Introduction

Hospital readmissions pose a significant challenge within the healthcare landscape, with profound implications for patient outcomes1 and healthcare costs2. In the United States, annual expenditures on hospital readmissions exceed $50 billion, with Medicare bearing a substantial portion of these costs2. As a crucial quality metric for healthcare institutions, hospital readmissions can indicate potential issues at discharge or post-hospital care deficiencies, prompting national healthcare reform initiatives to reduce readmission rates by enhancing patient care quality3. Research has shown that 1-year readmission rates for stroke patients can vary widely, ranging from 30–62.2%[3], with approximately 16% experiencing two or more readmissions within 30 days of their initial stroke incident4. Patients with stroke readmissions face elevated mortality risks, increased disability levels, and greater healthcare resource utilization compared to those without readmissions1.

The complexity of stroke hospital readmissions lies in its multifaceted nature, encompassing patient, organizational, and environmental factors. While previous reviews have explored and found common associated factors such as advanced age, they were limited by a small number of studies (10–24 studies) leading to narrow scopes of inquiry, typically covering only a fraction of potential influences3,5,6,7. Consequently, the narrower scope of these reviews left less frequently studied or emerging predictors of stroke readmission insufficiently explored. Of the four existing reviews on stroke readmissions, three (Litchman et al., 2010; Zhong et al., 2016; Rao et al., 2016) were published nearly a decade ago, and the most recent (Deng et al., 2021) focused solely on 30-day readmissions3,5,6,7. To our knowledge, this is the first systematic review in nearly ten years to examine 90-day and 1-year readmissions alongside 30-day readmissions. Conflicting findings for 30-day readmission factors, such as atrial fibrillation reported as significant by Deng et al. (2021) but not by Zhong et al. (2016) and coronary artery disease, found significant by Zhong et al. (2016) but not by Deng et al. (2021), further underscore the need for a more comprehensive investigation5,7. To address these gaps, our research conducts a rigorous systematic review approach, synthesizing findings from over 100 studies across 18 countries to examine 132 factors associated with 30-day, 90-day, and 1-year ischemic stroke readmissions. Additionally, a meta-analysis was performed to identify factors specifically associated with 30-day readmissions in the United States.

Method

Literature search

Our literature review on factors associated with ischemic stroke readmissions involved a systematic search of PubMed and Web of Science for studies published between January 1, 2000, and February 5, 2024. PubMed was chosen for its extensive biomedical literature, making it a valuable resource for stroke readmission topics. Web of Science was selected for its multidisciplinary scope and citation tracking capabilities. We conducted an electronic search using predefined search terms in article titles or abstracts, specifically “Readmission” and “Stroke” or “Cerebral Infarction” or “Brain Infarction.” In PubMed, our search query was structured as follows: “readmission [Title/Abstract] AND (stroke [Title/Abstract] OR ‘cerebral infarction’ [Title/Abstract] OR ‘brain infarction’ [Title/Abstract]).” In Web of Science, we used the following query format: “readmission AND stroke (Topic) OR readmission AND ‘cerebral infarction’ (Topic) OR readmission AND ‘brain infarction’ (Topic).”

Inclusion and exclusion criteria

Eligible studies focused on adults aged 18 years and older and examined associated factors for ischemic stroke or stroke readmissions within a 30-day, 90-day, or 1-year timeframe. Studies on stroke readmissions were included, as ischemic stroke is the most common stroke subtype. The exclusion criteria encompassed non-English studies; studies that excluded ischemic stroke; those focused on stroke prevention interventions such as carotid artery stenting (CAS) or carotid endarterectomy (CEA); studies examining readmissions solely related to specific patient sub-groups such as individuals with diabetes; trend analyses, literature reviews, meta-analyses, opinion pieces, abstracts, and posters; studies that did not stratify stroke in their analysis when other diseases were involved; studies not stratifying stroke readmission and mortality; and investigations into postpartum or preoperative stroke. When studies provided data stratified by stroke subtypes, we prioritized ischemic stroke data. If stroke subtypes were reported but not stratified, we confirmed that ischemic stroke was the most common subtype represented. Studies with a timeframe of 28 to 31 days were categorized as 30-day readmission studies. The outcome measures were defined as factors associated with readmission within a 30-day, 90-day, or 1-year timeframe.

Data extraction and quality assessment

Identified studies were imported into Microsoft Excel to create the initial dataset. Duplicate records and non-English articles were removed. The remaining titles and abstracts were screened by EM using the predefined inclusion and exclusion criteria. Full-text eligibility concerns were reviewed independently by NH, VA, and ZR, with discrepancies resolved through discussion and consensus. No automation tools were used in the selection process.

Risk of bias and quality assessments were performed by EM using the appropriate tool for each study design. Oversight and guidance were provided by coauthors with relevant methodological and clinical expertise (NH, ZR, VH, SZ, and ZT). Discrepancies were resolved through group discussion and consensus. For quality assessment, we used the Cochrane Risk of Bias Tool for randomized controlled trials (RCTs), the Risk Of Bias In Non-randomized Studies - of Interventions (ROBINS-I) tool for non-randomized studies, the Newcastle-Ottawa Scale (NOS) for cohort and case-control studies, and the Modified Newcastle-Ottawa Scale (Modified NOS) for cross-sectional studies (refer Tables S4–S11, provided in full in the supplementary information).

Data synthesis and statistical analysis

Study characteristics, including country of origin, readmission rate range, mean, and standard deviation, were reported. Studies lacking explicit readmission rates were excluded from calculations of the mean and standard deviation of readmission rates8,9,10,11,12,13,14,15,16,17,18,19. A pie chart depicting the number of studies per country is provided. Tables organized by readmission timeframe, categorizing the studies by design, sample size, readmission rate, readmission type, population, stroke type, country, model used, and identified factors, are included in the supplementary information (see Tables S1–S3).

A meta-analysis of 30-day ischemic stroke readmission factors in the U.S. was performed using R Studio with the metafor package. Studies were selected from the final dataset if they reported 30-day readmission factors in U.S. populations. Only factors investigated in at least five independent studies were included in the quantitative synthesis; this threshold was applied to ensure sufficient data for meaningful meta-analysis. Log odds ratios for 30-day readmissions were used for forest plots, with reported odds ratios converted where necessary. For studies that lacked odds ratios but provided frequency data, log odds ratios were manually calculated. A random-effects model accounted for heterogeneity, assessed using Chi-squared and I² statistics. For studies with substantial heterogeneity (P ≤ 0.10 and I² > 50%) where the source of heterogeneity could not be identified, pooled estimates were presented but interpreted with caution. Publication bias was evaluated using inverted funnel plots and Egger’s test.

Results

In our systematic search across PubMed and Web of Science databases, we initially identified 2,922 relevant studies. After removing 37 non-English studies and 1,072 duplicates, our dataset was reduced to 1,813 unique studies. Abstract screening excluded 1,675 studies due to incorrect disease focus, literature reviews, meta-analyses, or opinion pieces, leaving 138 potentially eligible studies for full-text retrieval. Upon further review, we excluded two studies involving participants under 18 years old20,21 and another study exclusively focused on subtypes of ischemic stroke1 resulting in a final inclusion of 135 studies. Figure 1 shows the PRISMA flow diagram.

Descriptive characteristics of the included studies

Of the analyzed studies, 104 (70%) focused on 30-day readmissions1,4,8,9,11,12,13,14,15,17,19,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111,112,113,114, 21 (14%) on 90-day readmissions10,18,24,36,71,73,86,91,94,115,116,117,118,119,120,121,122,123,124,125,126, and 23 (16%) on 1-year readmissions16,35,60,66,72,116,121,127,128,129,130,131,132,133,134,135,136,137,138,139,140,141,142 (some studies examined more than one readmission timeframe). Publication trends indicated a notable increase, with seven studies published between 2000 and 2010 and 128 published between 2011 and 2024. The study spanned 18 countries, with the United States leading (84), followed by Taiwan (10), China (9), Australia (7), Denmark (4), and Sweden (3). Other countries, including Canada, England, Japan, Norway, Singapore, and South Korea, contributed 2 studies each, while France, Israel, Italy, Malaysia, Portugal, and Turkey each had 1 study (see Fig. 2).

Fig. 1
figure 1

PRISMA flow diagram.

Full size image
Fig. 2
figure 2

Percentage of studies per Country.

Full size image

Readmission rates

The 30-day readmission rates ranged from 1.7 to 28.8% (mean 11.0%, SD = 4.2%), 90-day readmission rates ranged from 8.6 to 31.5% (mean 19.2%, SD = 6.4%), and 1-year readmission rates varied between 11.6 and 67.0% (mean 40.6%, SD = 13.8%). The 30-day readmission rates varied across different countries, with the United States rates ranging from 2.9 to 21.0%. In comparison, readmission rates in other countries were 5.0–15.1% in Australia, 7.0–28.8% in China, 9.1–11.5% in Denmark, and 8.1–15.5% in Taiwan.

Factors associated with 30-day readmissions

Examined readmission factors can be grouped into seven categories: sociodemographic, discharge location, patient social behavior, hospital level, chronic diseases, cardiovascular diseases, complex conditions, and stroke-related factors. Sociodemographic factors associated with increased 30-day readmissions included older age, which was significant in 16 studies: 15 of good quality1,25,35,48,49,57,68,76,78,83,84,93,94,99,111 and one of moderate quality106. Reliance on public health insurance (e.g., Medicare, Medicaid) or self-payment was significant in 11 studies: nine of good quality37,38,41,64,80,83,90,101,111 and two of moderate quality61,63. Unemployment or retirement was significant in two good quality studies66,101, and socioeconomic disadvantage status was significant in three good quality studies37,83,84.

Discharge location and hospital level factors associated with significantly higher readmission rates included discharge to a nursing home (four good quality studies)38,53,64,83; discharge to a rehabilitation center (five good quality studies)34,38,53,64,83; care from a neurologist (three good quality studies)15,93,108; urban hospital settings (four good quality studies)35,64,67,76; and patient transfers from hospitals or emergency departments (two good quality studies)37,84. Comorbidities significantly associated with elevated readmission rates included atrial fibrillation (two good quality studies)83,93, heart failure (seven good quality studies)42,76,78,80,83,93,101, and peripheral vascular disease (three good quality studies)1,76,83.

Stroke-related factors significantly associated with higher readmission rates included feeding tube presence (five good quality studies)1,34,42,64,68; intracerebral hemorrhage (five good quality studies)25,34,35,84,111; subarachnoid hemorrhage (two good quality studies)25,35; a National Institutes of Health Stroke Scale (NIHSS) score between 5 and 23 (six good quality studies)35,38,64,65,81,93; and medical complications during hospitalization (two good quality studies)59,102; a history of more than two prior hospitalizations (four good quality studies)57,68,101,105, and low functional independence (seven high quality studies)29,39,53,59,73,101,103. Conversely, effective discharge planning, supported by two good quality studies92,107, three moderate quality studies36,54,98, and two low quality studies70,97, thrombolytic therapy (two high quality studies)38,83, and post-primary care visits (two high quality studies)50,71 were significantly linked to lower readmission rates. Notably, factors such as weekend admissions58,84,85,  tobacco use23,57,83,99, chronic liver disease78,80,83, obesity57,80,83,93,142, and the COVID-19 period44,55 showed no significant impact on 30-day readmissions. The Fig. 3 below represents only those associated factors examined in at least five studies.

Fig. 3
figure 3figure 3

Number of Studies Reporting Associated Factors for 30-day Readmission.

Full size image

Meta-analysis results

We conducted a meta-analysis on factors associated with 30-day readmissions in the United States (Figs. 4 and 5). Our findings indicate that the odds of readmission increased with the presence of atrial fibrillation (OR, 1.24 [95% CI, 1.12–1.36]), cancer (OR, 1.50 [95% CI, 1.19–1.89]), chronic obstructive pulmonary disease (OR, 1.20 [95% CI, 1.15–1.26]), congestive heart failure (OR, 1.36 [95% CI, 1.22–1.52]), coronary artery disease (OR, 1.61 [95% CI, 1.24–2.10]), dementia (OR, 1.42 [95% CI, 1.12–1.79]), diabetes (OR, 1.19 [95% CI, 1.12–1.26]), peripheral vascular disease (OR, 1.28 [95% CI, 1.13–1.45]), and seizure disorder (OR, 1.11 [95% CI, 1.04–1.18]).

Conversely, certain factors were associated with decreased odds of readmission: administration of tissue plasminogen activator (tPA) (OR, 0.92 [95% CI, 0.89–0.95]), having private insurance compared to other types of insurance (OR, 0.70 [95% CI, 0.66–0.75]), being White compared to other races (OR, 0.91 [95% CI, 0.84–0.98]), and hypertension (OR, 0.92 [95% CI, 0.88–0.97]).

The following factors were not significantly associated with readmissions: being female (OR, 1.00 [95% CI, 0.98–1.03]), depression (OR, 1.06 [95% CI, 0.93–1.20]), dyslipidemia (OR, 0.77 [95% CI, 0.57–1.05]), dysphagia (OR, 1.30 [95% CI, 0.97–1.75]), prior stroke (OR, 1.45 [95% CI, 0.99–2.10]), having ischemic stroke as the stroke type (OR, 0.98 [95% CI, 0.90–1.06]), and discharge to home compared to other locations (OR, 0.75 [95% CI, 0.55–1.02]). Figures 4 and5 present the forest plot findings, with additional panels included in the supplementary information as Figure S1. Figure S2 shows the inverted funnel plots, and the results of Egger’s test are also included in the supplementary information. Publication bias was ruled out through Egger’s test and the inverted funnel plots.

Fig. 4
figure 4

Meta-Analysis Results: Factors Contributing to 30-day Readmissions in the U.S. Cardiovascular Disease Factors, Atrial Fibrillation (yes vs. no (ref)).

Full size image

Factors associated with 90-day readmissions

For 90-day readmissions, advanced age115,120,125,126, higher Charlson Comorbidity Index (CCI)118,120, lower functional independence73,115,116,122, and prior hospitalizations36,120 correlated with increased readmission rates. Conversely, effective discharge planning86,91,119 was linked to reduced readmissions during this timeframe.

Fig. 5
figure 5

Meta-Analysis Results: Factors Contributing to 30-day Readmissions in the U.S. Stroke Related Factors, Tissue plasminogen activator (yes vs. no (ref)).

Full size image

Factors associated with 1-year stroke readmissions

Older age35,72,128,131,132,133,136,139,141, lower income35,66,132, discharge to a rehabilitation center121,138, prior stroke136,138, diabetes72,129,131, high comorbidity burden (indicated by a high Charlson Comorbidity Index)35,60,132,139, prolonged length of stay35,72,129,136,141, non-ischemic stroke types35,60, lower functional independence128,129,136,138,139, and medical complications during the index hospitalization136,139 were associated with increased 1-year readmissions. Conversely, discharges to home129,134,139 and higher hospital rankings35,141 were linked to reduced 1-year readmission rates.

Factors associated with both 30-day, 90-day, and 1-year stroke readmissions

Advanced age1,25,35,48,49,57,68,72,76,78,83,84,93,94,99,106,111,115,120,125,126,128,131,132,133,136,139,141, reduced functional independence29,39,53,59,73,101,103,116,122,128,129,136,138,139, high comorbidity burden (CCI)35,60,78,90,132,139, and prior hospitalizations/ED visits36,57,60,68,101,105,120 were associated with increased readmission rates at 30-day, 90-day, and 1-year intervals.

Discussion

Our study identified a multitude of factors contributing to 30-day stroke readmissions, ranging from socio-demographic variables to clinical indicators and healthcare system-related factors. Older age, reliance on public health insurance, non-home discharge locations, neurologist care, multiple comorbidities, and low functional independence at discharge were associated with increased readmissions. Conversely, effective discharge planning, thrombolytic therapy administration, and post-primary care visits were linked to lower readmission rates, while tobacco use, obesity, weekend admissions, and the COVID-19 period showed no significant impact on 30-day readmissions. In cases of conflicting associated factors, our meta-analysis found that atrial fibrillation (OR, 1.24 [95% CI, 1.12–1.36]) and coronary artery disease (OR, 1.61; 95% CI, 1.24–2.10) significantly increased the odds of 30-day readmissions in the United States. Our results also found that dyslipidemia (OR, 0.77 [95% CI, 0.57–1.05]) and gender (OR, 1.00 [95% CI, 0.98–1.03]) were non-significant factors in the United States, consistent with the conclusions of Deng et al.7.

Moreover, we found that the average 30, 90, 365-day readmission rates were 11.0%, 19.2%, and 40.6% which are similar to other studies3,5,7. The 30-day, 90-day, and 1-year readmissions rates had wide ranges which can be explained by a variety of things particularly like different patient populations, sample sizes, and study types. Hospitals vary in the demographic characteristics of their patients, the types of health conditions they treat, and the prevalence of social risk factors such as food insecurity, housing instability, and low income. Research has shown that adjusting for these social risk factors significantly impacts hospital performance evaluations and helps reduce disparities in Medicare penalties143. Since 2019, Medicare has accounted for these factors by comparing each hospital’s readmission rate to that of peer institutions143,144. Hospitals are penalized only if their risk-adjusted readmission rate exceeds the median rate of their peer group, with adjustments made for patient age, sex, and comorbidities143,144.

Contextualizing these findings within the broader healthcare landscape unveils the pivotal role of resource allocation, policy interventions, and patient engagement strategies in mitigating readmission risks. Effective discharge planning, highlighted as a cornerstone strategy, emphasizes the importance of seamless care transitions and robust post-discharge support mechanisms. Effective discharge planning also necessitates clear communication of discharge plans, proactive scheduling of follow-up appointments, and coordination of therapies tailored to individual patient needs23,50,71,92,98. Particularly for high-risk patient cohorts, such as older individuals, those with complex medical histories, and patients treated by neurologists (likely due to the intricate nature of cases they oversee)15,42,93 personalized care coordination emerges as paramount for reducing risk of readmission.

Beyond the confines of hospital walls, the imperative for comprehensive coordinated care extends to diverse care settings, including nursing facilities, rehabilitation centers, and home health care services. Early post-discharge follow-ups, coupled with empathetic communication54 and patient education initiatives51 emerge as critical interventions in bolstering patient engagement and reducing adverse outcomes. A casual and optimistic communication style, especially regarding expected length of stay, has proven to be more effective in engaging patients and reducing discharge against medical advice incidents100. Additionally, empowering patients through lifestyle modifications114 and early rehabilitation interventions holds promise in mitigating readmission risks and fostering functional independence24. While high-intensity physical therapy is the preferred and recommended approach, low-intensity physical therapy also demonstrated a significant reduction in readmission rates for patients who received it as compared to those who did not24.

In navigating the challenge of high readmission rates, it is important to adopt a multifaceted approach. This includes enhancing discharge planning protocols, conducting rigorous patient risk stratification, evaluating the nurse working environment (as improved environments have been shown to decrease readmission rates57, and continually assessing care delivery processes. Embracing a culture of quality improvement and benchmarking against best practices can further catalyze efforts to optimize care delivery and enhance patient outcomes in stroke management.

Future directions

Current research on readmissions primarily emphasizes patient-level factors and care transitions as a key hospital-level factor. However, few studies have explored what distinguishes hospitals with lower stroke readmission rates. Future research should focus on additional hospital-level factors to identify strategies that could further reduce readmission rates, beyond improving care transitions and nursing work environments.

Limitations

While this comprehensive review aimed to identify factors influencing 30-day, 90-day, and 1-year stroke readmissions, several limitations should be noted. First, because the search was finalized in February 2024, more recent studies may not have been captured. Second, methodological heterogeneity among included studies may have introduced variability in the findings and their interpretation. Third, the overrepresentation of studies from the United States may limit the generalizability of our conclusions to a broader international context. Finally, restricting inclusion to English-language, peer-reviewed publications may have introduced publication and language bias, potentially excluding relevant studies published in other languages and favoring those with positive findings.

Conclusion

In summary, our analysis of 135 publications highlights diverse factors influencing 30-day, 90-day, and 1-year ischemic stroke readmissions, including sociodemographic characteristics, clinical conditions, and healthcare system elements. Key contributors to increased 30-day readmissions include older age, heart failure, atrial fibrillation, diabetes, depression, and higher NIHSS scores. In contrast, effective discharge planning, post-primary care visits, and thrombolytic therapy administration reduce readmission rates. These modifiable factors present opportunities to reduce readmission for ischemic stroke patients.