Direct‐to‐Angio Without Any Repeat Neuroimaging for Planned Endovascular Therapy in Patients Transferred From Remote Primary Stroke Centers: A Propensity‐Matched Study.

Abstract

Background

Repeat imaging when regional and remote patients with stroke arrive at a comprehensive stroke center can delay endovascular thrombectomy. We examined outcomes amongs patients transferred for endovascular therapy from nonmetropolitan primary stroke centers.

Methods

In this prospective observational study patients who were transferred from remote nonmetropolitan hospitals with large vessel occlusion were recruited between 2020–2023. The control group was defined as patients with repeat neuroimaging at the comprehensive stroke center in the radiology/emergency department. Direct‐to‐angio (direct to angiography) included patients who proceeded directly to the angiography suite without any repeat neuroimaging or routine flat panel computed tomography. Logistic regression with propensity matching was performed to assess factors associated with 3‐month independent outcome (modified Rankin scale score 0–2). A secondary analysis was performed to assess factors associated with recanalization.

Results

Between June 2020 and February 2023, 227 patients with large vessel occlusion were transferred for endovascular clot retrieval. A total of 47 (26%) patients recanalized by time of arrival and 180 had persistent large vessel occlusion. Primary stroke centers were a median distance of 185 km from the comprehensive stroke center (interquartile range 130–256). After propensity matching 138 patients remained. Characteristics between direct‐to‐angio and control groups were similar with regard to age (68 versus 66), primary stroke centers National Institutes of Health Stroke Scale score (15 versus 13), and onset to referral (200 versus 246 min). Direct‐to‐angio increased independent functional outcome at 3 months (adjusted odds ratio, 2.2 [95% CI, 1.0–4.9]; P = 0.05). Direct‐to‐angio resulted in shorter door‐to arterial puncture time (43 versus 77 min, P<0.001) and a higher likelihood of receiving endovascular therapy (100% versus 65%, P<0.001).

Conclusion

In patients who have already received telestroke review at the primary stroke centers, our results suggest a direct‐to‐angio approach is likely to result in better outcomes.

Nonstandard Abbreviations and Acronyms

ASPECTS

Alberta Stroke Program Early CT Score

CSC

comprehensive stroke center

CT

computed tomography

Direct‐to‐angio

direct to angiography

DSA

digital subtraction angiography

DT3

delay time < 3 seconds

EVT

endovascular thrombectomy

LVO

large vessel occlusion

NIHSS

National Institutes of Health Stroke Scale

PSC

primary stroke center

Clinical Perspective

What Is New?

• There is accumulating evidence that the direct‐to‐angio strategy can improve outcomes for patients with large vessel occlusion. This research provides a unique perspective into how a direct‐to‐angio strategy modulated care and outcome among patients referred as part of a remote telestroke network in Victoria, Australia.

• Direct‐to‐angio increased independent functional outcome at 3 months. Direct‐to‐angio resulted in shorter door to arterial puncture time and a higher likelihood of receiving endovascular therapy.

What Are the Clinical Implications?

• Adoption of a direct‐to‐angio strategy will likely result in better outcomes among patients with large vessel occlusion who were already assessed via a telestroke network.

Patients from remote locations with stroke and large vessel occlusion experience unique challenges. This is, in part, due to longer delays in receiving reperfusion therapy. ¹ For each hour reperfusion is delayed there is a 5%–10% decreased likelihood of achieving independent outcome for patients with large vessel occlusion (LVO). ² Patients who present to nonurban hospitals are also less likely to receive key therapeutic interventions. ¹ Systems of care are therefore vital to ensure that all regions are provisioned, with models including direct bypass to the comprehensive center ³ , ⁴ , ⁵ and drip‐and‐ship. ⁶ , ⁷ The drip‐and‐ship model involves initial imaging and medical therapy at the primary stroke center (PSC), with transfer to the comprehensive center (CSC) if endovascular therapy (EVT) is necessary. Telemedicine may also optimize the administration of thrombolytics and early referral and transfer. ⁸ Even when systems are well optimized, patients who are transferred long distances may still experience deterioration during transport.

After long transport patients may have repeat imaging performed in usual radiology units, on‐table flat‐panel computed tomography (CT) in the angiography suite, or no repeat imaging at all. Due to the importance of early reperfusion, it is vital that systems are in place at the CSC that minimize delay to EVT. The direct‐to‐angiography (direct‐to‐angio) strategy presents nuanced clinical advantages and limitations. Although this approach can significantly reduce door‐to‐groin times and improve functional outcomes, ⁹ , ¹⁰ , ¹¹ , ¹² , ¹³ resource demands are significantly increased. The mandatory digital subtraction angiography (DSA) protocol inherent in the direct‐to‐angio strategy also subjects all patients to invasive imaging, including those who may have experienced spontaneous recanalization during transport. With this in mind, the ANGIOCAT trial (Evaluation of Direct Transfer to the Angiography Suite vs, CT Suite in Endovascular Treatment: Randomized Clinical Trial) demonstrated that procedural complications remain rare among patients with “false‐positive” activations, mitigating potential concerns about the safety of unnecessary interventions. ¹²

The clinical profile and diagnostic certainty of transferred patients differ from those presenting directly to the CSC. In the transfer cohort, LVO has already been confirmed through initial neuroimaging, and therapeutic interventions such as systemic thrombolysis may have been initiated at the PSC. Two randomized controlled trials pooled mothership and transferred patients in their comparison ¹² , ¹³ with other evidence being largely observational. ¹⁴ , ¹⁵ , ¹⁶ , ¹⁷ , ¹⁸ Meta‐analysis showed mixed results for an improvement in functional outcome with a direct‐to‐angio protocol in the transferred population. ¹⁰ , ¹¹ , ¹³ Heterogeneity of study designs and patient population was a significant limitation. The benefits of direct‐to‐angio may be attenuated in patients with prolonged transfer times, ¹⁸ although this remains an area of ongoing investigation. Furthermore, there remains a lack of consensus guidelines regarding patient selection for direct‐to‐angio approaches in the transfer population. ¹⁹ , ²⁰

The resource‐intensive nature of direct‐to‐angio protocols and the challenges in predicting vessel recanalization highlights the need for optimized workflows that can maintain efficiency while accommodating unique patient scenarios. Hartog et al ²¹ highlight that although door‐to‐puncture times show minimal variation between hospitals, there is substantial interpatient variability within individual centers. This suggests that standardized in‐hospital protocols could help achieve more consistent treatment times across all patients. Prediction of recanalization on arrival may be inaccurate and contribute to significant interpatient variability. Current prognostic models demonstrate minimal spontaneous recanalization rates in internal carotid artery occlusions) ²² ; however, prediction of middle cerebral artery vessel recanalization is more difficult. A single center analysis was able to correctly identify 64% of patients with recanalization using a one‐third intertransport National Institutes of Health Stroke Scale (NIHSS) improvement. ²³ Distal occlusions, ²⁴ , ²⁵ shorter thrombi, ²⁴ and clinical improvement have been associated with recanalization during transport.

Our aim is to evaluate outcomes associated with imaging protocol in patients with confirmed LVO who were transferred from remote PSCs. We will compare 2 groups: patients who underwent repeat imaging upon arrival and those who proceeded directly to the angiography suite without additional neuroimaging or routine flat panel CT.

Methods

The data in this study are available from the corresponding author on reasonable request and with appropriate ethical approval.

Patients from remote PSCs with LVO who were referred for EVT were recorded between June 2020 and February 2023. The CSC providing EVT was Royal Melbourne Hospital Victoria and PSCs were 14 referral hospitals across Victoria and Tasmania. Patients were referred by stroke physicians after review within the statewide telehealth network. Multimodal stroke imaging was performed at the primary stroke center (CT brain, CT angiography, CT perfusion) as part of the telehealth consult. All management of patients was at the discretion of the treating physicians.

All transferred patients were reviewed for whether they received repeat neuroimaging (CT with or without CT angiography or perfusion). The control group was defined as patients with repeat neuroimaging at the CSC in the radiology/emergency department. Direct‐to‐angio included patients who proceeded directly to the angiography suite without any repeat neuroimaging or routine flat panel CT. Clinical and imaging data were extracted from the MOSES (The Monitoring of Stroke Endovascular Services) study registry. This is an ongoing registry of patients being considered for EVT at the study center. Baseline clinical data recorded include, age, sex, NIHSS score, treatments received, premorbid modified Rankin scale score, ²⁶ and time from symptom onset. Three‐month outcome and safety outcomes (hemorrhage and mortality) were also collected. All data were collected in compliance with local ethics committee and institutional guidelines and patients provided consent for their data to be collected.

CT perfusion was analyzed by commercial software MIStar (Apollo Medical Imaging Technology, Melbourne, Australia). Automated maps were generated for delay time (a delay and dispersion corrected Tmax), cerebral blood flow, cerebral blood volume, and mean transit time. A delay time of >3 seconds was set as the severe hypoperfusion threshold ²⁷ and relative cerebral blood flow <30% for the ischemic core. The CT scanner and imaging protocol varied from site to site. Alberta Stroke Program Early CT Score (ASPECTS) was scored by the primary author (M.V.). Recanalization was defined as either partial or complete recanalization confirmed on repeat CT or as clot migration on DSA (treatment in cerebral ischemia 2b or above), that resulted in aborting the planned EVT. Patients with clinically presumed (documented without angiographic imaging) were also not included in the primary analysis. Safety outcomes included any intracerebral hemorrhage, symptomatic intracerebral hemorrhage, ²⁸ and death.

Statistics

Baseline patient characteristics were summarized using medians and interquartile range (IQR). In‐table comparisons were performed using χ2 test, Mann–Whitney U‐test, or analysis of variance as appropriate. Statistics were performed using SPSS (Version 29, IBM statistics). For the binomial regression analysis performed, the linearity assumption was tested via a Box‐Tidwell procedure with Bonferroni correction applied. Continuous variables were assessed for linearity to the logit of the dependent variable. Sample size was determined by comparing effect sizes reported in previous research. ¹⁴ , ¹⁵ , ¹⁶ There were no missing data for any of the variables included in the analysis.

The primary analysis related to factors associated with 3‐month independent outcome (modified Rankin scale score 0–2) among patients transferred for consideration of EVT from remote sites (PSCs). To control for significant group differences, propensity‐based matching was performed to correct for the variables associated with repeat imaging. Binomial logistic regression was used with the dependent variable set as functional independence at 3 months (modified Rankin scale score 0–2). Independent variables included age, sex, distance traveled, onset to CSC referral, lysis administration, PSC NIHSS score, PSC ASPECTS, and whether direct to EVT was performed. To reduce the heterogeneity of the direct‐to‐EVT group, analysis was performed with propensity matching. A sensitivity analysis was performed to assess how the inclusion of recanalized patients and the exclusion of posterior circulation LVO influenced the primary outcome measure.

Propensity matching was performed using calipers of width equal to 0.2 of the pooled SD of the logit of the propensity score. Independent variables included in the propensity score were age, distance traveled, onset to CSC referral, lysis administration, PSC NIHSS score, and PSC ASPECTS. Patients were then matched with their nearest neighbor using an algorithm. In addition, we assessed the association between time from PSC imaging to CSC arterial puncture and clinical outcome. A probability function was plotted in RStudio using a binomial model with time taken from PSC imaging to CSC arterial puncture set as the independent variable.

A secondary analysis was performed to assess factors associated with recanalization among patients transferred for consideration of EVT from remote sites. Binomial logistic regression was utilized with recanalization as the dependent variable. Independent variables included thrombolysis, time from stroke onset, sex, age, distance traveled, time traveled, occlusion location, NIHSS score percentage change, and age. Occlusion location was entered as a categorical variable, with 3 levels: internal carotid artery/tandem/basilar, first segment middle cerebral artery, and second segment middle cerebral artery/posterior cerebral artery.

Results

Between June 2020 and Feb 2023, 227 patients with LVO were transferred for EVT. A total of 47 (26%) patients recanalized by time of arrival and 180 had persistent LVO. The median age of remote transfers was 69 (IQR 58–79). Median PSC NIHSS score was 14 (IQR 8–18) and time from onset was 209 minutes (IQR 130–428). PSCs were a median distance of 185 km from the CSC (IQR 130–256).

A total of 93 patients went directly to the endovascular suite without repeat imaging and 87 received repeat imaging. A study outline is shown in Figure 1. Patients who went directly to EVT had a shorter time from onset to referral (270 versus 187 min), were referred from closer PSCs (226 km versus 161 km), and had higher PSC NIHSS scores (12 versus 16) (Table S1). After propensity matching (using the defined criteria), 69 patients were able to be matched in each group. Matched groups were similar with regard to clinical and imaging variables except for distance from CSC (Table 1). In the direct group, 2 patients underwent flat panel CT imaging, which did not alter the decision to proceed to EVT.

Figure 1.

Study summary. CTA indicates computed tomography angiography; DSA, digital subtraction angiography; EVT, endovascular therapy; LVO, large vessel occlusion; and mRS, modified Rankin scale.

Table 1.

Clinical and Imaging Variables

	Repeat imaging (N = 69)	Direct to EVT (N = 69)	P value
Age, y (median, IQR)	66 (54–78)	68 (57–80)	0.430
Male sex, n (%)	40 (58)	39 (57)	0.845
NIHSS score at onset, median (IQR)	13 (9–18)	15 (11–18)	0.072
NIHSS score change, median (IQR)	0 (‐3,2)	0 (‐2,5)	0.258
Onset to referral, min, median (IQR)	246 (145–550)	200 (137–358)	0.149
Distance, km, median (IQR)	226 (130–258)	161 (130‐226)	0.022
Transport time, min, median (IQR)	192 (152–257)	180 (149‐223)	0.095
Afterhours CSC arrival, N (%)	30 (43)	41 (59)	0.061
Premorbid mRS score 0—1, n (%)	63 (91)	66 (96)	0.301
Thrombolysis, n (%)	34 (49)	35 (50)	0.865
Tenecteplase, n (%)	26 (38)	27 (39)	0.861
EVT, n (%)	45 (65)	69 (100)	<0.001
Occlusion site, n (%)			0.908
M1	23	35
M2	13	4
ICA	16	11
Tandem	11	9
Basilar	5	9
PCA	1	1
Remote PSC CT
ASPECTS, median (IQR)	9 (7–10)	8 (7–10)	0.496
Ischemic core, ml, median (IQR)	15 (3–42)	18 (5–32)	0.341
DT3 volume, ml, median (IQR)	80 (50–118)	100 (60–125)	0.341
Repeat CSC CT
ASPECTS, median (IQR)	7 (4–10)	n/a	n/a
Ischemic core, ml, median (IQR)	26 (3–94)
DT3 volume, ml, median (IQR)	112 (66–178)

Propensity matched repeat imaging vs direct to EVT.

ASPECTS indicates Alberta Stroke Program Early CT Score; CSC, comprehensive stroke center; CT, computed tomography; DT3, delay time > 3 seconds; EVT, endovascular therapy; ICA, internal carotid artery; IQR, interquartile range; M1, first segment middle cerebral artery; M2, second segment middle cerebral artery; mRS, modified Rankin Scale; NIHSS, National Institutes of Health Stroke Scale Score; PCA, posterior cerebral artery; and PSC, primary stroke center.

Of the 69 matched patients who had repeat imaging, 24 did not receive EVT. The documented reason for EVT abandonment was established as infarct in 23 cases and hemorrhage in 1 case. Imaging characteristics of patients with LVO who did not receive EVT are summarized in Table S2.

Factors Associated With Independent Outcome (modified Rankin Scale Score 0–2)

Patients who were taken direct‐to‐angio had a higher likelihood of functional independence (modified Rankin scale score 0–2) at 3 months (48% versus 33%; adjusted odds ratio, 2.2 [95% CI, 1.0–4.9]; P = 0.05). Modified Rankin Scale score outcomes for patients who had repeat imaging versus a direct‐to‐angio strategy are summarized in Figure 2. Age, distance, and high PSC ASPECTS were significantly associated with functional outcomes. PSC NIHSS score, lysis, and time from onset were not significantly associated with outcome. Odds ratios for each independent variable are summarized in Figure 3. There were no standardized residuals above 3 SDs and all cases were kept in the analysis. Patients who underwent repeat imaging had longer CSC door to arterial puncture time (43 versus 77 min, P<0.01) and similar rates of revascularization post EVT (Table 2). The relationship between outcome and time and time to EVT commencement has been summarized in Figure 4. The included sensitivity analyses were not able to achieve a statistically significant result (Tables S6 and S7)

Figure 2.

Propensity matched repeat imaging versus direct to EVT; mRS scores 3 months. Distribution of the scores of the modified Rankin Scale at 3 months (propensity analysis). Patients who were taken directly to EVT had a higher likelihood of functional independence (mRS score 0–2) at 3 months (48% vs 33%; adjusted OR, 2.2 [95% CI, 1.0–4.9]; P = 0.05). Groups matched using propensity score (age, distance traveled, onset to CSC referral, lysis administration, PSC NIHSS score, and PSC ASPECTS). ASPECTS indicates Alberta Stroke Program Early CT Score; CSC, comprehensive stroke center; EVT, endovascular therapy; mRS, modified Rankin Scale; NIHSS, National Institutes of Health Stroke Scale; OR, odds ratio; and PSC, primary stroke center.

Figure 3.

Forest plot. Binomial logistic regression: independent outcome (mRS score 0–2). Forest plot demonstrating odds ratio for independent variables predicting independent outcome (mRS score 0–2). Higher odds indicate a higher likelihood of functional independence. ASPECTS indicates Alberta Stroke Program Early CT Score; CSC, comprehensive stroke center; EVT, endovascular therapy; mRS, modified Rankin Scale; NIHSS, National Institutes of Health Stroke Scale; and PSC, primary stroke center.

Table 2.

Secondary Outcomes and Safety

	Repeat imaging (N = 69)	Direct to EVT (N = 69)	P value
Door to puncture (median, IQR)	77 (53–118)	43 (35–50)	<0.001
TICI 2b or above (n, %)	38 (84)	60 (87)	0.706
Any intracerebral hemorrhage (n, %)	10 (14)	13 (19)	0.493
Symptomatic intracerebral hemorrhage (n, %)	4 (6)	6 (8)	0.431
Mortality (n, %)	17 (25)	15 (22)	0.687

Propensity matched repeat imaging vs direct to EVT.

EVT indicates endovascular therapy; IQR, interquartile range; and TICI 2b or above, treatment in cerebral ischemia 2b or above >50% reperfusion of the involved territory.

Figure 4.

Probability function. Likelihood of independent function (mRS score 0–2) at 3 months compared to time from PSC imaging to CSC arterial puncture. Probability function using a binomial model for independent function (mRS score 0–2) with time taken from PSC imaging to CSC arterial puncture set as the independent variable. CSC indicates comprehensive stroke center; mRS, modified Rankin Scale; and PSC, primary stroke center.

Safety Outcomes

Safety measures such as hemorrhage and mortality were similar between groups (Table 2). Although patients who recanalized were not included in the primary analysis, adverse events related to DSA were also recorded for this group. Ten of 47 patients with recanalization received DSA on arrival. None of these patients experienced symptomatic hemorrhage. As a precaution, 1 patient remained intubated for an extended period after DSA due to concerns regarding an airway injury during intubation. After observation, no injury was observed, and the patient was successfully extubated without complication. Two patients had small amounts of bleeding observed from the groin access site that resolved spontaneously without further intervention. No post hoc adjustments were performed for safety outcomes.

Factors Associated With Recanalization on Arrival

A total of 47 patients had recanalization during transport. Patients who recanalized had lower NIHSS scores (median 9, IQR 6–14) and were more likely to receive thrombolysis (81% versus 51%, P<0.01). With regard to occlusion type, second segment middle cerebral artery occlusion was more likely to recanalize and there was a trend to less recanalization with ICA and tandem occlusion (Table S4). Ten patients with recanalization were identified on DSA after going direct to the angiography lab.

A logistic regression model was performed to assess the effects of age, distance traveled, time traveled, time from onset, occlusion location, and thrombolysis, on the likelihood of recanalization during transfer. The regression model explained 28% of the variance in recanalization and was statistically significant, χ ² (9) = 44, P<0.001. The model correctly classified 81.1% of cases. The area under the receiver operating characteristic curve was 0.792 (95% CI, 0.722–0.863) (Figure S2). Thrombolysis and occlusion location were the only statistically significant independent variables. Thrombolysis (odds ratio = 4.6, P<0.01) first segment middle cerebral artery occlusion (odds ratio = 7.1, P<0.01), and second segment middle cerebral artery/posterior cerebral artery occlusion (odds ratio = 13.7, P<0.01) were associated with a higher likelihood of recanalization. Variable odds are summarized in Figure S1.

Discussion

Our major finding was markedly increased odds of independent functional outcome at 3 months in patients with LVO who were taken direct‐to‐angio after transfer from PSC. Patients who were transferred direct‐to‐angio had a shorter door‐to‐arterial puncture time (43 versus 77 min) and were more likely to ultimately receive EVT (100% versus 65%). These findings suggest that repeat imaging and essentially repeat triage after arrival at the CSC is likely to result in poorer overall patient outcomes. Delay to recanalization resulted in approximately a 5%–10% reduction in the chance of independent function per hour. Safety outcomes were otherwise similar between each group.

Poorer outcomes in patients with repeat imaging were likely due to a combination of both delayed reperfusion and reduced EVT utilization. The main reason cited for the abandonment of EVT was established ischemic change. Based on the median imaging characteristics of those who did not receive EVT (ASPECTS of 4 and ischemic core of 100 mL), these patients would have fallen outside of guidelines based on DAWN (Clinical Mismatch in the Triage of Wake Up and Late Presenting Strokes Undergoing Neurointervention With Trevo) ²⁹ and DEFUSE3 (Endovascular Therapy Following Imaging Evaluation for Ischemic Stroke 3) ³⁰ criteria. One explanation for the worse outcomes in the repeat‐imaging group is that even these patients with larger ischemic cores may have benefited from EVT. Indeed, multiple large core trials ³¹ , ³² , ³³ , ³⁴ , ³⁵ , ³⁶ have demonstrated that EVT improves outcome for patients with larger areas of ischemic injury.

Many retrospective studies of direct‐to‐angio have demonstrated improved time‐to‐reperfusion ¹⁴ , ¹⁵ , ¹⁶ , ¹⁷ and functional outcomes. ¹⁵ , ¹⁸ ANGIOCAT ¹² was a randomized trial comparing direct‐to‐angio; however, it also included patients presenting primarily to the CSC and was restricted to 6 hours from onset. Comparing trials is difficult due to the significant heterogeneity of inclusion criteria, transferred distances, and the statistical methods used for the correction of group differences. Our center differs significantly from previous research due to longer onset to CSC arrival (409 min median) and more widespread use of CT perfusion triage at the PSC. Patients had excellent CT perfusion profiles at the PSC (Ischemic core median 16.5 mL) and high ASPECTS (8–9). This needs to be considered when accounting for the safety and benefits of a direct‐to‐angio approach in other healthcare systems.

Propensity‐based matching was chosen to avoid noncausal (spurious) observations that may arise from using weights to correct for the time‐dependent differences. ³⁷ Using our defined approach, we achieved a reasonable balance between the measured group differences. The remaining differences (such as higher NIHSS scores and more proximal middle cerebral artery occlusion) in the direct‐to‐angio group, in theory, should have favored better outcomes in those who underwent repeat imaging. Patients who could not be matched in the repeat imaging group had milder stroke syndromes with lower NIHSS scores, whereas those who were not matched in the direct‐to‐angio group had shorter transfer times from the PSC (Table S1). It is important to acknowledge these differences when applying our results, and further analysis may be necessary to assess a direct‐to‐angio approach in patients with lower NIHSS scores.

Repeat imaging can also be performed to assess whether recanalization has occurred and if EVT can be abandoned. Patients with distal occlusion and who had received thrombolysis therapy were the most likely to recanalize before CSC arrival. Prediction of recanalization is difficult, and it is not possible to predict which patients are more likely to recanalize with accuracy. Although our recanalization model resulted in an area under the curve of 0.79, this requires external validation. Even at this level of discrimination, applying the model in the clinical setting would result in an excessive number of patients with persistent occlusion receiving repeat imaging. When examining the receiver operating characteristic curve (Figure S2), optimizing a prediction model to identify 80% of patients who recanalize would result in falsely predicting recanalization in 40% of patients with occlusion. NIHSS score improvement (>2/3) was also not significantly associated with recanalization; however, these patients could possibly be excluded from EVT based on their improvement alone rather than any imaging evidence.

The main limitation of the study is the observational design and single network of hub and spoke sites. Without a randomized approach, it is not possible to assess for unmeasured variables that physicians may have considered when deciding whether patients required repeat imaging. Propensity matching as well as prospective ascertainment were used to mitigate these disadvantages. Being a single‐center study, the patients transferred had excellent favorable imaging with high ASPECTS scores and high rates of thrombolysis use at the PSC. Our patient cohort reflects practice prior to the publication of large core EVT trials. ³¹ Notably, patients transferred with lower ASPECTS are more likely to deteriorate during transportation, likely due to faster core growth. ³⁸ , ³⁹ , ⁴⁰ , ⁴¹ , ⁴² We had relatively few patients in our cohort.

Furthermore, the randomization of patients would facilitate the evaluation of clinical outcomes in individuals undergoing direct‐to‐angio who received diagnostic DSA without subsequent EVT. Our sensitivity analysis incorporating these patients did not reach statistical significance (Table S6). Future results from ongoing randomized trials such as DIRECT ANGIO (Effect of DIRECT Transfer to Angiosuite on Functional Outcome in Severe Acute Stroke) and WE‐TRUST (Workflow Optimization to Reduce Time to Endovascular Reperfusion for Ultra‐Fast Stroke Treatment) are awaited ⁴³ to clarify these outcomes.

Conclusion

Repeat imaging, and thus repeat triage, after arrival at CSC is likely to result in poorer overall patient outcomes. Direct‐to‐angio resulted in 2.5 times increased odds for independent functional outcome at 3 months. Patients had a shorter door‐to‐groin time (43 versus 77 min) and were more likely to ultimately receive EVT (100% versus 65%). Precise prediction of successful early recanalization is difficult and likely to result in unnecessary repeat imaging; however, it is associated with thrombolysis and more distal occlusion location. Although our findings indicate that a direct‐to‐angio approach may offer superior outcomes, additional randomized controlled trials are awaited to establish a definitive clinical recommendation.

Conflict of Interest Statement

None declared.

Sources of Funding

Australian Government Research Training Program Scholarship funded PhD research of the first author.

Supporting information

Table S1: (Unmatched) Repeat imaging vs direct to EVT. Clinical and imaging variables.

Table S2: CSC Imaging variables in participants with EVT abandonment (propensitymatched group).

Table S3: (Unmatched) Repeat imaging vs direct to EVT. Outcomes.

Table S4: Clinical variables recanalized on arrival

Table S5: Imaging variables recanalized on arrival

Table S6: Sensitivity Analysis‐Propensity analysis with recanalization patients included.

Table S7: Sensitivity Analysis‐Propensity analysis with posterior circulation occlusion removed.

Figure S1: Forest plot. Binomial Logistic regression: recanalization prediction

Figure S2: ROC curve: recanalization model