Exploring Predictors of Long-Term Care Facility Admissions in Stroke Survivors: Insights from a Taiwanese Hospital-Based Study

Introduction

Acute stroke constitutes a significant risk factor for admission into long-term care facilities (LTCFs), primarily attributable to sudden functional impairments in activities of daily living.^1–3 For those non-disabled before stroke, a disabling stroke has a relatively greater impact on the care burden and socioeconomic needs. Data from the literature suggest that the incidence of LTCF admissions among stroke survivors varies widely, ranging from 7% to 39%.² Notably, the Taiwan Stroke Registry reports a rate of 7%.⁴ Extended hospital stays often occur while preparing for transition to LTCF, not only increasing healthcare costs but also negatively affecting patients’ psychological and physical health.^1,5–8 Although numerous studies have sought to identify predictive risk factors for LTCF placement post-stroke, a definitive predictive model remains elusive.^2–4 Early identification and intervention targeting modifiable risk factors, such as early rehabilitation to enhance functional recovery, can reduce the need for long-term institutionalization and increase the chances of home return.^9,10 For non-modifiable risk factors, early recognition of patients at higher risk for LTCF admission enables timely discharge planning and coordination of appropriate care settings. Addressing these risk factors proactively leads to more efficient healthcare resource utilization and smoother transitions to suitable post-acute care residency.

Hence, our research focuses on identifying the risk factors for long-term institutionalization in stroke patients, specifically targeting those who transitioned from independence to disability as a result of the stroke.

Materials and Methods

Patient Selection and Definitions

This study sourced data from the hospital-based prospective stroke registry at Chi-Mei Medical Center, a teaching hospital with more than 1200 beds in Southern Taiwan. The stroke registry has been approved by the Ethics Committee of Chi-Mei Medical Center and conformed to the criteria of the nationwide Taiwan Stroke Registry.¹¹ In brief, the registry enrolled patients admitted within 10 days of stroke onset, using International Classification of Diseases-10 codes: I63 for ischemic stroke, G45 for transient ischemic attack, I61 and I62 for nontraumatic intracerebral hemorrhage, and I60 for nontraumatic subarachnoid hemorrhage. Patient characteristics, including demographic data, medical history, comorbidities, stroke severity, treatments, hospital course, and complications, were collected according to a pre-defined system. Stroke severity was assessed with National Institutes of Health Stroke Scale (NIHSS). Functional status before the stroke event and at discharge was categorized with the modified Rankin Scale (mRS) and Barthel index scores. The mRS ranges from 0 (no symptoms) to 6 (death), indicating different levels of disability and dependency. The Barthel Index is an ordinal scale that measures a person’s ability to perform activities of daily living, with scores ranging from 0 to 100, where higher scores indicate better functional ability. We restricted our cohort to patients aged 18 years or above and diagnosed with any type of stroke, either first-ever or recurrent, within the time frame of January 1, 2017, to December 31, 2022. Stroke was defined in accordance with the recommendations of the World Health Organization and categorized into either ischemic or hemorrhagic stroke, confirmed through computed tomography and/or magnetic resonance imaging. Patients who had a pre-admission mRS score of 3 or higher, or who were already residing in an LTCF before the stroke event, as well as those who had an mRS score of 2 or less at the time of discharge, or who died prior to hospital discharge, were excluded. For stroke survivors, the discharge disposition was recorded as home, nursing home, respiratory care ward, transition to hospital, or Post-Acute Care. Since 2014, Taiwan’s National Health Insurance has implemented a Post-Acute Care program for stroke patients, aimed at enhancing functional recovery through rehabilitation in regional or community hospitals. We defined nursing home and respiratory care ward as long-term care facilities (LTCF), and the rest as the non-LTCF group.

This study followed the Declaration of Helsinki and was approved by Chi-Mei Medical Center’s Institutional Review Board. As the study was retrospective in nature, the need for informed consent was waived. Patient data confidentiality was strictly maintained.

Statistical Analysis

Continuous variables are delineated as medians accompanied by interquartile ranges or mean accompanied by standard deviation, whereas categorical variables are expressed as frequencies and percentages. LTCF group and non-LCTF group were compared using the Mann–Whitney U-test for continuous data and either Pearson’s chi-square or Fisher’s exact test for categorical data. We assessed changes in Barthel Index scores from admission to discharge within groups and differences between LTCF group and non-LCTF group. Box plots visualized score distributions, central tendency, and variability. Logistic regression was employed to estimate the associations between LCTF admission and variables, presented as odds ratios (ORs) along with 95% confidence intervals (CIs). Significant predictors of LCTF admission (p < 0.05) identified in univariable analyses were further scrutinized using multivariable logistic regression. All statistical procedures were conducted using SAS software (version 9.4; SAS Institute, Cary, NC, USA), with a significance level set at p < 0.05.

Results

Our study, conducted from 2017 to 2022, initially included 6967 hospitalized stroke patients. After applying exclusion criteria, 4940 patients were excluded, leaving 2027 eligible for further analysis (Figure 1). Among the 2027 stroke patients enrolled in the study, 343 resided in LTCFs post-discharge (LTCF group), while 1684 did not (non-LTCF group).

Figure 1 Flowchart of patient selection in this study.

The baseline characteristics indicated that the LTCF group was generally older (p < 0.001) and had a higher proportion of NG tube (p < 0.001) and Foley catheter (p < 0.001) placement at discharge compared to the non-LTCF groups (Table 1). However, no significant differences were observed in comorbidities between the two groups, apart from dyslipidemia (p = 0.021) and atrial fibrillation (p = 0.035).

Table 1 Demographic and Clinical Characteristics of Included Patients

Table 2 showed that patients in the LTCF group exhibited higher discharge mRS scores and NIHSS scores, along with lower Barthel Index scores and GCS scores. The mRS scores prior to the stroke event did not significantly differ between the LTCF group and the non-LTCF group.

Table 2 Stroke Severity Scales and Functional Status of Included Patients

As depicted in Figure 2, Barthel Index scores remained statistically unchanged from admission to discharge in stroke survivors residing in LTCFs (p = 0.329), whereas a significant increase was observed in those not in LTCFs (p < 0.001). Comparative analysis revealed that the LTCF cohort had lower Barthel Index scores both at admission and discharge, relative to the non-LTCF cohort (p < 0.001).

Figure 2 Barthel Index Scores at Admission and Discharge for Patients Based on LTCF Admission. Abbreviation: LTCF, long-term care facility.

In univariable logistic regression, those characteristics associated with discharge to LTCFs included age (p = 0.006 for ≥ 80 years), atrial fibrillation (p = 0.035), stroke subtype (p < 0.001 for hemorrhagic vs ischemic stroke), higher discharge mRS score (p < 0.001), lower Barthel Index score (p < 0.001), NIHSS score ≥8 (p < 0.001), GCS score ≤12 (p < 0.001), NG tube at discharge (p < 0.001), Foley catheter at discharge (p < 0.001), length of hospital stay (p < 0.001 for ≥ 14 days) and ICU admission (p = 0.007) (Table 3). Dyslipidemia was associated with decreased likelihood of LTCF disposition (p = 0.022).

Table 3 Associations Between LTCF Admission and Selected Variables.

On multivariable analysis, discharge mRS score 5 (OR 3.22 [1.72–6.05]), discharge Barthel Index (OR 0.97 [0.96–0.98]), NG tube at discharge (OR 1.87 [1.29–2.71]) length of hospital stay (OR 1.49 [1.01–2.19]) and were significantly associated with LTCF admission.

Discussion

This retrospective study focuses on the clinical attributes associated with the discharge destination of acute stroke patients. We found that a discharge mRS score of 5, the discharge Barthel Index score, nasogastric tube placement, and a hospital stay of more than 14 days were significantly associated with admission into LTCFs. These factors serve as valuable clinical indicators, aiding physicians in identifying patient subgroups that may require proactive intervention and early discharge planning.

The previous study indicated that the association between pre-stroke mRS scores and stroke outcomes may be confounded by multiple factors.¹² Therefore, we specifically targeted stroke patients with pre-stroke mRS scores of 2 or lower and post-stroke mRS scores of 3 or higher. By selecting individuals who experienced a significant decline in functional status, we aimed to identify patients at increased risk for LTCF admission. This approach allows us to focus on identifying key risk factors and prioritizing patients who may benefit most from early interventions, ensuring that our findings have direct clinical relevance.

Our findings indicate that both post-stroke mRS scores and Barthel Index scores are crucial factors for admission into LTCFs for stroke survivors. These results are consistent with previous studies, which have shown that these indicators are reliable and well-established tools for assessing functional disability^13,14 and predicting the need for institutionalization.^8,15 The difference between the admission and discharge scores of the Barthel Index is minimal in our cohort, and this observation underscores the importance of early assessment for facilitating smoother discharge transitions based on patients’ needs.

The use of NG tubes is common among stroke patients with severe dysphagia, reflecting the severity of the disease.¹⁶ Swallowing difficulties may lead to increased risk of aspiration pneumonia and mortality,^16,17 and notably, it is associated with an increase in institutionalization rates.^8,18 The use of NG tube feeding not only has implications for patient care but also exerts a negative effect on caregivers’ well-being, creating an additional barrier to a smooth transition from hospital to community settings.^19,20 Active swallowing training may increase the likelihood of NG tube removal and reduce the risk of aspiration pneumonia.²¹ Therefore, the use of NG tubes in stroke patients serves as an important consideration in rehabilitation and discharge planning, often requiring a multidisciplinary approach to manage both the clinical and social implications effectively.²²

Previous literature demonstrated that a longer length of hospital stay is independently associated with an increased likelihood of institutionalization, which is consistent with our finding.^8,23 The determinants of prolonged length of stays are multifactorial and complex, with some non-modifiable factors such as the severity of stroke and preexisting comorbidities.²⁴ Of note, the time-consuming process of finding suitable care facilities, as well as their limited availability, are also key factors that extend the length of stay. This emphasizes the importance of prompt evaluations for LTCF placement. Effective inter-communication among stroke care teams and the early establishment of family expectations will be a key step.^25–27

Finally, age showed a significant association with discharge dispositions in our preliminary analysis, consistent with previous studies linking higher age to increased chances of LTCF admission.^2,8 However, when we adjusted for confounding variables such as mRS scores and the Barthel Index scores, age was no longer statistically significant. Given the potential complexities and challenges in caring for the elderly after a stroke,²⁸ this observation may suggest that functional disability, rather than age, is the more crucial factor in decision-making for LTCF transition.^29,30

Our study has several limitations. Firstly, this research was a retrospective analysis conducted in a single medical center and targeted specific stroke patients, potentially limiting the generalizability of the results. Secondly, data extracted from the electronic medical records in Chimei Medical Center’s stroke registry may be subject to missing data, including laboratory data and clinical care metrics, or potential misdiagnosis. A total of 8.8% of the patients were excluded due to missing data, which may have introduced selection bias. Additionally, pre-stroke mRS and Barthel index scores were obtained through patient or caregiver interviews, which may introduce self-report bias. Thirdly, our study may be subject to selection bias due to the specific inclusion criteria, which might not fully represent the broader stroke patient population. Additionally, while our focus was on discharge arrangements and institutionalization, key factors such as socioeconomic status and family support systems, which could significantly impact prognosis, were unavailable. Finally, although we assessed overall functional outcomes, we did not explore specific aspects of functional impairment, such as limb mobility, language skills, swallowing function, and cognitive abilities. As a result, our study may not fully capture the complexity of factors influencing LTCF admission.

Conclusion

In conclusion, this study identifies critical clinical factors—including discharge mRS and Barthel Index scores, as well as nasogastric tube placement and length of hospital stay—that are significantly associated with LTCF admission in stroke patients. These findings highlight the need for early assessments and comprehensive discharge planning to improve patient outcomes. Further research is essential to validate these results and explore other potential determinants, particularly socioeconomic variables, to create a more robust predictive model for LTCF admission. Additionally, studies may explore early and prompt interventions to improve this vital patient-centered outcome.