INTRODUCTION

The transition from hospital to home is a high-risk period for patients.1,2 Challenges to a safe transition include lack of communication regarding self-care needs, poor understanding of home care strategies, and difficulties adjusting to new medications.3,4 Many of these challenges can be addressed through care coordination, or the deliberate organization of a patient’s care activities among providers.5 Care coordination enhances quality of care, reduces costs, and improves patient quality of life.6 Several care coordination interventions have successfully reduced hospital readmissions,7,8,9 but very few have been disseminated to diverse settings.3,10 Improving care coordination is a priority in the Veterans Health Administration (VHA), but there is little consensus on how to best disseminate care coordination interventions across large health care systems.7,11,12

Adapting interventions to local contexts improves program implementation.13,14,15 Implementation science tools and frameworks provide a standardized structure for understanding how contextual differences at diverse sites may affect implementation of an intervention.16,17 Created by national experts and informed by prior implementation science work, the Practical Robust Implementation and Sustainability Model (PRISM)15 provides a framework for applying implementation science principles. PRISM describes six key domains to evaluate in disseminating evidence-based interventions: implementation and sustainability infrastructure of the receiving organization, organizational perspective of the intervention, patient perspective of the intervention, organizational characteristics, external environment, and patient characteristics.15 Although PRISM has shown promise as a useful framework for exploring barriers and facilitators in case studies,18,19 questions remain about how to successfully operationalize the framework in multisite studies.

This project sought to address these two gaps: the lack of published data on experiences using rigorous implementation science methods to inform dissemination of promising care coordination interventions, and how to operationalize PRISM for this purpose. PRISM guided the implementation of the rural Transitions Nurse Program (TNP) across five diverse VHA facilities. TNP is a multi-component, nurse-led intensive care coordination intervention designed to improve care transitions for rural Veterans who are hospitalized at VHA hospitals and subsequently discharged to their rural residence and care setting.20,21 The transition from hospital to home is especially challenging for rural patients, who may experience medication gaps due to limited availabilities of medication in rural areas, inadequate discharge plans, lack of communication between hospital and primary care providers, and difficulty obtaining follow-up care.20 Like most successful care coordination programs, TNP addresses multiple barriers to an ideal transition of care and encompasses both pre- and post-discharge components.4 The TNP intervention consists of four core components carried out by a Transitions Nurse (TN) based at an urban VA medical center. These include meeting with the Veteran in the hospital to assess and address readmission risks, partnering with inpatient medical teams to identify post-discharge needs and coordinate a follow-up with a VA primary care provider, engaging with the rural Patient Aligned Care (PACT) team by sending hospital records and calling to discuss patient needs, and a follow-up phone call with the Veteran to revisit goals set during hospitalization and assess current symptoms. In this paper, we discuss how we operationalized PRISM to identify barriers, facilitators, and important contextual factors prior to TNP implementation. We then describe how we tailored our implementation efforts to address the most common barriers, facilitators, and contextual factors at each site. Our findings may apply to dissemination of care coordination interventions in a variety of settings.

METHODS

Pre-implementation Assessment Design

We operationalized PRISM by creating definitions for each of the six PRISM domains. We identified data categories and sources necessary to assess each domain. We collected and analyzed pre-implementation data with the goal of learning about contextual factors at each site.

We applied a rapid ethnographic approach to assess context using a variety of methods.22 We designed key informant (KI) interviews to elicit information on organizational perspective of the intervention, organizational characteristics, implementation and sustainability infrastructure, external environment, and barriers and facilitators to TNP implementation and success (Appendix, Key Informant Interview Guide). We pilot tested the KI interview guide among colleagues to ensure that questions were easily understood. KI interviews were conducted over the phone with participants at both VA hospitals and PACT clinics, recorded, and transcribed verbatim. We designed observations of discharge processes and follow-up appointments to assess patient characteristics, implementation and sustainability infrastructure, and organizational characteristics. We piloted the observation process locally prior to site visits. Observers took detailed notes on process and context of discharge planning, including the types and nature of interactions observed. Observations were conducted at both VA hospitals and PACT clinics. Veteran interviews were intended to gather information on patient perspective of the intervention and patient characteristics.21 Finally, the brainwriting premortem activity was designed to identify barriers to program success. Brainwriting is a silent, written group brainstorming activity. The brainwriting premortem was developed and piloted in Denver to identify reasons TNP might fail. Each brainwriting session was followed by a discussion, which was recorded, and transcribed verbatim.23

Data Analysis

We used inductive-deductive framework analysis to identify themes across KI interviews, observation notes, Veteran interviews, brainwriting papers, and transcripts from post-brainwriting discussions. Deductive codes were based on the six PRISM domains and factors that may be encapsulated within those domains (15; Table 3), barriers, and facilitators. Qualitative analysts created emergent codes to label additional contextual factors, barriers, and facilitators. Intercoder consensus was built through team discussion by resolving points of disagreement after analysts independently coded the same 11 documents, representing 12% of all qualitative documents.24 Atlas.ti V 7.5.1825 was used to manage qualitative data.

Identification of Adaptations Using Qualitative Data

Key contextual factors, facilitators, and barriers were compared across sites. We shared these data with sites at an in-person training session in Denver 1 month after the completion of data collection. Data were used to brainstorm actionable adaptations to the implementation process. TN’s from all sites continued to discuss potential adaptations on weekly phone calls with facilitators from the Denver TNP team. TNP received quality improvement status from the VA Office of Rural Health (ORH), which funded this project. All findings were used for operations purposes. Participants were informed that their participation was voluntary and assured that their contributions would remain confidential.

RESULTS

Our efforts to operationalize PRISM resulted in working definitions of each of the six domains. Table 1 shows the data categories and types of data collected to assess each domain. Table 2 shows the number of each qualitative data type collected. We identified themes related to five of the six PRISM domains using data from all sources. We were unable to identify themes related to “patient perspective of the intervention” due to a limited number of interviews with Veterans. PRISM themes highlight potential barriers and facilitators to TNP implementation and identify areas in which TNP can improve transitional care for rural Veterans. Quotes to illustrative each theme are provided in Table 3.

Table 1 PRISM Working Definitions and Types of Data Collected
Full size table
Table 2 Number of Each Qualitative Data Type Collected at Each Site
Full size table
Table 3 Illustrative Quotes for Themes Related to PRISM Domains
Full size table

Themes Related to Organizational Characteristics

Disconnect Between Primary Care Teams and Hospital Inpatient Teams

Hospital-based participants noted a lack of knowledge of workload, resource availability, and processes in primary care. They did not know what follow-up capabilities were in place when recommending future care for hospitalized patients. Participants in primary care settings described a disconnect between hospital expectations for discharge follow-up and the reality of what they could provide. They described confusion around who to contact for additional information, or difficulties and frustrations when contacting inpatient hospital teams (quote A1, Table 3).

Areas for Improvement in Current Transition of Care Processes

Given the disconnect between hospital and primary care settings, participants at all sites described areas for improvement in transitional care processes. One site stressed the need for better medication education since pharmacy was short-staffed. The remaining sites emphasized poor coordination of discharge supplies and follow-up services, and communication barriers within and between services. Poor coordination was reported to result in challenges for patients trying to obtain medication, supplies, or follow-up care (quote A2, Table 3).

Themes Related to Organizational Perspective of the Intervention

Positive to Mixed Reaction to the Program

Many participants expressed strong enthusiasm for the program. They felt that TNP addressed a clear gap in transitional care for rural Veterans and thought that TNP had the potential to improve communication between hospitals and PACT sites. Some participants stated that positive program outcomes would depend on how TNP was rolled out, what kind of support it received, and who was selected for the TN position. In particular, participants felt success hinged upon the ability of the TN to provide clear education about the program, patient eligibility, and how the TN could augment existing work processes (quote B1, Table 3).

Concerns About Work Duplication

One of the most common concerns about the program was that it would duplicate work, overlapping with existing programs. For example, participants at rural clinics worried that the TN follow-up call would duplicate the PACT nurses’ post-discharge call. Hospital staff worried that the TN’s discharge planning duties would overlap with the responsibilities of other roles involved in discharge (quote B2, Table 3).

Concerns That One Nurse Could Not Meet the Demand for the Program

Participants at most sites emphasized that there were many more high-risk rural patients than one nurse could enroll in the TNP. Thus, they worried that the TN would be overwhelmed and overburdened. Some participants stated the necessity for the TN to say “no” when necessary, and others suggested that the TN clearly delineate eligibility criteria (quote B3).

Themes Related to Implementation and Sustainability Infrastructure

Weak Infrastructural Support for Coordination of Care Between Hospitals and PACT Sites

Participants at most sites described communication difficulties between VA tertiary and primary care sites. In some cases, medical records were not shared because the hospital and primary care site were in different VA regions. This made it difficult for inpatient hospital teams to share information, and for participants in primary care to understand the details of hospitalizations. In other cases, there was a lack of familiarity with the needs of hospital or primary care clinicians and a lack of knowledge of who to contact with questions. This caused frustration when additional information about a patient was needed. This theme is related to the theme disconnect between primary care teams and hospital inpatient teams (quote C1, Table 3).

Concerns About Infrastructural Support for the Program

Sites were concerned about infrastructural support, including workspace, resources, and program sustainment after grant funding for the transition nurse position ended. Participants suggested that the Transitions Nurse build relationships with colleagues with strong institutional knowledge of care transition processes. Some participants described strong leadership interest in TNP and long-term plans to expand the TN position, while at other sites, participants described a need for positive program outcomes for continued program support (quote C2, Table 3).

Themes Related to Patient Characteristics

Difficulty Contacting Patients

All sites noted difficulty reaching rural Veterans as they often did not answer their phone, had poor telephone reception, or were hard of hearing. Some participants stated that Veteran contact information is often outdated in hospital records, or that patients sometimes return home with friends or family with different telephone numbers. Others suggested that some Veterans do not want to be contacted and might be unresponsive (quote D1, Table 3).

Transportation Challenges

Most sites also described a lack of reliable transportation among rural Veterans. Participants felt that low socioeconomic status of a large portion of the rural Veteran population was an important contributor to transportation issues. Participants noted that some rural Veterans live far from their VHA primary care facility and choose not to attend follow-up appointments (quote D2, Table 3).

Themes Related to External Environment

Impact of the Veteran’s Choice Program

Participants at most sites discussed the potential impact of the Veteran’s Choice Program, which allows Veterans to seek care outside of the VA. This program has increased non-VA healthcare utilization by Veterans. They described the need for a program like TNP to coordinate care between VA and community settings and wondered if TNP could fill this gap. They noted that many Veterans utilize community resources, but it was difficult to know which community services the Veteran qualified for or received (quote E1, Table 3).

Difficulties Recruiting and Retaining Providers in Rural Areas

Difficulties recruiting and retaining providers in the VA in rural settings were an important structural barrier, creating large patient panels for existing providers and limiting the availability of follow-up appointments. This was sometimes so impactful that the VA had to contract with outside providers to provide primary care for rural Veterans. Participants at contracted clinics expressed concern that TNP would increase their work load and necessitate contract re-negotiation (quote E2).

DISCUSSION

Using PRISM to evaluate site context yielded important insights of potential barriers and facilitators to implementation of a care coordination program and helped identify crucially important adaptations. We could anticipate potential challenges, such as perceived work duplication, concerns that one nurse could not meet demand for the program, weak infrastructural support for care coordination, and difficulties related to patient characteristics.

At sites concerned with role duplication, we helped TN’s create clear role descriptions and brainstorm ways to utilize existing infrastructure. At sites concerned that the TN would be overwhelmed with the number of eligible patients, we helped the TN’s modify program enrollment criteria. Modifications to enrollment criteria were the most prevalent adaptation in the implementation process. At sites concerned with poor communication structure between hospitals and PACT sites, we recommended that TN’s provide informational sessions at PACT sites to engage stakeholders and initiate relationships. This adaptation required more time and resources than others, but TN’s visited at least some PACT clinics. Making these adaptations early in the implementation process helped to roll out TNP quickly at each site. Some TN’s met enrollment goals within 2 months of beginning enrollment.

Importantly, while some of our findings were anticipated, such as concerns about infrastructural support and perceptions of imperfect transitional care processes, many findings were novel and unanticipated by the TNP team. For example, we did not anticipate interest in a transitional care program to work with non-VA hospitals, nor concerns about transportation for rural Veterans. We also did not foresee the extent to which clinicians we spoke to were pleasantly surprised by the time and effort expended in understanding processes, culture, and context at each expansion site. However, we could not address all potential barriers during implementation, such as transportation challenges and staffing issues at PACT clinics because these issues exceed the scope of the TN role. Understanding these potential barriers allowed us to help the TN’s plan for these challenges.

Previous studies have used PRISM to assess barriers and facilitators prior to Implementation,18,19 but only in single-site programs. These studies found that participants felt that the intervention needed to be compatible with current processes and consider clinician time constraints along with unique patient needs. Our study corroborated this finding in a multisite pre-implementation evaluation. Using PRISM in a multisite implementation of a care coordination intervention provided the opportunity to identify barriers that are generalizable to implementation of care coordination interventions across sites with very different characteristics. We build on previous studies by identifying several facilitators and barriers related to five of six PRISM domains. Further, our study shows that PRISM can successfully integrate qualitative data from a variety of sources to identify actionable themes.

Many of the themes in our data may be applicable to other care coordination interventions within the VHA. For example, challenging communication infrastructures and difficulties meeting program demands are certainly not unique barriers to TNP. Our findings reveal challenges with critical steps to safe care transitions, including coordination with community and social support resources, coordinating care among team members, and outpatient follow-up.4 Knowing how to assess and effectively respond to such barriers is a critical step in the implementation process. The examples provided here were successful strategies for implementation and adaptation of an evidence-based care coordination intervention. Future studies should try to replicate these results to increase generalizability for care coordination interventions. We are currently implementing TNP at five additional sites and continue to streamline the use of PRISM for pre-implementation assessment.

Limitations and Next Steps

This evaluation was conducted by qualitative analysts, and methods may be difficult to replicate without similar resources to conduct site visits. We were unable to thoroughly assess patient perspective of the intervention due to a limited number of interviews with Veterans. We did not have the resources to collect data of comparable detail to that collected at VA hospitals and PACT clinics. Consideration of patient perspective is a key domain of PRISM, and future work should assess patient level barriers and facilitators related to care coordination programs. This study was conducted in the VA, and some of the contextual factors, facilitators, and barriers identified may be unique to the VA. Future studies should determine which contextual elements are present only in the VA, and which contextual elements influence the implementation of care coordination interventions in the community. Finally, there is no gold standard for validating our assessment. Further experience across future projects will help to refine our approach.

CONCLUSIONS

The use of an implementation science framework is crucial to multisite implementation of complex interventions. Using PRISM, we identified barriers and facilitators to TNP implementation, as well as contextual information that allowed us to adapt implementation to local settings. Results guided adaptations to program delivery and materials. Sites acted upon findings early in the implementation process. Common barriers and facilitators across TNP sites are likely generalizable to VHA care coordination interventions.