State of IS Integration in the Context of Patient-Centered Care: A Network Analysis and Research Directions

For more than a decade, healthcare reform has emphasized coordinated “patient-centered care”. To that end, policymakers have invested in integration of healthcare providers’ information flows. Research has studied healthcare providers’ information needs but overlooked communicative exchanges among participants in coordinating treatment plan decisions. Consequently, although medical literature asserts that patients should depend on information exchange with healthcare providers to enable participation in treatment plan decisions, the assertion has not been tested. In this paper, the authors conduct an empirical study to elucidate the structure of actors’ communications in support of their information dependencies. The findings illustrate that although patients are well connected through personal contact with healthcare providers, patients are disenfranchised from integrated healthcare information systems (IS) and the potential of IS to support patients’ participation in coordinated “patient-centered care” decisions. Furthermore, knowledge asymmetry between patients and healthcare providers should be considered in the selection and design of healthcare IS. information systems (IS) should support the coordination of essential information from all pertinent actors. However, we found that the assertion has not been systematically assessed in prior research. Therefore, we undertook a systematic empirical study to develop and substantiate a conceptual framework of information dependencies and patterns of information DOI: 10.4018/jhisi.2011010101 2 International Journal of Healthcare Information Systems and Informatics, 6(1), 1-18, January-March 2011 Copyright © 2011, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited. flows in the context of patient-centered care for diabetics in one healthcare system. We develop the initial conceptual framework based on a review of the literature and refined through expert informants. We then substantiate the conceptual framework through face-to-face interviews with primary care physicians, clinical diabetes educators, and diabetes specialists in context. The findings shed new light on the state of IS integration in the context of patient-centered care, and highlights implications of providing patients with access to electronic medical records that should be considered in the selection and design of healthcare IS. The basic premise of our study is that a mantra of healthcare reform for more than a decade has been “patient-centered care” in which patient-specific treatment plan decisions should be coordinated among all of the pertinent actors – patients and healthcare providers – over the course of patients’ lifecycles (ACP, 2007; Bergeson & Dean, 2006; Porter & Teisberg, 2007; Wagner et al., 1996). A central tenet of patient-centered care is that integrated IS should support the coordination of essential information from all pertinent actors. Integrated information flows are essential to coordinating treatment plan decisions over the course of patients’ lifecycles. The process of enacting patient-centered care involves diagnosing symptoms, identifying potential treatments, projecting possible health outcomes and deciding on plans of action (D’Cruz, 2008). In the process, healthcare providers and patients often face uncertainty for two reasons (Charles et al., 1997; Whitney et al., 2004). Firstly, symptoms are often indicative of numerous and potentially interacting medical conditions that are difficult to distinguish with certainty. Secondly, several alternative treatment options often exist with different possible impacts on patients’ physical and psychological wellbeing and no certain right or wrong answer. Providers and patients seek information, therefore, to help them to overcome uncertainty and decide on plans of action (Charles et al., 1997; Whitney et al., 2004; Daft & Lengel, 1986; Goodhue et al., 1992; Wybo & Goodhue, 1995). However, “while the healthcare provider possesses better knowledge regarding the expected effectiveness of health care in improving health status, the individual [patient] knows best how improvements in health status affect his or her wellbeing” (Hurley et al., 1992, p. 4). Therefore, to make coordinated patient-centered treatment plan decisions requires the integration of healthcare providers’ technical knowledge and patients’ intimate knowledge of their wellbeing (Charles et al., 1997; Flynn et al., 2006; Gafni et al., 1998; Hurley et al., 1992; Von Korff et al., 1997). To that end, “information exchange between patients and health professionals is fundamental to achieving patient participation in decision-making...[which] requires the exchange of all information relevant to decisionmaking” (Bugge et al., 2006, p. 2065). Because they depend on each other for information essential to treatment plan decisions, patients and healthcare providers can be viewed as a network of interdependent actors. Prior studies have conceptualized the healthcare system as a network of actors, each of which possesses specialized knowledge in support of treatment plan decisions (e.g., Ellingsen & Obstfelder, 2007; Hanseth et al., 2006; McGrath, 2002; Ramiller, 2007; Timpka et al., 2007). Specialization is necessary because individual actors are unable to acquire, store and process information in all areas of knowledge (Grant, 1996). Actors lacking resources, such as information related to specialized knowledge, on which they depend to achieve objectives, such as deciding a treatment plan, coalesce into social networks (Borgatti & Cross 2003; Gasson, 2006; Tillquist et al., 2002). Social networks consist of interdependent human actors with links or ties that facilitate coordination, i.e., the exchange of resources, such as information, through actor-to-actor interactions. The structure of a knowledge-intensive social network such as a healthcare system is constituted in the patterns of actors’ dependencies and the communicative exchanges they enact to coordinate the flow of information. Communicative exchanges and information flows between actors in the social network International Journal of Healthcare Information Systems and Informatics, 6(1), 1-18, January-March 2011 3 Copyright © 2011, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited. can occur through (1) personal contact, such as face-to-face and telephone communication (e.g., Coiera, 2000), and/or (2) IS-supported contact, such as email communication, and Internet applications and mobile devices that facilitate the recording, transmission and analysis of biological information such as blood sugar and blood pressure levels between patients and their healthcare providers (e.g., Korhonen & Bardram, 2004; Steinbrook, 2008). The literature suggests that provider-to-provider communication in support of clinical practice is typically enacted through both personal contact and IS-supported contact (e.g., Coiera, 2000). Other studies advocate for both personal contact and IS-supported communication between patients and healthcare providers in support of patient-centered care (e.g., Bergeson & Dean, 2006). An understanding of the structure of networks constituted in the information flows between actors is considered essential to informing design and selection of IS that can support actors’ information requirements, and assessing “whether the systems we build will work the way we want them to in context” (Tillquist et al., 2002, p. 92). However, prior literature has analyzed the information requirements of healthcare organizations and individual actors without taking into account information exchange constituted through communicative interaction among all actors pertinent to care decisions (e.g., Clarke et al., 2003; Grimson et al., 2000; Huq & Martin, 2006; Keshavjee et al., 2006; Khoumbati et al., 2006; Shakir & Viehland, 2005; van Merode et al., 2004; Xu et al., 2000). Therefore, notwithstanding the “patient-centered care” mantra of healthcare reform, we still do not understand the structure of the network in support of patient-centered care as constituted in information flows among all pertinent actors. To ameliorate this gap, we undertook an empirical study in Ontario, Canada to elucidate the nature of actors’ communications in support of their information dependencies and the consequent structure of the patient-centered healthcare social network. The context of healthcare for Canadians with diabetes provides a poignant example of patient-centered care that relies on the active participation of patients and numerous healthcare providers in treatment plan decisions. As discussed in this paper, scholars find that active participation of diabetic and pre-diabetic (i.e., at risk) persons can substantially mitigate escalating medical complications and associated treatment costs (Homer et al., 2004; Testa & Simonson, 1998; Wagner et al., 2001) – costs that amounted to CAD $6 billion of the $54 billion Canada spent on all public healthcare in 2000 (PTMH, 2000). However, patients’ active involvement in treatment plan decisions must be informed through communication with healthcare specialists. Such a flow of information among actors within a patient-centered healthcare system has been advocated (e.g., Bergeson & Dean, 2006; Porter & Teisberg, 2007; Wagner et al., 1996) but has not been systematically assessed. This paper proceeds as follows. First, we elaborate on the nature of patient-centered healthcare information flow among actors within the context of the chronic condition diabetes mellitus. This provides the basis to draw on social network theory to postulate hypotheses regarding how actors communicate in support of coordinated treatment plan decisions. We then outline the conceptual framework and methodology adopted to test the stated hypothesis, and follow with an analysis and discussion of findings. The paper concludes with implications for researchers and practitioners in the quest to improve information flows in support of patient-centered care. THEORETICAL FOUNDATION The context of patient-centered care can be conceptualized as a knowledge-intensive social network in which information essential to deciding treatment plans is distributed among actors with specialized knowledge, and effective patient-centered care depends on coordinating the exchange of information among all pertinent actors. As depicted in Figure 1, Abidi (2008) asserts that an understanding of information 4 International Journal of Healthcare Information Systems and Informatics, 6(1), 1-18, January-March 2011 Copyright © 2011, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited. coordination in support of healthcare delivery requires analysis of the stakeholders, the activities they undertake to achieve their healthcare goals (i.e., workflow), the knowledge sets that are the source of information required for care decisions, and the communications and technologies that support information flows. Patient-centered care decisions depend on the exchange of information about these dimensions and held within each dimension. The exchange of information through communicative interactions is a dynamic process. Factors influencing the dynamic process of information exchange are depicted in the Borgatti and Cross (2003) dynamic model of learning, as depicted in Figure 2. The model states that the success of information exchange is a function of the dynamic process of asking for the information and of relational conditions between the actors. The process of asking for information and providing information links actors in the network. The nature of actors’ information requirements (i.e., dependencies) provides an impetus for a relationship or tie between actors. Relational ties, in turn, affect the success of the information exchange. We use the case of healthcare for diabetics to highlight the significance of information flow among actors in support of coordinated treatment plan decisions. Diabetes mellitus is a chronic condition that tends to amplify medical complications over patients’ lifecycles (D’Cruz, 2008). Patients diagnosed as pre-diabetic (i.e., “at-risk”) bring their risk factors under control through dietary changes, exercise and periodic monitoring of their condition. Undiagnosed diabetics in the population are at severe risk of escalating medical complications because they are unaware of their need to control their blood glucose, blood pressure and cholesterol, for example, to mitigate escalation of diabetes. The diagnostic process involves physicians, nurses and medical lab tests among others. Patients diagnosed with diabetes require regular monitoring, pharmaceuticals such as insulin, and consultations with physicians in an effort to prevent the onset of diabetes-related medical complications such as stroke, heart attack, vision problems and foot disorders (O’Reilly et al., 2007). Patients with escalated diabetes experience diabetes-related organ damage that can lead to escalated medical complications, disabilities and death. Patients’ overarching objective is to manage their diabetic risks and conditions in order to mitigate escalation of medical complications over time. Escalation of diabetes-related conditions can be substantially mitigated provided that patients implement dietary changes, regular blood glucose monitoring and insulin treatment, for example (ADA, 2003; Homer et al., 2004; Testa & Simonson, 1998; Wagner et al., 2001). Management of diabetes is necessarily patient-centered because substantial segments of the management — diet, exercise, selfmonitoring and medication use, for example Figure 1. Healthcare Knowledge Management dimensions (adapted from Abidi, 2008) International Journal of Healthcare Information Systems and Informatics, 6(1), 1-18, January-March 2011 5 Copyright © 2011, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited. — rely on the actions of patients over their lifetimes (Bodenheimer et al., 2002; Porter & Teisberg, 2007; Wagner et al., 1996). However, numerous healthcare providers with specialized knowledge (e.g., physicians, nurses, dieticians, medical laboratory technicians, pharmacists) play a role in informing treatment plan decisions (D’Cruz, 2008). Thus, patient-centered care depends on knowledge held by patients, providers, organizations, the medical community and knowledge about medical resources, processes, metrics used to assess healthcare delivery, and the network of actor relationships (Abidi, 2008). Actors involved in patient-centered healthcare (i.e., patient and healthcare providers) share their information towards coordinated treatment plans for a specific patient. Within this context, actors communicate with each other to reduce uncertainty, thereby making their decision environments more predictable (Te’eni, 2001). Communication among networked actors supplements information exchange by providing opportunities for clarification and sense-making regarding the potential impacts of alternative treatment strategies for a specific patient (Grabowski & Roberts, 1999). An important aspect of social network theory is the presence or absence of ties that support communication among actors, measured by means of network density and connectivity, and the structure of ties has a direct link with how information is exchanged (Borgatti & Cross, 2003; Marsden, 1990). Communication and information exchange among the actors benefits from network closure, referring to a dense network in which all actors are cohesively connected to each other (Coleman, 1988). Here, information exchange is leveraged when actors tend to “close” the network by creating and maintaining a dense network of relationships. The closure of the network structure encourages every individual to play an active role in the network and to add to its richness by exchanging his or her information (Akgun et al., 2005; Lee & Choi, 2003). Information exchange between actors in the social network can occur through personal contact and through IS-supported contact. The literature suggests that healthcare providers typically exchange information with each other in support of clinical practice by means of both personal contact and IS-supported contact (e.g., Coiera, 2000). Other studies advocate for both personal contact and IS-supported information exchange between patients and healthcare providers. For example, the typical patient’s physician office visit involves face-to-face communication (i.e., personal contact) but “can only address a portion of patients’ needs” (Johnson & Ambrose, 2006, p. 109). Physicians may not have time, for instance, to explain that an antibiotic prescribed to be taken three times a day must be taken every eight hours in order to maintain the desired blood levels of the medication. Hence, “a major cause of noncompliance is lack of comprehension of the treatment, fueled by lack of information” (Johnson & Ambrose, 2006, p. 110). To complement information exchange in face-to-face physician office visits, the literature advocates the use of ubiquitous healthcare information systems (UHIS) (e.g., Bonato, 2003; Dishman, 2004; Korhonen & Bardram, 2004; Steinbrook, 2008). UHIS refer Figure 2. Dynamic Model of Learning in Intentional Search (adapted from Borgatti & Cross, 2003) 6 International Journal of Healthcare Information Systems and Informatics, 6(1), 1-18, January-March 2011 Copyright © 2011, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited. to pervasive and ubiquitous consumer technologies such as email, Internet, mobile phones and PDAs (personal digital assistants) that facilitate information flows between patients and their healthcare providers in support of prevention, diagnosis, and treatment. For example, Internet applications such as Microsoft HealthVault and Google Health enable patients to record information about their medical indications (e.g., weight and blood pressure) over time and aggregate data from healthcare providers in order to provide up-to-date information to physicians, specialists and other healthcare providers in support of diagnosis and treatment (Steinbrook, 2008). PDAs and mobile phones offer a mobile platform for logging biological information (e.g., blood glucose, pressure) and subsequently providing real-time guidance to the patient in support of compliance (e.g., alerting the patient that he/she needs insulin) and/or submitting the information wirelessly to healthcare providers in support of diagnoses and treatment plan decisions (Korhonen & Bardram, 2004). Email enables patients to follow-up their physician office visits with a quick question (Practice Solutions, 2009). Based on the preceding theoretical foundation, within the context of patient-centered health care, we can state the following hypotheses: Hypothesis 1: Healthcare providers in a patientcentered context maintain direct ties with each other through (a) personal contact and (b) IS-supported contact in order to exchange information in support of coordinated treatment plan decisions. Hypothesis 2: Patients and healthcare providers in a patient-centered context maintain direct ties with each other through (a) personal contact and (b) IS-supported contact in order to exchange information in support of coordinated treatment plan decisions. In order to assess our hypotheses, we undertook an interpretive field study (Klein & Meyers, 1999) in the context of healthcare for diabetics in Ontario, Canada. METHODOLOGY The purpose of an interpretive field study is to develop researchers’ understanding within subjects’ context (Klein & Meyers, 1999). It enables us to develop a rich understanding of the context of information exchange among actors in support of patient-centered care for diabetics. Interpretive study is guided by and couched within a theoretical framework that informs researchers’ initial conceptualization. To that end, we developed an initial conceptual framework of the actors involved in patient-centered care for diabetics, their information dependencies and their pattern of interactions to satisfy their dependencies through both literature survey and expert informants. We subsequently refined and substantiated the conceptual framework through face-to-face interviews with primary healthcare providers in an iterative process that continued until consensus emerged among the interview subjects. The resulting conceptual framework, in the form of a dependency network diagram, emerged from interviews with seven expert healthcare providers: two primary care physicians, a clinical pharmacist, two certified diabetes educators, and two diabetes specialists. All of the informants were experts that had both clinical and research experience in the context of information flows in support of patient-centered care for diabetics. The protocol employed in our research is elaborated as follows. Interpretive Field Study: Conceptual Framework Tillquist et al. (2002) draw upon a rich history of engineering data flow diagrams, process models, entity-relationship diagrams, and state transition diagrams to specify a representation methodology, dependency network diagramming (DND), which enables detailed qualitative analysis of actors’ information dependencies and interactions that constitute the structure of a network. Therefore, we adopted the protocol of dependency network diagramming to develop our conceptual framework. In the DND protocol, researchers identify roles and actions International Journal of Healthcare Information Systems and Informatics, 6(1), 1-18, January-March 2011 7 Copyright © 2011, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited. that collectively refer to the nature of actors’ purposeful participation in the actor network. A role, encapsulates a set of actions performed by an actor or actors to achieve a goal. An action is the means or procedure for the manipulation of resources to achieve a desirable or suitable goal. A resource is anything perceived as valuable by actors in support of performing their roles, such as information, specialized skills, materials, tools or equipment. When actors are unable to achieve their goals without relying on the outputs, resources or skills of another role, a dependency exits between the roles. A dependency forms the impetus for actors to establish network ties. Coordination refers to how actors interact and exchange information (i.e., communicate) to satisfy dependencies, thereby forming ties between actors. DND arranges these constructs – roles (actors, activities, goals), dependencies and coordination – diagrammatically to depict the structure of a network. We followed the protocol specified by Tillquist et al. (2002) for developing a DND that is parsimonious yet sufficiently detailed to facilitate analysis, as follows. 1. Scope: The scope of the diagram should encompass the roles involving communication to satisfy actors’ patient-specific information dependencies. 2. Activities: Concentrate on the essential units of activities for the purpose of analyzing dependencies and communications. Thus, we modeled a composite of activities as a single activity when they produced a single goal within the same role. Otherwise, we modeled activities separately. 3. Goals: Combine like goals that utilize the same supporting activities within a role. 4. Dependencies: Depict only one dependency from one role to another unless the dependencies are unrelated. In order to identify the components of DND – specifically the actors, their roles, actions, goals and information exchanges – we developed and refined the initial DND diagram in four phases. First, we conducted a detailed search of the medical literature (e.g., including but not limited to Annals of Internal Medicine, Diabetes Care, Journal of the American Medical Association) and medical informatics literature (e.g., including but not limited to International Journal of Medical Informatics, Journal of the American Medical Informatics Association, Journal of Biomedical Informatics). Second, two rounds of semi-structured face-to-face interviews were conducted with a primary care physician who has been involved in design and development of medical information systems. Third, we walked through the DND with a medical specialist in chronic diseases during a one-hour face-to-face interview. Fourth, the DND was sent to an independent expert – a clinical pharmacist experienced in care for patients with diabetes – for review and comments. We concluded the process of developing the DND when no further changes were identified (Miles and Huberman 1994). With our initial conceptualization informed by the literature and by expert informants, we undertook the next phase of interpretative field study. Interpretive Field Study: Face-to-Face Interviews Following the methodology of interpretive field study, we next sought to enhance and refine our initial conceptualization within subjects’ context (Klein & Meyers, 1999). To that end, we refined and substantiated the DND through in-context face-to-face interviews with each of the primary healthcare providers involved in advising patients: a primary care physician, two certified diabetes educators, and two diabetes specialists. Our first interview subject was a primary care physician because the primary care physician is considered the most significant actor in patient-centered care for diabetics (Charles et al., 1997; Porter & Teisberg, 2007), and it is the primary care physician that typically refers patients to diabetes educators and specialists, and provides prescriptions that are filled by pharmacists. Through face-to-face 8 International Journal of Healthcare Information Systems and Informatics, 6(1), 1-18, January-March 2011 Copyright © 2011, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited. interviews of 60 to 120 minutes duration conducted within healthcare providers’ context, we obtained data to elaborate, amend and refine our initial conceptualization (i.e., DND conceptual framework). Interview subjects were provided a copy of the emerging DND diagram and asked to confirm, disconfirm, amend or refine the diagram to best represent the information dependencies and information flows among all pertinent actors in patient-centered care for diabetics. We asked all interview subjects to enhance and refine our understanding of all of the roles and information flows that typically support patient-centered care for diabetics. We further substantiated the data by interviewing actors in each of the primary roles involved in advising patients, as previously discussed. In accordance with the process of interpretive study, we concluded the process of refining the conceptual framework when the data converged; that is no further changes were identified (Miles & Huberman, 1994). Convergence occurred after five interviews, and our subjects included a primary care physician, two certified diabetes educators, and two diabetes specialists. All of the participants were experts that had both clinical and research experience in the context of information flows in support of patient-centered care for diabetics. Analytical Procedures The process of developing a DND is analytical in nature (Tillquist et al., 2002). In the process of developing the DND, described above, a general conceptualization of the network structure emerges based on the pattern of actor-to-actor information exchange. However, to assess our hypotheses we sought to further evaluate the extent to which the pertinent actors maintain direct ties with each other in order to exchange information in support of coordinated treatment plan decisions, and the pattern of communicative exchange through personal contact and IS-supported contact. To that end, we used the final DND to inform a structural approach to social network analysis (SNA) (Borgatti et al., 2002; Gabbay & Leenders, 2001). Whereas the “tie approach” to SNA focuses on the dyadic relationship between a given ego and alter (i.e., subject and object actors) in order to assess the strength of ties, the “structural approach” to SNA focuses on the structure of relationships in which the ego is embedded in order to identify holes (or conversely, closure) in the network structure (Borgatti et al., 2002). Following the Borgatti et al. (2002) procedure for structural analysis, we arranged all seven actor roles identified in the DND (i.e., patients, primary care physicians, clinical professionals including diabetes educators, specialists, pharmacists, and the medical labs and insurers that provide technical and operational support) in a 7x7 matrix and coded a “1” in each cell of the matrix for which information flowed from one actor (i.e., rows) to another (i.e., columns), a “1” in each cell of the matrix for information flowed in the reciprocal direction (i.e., from actors in columns to actors in rows), and a zero in the remaining cells of the matrix. The matrix was then subjected to network density and centrality (i.e., geodesic distance) analysis and to visual analysis using UCINET software (Borgatti et al., 2002). Network density and centrality measures enabled us to evaluate the extent to which actors are cohesively connected to each other. The visual analysis enabled us to differentiate ties (i.e., patterns of actor-to-actor information flows) based on actors’ use of personal contact and/or integrated IS-supported contact. Supporting “integrated IS” include available systems to coordinate and integrate all actors’ participation, and exclude systems available to only a subset of pertinent actors (e.g., interactive pharmacy websites that are not available to patients from all pharmacies). The DND and the SNA analyses enabled us to assess our hypotheses as elaborated next.