Direct Transport to a Thrombectomy Center Accelerates Acute Stroke Reperfusion Therapies in Nonurban Settings

Abstract

Background

In the United States, the impact of bypassing the nearest local stroke center to facilitate direct transport of nonurban patients to a thrombectomy‐capable center (TCC) for mechanical thrombectomy (MT) remains unclear. We compared intravenous thrombolysis and MT treatment times between patients transferred directly to a TCC and those transported initially to a local stroke center and undergoing MT.

Methods

In this retrospective observational study within an academic health care system with integrated telestroke and transport services, we included consecutive nonurban patients (ground transport ≥15 minutes to TCC) who were transported via emergency medical services and underwent MT between January 2021 and October 2024. Patients were categorized into those transported directly to TCC or those initially taken to 1 of 12 telestroke‐supported local stroke centers and subsequently transferred to TCC. These groups were compared using inverse probability of treatment weighting. Outcomes included first medical contact–to–intravenous thrombolysis, first medical contact–to–arterial puncture, and first medical contact–to–arrival times, among others. All times are expressed as median (IQR).

Results

Among 304 patients (transported directly to TCC, 174; initial transport to local stroke center, 130; median age, 72 [IQR, 62–82] years; median National Institutes of Health Stroke Scale score, 16 [IQR, 12–22]; 36.5% air transported), first medical contact–to–intravenous thrombolysis (75 [IQR, 65–89] versus 86 [IQR, 72–110] minutes; P<0.001), door‐to‐needle time (40 [IQR, 28–45] versus 63 [IQR, 47–80] minutes; P<0.01) and first medical contact–to–arterial puncture (117 [IQR, 103–138] versus 197 [IQR, 164–271] minutes; P<0.001) times were shorter in transport directly to TCC, despite longer FMC‐to‐arrival (36 [IQR, 30–43] versus 27 [IQR, 22–36] minutes; P<0.001) and door‐to‐puncture (74 [IQR, 65–93] versus 25 [IQR, 16–61] minutes; P<0.001) times.

Conclusions

Among nonurban patients undergoing MT, direct transport to a TCC accelerated treatment times and may improve clinical outcomes. Our study highlights inefficiencies in regional systems of care.

Nonstandard Abbreviations and Acronyms

DIDO

door in–door out

DTN

door to needle

DTT

direct to TCC

FMC

first medical contact

IPTW

inverse probability of treatment weighting

IVT

intravenous thrombolysis

LSC

local stroke center

LVO

large‐vessel occlusion

MT

mechanical thrombectomy

PSSS

Prehospital Stroke Severity Scale

RACECAT

Effect of Direct Transfer to an Endovascular Treatment–Capable Stroke Center on Outcomes in Patients With Acute Ischemic Stroke and Suspected Large Vessel Occlusion Identified by Emergency Medical Services

STRATIS

Systematic Evaluation of Patients Treated With Neurothrombectomy Devices for Acute Ischemic Stroke

TCC

thrombectomy‐capable center

TLSC

initial transport to local stroke center

TRIAGE‐STROKE

Transport Strategy in Patients With Suspected Acute LVO

VN

vascular neurology

Clinical Perspective.

What Is New?

• In this retrospective observational study of 304 nonurban patients with suspected large‐vessel occlusion strokes treated within a large US‐based academic health care system with integrated telestroke and transport services, direct transport to a thrombectomy‐capable center rather than initial transport to a nearby local stroke center reduced the time from first medical contact to mechanical thrombectomy by 80 minutes and to intravenous thrombolysis by 11 minutes, despite longer prehospital transport times.

• By focusing on first medical contact–to–treatment metrics rather than in‐hospital intervals alone, this study offers system‐level insight into prehospital triage and destination decision‐making in the United States, in contrast with prior European studies centered primarily on in‐hospital workflow metrics.

What Are the Clinical Implications?

• Our study demonstrates that in nonurban US settings, direct emergency medical services transport to a thrombectomy‐capable center may expedite both thrombolysis and thrombectomy by reducing first medical contact–to–treatment delays across the continuum of acute stroke care, thereby identifying system strengths and inefficiencies and informing regional clinical practice and emergency medical services transport policy.

Acute ischemic stroke treatment includes administration of intravenous thrombolysis (IVT) and mechanical thrombectomy (MT) for eligible patients. ¹ While these treatments are highly effective, they are extremely time sensitive. ² , ³ Moreover, centers that perform MT (thrombectomy‐capable centers [TCCs]) are relatively few and clustered in urban locations limiting or delaying access to this treatment for a large majority of the nonurban US population. ⁴ Recent observational studies show that even among well‐organized stroke networks, patients with large‐vessel occlusion (LVO) transported from a local stroke center (LSC) to a TCC after diagnosis of LVO have worse functional outcomes compared with those presenting directly from the field to the TCC. ⁵ This difference is primarily driven by shorter times to MT among patients presenting directly to a TCC despite a modest delay in delivery of IVT. However, 2 randomized controlled trials in Europe showed that while a strategy of bypassing an LSC on the basis of stroke severity reduced time to delivery of MT, it substantially delayed IVT administration times, negating the benefit of direct transport to a TCC over transporting them to the nearest LSC. ⁶ , ⁷ Since delivery of these time‐sensitive treatments relies on workflow efficiencies at individual treating facilities and regional systems of care, we sought to determine the impact of direct transport of patients with suspected LVO from the field to a TCC by emergency medical services (EMS) compared with transport to the nearest LSC followed by interfacility transport to a TCC on stroke treatment time metrics within a large health care system with an integrated telestroke and transport network.

Methods

This is a retrospective analysis of a prospectively maintained institutional acute ischemic stroke registry including consecutive patients with stroke undergoing MT at 2 TCCs from January 2021 through October 2024. Our analysis was conducted according to the Strengthening the Reporting of Observational Studies in Epidemiology criteria for observational studies (Tables S1–S3). ⁸ The data that support the findings of this study are available from the corresponding author upon reasonable request. This study was approved by the Institutional Review Board of University of Pittsburgh School of Medicine, and the requirement for informed consent was waived due to the retrospective nature of the analysis.

Study Setting

Our telestroke network, overseen by our vascular neurology (VN) team (attending and/or fellow) delivers acute stroke treatment services to >40 facilities. Among these, we had comprehensive data (including prehospital, in‐hospital, and interfacility transport) for 2 TCCs that maintain dedicated, in‐person, continuous availability of a stroke team member (neurology trainee, advanced practice provider, or attending) and 12 surrounding LSCs that refer patients with LVO to these TCCs (Figure 1).

Figure 1. Map of western Pennsylvania.

LSC indicates local stroke center; and TCC, thrombectomy‐capable center.

The details of prehospital care and in‐hospital care when patients present directly to a TCC were previously published. ⁹ Among patients who have a high suspicion of LVO (modified Rapid Arterial Occlusion Evaluation score ≥5, Cincinnati Prehospital Stroke Scale score ≥2), the EMS protocol encourages direct transport to a TCC if it can be reached within 45 minutes regardless of their IVT eligibility. ¹⁰ At the TCCs, the in‐person VN team member determines IVT eligibility for patients in the emergency department in conjunction with the emergency department team (MD or advanced practice provider) while concurrently activating the neuroendovascular team. At the included LSCs, when there is a patient with suspected acute ischemic stroke, the emergency department staff activates our telestroke services via a telephone call or a Health Insurance Portability and Accountability Act–compliant web‐based digital application. Cerebrovascular imaging, most commonly computed tomography angiogram of the head and neck to screen for LVO is routinely performed for all patients as a part of standard acute stroke evaluation. Patients who qualify for IVT administration are routinely evaluated via a 2‐way audio/video interactive telemedicine platform, while patients with an LVO who do not qualify for IVT are evaluated via telephone consultation to coordinate rapid transfer to a TCC. A central medical command facility in conjunction with our institution’s transfer service help coordinate timely interfacility transport to the appropriate destination. A substantial proportion of patients who receive index stroke imaging at the referral facility will be transported directly to the neuroendovascular angiography suite, while a few may be routed to the emergency department to obtain additional imaging/evaluations as deemed clinically appropriate. ¹¹

Patient Selection

We included consecutive patients who presented via EMS to the LSC (when the LSC was the closest stroke center) or TCC with a prearrival stroke alert notification that underwent MT. A recent analysis of travel time indicates that in the urban United States, the median ground transportation time to the nearest stroke center and total time to bypass the nearest stroke center to arrive at a thrombectomy center are 9.9 minutes and 19.8 minutes, respectively. ¹² To target a nonurban population and eliminate patients transported directly to a TCC solely because it was the closest stroke center, we excluded those who arrived at the TCC with a ground transport time <15 minutes. The patients who were diagnosed with an acute ischemic stroke during their hospitalization or were deemed ineligible for emergent MT upon presentation to a TCC were also excluded (Figure 2). One LSC had intermittent coverage of TCC capabilities; patients were selected only if they presented when there were no MT capabilities at that center. We collected the time of first medical contact (FMC) by EMS personnel, transport times and distances, in‐hospital stroke time metrics including door in–door out (DIDO) times, demographic variables, baseline National Institutes of Health Stroke Scale score, and imaging characteristics.

Figure 2. Flowchart for patient selection.

DTT indicates direct transport to a thrombectomy‐capable center; ED, emergency department; MT, mechanical thrombectomy; NIHSS, National Institutes of Health Stroke Scale; TCC, thrombectomy‐capable center; and TLSC, initial transport to a local stroke center.

Outcomes

Our primary outcomes were FMC–to–arterial puncture and FMC‐to‐IVT times. Our secondary outcomes included (1) FMC to arrival at the initial stroke facility, (2) time from hospital arrival (either LSC or TCC) to administration of IVT (door‐to‐needle [DTN]), (3) time from hospital arrival (TCC) to arterial puncture (door to puncture), (4) travel time to the first medical facility, (5) good functional outcomes (modified Rankin Scale score, 0–2) at discharge, and (6) in‐hospital death. We also report DIDO, defined as the time from arrival to transfer from the referring LSC. According to the transport strategy adopted by EMS personnel, our cohort was divided into 2 groups: (1) patients who were transported directly from the field to 1 of the TCCs (DTT group) and (2) patients who were initially transported to the closest LSC with subsequent transport to a TCC (TLSC group).

Statistical Analysis

We performed inverse probability of treatment weighting (IPTW) analysis to compare our primary outcomes in the DTT and TLSC groups by first calculating the standardized mean difference for baseline characteristics. Covariate balance was evaluated using standardized differences and kernel density plots. ¹³ Baseline variables with a standardized mean difference of >0.1 were considered unbalanced and included in the logistic regression model to calculate the propensity score, provided they occurred before the decision for the initial destination. Since IVT administration occurred after the initial decision for transport destination was made, it was not adjusted for in our model. Additionally, rates of Prehospital Stroke Severity Scale (PSSS) acquisition and air transport were expected to be more frequent in the DTT group and should not impact treatment times in the TLSC. These were excluded to prevent distortion of the causal relationship. We then calculated the average treatment effect of population using the propensity score and created the IPTW‐adjusted population on the basis of average treatment effect of population weights (Figures S1–S3). ¹⁴

To account for the pseudo‐population’s inflated sample size and induced correlation, we used the robust “sandwich” variance estimator to provide reliable variance and CIs. We further fitted a linear regression or binary logistic regression, weighted by average treatment effect of population as appropriate, to assess the significance of associations with outcomes. To address skewness in continuous outcome variables, we applied a log transformation and analyzed them with linear regression using the gaussian method, weighted by average treatment effect of population. ¹⁵ Statistical significance and CIs were determined using the robust “sandwich” variance estimator, with a 2‐sided statistical significance defined as P<0.05. For clarity of presentation, continuous outcomes were summarized as weighted medians and interquartile ranges (IQRs) to reflect the adjusted pseudo‐population distribution, whereas binary outcomes were reported as observed counts with IPTW‐weighted adjusted odds ratios and corresponding P values derived from weighted regression models. All statistical analyses were performed using R version 4.1.1 (R Foundation for Statistical Computing, Vienna, Austria) R studio software.

Results

In the overall cohort, median age and baseline National Institutes of Health Stroke Scale were 72 (IQR, 62–81) years, and 16 (IQR, 12–22), respectively. A total of 55.9% were women, 29.3% received IVT, and 49.7% presented to the initial hospital within 4 hours of last‐known‐well time. In the unweighted population, the patients in the DTT group were less likely to receive IVT (26.4% versus 33.1%) and had a lower rate of atrial fibrillation (26.4% versus 32.3%) and lower baseline Alberta Stroke Program Early Computed Tomography Score (8 versus 9). A higher proportion of DTT patients had PSSS acquisition (67.8% versus 33.8%), air transport to the initial receiving facility (58% versus 7.7%), tandem occlusions (29.3% versus 11.5%), and proximal site of anterior circulation intracranial occlusions (Table 1).

Table 1.

Baseline Characteristics

	Overall, n=304	Pre‐IPTW adjustment		SMD	Post‐IPTW adjustment		SMD
		DTT, n=174 (57.2%)	TLSC, n=130 (42.8%)		DTT,	TLSC,
					n=174.3	n=129.2
Age, y, median (IQR)	72 (62–81)	72 (61–80)	73 (62–81)	0.04	72 (61–80)	71 (61–81)	0.02
Female, n (%)	170 (55.9)	95 (54.6)	75 (57.7)	0.06	95.9 (55.0)	72.9 (56.4)	0.04
Atrial fibrillation, n (%)	88 (28.9)	46 (26.4)	42 (32.3)	0.1	51.3 (29.4)	38.6 (29.9)	0.01
Diabetes, n (%)	91 (29.9)	51 (29.3)	40 (30.8)	0.03	52.3 (30.0)	36.3 (28.1)	0.04
Hyperlipidemia, n (%)	190 (62.5)	110 (63.2)	80 (61.5)	0.03	110.5 (63.4)	78.3 (60.6)	0.06
Hypertension, n (%)	222 (73.0)	124 (71.3)	98 (75.4)	0.09	125.2 (71.9)	94.0 (72.7)	0.02
History of previous stroke, n (%)	50 (16.4)	30 (17.2)	20 (15.4)	0.05	32.5 (18.6)	17.7 (13.7)	0.1
Location, n (%)				0.2			0.04
Internal carotid artery, n (%) M1	221 (72.7)	132 (75.9)	89 (68.5)		127.3 (73.0)	95.9 (74.2)
M2	63(20.7)	33 (19.0)	30 (23.1)		36 (20.7)	25.1 (19.4)
DMVO	8 (2.6)	4 (2.3)	4 (3.1)		4.6 (2.6)	3.1 (2.4)
Basilar artery	12 (7.2)	5 (2.9)	7 (5.4)		6.4 (3.7)	5.1 (4)
Tandem, n (%)	66 (21.7)	51 (29.3)	15 (11.5)	0.5	37.7 (21.7)	27.4 (21.2)	0.01
ASPECTS, median (IQR)	9 (7–10)	8 (7–10)	9 (7–10)	0.2	9 (7–10)	9 (7–10)	0.03
NIHSS, median (IQR)	16 (12–22)	17 (13–21)	16 (11–22)	0.09	16 (13–21)	17 (11–22)	<0.01
PSSS acquisition before arriving at the first medical facility, n (%)	162 (53.2)	118 (67.8)	44 (33.8)	0.7	120.2 (69.0)	42.4 (32.8)	0.8
Mode of transport to the first medical facility, n (%)				1.3			1.3
Air	111 (36.5)	101 (58.0)	10 (7.7)		102.7 (58.9)	9.1 (7.1)
Ground	193 (63.4)	73 (42.0)	120 (92.3)		71.5 (41.1)	120.1 (92.9)
Last known well to first medical facility ≤4 h, n (%)	151 (49.7)	84 (48.3)	65 (50.0)	0.03	91.0 (52.2)	57.7 (44.6)	0.1
Last known well to first medical contact, min, median (IQR)	207 (57.3–513.5)	210.00 (46.00–576.00)	206.5 (75–502.7)	0.08	275.38 (50.6–605.3)	184.00 (70.8–466.6)	0.1
IVT use, n (%)	89 (29.3)	46 (26.4)	43 (33.1)	0.14	51.0 (29.3)	38.0 (29.4)	<0.01

ASPECTS indicates Alberta Stroke Program Early Computed Tomography Score; DMVO, distal medium‐vessel occlusion; DTT, direct to thrombectomy‐capable center; IPTW, inverse probability of treatment weighting; IQR, interquartile range; IVT, intravenous thrombolytics; M1, first segment of the middle cerebral artery; M2, second segment of the middle cerebral artery; NIHSS, National Institutes of Health Stroke Scale; PSSS, Prehospital Stroke Severity Scale (either modified Rapid Arterial Occlusion Evaluation or NIHSS); SMD, standardized mean difference; and TLSC, initial transport to local stroke center.

Post‐IPTW, all variables used to construct the IPTW sample were balanced. Median FMC‐to‐IVT was significantly shorter in DTT compared with TLSC (75 [IQR, 65–89] versus 86 [IQR, 72–110] minutes; P<0.001). Similarly, FMC–to–arterial‐puncture time was shorter in DTT (117 [IQR, 103–138] versus 197 [IQR, 164–271] minutes; P<0.001). In contrast, FMC to arrival at the initial stroke facility was longer (36 [IQR, 30–43] versus 27 [IQR, 22–36] minutes; P<0.001), as was door‐to‐puncture time (74 [IQR, 65–93] versus 25 [IQR, 16–61] minutes; P<0.001). DTN time was also significantly shorter in the DTT group (40 [IQR, 28–45] versus 63 [IQR, 47–80] minutes; P<0.01; Figure 3, Table 2).

Figure 3. Median time from stroke onset (the time of last seen well) to arterial puncture.

CT indicates computed tomography; DTT, direct transport to a thrombectomy‐capable center; FMC, first medical contact; IVT, intravenous thrombolysis; and TLSC, initial transport to a local stroke center.

Table 2.

Time Metrics and Outcomes After IPTW Adjustment

	DTT	TLSC	P value	aOR (95% CI)
FMC to first medical facility, min, median (IQR)	36 (30–43)	27 (22–35)	<0.01	N/A
DTP, min, median (IQR)	74 (62–91)	25.00 (16–62)	<0.01	N/A
FMC to puncture, min, median (IQR)	117 (103–138)	197 (164–271)	<0.01	N/A
Door to needle, min, median (IQR)	40 (28–45)	63 (46–79)	<0.01	N/A
FMC to needle, min, median (IQR)	75 (65–89)	86 (72–110)	0.02	N/A
Travel time to first medical facility, min, median (IQR)	20 (16–25)	11 (8–17)	<0.01	N/A
mRS score 0–2 at discharge, n (%)	41 (23)	20 (15.3)	0.045	1.89 (1.02–3.45)
Discharge death, n (%)	18 (10.3)	21 (16.2)	0.1	0.56 (0.28–1.14)
mRS score 0–2 at 90 d, n (%)	56 (32.2)	36 (27.7)	0.4	1.25 (0.74–2.13)

aOR indicates adjusted odds ratio; DTP, door to puncture; DTT, direct to thrombectomy‐capable center; FMC, first medical contact; IPTW, inverse probability of treatment weighting; IQR, interquartile range; mRS, modified Rankin Scale; N/A, not applicable; and TLSC, initial transport to local stroke center.

Since PSSS acquisition plays an important role in transport destination decision‐making by EMS personnel in our system, we report baseline characteristics of each group stratified by the presence of PSSS. In the DTT group, patients with PSSS were more likely to be transported via air (72% versus 28.6%; P<0.001), had a higher median baseline Alberta Stroke Program Early Computed Tomography Score (9 [IQR, 8–10] versus 8 [IQR, 7–10]; P=0.04), shorter median door‐to‐puncture time (71 [IQR, 62–86] versus 79.5 [IQR, 67.5–99]; P=0.02), and longer median travel distance (29.6 [IQR, 17–39] versus 12.2 [IQR, 9–17] miles; P<0.001) compared with patients who did not have a PSSS (Table S2).

In the TLSC group, patients with PSSS had longer travel distance (9.7 [IQR, 6–13] versus 5.6 [IQR, 2–9] miles; P=0.001). The median DIDO time among LSC patients was 107 (IQR, 87–138) minutes (Table S3). Finally, while in‐patient death was similar between groups with a nonsignificant trend favoring DTT (adjusted odds ratio, 0.56 [95% CI, 0.28–1.14]; P=0.1), the rate of good outcomes (modified Rankin Scale score, 0–2) at discharge was higher among DTT patients (adjusted odds ratio, 1.89 [95% CI, 1.02–3.45]; P=0.045) without any significant difference in the rate of 90‐day modified Rankin Scale score of 0 to 2 (adjusted odds ratio, 1.25 [95% CI, 0.74–2.13]; P=0.4; Table 2).

Discussion

Our real‐world study shows that in a predominantly nonurban population undergoing MT, a strategy of direct transport by EMS to a TCC was associated with a significant acceleration in delivery of both IVT and MT. While most previous studies on timeliness of acute stroke reperfusion therapies have been limited to the assessment of in‐hospital time metrics, our emphasis on FMC to treatment time metrics broadens this scope. This approach allows us to examine the strength of our regional systems of care that heavily rely on prospectively collected prehospital quality metrics such as appropriate use of LVO screening tools in the field, appropriate triage, EMS prenotification, and timely transport of patients with suspected LVO to the optimal stroke facility. Furthermore, by measuring the time metrics at the level of individual facilities such as DTN and DIDO times, our study identifies critical workflow inefficiencies across a regional network of TCCs and LSCs.

The overarching finding of our study showing acceleration of delivery of MT among patients presenting directly to TCC aligns with previously published studies with some key differences. In the STRATIS (Systematic Evaluation of Patients Treated With Neurothrombectomy Devices for Acute Ischemic Stroke) registry, a hypothetical model showed that bypassing a non–endovascular‐capable center could modestly reduce or delay delivery of IVT among eligible patients, while it significantly accelerated delivery of MT. ⁵ This study did not provide information on the proximity of the patients presenting directly to the thrombectomy center and cannot exclude that the benefit in outcomes among directly presenting patients was in part related to the distance/transportation times to the endovascular center, which might also have been the nearest stroke center. In our study, most patients arriving at the TCC did so via air transport and had a higher rate of PSSS assessment, suggesting that the decision for direct transport to the TCC was primarily based on the field determination of stroke severity rather than proximity to the TCC. This inference is reinforced by the exclusion of patients who presented directly within a 15‐minute driving distance and that the FMC–to–arrival at the initial stroke facility time was longer in the DTT group compared with the TLSC group. In contrast, the EMS scene–to–door arrival times were shorter among directly presenting patients in the STRATIS registry, arguing in favor of the possibility that the endovascular center may have been the closest stroke center for many of the directly presenting patients in that study. This difference becomes relevant as we focus on optimizing transport destination in a substantial population of the United States that does not have rapid access to a TCC. ¹⁶ Additionally, in the RACECAT (Effect of Direct Transfer to an Endovascular Treatment–Capable Stroke Center on Outcomes in Patients With Acute Ischemic Stroke and Suspected Large Vessel Occlusion Identified by Emergency Medical Services) trial, the median time from stroke onset to the first hospital arrival was 54 minutes longer in the direct transported group compared with patients transferred to the LSC first, suggesting (though not reported) that air transport was rarely used. ⁷ In contrast, this difference was only 9 minutes in our study. While this may have to do with the geographic location of the treatment centers, it can also be attributed to patients with suspected stroke symptoms often being directly transported by air to a TCC. ¹⁷ This strategy has been shown to reduce thrombolytic treatment times ¹⁸ and may have contributed to acceleration of delivery of IVT and MT among patients presenting directly to TCC compared with patients transferred from an LSC, a unique finding that distinguishes our analysis from previous studies.

Another key difference between our study and others is the performance of the LSCs. The median DTN at LSCs in our study was >60 minutes, which is substantially higher than that in the RACECAT and TRIAGE‐STROKE (Transport Strategy in Patients With Suspected Acute LVO) trials. ⁶ , ⁷ While many factors may contribute to stroke outcomes, DTN has been used a surrogate marker of the quality of acute stroke care, and a substudy from the RACECAT trial showed that delay in systemic IVT among patients transported to an LSC was associated with suboptimal outcomes, favoring direct transport to a TCC over a drip‐and‐ship transfer paradigm. ¹⁹ The study also showed that centers that were staffed via telestroke expertise had higher DTN time compared with ones with an in‐person VN specialist. While telestroke has been critical in expanding the availability of VN expertise to remote centers, ²⁰ some studies have shown delays in DTN time associated with telestroke, primarily owing to time taken from hospital arrival to telestroke team activation at these centers. ²¹ In our study, all our LSCs were staffed via telemedicine, which may have contributed to substantial prolongation of DTN. Nonetheless, a recent study showed that targeted quality improvement efforts led to a significant reduction of DTN times across a multicenter telestroke program. ²² This emphasizes the need for continuous quality improvement efforts at telestroke centers. Another distinguishing characteristic of our study is that the most common thrombolytic agent used among patients undergoing IVT was tenecteplase, potentially accelerating the DIDO times due to its ease of administration compared with alteplase. ²³

The median DIDO time in our study was 107 minutes, significantly longer than in the RACECAT and TRIAGE‐STROKE trials. However, a recent US nationwide analysis indicated that overall median DIDO times among patients eligible for MT was 132 minutes. ²⁴ While this study identified several demographic and geographic disparities in DIDO times, these delays can be multifactorial, including lack of prearrival notifications, inconsistencies in acute stroke imaging workflow, and delays in telestroke consultation and securing timely interfacility transportation, all contributing to prolonged DIDO times. ²⁵ A recent study showed that establishing a standardized transfer protocol can reduce median DIDO times to as low as 64 minutes. ²⁶ In our study, most LSC patients underwent protocol‐based interfacility air transport directly to the angiography suite without requiring reimaging at the TCC, evidenced by short median interfacility transport and door‐to‐puncture times. ¹¹ This workflow significantly accelerates the delivery of MT among LSC patients, another striking difference compared with the STRATIS registry, which excluded patients who underwent air interfacility transport, and a significant proportion underwent repeat imaging, as evidenced by prolongation of picture‐to‐puncture times. ⁵

Our study has several limitations inherent to its retrospective, observational nature. First, our cohort was limited to patients who ultimately underwent MT for LVO. As a result, patients who may have recanalized after receiving IVT at an LSC and did not require MT upon transfer to a TCC were excluded, and clinical outcome data in this group were not collected. This study design may underrepresent the potential clinical benefits among patients initially routed to an LSC, and any conclusions about outcomes should be interpreted with caution. Nevertheless, the objective of our study was to evaluate the impact of a prehospital transport strategy on the treatment time metrics, specifically to investigate if direct transport to a TCC was associated with delays in IVT, as observed in previous studies. In contrast, we found that direct transport to a TCC was associated with a modest acceleration of IVT administration (shorter FMC‐to‐IVT time) as well. Next, although the overall rate of PSSS acquisition was relatively low, it was higher among DTT compared with TLSC. Notably, the median prehospital transport time was longer in DTT despite a higher rate of use of air transport. Although we did not collect individual‐level data on the distance from the field to the TCC for each patient in TLSC, this pattern suggests that the majority of the EMS‐driven triage decisions were in accordance with statewide EMS transport protocols and/or guided by centralized medical direction that supports bypassing the LSC to prioritize transport to a TCC (provided this center can be reached within 45 minutes) among patients with a suspicion of an intracranial LVO. Next, although detailed information on individual triage decisions was not available, our findings highlight inconsistencies in the use of the PSSS, further underscoring the need for standardized education, systematic implementation, and consistent use of the PSSS in the field. Moreover, there was a substantial proportion of patients who were excluded from our analysis due to missing prehospital data. While we do not have information on these patients, we emphasize the need for more robust prehospital stroke data collection as a part of continuous quality improvement. In our study, we did not collect data on patients with intracranial hemorrhage or stroke mimics, which could confound the utility of EMS‐driven stroke center triage determination. Additionally, the EMS personnel did not routinely engage a VN specialist, an approach that was successfully adopted in the RACECAT and TRIAGE‐STROKE trials. Early engagement of VN with EMS personnel can be leveraged to accelerate treatment times both at the LSC and TCC and could also potentially minimize overtriaging of patients with stroke mimic. We did not restrict our population on the basis of the last‐known‐well time or location of the intracranial occlusion. However, the timeliness of MT has been proven across all time periods, and its utility has been demonstrated in a broad range of patients. ¹

Our findings highlight key differences in stroke care workflows compared with previous international studies, reflecting the unique challenges and strengths of US‐based systems. We identified delays in DTN and DIDO times but also noted strong prehospital triage and faster transport largely due to system integration and frequent air transfers. These results support adopting system‐specific transport strategies rather than a “one‐size‐fits‐all” model. Strengthening VN collaboration in the field and implementing real‐time dashboards for EMS could further improve triage decisions and protocol development. Despite the relatively modest sample size, our study provides pragmatic insights into the optimization of stroke systems of care within our health care setting, with potential applicability to similar systems across the United States.

Sources of Funding

None.

Disclosures

R.G.N. reports consulting fees for advisory roles with Anaconda, Biogen, Cerenovus, Genentech, Philips, Hybernia, Imperative Care, Medtronic, Phenox, Philips, Prolong Pharmaceuticals, Stryker Neurovascular, Shanghai Wallaby, and Synchron; and stock options for advisory roles with Astrocyte, Brainomix, Cerebrotech, Ceretrieve, Corindus Vascular Robotics, Vesalio, Viz‐AI, RapidPulse, and Perfuze. R.G.N. is one of the principal investigators of the ENDOLOW (Endovascular Therapy for Low National Institutes of Health Stroke Scale Ischemic Strokes () trial. R.G.N. is the principal investigator of the DUSK (Combined Thrombectomy for Distal Medium Vessel Occlusion Stroke) trial. R.G.N. is an investor in Viz‐AI, Perfuze, Cerebrotech, Reist/Q’Apel Medical, Truvic, Vastrax, and Viseon. All other authors have nothing to disclose.

Supporting information

Tables S1–S3

Figures S1–S3