Danny R. Rose, Jr. MD
Boulouis G, Siddiqui K, Lauer A, Charidimou A, Regenhardt R, Viswanathan A, et al. Immediate Vascular Imaging Needed for Efficient Triage of Patients With Acute Ischemic Stroke Initially Admitted to Nonthrombectomy Centers. Stroke. 2017
The landmark publication of multiple positive endovascular thrombectomy (EVT) trials in 2015 was a pivotal moment for treatment of acute ischemic stroke. The most significant development in acute stroke treatment in the nearly twenty years since the FDA approval of tissue plasminogen activator in 1996 has led to much discussion with respect to improving stroke systems of care to be able to provide this treatment to as many eligible patients as possible. Reflecting this new development in acute stroke treatment, the American Heart Association released a focused update to their guidelines on acute stroke treatment that recommended endovascular therapy be offered to patients who present within 6 hours of last known normal and have a favorable imaging profile and a National Institutes of Health Stroke Scale (NIHSS) of 6 or greater.
Just as the time-sensitive nature of intravenous thrombolytic administration led to the development of prehospital stroke scales and the stroke alert process, the most effective way to triage and treat patients with suspected emergent large vessel occlusions (LVO) amenable to endovascular treatment is a topic of ongoing research and debate. An important facet of this discussion concerns the most effective method to triage and transfer patients with suspected LVO to a thrombectomy-capable stroke center. A cohort by Sarraj et al. presented at the 2017 International Stroke Conference showed comparably good outcomes for patients transferred to thrombectomy-capable centers as compared to patients who presented directly to the facility, suggesting that the “drip and ship” transfer paradigm can be successfully augmented to accommodate endovascular therapy.
One of the unique challenges this presents is finding the most efficient method to rapidly identify patients who could potentially benefit from endovascular intervention. Many prehospital scales have been developed or tailored to identify the subset of acute ischemic stroke patients that may have large vessel occlusions (RACE, CPSSS, PASS, etc.). This is an integral first step in the process of overhauling stroke systems of care to accommodate endovascular treatment and may be the most useful triage tool for non-thrombectomy-capable centers who also lack access to rapid vascular imaging. This method of triage and transfer carries inherent inefficiencies as it does not definitively identify the presence or absence of LVO, resulting in over-triage and futile transfers for a subset of identified patients.
A study by Boulouis et al. published in the August issue of Stroke sought to investigate the clinical and imaging factors associated with greater likelihood of endovascular thrombectomy (EVT) for patients transferred to their thrombectomy-capable center from referring hospitals and the potential effect of implementing baseline vascular imaging prior to transfer. Their data was obtained from a cohort of patients transferred from 30 referring hospitals from 2010–2016. They developed mathematical models to assess the sensitivity, specificity, positive (PPV) and negative predictive values (NPV) for triage using criteria for the “base case” (patients who meet AHA guidelines for consideration for endovascular treatment), as well as two additional models taking into account LVO as identified using the dense vessel sign on baseline noncontrast head CT imaging (model 1), as well as a model looking at all patients who were found to have evidence of LVO on vascular imaging after transfer (model 2).
Applying the AHA guideline criteria resulted in a sensitivity of 92%, specificity of 53%, PPV of 28% and NPV of 93%. Using both models in a simulation taking into account varying minimum NIHSS cutoff values, the best sensitivity and specificity trade-off was found using model 2 and a minimum NIHSS value of 9 (sensitivity=91%; 95% CI, 0.83–0.95; and specificity=80%; 95% CI, 0.75–0.83). Unsurprisingly, the more restrictive model using the dense vessel sign slightly improved specificity at the cost of a significant reduction in sensitivity. Their model suggests that the implementation of baseline vascular imaging would result in the prevention of 2 futile transfers and 5 EVT procedures for every 10 patients transferred for EVT evaluation, roughly doubling the likelihood that a transferred patient will be treated with EVT as compared to the “base case” model.
This study clearly shows the utility of implementing baseline vascular imaging to improve the efficiency of triage in acute ischemic stroke patients. Given the wide disparity of resources in stroke care systems nationwide, the limitation of this study being conducted from one center is an important one to consider. Additional considerations include the potential delay in treatment and subsequent worsening of outcomes that could result from obtaining vascular imaging in a referring hospital where such processes may be less streamlined as compared to thrombectomy-capable centers. Quality improvement research investigating baseline vascular imaging is ongoing and will likely result in significant changes in stroke triage, both at referring and thrombectomy-capable centers. Thrombectomy-capable centers and their referring hospitals will need to work together in order to develop an efficient method of LVO triage that reduces futile transfers while maintaining a high sensitivity without causing significant delays in treatment.