioning

Our initial rough realist theory suggested that partnerships can be characterised along a spectrum of integration, from full integration (i.e. mergers) to more informal endeavours involving fewer people (i.e. buddying, clinical networks).61 Such differences between collaborative types (e.g. buddying vs. alliances) are also reflected in our findings as changes to contextual elements (i.e. whether or not they are mandated), and these changes can affect the task complexity, perceived legitimacy, faith and initial trust. Our review identifies how it is possible to trace how these changes affect implementation. For example, a voluntary buddying arrangement is likely to be relatively simple to implement, as it is unlikely to be perceived as threatening by staff, does not involve many organisations and, although perhaps not supported by formal legal agreements, is likely to involve partner self-selection. This arrangement is likely to start with a high degree of initial trust between partners and initial faith in the process. On the other hand, ICSs, which are now mandated to come into force in 2021, involve a significant number of large organisations coming together in a cross-sector manner, including local councils, primary and social care, and acute care,33 and this drastically increases task complexity by being cross-sector and having many involved organisations. Additional challenges include that it is likely such a move may be perceived as a threat by staff, reducing faith by reducing its legitimacy in their eyes. An ICS is also likely to increase difficulty of effective communication by having both great breadth and depth of organisations involved, making trust building more difficult, and there may be reduced initial trust from outset by being set in a local context of pre-existing competitive attitudes. In addition to that, being given a rapid timeline while having to overcome prior differences and conflicts poses an additional challenge, which requires a strict patient-centred focus shared across the system to overcome.

Our programme theory suggests that formalisation through contracts may be one means of enabling collaborative behaviour in such a situation where initial trust is likely to be low or complexity very high. The following section explores further how trust and its relationship to risk threshold may be replaced by confidence as a primary driver for collaborative behaviour in mandated or integrative collaboration types.

Trust versus confidence in integrative and mandated partnership types

In the gleaning phase of our realist synthesis, we identified a common CMOC that suggested that the formalisation of an arrangement through contracting facilitated trust relationships by having potential to act as a buffer where trust may otherwise be lacking.138 Inversely, trust may also be undermined if too much collaborative behaviour is mandated through contract, as organisations assume that the partner will act collaboratively as a result of contractual obligation. We also found evidence that, with either mandated or otherwise more integrative types of partnerships (e.g. mergers), trust may not be the primary determinant of collaborative behaviour, as that role would instead shift to confidence.138

Our updated review sought to further understand the relationship between confidence, formalisation, risk tolerance and trust. Although it was difficult to identify relevant information, our included literature identifies an inherent mistrust taking place in organisations involved in mandated partnerships, with perceptions of ‘taken over’, atmospheres of ‘them and us’ and domination of powerful partners resulting in a lack of trust and partnership strength.164 Next to this lack of trust in mandated collaborations, formalisation was seen by sources as a means for risk management. Use of controls and contracts ‘to minimise uncertainties of behaviour by partners’ had value in instances where trust was low.44 Another source put forward the notion that formalisation through contract is ‘primarily about managing risk, trying to situate the risk with the organisation/s most able to mitigate it, and giving them the power to do so’.187 Therefore, we have added the concept of confidence, built through formalisation, to our programme theory diagram as a determinant of risk tolerance (see Figure 12). Greater confidence (context) will thereby increase risk tolerance (mechanism) for engaging in collaborative behaviour (outcome) (see Figures 12 and 14).

FIGURE 14 Depiction of ‘causal web’ formed by CMOC chains of how and why health-care collaborations work. Light blue and dark blue elements relate to collaborative functioning. Adapted with permission from Aunger et al.138 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: https://creativecommons.org/licenses/by/4.0/. The figure includes minor additions and formatting changes to the original text. (Figure not reproduced here; see the original article.)

However, while some sources were espousing formalisation as a means for improving trust, it was evident that other sources were warning that overformalisation could undermine the trust-building process:

[The] contribution [of contracts] is less in subsequent phases once trust grows, not least because early contracts cannot anticipate every eventuality.44

This sentiment is echoed by the following quote from Dickinson and Glasby:22

. . . although a lot of the joint commissioning processes described to us were formalised and structural, people often recognised that joint working is essentially relational (based on informal conversations and interactions).22

These findings indicate that voluntary collaborations should be careful not to overly rely on structural means of obligating collaborative behaviour where they should be focused on building relationships. Memoranda of understanding, although generally non-binding, increase trust between collaborators and thereby reduce perception of the risk taken on when engaging in collaborative behaviour.178 We hypothesise that these non-binding contracts, such as memoranda of understanding, may be most appropriate as a tool to increase risk tolerance for less integrative or voluntary arrangements, where there is a risk of misattribution to obligation rather than genuine collaboration with the use of binding contracts, as Casey puts forward ‘there is a need for a balance between power-sharing and control, between processes and results, between continuity and change and between interpersonal trust and formalized procedures’.60

A novel means of depicting context–mechanism–outcome configuration chains: ‘causal webs’

Our review findings suggest that CMOCs formed chains of generative causality as a result of the outcome of one CMOC becoming the context for another. For example, a larger organisational size (context) leads to greater task complexity (mechanism), which affects people’s faith in the process (outcome).160 However, further down the chain, faith is a context in which collaborative behaviour (mechanism) occurs, leading to improved synergy (outcome) and, therefore, the ‘chains of causality’ emerged naturally from the linkages between CMOCs. As there is no common diagrammatical standard for how to depict CMOC chains in realist syntheses, we found that it was appropriate to depict our findings in the ‘causal web’ shown in Figure 14. In Figure 14, it should be simple to follow the aforementioned example of CMOC chain in the figure from organisational size (context 1) through to collaborative behaviour (outcome 2). This mode of representation draws similarities with the concept of a ‘context map’ put forward by Renger et al.215 for use with realistic evaluations; however, our approach expands a context map by including both mechanism and outcome in the web with their own means of representation. Yet, it is important to note that the diagram does not depict how specific contexts alter the mechanisms, only which mechanisms are attached to which contextual factors, and which outcomes are attached to which mechanisms.215 For specific dynamics, Figure 12, updated from our version in the theory gleaning phase based on the present refinements, depicts the key mechanisms and dynamics underlying the how of the workings of the CMOCs we identified.

Conclusion

Although many existing theories have delved into what underlies the process of IOCs in health care, this chapter builds on our knowledge of not only what but also how and why these elements work, and this was achieved by gleaning and refining CMOCs based on 86 papers to produce a refined realist theory. The theory has surfaced the inter-related roles of trust and risk tolerance, faith, task complexity, interpersonal communication, cultural integration and perception of progress, and how these elements drive collaborative behaviour. Likewise, the theory demonstrates that, in mandated or highly integrative collaborations, the locus may be shifted from trust towards contractual obligation and a sense of confidence that the partner will act collaboratively. The theory also builds on similar existing findings in the literature from other authors by linking contexts, mechanisms and outcomes together into a long causal chain, as this allows an understanding of how initial environmental and interorganisational conditions set levels of trust, faith and task complexity, and how these mechanisms can be managed later in the process of collaborating. These chains of CMOCs were situated within a ‘web of causality’, which allowed us to depict how distant contextual items and their mechanisms work to affect more long-term outcomes, including collaborative behaviour. This version of our programme theory will be tested further in the following chapter, which is based on findings from the empirical side of the project.