Stroke Care and Outcomes Among Patients Experiencing Homelessness: A Systematic Review and Meta-Analysis

Abstract

Introduction:

Homelessness is associated with an increased risk of stroke, likely secondary to uncontrolled vascular risk factors and poor access to preventive care and chronic disease management. Its specific impact on acute stroke care and outcomes remains poorly understood. In this study, we conducted a systematic review and meta-analysis to summarize the available evidence.

Methods:

We searched PubMed, EMBASE, PsycINFO, Scopus, and Web of Science from 2000 to 2024 without language or geographical restrictions. Original observational studies of adults with acute stroke experiencing homelessness were included. Studies in animals and children, and studies without clear characterization of housing status and stroke outcomes were excluded. We compared post-stroke mortality, use of cerebrovascular diagnostics and therapeutics (including intravenous thrombolysis [IVT], mechanical thrombectomy, and vessel imaging), discharge destinations, and length of hospital stay between patients experiencing homelessness (PEH) and housed patients. We conducted a meta-analysis by pooling relative risk effect size.

Results:

Titles and abstracts of 1,598 articles were screened by two independent reviewers, and 80 studies were selected for full-text review. Seven studies comprising 620,327 (86.5% male, 54.5 average age) PEH and 3,035,234 (52.7% male, 57.4 average age) housed patients were included in the systematic review. Studies included in the meta-analysis were conducted in the United States. All seven studies were rated as moderate to high quality. Pooled analysis did not reveal differences in in-hospital mortality between PEH and housed patients (risk ratio [RR]; 1.10 95% confidence interval [CI] 0.82–1.48). PEH with ischemic stroke were receiving IVT less frequently compared to housed individuals (RR 0.86, 95% CI 0.77–0.97). Two studies reported on discharge destinations, both indicating that PEH were more likely to be discharged to self-care or leave against medical advice.

Conclusions:

Patients experiencing homelessness with stroke were less likely to receive IVT and more often discharged without support. Further research is needed to identify long-term stroke outcomes and address disparities of post-stroke care to improve outcomes in PEH with stroke.

Registered:

Prospero (Registration No. CRD42024582119).

Introduction

Significant disparities in acute cerebrovascular care are largely driven by social determinants of health (SDOH).¹ SDOH are defined by the World Health Organization (WHO) as the conditions in which people are born, grow, work, live, and age, and the broader systemic forces shaping these conditions, including economic policies, social norms, and political systems.² Housing status is a key SDOH, and several studies have shown that homelessness is associated with adverse health outcomes, including premature aging, increased mortality, and both physical and cognitive disabilities.³ Mortality among patients experiencing homelessness (PEH) is estimated to be three times higher than mortality in the general population.⁴ In addition, PEH have higher reported rates of hypertension, diabetes, dyslipidemia, tobacco use, and substance use disorders, along with reduced access to routine care necessary to manage these chronic conditions.⁵ Previous studies have suggested that homelessness contributes to increased cardiovascular morbidity and mortality.^6,7,8,9 However, literature on the specific impacts of homelessness on acute stroke care and outcomes is sparse. In this systematic review and meta-analysis, we aim to systematically evaluate the available literature and synthesize existing evidence on the impact of homelessness on acute stroke care and outcomes.

Methods

Search Strategy

Literature search design was conducted by a health sciences librarian (B.O.) with experience in conducting and documenting searches for systematic reviews. The search was developed using both natural language and controlled vocabulary to capture the concepts of PEH and cerebrovascular disorders. The search was developed for MEDLINE(R) ALL (via OVID) and then translated to the following databases: Embase and Embase Classic (via Ovid), APA PsycInfo (via Ovid), Scopus, and Web of Science Core Collection (SCI-Expanded, SSCI, AHCI, CPCI-S, BKCI-S, BKCI-SSH, ESCI, CCR-Expanded, IC) search syntax of search conducted in MEDLINE(R) ALL (via OVID)). Additional data available in eMethods for search syntax. The search was conducted on September 20th, 2024, and limited to articles published on January 1, 2000, forward, to account for contemporary systems of care for acute cerebrovascular events, no other limits were applied to the search. A systematic search of the literature to identify relevant publications without language and geographical restrictions was performed by two independent reviewers (J.M. and M.S.).

Eligibility Criteria and Study Selection

We used broad inclusion criteria for our search strategy, including all types of strokes. Studies were included if the following criteria were met: (1) original observational design (cross-sectional, retrospective and prospective cohort, and case-control studies), (2) PEH hospitalized for acute stroke (ischemic and hemorrhagic), and (3) focused on stroke outcomes including mortality, utilization of mechanical thrombectomy and intravenous thrombolysis, vessel imaging, length of hospital stay, and discharge destination. Animal studies, studies in children, and studies without clear characterization of housing status and stroke outcomes were excluded. Homelessness was operationalized by identifying if each study included patients that had experienced homelessness at or before the time of the study. PEH were captured via different methods for each study (additional data available in eTable 1). Most studies relied on hospital staff to report the housing status of each patient at the time of admission or discharge. To avoid inclusion of studies with similar populations, final studies were reviewed for similar authors, patient characteristics, and results. Similar studies were included in the systematic review but excluded for meta-analysis.

Data Extraction and Quality Assessment

The following information was extracted from the articles: author names, year of publication, geographical location, number of participants, race of participants, sex, age, homelessness status, diagnostic and therapeutic procedures related to acute stroke care, discharge destination, length of hospital stay, and mortality. We were unable to assess stroke severity as this data does not exist in the literature in this population currently. In each eligible study, descriptive and quantitative data on the association between homelessness and clinical stroke outcomes were obtained. Methodological quality of included observational studies was assessed using previously published criteria in the Newcastle-Ottawa quality assessment scale (NOS) for cohort studies which evaluates studies across three domains: selection of study groups, comparability of groups, and outcome assessment, with higher scores indicating higher quality (additional data available in eTable 2).¹⁰ The primary outcome of mortality was used as a basis for answering outcome assessments questions.

Standard Protocol Approvals, Registrations, and Patient Consents

This systematic review and meta-analysis were conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses Guidelines (PRISMA). All individual studies included in the systematic review received informed consents from their participants. This systematic review protocol was registered with PROSPERO (Registration No. CRD42024582119).

Statistical Analysis

Articles were included for meta-analysis when three or more studies reported on a specific outcome. Four studies were identified and included in the meta-analysis for mortality in stroke patients. For binary outcomes, relative risk measures including risk ratios (RRs), and adjusted hazard ratios (AHR) were used to calculate pooled effect estimates for the association between homelessness and stroke outcomes. Standard errors for individual measures were derived using natural logarithm transformation. When studies reported multiple relative risk estimates, we extracted those from multivariable models with the highest level of adjustment. In studies where RRs were not directly reported, we calculated them using the prevalence of the outcome in PEH and housed groups. The standard error (SE) of the log-transformed RR was calculated using the reciprocals of the event counts minus the reciprocals of group totals, followed by taking the square root. For studies reporting differences, RR was calculated using event rates. Event rates were converted to proportions to calculate RR. The 95% CI was calculated using event rate and sample sizes. Both fixed-effects and random-effects models were used to estimate pooled associations. Initially, we visually compared effect estimates and confidence intervals across studies to assess consistency. Heterogeneity was determined by the I² statistic and corresponding p-values.¹¹ To further account for between-study variance and minimize false-positive rates, the Hartung-Knapp adjustment was applied.¹² For each study included in the quantitative synthesis, we extracted effect estimates from the multivariable model with the highest degree of covariate adjustment to reduce potential confounding. All studies contributed to pooled analyses adjusted for traditional cardiovascular risk factors—such as hypertension, diabetes, dyslipidemia, and tobacco—along with demographic characteristics. Although some residual confounding is unavoidable given the complex clustering of comorbidities among people experiencing homelessness, the use of maximally adjusted estimates allows the association between homelessness and stroke outcomes to be evaluated with the greatest methodological rigor allowed by the available data. Publication bias was evaluated using funnel plots, and asymmetry was tested using Egger’s regression (additional data available in eFigure 1 and eFigure 2).¹³ All analyses were conducted using RStudio (version 2024.12.1).

Data Availability

Data not provided in the article because of space limitations may be shared (anonymized) at the request of any qualified investigator for purposes of replicating procedures and results.

Results

Our literature search yielded 1,598 articles. After abstract and title screening and the removal of duplicates, 80 studies were identified for eligibility assessment by both reviewers. A PRISMA flow diagram of the search strategy is provided in Figure 1. A total of seven observational studies comprising 620,327 patients experiencing homelessness (86.5% male, 55.1% White) were included, with mean or median ages ranging from 39 to 65 years across studies (Table 1). Three of the seven studies were cross-sectional and four were retrospective longitudinal cohort studies. Details of individual study results are presented in Table 2. Two studies assessed mortality and vessel imaging for overall stroke, ischemic, and hemorrhagic stroke,^4,5 while other studies solely focused on overall stroke (including all stroke types), ischemic stroke alone, recurrent stroke, or incident stroke. No other studies focused on the difference in outcomes between hemorrhagic and ischemic stroke, and thus a pooled analysis on outcomes per stroke subtype was not performed. Three studies surrounding post-stroke mortality were available for inclusion in meta-analysis. Three out of four observational studies reported that PEH were less likely to receive IVT. One out of two studies reported PEH were less likely to undergo mechanical thrombectomy. Two out of two studies reported PEH were less likely to receive vessel imaging. Neither mechanical thrombectomy nor vessel imaging were able to be assessed on pooled analysis due to the limited number of studies. Two out of two observational studies reported PEH were more likely to discharge to self-care or leave against medical advice (AMA). One out of three studies reported longer length of hospital stay for PEH. Length of hospital stay could not be evaluated in the pooled analysis due to the limited number of studies reporting this outcome after excluding one study from the analysis given the restricted patient population.¹⁴ Two other studies are reported in the systematic review but are excluded from quantified analyses given their overlapping patient populations and lack of control group.^15,16

Figure 1.

PRISMA Flowchart

Table 1.

Characteristics of Studies Included in the Systematic Review

Study	No. Participants (%)	Age, Mean (SD)	Male Sex, n (%)	White Race, n (%)	Study Design	Study Population	Study Location	Stroke Type
Asaithambi et al, 202114	PEH: 514 (0.1) Housed: 364,408 (99.9)	PEH: 54.7 (10.2) PEH: 68.6 (14.7)	PEH: 409 (79.6) Housed: 183,161 (50.3)	PEH: 290 (59.3) Housed: 263,204 (72.9)	Retrospective	Patients >18 years old admitted with ischemic stroke who received IVT in the National Inpatient Sample in PEH vs housed patients	United States	Ischemic stroke
Khan et al, 20233	PEH: 3,134 (0.5) Housed: 645,810 (99.5)	Median (IQR) PEH: 39 (33–42) PEH: 38 (32–42)	PEH: 1,790 (67.1) Housed: 330,462 (51.1)	PEH: 1,344 (47.8) Housed: 279,209 (51)	Cross Sectional	Stroke related hospitalizations in young adults age 18–44 years old using National Inpatient Sample in PEH vs housed patients	United States	Ischemic stroke and hemorrhagic stroke
Lin et al, 202415	PEH: 565,608 (100.0) Housed: N/A	PEH: 51.8 (14.8)* Housed: N/A	PEH: 500,446 (88.5) Housed: N/A	PEH: 324,204 (57.3) Housed: N/A	Retrospective	Veterans >18 years old with homelessness and incident stroke in the Veteran Affairs EHR	United States	Incident stroke
Miyawaki et al, 20215	PEH: 2,123 (1.3) Housed: 164,982 (98.7)	PEH: 64.5 (14.5) Housed: 69.6 (14.5)	PEH: 1,155 (54.4) Housed: 90,589 (54.9)	PEH: 406 (19.1) Housed: 108,648 (65.9)	Cross Sectional	Outcomes of patients > 18 years old admitted with acute myocardial infarction or stroke at safety net versus non-safety net hospitals in PEH and housed patients	United States (Florida, Maryland, Massachusetts, and New York)	Ischemic stroke and hemorrhagic stroke
Montgomery et al, 202416	PEH: 15,566 (100.0) Housed: N/A	PEH: 63.3 (8.4) Housed: N/A	PEH: 14,595 (93.8) Housed: N/A	PEH: 9,163 (58.9) Housed: N/A	Retrospective	Veterans >18 years old experiencing homelessness with incident stroke who did and did not have recurrent stroke	United States	Ischemic stroke, TIA, ICH
Nanjo et al, 202017	PEH: 8,492 (20.9) Housed: 32,134 (79.1)	Homeless: 39.0 (14.6) Housed: 38.8 (14.7)	PEH: 4,975 (58.6) Housed: 18,650 (58.0)	PEH: 6,084 (81.2) Housed: 17,670 (81.3)	Retrospective	Individuals >16 years old estimated prevalence, incidence, and one-year mortality for twelve CVDs using three EHRs in PEH vs housed patients	United Kingdom	Ischemic stroke, TIA, ICH
Wadhera et al, 20204	PEH: 24,890 (1.3) Housed: 1,827,900 (98.7)	PEH: 65.1 (14.8) Housed: 72.1 (14.6)	PEH: 13,438 (54.0) Housed: 977,240 (53.5)	PEH: 4,300 (17.3) Housed: 1,235,021 (67.6)	Cross Sectional	Adult patients aged >18 years old with principal discharge diagnosis of AMI, stroke, cardiac arrest, or heart failure in PEH vs housed patients	United States (Florida, Massachusetts, New York)	Ischemic stroke and hemorrhagic stroke

Abbreviations: Patients Experiencing Homelessness (PEH), Transient Ischemic Attack (TIA), Intracerebral Hemorrhage (ICH), Standard Deviation (SD), Cardiovascular Disease (CVD), Electronic Health Record (EHR)

Table 2.

Outcomes of Studies Included in the Systematic Review

Study	Mortality	Length of Stay (LOS) / Discharge Destination (DD)	Intravenous Thrombolysis (IVT) / Mechanical Thrombectomy (MT) / Vessel Imaging (VI)
Asaithambi et al, 202114	Mortality in PEH was 3.7% vs 8.7% in housed patients (p<0.01).	LOS: Mean LOS for PEH was 12.3 days (SD 15.6) and housed patients was 6.8 days (SD 3.2) (p<0.01). DD: 40.3% PEH discharged to “self-care” vs 33.8% housed patients (p=0.003).	IVT: Patient population in the database included 514 PEH and 364,408 housed patients who received IVT (p<0.01). MT: 6.0% PEH underwent mechanical thrombectomy vs 6.4% housed patients (p=0.85).
Khan et al, 20233	Mortality was 9.0% in PEH vs 9.8% in housed patients (p>0.05). Mortality increased in PEH 3% to11% (p=0.01) vs mortality decreased in housed patients 11% to 9% over time (p<0.01).	LOS: Mean LOS for PEH was 5 days (SD 3–13) and housed patients was 5 days (SD 2–11) (p>0.05). DD: 8.5% PEH left AMA vs 1.9% housed patients (p<0.01).	IVT: 3.2% PEH received IVT vs 4.0% housed patients (p<0.05). MT: 0.6% PEH underwent mechanical thrombectomy vs 1.0% of housed patients (p<0.05).
Lin et al, 202415	Mortality in PEH with incident stroke was 17.6% vs 10.8% in those with no stroke (p=0.09). PEH with incident stroke had higher odds of mortality (OR 2.95, 95% CI 2.62–3.31, p<0.01).	N/A	N/A
Miyawaki et al, 20215	Safety Net: No difference in mortality between PEH and housed patients for overall stroke (OR 1.02, 95% CI 0.81–1.30, p=0.85), hemorrhagic stroke (OR 0.91, 95% CI 0.65–1.27, p=0.57), or ischemic stroke (OR 1.18, 95% CI 0.82–1.58, p=0.38). Non-Safety Net: No difference in mortality between PEH and housed patients for overall (OR 0.58, 95% CI 0.21–1.59, p=0.29), hemorrhagic (OR 0.37, 95% CI 0.09–1.62, p=0.19), or ischemic stroke (OR 1.07, 95% CI 0.27–4.22, p=0.93).	N/A	IVT: There was no difference between PEH and housed patients receiving IVT at safety net (OR 0.90, 95% CI 0.38–2.11, p=0.80) or non-safety net hospitals (OR 0.90, 95% CI 0.38–2.11, p=0.80). VI Safety Net: PEH were less likely to undergo vessel imaging for overall stroke (OR 0.23, 95% CI 0.23–0.60, p<0.01), hemorrhagic stroke (OR 0.37, 95% CI 0.23–0.60, p<0.001) and ischemic stroke (OR 0.12, 95% CI 0.06–0.24, p<0.01). VI Non-Safety Net: There was no difference in undergoing vessel imaging between PEH and housed adults for overall stroke (OR 0.74, 95% CI 0.34–1.58, p=0.74), hemorrhagic stroke (OR 0.43, 95% CI 1.13–1.41, p=0.43), or ischemic stroke (OR 1.06, 95% CI 0.38–2.95, p=0.91)
Montgomery et al, 202416	Recurrent stroke in PEH had increased odds of mortality compared to incident stroke (OR 1.55, 95% CI 1.38–1.73, p<0.01).	N/A	N/A
Nanjo et al, 202017	Crude one-year mortality for cerebrovascular disease was 17.8% in PEH vs 16.1% in housed patients (aHR 1.49, 95% CI 0.9–2.45, p=0.12).	N/A	N/A
Wadhera et al, 20204	Mortality in PEH was 8.9% vs 6.3% in housed patients for overall stroke (difference 2.5%, 95% CI 1.4–3.6, p<0.01). Mortality in PEH was 4.6% vs 3.0% housed patients for ischemic stroke (difference 1.6%, 95% CI 0.5–2.7, p=0.003).	LOS: No significant difference in mean LOS in PEH and housed patients for overall stroke (difference 1.69%, 95% CI 0.89–2.49).	IVT: 4.8% PEH received IVT vs 5.2% housed patients for overall stroke (difference −0.4%, 95% CI −1.0–0.3, p=0.28). 5.9% PEH received IVT vs 6.1% housed patients for ischemic stroke (difference −0.2%, 95% CI −1.1–0.7, p=0.69) VI: 2.9% PEH underwent vessel imaging vs 9.5% housed patients for overall stroke (difference −6.6%, 95% CI −7.5– −5.7, p<0.01). 2.1% PEH underwent vessel imaging vs 6.7% housed patients for ischemic stroke (difference −4.7%, 95% CI −5.3– −4.0, p<0.01). 5.1% PEH underwent vessel imaging vs 18.1% housed patients for hemorrhagic stroke (difference −13.0%, 95% CI −15.5– −10.5, p<0.01).

Abbreviations: Patients Experiencing Homelessness (PEH), Transient Ischemic Attack (TIA), Intracerebral Hemorrhage (ICH), Standard Deviation (SD), Intravenous Thrombolysis (IVT), Length of Stay (LOS), Discharge Destination (DD), Mechanical Thrombectomy (MT), Vessel Imaging (VI), Against Medical Advice (AMA), Cardiovascular Disease (CVD), Electronic Health Record (EHR), Acute Myocardial Infarction (AMI), International Classification of Disease (ICD), Veteran Affairs (VA)

^*Mean and SD are calculated from an approximation

Pooled analysis from three studies ^3,4,5 (Figure 2) reported that PEH do not have a significant difference of post-stroke in-hospital mortality compared to housed patients (RR 1.10, 95% CI 0.82–1.48). Heterogeneity was high (I²=94.7%, p < 0.0001) (Figure 2). Pooled estimates showed that PEH had lower risk of receiving intravenous thrombolysis compared to housed individuals (RR 0.86, [0.77–0.97]) (Figure 3). Heterogeneity was low (I²=12.9%, p= 0.3174) (Figure 3). Visual inspection of funnel plots showed a risk of publication bias for the associations between PEH and mortality or receipt of intravenous thrombolysis, as the studies appeared scattered around the pooled estimates (additional data available in eFigure 2).

Figure 2.

Pooled Estimate Association between Homelessness and Post Stroke Mortality

Figure 3.

Pooled Estimate Association between Homelessness Status and IV Thrombolysis Treatment

Discussion

PEH with stroke represent a critically understudied population, and further research is essential to elucidate disparities in their access to and quality of acute stroke care. This systematic review and meta-analysis demonstrate that PEH are less likely to receive intravenous thrombolysis and more likely to be discharged to self-care or leave the hospital AMA. Additionally, PEH were less likely to undergo vascular imaging during hospitalization.

Previous research has reported that PEH have three times the risk of mortality compared to the general population and increased mortality from cardiovascular diseases.^4,6,18 This population often faces a greater burden of vascular risk factors, including hypertension, diabetes, hyperlipidemia, and substance use disorders, which, in combination with adverse socioeconomic determinants of health and poor access to preventive care, predispose them to develop stroke and all-cause cardiovascular events.³ Additionally, PEH are exposed to adverse environmental conditions—including extreme heat and cold—that may increase their vulnerability to ischemic and hemorrhagic stroke.^19,20 In this study we did not observe a significantly increased acute post stroke mortality among PEH when we pooled data across three studies. However, we identified a significant degree of heterogeneity among these studies which may point to various methodologies for identifying PEH. An important limitation across all included studies is the risk of under-ascertainment and inconsistent documentation of homelessness. The definition of homelessness was not clearly defined by each study as they were retrospective in nature and used multiple large databases. In addition, most datasets relied on hospital staff coding or administrative indicators, which are known to under-capture housing instability and vary widely in sensitivity. Misclassification of PEH as housed individuals would bias results toward the null, potentially underestimating the true magnitude of disparities in stroke outcomes.

In this study, we observed that IVT was received less by PEH compared to housed patients. Multiple factors could contribute to reduced IVT use among PEH, including difficulty establishing last known well. Many of these patients may be found alone, making it difficult to identify a witness, family member, or caregiver to confirm an acute neurological change. PEH have made note of stigma due to their housing status, leading to barriers in proper health care access.²¹ Health care provider stigma about PEH can lead to these patients not receiving services out of fear of negative perception or discrimination from healthcare services.²¹ Other studies have found that the health care experiences of low-income adults may be affected by bias with one’s insurance status.²² This further shows the complex relationship between stigma, bias, and homelessness, which may act as a barrier for adequate and timely care. This is particularly important to consider when assessing acute stroke care treatments where time is of the essence. Social determinants, such as limited social support, racial and ethnic disparities, lower education, financial and food insecurity, and mental health comorbidities, may further hinder timely care and recovery. Additionally, disadvantaged populations are reported to be less likely to visit emergency departments, increasing the odds of delayed arrival.²³ Implicit bias may also contribute to underreporting of outcomes, and delayed treatment and hospital admission compared to housed patients with similar acuity.^24,25,26,27 Additionally, given the higher rate of substance use disorders in PEH, neurological symptoms may be prematurely misattributed to substance use.²⁸ Future studies are needed to assess why PEH receive less IVT and further explore possible disparities.

Quantitative analysis for discharge destination and mean length of hospital stay were not possible due to the lack of studies that assessed these outcomes. However, previous research has reported that PEH are more likely to be discharged to the streets or leave AMA after hospitalization for various reasons.²⁹ The reasons for leaving AMA are varied and include system-level, provider-level, and individual-level factors. Factors such as lack of trust in the healthcare system, inadequate post-stroke medical and social support, stigma among medical professionals, limited insurance coverage, and restricted access to rehabilitation or nursing facilities warrant further investigation. Studies in the systematic review reported mixed findings on mean length of hospital stay, but research suggests PEH have longer length of hospital stays and more discharge delays compared to housed patients.^30,31 Future studies should examine the frequency and reasons for AMA discharges and effects on mean length of hospital stay in PEH with stroke, as well as how to improve adequate discharge planning.

Across the studies included in this systematic review, 86.5% of individuals experiencing homelessness were male, compared with 52.7% in housed comparison groups. The marked overrepresentation of males in the homeless population across included studies underscores important sex-related differences in exposure patterns, comorbidity profiles, and access to care. Because very few studies reported sex-stratified outcomes, we were unable to evaluate whether the association between homelessness and stroke outcomes differs between males and females. Future studies with adequate sex-specific reporting are needed to clarify potential differential vulnerabilities and to inform tailored prevention strategies.

In this study, we applied multiple strategies to capture all available evidence regardless of language or geographic location of the studies. Heterogeneity was high in the mortality meta-analysis but low in IVT meta-analysis, but that does not eradicate the possibility of bias that could impact reported estimates. Additionally, because several studies used State Inpatient Databases (SID) or National Inpatient Sample (NIS) datasets with partially overlapping geographic and temporal coverage, we evaluated the potential for sample duplication. Studies contributing to the same meta-analytic model did not share identical state–year combinations, and datasets contributing to different models evaluated distinct outcomes, thereby limiting the possibility of double-counting within pooled analyses. In addition, although the NIS is constructed using a sample of SID hospitals, the sampling structure and lack of patient identifiers preclude direct duplication across datasets. Nonetheless, the possibility of partial overlap remains a methodological limitation.

The lack of a description on the operationalization of homelessness insinuates that standardized practice on reporting whether patients are experiencing homelessness varies. This is supported by literature findings of no universally agreed upon definition of homelessness³² and lack of consistency in collecting housing status data in routine public health studies.³³ This limitation underscores the urgent need for standardized, systematic collection of housing status in clinical and public health datasets to more accurately quantify and address cerebrovascular inequities in this population. The study populations were disproportionately male (87%) and White (55%) compared with national homelessness demographics, potentially limiting the generalizability of findings to unhoused women and racial and ethnic minority groups. In addition, given the secondary nature of the analysis, demographic variables were derived from existing records, and the gold standard of self-reported data could not be uniformly verified.

Despite these limitations, this is a robust and comprehensive literature review and meta-analysis on a limitedly studied topic. While hard outcomes such as mortality were consistently reported, key measures of stroke severity — such as the National Institutes of Health Stroke Scale (NIHSS) and the modified Rankin Scale (mRS) — were not reported. Future research should further assess these two outcomes as this is a significant gap in the literature and understanding of stroke outcomes in this population.

In this systematic review and meta-analysis, we synthesized the current literature on stroke outcomes in patients experiencing homelessness. We identified substantial disparities in post-stroke care, which underscores the need to reduce risk factors and improve care for patients experiencing homelessness. Further research is urgently needed in this underserved, growing and understudied population to identify key mediators in the disparities we identified and to inform interventions that reduce these disparities in care.