NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

McDonald KM, Sundaram V, Bravata DM, et al. Closing the Quality Gap: A Critical Analysis of Quality Improvement Strategies (Vol. 7: Care Coordination). Rockville (MD): Agency for Healthcare Research and Quality (US); 2007 Jun. (Technical Reviews, No. 9.7.)

Cover of Closing the Quality Gap: A Critical Analysis of Quality Improvement Strategies (Vol. 7: Care Coordination)

Closing the Quality Gap: A Critical Analysis of Quality Improvement Strategies (Vol. 7: Care Coordination).

Show details

5Conceptual Frameworks and Their Application to Evaluating Care Coordination Interventions

5A. Background

As noted in the previous chapters, a diverse set of large scale care coordination projects are being planned or are underway with the support of the Center for Medicare & Medicaid Services,32, 280, 281 the Department of Veterans Affairs,147 professional organizations, and foundations.188 The vast majority of healthcare systems and managed care organizations reported have disease management programs,64 though these programs vary significantly in their design.187, 282 In addition, the research literature includes numerous studies reporting evaluations of care coordination interventions. Efforts to identify optimal strategies for coordinating care have been impeded in part by the lack of conceptual frameworks to guide the evaluation of care coordination interventions, as well as by uncertainty regarding how best to measure coordination itself.50, 93, 119, 283 To evaluate the effectiveness of interventions for improving care coordination and the validity of instruments used to measure care coordination, the concepts related to care coordination need clarification.

This chapter discusses some of the literature on conceptual frameworks and related metrics either directly or potentially applicable to care coordination. The purpose of this discussion is two-fold: to provide brief descriptions of selected potentially useful frameworks and to demonstrate how these frameworks might be used to guide development, implementation, and evaluation of care coordination interventions. Thus, the purpose of this chapter is to show that theoretical thinking from different fields has the potential to enrich the study of care coordination interventions. The goal is not to develop one integrated theory, as that is a major step beyond the scope of the current work. However, the frameworks presented in this chapter show that there are multiple alternatives to hypothesizing how an intervention might cause (or not cause) a desired effect.

5B. Methodological Approach

Similar to the methods described in detail in Chapter 3, we used iterative searches to identify literature describing conceptual frameworks and associated empirical evidence related to care coordination. We reviewed theoretical work developed in the behavioral, organizational, and health services research fields, and adapted selected frameworks to care coordination. While there are many potential frameworks from these and other fields, we chose well-established frameworks that had previously been used in or adapted to the health care setting, offered relevant concepts based on our discussions with experts in the field, and/or provided complementary ideas for understanding care coordination. Finally, we also searched for measures related to care coordination and summarized key information about some example measures and describe their relationship to the frameworks. We focused on providing information of relevance to potential decisionmakers and others involved in care coordination.

Focusing the Conceptual Frameworks on Key Decisionmakers

After our review of the types of care coordination programs underway (Chapter 2) and care coordination interventions evaluated in systematic reviews (Chapter 4), we decided to focus our conceptual framework on two levels of decisionmaking related to care coordination: system-level policymakers and service-level decisionmakers. By system-level policymakers (e.g., State Medicaid directors, Medicare officials, health plan managers), we mean individuals who have responsibility for paying for health care services for large numbers of individuals (i.e., health plan enrollees, Medicare beneficiaries) and make decisions about how to coordinate care at a system level in ways that minimize their financial risks and maximize the health care outcomes of their population of patients. By service-level decisionmakers (e.g., a primary care doctor, managers of a multi-specialty clinic), we mean individuals who are involved in providing health care services to individual patients or a panel of patients, and therefore tackle care coordination at the service delivery level. Depending upon the particular local environment, they make decisions related to care coordination to maximize health care outcomes and profit. To varying degrees, both the systems and delivery decisionmakers have shared responsibility for making care safe, effective, patient centered, timely, efficient, and equitable (the six IOM goals for quality healthcare). As we discuss specific conceptual frameworks, we will explore how each framework could help inform decisions related to care coordination for both of these types of decisionmakers.

The patient, of course, offers another key perspective to consider as we explore and apply specific frameworks. A focus group of hospitalized patients found that patients perceived that clear communication reflected good care coordination, and had varying opinions about who has responsibility for coordination (e.g., doctor, nursing supervisor, nurse, patient, etc.).284 With mounting interest in consumer-driven health plans and patient- (and family-) centered care, it is likely that the patient will have an increasingly active role in health care decisionmaking.285, 286 The patient is often the decisionmaker who experiences issues of coordination failures, so all frameworks apply to this key participant in care.

5C. Results

Malone and Crowston have argued for the utility of interdisciplinary study of coordination that draws from the fields of computer science, organizational theory, economics, psychology, management science, linguistics and biology, and have even provided a first synthesis of some common themes in relationship to information technologies.117 While we see the value of integrating across fields to develop a common conceptual framework of care coordination, we recognize that such a goal goes beyond the scope of this report. As a result, we have limited our scope to four frameworks from the fields of behavioral science, organizational design, management sciences, and health care that seem particularly relevant to the questions posed by care coordination decisionmakers (Chapter 2). We required that frameworks selected from these different fields include potential cause and effect relationships among the concepts included in the frameworks.* In the following sections, we describe each framework and the concepts that are likely to be particularly useful to decisionmakers. We then present some examples of metrics related to care coordination, and how they relate to the frameworks and evaluations of care coordination. Finally, we apply the concepts from these frameworks and metrics to our findings from the previous sections of this report to show how such an approach might be useful to those designing and evaluating care coordination interventions and programs. The intent of the rest of this chapter is to understand what factors might enable well-coordinated care in a variety of scenarios.

Model 1: The Andersen Behavior Framework

Since health care delivery relies greatly on individuals, we searched the sociology and psychology literatures for relevant frameworks and found a useful one originally developed by medical sociologist Ronald Andersen and adapted over the past 35 years by many others.287 The original framework from the 1960's lays out the some of the most salient concepts for care coordination. Originally intended to predict and explain use of health care services by individuals, the Andersen behavior model has recently been applied to model clinician response to quality-based payment incentives.288 We adapt the purpose here to focus on the coordination behaviors of health care delivery participants (including patients and clinicians, as defined in Chapter 3). We also refer the interested reader to the first report of this Closing the Quality Gap series for a descriptions of behavioral change theories.289

Figure 3 shows the initial framework with our substitution of the word “coordination” for “use.”** Deceptively simple, the framework suggests that coordination of health services relates to three concepts: the participants' predisposition to coordinate care, the resources that enable or impede coordination, and the participants' need for coordination.

Figure 3. Andersen Behavior Framework.


Figure 3. Andersen Behavior Framework.

First, some participants might be more or less predisposed to coordinate care based on their own attitudes toward or knowledge about their role in coordinating care. The idea behind predisposing characteristics is that they are not easily altered. Another predisposing characteristic in the context of care coordination might be the structure of medical professions, which set certain expectations about who has responsibility for specific care activities. Shifting major responsibility to the patient for example, for coordination of their own care, would go against the norms of some care professionals. Numerous other predisposing characteristics have been shown or hypothesized to relate to improvements in care, potentially through effects on care coordination behaviors: incentives, climate and culture, staff expertise, leadership/commitment to quality improvement,109, 290293 pre-existing team/group or inter-clinician factors (e.g., team structure, collaborative practice),136, 164, 294298 and individual clinician characteristics (e.g., knowledge, attitudes, and skills).96, 99, 140, 299301 Since predisposing characteristics are difficult to change, more resources or creativity would likely be necessary if a clinician, patient or systems-level decisionmaker wanted to reduce a barrier related to these characteristics.

Second, enabling resources reflect the availability and access to the requisite information systems, organizational structures, or productive relationships with others providing care to the same patient. Enabling resources affect the ability of a participant to respond to the need for coordination. A key distinguishing feature of enabling resources (compared to predisposing characteristics) is that they may be changed by systems- or service-level decisionmakers. Interventions to improve care coordination typically involve changes to enabling resources (e.g., introduction of a protocol for handoffs or designating a nurse as a patient navigator at the service delivery level; and implementation of contracts with disease management organizations or changing payment policies at the systems level). More details about potential enabling resources will be covered in a subsequent section on organizational theory and design.

Classification of predisposing characteristics and enabling resources is a function of point of view. For example, a doctor working within a particular healthcare system would see a lack of an information system as a predisposing characteristic since he or she alone could not change the situation. However, the leader of the same system may make a choice about whether to invest in information systems, making the same factor an enabling resource. Thus, the service-level decisionmaker and the systems-level decisionmaker will have different views of predisposing characteristics versus enabling resources.

Third is the notion of the need for coordination. In Andersen's original model the need to utilize health care is based on the patient's health and functional state, and his/her perception of need for health care. Illness is therefore typically the trigger for using health care services. In the adapted care coordination framework, we assume that one or more of the participants must perceive a need for coordinating care in order to trigger actual coordination behaviors by the participants (e.g., exchanging information between two clinicians at the delivery level; setting up a registry to flag more complicated patients for intensive case management at the delivery or systems level). The need for coordination is likely a function of the patient when we consider the health care delivery level, and of the patient population when thinking about the system level (e.g., Medicare). Patients whose health requires the participation of multiple participants (e.g., several doctors for multiple chronic conditions, a rehabilitation therapist for post-stroke care, a social worker for connecting the patient to community resources, a pharmacist to help sort out Medicare Part D benefits, etc.) need more coordination of their care.

While this model is described using a general concept of coordination, it applies to any behavior related to coordination. In viewing coordination through a behavioral model, designers and evaluators of care coordination interventions might be motivated to ask, for example, "What behaviors need to change to improve coordination between medical and non-medical services? The focus is then on a specific element of coordination - coordination across services—presumably because patient complaints perhaps stimulated a desire to change the behavior related to this element of coordination. The situational analysis then would review predisposing and enabling factors that create barriers and opportunities for achieving the specific behavior change - coordinating more effectively among participants from medical and non-medical services - in order to design and test an appropriate intervention. Thus, application of this model involves potentially focusing on discrete elements of care coordination (e.g., smooth exchange of information, efficient planning and delivery of disparate services, education of patients about the care plan, adherence to treatment) and mapping out what behaviors need to change (e.g., the physician needs to describe to patient and support staff the non-medical service needs envisioned, and the support staff person needs to take responsibility for effectively linking the patient to the appropriate non-medical resources). The choices of appropriate interventions to improve coordination are likely to be more self-evident by breaking the analysis up into discrete coordination problems.

Model 2: Donabedian's Quality Framework

Well known to those involved in health care quality research, Avedis Donabedian described a framework for assessing the quality of care that is flexible enough to apply to many situations.302 Figure 4 illustrates the intuitive relationship between three related concepts. First, structures of health care are defined as the physical and organizational aspects of care settings (e.g., facilities, equipment, personnel, operational and financial processes supporting medical care, etc). Second, the processes of patient care sit in the middle of the diagram because they rely on the structures to provide resources and mechanisms for participants to carry out patient care activities. In addition, processes are performed in order to improve patient health in terms of promoting recovery, functional restoration, survival and even patient satisfaction. This latter concept is well-known as the outcomes of medical care.

Figure 4. Donabedian's Quality Framework.


Figure 4. Donabedian's Quality Framework.

In the context of care coordination, we note that health outcomes result from the medical care delivered to the patient and the patient's underlying characteristics. In focusing on the linkage between what is under the control of the medical profession and effects patient outcomes, Donabedian's framework purposely does not account for patient, economic or social factors outside of the care delivery system. In his seminal 1966 paper, republished recently, Donabedian states: “This is justified by the assumption that one is whether what is now known to be “good” medical care has been applied. Judgments are based on considerations such as the appropriateness, completeness and redundancy of information obtained through clinical history, physical examination and diagnostic tests; justification of diagnosis and therapy; technical competence in the performance of diagnostic and therapeutic procedures, including surgery; evidence of preventive management in health and illness; coordination and continuity of care; acceptability of care to the recipient and so on.”303 Thus, the framework has coordination of care listed in the process box, meaning that care coordination is expected to be influenced by the setting and other structure variables and to exercise causal effects on patient outcomes. Another take-home point from this framework is that the positioning of care coordination implies that it is one of many important care processes, and therefore does not act in a vacuum even at the level of service delivery. In focusing on care coordination, it is easy to lose sight of this important, though relatively obvious point. Coordinating care better is only beneficial if other aspects of care delivery are optimized as well.

For a given care delivery setting—for example a small office-based physician practice—the coordination process of information exchange (e.g., test results conveyed from laboratory to physician) depends on the structures in place (e.g., information system linked with lab, fax machine). To coordinate care better, the physician may consider a structural change - purchasing an information technology solution to receive and flag results that need action, or adding staff time to perform the same function. The process could also be modified through a standard protocol to guide how the information flows, and to designate who has responsibility for each step under specific circumstances. Outcomes relevant to the information exchange process could include patient satisfaction with communication, timeliness of care, and clinical outcomes dependent on the information conveyed (e.g., better control of clotting times based on changing anticoagulant drug dosing). At the systems level—for example, an integrated health care system, the structural change might be to create an anticoagulation clinic to co-locate testing, results reporting, and clinician visits. The coordinating process would be teamwork, and the outcomes would be the same as in the first case.

Model 3: The Organizational Design Framework

The organizational theory literature offers numerous relevant concepts for thinking about care coordination, and for simplifying the complexities of the effects of the actions of multiple participants on multiple coordination parameters. Many studies outside of and within health care have focused on the effects of factors associated with organizational decisions on coordination and organizational effectiveness. However, there is not a single established framework that seems fully applicable to the questions posed by care coordination decisionmakers. Instead, several key concepts offer important lessons to consider in designing and evaluating new approaches to care coordination. To present these concepts as accessibly as possible, we anchor our discussion using a framework of formal coordinating mechanisms from organizational design research. For those decisionmakers who have a span of control within one organization that provides integrated care to patients, the organizational design framework could readily apply, and offer a way to generate potential solutions appropriate to the particular demands of a care coordination problem. In contrast, for systems-level decisionmakers whose policies affect multiple organizations, the framework we present is largely illustrative of the types of failures that could occur among or within organizations participating in patient care. The organizational theory literature describes the relationships among organizations that together produce a good or a service; however, a detailed review of this information is outside the scope of this report. We direct the interested reader to Gittell and Weiss105 for a cogent detailed description of a “multi-level framework for coordination” applied to health care.

The general organizational design framework shown in Figure 5 characterizes organizations as information-processing systems, where the flow of information among participants is a function of the demands of the situation and the capabilities of the organization to move information to where it is needed. The framework presents three concepts that underpin choices about organizational design: information requirements, information-processing capacity, and the match or fit between these.304

Figure 5. Organizational Design Framework.


Figure 5. Organizational Design Framework.

First, different situations produce variable information requirements, as shown on the left side of Figure 5. Studies within and outside of healthcare suggest that several basic characteristics of the organization's task or in the case of health care, the specific patient care activities, have important implications for designing coordination mechanisms to facilitate information flow effectively: interdependence, uncertainty and complexity of patient care activities.130, 164, 304307

  • In order to successfully perform their respective care activities, participants often rely on one another for information or other resources. As the level of interdependence among participants increases, so do the demands for information among participants.130, 138 A higher level of information flow is required in situations of reciprocal interdependence, such as for complex patients and referrals between physicians. Information flow between physicians is bi-directional, thereby increasing the demands for timely exchange of information.
  • Uncertainty is ubiquitous in health care and results from a lack of information about what will happen in the future. The course of disease or treatment for a particular patient may be unpredictable. Participants working in situations of greater unpredictability tend to need to exchange information quickly and make numerous adjustments to meet changing patient care needs.
  • Complexity relates to the amount of information required to manage a patient or group of patients. For patients with multiple chronic conditions, there are increasing needs to collect and respond to more symptom, diagnostic and monitoring information. Complexity also increases with the number of participants from different organizations, professions, or geographical location that must be engaged in care activities.308

The second key concept shown in the middle of Figure 5 is that the capacity of the organization to provide information must match, or fit the demands for information by the participants carrying out the patient care activities. In other words, the designs of structures for information-processing affect the ability of the participants to get the information they need to carry out their respective patient care activities. A care coordination intervention, therefore, needs to be appropriate for the coordination problem. While there may be multiple approaches to designing a good fit between information requirements and organizational capacity, some approaches may be more cost-effective than others.304 Designing in more expensive forms of coordination may be necessary for the most interdependent, complex and uncertain situations as shown in Figure 6. But simpler interventions such as standardization (e.g., implementation of care pathways) may be effective enough for lower interdependence situations.

Figure 6. Schematic of relationships between situational characteristics and appropriate care coordination approaches.


Figure 6. Schematic of relationships between situational characteristics and appropriate care coordination approaches.

Thus, just as the characteristics that drive the amount, timing and types of information flow required for care activities vary by patient need and other situation characteristics, organizations can be designed with differing types of information-processing capacity, the third main concept shown on the right side of Figure 5. The movement of information is a function of decisions about the structure of the organization, with three main areas that designers can change: grouping of participants, structural linking between participants and operational processes, as shown in the right rounded rectangle in Figure 5.304 It is worth noting that these areas could easily be subsumed within the Donabedian framework's structure concept, but we have additional insight from organizational design that these particular areas relate directly to facilitating information exchanges, and therefore act as coordinating mechanisms, which in turn can be described by three areas of leverage.

  • Grouping involves putting participants together (or separating them into units). A multidisciplinary clinic aggregates various specialties into one setting making it possible for information and patients to move more easily between physicians. For example, a patient diagnosed with prostate cancer might be rapidly seen by his primary care physician, a urologist, and a radiation oncologist to determine the best course of treatment. If physicians are practicing more independently, such coordination of visits might take longer and information flow might be less reliable. For example, consider an elderly patient diagnosed with breast cancer in a small community requiring oncology care in a distant tertiary care hospital. She may have to coordinate the transfer of medical records and other critical health information across geographic and institutional barriers to facilitate her care. The information requirements in both situations are similar, but the organization of participants either facilitates or impedes information flow.
  • Various formal mechanisms can be used to coordinate care across organizational boundaries, and are referred to as structural linking. Examples of these mechanisms that operate mostly at the service delivery level include designating participants as liaisons (e.g., primary care physicians often fulfill this role based on their training and professional sense of responsibility to the patient), creating a coordinating committee comprised of participants from different groups (e.g., a guidelines committee), and hiring someone into an integrator role (e.g., a case manager to facilitate efficient care for particularly complex patients). At the system level, other higher powered mechanisms might be applicable, such as addition of an organization playing an integrator role (e.g., disease management vendors), or development of management structures where participants are accountable to more than one group (e.g., employer purchasing groups exercise some authority over health care providers through voluntary quality reporting requirements, while these same providers are accountable directly to their patients).
  • Operational processes include standardization, adjustment, monitoring and organizational support, which are defined in Table 17. For example, standardization uses formalized mechanisms that pre-specify the roles, responsibilities and activities of individuals, or specify intermediate outputs, or skill sets needed for specific activities. Practice guidelines, care maps and protocols are examples of standardization (Table 17).
Table 17. Operational processes.

Table 17

Operational processes.

Empirical evidence from outside of health care has shown that information requirement characteristics (interdependence, uncertainty, and complexity) correlate in predictable ways to organizational capacity (coordinating mechanisms such as grouping, linking, and operational processes). 130 Applying these concepts to the health care setting is largely limited to observational studies that have yielded somewhat mixed results. 8991, 109, 135, 297, 309, 311, 312 Figure 6 depicts the hypothesized relationships between the underlying situation or coordination problem (square box on left side of Figure 5) and some of the coordinating mechanisms (rounded corner box on the right side of Figure 5).

Wagner's Chronic Care Model as an Example of Organizational Design. To make some of these concepts more concrete, we describe how a commonly known model for effective chronic illness care reflects concepts of the organizational design framework. Wagner and colleagues have proposed and applied a model for that relies on “productive interactions” between an “informed, activated patient” and a “prepared, proactive practice team” to produce improved functional and clinical outcomes.187 The information processing requirements of the patient relate to the goal of making him or her well-informed and able to process information appropriately. Similarly, the practice team must be prepared through receiving adequate information. The information processing capacity can be titrated to fit each patient-practice team dyad's information processing requirements, based on a range of coordinating mechanisms all pitched under the umbrella of organization of health care in Wagner's model.149


Community linkages are an example of structural linking in the organizational design terminology, and consisted of use of a designated case manager and creating interactive web sites.


Self-management support takes the form of an operational process of standardization in the case of a tool kit with tracking forms and action plans.


Delivery system re-design could reflect structural linking when telemedicine for rural patients is implemented, or grouping when nurse educator is included in a planned diabetes visit, or the operational process of monitoring with group visits.


Decision support occurs as an operational process of adjustment in the case of a system that generates regular feedback for clinical teams on guideline compliance from registry data, or simply an organizational support to help facilitate other coordination mechanisms.


Clinical information systems also reflect an organizational support.

The examples provided for each of these five mechanisms come from a collaborative “Breakthrough Series” effort to innovate across 23 diverse health care organizations - including academic medical centers, community clinics, hospital-based programs, manage care, and safety net organizations.149 Other examples given could also be easily mapped to the organizational design concepts.

Model 4: The Relational Coordination Framework

From the management sciences field, Jody Gittell has introduced a framework of relational coordination to understand the dynamics present in teamwork or collaboration.103, 104 Relational coordination aims to focus attention on relationships between participants whose awareness of the relationship of their work to the overall goals and to others involved in patient care is crucial, particularly for service organizations like health care with highly uncertain, time-sensitive, and interdependent activities. Relational coordination is characterized (and measured) by the following: frequency, timeliness, and problem-solving aspects of communication among participants in care; helpfulness; shared goals and knowledge; and mutual respect. Gittell has conducted several studies to explore the links between relational coordination, organizational design, and performance (quality and efficiency)103105, 313 Figure 7 shows Gittell's Framework of Relational Coordination, and illustrates some of the hypothesized linkages. In particular, variability in outcomes from different interventions to improve care coordination may be due in part to differences in the effectiveness of these interventions in improving relationships among interdependent clinicians.294, 313

Figure 7. Relational Coordination Framework.


Figure 7. Relational Coordination Framework.

Several studies have investigated the relational aspects of coordination. Relational coordination has been linked to higher patient-perceived quality of care and reductions in length of stay for joint arthroplasty in hospital orthopedic departments.104, 313 In the hospital setting, Shortell et al. found that higher reported quality of caregiver interactions in intensive care units was strongly associated with lower risk-adjusted LOS, lower nurse turnover, perceived technical quality of care, and perceived ability to meet family member needs, but was not associated with risk-adjusted mortality.126 Studies of nurse-physician collaboration by Baggs et al, also in the intensive care setting, suggest that better interprofessional collaboration as reported by nurses may be associated with better patient outcomes314 and provider satisfaction.315 Finally, in a study of nurse, respiratory therapist, and physician collaboration in neonatal intensive care units, lower rates of certain morbidities were associated with higher collaboration but varied by clinician group, while lower mortality rates were associated with better respiratory therapist-reported collaboration only. When the collaborative scores of all clinician groups were evaluated simultaneously, however, the relationship between collaboration and mortality failed to reach statistical significance.124

Summary of Concepts From Frameworks

Table 18 summarizes the concepts from the four frameworks, organizing concepts into three general areas: baseline assessment of the setting, patient population, and other factors that might influence the amount of coordination needed; coordinating mechanisms related to each framework; and outcomes aimed for in each framework. Grouping the concepts into three main analytic areas may oversimplify the potential relationships between concepts. However, for those who are designing or evaluating a new intervention to improve care coordination, we recognize the need to have an analytic framework that provides a starting point (baseline assessment of current situation—depicted in each of the included figures of the conceptual frameworks by the boxes with square corners), options for interventions (coordinating mechanisms—depicted in each of the figures by boxes with rounded corners), and outcomes to monitor (depicted by shaded diamonds). Each of the concepts from the four frameworks applies to developing or studying a given approach to care coordination.

Table 18. Summary of relationship of concepts across frameworks.

Table 18

Summary of relationship of concepts across frameworks.

Thus, a general approach to applying concepts from theoretical frameworks involves: 1) assessment of the needs for coordination by reviewing baseline characteristics for a given practice setting and patient population, 2) identification of the options for improving coordination by reviewing potential coordination mechanisms and considering their fit with the demands of the particular circumstances, 3) selection and implementation of one of the alternatives, 4) evaluation to determine effects on coordination and outcomes of care, and 5) iteration if needed to test alternative solutions. Such application of the frameworks presented in this chapter may provide a useful way for systems level decisionmakers to characterize and assess specific approaches embedded in demonstration projects and used within health care delivery organizations. Likewise, the components list (described in Chapter 3) offers an approach for decomposing interventions at the clinician-patient service delivery level.

Measures Related to Care Coordination

Assessments of care coordination interventions report five types of measures: patient outcomes, cost outcomes, care delivery process measures, coordination mechanism measures, and patient/family perception of coordination. Both patient and cost outcomes measures (mortality, morbidity, functional status, costs, etc) are the end goals for improvements in care coordination. Assessing these outcomes is important in all evaluations of care coordination interventions. Care delivery processes generally capture the occurrence of recommended care activities that are expected to arise from appropriately coordinated work. Measures of care delivery processes are often intended to identify whether care practices (e.g., patient follow-up visits, intensification of medication) occurred in accordance with recommended guidelines.4, 132, 316 At the same time, they provide limited insight into the processes that facilitated the appropriate performance of these activities. In addition, guidelines upon which these measures are based are often disease-specific, and provide little information about how care is negotiated to manage multiple conditions.317

The last two categories of measures (coordination mechanisms and patient perception of coordination) relate more specifically to care coordination.

Table 19 illustrates examples of some of these measurement tools found in the literature and how they map to concepts from the frameworks. The measures in the table are subsidiary to outcomes measures, but are important for intervention design to determine what features of a design contribute to improvements in coordination. These measures need to be used along with outcomes measures to provide a full picture of the effectiveness of a care coordination intervention. The subsequent sections highlight some of the important methodological challenges in measuring concepts specifically related to care coordination.

Table 19. Instruments and measures related to care coordination mechanisms or patient/family perception of coordination.

Table 19

Instruments and measures related to care coordination mechanisms or patient/family perception of coordination.

Table 20. Suggested approaches for improving care coordination.

Table 20

Suggested approaches for improving care coordination.

Coordination mechanism measures reported in the literature focus on either measurement of information exchanges, relational coordination among participants, or enabling resources present in the care setting. Information exchange and relational coordination measures ideally assess both the occurrence of information transfer among participants and recognition or awareness of relevant information by the decisionmaker. Such measures would help test whether a particular coordinating mechanism used in organizational design matched the information-processing requirements of the patient care situation. Similarly, measures of relational processes would be able to assess both the occurrence of interaction among participants, and common understanding of care activities and individual roles in delivering care to help identify issues related to relational coordination.

Direct observation of these processes poses significant methodological and data collection challenges, and indirect (but more easily gathered) measures have therefore typically been used. Measures of clinical information exchange include use of medical record audits to identify written or reported evidence of information transfer (e.g., note in medical record of physician knowledge of other physician's involvement in patient care96, 143). Measures of relational processes have often relied on self-report by team members, which may or may not reflect actual collaborative practices. Further research is needed to understand how differences in perceptions of collaboration150, 160, 162, 314 and specific components of collaborative interactions136 may affect delivery of care.

Development of measures for interprofessional collaboration has also generally been conducted within the acute care settings where either teams or organizational units are well-defined. More recent efforts have attempted to measure collaboration in other settings, such as integrated long-term care settings.129 Given existing methodological and data collection challenges in measuring collaboration in well-defined units or teams, it is as yet unclear how clinician report-based measurement efforts may be extended to settings where interdependent clinicians are more loosely affiliated, not aware of one another, or not easily identified.

Patient-reported perceptions of coordination provide a proxy measure for the overall coordination performance of providers. In the setting of care transitions and often fragmented chronic illness care, patients are recognized as potentially the only “common thread” linking interdependent clinicians and settings318 and may represent the only perspective (and data source) from which coordination of care may be measured. These measures are also more aligned with a patient-centered focus in health care quality. They can be meaningful, for example, in identifying that patients are getting conflicting advice that is not resolved (e.g., poly pharmacy). However, patients are unlikely to be aware of many of the specific activities coordinated in their care.94, 162 As a result, these measures may provide limited value in identifying and monitoring the specific processes that interventions to improve coordination might seek to change.

Given the relative strengths and limitations of these approaches to measurement, it seems likely that use of a combination of these measurement approaches within studies (e.g, clinician and patient report of coordination, direct and indirect measures) are needed to achieve a more comprehensive understanding of care coordination.

For example, an intervention designer testing a collaborative care model might want to use two instruments to assess the perspectives of different participants—one for the patient and another instrument for the providers of care—in order assess whether the intervention functioned as expected. Without such an assessment, it is impossible to tease apart a lack of effectiveness in achieving collaboration versus a lack of impact on patient outcomes from effective collaboration. Thus, during the evaluation phase for a new intervention, the implementer could survey patients with Glasgow et al's “Patient Assessment of Chronic Illness Care” instrument, which includes items such as “Over the past 6 months, I was asked how my visits with other doctors were going” and “I was satisfied that my care was well organized”. For members of the collaborative care team, the implementer could use or adapt the survey instrument developed by Temkin-Greener and colleagues to assess interdisciplinary team processes and performance, which includes items related to relational coordination such as “When team members talk, we understand each other,” and “Others in my team have a good understanding of patient care plans and goals.” If the decisionmaking between two types of team members (e.g., nurse and physician) is considered particularly important, Baggs' "Collaboration and Satisfaction About Care Decisions (CSACD) would be another choice for an instrument. The findings from these surveys could help the intervention designer determine weaknesses in the intervention, that once ameliorated would increase the chances that the intervention would improve patient outcomes.

5D. Summary Answers to Key Questions

Research Question 8: What Concepts Are Important To Understand and Relate to Each Other for Evaluations of Care Coordination? What Conceptual Frameworks Could be Applied To Support Development and Evaluation of Strategies To Improve Care Coordination?

We identified four well-established frameworks that complement each other in terms of developing and studying care coordination interventions and programs. Taken together, the frameworks include a dozen concepts generally fitting into one of three domains: baseline assessment of the specific patient care situation, coordination mechanisms, and outcomes of care. The exact relationships between concepts (e.g., how much of the variation in use of health services is explained by enabling resources like availability of a clinic) is fairly well-developed for the original specification of the frameworks; however extensions of these frameworks to care coordination (e.g., how much of the variation in care coordination behaviors by the participants is related to predisposing characteristics like attitude toward collaboration) will need to be studied carefully.

These frameworks for care coordination provide evaluators of new interventions with a guide to understanding the relationships and connections between an intervention and patient outcomes. Developers and evaluators of interventions to improve coordination need to ask:

  • What are the coordination needs related to patient care? At the service level, this might entail an initial assessment of an individual patient that determine what needs to be coordinated based on the level of complexity and uncertainty related to the patient's clinical condition, insurance coverage, preferences, family support and other situation-specific factors. At the system level, this question could be posed for a population with an assessment of the range of coordination needs anticipated.
  • Who are the participants in care, and how are they dependent on each other for a given care situation? At the service level, the participants might be a primary care physician, office staff, and an adult patient who has several chronic illnesses and is also seeing two specialists. At the system level, care of a targeted group of high cost Medicare beneficiaries may include numerous participants, with varying levels of dependence on each other for information and services.
  • What are the enabling factors already in place (e.g., personnel resources, information systems)? Does the intervention or a part of it aim to add a new enabler (e.g., quality improvement strategy such as provider reminders) expected to improve coordination?
  • What are the predisposing factors that influence the motivation of those involved in coordination (e.g., attitudes, incentives)?
  • How is the intervention expected to change the coordination process of informational exchange? In other words, how does the intervention movement of necessary information across interfaces, such as different settings of care?
  • How is the intervention expected to change the coordination process of relational awareness? In other words, what does the intervention do to improve participant's understanding of the relationship of one individual's work to the overall goals and to that of others involved in patient care?
  • How are the interactions of these factors and coordination processes expected to affect clinical processes and patient outcomes (e.g., what is the hypothesis about why the intervention will work)?

Research Question 9: What Measures Have Been Used To Assess Care Coordination?

Studies of care coordination have evaluated patient outcomes, including changes in mortality, symptoms, unemployment, staying connected to services, and adherence to medication. Cost and utilization outcomes, including hospitalizations, emergency department visits, and clinic visits were included in a number of studies. Also, patient and family satisfaction were reported in some instances.

We also separately searched the literature for instrument development related to care coordination, and found 20 instruments and approaches. About half of the instruments are targeted at patient and family members, and ask about perceptions of care, including items about coordination (e.g., “treatment was planned with appropriated considerations of previous course of the disease”,321 “told me which nurse was primarily responsible for coordinating my care”).322 Two of the instruments derive their data from chart reviews to assess the information exchanged between physicians. Seven instruments survey physicians or members of a defined care team to assess collaboration and teamwork processes and performance. Two instruments evaluate resources and structures (e.g., community linkages) that support care coordination. One of these instruments is for systems that care for adults with chronic illness, and the other is for primary care practices that have adopted a “medical home” approach to pediatric care.

The measurement field related to care coordination is in the early phases of its development. It is as yet unclear what approach or combination of approaches to measurement will adequately capture the processes driving an intervention's effect, particularly outside well-defined care settings, where the challenges for coordination are most salient to the patient and families.

Research Question 10: How do These Frameworks Relate to Quality Improvement Strategies Evaluated in the Previous Closing the Quality Gap Series Reports?

The IOM Priorities Report6 highlighted care coordination as a cross-cutting topic, meaning that it related to the other areas prioritized for national action. As a result, the relationship between the conceptual frameworks for care coordination and our previous work on quality improvement strategies for some of the IOM priority conditions (e.g., hypertension, diabetes, asthma, etc.) merits some exploration. The quality improvement strategies evaluated in the previous reports from our Closing the Quality Gap series include patient education, self management, provider education, provider reminders, audit and feedback, relay of clinical data, organizational change (including disease management and case management), financial and regulatory incentives and are relevant to care coordination.325 Most of these strategies have been shown to improve health outcomes in randomized controlled trials or other fairly rigorous comparative study designs. They are often used in packages of several strategies together, so assessing the essential component or components is often not feasible.

These strategies share the objective of improving care through changing patient, provider or organizational behavior. To the extent that they influence behavior, they are most easily mapped into the Andersen behavior framework as changes to predisposing or enabling factors (e.g., financial incentives to alter a predisposing characteristic—one's underlying motivation, or provider education to enhance skills as an enabling resource for improving quality of care). In addition, many of the strategies relate to the organizational design and relational coordination frameworks (e.g., provider reminders as an operational process that improves information transfer; patient education and self-management aimed at enhancing communication between patient and physician). Finally, the organizational change quality improvement strategies are synonymous with care coordination interventions (e.g., case management, disease management, creation of multidisciplinary teams), based on our working definition.



We have focused our review on frameworks that might help intervention designers analyze their situation and tailor solutions based on an understanding of the cause and effect relationship of various concepts. As a result, we have not included many current models that bundle interventions together into holistic approaches to improve coordination. For example, the Wagner Chronic Care model has been applied widely as a package of approaches to activating patients and developing effective health care teams with adequate support from the community and the health care system. In the section on the third framework from organizational sciences, we describe how the Wagner model is consistent with organizational design concepts.


In each of the figures describing the conceptual frameworks relevant to care coordination, we adopt the following graphical convention: We use a box with square corners to depict the concepts related to the baseline assessment of the care coordination setting, patient population, and other existing factors in the health care environment. We use a box with rounded corners to depict the coordinating mechanisms, and we use a shaded diamond to depict the outcomes of care coordination.


  • PubReader
  • Print View
  • Cite this Page
  • PDF version of this title (977K)

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...