Impact of emergency medical services on the prognosis in acute ischemic stroke patients receiving reperfusion therapy or usual treatment: a prospective cohort study Authors: Junmin Tao, Xin Pan, Furong Li, Xiaowen Sui, Lijuan Qu, Yan Gao, Jun Li, Dandan Shi, Dong Chen, Hongling Zhao Published: 2026 ABSTRACT BackgroundThe impact of Emergency Medical Service (EMS) on the prognosis of patients with acute ischemic stroke (AIS) remains controversial.MethodsThis prospective cohort study enrolled AIS patients admitted to the emergency department of the Central Hospital of Dalian University of Technology between August 1, 2024, and July 31, 2025. Inverse probability weighting (IPW) adjusted for baseline characteristics, and modified Poisson regression determined the association between EMS transport and the primary outcome, defined as the modified Rankin Scale score of 0–1 at the 3-month follow-up. Post-hoc analyses compared differences in key clinical indicators between groups. Subgroup analyses for primary outcome explored variations among populations with distinct characteristics.ResultsAmong the 773 patients included, 236 were in the EMS-transported group and 537 in the self-transported group. There was no significant difference between the two groups regarding the possibility of good functional recovery [weighted risk ratio (RR) = 1.37, 95% confidence intervals (CI) = 0.98–1.93]. Post hoc analysis indicated shorter onset-to-door time (ODT) but longer door-to-needle time (DNT) in the EMS-transported group. Subgroup analysis revealed the lower possibility of good functional recovery in patients with a history of coronary heart disease (weighted RR = 4.74, 95% CI = 1.83–12.32; P for interaction = 0.007) and non-smokers (weighted RR = 1.63, 95% CI = 1.11–2.40; P for interaction = 0.039) via EMS transport.ConclusionsBased on IPW, EMS-transported group showed shorter ODT and longer DNT, with no significant functional improvement. Targeted pre-hospital education and risk-stratified response mechanism within hospitals may be warranted.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12873-026-01621-9. Introduction Stroke is a leading cause of death both in China and globally, imposing a heavy burden on families and society [1–3]. As the most common type of stroke, acute ischemic stroke (AIS) is a heterogeneous condition comprising various subtypes, however, the “time-sensitive” nature of its early treatment is a universal principle [4]. Therefore, every step from symptom onset to effective intervention is critical [5]. The mode of hospital admission, as the starting point of the AIS treatment chain, at present, patients are mainly admitted via two routes: Emergency Medical Services (EMS) (activated by calling 120) or self-transport. Different modes of admission may be closely associated with patient clinical outcomes by influencing pre-hospital delays, triage efficiency, and the timeliness of reperfusion therapy [6, 7]. Current medical systems emphasize the importance of EMS for patient transport [8]. The primary advantages of EMS lie in faster triage and higher reperfusion rates, its use in patients with AIS can significantly reduce pre-hospital delays and improve advance notification to receiving hospitals, thereby facilitating the timely initiation of reperfusion therapy [9, 10]. An observational study of 3555 patients showed that, compared with patients who arrived at the hospital on their own, those transported by EMS were more likely to receive timely evaluation [11]. Research on EMS and stroke care has primarily focused on the pre-hospital phase and key time points during the treatment process. Existing studies confirm that EMS transport can significantly reduce the onset-to-door time (ODT) and the onset-to-needle time (ONT), but not the door-to-needle time (DNT) [12, 13]. In addition, Relevant studies have shown that patients transported by EMS tend to present with more severe neurological deficits and a higher prevalence of cardiovascular diseases including atrial fibrillation (AF), coronary heart disease (CHD), and heart failure, which are often associated with worse clinical outcomes [13]. The evidence regarding the direct impact of EMS on functional outcomes remains inconsistent, and most studies are limited by inadequate adjustment for baseline confounding factors. It remains unclear whether the route of admission affects long-term functional recovery. To address this gap, it is necessary to further adjust for key confounding factors in prospective cohorts, systematically evaluate the independent impact of mode of admission on the prognosis of patients with AIS, and validate its clinical significance. Therefore, this study aimed to explore the effects of different admission routes (EMS transported vs. self-transported) on clinical outcomes in patients with AIS, with a focus on analyzing potential heterogeneity in this association across key subgroups. The findings may help clarify the comprehensive role of admission pathways throughout stroke treatment and provide a more targeted evidence-based foundation for optimizing stroke care systems. Methods The single-center, prospective cohort study enrolled patients clinically diagnosed with AIS in the emergency department of the Central Hospital of Dalian University of Technology between August 1, 2024 and July 31, 2025. All patients were routed to the hospital via the “Stroke Emergency Map” platform. Patients were included based on medical records and imaging examinations confirming an AIS diagnosis. Exclusion criteria: wake-up stroke or unclear onset time, transferred from another hospital, incomplete clinical information, and loss to follow-up at 3 months post-discharge. Data were collected using the electronic medical record system. The 3-month follow-up was conducted via telephone or face-to-face interview. Clinical outcomes The primary outcome was the proportion of patients with no disability, defined as a modified Rankin Scale (mRS) score of 0–1 at the 3-month follow-up. Secondary outcomes included: the proportions of patients with mRS scores of 0–2 and 0–3 at discharge and at the 3-month follow-up. For each outcome, scores within the defined range were considered favorable outcomes, and scores above that range were considered unfavorable. These definitions and thresholds were selected based on outcome measures from previous studies [14, 15]. Safety outcomes were all-cause mortality at discharge and at the 3-month follow-up. Statistic analysis Inverse probability weighting (IPW) was applied to balance baseline characteristics and create a comparable dataset. Propensity scores were estimated using 20 covariates, including age, sex, marital status, education level, family history, type of medical insurance, smoking, drinking, risk factors for diseases including hypertension, diabetes, dyslipidemia, AF, CHD and prior stroke, medication use for hypertension, diabetes, and dyslipidemia, body mass index (BMI), baseline National Institutes of Health Stroke Scale (NIHSS) score and pre-onset mRS score. The inverse probability weight for each individual was then calculated based on the propensity score. EMS-transported group were assigned a weight of 1/ propensity score, and the self-transported group were assigned a weight of 1 / (1 - propensity score). To address extreme weights, truncation was performed at the 99th percentile of the weight distribution. Balance between groups after weighting was assessed using standardized mean difference (SMD). Modified Poisson regression analyses were conducted, with results presented as weighted risk ratios (RR) and 95% confidence intervals (CI). To verify the robustness of the findings, a sensitivity analysis was conducted using propensity score matching (PSM). A post hoc analysis was carried out on the weighted data to compare the rates of intravenous thrombolysis (IVT), endovascular treatment (EVT), ODT, DNT, ONT and door-to-puncture time (DPT) among patients with different admission routes. Exploratory subgroup analyses were performed to investigate the association between admission routes and primary clinical outcome across different patient characteristics. A two-sided P value < 0.05 was considered statistically significant. All statistical analyses were performed using R software version 4.4.3. Results A total of 972 patients were screened, of whom 128 patients were excluded for not meeting the inclusion criteria (43 wake-up patients, 80 transfer patients, and 5 incomplete baseline clinical information), and 71 were further excluded for loss to follow-up. Ultimately, 773 patients were included in the study. Among them, 236 were in the EMS-transported group and 537 were in the self-transported group (Fig. 1). The mean age of the patients was 68.8 ± 11.64 years, and 36.6% were female. Table 1 presents the baseline characteristics of patients based on their mode of admission. Patients transported by EMS were significantly older than those who self-transported. They also had a higher prevalence of atrial fibrillation and higher scores on both the baseline NIHSS and pre-onset mRS. Conversely, a higher proportion of patients in the self-transported group reported smoking and drinking. After IPW, the covariates were well balanced between the two groups, with all SMDs below 0.1 (Table 1). The distribution of the weights before and after truncation is shown in Table S1. After truncation at the 99th percentile, the IPW range narrowed from 0.31 to 42.99 to 0.31–3.20, confirming that extreme weights were effectively mitigated. Fig. 1Study flow chart Study flow chart Table 1Baseline characteristics of included patientsCharacteristicAll patientsInverse probability weighting patientsSelf-transported (n = 537)EMS-transported(n = 236)P-valueaSelf-transported(n = 772.74)EMS-transported(n = 751.40)SMDbAge (years, mean ± SD)67.23 ± 11.5772.35 ± 11.01< 0.00168.90 ± 11.6468.91 ± 12.110.001Female [n (%)]185 (34)98 (42)0.072291.7 (37.7)271.6 (36.1)0.033 Marital status 0.0740.095 Single6 (1)1 (1)6.7 (0.9)2.5 (0.3) Married437 (81)177 (75)607.0 (78.6)580.6 (77.3) Divorced15 (3)6 (3)22.4 (2.9)17.9 (2.4) Widowed79 (15)52 (22)136.6 (17.7)150.5 (20.0) Education level 0.6630.063 No education15 (3)11 (5)21.2 (2.7)18.8 (2.5) Primary school150 (28)61 (26)223.5 (28.9)208.5 (27.7) Junior high school217 (40)94 (40)296.8 (38.4)282.5 (37.6) High school/vacational school104 (19)44 (19)154.9 (20.1)154.0 (20.5) College degree or above51 (9)26 (11)76.3 (9.9)87.7 (11.7)Family history [n (%)]25 (5)20 (8)0.05543.4 (5.6)49.0 (6.5)0.038 Medical insursance 0.3390.088 Employee Insurance425 (80)198 (84)627.1 (81.2)629.8 (83.8) Resident Insurance100 (19)36 (15)131.8 (17.1)112.8 (15.0) Commercial Insurance1 (1)0 (0)1.0 (0.1)0.0 (0.0) Self-funded11 (2)2 (1)12.8 (1.7)8.8 (1.2)Smoke [n (%)]197 (37)51 (22)< 0.001256.0 (33.1)246.0 (32.7)0.008Drink [n (%)]165 (31)52 (22)0.017212.1 (27.5)220.7 (29.4)0.042Hypertension [n (%)]383 (71)172 (73)0.721551.3 (71.3)532.0 (70.8)0.012Diabetes [n (%)]170 (32)69 (29)0.558248.4 (32.1)227.6 (30.3)0.040Dyslipidemia [n (%)]135 (25)53 (22)0.478190.7 (24.7)182.9 (24.3)0.008AF [n (%)]62 (12)69 (29)< 0.001117.8 (15.2)131.4 (17.5)0.061CHD [n (%)]62 (12)40 (17)0.05488.2 (11.4)92.4 (12.3)0.027Stroke [n (%)]156 (29)59 (25)0.285220.6 (28.5)196.8 (26.2)0.053Hypertension medications [n (%)]245 (46)114 (48)0.542361.0 (46.7)340.7 (45.3)0.028Diabetes medications [n (%)]130 (24)55 (23)0.857194.3 (25.1)176.2 (23.4)0.040Dyslipidemia medications [n (%)]51 (9)19 (8)0.61172.0 (9.3)71.9 (9.6)0.000BMI (mean ± SD)24.98 ± 3.7725.15 ± 3.510.55125.03 ± 3.8225.15 ± 3.470.032Baseline NIHSS score (mean ± SD)4.95 ± 4.3010.07 ± 7.66< 0.0016.84 ± 7.216.63 ± 6.010.009Pre-onset mRS score (mean ± SD)0.20 ± 0.550.34 ± 0.830.0080.25 ± 0.640.26 ± 0 0.670.008aContinuous variables for Mann-Whitney U test, categorical variables for chi-square test; bContinuous variables for weighted t test, and categorical variables for weighted chi-square test. EMS, Emergency Medical Services; SMD, standardized mean difference; AF, atrial fibrillation; CHD, coronary heart disease; BMI, body mass index; NIHSS, National Institute of Health Stroke Scale; mRS, modified Rankin Scale Table 1Baseline characteristics of included patientsCharacteristicAll patientsInverse probability weighting patientsSelf-transported (n = 537)EMS-transported(n = 236)P-valueaSelf-transported(n = 772.74)EMS-transported(n = 751.40)SMDbAge (years, mean ± SD)67.23 ± 11.5772.35 ± 11.01< 0.00168.90 ± 11.6468.91 ± 12.110.001Female [n (%)]185 (34)98 (42)0.072291.7 (37.7)271.6 (36.1)0.033 Marital status 0.0740.095 Single6 (1)1 (1)6.7 (0.9)2.5 (0.3) Married437 (81)177 (75)607.0 (78.6)580.6 (77.3) Divorced15 (3)6 (3)22.4 (2.9)17.9 (2.4) Widowed79 (15)52 (22)136.6 (17.7)150.5 (20.0) Education level 0.6630.063 No education15 (3)11 (5)21.2 (2.7)18.8 (2.5) Primary school150 (28)61 (26)223.5 (28.9)208.5 (27.7) Junior high school217 (40)94 (40)296.8 (38.4)282.5 (37.6) High school/vacational school104 (19)44 (19)154.9 (20.1)154.0 (20.5) College degree or above51 (9)26 (11)76.3 (9.9)87.7 (11.7)Family history [n (%)]25 (5)20 (8)0.05543.4 (5.6)49.0 (6.5)0.038 Medical insursance 0.3390.088 Employee Insurance425 (80)198 (84)627.1 (81.2)629.8 (83.8) Resident Insurance100 (19)36 (15)131.8 (17.1)112.8 (15.0) Commercial Insurance1 (1)0 (0)1.0 (0.1)0.0 (0.0) Self-funded11 (2)2 (1)12.8 (1.7)8.8 (1.2)Smoke [n (%)]197 (37)51 (22)< 0.001256.0 (33.1)246.0 (32.7)0.008Drink [n (%)]165 (31)52 (22)0.017212.1 (27.5)220.7 (29.4)0.042Hypertension [n (%)]383 (71)172 (73)0.721551.3 (71.3)532.0 (70.8)0.012Diabetes [n (%)]170 (32)69 (29)0.558248.4 (32.1)227.6 (30.3)0.040Dyslipidemia [n (%)]135 (25)53 (22)0.478190.7 (24.7)182.9 (24.3)0.008AF [n (%)]62 (12)69 (29)< 0.001117.8 (15.2)131.4 (17.5)0.061CHD [n (%)]62 (12)40 (17)0.05488.2 (11.4)92.4 (12.3)0.027Stroke [n (%)]156 (29)59 (25)0.285220.6 (28.5)196.8 (26.2)0.053Hypertension medications [n (%)]245 (46)114 (48)0.542361.0 (46.7)340.7 (45.3)0.028Diabetes medications [n (%)]130 (24)55 (23)0.857194.3 (25.1)176.2 (23.4)0.040Dyslipidemia medications [n (%)]51 (9)19 (8)0.61172.0 (9.3)71.9 (9.6)0.000BMI (mean ± SD)24.98 ± 3.7725.15 ± 3.510.55125.03 ± 3.8225.15 ± 3.470.032Baseline NIHSS score (mean ± SD)4.95 ± 4.3010.07 ± 7.66< 0.0016.84 ± 7.216.63 ± 6.010.009Pre-onset mRS score (mean ± SD)0.20 ± 0.550.34 ± 0.830.0080.25 ± 0.640.26 ± 0 0.670.008aContinuous variables for Mann-Whitney U test, categorical variables for chi-square test; bContinuous variables for weighted t test, and categorical variables for weighted chi-square test. EMS, Emergency Medical Services; SMD, standardized mean difference; AF, atrial fibrillation; CHD, coronary heart disease; BMI, body mass index; NIHSS, National Institute of Health Stroke Scale; mRS, modified Rankin Scale Baseline characteristics of included patients aContinuous variables for Mann-Whitney U test, categorical variables for chi-square test; bContinuous variables for weighted t test, and categorical variables for weighted chi-square test. EMS, Emergency Medical Services; SMD, standardized mean difference; AF, atrial fibrillation; CHD, coronary heart disease; BMI, body mass index; NIHSS, National Institute of Health Stroke Scale; mRS, modified Rankin Scale Clinical outcomes The mRS scores of the included patients at the 3-month follow-up are shown in Fig. 2. Compared with the self-transported group, patients in the EMS-transported group were less likely to achieve good functional recovery (mRS 0–1) during follow-up (unadjusted RR = 3.03, 95% CI = 2.29–4.05, adjusted RR = 1.43, 95% CI = 1.06–1.96) However, this association was not statistically significant after IPW (weighted RR = 1.37, 95% CI = 0.98–1.93) (Table 2). Regarding secondary outcomes, after multivariable adjustment and IPW, the EMS-transported patients were more likely to have poor neurological recovery (mRS 3–6) at discharge (adjusted RR = 1.37, 95% CI = 1.08–1.77, weighted RR = 1.43, 95% CI = 1.06–1.92), while the remaining secondary outcomes showed no significant associations (Table 2). Sensitivity analysis yielded consistent results, EMS transport was associated with poorer neurological recovery (mRS 3–6) at discharge (after PSM, RR = 1.70, 95% CI = 1.09–2.65), with no statistically significant differences observed for other clinical outcomes (Table S2, S3). Fig. 2Distribution of the modified Rankin Scale (mRS) score at follow-up. The mRS distribution shown in this figure represents the unadjusted raw cohort study, and there are differences in baseline characteristics between groups Distribution of the modified Rankin Scale (mRS) score at follow-up. The mRS distribution shown in this figure represents the unadjusted raw cohort study, and there are differences in baseline characteristics between groups Table 2Different admission routes and clinical outcomes among patientsOutcomeSelf-transported (n = 537)EMS-transported(n = 236)Unweighted patientsWeighted risk ratio (95% CI)bUnadjusted risk ratio (95% CI)Adjusted risk ratio (95% CI)aPrimary outcome mRS at follow-up 0–1465 (86.6)140 (59.3)refrefref 2–672 (13.4)96 (40.7)3.03 (2.29–4.05)1.43 (1.06–1.96)1.37 (0.98–1.93)Secondary outcomes mRS at follow-up 0–2485 (90.3)162 (68.6)refrefref 3–652 (9.7)74 (31.4)3.24 (2.41–4.58)1.23 (0.89–1.71)1.22 (0.81–1.84) mRS at follow-up 0–3507 (94.4)183 (77.5)refrefref 4–630 (5.6)53 (22.5)4.02 (2.71–6.26)1.18 (0.82–1.75)1.15 (0.67–1.97) mRS at discharge 0–1373 (69.5)94 (39.8)refrefref 2–6164 (30.5)142 (60.2)1.97 (1.68–2.36)1.13 (0.97–1.37)1.19 (0.96–1.47) mRS at discharge 0–2449 (83.6)122 (51.7)refrefref 3–688 (16.4)114 (48.3)2.95 (2.37–3.81)1.37 (1.08–1.77)1.43 (1.06–1.92) mRS at discharge 0–3492 (91.6)155 (85.7)refrefref 4–645 (8.4)81 (34.3)4.10 (3.01–6.04)1.37 (0.98–1.88)1.38 (0.91–2.09)Safety outcomesSurvival at discharge529 (98.5)211 (89.4)refrefrefDeath at discharge8 (1.5)25 (10.6)7.11 (3.30-18.28)2.014 (0.89–6.31)1.36 (0.50–3.70)Safety outcomesSurvival at follow-up520 (96.8)203 (86.0)refrefrefDeath at follow-up17 (3.2)33 (14.0)4.42 (2.53–9.06)1.35 (0.72–2.71)1.15 (0.55–2.39)aAdjustments were made for logistic regression analysis for age, sex, marital status, education level, family history, type of medical insurance, smoking, drinking, risk factors for diseases including hypertension, diabetes, dyslipidemia, atrial fibrillation, coronary heart disease, and prior stroke, medication use for hypertension, diabetes, and dyslipidemia, body mass index, baseline National Institutes of Health Stroke Scale score and pre-onset modified Rankin Scale (mRS) score. bModified Poisson regression using inverse probability weighting Table 2Different admission routes and clinical outcomes among patientsOutcomeSelf-transported (n = 537)EMS-transported(n = 236)Unweighted patientsWeighted risk ratio (95% CI)bUnadjusted risk ratio (95% CI)Adjusted risk ratio (95% CI)aPrimary outcome mRS at follow-up 0–1465 (86.6)140 (59.3)refrefref 2–672 (13.4)96 (40.7)3.03 (2.29–4.05)1.43 (1.06–1.96)1.37 (0.98–1.93)Secondary outcomes mRS at follow-up 0–2485 (90.3)162 (68.6)refrefref 3–652 (9.7)74 (31.4)3.24 (2.41–4.58)1.23 (0.89–1.71)1.22 (0.81–1.84) mRS at follow-up 0–3507 (94.4)183 (77.5)refrefref 4–630 (5.6)53 (22.5)4.02 (2.71–6.26)1.18 (0.82–1.75)1.15 (0.67–1.97) mRS at discharge 0–1373 (69.5)94 (39.8)refrefref 2–6164 (30.5)142 (60.2)1.97 (1.68–2.36)1.13 (0.97–1.37)1.19 (0.96–1.47) mRS at discharge 0–2449 (83.6)122 (51.7)refrefref 3–688 (16.4)114 (48.3)2.95 (2.37–3.81)1.37 (1.08–1.77)1.43 (1.06–1.92) mRS at discharge 0–3492 (91.6)155 (85.7)refrefref 4–645 (8.4)81 (34.3)4.10 (3.01–6.04)1.37 (0.98–1.88)1.38 (0.91–2.09)Safety outcomesSurvival at discharge529 (98.5)211 (89.4)refrefrefDeath at discharge8 (1.5)25 (10.6)7.11 (3.30-18.28)2.014 (0.89–6.31)1.36 (0.50–3.70)Safety outcomesSurvival at follow-up520 (96.8)203 (86.0)refrefrefDeath at follow-up17 (3.2)33 (14.0)4.42 (2.53–9.06)1.35 (0.72–2.71)1.15 (0.55–2.39)aAdjustments were made for logistic regression analysis for age, sex, marital status, education level, family history, type of medical insurance, smoking, drinking, risk factors for diseases including hypertension, diabetes, dyslipidemia, atrial fibrillation, coronary heart disease, and prior stroke, medication use for hypertension, diabetes, and dyslipidemia, body mass index, baseline National Institutes of Health Stroke Scale score and pre-onset modified Rankin Scale (mRS) score. bModified Poisson regression using inverse probability weighting Different admission routes and clinical outcomes among patients aAdjustments were made for logistic regression analysis for age, sex, marital status, education level, family history, type of medical insurance, smoking, drinking, risk factors for diseases including hypertension, diabetes, dyslipidemia, atrial fibrillation, coronary heart disease, and prior stroke, medication use for hypertension, diabetes, and dyslipidemia, body mass index, baseline National Institutes of Health Stroke Scale score and pre-onset modified Rankin Scale (mRS) score. bModified Poisson regression using inverse probability weighting Mortality at discharge was 10.6% (25/236, neurological death 25/25) in the EMS-transported group and 1.5% (8/537, neurological death 8/8) in the self-transported group. In unadjusted analysis, EMS-transported was associated with a higher likelihood of death at discharge (unadjusted RR = 7.11, 95% CI = 3.30-18.28). However, this association was not significant after multivariable adjustment (adjusted RR = 2.01, 95% CI = 0.89–6.31) or IPW (weighted RR = 1.36, 95% CI = 0.50–3.70) (Table 2). At follow-up, mortality occurred in 14.0% (33/236, neurological death 5/33) of the EMS-transported group and 3.2% (17/537, neurological death 8/17) of the self-transported group (unadjusted RR = 4.42, 95% CI = 2.53–9.06, adjusted RR = 1.35, 95% CI = 0.72–2.71, weighted RR = 1.15, 95% CI = 0.55–2.39) (Table 2). Sensitivity analysis results were consistent with these findings, no statistically significant association was observed between transport mode and all-cause mortality at discharge or follow-up (Table S3). Post hoc analysis Table 3 shows the differences in key indicators of the two groups’ admission methods after weighting. The ODT of the EMS-transported group was significantly shorter than that of the self-transported group, however, the DNT of the EMS-transported was longer than self-transported (weighted P < 0.05). The results of the sensitivity analysis are consistent with the above. Moreover, the sensitivity analysis results indicate that the ONT for the self-transported group is significantly longer than that of the 120-admitted group (weighted P < 0.05) (Table S4). Table 3Post hoc analysis among weighted patientsSelf-transported (n = 537)EMS-transported(n = 236)P-valueIVT [n (%)]95.08%95.07%0.965EVT [n (%)]18.11%20.64%0.507ODT [median, (IQR)]120 (60–210)90 (60–180)0.011DNT [median, (IQR)]23 (17–30)26 (20–34)0.006ONT [median, (IQR)]144 (88–230)133 (85–200)0.056DPT [median, (IQR)]137 (102–275)124 (107–273)0.792EMS, Emergency Medical Services; IVT, intravenous thrombolysis; EVT, endovascular thrombectomy; ODT, onset-to-door time; DNT, door-to-needle time; ONT, onset-to-needle time; DPT, door-to-puncture time Table 3Post hoc analysis among weighted patientsSelf-transported (n = 537)EMS-transported(n = 236)P-valueIVT [n (%)]95.08%95.07%0.965EVT [n (%)]18.11%20.64%0.507ODT [median, (IQR)]120 (60–210)90 (60–180)0.011DNT [median, (IQR)]23 (17–30)26 (20–34)0.006ONT [median, (IQR)]144 (88–230)133 (85–200)0.056DPT [median, (IQR)]137 (102–275)124 (107–273)0.792EMS, Emergency Medical Services; IVT, intravenous thrombolysis; EVT, endovascular thrombectomy; ODT, onset-to-door time; DNT, door-to-needle time; ONT, onset-to-needle time; DPT, door-to-puncture time Post hoc analysis among weighted patients EMS, Emergency Medical Services; IVT, intravenous thrombolysis; EVT, endovascular thrombectomy; ODT, onset-to-door time; DNT, door-to-needle time; ONT, onset-to-needle time; DPT, door-to-puncture time Subgroup analysis Fig S1 shows that the associations between admission routes and primary outcomes were consistent across the exploratory key group analyses, including those defined by age, sex, drinking status, history of hypertension, diabetes, dyslipidemia, AF, stroke, obesity, stroke severity and treatment methods. However, potential heterogeneity was observed in non-smoking patients, among whom EMS-transported was associated with poorer functional recovery (weighted RR = 1.63, 95% CI = 1.11–2.40; P for interaction = 0.039). Similarly, EMS-transported was associated with poorer functional recovery in patients with a history of CHD (weighted RR = 4.74, 95% CI = 1.83–12.32; P for interaction = 0.007) (Fig S1). The results of post hoc exploratory analyses for each subgroup are provided in Table S5. Compared with the self-transported group, the EMS-transported group tended to have a shorter ODT and a longer DNT, and no association was found between the length of ODT, DNT, and ONT and the presence of stroke risk factors (Table S5). Discussion Our study found that compared with self-transported patients, those transported by EMS showed no significant improvement in functional recovery for the primary clinical outcome. However, for secondary outcomes, the EMS-transported group had worse functional recovery (defined as a mRS 0–2 vs. 3–6 at discharge). Mortality did not differ significantly between the two groups. Although the ODT was significantly shorter in the EMS-transported group, the DNT was longer. Exploratory subgroup analysis indicated potential heterogeneity based on CHD history and smoking status. Among patients with a history of CHD, those in the EMS-transported group had worse 3-month functional recovery. A similar pattern was observed among non-smokers, in whom EMS transport was also associated with worse 3-month functional recovery. The association between stroke severity and EMS transport is widely recognized [16]. Advanced age and AF serve as direct factors influencing disease severity, previous studies have confirmed that patients transported by EMS tend to be older, present with more severe conditions, and exhibit a higher prevalence of AF, these findings are consistent with our results [11, 15]. This suggests that EMS transport constitutes a highly selective clinical pathway. To account for this selection bias and balance baseline covariates across groups, we performed IPW. IPW offers significant advantages in addressing sample size imbalances between groups and estimating marginal treatment effects [17, 18]. A retrospective study involving 2123 patients receiving IVT across 70 hospitals in Zhejiang, China, found no association between EMS transport and favorable 3-month outcomes (mRS score 0–2) [19]. However, a recent observational study of patients treated with alteplase within 4.5 h of symptom onset found that the rate of favorable outcomes (mRS score 0–2 at 3-month follow-up) was lower in the EMS-transported group. In the subgroup of patients with moderate-to-severe stroke (defined as NIHSS score 16–20), EMS transport was associated with favorable outcomes [12]. The statistical analyses in such studies relied solely on univariate logistic regression, failed to control for confounding factors, and lacked sufficient statistical power to definitively establish an association between EMS transport and patient outcomes. This study design addresses limitations of prior research by using the weighted population simulation randomized controlled trial design and conducting sensitivity analyses using PSW to validate result robustness. Additionally, the prospective collection of clinical outcome effectively reduced the potential bias. Furthermore, this study shows that the ODT was significantly shorter in the EMS-transported group than in the self-transported group, suggesting that EMS can effectively reduce pre-hospital delays. In fact, patients transported by EMS can be identified early, prioritized for hospitals with stroke treatment capabilities, and transported to the hospital quickly; in contrast, self-transported patients may experience pre-hospital delays due to misinterpretation of stroke symptoms, difficulties in selecting a hospital, and time lost during the journey. However, there was no statistical difference in ONT between the two groups, and the DNT was actually longer in the EMS-transported group. This may explain why the proportion of patients with an mRS score of 0–2 at discharge was lower in the EMS-transported group than in the self-transported group in this study; the longer delay in initiating in-hospital treatment in the EMS-transported group may have had a more significant impact on short-term outcomes at discharge. Previous studies have indicated that EMS transport alone cannot shorten DNT or ONT. EMS combined with pre-hospital notification can significantly shorten both DNT and ONT, and the 3-month neurological recovery is better [20]. However, as stroke treatment system mature, the implementation of a coordinated care platform—where patient information precedes the patient—has become more common. Within this context, we hypothesize that a potential underlying factor may be the “waiting for family members” phenomenon. Self-transported patients are typically accompanied by family who can sign informed consent forms immediately, whereas in China, EMS services limit the number of family members allowed in the ambulance, so EMS-transported patients may face delays while awaiting family arrival, which could postpone treatment. While this interpretation remains speculative, it suggests opportunities for process improvement. Taken together, these findings may highlight the need for specific policy adjustments and process optimizations to address the delays observed in EMS-transported patients. Approaches such as implementing remote or telephone informed consent, or evaluating the feasibility of implied consent principles in time-sensitive emergency like thrombolysis, may represent promising strategies to preserve the time advantage gained during pre-hospital EMS transport and prevent its erosion during subsequent care, thereby potentially improving patient outcomes. Since patient outcomes depend more on the efficiency of the entire care chain than on any single link, it is necessary to optimize each process specifically. We are currently exploring remote informed consent to further reduce DNT. Exploratory subgroup analysis indicated that patients with a history of CHD or those who were non-smokers had poorer 3-month neurological recovery after EMS transport. Although the results of this subgroup analysis are intended to raise hypotheses, in clinical practice, a confirmed stroke diagnosis complicates management for CHD patients, necessitating more cautious treatment decisions [21]. Previous studies have shown that the distribution of stroke risk factors differs between younger and older populations [22]. In this study, non-smokers were significantly older than smokers, although the weighted data balanced for age, non-smoking patients tended to have poorer tolerance to ischemia and hypoxia due to milder atherosclerosis, which may explain their worse prognosis. Post-hoc analysis of key subgroups further revealed that despite shorter ODT among the EMS-transported group, individuals with traditional stroke risk factors did not exhibit shorter ODT, DNT, or ONT compared to the general stroke population. These findings highlight the importance of targeted health education during pre-hospital screening of stroke-prone populations. Additionally, hospitals could consider implementing a risk-stratified response mechanism to prioritize treatment for high-risk stroke patients. Unlike previous studies, the value of this study lies in revealing the complexity of the entire stroke care chain. We found that while EMS significantly reduced ODT, it did not translate into better functional outcomes and was even associated with poorer functional recovery in secondary outcomes. Furthermore, results from subgroup analyses suggest that a history of CHD and smoking may modulate the association between EMS transport and prognosis, a point that has not been systematically explored previously. This study has several limitations. First, as a single-center study, the patient cohort may lack generalizability, and the findings may be applicable only to hospitals that use similar protocols, given the significant differences in hospital infrastructure and EMS coordination mechanisms across different countries and regions. Second, the study primarily included middle-aged and elderly patients, which imposes certain limitations on the study population and sample size. Third, subgroup analyses were not adjusted for multiple comparisons, which increases the risk of Type I error, therefore, they should be strictly regarded as exploratory and hypothesis-generating. Finally, despite the use of IPW to balance baseline characteristics, residual confounding and confounding by indication may still influence the study results. Our study provides new evidence for understanding the association between EMS and the prognosis of patients with AIS, while also suggesting several areas worthy of further exploration. First, given that there are still differences in the pathophysiology, prognosis and clinical features among various stroke subtypes, it would be highly meaningful to evaluate the impact of admission pathways on clinical outcomes for different stroke subtypes [23]. Second, regarding the hypothesis of “waiting for family members” identified in our study, future interventional studies involving remote informed consent could be designed to assess improvements in functional outcomes. Third, the potential heterogeneity observed in the subgroup analysis results regarding CHD and smoking requires further clinical validation. Conclusions In this prospective cohort study, we found that EMS transport did not improve 3-months functional recovery compared with the self-transported group based on IPW. Functional recovery at discharge was actually worse in the EMS-transported group (mRS 0–2 vs. 3–6). Despite a significantly shorter ODT in the EMS-transported group, the DNT was prolonged. Patients with CHD or those who were non-smokers experienced poorer outcomes when transported via EMS. Our study may highlight the need for enhanced pre-hospital health education targeting high-risk populations and the establishment of risk-stratified response mechanism. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1