(BEMI-S): Brain Emergency Management Initiative: Assessment of Embolectomy Transfer Times Using a Novel Stroke System Transfer Protocol

Abstract

Background and Objectives:

Recent studies of national stroke door-in-door-out times (DIDO) found that most transfers for acute interventions are not completed within the recommended time frame. There is a critical need for effective emergent transfer protocols to improve outcomes. The Brain Emergency Management Initiative (BEMI) is a stroke transfer protocol connecting acute stroke patients at spoke sites to a hub center for embolectomy. BEMI has been shown to significantly reduce transfer time metrics through rapid transit activation, CT head/CTA bundling, digital image sharing, standardized documentation, and remote patient admission. In this study we evaluated the sustainability of the BEMI impact on reduction of these transfer metrics.

Methods:

We assessed data for patients transferred for embolectomy in our stroke system. Patients were compared between 3 groups: before protocol (“pre-BEMI”), initial year of protocol (“BEMI”), and the contemporaneous 5 years of protocol usage (“BEMI-S”) to assess for sustainability. Time metrics assessed include DIDO, time-of-treatment-decision-to-groin-puncture (TDGP), as well as a safety outcome of symptomatic ICH (sICH) rates.

Results:

429 transfers were evaluated, with a final sample size of 271 patients (pre-BEMI n=31, BEMI n=32, BEMI-S n=208). There was significantly shorter median DIDO in the BEMI groups (DIDO: pre-BEMI median=143 minutes vs. BEMI=118, p=0.015; vs. BEMI-S=97, p=2.1e-07). DIDO also improved significantly from BEMI to BEMI-S groups (118 vs. 97 minutes; p=0.0005). TDGP is significantly reduced in the BEMI and BEMI-S groups as compared to the pre-BEMI group (TDGP pre-BEMI median=155 minutes vs. BEMI=130, p=0.01; vs. BEMI-S=125, p=4.15e-13), however is similar between the BEMI and BEMI-S groups (130 vs. 125 minutes, p=.89). Symptomatic ICH rates were similar before and immediately after BEMI implementation but significantly reduced in BEMI-S group. (sICH pre-BEMI 12.9%, vs. BEMI 15.6%, p=1, vs. BEMI-S 2.4%, p=0.037; BEMI vs. BEMI-S p=0.014).

Discussion:

The BEMI protocol significantly improved transfer (DIDO) by 46 minutes and treatment times (TDGP) by 25 minutes in our stroke network, showing continuous sustainability. sICH rates significantly lowered by over 10% with continued use of the BEMI protocol. Our protocol builds upon similar rapid transfer stroke protocols through incorporating unique features such as air transit, video telestroke specialist evaluation and rapid on-load protocols with uniform documentation.

Introduction

Acute stroke treatments, such as thrombolysis and endovascular thrombectomy (EVT), are highly time-sensitive and not available at all hospitals. Current guidelines recommend initiating these treatments as quickly as possible in eligible patients, as faster treatment is associated with better clinical outcomes^1,2,3. Only a limited population can access an EVT capable center locally, creating a gap in acute stroke treatment access⁴. Interfacility stroke networks help bridge this access gap through remote direction of thrombolytic administration and arranging transfers to centers capable of performing EVT. Delays in interfacility transfer for acute stroke patients is a common reason these patients may be excluded from EVT⁵.

Door-in-door-out (DIDO) time is a common quality metric that measures the duration between a patient’s arrival at a spoke healthcare setting and their departure for transfer to a hub EVT-capable center. Recent studies of national DIDO in the Get with the Guidelines Stroke registry revealed that less than 30% of transfers for acute interventions were completed within the recommended timeframe^6,7. Current recommendations set a goal of less than 90 minutes in at least 50% of these transfers⁸. Prolonged DIDO portends worse outcomes in large vessel occlusion (LVO) stroke patients⁹. This indicates a critical need for effective emergent transfer protocols to improve key time metrics which enable improved clinical outcomes.

The Brain Emergency Management Initiative (BEMI) is a stroke system transfer protocol connecting acute stroke patients at spoke sites to a higher-level-of-care hub center for EVT for telestroke transfer patients¹⁰. BEMI includes early activation of transport, immediate computed tomography of the head (CT head) and head and neck angiography (CTA) digital imaging transmission, standardized documentation, rapid transport on-load, and remote patient admission.

The University of California, San Diego stroke center designed the BEMI protocol utilizing parallel processing and specific personnel assignments. This protocol has been in operation for many years within our telestroke network for transfer to our Hub center. BEMI, similar to the term STEMI, is modeled after the ST-segment elevation myocardial infarction (STEMI) rapid transfer protocols with the goal to impart the same emergency nature of an acute stroke.^{10, 11} BEMI has been shown to significantly reduce DIDO and time-of-treatment-decision-to-groin-puncture (TDGP) on initial evaluation¹⁰. Preliminary investigation of BEMI was performed in the framework of a pilot study, warranting continued evaluation. In this study, we evaluated the multi-year sustainability of the BEMI program and its impact on reduction of key transfer metrics.

Methods

We retrospectively assessed prospectively collected data for consecutive patients transferred for embolectomy in our two-tier stroke network. Our two tier network consists of a single hub hospital system made up of two comprehensive stroke centers in San Diego, CA that are thrombectomy capable (Tier 1). We have several spoke sites within neighboring counties in southern California, with distances from 53–123 miles away from our hub site, which are not thrombectomy capable (Tier 2). These spoke sites are considered urban and suburban, there are no rural sites included. During the time period of this study patients were evaluated at four different spoke sites and underwent transfer to one of the two San Diego hospitals that make up our hub site. Our departmental telestroke database and additional chart review were evaluated by a single reviewer.

Demographic data evaluated included age, gender, race, stroke risk factors, and presenting National Institutes of Health Stroke Scale (NIHSS). Our primary outcome was DIDO time. Door in time is defined as time of patient arrival to the spoke site. Door out time is the timepoint the helicopter took off. Our secondary outcome is time-of-treatment-decision-to-groin-puncture (TDGP). Time of decision is the time at which it was made to initiate the pathway to acute treatment including both thrombolysis and EVT. This data is manually recorded by the teleneurologist for every case for all patients evaluated. The time of groin puncture is when this procedure was achieved for planned EVT. A safety outcome of symptomatic ICH (sICH) incidence is also evaluated, defined using the National Institute of Neurological Disorders and Stroke definition.

Based on helicopter emergency medical services (HEMS) data, most spoke sites are an average of a 25-minute flight time to the hub, whereas one site’s flight time is 50 minutes. To normalize this additional transport time, 25 minutes was subtracted for the more distant spoke. This manual normalization was applied in the TDGP calculations and was not relevant to nor did it affect the DIDO times that are calculated prior to lifeflight transfer. Only one way flight times are included.

The BEMI protocol involves optimizing processes at the spoke site, transport system and hub site. At the spoke site, a telestroke consultation is uniformly activated by emergency department clinicians for patients presenting with a new focal deficit within 6 hours of last known well time (LKW). If a patient presents within 6–24 hours after last known well time, a local spoke-only stroke code is activated with CT head and CTA completed, and a telestroke hub code is subsequently activated if a CTA is performed and a LVO is identified. Once a telestroke code has been activated by the emergency department provider, direct sign-out from the emergency department provider is given to the stroke specialist at the hub site by phone while the patient is in the CT scanner. CT head and CTA are completed for all patients and digitally transmitted to the stroke specialist. These images are automatically uploaded into a digital imaging viewer. There is no requirement for copying of images or additional records. Documentation is consolidated into single form (figure 1) which delegates specific tasks for concurrent execution by designated spoke staff. HEMS is activated when determined by the stroke specialist once clinical history, video exam and imaging is reviewed. This protocol was approved by HEMS leads with knowledge that some activations may be aborted if the transfer is subsequently called off. Transfers may be called off in cases in which upon review by the neurointerventionalist, the patient is deemed not to be an EVT candidate based on anatomical or technical reasons, or in which the patient or family has decided not to proceed with EVT. Rapid on-load protocols designed to expedite the process of safely loading a patient onto the helicopter are used for BEMI patients to hasten door-out time. Concurrently at the hub site, the telestroke provider facilitates remote admission, rapid bed assignment and patient registration. The hub stroke specialist accepts patients on behalf of the intensive care unit and can place orders for the patient prior to their arrival. Pre-notification of air transport arrival is given to the hub site stroke team who then meet HEMS on the helipad for immediate assessment while transporting to the procedural suite or CT scanner if deemed necessary on a case by case basis.

Figure 1: The BEMI protocol.

Workflow of BEMI protocol

Patients were assessed in 3 groups: before protocol implementation (“Pre-BEMI”: October 2013 - June 2016), initial protocol implementation (“BEMI”: June 2016 -October 2017) and a contemporaneous group (“BEMI-S”: July 2018 – September 2023) to examine sustainability of the protocol. The start date of the BEMI group was chosen based on when the protocol was introduced. The start date of the BEMI-S group was chosen a year after this to allow for maturation and longitudinal monitoring of the protocol, and after our pilot study of the protocol had been completed. For both the BEMI and BEMI-S groups the same standardized BEMI protocol was utilized (figure 1). Inclusion criteria were emergency telestroke consultation, helicopter transport, transfer for planned acute embolectomy with groin puncture achieved. Exclusion criteria were inpatient telestroke consultation, transport converted to ground, nonurgent transfers, incomplete data for the primary outcome.

Demographic categorical items were compared using pairwise Fisher’s test, whereas NIHSS and age were assessed using pairwise Wilcoxon test. DIDO was compared between groups using pairwise comparisons via Wilcoxon Rank Sum test with Holm correction. TDGP was compared using Wilcoxon signed rank test with continuity correction. sICH rates were compared using Pairwise Fisher’s Test. For all continuous variables, median times were used given skewedness of data.

Standard Protocol Approvals, Registrations, and Patient Consents

Data was acquired with IRB approval via our institutional quality improvement ACQUIRE pathway (approval #863). Patient consent was waived for this study as approved by the IRB given this project was completed as part of a quality improvement initiative requiring chart review of data collected as part of standard of care.

Data Availability Statement

Anonymized data not published within this article will be made available by request from any qualified investigator.

Results

429 total transfers were evaluated with 10 excluded for inpatient status, 100 excluded for nonurgent transfers, 30 that came by ground transport and 18 with incomplete data (figure 2). The final sample size is 271, in which the subgroups consist of pre-BEMI n=31, BEMI n=32, BEMI-S n=208. Analyses compared each BEMI group to the pre-BEMI group. Each group was comparable in demographics and vascular risk factors aside from median initial NIHSS. Comparable factors between all groups includes age (pre-BEMI median= 70.2, BEMI=69.7, BEMI-S=71.2), gender (pre-BEMI percent male=67.7%, BEMI=59.4%, BEMI-S=49.5%), comorbid diabetes (pre-BEMI=25.8%, BEMI=46.9%, BEMI-S=30.8%), hypertension (pre-BEMI=80.7%, BEMI=71.9%, BEMI-S=69.2%) and atrial fibrillation (pre-BEMI=48.4%, BEMI= 46.9%, BEMI-S=42.8%). Median NIHSS was higher in the BEMI groups (pre-BEMI median=10 points vs. BEMI=20, p=0.0063; vs. BEMI-S=17, p=0.0046) (table 1).

Figure 2: Total screened patients and exclusion criteria.

Process flow diagram of excluded patients

Table 1:

Demographics are comparable between groups aside from NIHSS which is significantly higher in the BEMI groups

	Pre-BEMI	BEMI	BEMI-S	P value pre-BEMI vs. BEMI	P value pre-BEMI vs. BEMI-S	P value BEMI vs. BEMI-S
Total (n)	31	32	208
Age (mean)	70.2	69.7	71.2	1	1	1
Male (n, %)	21 (67.7%)	19 (59.4%)	103 (49.5%)	0.69	0.25	0.69
Diabetes (n, %)	8 (25.8%)	15 (46.9%)	64 (30.8%)	0.31	0.68	0.31
Hypertension (n, %)	25 (80.7%)	23 (71.9%)	144 (69.2%)	1	0.64	1
Atrial Fibrillation (n, %)	15 (48.4%)	15 (46.9%)	89 (42.8%)	1	1	1
Initial NIHSS (median)	10	20	17	0.006	0.005	0.21

There were significantly shorter median DIDO in the BEMI groups (DIDO: pre-BEMI median=143 minutes vs. BEMI=118, p=0.015; vs. BEMI-S=97, p=2.1e-07). DIDO also improved significantly from BEMI to BEMI-S groups (118 vs. 97 minutes; p=0.0005). (table 2, figure 3). TDGP is significantly reduced in the BEMI and BEMI-S groups as compared to the pre-BEMI group (TDGP pre-BEMI median=155 minutes vs. BEMI=130, p=0.01; vs. BEMI-S=125, p=4.15e-13), however is similar between the BEMI and BEMI-S groups (130 vs. 125 minutes, p=.89) (figure 4). Symptomatic ICH rates were similar before and immediately after BEMI implementation but significantly reduced with continued use of the protocol (sICH pre-BEMI 12.9%, vs. BEMI 15.6%, p=1, vs. BEMI-S 2.4%, p=0.037; BEMI vs. BEMI-S p=0.014).

Table 2:

Primary, Secondary and Safety Outcomes

	Pre-BEMI	BEMI	BEMI-S	P value pre-BEMI vs. BEMI	P value pre-BEMI vs. BEMI-S	P value BEMI vs. BEMI-S
DIDO (median minutes)	143	118	97	0.015	2.1e-07	0.0005
TDGP (median minutes)	155	130	125	0.01	4.15e-13	.89
sICH (%)	12.9%	15.6%	2.4%	1	0.037	0.014

Figure 3: There is a significant decrease in DIDO times with BEMI use.

Violin plot demonstrating distribution of DIDO times in each group

Figure 4: Median TDGP is reduced in BEMI groups as compared to the pre-BEMI group, while BEMI groups are similar.

Bar graph displaying difference in TDGP between groups.

Discussion

The BEMI protocol has been implemented in our regional, multi-center, telestroke network over the last decade. This protocol was developed in response to our observation of inefficiencies in the acute stroke transfer process, in line with national data that most transfers for acute interventions are not completed within the recommended time frame. Barriers to transfer expediency that we sought to overcome with the BEMI protocol include discordant activation of multiple teams without streamlined communication and concurrent processing, delays in imaging transmission, delays in activation for emergent transfer and a lack of urgency surrounding stroke transfers. The goal of this protocol is to optimize stroke transfer and treatment times to facilitate improved patient outcomes for patients transferred to the hub center after telestroke consultation. In this exploratory analysis, with continued use of the BEMI protocol we have found improved key time metrics. This protocol instructs a culture of urgency and efficiency in each step of a stroke patient’s evaluation and transfer. This is accomplished through a protocolized approach with assigned roles, rapid telestroke neurologist evaluation, streamlined imaging transmission, consolidated documentation, helicopter transport and a rapid HEMS onloading protocol. A small pilot study of this protocol suggested that BEMI is effective in improving key transfer metrics¹⁰. In this larger longitudinal study, we have similarly observed that the implementation of the BEMI protocol has led to significant improvements in both transfer and treatment times within our stroke network over the past five years. Notably, transfer (DIDO) and treatment (TDGP) times are markedly reduced compared to the pre-protocol period. Specifically, DIDO times have improved by nearly 50 minutes. These promising results reflect not only the replication of our initial findings but also suggest significant improvements in the protocol’s efficacy over time.

Improved DIDO times demonstrate faster stroke patient transfers, while TDGP times reflect expedited acute stroke treatment. Although patient outcomes were not specifically evaluated in this study, it is well established that faster treatment leads to improved outcomes in stroke patients^1,2,3. In addition, a novel finding that was not evident in our previous study of the BEMI protocol is the noteworthy reduction in sICH rates in the BEMI-S group. As a safety outcome with large sample size variation between groups, this data remains exploratory. Possible explanations include faster transfer times leading to expedited evaluation and treatment of patients with LVO. It is also possible that this difference may be artifactual in the setting of concurrent quality improvement endeavors, or due to disparate sample sizes. This raises an important question for further study. We postulate that BEMI may improve patient outcomes by expediting key transfer metrics. Future investigation of BEMI’s impact on common outcomes data such as 90 day modified Rankin Scale and cost effectiveness are warranted.

Similar recent studies have identified that use of transfer protocols improve time metrics. McTaggart et al. evaluated transfer metrics after introducing a rapid transfer protocol from primary stroke centers (PSCs) to comprehensive stroke centers (CSCs). Their approach involved bundling CTA/CT head, CSC prenotification, and cloud-based image sharing for patients presenting with a Los Angeles Motor Score greater than 4. They observed a reduction in DIDO time from a median baseline of 104 minutes to 64 minutes in a cohort of 22 patients.¹². Similar to our TDGP findings, they also observed improvement in door-in-to-groin-puncture time. Howell et al. assessed 22 patients after their “Autolaunch” protocol was implemented for all LVO patients within 24h of LKW, reducing their median DIDO time from 109 to 85 minutes.¹³ The tested intervention was automatic ambulance dispatch prior to patient acceptance by CSC. Additionally, Gaynor et al. studied a rapid transfer protocol in 27 patients, in which the EMS crew remained on-site for physician determination of EVT candidacy. For those deemed eligible, the protocol included immediate transfer, early CSC notification, and preregistration. They found a reduction in median DIDO from 96 to 45 minutes¹⁴. Lastly, Kodankandath et al. implemented their “Stroke Rescue” program in 31 patients, increasing the percentage of transfers completed in under 60 minutes from 62% to 81%. This improvement was achieved by minimizing paperwork through direct MD-to-MD communication, utilizing a computer-aided dispatch program, and establishing a centralized transfer center¹⁵.

Although each of these protocols differs from the BEMI protocol, they individually share certain elements with BEMI including early CSC notification, rapid transit activation, remote imaging review, PSC CTA/CT head bundling, and integrated simplified documentation. They differed from our study in their small sample sizes as compared to our experimental group of 271 patients, use of a uniform EMS agency as opposed to our network that utilizes multiple agencies, and the use of pre-hospital triage LVO clinical screening tools or CTA confirmation as opposed to our telestroke neurologist video evaluation in which an exam and CTA assessment is completed in real time.

These studies did not report any significant decrease in sICH rates with the use of their rapid transfer protocols. In addition, each of these studies transferred patients exclusively by ground while BEMI utilizes air transport, given that our spoke sites are typically further away from the hub than in these other studies. In addition, our study has only included urban and suburban spoke sites, without any rural sites. This may also affect transfer times given additional benefit of air transport such as avoiding traffic congestion. A study in a rural hospital network comparing DIDO in ground versus air transport have found typically longer transfer times in air transported patients which may explain why our DIDO times are longer than those reported in some of these studies¹⁶ BEMI also streamlines additional hub site parameters including use of a telecommunications platform for imaging and information sharing with our multidisciplinary team including neurointerventionalist, remote admission and bed assignments which are not included in these other contemporary protocols, and may provide additional advantage over similar protocols. These differences highlight the need for customization of some aspects of transfer protocols for the stroke systems they are applied to.

The major limitations of our study include small sample size in two subgroups and unadjusted analyses. This study examines real world, prospectively collected data, thus there are limitations in the patient characteristic data available in all groups. For example we are unable to control for key factors such as specific vessel occluded, thrombolysis administration and successful endovascular reperfusion rates. Unadjusted analyses are therefore used thus our findings remain exploratory. In addition, although all demographics and most risk factors were comparable between groups, NIHSS was significantly higher in the BEMI and BEMI-S groups as compared to the pre-BEMI group. This suggests that the patients in the BEMI and BEMI-S groups had comparably more severe deficits. This further supports that the BEMI protocol expedites stroke transfers, even despite patients with more severe deficits with potential for higher acuity complications. Given the small sample size in the pre-BEMI group however, it is also possible this difference is spurious. Additionally, as this is an observational study of an intervention over time, our results may be confounded by concurrent quality improvement processes. Lastly, the center-specific and regional nature of our protocol presents limitations; its success relies heavily on the engagement of key stakeholders, including partner hospitals, helicopter transport companies, and the access to a telecommunications platform. Development and maintenance of these collaborative relationships may represent a significant barrier to broader adoption of similar protocols as they require significant time, resources and institutional commitment. Integral to the sustained success of our protocol is that we work to actively maintain engagement and protocol use through education sessions, stakeholder meetings, and feedback for any issues that arise during transfers. Similar strategies will be essential for the successful implementation of other acute transfer quality improvement initiatives.

Our study possesses several strengths, including a moderate sample size of the experimental group, prospective sequential data collection and a multi-center design involving transfers from distant and numerous spoke sites. Through use of prospectively collected data we anticipate minimal selection bias, however we acknowledge our exclusion criteria may introduce bias. For example, most patients converted to ground transport are for weather reasons, however some are due to high patient weight that was not compatible with safe helicopter transport. This may introduce complexity of patient loading that may result in delays. We anticipate minimal bias from inpatients as these patients follow an entirely different transfer protocol, patients excluded for incomplete data given this is a prospectively collected dataset, as well as for those excluded for nonurgent transfers in which BEMI protocol was applied in setting of patients without indication for acute EVT. The diversity of the size and scope of healthcare organizations within our stroke network enhances the generalizability of our findings. The distance between the hub and spoke sites and the application of air transport is also a unique feature of this study and may provide valuable insights for establishing protocols in suburban and rural locations in particular, where the density of CSC’s is on average lower⁴. A subsequent study including rural sites would be invaluable. Furthermore, current stroke guidelines advocate for the use of structured stroke systems and protocols such as BEMI¹. This alignment with established guidelines not only reinforces the value of our study in that it underscores the importance of implementing standardized protocols to improve stroke care.

While our study has demonstrated the sustainability of the association between BEMI protocol use and reduction of acute stroke/telestroke transfer metrics over time, further research is necessary to confirm a causal relationship between BEMI use and the improvement in key transfer time metrics, to understand its broader impact on patient outcomes, as well as reproducibility in other stroke networks. Future studies of BEMI should include additional patient baseline characteristics as well as adjusted analyses to further confirm our exploratory findings. The ideal acute stroke transfer protocol will have components that undoubtedly need to be customized to each telestroke system. Nonetheless, we assert that the core features of the BEMI protocol are likely to improve key time metrics across systems. For example, this study specifically evaluated air transfers but we hypothesize components of this protocol will be useful in improving even ground transport protocols. Specifically, we anticipate pre-assigned roles, standardized communication, remote clinical and imaging evaluation, and rapid on-load protocols will be universally beneficial. Ideally in future studies we would like to apply BEMI to other telestroke systems and assess its comparative efficacy. Concurrent quality improvement projects at other institutions may find it valuable to examine these individual components of their stroke transfer protocols to best optimize their workflow. Such investigations are warranted, to assess how to best form policy and guidelines surrounding best practice acute stroke transfer protocols. Overall, our findings highlight the potential of the BEMI protocol to serve as a model for stroke/telestroke transfer protocols. By improving critical time metrics, BEMI contributes to faster, more efficient stroke care, with the potential to enhance patient outcomes across diverse stroke systems.

Summary.

• The BEMI protocol significantly improved transfer (DIDO) and treatment times (TDGP) in our stroke network for telestroke transfers.

• Continued BEMI protocol usage shows sustainability for continued improvement of transfer and treatment times.

• BEMI may serve as a model for other stroke/telestroke system transfer protocols.