We addressed four Guiding Questions (GQ)

The goals of this Technical Brief are to 1) categorize and evaluate current knowledge regarding strategies to reduce length of stay (LOS) for medically complex, high-risk, or vulnerable patients at increased risk of extended LOS; 2) examine contextual factors (e.g., resources, costs, staffing, technology) that affect implementation of LOS-focused interventions; 3) identify emerging concepts or initiatives that may merit future research; and 4) develop a series of evidence maps to inform health systems' strategic efforts for LOS reduction in these populations.

GQ 1: What are the characteristics of interventions to decrease length of hospital stay, and how do they vary?

GQ 2: What are the contextual factors (e.g., resources, staffing, technology) that impact implementation of interventions to decrease hospital length of stay?

GQ 3: What is the current evidence addressing interventions to decrease hospital length of stay?

GQ 4: What future research is needed to close evidence gaps regarding interventions to decrease length of hospital stay?

PICOTS and inclusion criteria

Evidence Base

Our search of the published literature identified 4,364 potentially relevant studies, of which we excluded 1,227 at the title level (not relevant). We excluded 2,725 studies during abstract screening for one of the following reasons: intervention, population, or care setting was not relevant, the study design did not meet our inclusion criteria (e.g., narrative review), the abstract did not address one of the Guiding Questions (GQs), key outcomes were not reported, or studies in the systematic review were either conducted solely outside the United States or 50 percent or more of the studies reporting hospital length of stay (LOS) were conducted outside the United States. The most common reason a described intervention was considered "not relevant" was because it was not a hospital or health system-led intervention. This resulted in full-text screening of 412 articles. We excluded 392 studies at the full-text level. Reasons for exclusion at this level were similar to reasons listed for the abstract level (see Appendix B). We also received three studies through the Agency for Healthcare Research and Quality (AHRQ's) Supplemental Evidence and Data submission process. We excluded all three studies because they were not systematic reviews.

We included 19 systematic reviews in 20 publications, 1 of which was identified in our grey literature search.

Challenges for Local Implementation

To gauge to what extent these interventions might be successfully implemented for reducing LOS and improving other outcomes in a local setting, hospital administrators benefit from details about the local context and implementation factors (e.g., process and resources required). However, SRs provided only limited information. Thirteen reviews (in 14 articles) all described interventions conducted in multiple types of hospitals, including academic medical centers, community hospitals, and less frequently, Veterans Affairs hospitals. Only five reviews reported whether all included studies were conducted in urban, suburban, or rural settings, and few reviews reported hospital bed size or affiliation with a health system. In addition, the process and resources used to support implementation were often not reported. Not all primary studies informing the systematic reviews provided details about the expertise of staff leading and implementing interventions. For instance, for discharge planning interventions, only one review specified that included study interventions were led by a nurse practitioner. Otherwise, systematic reviews simply indicated a provider (e.g., nurse, clinician) or multidisciplinary team led or participated in implementing the intervention. Availability of current resources, such as staff with particular expertise, will undoubtedly affect the feasibility of successfully implementing many interventions. For example, several systematic reviews evaluated geriatric assessment, which often involved specialized assessment by a geriatrician. However, hospitals or health systems may not necessarily have a geriatrician to lead this intervention and may instead engage staff members for training to deliver this intervention, which may affect ultimate success.

Ultimately, we did not find evidence that most interventions have been widely replicated or scaled with sufficient detail or context to adequately inform local implementation.

Trade-Offs and Implications

System-level interventions have the potential to create trade-offs between outcomes, such as LOS and postdischarge adverse outcomes (e.g., hospital readmission, mortality). All systematic reviews in our evidence base reported LOS, and most reported readmissions and mortality. However, the manner in which outcomes were measured varied. Not only is it important for studies to evaluate these outcomes collectively, but also to standardize the way outcomes are reported.

Our findings suggest that, at present, no existing intervention or approach can be implemented to decrease LOS for broad populations of medically complex, high-risk, or otherwise vulnerable patient populations. Attempting to implement an unfocused broad-based intervention across varied populations may have unintended consequences and lead to worse outcomes. Hospitals and health systems may need to carefully consider their own local contexts and populations when assessing whether particular interventions would be a good fit. Input from our Key Informants emphasized the importance of considering factors associated with social care needs and ways to address these needs when seeking to reduce LOS with a system-level intervention. Building relationships and establishing partnerships with community organizations may help hospitals and health systems leverage resources to support and manage needs of medically complex, high-risk, and vulnerable patients postdischarge.

Finally, interpretation of these findings should also consider several key additional factors. Systematic review design lags behind primary research. Thus, our evidence base may not reflect the most recent findings or evolving interventions yet to be synthesized in published reviews. Moreover, the evidence we reviewed was generated before the COVID-19 pandemic. It is likely that the pandemic has led hospitals to innovate in myriad ways that may affect hospital care, LOS, and other critical outcomes that we cannot yet assess. Additionally, we focused on medically complex, high-risk, and vulnerable patient populations, including those at high socioeconomic risk of poor medical outcomes, but found little direct evidence on socioeconomically disadvantaged patients. Emerging efforts to address longstanding social, economic, and health inequities may yield new insights on how to best design care to benefit these patients.

Overall, understanding the unique challenges and needs of a hospital or health system and its surrounding community may help inform the development of a strategic plan to implement a system-level intervention to reduce hospital LOS and provide high-quality care for the patient populations served.

Next Steps

Hospital Administrative Leaders Can Do the Following:

• Understand different populations with varying risk levels within hospitals attempting to reduce LOS.
• Explore specific interventions matched to medically complex, high-risk, and vulnerable populations with higher LOS.
• Maximize expertise of current staff when identifying and implementing system-level intervention (e.g., clinical pathways, geriatric assessment).
• Understand tradeoffs between reducing LOS in medically complex, high-risk, and vulnerable populations and other patient-centered outcomes (e.g., functional decline, patient experience, mortality, readmissions) and patient safety and quality metrics.
• Evaluate opportunities to support research and implementation of system-level interventions targeting medically complex, high-risk, or vulnerable populations.
• Work with policymakers to identify best approaches to reducing hospital LOS in U.S. healthcare delivery systems.

Researchers Can Do the Following:

• Conduct research focused on general medical and surgical ward inpatients.
• Provide sufficient operational context about how interventions were implemented in primary studies and evidence syntheses.
• Report details about local hospital settings where initiatives to reduce LOS have been implemented (e.g., patient volume, bed size, payer mix) in primary studies and evidence syntheses.
• Assess health information technology's role in supporting interventions to reduce LOS and identify opportunities to develop or adapt technology to support new initiatives.
• Include and subgroup patients facing severe social and economic barriers to achieving and maintaining wellness before, during, and after hospitalization in primary studies evaluating system-level interventions to decrease LOS.
• Evaluate how health systems can address the unique challenges pediatric populations' face, specifically those that are susceptible to many sources of social and economic vulnerability.
• Examine enhanced recovery programs and patient mobility programs in medically complex or otherwise vulnerable patient populations.
• Conduct well-designed systematic reviews, such as those assessing both the risk of bias of primary studies and providing the strategy for the literature search.

Policymakers Can Do the Following:

• Support new research and development with additional funding of both primary research and evidence synthesis applicable to the unique characteristics of the U.S. healthcare delivery system.
• Address the role of LOS — as both a metric and a concept — in value-based reimbursement systems.

This review does not provide cost information.

Limitations

Inconsistent Evidence on Effectiveness of Interventions

The evidence base highlights inconsistencies on the effectiveness of interventions to reduce LOS; no intervention demonstrated a clear direction of effect. For example, three systematic reviews evaluating discharge planning compared with usual care in either older adults or patients with chronic conditions found no difference between groups for LOS, while two found that discharge planning decreased LOS, and one found that discharge planning increased LOS. Similarly, findings varied for readmissions and mortality. Reviews examining geriatric assessment, decision support, interdisciplinary care, and telehealth also reported heterogeneous and sometimes inconsistent findings for quantitative syntheses. However, one SR examining case management and one SR assessing clinical pathways in patients with heart failure found a reduction in LOS, a lower risk of readmissions, and a lower risk or odds of mortality compared with usual care. The findings suggest that case management or clinical pathways may have a consistent direction of effect for these outcomes in patients with heart failure, but more research is needed.