Impact and determinants of door in–door out time for stroke thrombectomy transfers in a large hub-and-spoke network

Abstract

Introduction

The mantra “time is brain” cannot be overstated for patients suffering from acute ischemic stroke. This is especially true for those with large vessel occlusions (LVOs) requiring transfer to an endovascular thrombectomy (EVT) capable center. We sought to evaluate the spoke hospital door in–door out (DIDO) times for patients transferred to our hub center for EVT.

Methods

Individuals who first presented with LVO to a spoke hospital and were then transferred to the hub for EVT were retrospectively identified from a prospectively maintained database from January 2019 to November 2022. DIDO was defined as the time between spoke hospital door in arrival and door out exit. Baseline characteristics, treatments, and outcomes were compared, dichotomizing DIDO at 90 minutes based in the American Heart Association goal for DIDO ≤90 minutes for 50% of transfers. Multivariable regression analyses were performed for determinants of the 90-day ordinal modified Rankin Scale (mRS) and DIDO.

Results

We identified 194 patients transferred for EVT with available DIDO. The median age was 67 years (IQR 57–80), and 46% were female. The median National Institutes of Health Stroke Scale (NIHSS) was 16 (10–20), 50% were treated with intravenous thrombolysis at a spoke, and TICI 2B-3 reperfusion was achieved in 87% at the hub. The median DIDO was 120 minutes (97–149), with DIDO ≤90 minutes achieved in 18%. DIDO was a significant determinant of 90-day ordinal mRS (B = 0.007, 95% CI = 0.001–0.012, p = 0.013), even when accounting for the last known well-to-spoke door in, spoke door out-to-hub arrival, hub arrival-to-puncture, puncture-to-first pass, age, NIHSS, intravenous thrombolysis, TICI 2B-3, and symptomatic intracranial hemorrhage. Importantly, determinants of DIDO included Black race or Hispanic ethnicity (B = 0.918, 95% CI = 0.010–1.826, p = 0.048), atrial fibrillation or heart failure (B = 0.793, 95% CI = 0.257–1.329, p = 0.004), and basilar LVO location (B = 2.528, 95% CI = 1.154–3.901, p < 0.001).

Conclusion

Spoke DIDO was the most important period of time for long-term outcomes of LVO stroke patients treated with EVT. Targets were identified to reduce DIDO and improve patient outcomes.

Introduction

The established benefits of reperfusion therapies, notably intravenous thrombolysis (IVT) and endovascular thrombectomy (EVT), have been extensively documented in the literature.^1–5 Early EVT with medical therapy, compared to medical therapy alone, led to less disability at 3 months, especially when performed within 2 hours of symptom onset.^4,6,7 Prolonged door in–door out (DIDO) times may have a deleterious effect on outcomes. ⁸ Indeed, the effectiveness of reperfusion therapies exhibits a downward trend as the elapsed time since the onset of symptoms increases.^1,2,4,5 Importantly, DIDO times are frequently prolonged for acute ischemic stroke patients eligible for endovascular therapy.^3,8–12

Despite the growing interest in targeting this time epoch for improvement, a recent US registry study reported an average DIDO duration of 174 minutes.^12–15 Furthermore, a multitude of factors may influence DIDO times. ¹⁶ In this study we sought to evaluate the spoke hospital DIDO times in our own large hub-and-spoke network to evaluate and report our real-world experience. Since every region is different, we also aimed to evaluate the relationship of DIDO and outcomes in our system and identify independent determinants of DIDO to identify targets for improvement.

Methods

This study was compliant with the Health Insurance Portability and Accountability Act and was reviewed and approved by the hub site institutional review board. Informed consent was waived based on minimal patient risk and practical inability to perform the study without the waiver. ¹⁷ The data that support the findings of this study will be made available from the corresponding author upon reasonable request and pending approval of the local institutional review board.

Individuals who first presented with large vessel occlusion (LVO) to a spoke hospital and were then transferred to the hub for EVT were retrospectively identified from a prospectively maintained database from January 2019 to November 2022. Our granular database includes demographic characteristics, presentation details, imaging findings, medical history, treatment details, and long-term outcomes. The local hub-and-spoke system accepted transfers from 32 spoke hospitals during the time of the study.^18–20 DIDO was defined as the time between spoke hospital door in arrival and door out exit and was collected for consecutive patients.

National Institutes of Health Stroke Scale (NIHSS) score was determined by the hub neurologist, with higher numbers reflecting increased clinical stroke severity. ¹⁴ All patients underwent head computed tomography (CT) and head and neck CT angiography (CTA) at spoke hospitals before transfer.^21,22 Alberta Stroke Program Early CT score (ASPECTS) and presence of LVO on CTA were determined by a vascular neurologist and confirmed by a neuroradiologist.^23,24 LVO was defined as occlusion of the internal carotid artery (ICA) terminus, first (M1) and proximal second (M2) segments of the middle cerebral artery, or the basilar artery.^25,26 Cervical ICA disease was defined as severe stenosis (>70%) or occlusion related to atherosclerosis or dissection.^27,28 IV thrombolysis treatment decisions at spokes were guideline-based at the discretion of a vascular neurologist through telemedicine.^29,30

EVT treatment decisions at the hub were at the discretion of a vascular neurologist and a neurointerventionalist.^31,32 Thrombolysis in Cerebral Infarction (TICI) scores were determined by a neurointerventionalist using the modified scale: 2a partial filling <50%, 2b partial filling ≥50%, 3 complete perfusion. ³³ Adequate reperfusion was defined as TICI 2B-3. ³⁴ Symptomatic intracerebral hemorrhage (sICH) was defined as any symptomatic intraparenchymal, intraventricular, or subarachnoid hemorrhage during hospitalization. Discharge modified Rankin Scale (mRS) was noted by the discharging vascular neurologist. 90-day mRS score was obtained by clinical follow-up or telephone call by certified assessors.^35,36

Median values with interquartile range (IQR) were reported for continuous and ordinal variables. Percent and count were reported for categorical variables. For our initial comparisons, DIDO was dichotomized at 90 minutes based in the American Heart Association goal for DIDO ≤90 minutes for 50% of transfers. Multivariable regression analyses were performed for determinants of 90-day ordinal mRS and continuous DIDO. Independent variables were selected for inclusion based on prior literature and hypothesized relationships based on clinical experience. Unadjusted and adjusted regression coefficients and p-values were reported. Two-tailed p-values <0.05 were interpreted as statistically significant. All statistics were performed with SPSS version 23.0 (IBM Corp).

Results

We identified 194 patients transferred for EVT with available DIDO. The median age was 67 years (IQR 57–80), and 46% were female. The racial composition was Black (4%), Asian (3%), and White (70%), and 5% had Hispanic ethnicity. Occlusions were of the ICA terminus (12%), M1 (59%), M2 (24%), and basilar (4%) arteries. The complete baseline characteristics are presented in Table 1.

Table 1.

Baseline characteristics of the study cohort and comparison between patients with DIDO ≤90 minutes and DIDO ≥90 minutes.

Variable	Total	DIDO ≤90	DIDO >90	p
Age	67 (57, 80)	65 (56, 75)	67 (58, 81)	0.220
Female sex	90 (46%)	15 (44%)	75 (47%)	0.460
Pre-stroke mRS 3–5	9 (5%)	2 (6%)	7 (4%)	0.478
Black	8 (4%)	0 (0%)	8 (5%)	0.207
Asian	5 (3%)	2 (6%)	3 (2%)	0.211
White	135 (70%)	23 (68%)	112 (70%)	0.467
Hispanic	10 (5%)	1 (3%)	9 (6%)	0.450
Hypertension	127 (66%)	19 (56%)	108 (68%)	0.137
Atrial fibrillation	59 (30%)	5 (15%)	54 (34%)	0.020
Coronary artery disease	35 (18%)	8 (24%)	27 (17%)	0.245
Carotid stenosis	4 (2%)	1 (3%)	3 (2%)	0.540
Diabetes mellitus	43 (22%)	7 (21%)	36 (23%)	0.505
Dyslipidemia	96 (50%)	21 (62%)	75 (47%)	0.082
Heart failure	24 (12%)	1 (3%)	23 (14%)	0.049
DVT/PE	9 (5%)	1 (3%)	8 (5%)	0.511
Obesity/overweight	110 (57%)	19 (56%)	91 (57%)	0.531
Previous stroke/TIA	32 (17%)	5 (15%)	27 (17%)	0.493
Chronic renal insufficiency	20 (10%)	3 (9%)	17 (11%)	0.521
Smoker	27 (14%)	6 (18%)	21 (13%)	0.325
Sleep apnea	10 (5%)	2 (6%)	8 (5%)	0.550
Migraine	11 (6%)	2 (6%)	9 (6%)	0.606
Dementia	3 (1.5%)	0 (0%)	3 (2%)	0.559
Depression	30 (16%)	6 (18%)	24 (15%)	0.434
ICA terminus occlusion	24 (12%)	6 (18%)	18 (11%)	0.223
M1 occlusion	115 (59%)	23 (68%)	92 (58%)	0.184
M2 occlusion	47 (24%)	4 (12%)	43 (27%)	0.044
Basilar occlusion	7 (4%)	0 (0%)	7 (4%)	0.254
Large artery atherosclerosis	27 (14%)	6 (18%)	21 (13%)	0.330
Cardioembolic	96 (50%)	12 (35%)	84 (53%)	0.047
Dissection	11 (6%)	3 (9%)	8 (5%)	0.302
Hypercoagulability	12 (6%)	2 (6%)	10 (6%)	0.644
Other (vasculopathy or hematologic)	3 (2%)	1 (3%)	2 (1%)	0.443
Cryptogenic	44 (23%)	10 (29%)	34 (21%)	0.213

The median LKW-to-spoke door in time was 90 minutes (IQR 45–323). At the spoke hospitals, the median NIHSS was 16 (IQR 10–20), and 50% were treated with intravenous thrombolysis. The median DIDO was 120 minutes (97–149), with DIDO ≤90 minutes achieved in 18%. At the hub, the median hub arrival-to-puncture time was 23 minutes (IQR 11–62). TICI 2B-3 reperfusion was achieved in 87%. sICH occurred in 7%. A discharge mRS 0–2 was achieved in 19%, and a 90-day mRS 0–2 was achieved in 48%. The complete details of times, treatments, and outcomes are presented in Table 2.

Table 2.

Times, treatments, and outcomes of the study cohort and comparison between patients with DIDO ≤90 minutes and DIDO ≥90 minutes.

Variable	Total	DIDO ≤90 minutes	DIDO >90 minutes	p
LKW-to-spoke door in	90 (45, 323)	55 (40, 203)	96 (46, 338)	0.217
Spoke door in-to-spoke door out	120 (97, 149)	78 (71, 84)	126 (108, 161)	<0.001
Spoke door out-to-hub arrival	39 (26, 58)	45 (32, 69)	35 (24, 56)	0.108
Hub arrival-to-puncture	23 (11, 62)	28 (12, 42)	22 (11, 65)	0.984
Hub puncture-to-first pass	19 (12, 28)	18 (11, 28)	19 (12, 28)	0.716
Telestroke consult	163 (84%)	29 (85%)	134 (84%)	0.529
NIHSS	16 (10, 20)	17 (14, 20)	16 (10, 20)	0.316
IVT at spoke	97 (50%)	20 (59%)	77 (48%)	0.173
TICI 2B-3	162 (87%)	32 (100%)	130 (84%)	0.008
sICH	13 (7%)	2 (6%)	11 (7%)	0.594
CMO during admission	31 (16%)	3 (9%)	28 (18%)	0.159
CMO Day 0 or 1	7 (4%)	1 (3%)	6 (4%)	0.645
Discharge mRS 0–2	37 (19%)	8 (24%)	29 (18%)	0.304
Discharge mRS 0–3	79 (41%)	20 (59%)	59 (37%)	0.015
Discharge home	44 (23%)	10 (29%)	34 (21%)	0.207
90-Day mRS 0–2	79 (48%)	17 (63%)	62 (45%)	0.066
90-Day mRS 0–3	102 (62%)	22 (82%)	80 (58%)	0.016

In multivariable modeling for determinants of 90-day ordinal mRS, DIDO was a significant independent determinant (B = 0.007, 95% CI = 0.001–0.012, p = 0.013). Other independent determinants of 90-day mRS included age (B = 0.058, 95% CI = 0.035–0.081, p < 0.001), NIHSS (B = 0.114, 95% CI = 0.064–0.164, p < 0.001), TICI 2B-3 (B = −1.442, 95% CI = −2.498–0.387, p = 0.007), and sICH (B = 1.770, 95% CI = 0.227–3.313, p = 0.025). The model further accounted for the last known well-to-spoke door in time, spoke door out-to-hub arrival time, hub arrival-to-puncture, puncture-to-first pass time, and intravenous thrombolysis (Table 3).

Table 3.

Determinants of 90-day mRS in univariable and multivariable modeling.

Variable	Univariable estimate	95% CI	p	Multivariable estimate	95% CI	p
LKW-to-spoke Door in	0.001	0.000, 0.002	0.091	0.000	−0.002, 0.001	0.949
Spoke door in-to-door out	0.006	0.001, 0.011	0.013	0.007	0.001, 0.012	0.013
Spoke door out-to-Hub arrival	0.009	0.000, 0.018	0.047	0.004	−0.007, 0.016	0.457
Hub arrival-to-puncture	0.003	−0.004, 0.009	0.384	−0.002	−0.011, 0.006	0.598
Hub puncture-to-first pass	0.016	−0.001, 0.033	0.062	0.009	−0.012, 0.030	0.411
Age	0.046	0.027, 0.065	<0.001	0.058	0.035, 0.081	<0.001
NIHSS	0.103	0.063, 0.143	<0.001	0.114	0.064, 0.164	<0.001
IVT at spoke	−0.814	−1.323, −0.306	0.002	−0.635	−1.357, 0.087	0.085
TICI 2B-3	−1.311	−2.115, −0.506	0.001	−1.442	−2.498, −0.387	0.007
sICH	1.867	0.591, 3.143	0.004	1.770	0.227, 3.313	0.025

In multivariable modeling for determinants of DIDO, Black race or Hispanic ethnicity (B = 0.918, 95% CI = 0.010–1.826, p = 0.048), atrial fibrillation or heart failure (B = 0.793, 95% CI = 0.257–1.329, p = 0.004), and basilar LVO location (B = 2.528, 95% CI = 1.154–3.901, p < 0.001) were all significant determinants (Table 4).

Table 4.

Determinants of spoke DIDO in univariable and multivariable modeling.

Variable	Univariable Estimate	95% Cl	P	Multivariable Estimate	95% Cl	P
Black or Hispanic	0.626	−0.264, 1.515	0.168	0.918	0.010, 1.826	0.048
Atrial fib or heart failure	0.508	−0.009, 1.024	0.054	0.793	0.257, 1.329	0.004
M2 occlusion location	0.323	−0.248, 0.893	0.267	0.546	−0.031, 1.123	0.064
Basilar occlusion location	2.090	0.747, 3.432	0.002	2.528	1.154, 3.901	<0.001

Discussion

In our study, only 18% of patients achieved DIDO ≤90 minutes, falling short of the American Heart Association goal of 50%. ³⁷ The median DIDO time was 120 minutes (IQR 97–149). This is similar to reports from other studies.^12,38–41 We found that DIDO was a significant determinant of 90-day ordinal mRS. Furthermore, we identified several determinants of DIDO that require further investigation.

A prior study demonstrated that transferring patients from another hospital is associated with elevated mortality rates in individuals undergoing EVT. ¹⁰ DIDO has been shown to be related to poor outcomes in another study of patients presenting with LVO. ⁸ These data corroborate our findings and stand to reason since delays from symptom onset have been associated with reduced efficacy of reperfusion therapies.^1,2,4,5 Importantly, even when accounting for other time periods and variables, prolonged DIDO was an independent determinant of worse outcomes in our cohort.

Black race or Hispanic ethnicity was noted to be a significant determinant of longer DIDO in our study. This observation was also made in our prior study examining delays in imaging. ⁴² Furthermore, racial and ethnic disparities were noted in a US registry-based study that evaluated DIDO times for interhospital stroke transfers. ¹² These data illustrate that disparities in stroke care and outcomes remain a major concern. There are multiple levels to address to understand this health crisis fully. Health literacy, treatment adherence, mistrust, and patient preferences all fall under patient-level factors.^43,44 Provider-level factors, including but not limited to unconscious bias, lack of representation of minority providers, and prejudice, may all contribute to disparities in stroke care. Finally, cultural/societal norms, health insurance, allocation of resources, and policies may lead to systemic racism. ⁴⁵ In a study using the Nationwide Inpatient Sample between 2006 and 2016, it was noted that Black patients may be less likely to receive EVT. ⁴⁶ Despite the challenges we face with this crisis, a recent study based on Get With The Guidelines-Stroke database (2012–2019) did show that disparities in EVT use between Black versus non-Hispanic whites have narrowed over time. ⁴⁷

Basilar LVO was another independent determinant of prolonged DIDO identified in our study. Diagnosing basilar LVO is challenging.^48,49 Delayed ED presentation and missed diagnosis may be attributed to a lack of sudden symptom onset and atypical stroke symptoms. ⁵⁰ Prehospital stroke screening tests, such as the Face Arm Speech Test (FAST), have been shown to miss 40% of strokes involving the posterior circulation compared to 10% in the anterior circulation. ⁵¹ The addition of “B” for balance and “E” for eyes to FAST has been shown to reduce the proportion of missed posterior circulation strokes. ⁵² The NIHSS also tends to underestimate the severity of posterior circulation strokes. To improve prognostic accuracy, the posterior NIHSS (POST-NIHSS) was developed for patients presenting with mild or moderate posterior circulation symptoms. This scale has been shown to have a higher prognostic accuracy than the traditional NIHSS but is not widely used in current clinical practice. ⁵³

In a meta-analysis that assessed the efficacy of EVT in anterior circulation LVOs, 33% of patients had known atrial fibrillation at the time of presentation. 54 out of 59 patients who had atrial fibrillation in our study were noted to have a DIDO >90 minutes. This may be due to the additional time it takes to stabilize a patient in rapid ventricular response before transfer.^54,55 On the topic of medical management delays, it is also worth noting that intravenous thrombolysis did not prolong DIDO times, consistent with prior studies.^12,56 This provides further support that thrombolysis should not be withheld from patients who require transfer for EVT.

Other independent predictors of extended DIDO have been reported in the literature and include greater age, symptom onset-to-door in time exceeding 6 hours, right-sided occlusions, and M2-occlusions. ⁵ Additionally, non-stroke certified centers were found to have longer DIDO times. ⁴¹ While our own prior research supports the utility of routinely obtaining CTA at spoke hospitals, prior research has shown that the time from CTA to ambulance notification is a primary contributor to longer DIDO times. This can potentially be modified through process and workflow improvements in the spoke hospital ED.^3,5,57 CTA added nearly 40 minutes to DIDO in a study of 191 patients; interestingly bundling CT with CTA did result in reduced time.^3,41

Optimizing DIDO time starts in the pre-hospital setting, with emergency medical services playing a key role. Defined algorithms and protocols for rapid dispatch, clinical assessment, and transport strategies are essential. Rapid assessment and imaging on arrival to the spoke emergency department, with fast administration of intravenous thrombolysis when appropriate, are also goals to improve DIDO. Furthermore, early communication with the receiving thrombectomy capable center and expedited transport handoffs are all helpful.^37,58 The use of artificial intelligence and other novel technologies has also been associated with improved DIDO. Software can be utilized to automatically identify suspected LVO through CTA imaging and alert on-call stroke teams, reducing patient transfer times and resulting in earlier initiation of EVT.^59–61

This study has several limitations. It is firstly a retrospective study, which may increase the risk of selection bias. The generalizability of the results may also be limited as study participants were all treated in a hub-and-spoke network in the New England region. However, each region is different, and individual studies are needed to identify goals for process improvement. There is also an underrepresentation of minority racial/ethnic groups based on the local population, with 70% of the study participants being White. The proportion of basilar LVOs in the study may also limit definitive conclusions.

Conclusion

In this real-world analysis of a large hub-and-spoke network, spoke DIDO was the most important period of time for long-term outcomes of LVO stroke patients treated with EVT. Determinants of DIDO were identified which require further investigation and motivate system changes that may reduce DIDO and improve patient outcomes.