Article Text
Abstract
Background Despite the increased focus on improving patient’s postacute care outcomes, best practices for reducing readmissions from skilled nursing facilities (SNFs) are unclear. The objective of this study was to observe processes used to prepare patients for postacute care in SNFs, and to explore differences between hospital-SNF pairs with high or low 30-day readmission rates.
Design We used a rapid ethnographic approach with intensive multiday observations and key informant interviews at high-performing and low-performing hospitals, and their most commonly used SNF. We used flow maps and thematic analysis to describe the process of hospitals discharging patients to SNFs and to identify differences in subprocesses used by high-performing and low-performing hospitals.
Setting and participants Hospitals were classified as high or low performers based on their 30-day readmission rates from SNFs. The final sample included 148 hours of observations with 30 clinicians across four hospitals (n=2 high performing, n=2 low performing) and corresponding SNFs (n=5).
Findings We identified variation in five major processes prior to SNF discharge that could affect care transitions: recognising need for postacute care, deciding level of care, selecting an SNF, negotiating patient fit and coordinating care with SNF. During each stage, high-performing sites differed from low-performing sites by focusing on: (1) earlier, ongoing, systematic identification of high-risk patients; (2) discussing the decision to go to an SNF as an iterative team-based process and (3) anticipating barriers with knowledge of transitional and SNF care processes.
Conclusion Identifying variations in processes used to prepare patients for SNF provides critical insight into the best practices for transitioning patients to SNFs and areas to target for improving care of high-risk patients.
- healthcare quality improvement
- process mapping
- qualitative research
- transitions in care
- health services research
Data availability statement
Data sharing not applicable as no datasets generated and/or analysed for this study. All data relevant to the study are included in the article or uploaded as supplementary information. Additional information may be obtained by emailing kirstin.manges@pennmedicine.upenn.edu.
Statistics from Altmetric.com
- healthcare quality improvement
- process mapping
- qualitative research
- transitions in care
- health services research
Introduction
Healthcare systems are increasingly seeking solutions to enhance the transition between hospitals and skilled nursing facilities (SNFs) to improve patient outcomes and comport with growing financial incentives. For example, the Bundled Payments for Care Improvement (BPCI) programme holds hospitals responsible for the costs associated with an ‘episode’ of care triggered by a hospitalisation, and hospital readmission accounts for a substantial proportion of overall episode costs.1 Similarly, the newly implemented SNF Value-Based Purchasing programme is leveraging financial penalties on SNFs with elevated hospital readmission rates.2
Despite the important role of postacute care in healthcare systems, best practices for reducing readmissions from SNFs are unclear.3 Published examples of successful transition of care interventions require significant investment of new personnel in SNFs and are unlikely to be scalable or sustainable.3–6 Instead, since there is striking variation in risk-adjusted readmission rates across hospitals and SNFs, identifying what distinguishes high-performing hospitals and SNFs (those with low risk-adjusted readmission rates) from low-performing hospitals might provide fruitful insights.7 Unfortunately, we were unable to identify hospital characteristics from administrative data that reliably distinguish high and low performance on short-term readmission rates from SNFs.8 Additionally, prior qualitative studies have identified that individual patients, caregivers and their healthcare team lack understanding of the process for discharging patients to postacute care, which further adds to the complexity of improving hospital to SNF transitions.9–14
Although these studies collectively suggest a need to improve the hospital to SNF discharge planning process,3–14 there is a knowledge gap with respect to the most effective mechanisms or care processes used by hospitals and SNFs to prepare patients for the transition to SNF. In this study, we sought to observe and identify transitional care processes for patients making this transition in high-performing and low-performing hospitals and their most frequently affiliated SNF partners. Our goal was to understand key process differences to provide critical insights for redesigning the hospital to SNF transition.
Methods
Design overview
We used a rapid ethnographic approach with intensive multiday observations to describe transitional care processes for preparing older adults for discharge from the hospital to an SNF. Trained qualitative researchers (RA, CL, ML, EG) conducted observational site visits at four hospitals and their most frequently used SNF partner(s) (determined by hospital staff) between August 2018 and October 2018. Observations, interviews, clarifying conversations and collecting artefacts allowed us to capture rich data around the social-cultural complexities of preparing patients for discharge from a wide range of perspectives.15 16 Through direct observations, we documented detailed descriptions of activities, behaviours and interactions as they occurred in the practice environment.15 16 Standards for Reporting Qualitative Research Guidelines were used to enhance transparency.17
Sampling strategy
We used a convenience maximum variation sampling strategy to describe differences in processes between high-performing and low-performing hospitals.7 18 To identify high-performing and low-performing hospitals, we used a previously published sample consisting of all veterans enrolled in the Veterans Health Administration (VHA) and discharged from any hospital to any SNF in the years 2012–2014.8 19 We then identified a sampling frame containing the top (n=20 high performing) and bottom (n=20 low performing) non-VHA hospitals as defined by mean risk-adjusted 30-day readmission rates to SNFs. Since 85% of all veteran transitions were between non-VHA hospitals and non-VHA SNFs under the Medicare benefit,8 we limited our sample to non-VHA hospitals.
To recruit sites, our study team contacted the case management directors and lead hospitalist at all 40 hospitals. We stratified the hospitals by characteristics in table 1, and chose four hospitals (n=2 high performing, n=2 low performing) who were willing to participate based on variability in these characteristics. Then, we asked hospitals to identify their most frequently used SNFs and obtained permission for observations. We also did not limit our observations to veteran discharges. Sample size was determined by the resources available to conduct the study.
Data collection
Two to three trained qualitative researchers observed the inpatient team discharge process for several patients undergoing the hospital-SNF transition at each site. At each hospital, we observed the discharge process on a medical surgical unit, and another unit based on local sites preference (ie, oncology, neurology, orthopaedic) over a 2-day period (mean=28 hours). The primary informant for direct observations was the discharge coordinator, defined as a social worker, discharge navigator, transitional care specialist or case manager responsible for coordinating discharges. Similarly, we visited corresponding SNFs for approximately 6–8 hours to interview key informants (eg, SNF liaisons, administrators, nurses) and observe a new SNF admission. Additional roles were observed based on local site recommendations. Observation fieldnotes and memos were kept to record descriptions of the environment, typical workflow, events and interactions observed (ie, physical layout, interprofessional huddles, rounding, charting).15 16 20 21 At the end of each site visit day, observers met to cross-check data, reflect and triangulate their experiences in field memos.16 20 Observing multiple hospitals, units, provider roles and SNFs allowed for a breadth and range of experiences around the hospital-SNF transition process.15 16
Data sources
We collected three main data sources: observations, interviews and artefacts to gather rich detailed descriptions of the hospital-SNF discharge process.
Observations
We used a structured tool and unstructured fieldnotes to record observation data. The hospital-SNF observation tool (online supplemental appendix A) was developed using the Ideal Transitions of Care Framework22 and included a process checklist to record actions observed or discussed, roles involved and provided follow-up probes. Additionally, open fieldnotes were taken about observations regarding the environment, activities, dynamics and roles in the discharge process.15 16 20 21 Since the observers were unblinded, using a structured observation tool step helped to limit bias in data collection and allowed for consistent cross-checking across observers, units and hospitals; while still providing rich descriptions of the observations.15 16 20 21 Prior to data collection, we piloted the observational tool at a hospital and SNF that were not included in the sample.
Supplemental material
Interviews
The interviews were semi-structured and opportunistic, taking advantage of moments when informants could answer detailed questions without impeding patient care. We asked informants to walk through the process and share their perspective on what worked well and did not when discharging patients to postacute care. As much as possible, informants’ responses were recorded verbatim on the observation tool.
Artefacts
We collected discharge checklists, training information, patient’s hospital record for SNF and discharge instructions/educational materials. Additionally, descriptive notes of other artefacts were recorded, including: communication boards (ie, whiteboards in conference rooms, on units or in patients’ rooms), lists of patients discharging and care coordination notes. All artefacts were de-identified.
Data analysis
Using thematic analysis,22 we (1) described the overall process used to prepare patients for SNFs and (2) explore the differences in subprocesses between high-performing and low-performing hospitals. All data were de-identified and organised by hospital-SNF site in Atlas.ti7. Data familiarisation started with open-coding to capture emergent contextual features and processes used to prepare patients for SNFs. For each site, we created flow-maps of the hospital-SNF discharge and developed a unifying flow-map to organise the data.23 24 These were used to identify cross-cutting patterns influencing hospital-SNF transitions within, and across, site.23–25 Through iterative team discussions, (re)coding and thematic mapping, we identified five stages (or major components of the larger process) that occurred across all hospital-SNF pairs. Then, using a codebook informed by the Ideal Transition in Care Framework,22 we used blinded and non-blinded focused-coding and constant comparison techniques to identify themes driving the differences between performance levels.22 Data triangulation, memos/audit trails, open and theory-driven codes and result tables were used to enhance rigour and trustworthiness.15–17 23–27 Online supplemental appendix B provides a detailed analysis description.
Findings
The final sample included 148 hours of observations with 30 providers across four hospitals (n=2 high performing, n=2 low performing) and corresponding SNFs (n=5). Description of the hospitals and their affiliated (SNFs) are provided in table 1, with additional descriptions in online supplemental appendix C. We identified five major transitional care process stages for preparing hospitalised patients for postacute care: recognising, deciding, selecting, negotiating and coordinating (figure 1). The stage definitions and distinguishing subprocesses of hospital high performers are outlined in table 2. During each stage, we found that high-performing sites differed from low-performing sites by focusing on: (1) earlier, ongoing, systematic identification of high-risk patients; (2) discussing the decision to go to SNF as a team-based iterative process and (3) anticipating barriers with knowledge of transitional and SNF care processes. Notably, we expected that the differences between high and low performance would be in the ‘coordinating’ domain, where most processes captured by the Ideal Transitions of Care framework22 occur. However, we found differences between high and low performers across each of the five stages (table 2). We describe the stages and distinguishing differences between high-performing and low-performing sites.
Recognising
Participants identified ‘recognising needs’ as the first stage of preparing patients for postacute care. For recognition to occur, there needed to be a trigger—usually through an assessment—to alert the team, patient and/or caregivers that postacute care may be appropriate. Although the trigger for needing postacute care could be identified at any time during hospitalisation (by any key stakeholder), all sites emphasised that ‘discharge planning begins at admission’. Across sites, we observed a coordinator (ie, social worker, nurse or case manager) who screened new admissions for postacute care needs. Examples of assessment content included: insurance status, living arrangements, emergency contacts/social support, equipment (eg, wheel chair, oxygen), prior home health or an SNF use and expected transportation home. However, the specific content, who performed the assessment, as well as who triggered recognition that postacute care might be needed, was not standardised both across and within hospitals.
Distinguishing ‘recognising’ subprocesses
Although all hospitals gathered initial assessments, the low-performing sites described fewer monitoring actions throughout the patient’s hospitalisation. In contrast, the high-performing sites reported consistently communicating patient’s ‘postdischarge deposition plans’ at handoffs (across shifts and units) and during interdisciplinary rounds. Likewise, while all sites discussed ‘starting discharge planning on admission’ and reported screening for postacute needs, the high-performing sites emphasised the importance of not treating the assessment as ‘checking off a box’, but rather information to ‘act on’ during the inpatient stay. The high-performing sites engaged the full team—especially physicians and physical therapy (PT)—in recognising patient’s potential postacute care needs. Additionally, during the initial assessment, the high-performing sites performed triaging to identify those ‘at most risk for hospital readmission’ to ensure appropriate actions were taken. For example, the high-performing sites had standing orders for PT consults, which would trigger automatically for patients meeting specific admission assessment criteria. Furthermore, when staff identified patients who might need postacute care and could be a ‘difficult placement’ they would consult the ‘transition in care team’ (TCT) if available. These teams specialised in complex care coordination cases and included a geriatric trained physician or nurse practitioner, as well as a nurse navigator and/or social worker.
Deciding level of care
The second stage is deciding the patient’s anticipated level of care needed postdischarge. There were large variations regarding the care subprocesses used to decide postacute supports. After recognising the potential need for care posthospitalisation, the team gathered additional information to explore options. Gathering information facilitated sensemaking among the team and patient-caregiver about the situation. Additional alignment activities involved the process of developing a shared mental model among all key stakeholders. In most cases, alignment occurred through morning rounds or huddles, although hospitals varied in who was actively involved in these events. However, all sites described using care coordination notes to keep the team ‘aware’ of where the patient was in the process. Informants identified important content to become aligned on, including: anticipated medical needs, patient-caregiver preferences, capacity to self-manage care at home, functional status, social support available, financial options and psychosocial history. Last, the key stakeholders would make a choice about postacute care preferences and needs. The sites varied in who was involved in deciding the level of postacute care needed, when the decision happened, as well as where the decision process occurred (eg, in the patient room or hallway, during interdisciplinary or physician-led rounds).
Distinguishing ‘deciding’ subprocesses
The high-performing sites emphasised the importance of consensus building and making a team-based decision with the patient and their caregiver about postacute care. The decision-making process at low-performing sites tended to be ‘driven’ by specific insurance company SNF admission criteria (such as a PT assessment), unilateral patient preference or unilateral clinician preference. The high-performing sites stressed the importance of continuously evaluating patient status to identify patients’ goals and potential need for postacute care. These discussions happened ‘early and often’ to prevent patients and family members from being ‘surprised’. It also allowed the team time to identify multiple options for and anticipate potential barriers to discharging patients with different levels of postacute care supports (ie, home with home health, palliative care, SNF). Again, if the situation was anticipated to be ‘complex’, the high-performing hospitals used TCTs to facilitate these transitions with more time-intensive discussions.
Selecting
The next stage involves selecting the SNF provider. This stage was largely standardised across all sites. Selecting involved identifying facility options, presenting potential facilities to patient-caregivers and the patient-caregivers choosing their preferences. The coordinators described first checking with the patient-caregivers to see if they have a facility of choice, a geographic area to target for finding a facility or have identified SNFs to exclude from their options. Based on these initial ‘constraints’ set by patients and caregivers, a coordinator identified a list of potential options and verified if the SNF accepted their insurance. Additionally, coordinators ‘tried to match’ the patient’s level of acuity to the SNF (eg, capacity of the facility to give certain medications, provide wound care), and ensure transportation to the facility is available. Then, the coordinator provided the patient-caregivers with a facility list and asked them to pick their top one to three sites. All sites used a paper list and provide information about Medicare’s Nursing Home Compare website (as required by regulation).28
Distinguishing ‘selection’ subprocesses
Overall, the low-performing hospitals described relying heavily on their prior experiences working with SNFs, focusing on SNFs that would accept the patient’s insurance, and offering fewer options to patients. Although both high-performing and low-performing hospitals had preferred SNFs, the high-performing hospital participants emphasised to patient-caregiver the advantages of having staff across the care continuum (eg, clinicians who provide care in both the hospital and SNF) to support ‘healthcare team continuity’ across settings. Likewise, the informants from the high-performing hospitals highlighted the importance of identifying patients’ preferences earlier, so they could assist with providing additional information on cost or quality. For example, the participants from the high-performing sites discussed providing ‘tailored cost estimates’ to patients-caregivers, while low-performing participants provided general cost estimates.
Negotiation of fit
The fourth stage is fit negotiation, or the process by which the inpatient healthcare team works to determine availability and applicability of the patient to receive care from a specific SNF. After top preferences were selected, the coordinator worked with SNF staff to identify openings and to see if the SNF was ‘willing to take the patient’. The process of accepting a patient began with a referral from the hospital, followed by the SNF assessing the patient’s medical history and needs, medications, social needs and insurance status. SNFs with liaisons used the liaison to assess the patient in person. SNFs without liaisons sometimes assessed patients in person if something in the referral was unclear or ‘concerning’.
Distinguishing ‘negotiation’ subprocesses
Although admission criteria were similar across SNFs, the negotiation process varied. Low-performing hospitals reported that SNF staff had more leverage in accepting patients compared with the high-performing hospitals. In general, the high-performing hospitals had stronger hospital-SNF administrator relationships and were able to better anticipate SNF questions. For example, these hospitals had administrator meetings with preferred facilities to expedite transitions for patients with extended lengths of stays. The high-performing hospital SNF counterparts consistently stated they liked accepting patients from these hospitals because they had their ‘ducks in a row’, were transparent, and knew patients would be appropriate for their facility. In contrast, the SNF counterparts at the low-performing hospitals reported frequently feeling like the hospital was ‘trying to hide something’, and discussed needing to be vigilant with chart reviews and personally seeing ‘questionable patients’ prior to admitting. In particular, both low-performing hospitals and their associated SNFs focused on documenting PT evaluations and patient functional status as essential for making an admission decision, ensuring that the ‘PT paperwork documenting function’ was a reoccurring theme in low-performing hospitals. Likewise, the low-performing sites also reported frequent delays due to missing paperwork and/or assessments.
Coordinating with SNFs
After SNF selection, the inpatient team began coordinating actions to prepare the patient for discharge. The process included assessing medical readiness, arranging needed postacute care resources and communicating patient needs to patient-caregiver and SNF. Across all sites, the physicians and advanced practice practitioners were charged with the task of deciding medical readiness for discharge. The sites discussed the patient’s readiness for discharge to the SNF during morning rounds or interprofessional discharge huddles. At all sites, interprofessional lists—either using a whiteboard or electronically—were kept in a central place to create situational awareness of ‘where the patient was’ in their discharge trajectory. Another important component of coordinating was ensuring that the needed postacute care resources were arranged. This included medication reconciliation, setting up transportation, educating the patient and creating an SNF discharge packet.
Distinguishing ‘coordination’ subprocesses
The high-performing hospitals worked ahead to ensure postacute care resources were arranged, while the low-performing sites tended to wait until the day of discharge. The high-performing sites reported fewer delays in the discharge process, which they ascribed to cross-training staff on specific tasks, standardising discharge planning and care documents, and eliminating bottlenecks like arranging transportation (table 2). At the low-performing sites, the physicians met with the coordinator one-on-one to discuss the plan, while at the high-performing sites these discussions frequently happened during interprofessional morning huddles and/or rounds. Both high-performing sites had a pharmacist rather than physician-led medication reconciliation process. The high-performing sites also had a pre-SNF admission process, where SNF staff had the opportunity to start their admission prior to the patient leaving the hospital. Notably the low-performing sites expressed frustration about waiting on other team members ‘to do their part’ including physicians signing notes, putting in PT notes, arranging timely transportation and challenges with medication reconciliation.
Discussion
Our results outline a taxonomy for understanding the transition of care process for preparing patients to receive postacute supports in an SNF. We identified five stages for hospital-SNF planning: recognising, deciding, selecting, negotiating and coordinating. Notably, the majority of literature examining discharge to home transitions focuses on coordinating care appropriately.4–6 22 However, our findings illustrate that coordinating care is only one critical component of preparing patients for hospital discharge, particularly to SNFs.
Transitions from hospitals to SNFs are potentially different from discharges to home. The patient population being discharged to SNF tends to have a higher level of acuity (eg, older, medically complex/comorbid and cognitively physically frail),29 30 and requires higher engagement of caregivers and involvement by the interprofessional team compared with a typical inpatient admission.9 10 Although there are transition in care frameworks to guide hospital to home discharges, these tend to focus on patient education and care coordination activities as the critical process components.31–36 Likewise, there are general guidelines for discharge planning,36 as well as for hospital to nursing home transfers,37 but no best practice guidelines for SNF transitions. Building on prior work, our findings describe critical steps in addition to care coordination, and identify the mechanisms to best prepare patients for the transitions to SNFs. Clinicians and researches can use these findings to further identify new target points for redesigning care, develop high-impact interventions and establish evidence-based guidelines to improve hospital to SNF transitions (table 3).
This study has practical implications for improving the hospital-SNF discharge process. Our findings support prior work emphasising the importance of early identification of patients at high-risk for needing postacute care.10 Tools early in development might be useful for identifying patients who might warrant postacute care and what type of services are needed.38 39 However, our findings suggest that decisions made through an iterative, team-based process are most effective for preparing patients for successful postacute care transitions. Yet much of the current literature8–14 22 36 and decision support tools (ie, Nursing Home Compare28) focus on the SNF selection process after the type of services have been decided. Interestingly, we know little about whether and how patients are informed about the differences between settings.30 40 Because being discharged with postacute care services can be a pivotal life event for older adults, it is important to embrace this transition as an opportunity for ‘current life planning’. Given the complex nature of this decision, healthcare teams should engage patients and their caregivers as active partners in the process by discussing critical issues directly related to patient’s values, anticipated quality of life, healthcare goals and care preferences.30 40–43
An important driver of performance was hospitals’ ability to anticipate barriers to discharging patient to SNFs. Similar to the existing literature, we found that high-performing hospitals had pharmacist-led medication reconciliation,44 45 arranged transportation earlier9 and cross-trained staff to complete critical tasks like discharge paperwork.23 Additionally, we identified that high-performing hospitals also used standing order sets of PT that are triggered for high-risk patients, conducting hospital-SNF administration rounds on extended stay patients and implementing a pre-admission process to SNFs, and overall were better at anticipating SNFs questions/concerns. Notably, many of these subprocesses require the hospital to understand potential barriers from an SNF perspective. Since few inpatient providers have ever set foot in an SNF, this further highlights the need for cross-training across hospital and SNFs.
Although we present the stages in a linear fashion, they are interconnected, as illustrated by the multiple feedback loops in the flow-map (figure 1). The themes were clustered together in this fashion to make the process modular: it is possible that facilities could perform these stages and associated actions in different orders. For example, for some patient populations where procedures are planned (eg, knee replacement), the recognition, deciding, selecting and negotiation stages could occur prior to hospitalisation. To improve efficiency of hospital-SNF transitions, future work should explore how the identified stages and strategies can be best tailored to fit specific population and/or health system needs.
Our findings should be considered within their derived context. Strengths of our study include the purposive sample, multiple observer team and the rigorous approach used to triangulate, analyse and identify themes. However, the small sample size and short time-frame of visits limited our ability to capture changes over time and to understand all relevant social-cultural factors. Although we sought a diverse sample (online supplemental appendix C), we cannot rule out factors that might have been different during that time-period (ie, staff out with illness, day vs weekend shifts, preparing for a Joint Commission visit) or other contextual factors that may vary between site (ie, financial resources, policies, work environments). Additionally, performance was defined in terms of veteran outcomes, which may not be representative of all Medicare beneficiaries. However, a national sample was used to identify sites, and there are no data suggesting practices vary for veterans versus non-veterans, particularly since we focused on Medicare beneficiaries in this study. Last, due to the nature of thematic analysis, the identified themes are one interpretation of the data. We recommend clinicians use data in the tables and appendices to identity findings relevant to their local institution.
Conclusion
This study highlights the many complex issues around how hospitalised patients are prepared for SNF in clinical practice. Hospitals are increasingly looking to identify strategies to more cost-effectively and efficiently prepare patients for a safe discharge to SNFs. By identifying variations in processes used to prepare patients for SNF between high-performing and low-performing hospitals, we provide critical insight into the best practices for preparing patients for the transitions to SNFs and areas to target for improving care.
Data availability statement
Data sharing not applicable as no datasets generated and/or analysed for this study. All data relevant to the study are included in the article or uploaded as supplementary information. Additional information may be obtained by emailing kirstin.manges@pennmedicine.upenn.edu.
Ethics statements
Ethics approval
The study was approved by the local IRB at the study team’s primary site: seven sites ceded to the primary IRB, while two sites obtained separate approvals.
References
Footnotes
Twitter @Kirstin_Manges, @BBurkeMD
Contributors Study concept and design: KM, RA, CL, ML and REB. Data collection: RA, CL, ML and EG. Analysis and interpretation of data: KM, RA, CL, ML, EG and REB. Preparation of the manuscript: KM, RA, CL, ML, EG and REB.
Funding This study was funded by Agency for Healthcare Research and Quality (T32HS026116-02) and VA Health Services Research & Development (HSR&D) (Career Development Award).
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.