Mobile Stroke Unit and Mechanical Thrombectomy Workflow: A Single Center 5‐Year Experience

Abstract

Background

Mobile stroke unit (MSU) has been demonstrated to significantly reduce time to treatment and increase the chances of early intravenous thrombolysis with positive effect on clinical outcomes. The evidence of its impact on the treatment time metrics and outcomes in endovascular treatment‐eligible patients outside of large clinical trials is limited. We sought to investigate the potential workflow benefits of MSU‐based care compared with emergency medical services‐based care for endovascular treatment‐eligible patients.

Methods

This is an observational, cross‐sectional study, based on a retrospective review of a single‐center, prospectively maintained, mechanical thrombectomy database spanning June 2018 to November 2023. Patients receiving endovascular treatment for large‐vessel occlusion strokes were divided in 2 groups: MSU‐transported and emergency medical services‐transported (mothership) presenting within MSU operating days/hours. Treatment time metrics and clinical outcomes were compared.

Results

A total of 565 patients who fit inclusion criteria were identified: 66 were transported via MSU and 499 were transported via emergency medical services. MSU‐transported patients were more likely to be within treatment window for intravenous thrombolysis and to bypass multimodal imaging. The co‐primary end points (door to angio, door to puncture, and door to reperfusion times) favored the MSU‐transported patients (41 versus 62 minutes, 58 versus 82 minutes, 96 versus 127 minutes; P<0.001). The times from last known well to puncture were significantly shorter in the MSU‐transported group (237 versus 389 minutes, P = 0.021). Functional outcomes at 90 days and rates of intracerebral hemorrhage were similar between the 2 groups.

Conclusion

In this single‐center analysis, MSU‐transported patients demonstrated improved time metrics, with shorter treatment times when compared with emergency medical services‐transported patients. Outcomes and safety parameters did not differ between the 2 groups.

Nonstandard Abbreviations and Acronyms

EVT

endovascular treatment

IVT

intravenous thrombolysis

LKW

last known well

LVO

large vessel occlusion

MSU

mobile stroke unit

Clinical Perspective

What Is New?

• This study is one of the few to show the positive impact of mobile stroke unit‐related care in workflow metrics in endovascular therapy‐eligible patients with ischemic stroke outside of randomized trials.

• All in‐hospital workflow related metrics, including times from emergency department door to computed tomography, door to angio, door to puncture, and door to reperfusion were faster in mobile stroke unit‐transported patients compared with emergency medical services‐transported patients.

What Are the Clinical Implications?

• The positive impact of mobile stroke unit‐related care in delivering faster endovascular treatment for eligible patients is significant and requires further corroboration to help determine its final impact in outcomes.

Endovascular therapy (EVT) has become the standard of care for eligible patients with large vessel occlusion (LVO)‐related ischemic strokes. ¹ Early intervention/reperfusion is critical considering its association with improved clinical outcomes. ² , ³ , ⁴ , ⁵ Only about 10% to 15% of patients with ischemic stroke in the United States are treated with intravenous thrombolysis (IVT) and/or mechanical thrombectomy and approximately 1% get treatment within the first hour after the ictus, the so called “golden‐hour,” when the chances of favorable outcome are maximized. ⁶ , ⁷ The emphasis on fast access to treatment ⁸ coupled with the consistent delays in the LVO triage workflow ⁹ are corroborated by the efforts led by the American Heart Association/American Stroke Association in setting more aggressive targets for timely treatment with door‐to‐device times within 90 minutes for direct‐arriving patients (60 minutes for transfers) in 50% or more of eligible patients with acute ischemic stroke. ¹⁰

Strategies to expedite diagnostic testing and treatment delivery are increasingly proposed, including the use of mobile stroke units (MSUs), where an ambulance equipped with computed tomography (CT) scan and telemedicine capabilities is dispatched with the goal of treating IVT‐eligible patients in the prehospital setting. Faster triage and treatment times are also expected to translate into better outcomes for LVO‐related patients who might be candidates for EVT. Nonrandomized data from an Australian study revealed lower median time from door to puncture (33 versus 50 minutes), favoring MSU versus controls (directly presenting or transfer patients). ¹¹ The BEST‐MSU (Benefits of Stroke Treatment Delivered by a Mobile Stroke Unit Compared with Standard Management by Emergency Medical Services) trial demonstrated that in patients presenting within 4.5 hours of last known normal (23.7% undergoing EVT), MSU‐treated subjects had overall improved clinical outcomes compared with emergency medical services (EMS) care, with MSU patients having access to earlier endovascular treatment (76 versus 94 minutes from door to puncture). ¹² Other process time metrics, like last known well (LKW) to puncture and recanalization, were not significantly different between the 2 groups. ¹³ In this study, we sought to investigate the potential workflow benefits of MSU‐based care compared with EMS‐based care for EVT‐eligible patients over the first 5 years of MSU activity in our comprehensive stroke center. Our hypothesis is that MSU‐transported patients with LVO‐related ischemic stroke may have faster time metrics when undergoing mechanical thrombectomy.

Methods

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Mobile Stroke Unit

The local MSU is equipped with a mobile CT scanner (Samsung CereTom 8‐slice) and staffed by an EMS driver, an emergency medicine registered nurse, a paramedic, and a CT technician. The EMS dispatcher screens calls for intervenable stroke symptoms and dispatches the nearest EMS to the scene if the last known well is <24 hours. The EMS and MSU have similar activation guidelines and are concomitantly activated. Both entities have the autonomy to cancel the alert if the diagnosis indicates a stroke mimic or if no stroke symptoms are observed. At our center, EMS providers notify our hospital stroke team of patients in the field with suspected stroke whenever possible; however, prenotification data were not captured for non‐MSU patients in this analysis. Approximately 80% of patients treated by non‐MSU EMS providers are treated by our institution's EMS providers, with the remainder of patients treated by other local EMS groups. Within the MSU, remote evaluation of the patient by a vascular neurology specialist (fellow or faculty) is facilitated by a 2‐way video telemedicine platform. The MSU has the provision of performing all standard point‐of‐care testing, teleradiology capabilities (noncontrast head CT, but no CT angiography [CTA]) to securely export images to our comprehensive stroke center, carries thrombolytics along with antihypertensive medications, and covers 132 square miles across the city of Atlanta. The MSU went operational on May 2018 initially with 6 days/week and later expanding to 7 days/week coverage, from 8:00 a.m. to 8:00 p.m., and with a fully dedicated staff.

Study Population and Measures of Outcome

This is a cross‐sectional study/retrospective review of a prospectively collected mechanical thrombectomy registry at a tertiary academic medical center from November 2018 to November 2023. All consecutive patients who arrived at our comprehensive stroke center during the same MSU operational days and hours and underwent EVT for LVO‐related ischemic stroke were included. Patients were divided in 2 groups: MSU and EMS‐transported (mothership) patients. Baseline, procedural, and imaging characteristics were compared between the 2 groups. Time points captured in this study included time of LKW, symptom recognition, emergency department (ED) door, imaging, angio suite arrival, arterial puncture, and reperfusion (expanded Thrombolysis in Cerebral Infarction score 2b‐3). ¹⁴ Treatment outcomes included degree of reperfusion (expanded Thrombolysis in Cerebral Infarction) and modified Rankin Scale score at discharge and 3 months (modified Rankin Scale≤2 considered good outcomes). Safety outcomes were symptomatic intracerebral hemorrhage (hemorrhage on CT/magnetic resonance imaging associated with a ≥4‐point increase in the National Institutes of Health Stroke Scale score), any evidence of parenchymal hematoma (PH‐1 and PH‐2) or subarachnoid hemorrhage in post‐EVT CT scans.

Statistical Analysis

Descriptive and Outcome Analyses

Baseline data were described as median (quartiles), when numeric, and absolute and relative frequencies when categorical. For the univariate analysis of logistic outcomes, Wilcoxon rank sum test, chi‐square test or Fisher's exact test were used, as appropriate.

The co‐primary end points were the time interval between door to imaging (repeat imaging in the ED for MSU patients), door to puncture, and door to reperfusion. Secondary endpoints included time from LKW and symptom recognition to door and arterial puncture. Subsequently, we performed linear regression analyses for IVT treatment among patients with LKW‐recognition times <180 minutes and for door‐to‐puncture time among patients with door‐to‐puncture time <500 minutes. Both models were adjusted for presence of witnessed symptom onset and time from LKW to symptom recognition. Such variables were chosen because they change the likelihood of MSU dispatchment and may directly or indirectly influence the outcomes of interest.

Mediation Analysis

To estimate how much of the MSU benefit on door‐to‐puncture time was explained by the time to imaging completion, we performed a mediation analysis. This way, we could estimate the average causal mediated effect, corresponding to the part of the effect explained by the time to imaging, and the average direct effect, which constitutes the nonmediated part of the total effect. We used nonparametric bootstrap with 1000 simulations through the mediation package for such analysis. ¹⁵

Missing Data

Missing data were handled through multiple imputation by chained equations via the mice package for all analyses. ¹⁶ An alpha of 0.05 was used to identify statistical significance. All analyses were conducted in R Software v4.3.3

This manuscript abides by the Strengthening the Reporting of Observational Studies in Epidemiology recommendations.

Results

Among patients undergoing EVT during MSU operating days/hours, 565 were included (66 were in MSU and 499 in EMS groups). Median overall age was 66 years, 55% were male. Median baseline Alberta Stroke Program Early CT Score and National Institutes of Health Stroke Scale score were comparable between groups (Table 1).

Table 1.

Descriptive and Univariate Analysis of MSU and Mothership (EMS) Patients

Characteristic	Mothership N = 499	MSU N = 66	P value
Age, y	66 (56, 76)	66 (47, 77)	0.318
Sex (Male)	276 (56%)	36 (55%)	0.968
Hypertension	378 (76%)	45 (68%)	0.183
AF	118 (24%)	18 (27%)	0.517
NIHSS score	16 (11, 21)	16 (11, 21)	0.839
ASPECTS	8 (7, 10)	9 (8, 10)	0.122
Prestroke mRS score	0 (0, 1)	0 (0, 1)	0.815
Occlusion location			0.392
ACA	7 (1.4%)	1 (1.5%)
Basilar	24 (4.8%)	2 (3.0%)
Extracranial ICA	41 (8.3%)	5 (7.6%)
Extracranial vertebral	1 (0.2%)	0 (0%)
ICA petrous or cavernous	15 (3.0%)	0 (0%)
ICA‐T	55 (11%)	7 (11%)
Intracranial vertebral	5 (1.0%)	0 (0%)
MCA M1	208 (42%)	26 (39%)
MCA M2	114 (23%)	25 (38%)
MCA M3	16 (3.2%)	0 (0%)
PCA	10 (2.0%)	0 (0%)

ACA indicates anterior cerebral artery; AF, atrial fibrillation; ASPECTS, Alberta Stroke Program Early CT [Computed Tomography] Score; EMS, emergency medical services; ICA, internal carotid artery; ICA‐T, internal carotid artery terminus; MCA, middle cerebral artery; mRS, modified Rankin Scale; MSU, mobile stroke unit; NIHSS, National Institutes of Health Stroke Scale; and PCA, posterior cerebral artery.

Witnessed onset of stroke was more common in the MSU group (42% versus 27%, P = 0.007) and time from symptom recognition to ED door was shorter for EMS‐transported patients (84 versus 103 minutes, P = 0.001). More patients in the MSU group were within the IVT window and received alteplase (tissue‐type plasminogen activator) or tenecteplase (45% versus 24%, P<0.001). MSU patients were more likely to bypass further imaging and go direct to angiography upon hospital arrival (12% versus 2.2%, P<0.001), and more patients in the EMS group had multimodal imaging before EVT (95% versus 83%, P = 0.001).

Time Metrics

In terms of the co‐primary end points, all time metrics favored MSU‐transported patients, including door to angio (41 versus 62 minutes, P<0.001), door to puncture (58 versus 82 minutes, P<0.001) and door to reperfusion (96 versus 127 minutes, P<0.001). Secondary end points indicated that the LKW‐to‐arterial puncture times were shorter in the MSU group (237 versus 389 minutes, P = 0.021), which also had significantly faster times from ED door to CT and multimodal imaging completion (9 versus 17 minutes, and 17 versus 27 minutes, respectively, P<0.001) (Table 2). A sensitivity analysis excluding direct‐to‐angio patients indicated that all time metrics, except for LKW‐to‐puncture time, remained faster in the MSU group. Results were unchanged after adjustment for witnessed symptom onset (yes/no) and time from last known normal to symptom recognition. (Table 3)

Table 2.

Time Metrics, Clinical and Safety Outcomes

Characteristic	MSU, N = 66	Mothership, N = 499	P value
Witnessed onset	28 (42%)	132 (27%)	0.007
Logistics
tPA/TNK administered	30 (45%)	119 (24%)	<0.001
Outside time window for IVT	23 (35%)	261 (52%)	0.008
Direct to angiography	8 (12%)	11 (2.2%)	<0.001
Multimodal imaging	55 (83%)	472 (95%)	0.001
Time from recognition to (min)
ED door	103 (81–147)	84 (52–123)	0.001
Arterial puncture	165 (136–210)	176 (136–226)	0.312
Time from LKW to (min)
Symptom recognition	20 (0–420)	90 (0–540)	0.088
Door	172 (99–650)	275 (90–714)	0.479
Puncture	237 (149–734)	389 (184–826)	0.021
Arterial puncture (categorical)			0.040
≤ 4.5h	36 (56%)	198 (43%)
> 4.5h	28 (44%)	266 (57%)
Time from door to (min)
CT	9 (4–15)	17 (12–28)	<0.001
Multimodal imaging completion	17 (13–23)	27 (20–37)	<0.001
tPA/TNK	−15 (‐21–‐8)	34 (27–47)	<0.001
Angiography suite	41 (31–56)	62 (48–83)	<0.001
Arterial Puncture	58 (50–75)	82 (65–104)	<0.001
Reperfusion	96 (80–132)	127 (101–162)	<0.001
Procedural and clinical outcomes
Final eTICI score			0.882
0	0 (0%)	11 (2.2%)
1	1 (1.5%)	1 (0.2%)
2a	0 (0%)	7 (1.4%)
2b50	4 (6.2%)	20 (4.0%)
2b67	9 (14%)	100 (20%)
2c	16 (25%)	81 (16%)
3	35 (54%)	279 (56%)
Discharge mRS score	3 (1–4)	3 (1–4)	0.198
Discharge mRS score 0–2	26 (44%)	178 (39%)	0.412
Day 90 mRS score	3 (1–5)	3 (1–5)	0.455
Day 90 mRS score 0–2	22 (45%)	162 (43%)	0.822
Parenchymal hematoma			0.781
Absent	63 (95%)	466 (95%)
PH‐1	3 (4.5%)	18 (3.7%)
PH‐2	0 (0%)	8 (1.6%)
Subarachnoid hemorrhage	4 (6.1%)	53 (11%)	0.227
Symptomatic ICH	1 (1.5%)	11 (2.3%)	>0.999
90‐d mortality	86 (23%)	10 (20%)	0.691

CT indicates computed tomography; ED, emergency department; eTICI, expanded thrombolysis in cerebral infarction; ICH, intracerebral hemorrhage; IVT, intravenous thrombolysis; LKW, last known well; mRS, modified Rankin Scale; MSU, mobile stroke unit; TNK, tenecteplase; and tPA, tissue plasminogen activator

Table 3.

Linear Regression Model for Door‐to‐Puncture Time (Minutes), DTA Not Included

Characteristic *	Beta (95% CI)	P value
Mothership	—
Mobile stroke unit	−23.3 (−34.7 to −11.9)	<0.001

^*Model adjusted for witnessed symptom onset (yes/no) and time from last known normal to symptom recognition

DTA indicates direct to angio.

Mediation analyses showed that the association between MSU and lower door‐to‐puncture time was substantially explained by a difference in rates and time to completion of multimodal imaging, which was responsible for 48% (95% CI, 26%–71%) of the difference between MSU and EMS. Direct to angio explained a small part of this difference, mediating a proportion of 5% of the MSU effect on door to ‐puncture (95% CI, 0%–14%). The bulk of this observed benefit was mainly due to shorter times to multimodal imaging completion observed in MSU patients (Table 4).

Table 4.

Mediation Analysis for DTA and Multimodal Imaging as Mediators for MSU Benefit in Door‐to‐Puncture Time (Minutes)

	DTA		MMI (Yes/No)		Time to MMI completion		Time to any imaging completion
Effect	Estimate (95% CI)	P value	Estimate (95% CI)	P value	Estimate (95% CI)	P value	Estimate (95% CI)	P value
ACME	−1.24 (−3.16 to 0.24)	0.076	−1.75 (−4.60 to −0.14)	0.062	−11.63 (−17.65 to −5.85)	<0.001	−10.75 (−16.85 to −4.65)	<0.001
ADE	−22.56 (−32.32 to −12.73)	<0.001	−20.71 (−31.06 to −11.15)	<0.001	−10.66 (−18.27 to −2.97)	0.003	−11.82 (−19.81 to −4.87)	<0.001
Total effect	−23.80 (−33.29 to −14.16)	<0.001	−22.46 (−32.61 to −12.91)	<0.001	−22.29 (−31.80 to −12.55)	<0.001	−22.57 (−32.64 to −13.48)	<0.001
Proportion mediated	0.05 (−0.01 to 0.14)	0.132	0.08 (0.01 to 0.21)	0.012	0.53 (0.31 to 0.80)	<0.001	0.48 (0.26 to 0.71)	<0.001

ACME indicates average causal mediated effect, corresponding to the part of the effect explained by the mediator; ADE, average direct effect (part of the total effect not mediated by the mediator of interest); DTA, direct to angio; MMI, multimodal imaging; and MSU, mobile stroke unit.

Procedural and Clinical Outcomes

There was no difference in final degree of reperfusion between the 2 arms. The median modified Rankin Scale score and rate of patients with favorable outcome (modified Rankin Scale score = 0–2) at 3 months were similar between the 2 groups (45% versus 43%, P = 0.822). Similar results were observed after the exclusion of direct‐to‐angio patients. There was no difference in hemorrhagic complications or mortality between the 2 groups.

Discussion

MSU‐derived thrombectomies represented nearly 12% of all EVT cases within the present cohort. MSU patients were found to have higher rates of IVT and improved workflow logistics compared with EMS‐transported individuals.

MSUs have been increasingly adopted since their inception in Germany in 2003. ¹⁷ Their routine use has been part of the prehospital stroke care in many large centers around the globe, mostly in the United States. Clinical trials have demonstrated that acute stroke care via MSU is associated with increased odds of fast delivery of IVT, mostly driven by swift imaging and clinical assessment, leading to higher odds of treatment within the first hour from ictus. ¹² , ¹⁸ , ¹⁹ Early treatment, specifically when delivered within the golden hour, has been consistently shown to translate into better functional outcomes, with an almost 2‐times higher chance of excellent outcome compared with treatments delivered later. ⁷ , ²⁰

In a randomized trial on the effectiveness of MSU (PHANTOM‐S [Prehospital Acute Neurological Treatment and Optimization of Medical Care in Stroke]) the chances of receiving IVT were 12% higher than in standard care, and treatment delivery was shortened by 25 minutes on average, ¹⁹ with higher rates of MSU patients treated within the first hour. ²¹ The same effect was demonstrated in the BEST‐MSU trial comparing MSU‐ to standard EMS‐delivered care, where the median time from LKW to IVT was 36 minutes shorter than standard EMS‐based care, with a 10‐fold greater odds of treatment being administered within the golden hour. ¹² , ²⁰ The improved treatment logistics were found to result in better clinical outcomes in MSU‐based IVT patients, validating the positive impact of prehospital stroke treatment delivered via MSU. ¹² , ²⁰ , ²² These randomized trials did not demonstrate a consistent positive impact in EVT‐related time metrics, as evidenced in a recent meta‐analysis ²³ and, although BEST‐MSU was able to demonstrate improved clinical outcome for the EVT‐eligible patients, there is little data outside of clinical trials. ¹³

In an analysis of the first‐year experience with MSU care in Australia, besides demonstrating MSU benefits toward IVT‐related logistics, a significant positive impact in EVT triage was noted, with 17 minutes faster median times from ED arrival to puncture (door to puncture), a benefit at least partially explained by the use of CTA in 80% of the MSU patients. ¹¹ In the BEST‐MSU trial EVT patients, the ED door‐to‐puncture time was faster in the MSU‐transported patients with a mean difference of 18 minutes favoring MSU patients. ¹³ These findings are very consistent with our present analysis (17 minutes faster door‐ to ‐puncture). However, in contrast with our results, BEST‐MSU found no difference in the LKW‐to‐puncture and LKW‐to‐recanalization times between the 2 groups. ¹³ In the present study, although time from symptom recognition to ED door was longer in the MSU, we observed a significantly improved time from LKW–as well as symptom recognition–to procedure initiation and to reperfusion in MSU‐treated patients. This benefit is in large part related to the expedited workflow in patients arriving via MSU and a prenotification system including ED personnel and endovascular team setting in motion the necessary cascade of events that will lead to intervention. This might help explain the improved metrics observed in our study compared with the BEST‐MSU results.

In exploring the potential explanations for better time metrics in MSU‐treated patients, a sensitivity analysis, excluding direct‐to‐angio cases, indicated that workflow benefits conferred by MSU care remained. Despite being able to quantify that time to completion of multimodal imaging was responsible for a significant proportion of the time metrics difference between MSU‐ and EMS‐treated patients (presumably due to already established intravenous access and no need for examination before scans), most of the MSU logistical benefit was conferred by other reasons not accounted for in this study and likely include more detailed prenotification and ability to contact family for relevant collateral information, before ED arrival. ED prenotification is routinely done in our institution for all incoming patients with stroke (EMS and MSU); however, MSU has the advantage of providing more detailed screening as well as pre‐hospital examination done by a neurologist, with earlier detection of possible LVO and hence earlier notification of the on‐call endovascular team. The usefulness and impact of multimodal imaging for EVT‐eligible patients may change in the future, especially given the recent evidence for remained benefit of mechanical thrombectomy in patients with large core infarcts. ²⁴ Whether differences in impact will be observed in MSU versus EMS patients remain to be determined. In our study no patients had CTA performed in the MSU; it would be reasonable to expect even higher benefit in EVT time metrics if CTAs are performed prearrival, as previously reported. ¹¹ Prenotification of the endovascular team prearrival would likely account for this benefit. In the present study, the absence of prehospital CTAs could have underestimated the number of patients who received IVT in the MSU and that had recanalized by the time of the ED CTA; however, this would unlikely affect the present results because the majority of patients underwent a CTA after ED arrival.

The present study was not powered to evaluate differences in clinical outcomes and was particularly limited by a relatively small MSU‐derived subgroup sample size, including the differences in the frequency of multimodal imaging. The limitations inherent to retrospective design are applicable to our analyses. We did not control for geospatial aspects of stroke alert activation; therefore, the time differences between the MSU‐ and EMS‐based care related to LKW and symptom recognition could be biased.

In conclusion, in our single‐center analysis, the MSU was responsible for a significant proportion of thrombectomies in patients who presented directly to our comprehensive stroke center within the study period. MSU transport was associated with improved workflow leading to shorter times to treatment. Safety and clinical outcomes were comparable among MSU‐ and EMS‐transport groups, supporting the need for further investigation of the direct‐to‐angio approach for patients with LVO.

Sources of Funding

This study was not funded.

Disclosures

B.L.: N/A. R. G. N. consultant: Cerenovus, Medtronic USA, Inc., Perfuze, Brainomix, Shanghai Wallaby, Stryker, Genentech, Perfuze, Genentech, Hybernia, Ceretrieve, Ceretrieve, Prolong Pharmaceuticals, Brainomix, Phenox, NeuroVasc Technologies, Inc., Anaconda, Vesalio, Imperative Care, Inc, RapidPulse, phenox Inc., Stryker Corporation, RapidPulse, Prolong Pharmaceuticals, Cerebrotech, Corindus Vascular Robotics, Philips, Viz‐AI, Biogen, Inc., Cerenovus, Astrocyte, Imperative Care, Vesalio, Corindus Inc., Biogen, Inc., Anaconda Biomed, Viz‐AI, Medtronic USA, Inc.; Ownership/partnership/ principal: Corindus Inc., Brainomix, Ceretrieve, Quantanosis AI, Brain4Care, Viseon, Inc., Viz‐AI, Piraeus Medical, Viz‐AI, Perfuze, Vesalio, Cerebrotech, Truvic, Perfuze, Reist/Q?Apel Medical; Research: Stryker, Cerenovus. P. N. M.: N/A. X. G.: N/A. N. A. B.: N/A. M. A. T.: N/A. A. R. A‐B. consultant: Stryker, Johnson and Johnson. D. N. N/A. M. R. F. expert witness: Franke & Solloum, PLLC, Spencer Fane LLP. C. F.: N/A. J. A. G. consultant: Imperative Care, Inc, NTI, Cognition; Research: National Institute of Neurological Disorders and Stroke, Uniformed Services University‐Surgical Critical Care Initiative, Emory Neurosurgery Catalyst, Georgia Research Alliance, Emory Medical Care Foundation. J. D. Ownership/partnership/principal: Gravity medical technology; Salary: Emory School of Medicine. D. C. H. consultant: Chiesi USA, Inc., Vesalio, Poseydon Medical, Brainomix, Cerenovus, Stryker; Ownership/partnership/principal: Viz AI. R. G. N. serves on the Editorial Board of S:VIN. Editorial Board Members are not involved in the handling or final disposition of submiss.

Supporting information

STROBE Statement—checklist of items that should be included in reports of observational studies