Abstract
Introduction:
Changes in long-term survival after stroke at the population level are incompletely characterized, particularly using recent data and stratified by stroke type.
Patients and methods:
This nationwide, register-based cohort study used data from the Swedish Stroke Register (Riksstroke) and the Swedish Causes of Death Register. Adults (⩾18 years) hospitalized with ischemic stroke or intracerebral hemorrhage in Sweden during either 2013 or 2020 were included. Patients were followed for 3 years after the index stroke. The primary outcome was the difference in all-cause mortality within 3 years. Survival was assessed using Kaplan–Meier analyses and Cox proportional hazards regression, stratified by stroke type.
Results:
The study included 42,926 patients. Among patients with ischemic stroke, 3-year survival increased from 63.8% in 2013 to 69.3% in 2020, corresponding to an absolute difference of 5.5 percentage points, with survival differences increasing progressively during follow-up. Ischemic stroke in 2020 was associated with a lower risk of 3-year mortality compared with 2013 in both unadjusted (hazard ratio (HR), 0.83; 95% CI, 0.80–0.86) and multivariable analyses (adjusted HR, 0.92; 95% CI, 0.89–0.96). Survival improvements after ischemic stroke were most pronounced among patients aged 75 years or older. In contrast, long-term survival after intracerebral hemorrhage did not change between cohorts (3-year survival, 50.9% in both years; adjusted HR, 0.97; 95% CI, 0.89–1.05).
Conclusions:
In this nationwide cohort study, 3-year survival after ischemic stroke improved between 2013 and 2020, particularly among patients aged 75 years or older, whereas long-term survival after intracerebral hemorrhage remained unchanged. These findings suggest a contemporary population-level improvement in long-term survival after ischemic stroke over a relatively short calendar period and indicate that survival differences between cohorts increased progressively over long-term follow-up.
Keywords
Introduction
Stroke remains one of the leading causes of death and long-term disability worldwide, with both ischemic stroke and intracerebral hemorrhage contributing substantially to the global burden of disease. 1 Between the 1990s and mid-2010s, the short-term survival2–6 and long-term survival (>1 year)7,8 after ischemic stroke improved, whereas short- and long-term survival after intracerebral hemorrhage remained largely unchanged.5,6 However, it remains unclear whether the substantial advances in acute management and secondary prevention introduced since the mid-2010s have translated into sustained improvements in long-term survival at the population level, and whether short-term survival gains have increased progressively with longer follow-up.
Nationwide stroke registers with high coverage and complete mortality follow-up enable robust comparisons of survival over time under real-world conditions within the same healthcare system. Sweden (population 10.4 million), with its population-based stroke register and comprehensive national mortality data linked with personal identification numbers, offers a particularly suitable setting for this type of analysis.
The objective of this study was to examine changes in 3-year all-cause mortality after stroke between 2013 and 2020, stratified by stroke subtype. Since its establishment, the Swedish Stroke Register (Riksstroke) has performed two nationwide 3-year long-term follow-up studies (2013 and 2020 cohorts) which therefore determined the study years based on predefined nationwide follow-up initiatives rather than conventional decade-based comparisons.9,10 By focusing on long-term outcomes and comparing stroke survival with age- and sex-standardized survival in the general population, this study aimed to characterize changes in long-term survival over time in the context of contemporary stroke management.
Methods
Study design and data sources
This nationwide, register-based cohort study compared long-term survival after stroke between two calendar-year cohorts 7 years apart. The selected years reflect predefined nationwide 3-year follow-up cohorts in Riksstroke, enabling consistent long-term outcome assessment using identical methodology. Data were obtained from Riksstroke, a national quality register with prospectively collected data on acute stroke care, covering hospitalized stroke patients from all 72 acute care hospitals in Sweden. The register had a coverage of 91% and 87% for all first-ever stroke cases in 2013 and 2020, respectively. Coverage estimates were based on comparison with the Swedish National Patient Register capturing all administrative data on hospitalized stroke cases.11,12 Two cohorts were constructed, comprising patients registered between 1 January and 31 December of 2013 and 2020, using identical case definitions, inclusion and exclusion criteria, variable definitions, and follow-up procedures. Survival status and dates of death are continuously obtained through linkage between Riksstroke and the Swedish Causes of Death Register 13 providing near-complete ascertainment of mortality. Population reference survival was derived from age- and sex-specific life tables from Statistics Sweden 14 for 2013 and 2020, with additional life tables from adjacent calendar years (2019 and 2021) used to contextualize background mortality around 2020 during the COVID-19 pandemic.
Participants and follow-up
Eligible participants were adults (age ⩾ 18 years) with a primary diagnosis of ischemic stroke (ICD-10 I63) or spontaneous intracerebral hemorrhage (ICD-10 I61) recorded in the Swedish Stroke Register. Patients with subarachnoid hemorrhage (ICD-10 I62) were excluded. For individuals with more than one stroke admission during the same calendar year, only the first event was included. The primary exposure was calendar year of stroke (2013 vs 2020), and the primary outcome was all-cause mortality within 3 years after the index stroke. Follow-up extended through 31 December 2016 for the 2013 cohort and 31 December 2023 for the 2020 cohort. Patients were followed from the index stroke date recorded in the register and continued until death or censoring at 3 years after the index event. In addition, for patients with a partially missing index date (known month and year but unknown day), the date was imputed as the 15th day of the month.
Variables
Baseline variables were predefined and extracted from standardized register items. Demographic variables included age and sex. Prestroke characteristics included functional status assessed by the modified Rankin Scale (mRS), as well as living conditions and co-habitant status. Medical history and vascular risk factors comprised previous stroke, transient ischemic attack or amaurosis fugax, atrial fibrillation, hypertension, diabetes mellitus, and current smoking. Medication use prior to admission and at hospital discharge included statins and antithrombotic therapy, categorized as single-/dual antiplatelet therapy or oral anticoagulation. Clinical status at admission included level of consciousness (alert, drowsy, or comatose), stroke severity assessed by the National Institutes of Health Stroke Scale (NIHSS), admission to stroke unit and length of hospital stay. Acute treatments for ischemic stroke included intravenous thrombolysis, endovascular thrombectomy and decompressive hemicraniectomy. Detailed data on specific acute interventions for intracerebral hemorrhage, such as anticoagulation reversal and neurosurgical procedures, as well as time-to-treatment metrics for ischemic stroke, were not consistently available across both cohorts and were therefore not included in comparative analyses.
Statistical analysis
Baseline characteristics were summarized using descriptive statistics, with continuous variables presented with means with standard deviations or medians with interquartile ranges, as appropriate, and categorical variables with frequencies and percentages. Survival rates were estimated using Kaplan–Meier’s method and compared between cohorts using log-rank tests, stratified by stroke subtype. Because the primary outcome was all-cause mortality, competing risks between specific causes of death were not modeled, as death from any cause represents the clinically relevant endpoint in this context. Cox proportional hazards regression models were fitted separately for ischemic stroke and intracerebral hemorrhage to estimate hazard ratios (HRs) with 95% confidence intervals (CIs) for 3-year mortality associated with calendar year. The proportional hazards assumption was assessed using scaled Schoenfeld residuals. Several covariates showed evidence of non-proportionality, which is expected in large population-based cohorts. Cox proportional hazards models were therefore retained, and HRs were interpreted as average effects over the follow-up period.
Sequential Cox proportional hazards regression models were constructed with increasing adjustment. Covariates included in the fully adjusted model were selected a priori based on established clinical relevance and prior evidence of association with long-term mortality after stroke, and comprised age, sex, prestroke mRS, level of consciousness at admission, atrial fibrillation, diabetes mellitus, hypertension, and previous stroke. Primary Cox proportional hazards regression analyses were conducted as complete-case analyses for the included covariates. Among patients with ischemic stroke, interaction terms between calendar year and age group as well as sex were included in Cox models to assess effect modification. For intracerebral hemorrhage, we performed an additional sensitivity analysis excluding patients on oral anticoagulation at admission to determine if shifts in anticoagulant-associated intracerebral hemorrhage influenced survival trends. Population reference survival was derived from publicly available age- and sex-specific life tables from Statistics Sweden (Supplemental Appendix). 14 NIHSS was not included in the primary multivariable models because of substantial missingness, particularly among patients with intracerebral hemorrhage. Because of this high degree of missingness, analyses incorporating NIHSS would have resulted in substantial loss of observations and potential selection bias. However, a sensitivity analysis was conducted restricted to patients with available NIHSS scores. Level of consciousness at admission was therefore retained in the primary models as an available indicator of initial stroke severity. All analyses were conducted using SPSS version 30.0 (IBM Corp) and R version 4.5.2, with statistical significance defined as a two-sided p < 0.05.
Ethics
The study was approved by the Swedish Ethical Review Authority (Dnr 2024-08008-01) and conducted in accordance with the Declaration of Helsinki. In Riksstroke, patients are informed about registration and given the opportunity to opt out; therefore, individual informed consent was not required. Reporting followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines.
Results
Study population and baseline characteristics
A total of 42,926 patients were included in the study, comprising 23,560 patients in the 2013 cohort and 19,366 patients in the 2020 cohort. Of these, ischemic stroke accounted for 87.1% in 2013 and 86.8% in 2020, whereas intracerebral hemorrhage accounted for 12.9% and 13.2%, respectively. Patient selection for the study population is summarized in Supplement Figure S1.
Compared with 2013, patients in the ischemic stroke 2020 cohort were more often independent before stroke and were more frequently living at home without home care services. Cardiovascular risk factor profiles were largely similar between cohorts, whereas endovascular thrombectomy was more frequently performed in 2020. Baseline characteristics stratified by stroke subtype and cohort year are presented in Table 1.
Demographics on the 2013 and 2020 cohorts.
Data presented as numbers (%), mean (SD) or median (IQR). Abbreviations: DOAC = Direct Oral Anticoagulant; DAPT = Dual Antiplatelet Therapy; IS = Ischemic Stroke; ICH = Intracerebral Hemorrhage; mRS = Modified Rankin Scale; NIHSS = National Institutes of Health Stroke Scale; OAC = Oral Anticoagulant; SAPT = Single Antiplatelet Therapy; TIA = Transient Ischemic Attack; VKA = Vitamin K Antagonist.
NIHSS values are reported for patients with available data only; a substantial proportion of observations were missing, which should be considered when interpreting median values (see text).
Only IS.
Patterns of secondary preventive therapy in survivors at discharge differed between cohorts (Supplement Table S1). Among patients with ischemic stroke, the use of dual antiplatelet therapy increased markedly in 2020 compared with 2013 (22.6% vs 3.9%), while the use of oral anticoagulation increased in parallel with a pronounced shift from vitamin K antagonists to direct oral anticoagulants. Similar but less-pronounced trends were observed among patients with intracerebral hemorrhage at admission, alongside higher stroke unit admission rates and shorter hospital stay for both groups in 2020 (Supplement Table S2).
Most variables had ⩽1.4% missingness, except prestroke functional status (3.6%) and smoking (11.5%). Stroke severity assessed by NIHSS was frequently missing, particularly among patients with intracerebral hemorrhage, with missing values in 66.9% of cases in 2013 and 56.0% in 2020, and in 49.0% and 31.8% of ischemic stroke cases, respectively.
Survival
Patients with ischemic stroke in the 2020 cohort had higher long-term survival compared with the 2013 cohort (Figure 1(a)). Three-year survival increased from 63.8% (95% CI, 63.2–64.5) in 2013 to 69.3% (95% CI, 68.6–70.0) in 2020, corresponding to an absolute difference of 5.5 percentage points (log-rank p < 0.001). In contrast, survival curves for intracerebral hemorrhage largely overlapped, with identical 3-year survival of 50.9% (95% CI, 49.2–52.8) in 2013 and 50.9% (95% CI, 49.0–52.9) in 2020 in both cohorts (log-rank p = 0.88) (Figure 1(b)). The proportion of patients with partially imputed dates was low (n = 959), and sensitivity analyses using alternative imputed dates yielded similar results (Supplement Table S3).
Kaplan–Meier estimates of 3-year survival after stroke. Kaplan-Meier curves show survival probability over time since index stroke for patients with ischemic stroke (Panel a) and intracerebral hemorrhage (Panel b) in the 2013 and 2020 cohorts. Numbers at risk are shown below the x-axis. Differences between cohorts were assessed using log-rank tests.
Among patients with ischemic stroke, absolute differences in survival between cohorts were modest early after the index event but increased progressively with longer follow-up. Survival improved from 89.0% to 90.3% at 30 days (absolute difference, 1.3 percentage points), from 84.7% to 86.1% at 90 days (1.4 percentage points), and from 77.1% to 79.9% at 1 year (2.8 percentage points), before reaching the largest difference at 3 years (63.8% to 69.3%; 5.5 percentage points). Kaplan-Meier estimates with 95% confidence intervals at all time points are provided in Supplementary Table S4. No corresponding temporal increase in survival differences was observed for intracerebral hemorrhage at any time point (Supplemental Figure S2). Survival after ischemic stroke was higher in the 2020 cohort compared with 2013 across subgroups defined by first-ever versus recurrent stroke and by sex, with similar relative improvements observed among men and women (Supplement Figure S3).
Age-stratified Kaplan-Meier analyses further demonstrated that the observed survival improvement after ischemic stroke was most pronounced among older patients (Figure 2(c) and (d), Supplement Figure S4). Significant separation of survival curves between the 2013 and 2020 cohorts was observed among patients aged 75–84 years and those aged 85 years or older, whereas survival differences were small and not statistically significant among patients younger than 75 years (Figure 2(a) and (b)). In contrast, no differences in survival between cohorts were observed across any age group among patients with intracerebral hemorrhage, with largely overlapping survival curves in all age strata (Figure 3(a) to (d), Supplement Figure S5). Kaplan-Meier curves for the entire stroke population are presented in Supplement Figure S6. Additional exploratory analyses further subdividing patients younger than 65 years into < 55 and 55–64 years were performed, and no differences were observed between 2020 and 2013 (Supplement Figure S7).
3-year survival after ischemic stroke by age group in 2013 and 2020. Kaplan-Meier curves showing all-cause survival up to 3 years after ischemic stroke, stratified by age group. Panel a: patients aged ⩽ 64 years, Panel b: aged 65–74 years, Panel c: aged 75–84 years, and Panel d: aged ⩾ 85 years. Survival differences between cohorts were not statistically significant in younger age groups (Panel a: p = 0.09; Panel b: p = 0.41), whereas significantly higher survival was observed in the 2020 cohort among patients aged 75–84 years and ⩾85 years (Panels c and d; both p < 0.001). p-values were derived from log-rank tests.
Three-year survival after intracerebral hemorrhage by age group in 2013 and 2020. Kaplan-Meier curves depicting all-cause survival up to 3 years following intracerebral hemorrhage, stratified by age group. Panel a: patients aged ⩽ 64 years, Panel b: aged 65–74 years, Panel c: aged 75–84 years, and Panel d: aged ⩾ 85 years. No statistically significant survival differences were observed between cohorts in any age group.
Expected 3-year survival in an age- and sex-standardized general population corresponding to the study population was 84.2% in 2013 and 85.6% in 2020. The observed 5.5-percentage point survival gain exceeded the 1.4-percentage point change in expected population survival between 2013 and 2020. Expected survival in 2020 was slightly lower than in adjacent calendar years, with estimates of 86.7% in 2019 and 86.8% in 2021. When stratified by age and year, expected 3-year survival among individuals younger than 75 years was 96.1% in 2013 and 96.1% in 2020. Among individuals aged 75 years or older, expected survival was 79.5% in 2013 and 81.0% in 2020, with an absolute difference of 1.5-percentage points. 14
Cox proportional hazards regression
Ischemic stroke
In unadjusted Cox regression analyses among patients with ischemic stroke, stroke occurring in 2020 was associated with a lower risk of 3-year mortality compared with 2013 (HR, 0.83; 95% CI, 0.80–0.86). This association remained after adjustment for age and sex and was attenuated but persisted after further adjustment for prestroke functional status, level of consciousness at admission, atrial fibrillation, diabetes, hypertension, and prior stroke, with a fully adjusted HR of 0.92 (95% CI, 0.89–0.96) (Table 2). Results from the fully adjusted model are shown in Supplement Figure S8. In a sensitivity analysis restricted to patients with available NIHSS data (n = 20,677), additional adjustment for stroke severity did not materially alter the results. The association between stroke year and mortality remained significant (HR 0.90, 95% CI 0.86–0.95). Higher NIHSS was independently associated with increased mortality (HR per point 1.06, 95% CI 1.06–1.06). Full results are presented in Supplementary Table S5. Among patients with ischemic stroke an interaction was observed when age was analyzed in predefined categories (p for interaction = 0.036). The survival benefit in 2020 was most evident among patients aged 75–84 years. No interaction by sex was observed (p for interaction = 0.85), presented in Supplementary Table S6.
Hazard ratios for 3-year mortality after ischemic stroke (IS) and intracerebral hemorrhage (ICH) by stroke year (2020 vs 2013).
Intracerebral hemorrhage
Among patients with intracerebral hemorrhage, stroke year was not associated with 3-year mortality in unadjusted analyses (HR, 1.01; 95% CI, 0.93–1.08) or after sequential multivariable adjustment. In the fully adjusted model, the HR for 2020 versus 2013 was 0.97 (95% CI, 0.89–1.05; Table 2). The fully adjusted model is presented in Supplement Figure S9. In a sensitivity analysis excluding patients receiving oral anticoagulants at admission, results were consistent with the main analysis (HR, 0.95; 95% CI, 0.86–1.05; Supplementary Table S7).
Discussion
In this nationwide cohort study, 3-year survival after ischemic stroke improved substantially between 2013 and 2020, with an absolute increase of 5.5 percentage points. In contrast, long-term survival after intracerebral hemorrhage did not improve, providing an internal negative control and suggesting a stroke subtype-specific improvement rather than a general shift in case mix or background mortality. Unlike prior registry studies demonstrating gradual survival gains over decades, this study suggests a contemporary improvement in 3-year survival over only 7 years.
Our findings align with prior population-based studies demonstrating gradual improvements in short- and long-term survival after ischemic stroke over extended periods, often spanning 15 to 20 years,7,8,15 with little or no corresponding improvement after intracerebral hemorrhage.2–4,6 Contemporary national data support the external validity of our findings; for example, a nationwide French study reported a 1-year survival of 79.2% after ischemic stroke in 2022, closely aligning with the 79.9% observed in our 2020 cohort. 16 Population-based studies from the early 2000s reported substantially lower 3-year survival rates of approximately 61%, underscoring the magnitude of survival gains over time. 17 To our knowledge, no prior nationwide, population-based study has reported a comparable improvement in long-term survival after ischemic stroke over such a short interval in the contemporary treatment era. The observed 5.5-percentage point survival gain exceeded the 1.4-percentage point change in expected population survival between 2013 and 2020. 14 Because expected survival in 2020 was modestly lower than in adjacent years, likely due to the Coronavirus disease 2019 (COVID-19) pandemic, the true improvement may be slightly underestimated. Age-stratified analyses further showed that expected survival in the general population among individuals aged 75 years or older increased only modestly, indicating that changes in background mortality in this age group were small compared with the substantially larger survival gains observed after ischemic stroke.
Several developments in ischemic stroke care during the study period may plausibly be associated with the observed improvement in long-term survival. In addition to broader implementation of endovascular thrombectomy and increased admission to dedicated stroke units, pharmacological secondary preventive management after the acute event evolved substantially. However, endovascular thrombectomy was performed in only a minority of patients during the study period and therefore is unlikely to account for a large proportion of the observed population-level improvement in survival. Consistent with prior observations, absolute survival differences between cohorts were modest early after ischemic stroke. 18 In our study, these differences widened over time, suggesting that factors influencing outcomes during the subacute and chronic phases become more apparent with extended follow-up (Figure 1(a)). In this context, higher use of dual antiplatelet therapy, direct oral anticoagulants, and antihypertensive and lipid-lowering treatments at discharge in the later cohort is consistent with more intensive secondary prevention (Supplement Table S1). Importantly, these improvements likely reflect cumulative, system-level changes across the entire stroke care continuum rather than the effect of any single intervention.
Notably, Kaplan-Meier analyses suggested that the survival gains were driven primarily by patients aged 75 years or older, indicating that recent improvements have primarily benefited older individuals. This pattern was supported by formal interaction analyses, which indicated that the association between calendar year and mortality differed across age groups, with the most pronounced relative improvement observed among patients aged 75–84 years. This shift likely reflects increased adoption of evidence-based acute and preventive therapies in a population historically underrepresented in randomized trials. Early thrombolysis trials included few older patients, contributing to initial uncertainty regarding safety and benefit in this group. 19 Subsequent trials and meta-analyses, however, have supported the use of thrombolysis in patients older than 80 years,20,21 and even among patients older than 90 years. 22 During the study period, endovascular thrombectomy increasingly became part of routine acute stroke care, and emerging evidence and evolving clinical experience supporting its use in older patients may have influenced treatment decisions toward broader inclusion of elderly individuals,23,24 representing a further potential contributor to improved outcomes in this age group. Changes in secondary prevention for atrial fibrillation further illustrate the broader adoption of evidence-based therapies during the study period. Practice transitioned from vitamin K antagonists to direct oral anticoagulants (Table 1). Although early evidence in elderly patients were limited in 2013, 25 later randomized and observational studies demonstrated favorable safety and effectiveness in elderly and high-risk populations.26,27 While causality cannot be inferred, expanded use of guideline-supported therapies may have contributed to the observed age-specific survival gains. National data further support this interpretation, showing that ischemic stroke incidence in Sweden has declined most sharply in individuals aged ⩾ 75 years, with recurrent events decreasing at nearly twice the rate of first-ever events. 28 Taken together, these findings are consistent with improvements in ischemic stroke care during the study period, with the greatest gains observed among older patients, although causal inferences cannot be made from the present observational data.
In contrast to ischemic stroke, long-term survival after intracerebral hemorrhage did not improve between 2013 and 2020, consistent with prior reports.5,6 This finding underscores the persistent lack of effective disease-modifying therapies for intracerebral hemorrhage and highlights a widening gap in long-term outcomes between ischemic stroke and intracerebral hemorrhage. Recent calls for action have emphasized the need to move beyond isolated interventions toward structured care bundle approaches and rigorous evaluation of multimodal treatment strategies for intracerebral hemorrhage.29–31
The strengths of this study include its nationwide design, near-complete coverage of hospitalized stroke cases in Sweden, and virtually complete ascertainment of mortality through linkage with the national Causes of Death Register. The use of two cohorts derived from the same registry, applying identical inclusion criteria and follow-up procedures, enables a robust head-to-head comparison of long-term outcomes over time.
Limitations
Several limitations should be acknowledged. First, the observational design precludes causal inference, and residual confounding cannot be excluded. Second, stroke severity assessed by NIHSS was incompletely recorded, particularly among patients with intracerebral hemorrhage, and was therefore not included in the primary multivariable analyses to avoid biased analyses. Consequently, changes in unmeasured stroke severity or case mix between calendar years, including a higher proportion of milder ischemic strokes in the later cohort, represent a plausible alternative explanation for part of the observed survival improvement. Importantly, in a sensitivity analysis restricted to patients with available NIHSS data, adjustment for stroke severity did not materially change the association between calendar year and mortality. Although this analysis was limited by substantial missingness and potential selection bias. Third, information on post-acute care, rehabilitation intensity, and long-term secondary prevention beyond discharge was not available, limiting the ability to directly assess mechanisms underlying the observed survival trends. Fourth, the COVID-19 pandemic may have affected healthcare-seeking behavior, emergency transport, and hospitalization patterns during 2020, potentially influencing which patients were admitted and registered. If patients with milder symptoms were less likely to present to hospital during parts of the pandemic, the observed survival improvement may be underestimated. Conversely, excess mortality related to COVID-19 during follow-up could have attenuated survival in the 2020 cohort. However, this effect is likely small as the coverage of the register was 87% in both 2020 and 2021, representing a 2 percentage point decrease from 2019 when compared with official administrative data. 12 Fifth, we also did not examine cause-specific mortality, and the relative contribution of recurrent stroke and other causes of death could therefore not be determined. All-cause mortality was used as a robust and clinically relevant outcome that avoids potential misclassification inherent in cause-of-death coding. The relatively short observation period should also be considered, as longer follow-up may provide additional insights into survival differences over time, particularly in the presence of competing risks in this predominantly older population, where non-stroke-related mortality may attenuate observed differences. Temporal changes in population composition may also have influenced the observed trends. Sweden underwent demographic shifts during the study period, including increased immigration and a lower proportion of cohabiting individuals, which may reflect differences in social support and frailty not fully captured by prestroke mRS. In addition, we lacked information on ethnicity, migration background, and individual-level socioeconomic factors, which may influence stroke risk, access to care, and outcomes. Finally, although register coverage remained high and stable over time, the observed decrease in registered stroke cases may reflect both true changes in incidence and differences in case ascertainment, which should be considered when interpreting temporal trends. Given the organization of stroke care and universal health care access in Sweden, the generalizability of these findings to health systems with different structures or resource availability may be limited.
Future directions
This study points to several priorities for future research. Whether similar temporal patterns are observed in low- and middle-income countries, where stroke systems of care and access to secondary prevention differ substantially, remains uncertain and warrants further investigation. Continued monitoring of stroke epidemiology using national stroke registries will be essential to detect changes in long-term outcomes over time. Data from other countries and regions are needed to assess the generalizability of these trends across different health care systems. Further analyses should focus on underlying factors driving changes in survival, including secondary prevention, post-acute care, migration patterns and socioeconomic conditions. 32 In addition, it will be important to determine whether observed gains in long-term survival are accompanied by corresponding improvements in functional outcomes and disability after stroke. Together, these considerations underscore that prognosis after stroke is a moving target, shaped by ongoing changes in treatment and population characteristics, and that sustained, population-based follow-up will be crucial to inform future stroke care.
Conclusion
In this nationwide cohort study, 3-year survival after ischemic stroke in Sweden improved between 2013 and 2020 over a relatively short calendar period, whereas long-term survival after intracerebral hemorrhage remained unchanged. These findings suggest a contemporary improvement in population-level long-term survival after ischemic stroke, particularly among patients aged 75 years or older, while underscoring the persistent lack of progress in survival outcomes after intracerebral hemorrhage. Continued monitoring of stroke outcomes at the population level will be important to determine whether these trends persist as stroke care continues to evolve.
Supplemental Material
sj-docx-1-wso-10.1177_17474930261448295 – Supplemental material for Changes in long-term survival after stroke between 2013 and 2020: A Nationwide cohort study from the Swedish Stroke Register
Supplemental material, sj-docx-1-wso-10.1177_17474930261448295 for Changes in long-term survival after stroke between 2013 and 2020: A Nationwide cohort study from the Swedish Stroke Register by Marcus Johnsson, Stefan Sennfält, Trine Apostolaki-Hansson, Mats Pihlsgård, Johan Wassélius, Bo Norrving, Teresa Ullberg and Björn M Hansen in International Journal of Stroke
Footnotes
Contributors
All authors contributed to the conception and design of the study. Data curation and statistical analyses were conducted by M.J. and M.P. All authors participated in the interpretation of the results. Study supervision was provided by T.A-H., T.U., J.W., B.N., and B.H. The first draft of the manuscript was prepared by M.J., and all authors critically revised the manuscript for important intellectual content. All authors approved the final version of the manuscript.
Declaration of conflicting interests
The authors declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: B.N. has received honoraria for Data Safety and Monitoring Board work for the HOVID trial Symbec-Orion, Moleac and Merck. T.U. received speaker’s honoraria from Siemens Healthineers. JW is a founder and shareholder of Uman Sense AB and has received speaker honoraria from Simens Healthineers, BALT group and Medtronic Inc. The other authors declare no competing interests.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was funded by grants from Regional ALF grants (JW, TU, BMH and TAH), the Crafoord Foundation (JW), VINNOVA (JW), SUS Stiftelser & Fonder (JW and BMH), and by the Bundy Academy (TU). None of the funding bodies had any involvement in the planning of methodology, data retrieval or analysis in the study.
Use of artificial intelligence tools
During the preparation of this manuscript, the authors used OpenAI’s ChatGPT-5 (last accessed 8 March 2026) exclusively for language refinement after the initial draft had been completed. All text was critically reviewed and approved by the authors, who take full responsibility for the accuracy and integrity of the manuscript.
ORCID iDs
Data availability
Access to aggregated data may be granted upon reasonable request and in accordance with applicable ethical and regulatory approvals.
Supplemental material
Supplemental material for this article is available online.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
