For patients with chronic conditions, health management is a continuous effort [1]. Management of chronic conditions entails numerous activities in both clinical (eg, developing plans of care) and daily living settings (eg, assessing medication side effects, controlling food intake) [2]. Activities across these settings interact [3-5], creating a high potential for a gap to occur if there is an inconsistency or disconnect between controlled clinical settings and complex daily living environments. We conceptualized gap as a break in continuity in the performance of health-related activities across clinical and daily living settings. A gap between health-related activities across settings is a latent phenomenon that can be identified or addressed only after precursors have surfaced or effects have been manifested (eg, when suboptimal health outcomes have occurred). As an example of a gap [4], medication prescription is a part of the clinical workflow that requires coordination with the patient’s daily living environment. Not incorporating daily living information to tailor therapy could lead to nonadherence. The study by Ozkaynak et al [4] provides more information on the definition of a gap. Identifying the specific points where a health care regimen can or must be modified is possible only if all health-related activities in both settings are considered part of a health care continuum [4-8]. Adding a gap construct to known models and frameworks may allow for a better comparison between theoretical and real-world management systems for patients.

This study aims to understand why these gaps occurred, what effects they produced, and preventative or remedial actions. The work was guided by the Infinicare framework [4], a theoretical perspective that highlights the need for integration of health-related activities across care delivery settings and accounts for patient-specific contextual features. The Infinicare framework has 3 principles. First, clinic-based health-related activities shape or influence activities in the context of daily living. Second, health-related activities within daily life inform activities in the context of the clinical setting [4]. Third, the Infinicare framework highlights the importance of context and operationalizes it by incorporating 4 different contextual dimensions: physical, organizational, social, and cultural. These 4 dimensions of context interdependently affect health-related activities in both clinical and daily living settings. The Infinicare framework posits that health and health care have no strict temporal or physical boundaries.

Anticoagulation therapy with warfarin is a particularly rich example for understanding gaps between clinics and daily life [7]. It requires strict adherence by patients and collaboration with providers, as frequent dosing adjustments are often necessary [40]. Food and alcohol affect dosing; therefore, a strict limitation of foods high in vitamin K and limiting alcohol intake is required. Moreover, contextual factors such as family support, ability to purchase food, and avoidance of specific foods that can affect therapy must be considered. Monitoring the therapeutic dose of warfarin is achieved by measuring blood levels for clotting. The international normalized ratio (INR) is a measure that evaluates the therapeutic range of anticoagulation [41]. Patients require testing in the clinic setting as often as several times a week, as home anticoagulant monitoring was not available at the site studied. Warfarin or Coumadin interacts with common antibiotics; therefore, communication is essential by the patient with their non–clinic-based providers (eg, dentist, podiatrist) ordering an antibiotic. Good communication between the provider and patient about testing results and dosing decisions is crucial. A wrong dose can result in serious bleeding, thromboembolic events, and medical complications from the underlying medical problem for which the anticoagulant was ordered. Examining gaps in optimal treatment can reveal clinician, patient, and system-related factors leading to suboptimal dosing and inform interventions.

We recognize that the dynamics of other chronic conditions (eg, diabetes, HIV) may differ from anticoagulation management. The findings and design guidelines based on anticoagulation, however, have the potential to be translated where HIT supports self-management activities in individualized therapy plans across patient conditions.

We used a qualitative study design to examine and characterize gaps between health-related activities across diverse settings. Qualitative design allows for exploring and gathering rich descriptions of the phenomena of interest (ie, gap) in a context. Data were collected from patients and providers (clinical pharmacists) in a single academic, hospital-based anticoagulation clinic through interviews and journaling. Data were analyzed qualitatively. This study was approved by the Colorado Multiple Institutional Review Board.

The Infinicare framework [4] shaped this study in 2 ways: (1) it guided the development of the data collection tools by ensuring that data were collected on basic components (eg, health-related activities in daily living and clinical settings and social, organizational, physical, and cultural context); and (2) guided data analysis through use of select Infinicare constructs (eg, daily routines, challenges of self-management, clinicians’ understanding of the patient’s daily living environment) for the initial codes.

Patients were recruited from the anticoagulation clinic of an urban, western mountain health system in the United States. The clinic provides outpatient care and is staffed by 4 clinical pharmacists. The primary medication used by the clinic was warfarin, an anticoagulant in pill form. Most clinic visits were face-to-face; however, phone visits also took place.

Exclusion criteria were individuals younger than 20 years of age. In an effort to approximate a real-world setting, participants with comorbid medical conditions, mental health conditions, substance use disorders, or pregnancy were eligible for inclusion. We used a stratified sampling strategy to ensure that we recruited older adult patients as well as younger patients of both genders and those both new and experienced with anticoagulation therapy. One participant’s first language was Spanish, for which a bilingual translator was used for data collection.

Fliers were posted in the clinic with information about the objective of the study, sponsors, principal investigator, institutional review board approval number, and contact information for the research team. Interested participants called the dedicated phone number to provide their contact information. This information was accessed only by the research team, who called to verify that the prospective participant met the inclusion criteria.

Data were collected (Textbox 1) first from primary interviews with patients, followed by patient journaling on tablet computers and exit interviews. We conducted provider interviews as the last step in data collection. Data from exit interviews with patients and provider interviews were collected to obtain both views. Figure 1 summarizes the data collection methods and how the different methods related to each other. Complete data (primary interviews, journaling, and exit interviews) were obtained from 90% (35/39) of patients (4 completed only the primary interviews). Four providers participated in the in-person interviews. Each patient interview (primary and exit) was conducted in person by the first (MO) and third authors (KH). These interviews occurred in a private meeting room in the College of Nursing, which is in the same campus as the anticoagulation clinic.

Those conducting the interviews used a predetermined list of questions; however, participants were also asked follow-up questions for clarification and to gain additional insights. Primary patient interviews provided information related to: (1) the care they received in the anticoagulation clinic; (2) their social, cultural, organizational, and physical environment; (3) daily health-related activities; (4) tools and technology they used to support these activities; (5) challenges and facilitators while conducting these activities; and (6) manner in which clinic-based health-related activities shape or influence activities in daily living contexts and health-related activities within daily life inform activities in the clinical context. Journaling provided data about health-related activities, the people involved in the activities, the temporal organization of the activities at home and other daily living settings, and the social, organizational, cultural, and physical contexts in which the activities occurred. Journaling also partially mitigated the recall bias. Journaling was accomplished by a tablet application that made voice entry possible. The application also moved all audio recordings to secure servers to facilitate data collection and analysis. Patients were provided approximately 20 min of training on the use of the tablet and the app. During the training, participants were instructed on how the journaling app worked, and the researcher reviewed the 11 guiding questions with participants. Participants were asked to journal at least every 2 days for a month and provided the telephone number of the research line in the event they needed assistance. Patients were given an option to keep the tablet computer as a thank for participating. Exit interviews were conducted with the patients after they completed journaling and explored issues that had been mentioned in the primary interview and journals. Exit interviews (Step 3) were conducted in person in the same room as primary interviews, with information used for triangulation. Patient primary and exit interviews averaged 48 and 20 min, respectively. Provider (pharmacist) interviews (n=4) lasted approximately 73 min. During this time, health-related daily living activities reported by the patients were explored and a description of the clinical processes, clinical policies, development of therapy plans, and provider perspectives dealing with challenges related to adherence of the patient.

We received consent from 39 patients for the primary patient interviews, journaling, and exit interviews and received consent from the pharmacists for provider interviews. For all of the data collection, any reference to a personal identifier (eg, names of people, organization, etc) was removed.

Interviews were transcribed verbatim professionally. The journal entries were reviewed and transcribed by KH. Qualitative analysis was based on a theory-driven approach [42] and the framework method of analysis by Gale et al [43]. The main concepts of the Infinicare framework [4] provided a priori codes, with additional codes identified inductively when textual elements did not fit within these predetermined codes [44,45]. The advantage of using an overarching framework was to increase the efficiency of the data analysis process and to build on prior studies. The advantage of deriving additional codes based on collected data allowed for better integration of the participant’s experiences.

Data from the 4 collection methods were coded, main themes were identified, and relationships between the main themes were examined. Data were managed using DEDOOSE 5.1.18. The analysis included independent coding of approximately 10% of the data by MO and KH. During this coding, various code ideas were discussed in light of the broader study purpose, such as (1) the process and contextual components associated with care for anticoagulation management and (2) the potential disconnect between health-related activities in the clinical setting and outside the clinical setting. Coding was then reviewed by 3 team members (MO, KH, and RV) until consensus was reached, following which a codebook was developed [46]. The codebook included codes, the concepts that were represented by the codes, specifications (definition, inclusion, and exclusions as applicable), and examples. We also allowed simultaneous coding, meaning that multiple codes could be applied to a piece of text [47]. KH applied the codebook to the remainder of the data. MO then examined embedding the codes within the overarching categories, based on their content.

Study participants ranged in age from 25 to 83 years and had been receiving anticoagulation therapy for 2 weeks to 26 years (Table 1). Most participants were women and White. Educational attainment varied, although most had a minimum education of a college or associate degree (30/39, 76%). Most participants lived with someone. The median household income was US $60,000; 3 of the 4 providers had a PharmD degree, the other had a BS degree. The average experience of the providers was 15 year.

The gaps are the operationalizable aspect of the Infinicare framework and can be more directly used to guide HIT development. Our analysis revealed 3 aspects of gaps: reasons, consequences, and strategies to overcome these gaps (Textbox 2).

The reasons for gaps were from 3 areas related to therapy recommendations: misfit with the patient’s daily routine, misfit within the context of daily living, and information was not transferred across settings.

We grouped the consequences of gaps into 3 themes: cognitive and physical workload on the patient, poor patient satisfaction, and compromised adherence to the therapy plan.

Five strategies were used to cope with the challenges of gaps: (1) patient-generated strategies, (2) routines, (3) collaborative therapy management, (4) social environment, and (5) tools and technologies.

This study examined chronic disease–related challenges by focusing on gaps between therapy plan development (ie, clinic based) and implementation (ie, daily living based), using the case of anticoagulation therapy. Gaps were latent and not directly observable, but their effects were evident. We ensured the collection of rich data on gaps by employing multiple data collection methods that inform each other. We also collected data from both patients and providers. The use of a framework (ie, Infinicare) ensured that only relevant data were collected on all aspects of the gaps. The reasons for and consequences of gaps as well as strategies to overcome them were identified. The reasons for gaps can prompt needs assessments and inform systematic interventions (eg, informatics, policy, educational, operational) that narrow these gaps. The consequences should also be taken into account when designing and implementing interventions. Moreover, these consequences can serve as evaluation criteria for interventions. Interventions should be congruent with the 5 reported categories of strategies for bridging the gaps. The reported strategies are a jumping-off point, and other strategies may be developed.

Understanding the reasons for the consequences of and strategies to overcome gaps can also help integrate the concept of gap into the current and future conceptual models and frameworks of health management. These models and frameworks can better explain the factors that affect health management.

Previous studies [26,48] that focused on the individualization of health care services and the transition of care have highlighted some of the themes discussed in this paper, such as collaboration among clinicians, patients, and family members; social support; provider awareness and knowledge of patient situations; understanding patients’ cultural practices and beliefs; understanding places in which the patient lives; emphasis on the patient and family experience; focus on home and community-based care; emotional support and alleviation of fear and anxiety; and encouragement of continued social roles. Moreover, studies that focused on self-management activities and adherence work stressed the importance of daily living contexts [2], discrepancies between clinicians’ and patients’ perspectives on self-management [49], and localization of therapy management [20]. However, this study went beyond previous findings by identifying the concept of the gap. Conceptually, gaps can provide more insights to researchers by better explaining incongruence among work system elements [2] in chronic disease management. This study provides a foundation for collaborative HIT design guidelines by highlighting relationships among the precursors to, consequences of, and counter measures against cross-setting gaps.

The 3 reasons identified for gaps underscored the need for clinician access to HIT that would incorporate more information about the patient’s daily living environment. Current consumer informatics technologies (eg, personal trackers, food log applications) support data collection on only a limited number of activities of daily living (eg, exercise, food consumption); however, these data are not always leveraged in developing therapy plans taking into consideration a patient’s routines and context [50,51]. This study shows that other types of routines (eg, social life) also need to be captured and accounted for. Moreover, there is a need to capture the interplay between different types of routines (eg, social life, food consumption, medication administration). The use of such rich data in creating therapy plans would require better integration of patient-generated data into EHRs. Clinical decision support systems that use machine learning algorithms can harness historical clinical and patient-generated data and assist in developing individualized therapy plans [52,53]. Mobile technologies provide opportunities to better capture essential information in each clinical and daily living setting and information that can then be retrieved as needed across settings [54-56]. We also contend that, instead of prioritizing either clinical or daily living settings, technologies should foster collaborative dialog. This collaborative dialog should include (1) joint clinician-patient interpretation of patient-generated data [39] and (2) discussion of patient-driven therapy plan translation options [20]. The reasons for gaps can guide clinicians’ question-asking and decision-making behavior. Focusing on gaps can help inform interventions with (1) technology, (2) policy, (3) educational, and (4) operational components that would support management in chronic disease. In fact, any collaborative HIT intervention that would narrow or close the gaps could need all 4 components. In addition, HIT designs can also be developed to promote sustained collaborative relationships between clinicians and patients [32,52]. There are no explicit guidelines on how to design and implement a truly collaborative HIT. This paper creates a starting point for the design and implementation of collaborative HIT. Design efforts should ensure that both clinicians and patients have tools and approaches for collecting, interpreting, and sharing information. Therefore, some of the potential next steps in designing and implementing collaborative HIT would be to create standard terminology for daily living activities (eg, food-related practice), developing machine-readable representations for daily living activities, and visualizations of these activities for both patients and providers.

Our analysis revealed patient-generated strategies for informing collaborative HIT design [17], but each strategy has attendant risks with the potential for unintended consequences. For this reason, these strategies should be vetted and implemented in consultation with clinicians, moreover because CTM (as we reported in the Results section) requires seamless, cross-setting information flow. HIT should be designed to be activated by patients and clinicians alike.

Owing to its outsized impact on patients’ decision-making processes, the social environment can both open and close gaps in chronic condition care and its maintenance. As a result, HIT should capture and communicate relevant social environmental information to clinicians to inform therapy plans [57,58].

Developing routines was reported as an effective strategy by both clinicians and patient participants. Today’s consumer technologies (eg, smartphones) partially support establishing routines; however, clinicians’ understanding of and contribution to these routines is limited, and some clinicians’ suggestions may not work for all individuals. Future collaborative HIT (eg, sensor-based technologies) should and can support clinicians’ understanding of patients’ existing routines and provide therapy plan options that can better fit in the context of patients’ home environments. Moreover, as both patients and clinicians report that routines are effective, new technologies can help generate routines to support health management in daily living settings. Collaborative HIT can play a role in identifying the reasons for the formation of gaps, monitoring the consequences of gaps, and supporting the implementation of gap-mitigating strategies.

An understanding of individuals’ daily living contexts can help providers better support patients’ chronic disease management by eliminating the 3 reasons for gap formation, as also highlighted by previous research [4,5,17,20]. Collaborative HIT can link currently disconnected clinical and consumer technologies to flesh out contextual factors [59] and attempt to neutralize potential discrepancies in social determinants of health.

The strength of this study is its theory-driven approach. The Infinicare framework informed data collection or analysis and focused the researchers on the cross-setting relationships among health-related activities. This study demonstrates the utility of Infinicare in both examining clinician-patient collaboration and informing collaborative HIT design and implementation. Moreover, this study extended the concept of the gap by providing a case showing how gaps can inform HIT design in the context of anticoagulation therapy. Another strength of this study is its use of multiple data collection methods. We used tablet-based journals complemented by interviews. Journaling allowed for moment-by-moment collection of critical data about patients’ health management activities, and interviews provided rich information about patients’ contexts.

This study is limited in that it drew data from a single outpatient anticoagulation clinic that uses warfarin and does not have home monitoring (nonetheless, this is likely a typical US anticoagulation clinic). Although we captured rich information on the causes and consequences of gaps and strategies for avoiding them, other clinical settings may reveal additional gaps. In addition, the study design leaned toward the patient’s perspective rather than the clinician’s perspective (eg, discussion of gap-related consequences did not address clinician workload). It is likely that there are more consequences to clinicians (eg, in terms of workload, patient education) than we reported. In our future work, we will seek to better understand the gaps from a clinician’s perspective. To minimize researcher bias, we collected data from 3 resources and triangulated them across data sources. Data were collected and analyzed by multiple researchers.

The study used multiple methods to explore the idea of gaps in a way to inform collaborative HIT. When designing and implementing these technologies, health-related activities conducted in daily living settings and clinical activities should be well connected and together they should be examined as a single workflow. Collaborative HIT can support clinicians and health systems to make decisions toward narrowing or eliminating the gaps can potentially improve patient and organizational outcomes.