A high-risk population for stroke recurrence exists among home-based stroke survivors discharged from an acute care hospital: A retrospective analysis

Abstract

Objectives

To examine whether the factors determining discharge destination after acute-phase treatment for stroke differ based on recurrence risk levels.

Design

Retrospective study.

Setting

General acute care hospital.

Participants

Acute ischemic stroke survivors (n = 1219).

Main outcome measures

Patients were stratified using the Stroke Prognosis Instrument II (SPI-II) and evaluated through forced-entry multiple logistic regression analysis. Among the independent variables, the primary outcome measure was the modified Rankin Scale (mRS) at discharge. Covariates included age, sex, and histories of diabetes, cerebral infarction, cerebral hemorrhage, coronary artery disease, hypertension, and congestive heart failure. The dependent variable grouped participants into those discharged home and those discharged to a medical facility.

Results

Among the 1219 included participants, 914 were classified into the home care group and 305 into the medical facility care group. SPI-II-based stratification revealed that 78.665% of the home care group had a moderate or higher risk of stroke recurrence. Multiple logistic regression analysis demonstrated that mRS at discharge was a significant factor across all stratified models, while age was a significant factor only within the moderate-risk group.

Conclusions

Discharge decisions after acute-phase treatment were primarily influenced by short-term functional independence, as reflected by mRS, rather than recurrence risk levels. Thus, a substantial proportion of stroke survivors with a high recurrence risk transitioned to home care. These findings highlight the need to revise current medical and social welfare services and to develop targeted strategies for stroke recurrence prevention, based on a more detailed understanding of the living environments of stroke survivors.

Introduction

According to the World Health Organization (WHO), stroke is the second leading cause of death worldwide. ¹ Stroke significantly impairs patients’ activities of daily living (ADL), with more than two-thirds of stroke survivors experiencing lasting disabilities that require assistance. ² However, advances in imaging technology for cerebral infarction have expanded the applicable time window for intravenous tissue plasminogen activator (tPA) therapy and endovascular treatment (EVT), reducing stroke severity and significantly improving modified Rankin Scale (mRS) scores at 90 days. ³ This suggests that the widespread adoption of tPA therapy and EVT has led to early functional improvements.

Despite these advancements, the COMPASS trial, which aimed to expand home care for stroke survivors, found that none of its 871 discharged patients were able to fully manage all major risk factors. Additionally, 70.5% did not receive home care, and 77.2% did not attend outpatient treatment. ⁴ Stroke survivors generally have multiple risk factors, with recurrence rates ranging from 5.7% to 51.3%. ⁵ This indicates that many stroke survivors fail to meet their self-management needs after discharge, leading to an increased risk of recurrence.

From the perspective of acute-phase healthcare providers, the absence of a clearly defined discharge strategy distinguishing home care from medical facility care may contribute to this issue. Previous studies examining factors influencing discharge destination after acute-phase treatment have primarily aimed to identify patients eligible for home care and optimize public healthcare expenditures.^6,7 Prior research has reported that neurological severity during acute-phase treatment, age, and family caregiving capacity are significant factors in transitioning to home care. These findings indicate that short-term ADL independence heavily influences discharge strategies, and stroke risk factors do not lead to major strategy adjustments unless they become apparent during hospitalization.

Thus, the current discharge strategies may lack a focus on stroke recurrence prevention, potentially resulting in some stroke survivors transitioning to home care despite remaining at high risk because of acute-phase discharge strategies. Therefore, identifying high-risk groups among stroke survivors discharged to home care is crucial for ensuring appropriate medical and social welfare services. This study aims to retrospectively examine whether the factors determining discharge destination after acute-phase treatment differ based on recurrence risk levels.

Methods

Study design and setting

This study included 1563 patients with acute ischemic stroke who were admitted to a 733-bed regional core hospital in the Saitama Prefecture, Japan, between April 2017 and March 2021. The study was approved by the Ethics Committee (Approval No.: 1091) and was conducted in accordance with the Declaration of Helsinki. As this was a retrospective study, informed consent was obtained using an opt-out approach.

Study participants

The study included 1563 patients diagnosed with acute ischemic stroke (International Classification of Diseases, 10th Revision: I630, I631, I632, I633, I634, I635, I638, I639) between April 2017 and March 2021. Of these, 344 patients were excluded because of death during hospitalization (n = 121), pre-hospital residence in medical facilities (n = 146), or incomplete data (n = 77). Ultimately, 1219 patients were included in the analysis. Figure 1 shows the flowchart illustrating the selection process for the included patients. The 1219 included patients were classified into two groups based on their discharge destination after acute-phase treatment: home care group and medical facility care group.

Figure 1.

Flowchart of participant inclusion for analysis.

Data collection

Data were collected from medical records during hospitalization. The information extracted included age, sex, body mass index (BMI), and pre-admission mRS score. Data on stroke risk factors included histories of diabetes, cerebral infarction, cerebral hemorrhage, coronary artery disease, hypertension, and congestive heart failure. Additional clinical information included stroke classification based on the National Institute of Neurological Disorders and Stroke III (NINDS III), administration of tPA, and EVT.

Discharge-related details, such as mRS at discharge, length of hospital stay, and discharge destination, were also collected. In this study, mRS at discharge was interpreted as an indicator of short-term functional independence. The mRS is the most widely used measure for assessing functional independence in contemporary stroke research, and data collection in this study was conducted based on mRS interpretation. ⁸ The rehabilitation intervention data, including the daily duration of physical therapy, occupational therapy, and speech therapy during hospitalization, were extracted. Medication prescriptions at discharge, specifically aspirin, cilostazol, and clopidogrel, were recorded.

The Stroke Prognosis Instrument II (SPI-II) was used to classify patients into low (0–3 points), moderate (4–7 points), and high (8–15 points) risk of recurrence (Supplementary Table S1). The SPI-II consists of the following parameters: age over 70 years (2 points), hypertension (1 point), diabetes (3 points), history of stroke (3 points), stroke as the index event (2 points), coronary artery disease (1 point), and congestive heart failure (3 points). Data related to the SPI-II were collected from medical records, based on information obtained through physician assessments and patient or family interviews. The SPI-II, composed of stroke risk factors, is widely used as a prognostic tool for short- and long-term stroke recurrence risk. It has also been utilized in observational studies to classify stroke patients according to recurrence risk and to investigate associations with novel clinical indicators.^9–11 The duration of individual rehabilitation interventions, including physical therapy, occupational therapy, and speech therapy, was calculated for each patient.

Statistical analysis

Comparisons between the home care and medical facility care groups were performed. Continuous variables were expressed as median (interquartile range, min–max) based on Shapiro–Wilk test results. Variability was assessed using Bartlett's test for normally distributed data and Levene's test for non-normally distributed data. As all variables exhibited non-normal distribution, between-group comparisons were conducted using the Mann–Whitney U test. Binary variables were presented as counts and percentages, and compared using Pearson's χ² test.

To evaluate the factors influencing discharge destination, a forced-entry multiple logistic regression analysis was performed for the 1219 patients included in the study.

The dependent variable was discharge destination, dichotomized into home care or medical facility care. The independent variables included mRS at discharge, age, sex, and history of stroke risk factors (diabetes mellitus, ischemic stroke, cerebral hemorrhage, coronary artery disease, hypertension, and congestive heart failure).

In this model, mRS at discharge was defined as the primary explanatory variable of interest, while age, sex, and comorbidities were treated as covariates. Furthermore, to examine whether the relationship between neurological disability and discharge destination varied according to recurrence risk, the 1219 patients were stratified into three groups based on the SPI-II: low (n = 215), moderate (n = 619), and high (n = 385) risk. The same logistic regression model was then applied within each risk group. The impact of each factor in the model was assessed by calculating odds ratios (ORs). Model validity was evaluated using the model χ² test and the area under the receiver operating characteristic (ROC) curve (AUC). Statistical significance was defined as p < 0.05. All analyses were performed using JMP^® version 11.4 (SAS Institute Inc., Cary, NC, USA).

Results

Between April 2017 and March 2021, a total of 1563 patients with ischemic stroke were admitted to the hospital. Of these, 344 patients were excluded based on the exclusion criteria. Among the remaining 1219 patients, 914 were classified into the home care group, while 305 were in the medical facility care group (Figure 1).

The results of the stratification based on the SPI-II are shown in Supplementary Table S2. Among the 215 patients in the low-risk group, 195 were in the home care group, while 20 were in the medical facility care group. Among the 619 patients in the moderate-risk group, 445 were in the home care group, while 174 were in the medical facility care group. Among the 385 patients in the high-risk group, 274 were in the home care group, while 111 were in the medical facility care group. In the home care group, 78.665% of patients were classified as moderate or high risk, whereas in the medical facility care group, this proportion was 93.442%.

A comparison of hospitalization-related information between the home care and medical facility care groups is presented in Table 1. The median age was significantly lower in the home care group than in the medical facility care group (74 vs. 79 years, respectively; p < 0.001). The median length of hospital stay was also significantly shorter in the home care group than in the medical facility care group (19 vs. 45 days, respectively; p < 0.001). The median mRS score at discharge was lower in the home care group than in the medical facility care group (1 vs. 4, respectively; p < 0.001).

Table 1.

Comparison of hospitalization-related information between the home and medical facility care groups.

Characteristics	Home care group N = 914	Medical facility care group N = 305	p-Value
Continuous variables	Median (IQR a )	Median (IQR a )
Age	74 (66.75–80)	79 (74–85)	<0.001
Length of hospital stay	19 (13–36)	45 (33–65)	<0.001
Body mass index	23.337 (21.145–25.956）	22.266 (20.005–24.732)	<0.001
Pre-admission mRS b	0 (0–1)	1 (0–5)	<0.001
Discharge mRS b	1 (1–2)	4 (3–4)	<0.001
Daily physical therapy duration (minutes/day)	65.714 (60–75.163)	69.333 (62.676–77.129)	<0.001
Daily occupational therapy duration (minutes/day)	63.529 (58.489–70)	62.752 (57.777–67.194)	0.031
Daily speech therapy duration (minutes/day)	40 (0–45)	39.332 (34.041–44.125)	0.008
SPI-II c total score	5 (4–8)	6 (4–8)	<0.001

^aInterquartile range.

^bModified Rankin Scale.

^cStroke Prognosis Instrument II.

^dTissue plasminogen activator.

^eEndovascular treatment.

The mean duration of physical therapy was significantly shorter in the home care group compared to the medical facility care group (65.714 vs. 69.333 minutes per day, respectively; p < 0.001). Conversely, the mean duration of occupational therapy was significantly longer in the home care group than in the medical facility care group (63.529 vs. 62.752 minutes per day, respectively; p = 0.031). Similarly, the mean duration of speech therapy was longer in the home care group than in the medical facility care group (40.000 vs. 39.332 minutes per day, respectively; p = 0.008).

The median SPI-II score was significantly lower in the home care group than in the medical facility care group (5 vs. 6, respectively; p < 0.001). Regarding sex distribution, the proportion of males was significantly higher in the home care group than in the medical facility care group (64.33% vs. 55.08%, respectively; p = 0.003). The NINDS III stroke subtype classification showed significant differences between the two groups (p < 0.001).

The rate of tPA administration was significantly lower in the home care group than in the medical facility care group (8.210% vs. 13.110%, respectively; p = 0.011), as was the rate of EVT (2.190% vs. 5.250%, respectively; p = 0.006). At discharge, the prescription rate of aspirin was higher in the home care group than in the medical facility care group (32.930% vs. 25.250%, respectively; p = 0.012), as was the prescription rate of clopidogrel (42.120% vs. 33.440%, respectively; p = 0.007). However, there was no significant difference in the cilostazol prescription rate between the two groups. The prevalence of diabetes, cerebral infarction, cerebral hemorrhage, coronary artery disease, hypertension, and congestive heart failure did not differ significantly between the two groups.

A forced-entry multiple logistic regression analysis was performed on the 1219 patients with acute ischemic stroke (Table 2). Age (OR = 1.040) and mRS at discharge (OR = 2.127) were identified as significant factors determining the discharge destination after acute-phase treatment. The model χ² test was significant (p = 0.006), and the AUC was 0.824, indicating good model fit (Figure 2).

Table 2.

Multivariate logistic regression analysis for all patients.

	OR a	95% CI b	p-Value
Age	1.04	1.023–1.058	<0.001
Sex (male)	0.812	0.672–1.273	0.629
Discharge mRS c	2.127	1.908–2.381	<0.001
History of diabetes	0.812	0.568–1.153	0.249
History of cerebral infarction	1.104	0.752–1.611	0.607
History of cerebral hemorrhage	1.283	0.630–0.252	0.478
History of coronary artery disease	0.992	0.640–1.525	0.974
History of hypertension	0.894	0.636–1.254	0.519
History of congestive heart failure	0.785	0.487–1.251	0.314

^aOdds ratio.

^bConfidence interval.

^cModified Rankin Scale.

Figure 2.

Receiver operating characteristic (ROC) curves and area under the curve (AUC) derived from multivariate logistic regression analyses stratified by SPI-II risk levels.

Note: SPI-II risk stratification was used to account for confounding by stroke recurrence risk and to evaluate the impact of individual factors, including discharge mRS, within each risk group. Multivariate logistic regression analyses were performed to identify factors associated with discharge destination. ROC curves were generated for each model, and AUC values were calculated to assess the discrimination performance within the sample.

To assess whether the factors influencing discharge destination differed by stroke recurrence risk levels, additional logistic regression analyses were performed for each SPI-II risk category. Among the low-risk group (215 patients), mRS at discharge (OR = 2.125) was identified as a significant factor (Table 3). The model χ² test was significant (p = 0.006), and the AUC was 0.762, indicating a moderate model fit (Figure 2).

Table 3.

Multivariate logistic regression analysis stratified by SPI-II risk level.

	SPI-II low-risk group			SPI-II moderate-risk group			SPI-II severe-risk group
	OR a	95% CI b	p	OR a	95% CI b	p	OR a	95% CI b	p
Age	0.987	0.939–1.042	0.623	1.062	1.030–1.097	<0.001	1.025	0.993–1.060	0.134
Sex (male)	1.768	0.545–7.087	0.373	0.836	0.532–1.317	0.44	0.889	0.527–1.509	0.663
Discharge mRS c	2.125	1.482–3.129	<0.001	2.443	2.095–2.878	<0.001	1.744	1.471–2.091	<0.001
History of diabetes	-	-	-	1.44	0.741–2.786	0.278	0.630	0.370–1.069	0.087
History of cerebral infarction	-	-	-	0.815	0.330–1.909	0.646	1.303	0.792–2.151	0.297
History of cerebral hemorrhage	-	-	-	1.322	0.263–5.101	0.703	1.232	0.529–2.785	0.62
History of coronary artery disease	5.228	0–21.098	0.998	0.619	0.259–1.386	0.259	1.271	0.743–2.172	0.378
History of hypertension	0.985	0.322–2.796	0.978	0.939	0.578–1.520	0.800	1.379	0.432–5.427	0.611
History of congestive heart failure	3.074	0–534.596	0.999	0.545	0.167–1.604	0.291	0.856	0.488–1.481	0.581

^aOdds ratio.

^bConfidence interval.

^cModified Rankin Scale.

For the moderate-risk group (619 patients), age (OR = 1.062) and mRS at discharge (OR = 2.443) were identified as significant factors (Table 3). The model χ² test was significant (p < 0.001), and the AUC was 0.856, indicating good model fit (Figure 2).

For the high-risk group (385 patients), mRS at discharge (OR = 1.744) was identified as a significant factor (Table 3). The model χ² test was significant (p < 0.001), and the AUC was 0.762, indicating a moderate model fit (Figure 2). Across all risk levels, none of the pre-existing stroke risk factors, such as a history of diabetes, hypertension, or cardiovascular disease, were identified as significant determinants of discharge destination after acute-phase treatment.

Discussion

The results of the forced-entry multiple logistic regression analysis demonstrated that mRS at discharge was a significant factor in determining the discharge destination after acute-phase treatment. While the mRS is widely recognized as a prognostic indicator for stroke, ⁸ its significance was maintained even after stratification by SPI-II, which reflects recurrence risk. In contrast, other stroke risk factors, including sex, history of diabetes mellitus, ischemic stroke, cerebral hemorrhage, coronary artery disease, hypertension, and heart failure, did not show significance either before or after stratification. This may reflect a trend in which short-term functional independence among stroke patients is increasingly prioritized, regardless of recurrence risk, as a result of efforts to optimize public healthcare expenditures in acute-phase treatment. In other words, unless stroke risk factors become clearly apparent and lead to a decline in functional independence during hospitalization, they may not be considered in the decision-making process for discharge destinations. This study highlights the existence of unresolved challenges in acute-phase discharge strategies from the perspective of stroke recurrence prevention.

In the home care group, 78.665% of patients were classified as having moderate or high recurrence risk according to the SPI-II. This result indicates that under the current discharge strategies, a substantial proportion of patients transitioned to home care despite having a significant risk of stroke recurrence. The prescription rate of antiplatelet therapy at discharge was significantly higher in the home care group, and all patients had scheduled outpatient follow-up appointments before discharge (data not shown). This suggests that healthcare providers involved in acute-phase treatment were somewhat aware of the recurrence risk and made efforts for ongoing interventions through pharmacotherapy and outpatient care.

However, the COMPASS trial reported that over 70% of 871 stroke survivors did not receive home or outpatient care. Given the high long-term recurrence rates among stroke survivors, it is conceivable that current acute-phase care practices may insufficiently incorporate strategies to enhance outpatient follow-up continuity after discharge.

Among the covariates, age was identified as a significant factor only within the moderate-risk group stratified by SPI-II. This finding suggests that the influence of age on discharge decision-making is not uniform but varies according to risk stratification. Furthermore, in the moderate-risk group, the odds of choosing home care increased by 6% with each additional year of age. This may imply that factors related to physical frailty and caregiving support associated with aging, ⁶ along with functional independence, influenced discharge destination decisions.

Thus, this study revealed that stroke risk factors are not sufficiently considered as determinants of discharge destination after acute-phase treatment. Accordingly, it is suggested that post-discharge living environments and regional healthcare systems play a more critical role in stroke recurrence prevention.

Nevertheless, stroke survivors living at home face various healthcare disparities that may significantly impact their recurrence risk. Large-scale prospective studies have reported that equitable access to existing healthcare systems is not guaranteed, due to factors such as race, sex, and geographic characteristics.^12,13 Stroke survivors experience inequities in multiple domains, including eligibility for rehabilitation, patient education, and access to acute-phase care.^14,15 While the European Stroke Organization recommends the management of risk factors such as hypertension and atrial fibrillation for stroke prevention, support measures have not yet been standardized across countries due to differences in healthcare system policies. ¹⁶ These findings suggest that stroke survivors living at home encounter multiple challenges, and that in environments where recurrence prevention measures are insufficient, patients at considerable recurrence risk may continue to receive home care.

In recent years, rehabilitation programs incorporating elements of cardiac rehabilitation have been introduced to ensure continuity of rehabilitation opportunities after discharge. A 12-week exercise program for home-based stroke survivors was reported to significantly improve physical function 6 months later. ¹⁷ Such stepwise exercise programs may serve as effective strategies for high-risk stroke survivors identified in this study.

This study was conducted at a 733-bed regional core hospital located in Ageo City, Saitama Prefecture, Japan. The number of hospitals (2.20) and hospital beds (432.71) per 100,000 population in this area aligns with the average values reported by the Organization for Economic Co-operation and Development (OECD). ¹⁸ In the multiple logistic regression analysis of 1219 patients, mRS at discharge was identified as the most influential factor in determining discharge destination. This finding is consistent with previous studies and supports the notion that discharge strategies in acute care hospitals are heavily dependent on short-term functional independence. Given the hospital's characteristics and the large sample size, the external validity of this study is considered high.

Several limitations of this study should be noted. First, manual data collection from medical records led to a relatively smaller sample size compared with some previous studies. Therefore, larger studies are needed to better understand the characteristics of stroke survivors based on healthcare support systems. Second, the presence of unmeasured confounding factors should be considered. In home environments, stroke survivors may have access to visiting services according to their needs. Human support for dietary management and medication adherence in these settings could indirectly influence stroke recurrence risk. ¹⁹ Sociodemographic characteristics such as educational background and income have also been reported to be closely associated with discharge destination after stroke treatment, and these aspects might have affected the results. ²⁰ These data are not routinely recorded in standard clinical documentation practices; and the single-center, retrospective design limited our ability to capture such information comprehensively. Future research should involve broader collaboration with administrative bodies and community healthcare institutions to systematically collect and analyze these important sociodemographic and environmental factors. Third, this study did not assess long-term outcomes after acute-phase treatment. Extended study periods are necessary to analyze outpatient follow-up histories and recurrence rates accurately. Future research should address these limitations and focus on developing clinical strategies to enhance self-management support for home-based stroke survivors.

Conclusion

This study demonstrated that mRS at discharge was significant in determining the discharge destination after acute-phase treatment. This relationship remained consistent even after SPI-II risk level stratification. While advancements in acute-phase treatment have led to short-term ADL independence becoming a primary criterion for discharge strategies, numerous challenges remain in the home care environment.

This study highlights that a significant proportion of patients with a high risk of stroke recurrence are discharged to home care, despite these challenges. This suggests the need for a revision of the medical and social welfare services currently provided to stroke survivors. Future research should focus on a detailed investigation of the home environment of stroke survivors to further clarify the characteristics of high-risk populations and develop targeted recurrence prevention strategies.