A Feasibility Study for Randomization of Code Stroke Imaging Strategies

The purpose of this study is to test feasibility of a comparative effectiveness framework for acute stroke imaging using prospective electronic health data. This is a prospective, cohort feasibility study of patients presenting to the Emergency Department with suspected acute ischemic stroke. The clinical stroke team will not be blinded to the imaging modality given the nature and purpose of the interventions/imaging. Knowledge of the imaging modality used and the knowledge gained from the resulting data will need to be considered for treatment decisions. Blinding will be maintained for data abstraction and analyses. Analysis will be on an "intent-to-scan" basis and all qualifying patients will be included in their assigned cohort.

All patients will receive standard-of-care therapy and management based on AHA/ASA clinical practice guidelines. For patients with emergent large vessel occlusions, this will include adjuvant endovascular therapy if criteria, as stipulated in the guideline, are met. Although this is referred to as a prospective study, no data collection will occur until after standard of care procedures.

Population: "Code-Stroke" (potentially-treatable) patients, defined as patients presenting with clinical syndrome suggestive of stroke within 24-hours of time last known well will be included.

Setting: four certified stroke centers (Ascension Seton Williamson, Ascension Seton Medical Center, Ascension Seton Hayes, and Dell Seton Medical Center)

Rationale: There is current equipoise in selection of first scan modality for code stroke. The time-dependent efficacy of treatment with thrombolytics increases diagnostic burden on the stroke team, particularly less experienced providers. In efforts to minimize door to needle time (DTN), stroke mimics may be inadvertently treated with thrombolytics, particularly when insensitive diagnostics are used. The investigators plan to examine feasibility of comparing CT-first versus MR-first strategies to guide the development of a future large multicenter comparison.

Methodology: The first-scan imaging modality will be assigned using two different schemes, a) clustered by hospital site and b) per-patient. Each scheme will be developed by a UT statistician or study team member, please see details in Randomization Schemes section below.

All practices will follow standard of care methods, no investigational procedures or interventions will occur as part of this study. A stroke alert is called prior to the arrival of the patient to the Emergency Department (ED). With this, patient information is communicated by the Emergency Medical Services team to the care team at the destination hospital to determine if the patient is potentially treatable (Code-Stroke).

Patient enrollment. Once a Code-Stroke is called, this will establish the patient eligible for the study and they will be enrolled. All stroke alerts are communicated via text message to the care team. The charge nurse leads a huddle prior to the arrival of each Code-Stroke patient. At the time the Code-Stroke is called, the first-scan imaging modality will be assigned and the patient will be enrolled.

Communication of assigned first-scan modality. The assigned first-scan imaging modality will be communicated via the ED charge nurse to all members of the care team including attending physicians, residents, and nursing staff. The mode of communication will be site-specific and include any of the following: verbal communication at initial pre-arrival stroke-alert huddle, sign posted at charge nurse desk that can be flipped to either "MRI" or "CT" typed in large letters, sign at scanners, sign at ED entrance. This will ensure the entire care team is aware and reminded of the first-scan modality assigned. This procedure was developed in close collaboration with Emergency Medicine attendings along with ED Medical and Nursing Directors at each participating hospital site.

Randomization Schemes: The prospective study includes two phases in which different randomization schemes will be evaluated for feasibility. Randomization schemes will be established a priori by the study Biostatistician. The Phases are as follows:

1. Multi-Center Hospital-Level Cluster Randomization - (Phase A) During the cluster randomized study period, everyone at the hospital is getting the same standard of care (diagnostic), the patient is not being randomized. The imaging modality used as first-scan for code stroke will be the hospital strategy and not determined on a per-patient basis. A randomization period equal to one-week in which the imaging modality will be either CT or MRI for all patients at the site. This will be implemented for a study length of 12 weeks. Within this 12-week period, 4-week blocks will be used to ensure balance in the assignment to either CT or MRI. Randomization schemes will be independent across the four participating sites and start will be staggered by one week.

This cluster randomization is designed to test feasibility of a comparative effectiveness framework using prospective electronic health data. This study design challenges the traditional research ethics paradigm. Rather than testing experimental, unproven interventions, the research question asks how interventions that are widely used compare, and deploys these interventions in ways consistent with good clinical practice at the hospital level.

1. Per-Patient Randomization - (Phase B) During the per-patient randomization, the first-scan imaging modality will be assigned to each eligible patient. A block size of 10 patients will be implemented to ensure balance in assignment to CT and MRI. The accrual of patients to meet this 10-patient block will be specific to each hospital site. Enrollment will be 5 weeks in duration. Randomization will be independent across sites. The assigned imaging modality will be communicated prior to each Code-Stroke and posted at the charge nurse station in the ED. With this, the entire stroke care team will know the assignment for the next Code-Stroke patient prior to their arrival.