The Generation of Integration: The Early Experience of Implementing Bundled Care in Ontario, Canada.

Policy Points Policymakers interested in advancing integrated models of care may benefit from understanding how integration itself is generated. Integration is analyzed as the generation of connectivity and consensus-the coming together of people, practices, and things. Integration was mediated by chosen program structures and generated by establishing partnerships, building trust, developing thoughtful models, engaging clinicians in strategies, and sharing data across systems. This study provides examples of on-the-ground integration strategies in 6 programs, suggests contexts that better lend themselves to integration initiatives, and demonstrates how programs may be examined for the very thing they seek to implement-integration itself. CONTEXT: By bundling services and encouraging interprofessional and interorganizational collaboration, integrated health care models counter fragmented health care delivery and rising system costs. Building on a policy impetus toward integration, the Ministry of Health and Long-Term Care in the Canadian province of Ontario chose 6 programs, each comprising multiple hospital and community partners, to implement bundled care, also referred to as integrated-funding models. While research has been conducted on the facilitators and challenges of integration, there is less known about how integration is generated. This article explores the generation of integration through the dynamic interplay of contexts and mechanisms and of structures and subjects.
METHODS: For this qualitative study, we conducted 48 interviews with program stakeholders, from organization leaders and managers to physicians and integrated care coordinators, across the hospital-community spectrum. We then used content analysis to explore the extent to which themes were shared across programs and to identify idiosyncrasies, followed by a realist evaluation approach to understand how integration was produced in structural and everyday ways in local program contexts.
FINDINGS: Integration was generated through the successful production of connectivity and consensus-the coming together of people, practice, and things, as perceived and experienced by stakeholders. When able, the programs harnessed existing cultures of clinician engagement, and leveraged established partnerships. However, integration could be achieved even without these histories, by building trust, developing thoughtful models, using clinicians' existing engagement strategies, and implementing shared systems and technologies. The programs' structures (from their scale to their chosen patient population) also contextualized and mediated integration.
CONCLUSIONS: This article has both practical and theoretical implications. It provides transferable insights into the strategies by which integration is generated. It also contributes conceptually to realist approaches to evaluation by advancing an understanding of mechanisms as contextually and temporally contingent, with the capacity to produce new contexts, which in turn generate new sets of mechanisms.

Integrated health care models are seen as a way to counter fragmented care delivery and rising system costs by bundling services and encouraging interprofessional and interorganizational collaboration. Integrated funding models (IFMs), also referred to here as bundled care, refer to an approach in which a predetermined payment is made to a group of providers to deliver agreed‐upon bundled health care services across a continuum for a specific procedure or diagnostic group within a specified time period.1, 2 Partnering organizations bear costs in excess of this payment but retain any savings accrued by lower expenditures, therefore incentivizing cost control through efficient care delivery, improved care transitions, and fewer rehospitalizations.3

In 2015, the Ministry of Health and Long‐Term Care (MOHLTC) in the Canadian province of Ontario chose 6 programs, each comprising multiple hospital and community partners, to implement IFMs. While integrated funding may have been an initial driver toward integration, in their early stages these programs were figuring out how to become integrated in multiple ways, from establishing shared systems and practices across partners to developing new, streamlined care pathways. For this qualitative study, we conducted 48 interviews with program stakeholders across the hospital‐community spectrum to understand how integration was generated in specific program contexts, in its many dimensions.

Ontario's IFM initiative is part of a larger legislative focus on the integration of health services. It was partly inspired by other jurisdictions that have already adopted a range of forms of integrated funding. In the United States, for instance, the Affordable Care Act allowed for a shift away from the traditional fee‐for‐service model and introduced integrated funding in the form of accountable care organizations and bundled payments for care improvement.4 Accountable care organizations are health care provider groups that take on the clinical and financial risks of caring for a group of beneficiaries selected on the basis of health care utilization,1, 5 and bundled payments for care improvement are funding for selected organizations based on a specific episode of acute care.6 Bundled‐funding programs have also been implemented in England (a national integrated care pilot) and the Netherlands.7 The Ontario initiative was also inspired by the success of a bundled care program undertaken by a health care organization in Ontario, which also participated in 1 of the 6 IFM pilots.

Scholarship on how integrated health systems work has ranged from a focus on identifying the levels, types, and degrees of integration;7, 8, 9 to the key components of integration;10, 11, 12 to the challenges and enablers of integration.8, 10, 11, 13, 14, 15, 16, 17, 18 Identified facilitators include trust building,11 existing capabilities expansion, leadership,8, 17, 19 physician engagement,14, 20 professionals’ capacity for role transformation,11 minimal patient and program complexity,11, 21 commitment of providers, timing and flexibility in implementation,17 and effective negotiation of privacy regulations and information technology (IT); all are factors that can become challenges when not in place.12, 22

However, comparatively less work has been done on the very meaning of integration itself. Integrated care typically has been defined as the coming together of system design and delivery for the purpose of delivering patient‐centered care, with emphasis placed on different aspects of this process.23, 24 At the same time, integration has been analytically distinguished from integrated care as process and outcome, with integration potentially resulting in integrated care. Kodner, for instance, defines integration as “a coherent set of methods and models on the funding, administrative, organisational, service delivery and clinical levels designed to create connectivity, alignment and collaboration within and between the cure and care sectors,” with the goal of benefiting patients.25 Singer focuses on patient‐centeredness, conceptualizing integrated care as patient‐tailored care that is continuous over time and coordinated and shared across professionals and systems.23 Kreindler's work uses a social‐psychological lens to analyze what integration means from stakeholders’ perspectives, concluding that “each site offered a unique interpretation that flowed from its existing strategies for social‐identity management.”26 Our article focuses on integration as a process that is continuously generated or stymied (sometimes simultaneously) by the dynamic interplay of contexts and mechanisms and of subjects and structures.

A realist evaluation approach lends itself well to the task of analyzing how integration is enacted, as it seeks to identify the underlying program theory or mechanism that leads to success. Realist evaluation also seeks to identify contextual influences on outcomes to discern “what works for whom, in what circumstances, in what respects and how.”27 Understanding how integration is generated seems to be particularly important to integrated care and related funding models, given mixed evidence on the effectiveness of these interventions.12, 17, 21, 28, 29 Yet, while realist evaluations have attempted to analyze the relationship between contexts (ie, environmental structures and conditions), mechanisms, and outcomes,30, 31 there has been little consensus on the application of the mechanism concept, understood as how program “resources … impinge on the stakeholders’ reasoning” in program implementation, or, simply put, what is it about programs that brings about their effects?32

A key point of contention has been the location of agency, the culturally mediated capacity to act.33 What produces integration (or the lack thereof)? Is it the relatively enduring social world (which includes program structures) in which people are embedded that generates change in the form of social agency? Is it individual psyche? While critical realists have suggested that both forms of agency must be accounted for separately when understanding mechanisms,34 proponents of realist evaluation have theorized integration mechanisms in more holistic terms as the interplay between programs’ structures and participants’ responses to them.35 We align ourselves here with the latter camp, as doing otherwise would be to assume that the individual's capacity to act is intrinsically and easily separable from his or her social context, rather than informing and mutually generating each other in always emergent ways and to contextually sensitive degrees.

Context and mechanism are similarly iteratively generative of each other, much like Pierre Bourdieu's habitus, a “structuring structure and a structured structure” that enables new responses while simultaneously circumscribing others.36 The iterative interaction of context and mechanism is aimed at a single overarching outcome—that of integration itself. Next we provide an understanding of how integration was produced through the successful generation of connectivity and consensus to form a more closely connected version of what preceded it. We use connectivity to refer to the coming together of systems, structures, and things and consensus to refer to the coming together of people, ideas, and practice as perceived and experienced by stakeholders. We then offer examples of practices that are potentially transferrable to other jurisdictions seeking to integrate care across health care organizations.

Methods

This article discusses the qualitative component of a mixed‐methods, real‐time, centrally conducted, provincial evaluation of the selected IFMs, using health administrative data, patient experience surveys, and qualitative interviews with key stakeholders conducted by researchers at the University of Toronto.

Context

In the 1990s, the province of Ontario moved from cost‐based to global budget funding for hospitals. Beginning in April 2012, health system funding reform was introduced for acute care hospitals to provide efficient, cost‐effective care. The expectation was that by April 2014, hospitals would receive 30% of their funding from global budgets, 30% through volume‐based payments for quality‐based procedures (QBPs) with existing evidence‐based pathways for acute care, and 40% from a population‐based model (Health Based Allocation Model).37 These policies are set by the provincial MOHLTC and implemented by regional authorities called local health integration networks (LHINs). LHINs are responsible for payments to hospitals and other institutional care settings, including nursing homes and home care service organizations. Until mid‐2017, home care itself was organized and paid for by community care access centers (CCACs), with 1 CCAC for each LHIN, which provided CCACs with global budgets. CCACs were responsible for assessing patients referred for community home care services, determining service levels, and then coordinating service contracts with independent home care service organizations. Integrating health care services in the context of these layered structures has been particularly challenging for Ontario, spurring a merger of LHINs and CCACs in mid‐2017. Integration remains a priority in the proposed transformation of the health care system.38

It was in this context that the MOHLTC, following a readiness assessment, chose 6 programs to pilot IFMs in June 2015. Each program consisted of partnerships across acute care and postacute care sectors. The programs could determine their clinical focus and the specific set of services that a patient received for an episode of care across providers for a fixed time period, not including physicians’ fees. An IFM coordinator typically helped ensure smooth patient transitions. This bottom‐up approach enabled significant heterogeneity across programs, allowing for an analysis of how integration was generated in different contexts.

Data Collection

We began the qualitative interviews by developing a recruitment strategy to identify suitable interview subjects across the 6 programs. To do this, we used a 3‐pronged selection strategy of reviewing each program's governance structure to identify key individuals, inviting the programs themselves to identify key players, and employing snowball sampling that asked interviewees to suggest others whose views they considered valuable. We chose 6 individuals from each program, including senior decision makers, managers, integrated care coordinators, and clinical champions from a range of partner organizations, to capture perspectives across the hospital to community and home care service continuum. Allowing program stakeholders to suggest suitable participants resulted in an interview participant pool that typically reflected what stakeholders themselves thought to be important. As such, our inability to include a physician champion at one program spoke to the uncertainty surrounding the role and challenges with engagement in general, whereas another program's suggestion that we include a patient advocate (the sole patient included) who had helped shape its model was a similar indication of what was valued.

The programs began implementation between October 2015 and February 2016, with the data collected between February and June 2016. We timed the interviews to capture the experience of early IFM conceptualization and rollout. Our questions focused on 3 core areas: (1) the development of the IFM program (what brought the partners together, how the partnership was fostered, how the clinical pathway was created, how financial decisions were made, and what role leaders played); (2) the program's implementation (how providers worked together to deliver the bundle, what new practices had to be learned, how trust was fostered across providers and organizations, and how they communicated); and (3) reflection on the initiative (what key challenges were encountered, what went well and why, and what support was needed to spread and/or scale the initiative). We also interviewed individuals from the MOHLTC and LHINs to understand the broader policy goals driving the initiative, the support offered to the programs, and the scope for sustainability, spread, and scale. In all, we conducted 48 one‐hour interviews, most by telephone given the programs’ geographic dispersal, but a few in person. The interviews were recorded and transcribed, and identifying names and other information were removed.

Analysis

We repeatedly read the transcripts to create open codes, which we then grouped into categories that we continuously amended as their latent meanings were explored. Using NVivo 10, we iteratively used content analysis39 to identify key themes. During this process, transcripts for each program were read and coded together to form a preliminary understanding of the unique context‐mechanism configurations at work within programs. We used a universal coding scheme, with certain codes being more pertinent to specific programs. Both the interviews and the analysis were conducted by an experienced qualitative researcher / social scientist who frequently conferred with the team members for support and feedback while coding. In order to validate our themes, we presented preliminary results to program participants and the MOHLTC on 2 occasions.

We used first‐order codes as preliminary data holders to organize the data (eg, program context, challenges, facilitators, perceived impact), with second‐order codes fleshing these out in greater detail (eg, under “facilitators” were previous collaboration, motivation, leadership, and openness to culture change, among others—codes that could sometimes be found under “challenges” too). Deconstructing the data in this way allowed us to assess the extent to which programs shared specific themes. We followed this with a process of holistic, context‐sensitive reconstruction, driven by a realist evaluation approach.

We adopted a realist evaluation approach because it is well suited to analyzing complex health care interventions involving multiple actors and nonlinear processes that require careful attention to context, mechanisms, and impacts.40 It is not a methodological procedure so much as a “logic of inquiry” that attempts to answer the question, “what works for whom and in what circumstances;” it is an exercise in “‘thinking through’ how a programme works.”27 Despite the lack of prescribed steps, realist evaluations typically have an explanatory focus, include mixed methods as appropriate, and examine context‐mechanism‐outcome configurations (sometimes presented in tabular form) that are iteratively tested and refined.41 While our approach was similar, we favored a nuanced approach to context and mechanism that regarded them as inextricably linked expressions of the other that overlapped and flowed into other contexts and mechanisms. In this article, we present our data as program‐specific case studies that typify specific mechanisms and also show how other programs relate to the concept being investigated.

Results

Program integration could be generated by building on existing structures and cultures, from already shared information systems to established cultures of collaboration and engagement. These acted as both contexts and mechanisms that mediated integration, as did more recently established program structures, from program scale to chosen patient population. However, integration could also be actively generated by newly fostered mechanisms working within the parameters of the current initiative, such as new partnership formation based on trust, thoughtful model development, a commitment to clinician engagement, and information‐sharing. Table 1 summarizes these mechanisms, which were present across the programs in various expressions and degrees. These mechanisms must be understood as interacting with one another in complex ways, as we explain next.

Table 1

Key Contexts and Mechanisms of Integration

1. Program Structure
Program scale, organization size, chosen patient population, and key program features mediated integration processes, with discrepancies in organization size, practice, and resources challenging integration.	“There's a new tool that they're [larger IFM program of which they are a part] introducing for the occupational therapists.… It's time over and above what they're doing already, their assessment.… We're trying to pick up okay, well, you know, is this tool that they're using for discharge, does it make sense that we change to this? Is this a good thing? And not just blindly do something.” (Program 1; 5)
2. Leveraging Existing Partnerships
Preexisting relationships between partners facilitated IFM implementation and encouraged more seamless sharing of data, practices, resources, and systems, as well as a close‐knit clinical team that spanned acute and postacute care sectors.	“[O]ur IT, finance, decision support, human resources, communications, patient experience, all of those have a joint vice‐president.… So the concept of this integrated funding wasn't a big kind of scary thing for us. Because we have a joint CFO, we were already used to the concept that … even though our budgets were all separate, we had somebody that had oversight to them.… [And] we've been striving to try to ensure that we have line of sights to each other's data.” (Program 2; 1)
3. Building Trust
Trust was fostered over time through frequent interprofessional and organizational interaction, with partners learning together and spending time in one another's professional worlds.	“[I]t was an eye‐opener to see how the nurses do the assessment, how they chart in the community, the lack of equipment at their fingertips. You know, if they want to do a blood pressure, the monitor is not just at their fingertips. You know, they've got to get it out of their bag and get it out of their car.” (Program 3; 3)
4. Developing Thoughtful Models
Model development was facilitated by working through risk scenarios, encouraging ground‐up input from clinicians and administrators, and accounting for perspectives across the hospital‐community spectrum.	“[W]e said, okay, what if volume is up 10%? Referral rate is exactly the same but just the volume is up.… What would it have meant for the hospitals, what would it have meant for the CCAC, what do we want to do? What do we do if the volume is the same but for two of the hospitals, everybody is on target with the referral but two of the hospitals are referring at a much higher rate than previously? … And then what if the volume is low?” (Program 4; 1)
5. Engaging Clinicians
Clinician engagement was generated by including clinicians in model development, drawing on integrated care coordinators with established relationships with physicians to foster buy‐in, and developing an engagement strategy led by formally appointed or informal clinical champions.	“[Physicians] were very excited to think that we would have an RT [respiratory therapist] following a COPD patient into the community.… And so if the RT clinical care coordinator wanted to call the respiratory therapist, they already have that relationship … [v]ersus, you know, a CCAC care coordinator where they don't have that relationship and don't have the confidence or the trust that they understand how to titrate oxygen or something.” (Program 5; 6)
6. Sharing Information
Information‐sharing was facilitated by innovations, such as real‐time data‐sharing across stakeholders, and challenged in the absence of shared systems, such as a common electronic medical record, and by different organizational interpretations of privacy regulations.	“Now physicians have the ability to see what's been happening to that patient since they left their inpatient bed. And again, that's helping the physicians to first of all become more aware of what happens in home. And so that is new information to them. Secondly, it's also helping them become more integrated and supportive of the in‐home care team.” (Program 6; 8)

Structuring Programs

The 6 IFM programs varied significantly as numerous organizations across the acute‐postacute continuum coalesced in heterogeneous configurations that ranged in size from a 2‐partner initiative to one that was LHIN‐wide, involving 15 partners. Three focused on chronic obstructive pulmonary disease (COPD) and congestive heart failure (CHF) (Programs 4, 5, 6). Others focused on cardiac surgery (Program 3), stroke (Program 1), and urinary tract infections (UTIs) and cellulitis (Program 2) (Table 2). Program structure, such as scale, parity in partner organization size and practice, chosen patient population, and key program features, both contextualized and mediated integration processes. Decisions regarding how to structure the IFMs were themselves informed by stakeholders’ perceptions of existing expertise and needs. The following case study illustrates the impact of discrepancies in an organization's culture and size. Throughout the case studies we use the format “Program #; #, #, #” to refer to interviews conducted with participants from each program (eg, “Program 1; 3,5,6” refers to 3 participants from Program 1). Given efforts to include heterogeneous partner perspectives within each program, the interviews typically include participants from various program partner organizations.

Table 2

The IFM Programs

Program	Program 1	Program 2	Program 3	Program 4	Program 5	Program 6
Clinical Condition	Stroke	UTI, cellulitis	Cardiac surgery	COPD, CHF	COPD, CHF	COPD, CHF
Key Partners	5	3	2	15	7	4
Key Features	Clinical collaboration tool; warm handoffs during transitions in care; potential use of telecommunications technologies to deliver health care services to patients at home.	Short‐term nursing interventions with approx. 14 nurses hired specifically for this intervention; full access to electronic health record both inside and outside the hospital; 1 contact number.	Integrated care coordinator who works with patients beginning at pre‐op; a 24/7 contact center; and telemonitoring in the home.	Integrated care coordinator; 24/7 telephone line; virtual team rounds; lead home care agency.	Care coordinators; a 24/7 access line for patients; remote consults enabled through technology and specialist follow‐up, including ambulatory rehabilitation.	eHomecare model (eShift/eClinic) for remote monitoring immediately after discharge; 24/7 telehealth; navigator; clinical care coordinator.
Expected Length of Bundle	104 days after discharge	60 days after discharge	30 days after discharge	60 days after discharge	60 days after discharge	60 days after discharge

Case Study: Program 1

Program 1 originally involved 6 partners across acute‐postacute sectors and included large hospitals, a postrehabilitation center, and 2 CCACs (and therefore 2 LHINs) that together developed a stroke‐based IFM. The organizations’ sizes and foci varied, with one being a designated regional stroke center and the others having stroke units or providing rehabilitation expertise. The program was the only one to cross LHIN boundaries, resulting in greater exposure to differences in LHIN risk‐taking cultures. The program's stakeholders agreed to manage patient transitions by substituting the integrated care coordinator role used by other programs with warm handoffs (Program 1; 3, 5, 6) because of a lack of consensus as to what the role entailed and concerns that it would increase costs and step on an existing unionized CCAC position (Program 1; 2, 5, 6). The decision unintentionally frustrated clinicians:

[W]e've got some really good clinical champions … [but] the clinicians are frustrated … I think that we, the steering committee, have failed the clinicians because we've insisted on perhaps having too much say in what they were looking at and what they were doing. (Program 1; 6)

The partners had largely been brought together based on their expertise and previous collaboration, and a decision was made to include a smaller hospital that had expressed interest in joining the model. Not all partners had previously collaborated with the smaller organization, and this lack of history, combined with differences in resources, clinical practice, and IT systems, posed early challenges. Stakeholders noted a lack of presence at planning tables, with one remarking that the organization in question was often “the last to contribute, the last to submit whatever is required” (Program 1; 6). It was also not part of the electronic rehabilitation referral system that the other partners shared.

At the same time, the organization in question found itself caught at the confluence of the need to change entrenched modes of practice and a lack of resources, time, and clinician buy‐in. It found itself struggling to put in place practices already shared by its larger counterparts and soon began wondering if the partnership was in fact a good fit:

There's a new tool that they're [larger IFM program of which they are a part] introducing for the occupational therapists.… It's time over and above what they're doing already, their assessment.… We're trying to pick up okay, well, you know, is this tool that they're using for discharge, does it make sense that we change to this? Is this a good thing? And not just blindly do something. (Program 1; 5)

Work relating to the model contributed to workloads, competed with other priorities, and had to be managed without additional resources, leading this participant to conclude that “something has to give … we're not robots” (Program 1; 5). Approximately a year later, this organization decided that it would no longer be a program partner.

These challenges cannot be understood in and of themselves as helping advance or thwart integration; after all, Program 1 had much that was shared by many of its partners, from strong leadership buy‐in and previous partnerships to healthy clinician engagement and shared platforms and practices. But if integration is understood as an ongoing process, the impact of these challenges on stakeholders’ perception of how well people and processes were coming together, at least during early implementation, must be acknowledged.

Other Experiences

Program 1 was by no means alone in experiencing the effects of discrepancies in structure and practice across partners. Program 4 was the most ambitious in scale, being LHIN‐wide with 15 partner organizations. It, too, ran into challenges coordinating care across differently sized, resourced, and located partners:

[W]e have a [name] clinic. So if I have an issue with a patient in the community, I have access to a team of expertise in managing COPD at my fingertips. Whereas not every site is going to have that.… So for example, if I need to bring a patient into [this] clinic, typically they live within a very short distance of the hospital.… But if we go out now to [area], well, (a) they don't have a clinic, and (b) their patients may be living an hour, 45 minutes, to a physician where they don't have buses or taxis readily available at their fingertips. So there are going to be different challenges as we spread this model across the LHIN. (Program 4; 5)

However, strategies such as local‐level clinical engagement across its multiple partner organizations helped Program 4 negotiate the complexities of scale.

Other structural variables affecting stakeholders’ experience of integration were partnership structures informed by hospital‐community relationship histories, chosen patient population, and modes of funding the new models. Most hospitals partnered with CCACs that had contracts with home care service organizations. A minority did so enthusiastically, having built good working partnerships over time. Others did so reluctantly, citing perceived inefficiencies inherent in working with a “middleman,” thereby requiring more work to be done to forge trusting relationships across teams. Two programs applied for community service status, allowing them to bypass their CCACs and directly contract with home care service organizations. These program stakeholders welcomed this decision despite the initial bureaucratic work it required.

The patient population chosen may also affect the complexity of the model's clinical and funding aspects. COPD/ CHF program participants tended to bemoan the clinical and financial complexity involved while simultaneously applauding the decision to tackle these conditions because of their well‐established evidence‐based pathways, greater need for process efficiencies, and therefore greater scope for rewards. For instance, the clinical and social patient complexity associated with chronic disease challenged programs’ ability to project costs for a typical patient. In many cases, funds that were anticipated to correlate with services used could only be tentatively allocated because the models were still being developed and the exact numbers of patients and services required were unknown. This caused some uncertainty for financial administrators even as their clinical counterparts forged ahead more resolutely with integration.

In addition, COPD and CHF were already funded QBPs that some believed would facilitate financial prognostication while being a reliable source of funding that could be tapped to fund new IFMs. However, some COPD patients, while theoretically qualifying for QBP funding at the outset, did not in reality qualify, given their comorbidities and longer than expected hospital stays. “Nobody comes in with just COPD or just CHF,” as one participant said (Program 5; 1), while another wondered whether they should have chosen something “much tighter” like stroke (Program 6; 5). While these clinical and financial considerations played a key role in structuring and mediating integration efforts, the focus was on clinical rather than financial integration. This was evidenced by program participants’ reference to the initiative as “bundled care” rather than “integrated funding models,” as used by the MOHLTC.

Leveraging Existing Partnerships

All 6 programs leveraged established partnerships in some form, although the breadth and depth of these relationships varied. Typically, preexisting deep connections between organizational partners facilitated IFM implementation by building on existing relationships, resources, and support structures.

Case Study: Program 2

Program 2 focused on UTIs and cellulitis and was one of the smaller IFM programs, with just 3 partner organizations involved: 2 hospitals and a CCAC. All 3 had a well‐established history of working together that created a sense of seamlessness for participants across the organizations. All 3 partners, for instance, had already been through a recent back‐office integration:

[O]ur IT, finance, decision support, human resources, communications, patient experience, all of those have a joint vice‐president.… So the concept of this integrated funding wasn't a big kind of scary thing for us. Because we have a joint CFO, we were already used to the concept that … even though our budgets were all separate, we had somebody that had oversight to them.… [And] we've been striving to try to ensure that we have line of sights to each other's data. (Program 2; 1)

The program was delivered by a team of nurses who, while based at the CCAC, needed direct access to both hospitals. The partners therefore decided to move a hospital‐based manager over to the CCAC to oversee the nurses while also providing a direct line of access to the hospital that would help the nurses liaise with physicians and promote physician trust and buy‐in. This helped both hospital and community workers become part of each other's world (Program 2; 5, 6). Such collaboration, the organization leaders suggested, would not have been possible without their close partnership.

[I]t was obvious when we went into this discussion that we would include all 3 partners. That this is kind of how we roll.… Bumps have [been overcome because we've] been so comfortable saying, yeah, I'll take that on, I'll get that out of the way, I'll make that change over here, I'll absorb those costs—knowing that there was a trust factor, right, and we were all in this together. (Program 2; 5)

Genuinely liking one another was also helpful: “The 3 CEOs and the 3 vice presidents of clinical have a very highly respected relationship. It's not competitive. It's we like each other actually” (Program 2; 6).

Program 2 was not alone in enjoying a long partnership history. As a Program 1 stakeholder similarly remarked, “We know each other, we really do. And not only that, I would absolutely tell you that we like each other” (Program 1; 6). Program 4 similarly had well‐established relationships with partners, including with its very active LHIN, an important relationship in a LHIN‐wide program.

[There has been] a memo of understanding between all 9 hospital corporations … probably since 2007 or 2008. And so their way of working together is well established. They do things … [l]ike we have one lab for all of our hospitals. One lab—think about that.… All of our accounts payable is done centrally … because that has been in place and because the hospitals have worked together, they came together relatively easily. They know each other really well. They meet together monthly … it's kind of been the culture [that they] work pretty well together. (Program 4; 7)

Existing partnerships, though, did not preclude organizations from experiencing the struggles of getting everyone on the same page. While another program similarly “latched onto an existing collaborative table that had all the right partners in it already” (Program 6; 1), it nevertheless struggled to come up with a meeting schedule for the steering committee that worked for everyone, hinting at uneven leadership buy‐in across partners.

Building Trust

The IFM, as participants noted, was “not a little tiny change. This is fundamentally changing everything” (Program 4; 7). It was changing not only how the health system was funded but also how care was provided, as well as team and organizational identity. Participants felt that it was a systemic change requiring correlative culture change and the development of common practice in both structural and mundane ways.

Case Study: Program 3. Program 3 focused on cardiac surgery and involved just 2 partners, a hospital and home care provider. This was the first time the 2 organizations had formally worked together. The hospital had decided to bypass the CCAC because it wanted “to cut out the middle man and go straight to a provider who could provide us with the services that we needed without increasing administrative costs, paperwork, etc.” (Program 3; 6). Each side was perceived by the other as being motivated and enthusiastic. But their relative newness to each other was an early challenge, given that sharing patient care and funding data required trust. As a (hospital) participant noted,

Initially … there seemed to be a lot of withholding from [partner organization] on what their policies looked like.… They didn't want them to go into the hands of other service providers in the community because of competition. So they had to trust us enough that we just want to see your policies to make sure that they coincide with what we are saying so that there's not big gaps in how we would provide dressing changes or IV therapies or whatever. (Program 3; 2)

Trust was developed slowly over time, with the community clinical team visiting the hospital to meet, talk, share information, and train with their hospital counterparts while at the same time familiarizing themselves with the hospital and its patients (Program 3; 2). Eventually the hospital participants went out into the community too, a process that helped them recognize the expertise of community home care. As a community‐based participant noted, “Getting out and working elbow to elbow really had a big impact” (Program 3; 1), while her hospital counterpart noted:

[I]t was an eye‐opener to see how the nurses do the assessment, how they chart in the community, the lack of equipment at their fingertips. You know, if they want to do a blood pressure, the monitor is not just at their fingertips. You know, they've got to get it out of their bag and get it out of their car. (Program 3; 3)

Over time and through frequent meetings, the hospital and community organization were no longer on different sides of the room; instead, as one participant noted, “they made friends.… [T]he IT folks will sit together in that group. They're clarifying issues that have come up during the week” (Program 3; 2). The community partner was seen as “very open and very forthcoming with ways to electronically integrate the patient's health record” (Program 3; 6). The community partner was also involved in hiring integrated care coordinators, helping to ensure that successful candidates understood community engagement (Program 3; 2).

Across programs, stakeholders first had to overcome perceived differences in work cultures. The energy and buzz of the hospital translated into a professional abruptness for community stakeholders, who in turn frustrated their hospital counterparts with their tendency to preface each meeting with small talk (Program 2; 5). Many hospital‐based clinicians had to overcome preconceptions of the community as a “black hole” (Program 4; 6) into which patients disappeared without appropriate follow‐up, while community partners had to overcome suspicions that the hospital wanted only to dump patients on them and that the IFM would benefit the hospitals alone (Program 6; 6).

[T]he hospital people's overall message to the community paternalistically was we look after really sick, really complicated people, and we do it with very high technology. And you people are lovely and nice and sweet, but you couldn't possibly do what we do. And the community‐based people said, you guys in hospitals are a comedy of errors. You have all of these resources, all of this infrastructure, much of which you only use a few hours a day and not on weekends or after hours. [After a long process‐mapping exercise from different perspectives,] there was a big eye‐opener for many of us who said, “You do that in homes? Like you go out into places like that? You go into unsafe places at all hours of the day and night?” (Program 4; 6)

Over time, physicians began discharging patients home sooner and feeling more comfortable doing so, because “they know who they're sending the patient home with” (Program 2; 1). Trust‐building not only facilitated interprofessional working relationships but also allowed organizations to be secure in their partners’ transparency.

Developing Thoughtful Models

The development of clinical pathways and a funding model was a complicated exercise that programs negotiated by being attentive to heterogeneous perspectives across professions and sectors, tweaking the pathway in response to on‐the‐ground concerns, and encouraging innovative modes of collaboration.

Case Study: Program 4

Program 4, the largest with 15 partners, included a lead organization that had previously pioneered bundled care in its own hospital and community, inspiring the current province‐wide IFM initiative. The program also focused on COPD and CHF. Its participants were keenly aware of the challenges posed by the program's scale and concomitant differences in partner organizations’ sizes, infrastructure, and leadership commitment:

It worked really well at [lead organization]. You had a very focused leader in [name] who was very hands‐on. In [city], we're seeing something very similar. But in [organization], their CEO is less hands‐on, maybe a bit more handing it off to the clinicians. So we're watching that one carefully. And we're going to [city], a totally different leadership team, different skill sets again [with] less of a hands‐on leader. (Program 4; 7)

Despite these contextual challenges, the program had the benefit of being able to lean on its previous experience to help develop its model. The original iteration had involved stakeholders across the acute care–community spectrum working to eliminate redundant steps in the care process, represented by Post‐it notes that wound around a large room (Program 4; 6). The current initiative included an integrated funding component that required further careful consideration. The program's participants closeted themselves in a LHIN conference room on a dreary December afternoon, painstakingly working through various scenarios:

[W]e said, okay, what if volume is up 10%? Referral rate is exactly the same but just the volume is up.… [W]hat would it have meant for the hospitals, what would it have meant for the CCAC, what do we want to do? What do we do if the volume is the same… [and] everybody is on target with the referral but 2 of the hospitals are referring at a much higher rate than previously? … And then what if the volume is low? (Program 4; 1)

Clinicians were also able to provide input into program logistics: “Well, you know, this visit doesn't need to be in person. We can do it by phone first” (Program 4; 6).

Many programs transformed existing structures in innovative ways to help develop clinical pathways and correlative costing models. Program 1, for instance, set up a clinical group, a finance group, and a third group with representatives of both, so that ideas presented were also financially justified. Various strategies were used to facilitate risk assessment, from developing multiple pathways for patients based on levels of complexity, to using risk classification tools, to spending long hours in meetings with partners working through risk scenarios. Potential clinical pathways and gaps were carefully considered, along with levels of existing physician engagement and supportive technology.

The first thing we did is process mapping with a whole bunch of providers. And the process map would have wrapped around the wall.… Then what we did is we parked that and we did the same thing with patients …, then we … did it with what the best practices are. We put all three of those together and then that created this giant view of the patients’ perspective, the providers’ perspective, and the best practice, and we looked at where the gaps were and what we needed to do differently in order to rectify the gaps. (Program 6; 6)

The development of models was challenged not only by patients’ clinical and social complexity, financial prognostication based on proxy calculations, and unreliable costing information, but also by differences in the organizations’ costing platforms, the lack of a shared electronic medical record, and even differences in the way partner organizations classified the same patient. The development of the model itself was therefore fraught with anxiety, and only long, thoughtful, and inclusive discussions within and across the organizations helped stakeholders arrive at common ground.

Engaging Clinicians

Clinician engagement was typically generated by including clinicians in model development, addressing their concerns, fostering their trust in the model, and devising an engagement strategy led by formally appointed or informal clinical champions.

Case Study: Program 5

Program 5 included an acute care hospital, a chronic care and rehabilitation center, and 5 community partners with different areas of expertise. It also focused on COPD and CHF. A strong history of physician engagement provided the context for the continued, active involvement of physicians in this program. Recent surveys of physician engagement had come back with scores of 89%, setting the benchmark for the province (Program 5; 1). Perhaps unsurprisingly in this context, respirologists and cardiologists were deeply engaged in this initiative, with the chief of medicine and respirology attending meetings regularly and primary health care team leaders also involved (Program 5; 3). This strong clinician engagement was fostered by strategies that helped incentivize physician engagement. These strategies included enhancing an existing CHF clinic, setting up a COPD clinic as a carrot (Program 5; 6), and employing respiratory therapists as clinical care coordinators.

[Physicians] were very excited to think that we would have an RT [respiratory therapist] following a COPD patient into the community.… And so if the RT clinical care coordinator wanted to call the respiratory therapist, they already have that relationship.… [v]ersus, you know, a CCAC care coordinator where they don't have that relationship and don't have the confidence or the trust that they understand how to titrate oxygen or something. (Program 5; 6)

Another strategy was employing formal and informal communications to spread word about the program. A director created a formal communication plan to spread awareness of the program aimed at different stakeholders (Program 5; 6); a physician leader incorporated news of the program in weekly physician newsletters; and a case manager went to all hospital floors to let people know about it (Program 5; 2). Both frontline clinicians and executive leaders therefore saw themselves in various capacities as clinical champions.

Physician engagement varied across the programs, as did strategies to encourage it. One influential physician, for instance, mandated participation, albeit in a context in which the participants had also long seen a need for such a program:

[I]t's [physician engagement] extremely high and was mandated by me to my colleagues that it had to be extremely high. They're basically under instructions that this is what is going to happen and you have to play ball with it. But that being said, you know, everybody's happier that our patients are going home with some form of structured care and follow‐up. (Program 3, 6)

Champions developed strategies to inform others about the program, and helped negotiate challenges such as the issue of engaging family physicians, often without readily available contact information (Program 4; 4, Program 6; 7).

A well‐developed plan for informing physicians about the model helped. Program 4, for example, had a LHIN primary care lead, as well as LHIN‐wide COPD and CHF leads, who met with locally appointed hospital leads and generated a plan for presenting the IFM to their teams. They drafted letters that they sent out to primary care physicians informing them about the program. When patients were discharged, the leads followed up with a physician survey to verify that they had knowledge of the program. They also ensured that the issue of billing was tackled head‐on:

[W]e've [leadership] learned now that the first thing you have to talk about is how you bill for it. So you talk about how you bill for it first, and then they'll listen. It's true. (Program 4; 7)

For hospital‐based specialists, participation in the program required a willingness to practice differently and to provide sustained care for a patient as a member of a team rather than as a solo professional. While some physicians welcomed these changes, others remained cautious, finding it difficult to change their entrenched modes of practice. For instance, physicians now had to coordinate with other health care providers, and some were skeptical about discharging their patients early and suspicious of the quality of care they would receive in the community (Program 2; 5). In addition, some were hesitant to order medication changes before seeing the patient or to act as the most responsible physician (MRP) for the first 5 to 6 days after the patient went home (Program 6; 3). Some feared the program would take away their decision‐making autonomy (Program 4; 1), increase their workload (Program 6; 3), and affect their billing.

Other clinicians, particularly those acting as IFM coordinators, enthusiastically championed the program. In fact, in many cases, participation in the program itself engendered engagement:

[I]n a very small way, everybody here in the hospital who cares for patients started to connect differently with their community colleagues. They wrote their notes about Mrs. [X] knowing that the person who was going to care for her in the community was going to see them and read them. Like you know, she has a problem hearing in her left ear. You want to be standing on her right side. And she fell last week so just watch X or Y.… So we got a greater sort of sense of connectivity. (Program 4; 1)

Patient information such as this could be communicated electronically, in person, and through telephone conversations, as we show next.

Sharing Information

Information‐sharing within hospital and home care teams and across partner organizations, while integral to enabling integrated systems, was a challenge for most programs. An information technology infrastructure had to be put in place before patients could even be recruited. Then the patients’ data had to flow smoothly from one site to another as they moved from hospital to community. Finally, a range of metrics, from costing to readmissions, had to be collected and shared across program partners, the LHIN , and the MOHLTC, as well as with the evaluators.

Case Study: Program 6

Program 6 consisted of 4 partners, including 2 hospitals, a family health team, and a CCAC. It, too, focused on COPD and CHF. A newly piloted electronic interface was a key facilitator of connectivity for the program, as it allowed everyone across the hospital‐community spectrum to remotely monitor the patient in real time after his or her discharge (Program 6; 1). However, the program initially struggled to engage physicians, requiring additional time and creative strategies to educate them about the tool, to encourage them to work in teams, and to follow their patients after discharge.

[A]t the start of the project, they [hospitalists] got the cool factor about the dashboard. But when the team said when would you like to see this, they said, “Well, we really don't. Call us when you need us.” Which kind of perpetuates the don't call me until there's a problem and then I'll kind of do it. But the team was very expert at kind of slipping in the dashboard so the physicians could see it before there was a crisis. (Program 6; 8)

Over time, however, the technology helped garner physician buy‐in:

Now physicians have the ability to see what's been happening to that patient since they left their inpatient bed. And again, that's helping the physicians to first of all become more aware of what happens in home. And so that is new information to them. Secondly, it's also helping them become more integrated and supportive of the in‐home care team. (Program 6; 8)

The patients also had 24/7 physician access by telephone and patient education information accessible on iPads at home (Program 6; 1, 2).

Systemic connectivity fostered and was fostered by the development of a collaborative culture. The providers thought of themselves as “really one team” and sported a single piece of identification, while the patients were presented with a single consent form instead of the multiple forms and signatures needed with each provider entering their home (Program 6; 2).

Other Experiences. Program 6's ease of connectivity was not always shared by other programs. Sharing information was complicated by the need to learn and migrate to new systems (Program 3; 3), and the lack of a uniform electronic medical record (Program 1; 6) and email system across partners (Program 5; 2, 6), requiring programs to work around this in order to relay information between hospital and community. Given the lack of a common platform, programs worried about possible service duplication (Program 4; 2), discharge process discrepancies (Program 4; 5), and the disconnection of primary care (Program 5; 5). In rural settings wireless connections could be disrupted, which in any case patients might not even have at home (Program 6; 2), thereby affecting patients and providers’ ability to interact with the larger health care team. Patients may also have limited data plans (Program 2; 2), or simply feel threatened by technology (Program 6; 4). Complicating these issues was the fact that “each hospital has a little bit of a different take on the privacy regulations” (Program 4; 1), affecting the sharing of data among partners.

When in place however, technology not only facilitated seamless patient transitions but also helped build trust between hospital and community. Information about the patient could be shared in real time through a secure email system set up between the hospital and the community partner (Program 5; 5) and through videoconferencing or the Ontario Telemedicine Network. The latter could, for instance, allow a physiotherapist in the patient's home to take a picture of a wound and send it to a coordinator or physician, who could provide feedback and rapidly initiate medication changes (Program 4; 3). A couple of programs found a supportive and responsive IT group working in tandem with the clinical and decision‐making groups to be particularly helpful (Program 3; 2, 3).

Conclusions

Our article demonstrates how integration has been generated through the successful confluence of people, practice, and things as perceived and experienced by stakeholders. When able, programs harnessed existing cultures of clinician engagement and leveraged established partnerships. Yet integration can be generated even without these histories, by building trust, developing thoughtful models, creating a well‐honed strategy of engaging clinicians, and sharing technologies. Program structures (from program scale to chosen clinical condition) also contextualized and mediated integration.

On one level, our findings confirm well‐documented facilitators and barriers of integration. Clear goals, effective leadership, integrated data systems, good communication, and shared values alongside other elements are facilitators found in many integrated care analyses.42 The relationships among program components are also somewhat familiar. As a recent synthesis of integrated care programs demonstrated, strong leadership fostered team trust, collaboration, and efficient information‐sharing, and flexibility in implementation and incentives for providers led to provider commitment to the model.17 A key contribution of our study, however, is its in‐depth analysis of how integration is both generated and stymied by the interplay of contexts and mechanisms in local contexts and histories.

Advancing Theory

The concepts of connectivity and consensus allowed us to analyze different dimensions of integration and are similar to other multidimensional approaches. Hilligoss and colleagues,43 for example, examined both the “cooperative and coordinative dimensions of alignment,” conceptualized as the will and the ability to work together, while Banfield and colleagues44 wrote about the “practical elements of integration,” co‐location, and information flows existing alongside the relational and cultural. While keeping these concepts separate may be analytically useful, it is in their very confluence that more fully fledged forms of integration reside.

Our use of the concept of context in this article refers to (1) structures and cultures that existed in organizational settings before IFM initiatives were formally implemented (such as existing collaborations and information systems) as well as (2) processes and cultures encouraged and implemented during the course of the program that became increasingly taken for granted and entrenched (eg, recently developed trusting relationships, a burgeoning culture of physician engagement), which became the backdrop for new mechanisms. Here, context is often merely mechanism congealed, a temporal precedent forming the temporary structural backdrop for newly emergent structures and practices.

We have advanced an understanding of integration mechanisms as not only contextually and temporally contingent but also inherently dynamic. As such, the mechanisms of integration are understood here as the contextually circumscribed processes through which stakeholders—from hospital and community care leaders to administrators and clinicians—generated consensus and connectivity. Mechanisms facilitating connectivity and consensus were themselves often tangled and overlapping, with one form inspiring and reinforcing another. Program 6, for instance, demonstrates how technology and clinician engagement interacted with and fostered each other. Program 4 demonstrates how well‐established partnerships (forming both context and mechanism) led to clinician engagement and thoughtful model development, even as model development itself is enabled by early clinician engagement. Mechanisms, therefore, interacted in complex ways with one another as much as they did with contexts. Even though the past might structure and predispose the present, integration can be generated in the present, too, by motivated stakeholders at the strategic and opportunistic confluence of resources and reasoning. Program 3, for instance, shows how trust‐building strategies effectively fostered familiarity and trust even between 2 organizations that had not previously worked together. Programs 1 and 4 offer examples of how even programs with willing collaborators and largely trusting relationships can be frustrated by differences in practice, culture, and resources. A specific component may also be context, mechanism, and outcome at different moments in time, albeit in qualitatively different ways, as in the case of Program 2, where already existing relationships resulted in even closer bonds.

Given the recursive relationships of these concepts, we interpret the “outcome” component of context‐mechanism‐outcome configurations in holistic terms, as integration itself. Doing so allows us to circumvent the need to arbitrarily assign the label of outcome to what may be a temporary milepost that over time leads to new processes and outcomes.

Practice Implications

The ways in which disparate contexts and mechanisms coincide vary, as seen in the case studies we discussed, because of their inherent rootedness in the local. As Kodner noted while attempting to define the “imprecise hodgepodge” of integrated care with its diverse actors and meanings, “[I]ntegrated care is like a country. It demands a culture of its own.”45

Accordingly, factors such as organizational size, program scale, and patient population mediated integration, and smaller organizations with fewer resources might find it harder to implement the systems and practices of their larger counterparts, making coordination difficult. Nonetheless, among other factors, a long history of partnership might help surmount these challenges. Although health conditions with defined end points and fewer comorbidities may be arguably better suited to this model because they facilitate pathway development and project costing, chronic conditions were seen as having promise for clinical and cost efficiencies. Previous scholarship is similarly divided as to whether surgical and well‐defined care episodes or chronic conditions are better suited to a bundled care approach.46, 47, 48

Nevertheless, our article has delimited transferable implications. Building on existing partnerships, including a wide range of cross‐sectoral perspectives, ongoing recalibration of the model in response to issues faced on the ground, committed leaders, willing clinicians, trusting interprofessional relationships, and seamless systems all were key to generating integration. More significantly, this work provides transferable insights into how these factors can be fostered. They were produced, for example, by organizing coeducational workshops and cross‐sectoral clinician work‐shadowing sessions, appointing physicians in positions of power to help persuade others, strategically employing integrated care coordinators with established relationships with physicians to build trust in the model, and establishing innovative planning committees that merged IT, finance, and clinical teams to encourage both connectivity and consensus.

Finally, this study has emphasized the need to closely examine the processes that underpin, promote, and frustrate integration. As such, in addition to evaluating a program's success and implementation processes based on traditional quantitative and qualitative criteria, from length of stay and readmissions to integration facilitators and barriers, we suggest that programs might also be assessed for the very outcome they seek to implement: integration itself as experienced and interpreted by those most closely engaged in it.

Limitations and Future Directions

This work captured the experiences of IFM stakeholders as they implemented their integration programs in real time. While it paints a vivid picture of the structures and processes that fostered integration, it remains temporally frozen, capturing the early experience of rollout. While this work is significant in its own terms, a follow‐up study would help contextualize and extend these insights. We are also keenly aware that incorporating patients’ and family members’ experiences of integrated care may affect how integration is conceptualized. While this work is limited to an understanding of integration from the perspectives of organization leaders and providers, follow‐up work will also examine patients’ and family members’ perspectives of integration efforts.