Highlights
-
Nearly half of EVT-treated stroke patients in Ontario, Canada, were transferred before treatment.
-
Transfer before EVT was associated with higher early mortality but lower long-term mortality beyond the first 6.3 months.
-
Transfer before EVT was associated with increased long-term care admission but no difference in recurrent stroke.
Introduction
Ischemic stroke is a leading cause of death and disability, with the burden of stroke projected to increase worldwide. Reference Fan, Li and Yu1 Endovascular thrombectomy (EVT) is a highly effective treatment to improve functional outcomes after disabling stroke. Reference Goyal, Menon and van Zwam2 With treatment time windows extended to 24 hours and the inclusion of patients with large infarct cores, the numbers of patients eligible for EVT continue to grow. Reference Albers, Marks and Kemp3–Reference Fayad5 Survivors of the first stroke are at a high risk of recurrent stroke with estimated rates of up to 10% at 5 years, Reference Amarenco, Lavallée and Monteiro Tavares6,Reference Vinding, Butt and Lauridsen7 but knowledge about long-term clinical outcomes and recurrent stroke rates after EVT is less well known. Understanding long-term outcomes is important for healthcare planning, particularly in optimizing long-term health service utilization in settings with limited healthcare resources and for improving patient selection to increase the proportion of patients who achieve favorable functional outcomes.
EVT is delivered within a system of care often involving inter-hospital transfers coordinated between emergency medical services, rapid access to a neurovascular stroke team and subsequent stroke unit care. Reference Heran, Lindsay and Gubitz8 Patients directly treated at a comprehensive stroke center (CSC) may have better 90-day outcomes compared to those treated after an inter-hospital transfer, Reference Romoli, Paciaroni, Tsivgoulis, Agostoni and Vidale9 but prior studies have been limited by small sample sizes and largely short-term follow-up windows and whether such transfers impact long-term outcomes is less understood.
We compared the long-term clinical outcomes between patients who were treated with EVT after an inter-hospital transfer versus those who presented directly to a CSC among community-dwelling adults hospitalized with ischemic stroke in Ontario, Canada. We hypothesized that patients who presented directly to a CSC would have a lower longer-term mortality, and through differences in system-level care and secondary prevention, have lower rates of recurrent stroke or repeat EVT compared with transferred patients.
Methods
Cohort identification
We conducted a population-based retrospective cohort study in Ontario, Canada, using linked administrative data that have been validated and used extensively for research purposes to identify the study cohort, exposure, covariates and outcomes. Reference Yu, Austin and Rashid10,Reference Taghdiri, Vyas and Kapral11 We included community-dwelling adults aged 18 to 105 years who were hospitalized through an emergency department, with a most responsible diagnosis of ischemic stroke between April 1, 2017 and March 31, 2023. The International Classification of Diseases, 10th Edition, Canada, codes I63, except I63.6, and I64 were used to identify a diagnosis of ischemic stroke. These codes have been demonstrated to have a high positive predictive value (97%) in identifying admissions for ischemic stroke. Reference Baldereschi, Balzi and Di Fabrizio12,Reference Hall, Mondor, Porter, Fang and Kapral13 We created unique episodes of care that accounted for inter-hospital transfers to avoid double-counting hospitalizations. Ontario has a universal-access health system with a single government payer. There is mandatory reporting of emergency department visits to the National Ambulatory Care Reporting System (NACRS) and for hospitalizations to be reported to the Discharge Abstract Database (DAD). 14 The province mandated the reporting of endovascular therapy (EVT) to the Canadian Institute for Health Information (CIHI) starting April 1, 2017. Thus, we have information on EVT use throughout the study period. Patients who did not have a valid health insurance number (non-Ontario residents at admission date), with invalid birth or death dates, missing information about emergency department triage acuity or those who had their stroke occur during an inpatient admission for another reason were excluded. In cases where a patient had multiple stroke hospitalizations during the study period, the first stroke episode with EVT treatment was used as the index event. The cohort creation flowchart is shown in Supplementary Figure 1. Case definitions for clinical variables used in this study, based on the administrative data, are listed in Supplementary Table 1.
Exposures and outcomes
The main exposure was whether a patient underwent an inter-hospital transfer prior to EVT, defined as any transfer from a non-CSC hospital to a CSC. A CSC was defined as a hospital with thrombolysis and EVT capability. A non-CSC hospital could be either a primary stroke center (PSC) which can administer thrombolysis only or a non-designated center (NDC) which has neither treatment available. The primary outcome was all-cause mortality at maximum follow-up. Secondary outcomes included in-hospital mortality, discharge to inpatient rehabilitation, recurrent admission for stroke, recurrent EVT and admission to long-term care. For the death and long-term care admission outcomes, the index date was the earliest admission date in the episode of care. For the outcomes of recurrent strokes and recurrent EVT, we used the discharge date from the hospital among those discharged alive as the index date.
Statistical methods
Baseline characteristics were summarized using descriptive statistics for participants with and without an inter-hospital transfer before EVT. Categorical and continuous variables were compared using standardized differences, with a difference of ≥ 0.1 indicating a potentially meaningful imbalance between groups.
To evaluate the impact of transfer status on clinical outcomes, inverse probability of treatment weighting was applied using stabilized average treatment effects weights. Balance in measured baseline variables was assessed before and after weighting using standardized differences to ensure comparability between groups. Reference Austin and Stuart15 Propensity score-weighted hazard ratios were calculated for all-cause mortality and long-term care admission. The propensity score model included the following variables: stroke severity based on the passive surveillance stroke severity (PaSSV) indicator, Reference Yu, Austin and Park16 age, sex, neighborhood income quintile, rurality, atrial fibrillation, dyslipidemia, coronary artery disease, prior stroke, peripheral vascular disease, hypertension, dementia, diabetes, thrombolysis during index admission and frailty status Reference Gilbert, Neuburger and Kraindler17 (see Supplementary Table 1 for case definitions of clinical variables used). Robust variance estimators were used. Reference Austin18 For the primary outcome of all-cause mortality, the proportional hazard assumption was violated, and we conducted a sensitivity analysis in which we allowed the hazard ratio for inter-hospital transfer to vary as a smooth function of time using methods described previously. Reference Austin, Fang and Lee19 We used cause-specific hazard models to model hazards of long-term care admission to account for the competing risk of death. For the secondary outcomes of admission to long-term care, recurrent ischemic stroke and recurrent ischemic stroke requiring EVT, we accounted for death as a competing risk and generated propensity score-weighted cumulative incidence function (CIF) plots. The proportional hazards assumption for the secondary outcomes was evaluated using the Schoenfeld residuals test from the weighted Cox model, with no significant deviation from proportionality observed for any of the secondary outcomes. Statistical significance was defined as a p-value < 0.05 and a 95% confidence interval (CI) that did not include 1. All statistical analyses were conducted using SAS v9.4 (SAS Institute Inc., Cary N.C.).
Privacy and ethics
Data were linked deterministically using unique encoded identifiers and analyzed at ICES (formerly the Institute for Clinical Evaluative Sciences). ICES is an independent, non-profit research institute whose legal status under Ontario’s health information privacy law allows it to collect and analyze healthcare and demographic data, without consent, for health system evaluation and improvement. The use of the data in this project is authorized under section 45 of Ontario’s Personal Health Information Protection Act (PHIPA) and does not require review by a Research Ethics Board.
Data availability
The dataset from this study is held securely in coded form at ICES. While data-sharing agreements prohibit ICES from making the dataset publicly available, access may be granted to those who meet criteria for confidential access after an application process.
Results
Baseline data
We identified 68523 patients hospitalized with ischemic stroke during the study period, among whom 5394 (7.9%) underwent EVT. The proportion who underwent EVT increased throughout the study period, from 5.5% of patients in 2017 to 9.7% in 2022 (Supplementary Figure 2).
Of the 5394 participants who underwent EVT, 3082 (57.1%) were in the direct-to-CSC group and 2312 (42.9%) were in the transfer group (Table 1). Patients in the direct-to-CSC group were older (mean age 72.2 vs 70.1 years, SD 0.15) and more likely to reside in an urban area (93.0% vs 82.8%, SD 0.35). The PaSSV indicator score was lower in the direct-to-CSC group (median (Q1–Q3) 6 (5–7) vs 7 (6–7), SD 0.20), indicating a more severe stroke syndrome. Patients who were transferred were less likely to have low frailty and more likely to have intermediate frailty, suggesting that they had a higher degree of frailty compared to the direct-to-CSC group. There were no significant differences in baseline history of hypertension, diabetes, atrial fibrillation, dyslipidemia or treatment with thrombolysis (SD < 0.01 for all). The median duration of hospital admission was 9 days in each group (SD 0.06).
Table 1. Baseline characteristics by inter-hospital transfer status
EVT = endovascular thrombectomy; CSC = comprehensive stroke center; PaSSV = passive surveillance stroke severity.
*The PaSSV score is an administrative data-based stroke severity measure with a theoretical range from −2.3 to 13.1 with higher PaSSV scores representing lower stroke severity. †Frailty Category was calculated from the Hospital Frailty Risk Score based on ICD-10 codes that characterized into low, intermediate or high risk of frailty.
Rates of primary and secondary outcomes by transfer status
After applying propensity score weighting using stabilized average treatment effect weights, baseline characteristics were well balanced between groups, as confirmed by the standardized differences after weighting (Table 2). The corresponding density plots (Supplementary Figure 3) demonstrated overlap in propensity score distributions between groups after stabilized average treatment effect weighting (sATEw).
Table 2. Means and prevalence before and after inverse probability of treatment weighting
EVT = endovascular thrombectomy; PaSSV = passive surveillance stroke severity score.
For the primary outcome of all-cause mortality, the all-cause unweighted rate of death per 100 person-years and 95% CI were similar between direct-to-CSC patients (16.7 [15.8, 17.6]) and transferred patients (16.4 [15.4, 17.5], p = 0.71). After weighting, mortality rates per 100 person-years remained similar (direct-to-CSC 16.2 [15.1, 17.3] and transferred patients 17.7 [16.3, 19.2], p = 0.11, see Table 3).
Table 3. Event rates per 100 person-years and adjusted hazard ratio and 95% confidence intervals (CI) comparing transferred patients to direct-to-CSC patients
CSC = comprehensive stroke center; EVT = endovascular thrombectomy.
*For all-cause mortality, the proportional hazard assumption was violated and no overall hazard ratio is reported, please see Figure 2 for the time-dependent Hazard Ratio.
The proportional hazards assumption was violated in the propensity score weighted Cox proportional hazards model, indicating that the hazard ratio changed over time (Figure 1). Thus, instead of reporting one adjusted hazard ratio, we showed the time-varying hazard ratio for inter-hospital transfer before EVT compared to direct-to-CSC patients and found that transferred patients had an increased rate of mortality in the early follow-up period (Figure 2). The 95% CI was above 1.0 until 6.3 months of follow-up time, indicating higher mortality in the transferred patients compared to the direct-to-CSC patients, and subsequently dropped below 1.0 until 34.6 months of follow-up (Figure 2).
Figure 1. Weighted Kaplan-Meier curve for all-cause death among ischemic stroke patients comparing transferred versus direct-to-CSC before EVT. EVT = endovascular thrombectomy; CSC = comprehensive stroke center.
Figure 2. Smooth hazard ratio of all-cause mortality comparing transferred versus direct-to-CSC patients in the propensity score weighted cohort modeled with restricted cubic splines. *Each knot represents equal numbers of outcomes for the intervals 5, 17, 112 and 610 days. CSC = comprehensive stroke center.
Unweighted rates per 100 person-years of admission to long-term care were similar between the direct-to-CSC patients (9.8 [9.1, 10.6]) and those transferred (10.6 [9.7, 11.6], p = 0.18). The weighted rates for the two groups were 9.7 per 100 person-years [8.9, 10.7] among direct-to-CSC patients and 11.2 [10.1, 12.5] among transferred patients, p = 0.05. There was a significant difference in the propensity score weighted cause-specific hazards model, with those who were transferred more likely to be admitted to long-term care once they survived the initial stroke (aHR 1.17, 95% CI: 1.03–1.33) (Supplementary Figure 4).
Among patients who survived the index hospitalization, recurrent ischemic stroke weighted rates per 100 person-years were similar between groups (2.5 [2.1–2.9] for direct-to-CSC vs 2.2 [1.8, 2.7] for transferred, p = 0.43, aHR 0.93, 95% CI 0.72–1.19). Similarly, recurrent EVT rates remained low across both groups, with a weighted event rate per 100 person-years of 0.5 [0.3, 0.7] for direct-to-CSC vs 0.5 [0.4, 0.8] for transferred patients, p = 0.73, aHR 1.14, 95% CI 0.67–1.93) (Supplementary Figure 4).
Discussion
We found in this population-based study that close to half of patients (42.9%) required an inter-hospital transfer before EVT. Patients brought directly to a CSC were older, had more severe strokes and were more likely to reside in urban areas. Inter-hospital transfer before EVT was associated with a time-varying association with all-cause mortality, with a higher hazard of death observed in the early post-procedural period among transferred patients. This difference attenuated after 90 days and reversed over time, favoring transferred patients by 4 years of follow-up.
Increasing EVT Rates Year on Year
EVT rates rose from 5.5% in 2017 to 9.7% in 2022, consistent with global trends. Reference Adcock, Schwamm and Smith20,21 Broader indications and improved access have driven this growth, suggesting that the number of patients requiring inter-hospital transfer before EVT will continue to increase.
Baseline Differences Between Direct to CSC and Transfer Groups
Those brought directly to a CSC were older and more likely to reside in urban areas suggesting potential disparities in access to EVT based on geography. Older patients presenting to non-CSCs may not have been given the opportunity for transfer and undergo EVT, though transferred patients were less often classified as low frailty and more often as intermediate frailty, indicating only modest differences in frailty overall and suggesting that a number of factors, including clinical judgment and logistical barriers in non-urban settings, likely influenced transfer decisions. Reference Kapral, Hall and Gozdyra22 Thrombolysis was also more common among direct-to-CSC patients (49.1% vs 36.4%), possibly due to paramedic pre-hospital transfer protocols or shorter pre-hospital times and better integration of thrombolysis and EVT pathways in urban stroke systems. Reference Holodinsky, Patel and Thornton23
Time-Varying Mortality Risk
The overall mortality rate among patients who received EVT was 16.6 per 100 person-years; however, mortality varied over time, with most deaths after EVT being front-loaded with the greatest number of deaths occurring in the early post-EVT period. This is lower than mortality rates in nationally representative stroke databases in Australia and New Zealand from the pre-EVT era, where an average mortality rate of 18.4 per 100 person-years has been observed. Reference Peng, Ngo, Hay, Alghamry, Colebourne and Ranasinghe24 Studies that have reported on EVT-treated patients have usually been limited by small sample size or shorter-term follow-up, Reference Beyeler, Weber and Kurmann25,Reference Gong, Huang and Kong26 and our study is novel in identifying that patients who were transferred and survived the initial post-stroke period had a survival benefit that emerged over the medium term follow-up, was sustained over time and was not offset by non-stroke related deaths. Infarct growth can be highly variable and is strongly associated with 90-day mortality outcomes. Reference Xu, Zhu and Guo27 Our findings may reflect unmeasured differences in patient selection, such as patients with poorer collateral supply being less likely to survive the acute post-EVT period given the delays introduced by inter-hospital transfers. The sustained decline in mortality in those surviving this acute period may reflect a predominance of “slow progressor” patients with better collateral circulation, who are more likely to tolerate ischemia and achieve improved long-term mortality outcomes. Reference Wouters, Seners and Yuen28 This has important practical implications for the organization of stroke care. While shorter transfer time is associated with favorable functional outcomes,Reference Seners, Khyheng, Labreuche, Lapergue and Pico29 direct-to-CSC transport strategies have not been observed to have the same mortality benefit in rural settings. Reference Pérez de la Ossa, Abilleira and Jovin30 Increased focus on trying to differentiate the “slow” versus “fast progressors” with imaging biomarkers Reference Mohammaden, Haussen and Pisani31 and combining this with access to neuro-protective agents as adjunctive treatments for acute ischemic stroke (32) could reduce inequities in outcomes post-EVT for patients who require transfer.
Secondary Outcomes; Long-Term Care Admission, Recurrent Stroke and EVT Rates
Patients who were transferred prior to EVT were more likely to be admitted to long-term care than direct to CSC patients once they survived the initial stroke. This suggests that although transferred patients were more likely to be alive at follow-up, they experienced worse functional recovery over time. While this may partly reflect longer time to treatment, which has been shown to correlate with poorer functional outcomes in EVT cohorts, Reference Mulder, Jansen and Goldhoorn33 other contributors may include delays in post-acute rehabilitation or differences in community-level pathways after discharge. Registry data from Get with the Guidelines – Stroke hospitals in the USA have demonstrated that transferred patients undergoing EVT are less likely to be discharged home. Reference Shah, Xian and Sheng34 Our study extends these findings by reporting on longer-term admission to long-term care which is an important patient-centered outcome. Recurrent ischemic stroke admission or stroke requiring EVT was a relatively rare event, indicating that the benefit of EVT is not eroded by a high rate of recurrent disabling stroke. The low rate of recurrent EVT, less than 1 per 100 person-years, aligns with findings from international cohorts. Reference Pirson, van Oostenbrugge and van Zwam35,Reference Lee, Kwak, Chung and Park36 There were no significant differences in recurrent ischemic stroke or EVT rates, suggesting that these long-term outcomes were similar across care pathways regardless of transfer status.
Limitations
The administrative databases we used lacked information on onset-to-puncture or door-to-puncture times, which are important determinants of outcome that are likely longer among transferred patients. Similarly, although we adjusted for thrombolysis using inverse probability of treatment weighting, we lacked detailed clinical and imaging information to further stratify outcomes by thrombolysis status. Administrative databases do not have information on diagnostic imaging, so we could not identify patients who had a large vessel occlusion and who did not undergo EVT. As this was an observational study, causal inferences cannot be made. Differences in outcomes were significant only in the weighted analyses, which, although well balanced, rely on the accuracy of the measured covariates and unmeasured confounding and residual selection bias therefore remain possible. While the generalizability of our findings may be limited in settings with markedly different healthcare funding structures, the results have important implications for systems seeking to optimize EVT delivery, particularly those adopting a hub-and-spoke model. Finally, while we observed higher mortality in transferred patients, this does not imply that inter-hospital transfers are harmful; rather, it underscores the challenges of delayed access and suggests that, in the absence of transfer, outcomes for EVT-eligible patients would likely have been even poorer.
Conclusion
In this population-based cohort study, we found that nearly half of patients undergo inter-hospital transfer before EVT, and this was associated with higher front-loaded mortality compared with direct to CSC patients. These results highlight the need to optimize pre-hospital stroke systems of care to minimize transfer-related delays and their impact on outcomes. As EVT indications continue to expand, there is a need to continue to optimize pre-hospital pathways and further explore the adjunctive potential of neuroprotective strategies for transferred patients to improve long-term outcomes.
Supplementary Material
The supplementary material for this article can be found at https://doi.org/10.1017/cjn.2026.10551.
Acknowledgements
This study was supported by ICES, which is funded by an annual grant from the Ontario Ministry of Health (MOH) and the Ministry of Long-Term Care (MLTC). This document used data adapted from the Statistics Canada Postal CodeOM Conversion File, which is based on data licensed from Canada Post Corporation, and/or data adapted from the Ontario Ministry of Health Postal Code Conversion File, which contains data copied under license from ©Canada Post Corporation and Statistics Canada. Parts of this material are based on data and/or information compiled and provided by the Ontario MOH and Canadian Institute for Health Information. The analyses, conclusions, opinions and statements expressed herein are solely those of the authors and do not reflect those of the funding or data sources; no endorsement is intended or should be inferred.
Author Contributions
Robert Murphy and Amy Yu contributed to all aspects of the development, data analysis, writing, editing and final approval of this manuscript. Zhixing Hong and Jiming Fang contributed by reviewing and analyzing the data. Peter Austin and Moira Kapral contributed to the development, writing and editing and final approval of this manuscript.
Funding Statement
AY holds a Canada Research Chair (Tier 2) in Data-driven design of stroke systems; AY has received operating grants from CIHR, Heart & Stroke, and the Brain Canada Foundation outside of the current study; and MKK holds the Sir John and Lady Eaton Chair of Medicine, University of Toronto, Toronto, Canada. Funding for this project is provided by the Ministry of Colleges and Universities’ Early Researcher Award program.
Competing Interests
The authors report no disclosures.
Open access
Target article
The Impact of Inter-Hospital Transfer before Endovascular Thrombectomy on Long-Term Outcomes after Acute Ischemic Stroke
Related commentaries (2)
Reviewer Comment on Murphy et al. “The Impact of Inter-Hospital Transfer Before Endovascular Thrombectomy on Long-Term Outcomes After Acute Ischemic Stroke”
Reviewer Comment on Murphy et al. “The Impact of Inter-Hospital Transfer Before Endovascular Thrombectomy on Long-Term Outcomes After Acute Ischemic Stroke”