PERSPECTIVE NARRATIVE

Improving care coordination is a top priority for the Veterans Health Administration (VA), the largest U.S. integrated health care system. Poor care coordination is associated with lower care quality, inadequate symptom control, and higher rates of medical errors and costs.1

Theoretical frameworks are fundamental to making systematic delivery system improvements in care coordination. They provide guidance in the design of care coordination initiatives as well as measures for assessing changes resulting from interventions. Many frameworks exist to guide practitioners in improving and evaluating the robustness of care coordination; however, application of these frameworks by delivery system managers seems limited, potentially due to insufficient opportunities to bridge the gap between theory and practice. In response, the VA State of the Art (SOTA) conference on Care Coordination brought together researchers and operational managers to assess the evidence for translating theory into practice.

Our SOTA Workgroup focused on theoretical frameworks and measures of coordination. Planning efforts were informed by a concurrent systematic rapid review (of theoretical frameworks) conducted by the VA Evidence-Based Synthesis Program (ESP), accompanied by 6 months of expert consensus-building.2 The workgroup process involved 1.5 days of in-person structured deliberation. In order to make findings more actionable, our workgroup decided to apply the theoretical frameworks to four major VA-based interventions (“use cases”) selected to represent various scopes and different stages of development, ranging from piloting to implementation.

We aim to make the theoretical frameworks identified through the SOTA a more practical resource for operational managers and researchers focused on care coordination. The use cases allow us to show how the theories discussed during the SOTA conference apply to real-life examples.

The Four Use Cases

A subset of the SOTA Workgroup familiar with the selected VA use cases (DZ, ML, DV, NA, MO) answered prompts created by the workgroup leaders to provide structured information. First, we wanted a description that established the intervention name, what motivated the need for an intervention, who and what the intervention aimed to coordinate, the stage of implementation in the VA system, specific implementation sites, evaluations conducted or planned, and results from the evaluations. Table 1 briefly summarizes this information for each of the four use cases—PIM: Patient Aligned Care Team (PACT) Intensive Management, which focuses on coordination within the primary care setting; CC-ICM: Care Coordination and Integrated Case Management, which focuses on coordination for complex patients without regard to setting; I-STEP: Improving Transplant Medication Safety through Technology and Pharmacist (original and expanded), which focuses on one aspect of coordination—medication management; and H2C: Hospital 2 Community, which focuses on transitions between care systems.

Table 1 Description of Four Use Cases
Full size table

We also asked our use case informants about what prior models—similar interventions and conceptual underpinnings—were used to inform intervention choices. PIM drew lessons from the Coleman et al. care transition model, the Camden hotspotter model, and the GRACE model for older adults.3 The CC-ICM started with a set of general principles and a toolkit from the Case Management Society of America.4 The third use case represents phased intervention design from the original I-STEP model5 to an expanded model for a broader patient population at risk of medication safety events. H2C is based on theory that highlights interactions between organizations and their environments, and the Community Partners in Care Model.6

To make the linkage between theory and practice accessible to a wide audience, we organized information about each use case into three SOTA-created conceptual domains—the context surrounding an intervention, an intervention’s locus (setting, level, and purpose), and elements of its design (mechanisms and types of interventions), as described in an ESP companion article.2 This exercise demonstrates linkages to a growing body of theory work that posits, often with empirical support, what interventions will work under what conditions. This body of theoretical work also underpins measurement development and testing, which is important for evaluating processes, intermediate outcomes, and patient experiences of care coordination interventions.

Table 2 provides an example of this domain mapping for the four use cases, and the following sections demonstrate our suggested domain-based approach to drawing useful insights from the theory (identified in the ESP review2) for these use cases. For each domain, we briefly describe what it is, note the number of theories from the ESP review that speak to the specific domain, and finally show applicability of the domain to one or more use cases through selected examples from the subset of theories available.

Table 2 Description of Use Cases Based on SOTA-Created Domains from Care Coordination Theoretical Frameworks Review
Full size table

Drawing from Theory to Inform and Interpret Use Cases: Context Domain

The context for an intervention refers to external factors that enable or thwart an intervention’s success at any stage of implementation.7, 8 At the stage of intervention design, some contextual factors may be completely impossible to change (e.g., for the PIM case, the electronic medical record system), while others, if modified, become part of the intervention instead of the context. For example, to expand I-STEP from transplant patients to other patients, the intervention team is collecting data to understand the broader context and what can be changed to avoid coordination failures that lead to medication safety events.

Based on the ESP review, 15 theoretical frameworks suggest questions to pose to understand the context for coordination. For the H2C, for example, Van Houdt’s framework suggests asking what are the interdependencies between policy for Veterans to access different providers (within and outside of the VA) and the current structures of care delivery (teams, organizations, and inter-organizational networks)?9 In turn, the theory suggests investigating factors that influence structures, such as, what information is required to manage the care of the H2C target population individually and as a group.9

The Calciolari et al. theoretical framework assesses antecedents to aligning “healthcare provision with evolving patient needs” both in the context of a single organization, such as the VA, and for coordinated “service delivery across health and social care providers.”10 For both the H2C and CC-ICM use cases, this framework points to “incentives and impediments” that influence attempts to integrate care at the organizational level: “institutional adequacy” which relates to people who can manage knowledge complexity (e.g., the different staffing models noted for CC-ICM), and resource commitment to integration (e.g., what resource commitments are available for a sustainable CC-ICM model), and “focus on results” (e.g., metrics and feedback, which may enable or present challenges for each use case situation).10, 11

Drawing from Theory to Inform and Interpret Use Cases: Locus Domain

Determinations regarding locus include deciding where the intervention is aiming to have an impact (i.e., setting including who is involved in caring for what specific patient populations, and scope in terms of levels from individuals [micro] to clinic teams [meso] to physician organizations or broader health systems [macro]) and what the intervention is trying to achieve (i.e., purpose[s]). For example, the locus of the CC-ICM intervention includes its setting which is system-wide; multiple levels including case managers or care managers (depending on patient need) who act as individuals or in teams; and purpose, which is to coordinate with both medical and social service providers and with veterans and their families. The ESP review links 28 care coordination theory articles to the locus domain.

Many of these theories offer opportunities to understand how locus drives the particular design of an intervention. For example, Gittell et al. describe how coordination communication networks can be evaluated at different levels (i.e., PIM: the meso level of activities within a team focused on primary care patients, while the others include macrolevel activities among organizations).12, 13 The Gittell et al. relational coordination theory posits that many of the same considerations for coordination operate at the different levels and in different settings, likely because of the common purpose of coordinating interdependent activity.12 Benzer et al. highlight some of the coordination challenges (and solutions) related to primary care and mental health clinical integration, a purpose that our PIM informants noted became more apparent after the intervention started enrolling patients (as a high proportion needed mental health care services).14

Theories also help identify important sets of questions related to locus for emerging interventions, such as the H2C with its broad locus of coordinating care across organizations within and outside the VA. For example, the SELFIE model includes a category “information and research,” which elicits questions about what is known about individual risk prediction of social resource needs (microlevel), opportunities for risk stratification (mesolevel), and barriers to information access (macrolevel)?15 If the purpose is to improve patient safety through better coordination, as in the case of I-STEP, the Hodgson et al. model raises questions about the patient’s need for interdisciplinary care, as a concept separate from the patient’s susceptibility to a medical mistake.16

Drawing from Theory to Inform and Interpret Use Cases: Design Domain

Design refers to how gains in coordination are anticipated or achieved. Choices for designing interventions can be approached conceptually in two complementary ways, as mechanisms for coordinating and as types of coordination. Mechanisms are action-oriented descriptions of expected cause and effect. Types of coordination differentiate aspects such as organizational and social integrations (more granularly, structural, functional, normative, interpersonal, and clinical integrations)17, or structures, processes, and outcomes of coordination.18 Another key concept is that there is not one ideal design, but rather the need to achieve a good fit between what requires coordinating and the menu of intervention options for the mechanisms and types to accomplish the results desired.17, 19 The four use cases demonstrate the range of mechanisms and types applied to the coordination challenges (Table 2).

Intervention leaders for the CC-ICM use case participated in the SOTA and have turned to several theoretical frameworks to further inform and refine their design choices. They are leveraging the Weaver et al. concept of multisystem teams to create a team of teams (e.g., the homelessness team, the primary care team, and the cardiology team) for their target populations, and plan to incorporate some of the coordinating mechanisms highlighted such as flow diagrams and tables of role responsibilities.20

For cases like PIM that depends on coordination within interdisciplinary teams as well as coordination among teams (e.g., primary care, VA specialists, and non-VA specialists), theories of team functioning (e.g., Lemieux-Charles et al.21), relational coordination (Gittell13), and programming and feedback (Young et al.22, 23) can be applied to ensure effective team functioning and coordination. Young et al. and Gittell et al. both demonstrate that both programming and feedback mechanisms of coordination are positively related to patient outcomes. Applying this finding to PIM’s coordination between primary care teams and other specialists would suggest identifying what activities can be specified in protocols, reminders, and checklists (forms of programming), and what relies on other mechanisms (such as team meetings, huddles) and processes for communication and joint problem solving (feedback).13, 22, 23 Additional guidance from Gittell would focus on interpersonal types of interventions that facilitate the development of common understanding and effective communication both within the PIM teams and between the PIM teams and other providers.13

CONCLUSION

As demonstrated by the structured tables for each use case prepared in partnership between researchers and practitioners from our SOTA Workgroup, it is feasible to set the stage for feedback loops between conceptually grounded research and practical needs of those developing and implementing interventions. Use cases have conceptual linkages that either explicitly or implicitly inform interventions, care coordination in practice, and evaluation and measurement planning. Throughout these steps, bridging theory and three conceptual domains (context, locus, and design) within the four use cases demonstrate the critical role of research and operational partnerships evident in the SOTA process. Others could take a similar approach of partnering to map their situation to the SOTA-created theory domains, and then turn to the ESP review (and other literature searches as needed) to identify applicable theories for enhancing the design and evaluation of coordination improvement interventions. To the extent that health care providers apply care coordination theory to inform and communicate findings from intervention efforts, such as the use cases described, both practice and theory-based work can advance in tandem to achieve the promise of better coordinated care.