RESPONSE BY LIMA ET AL TO LETTER BY HELDNER ET AL REGARDING ARTICLE, “FIELD ASSESSMENT STROKE TRIAGE FOR EMERGENCY DESTINATION: A SIMPLE AND ACCURATE PREHOSPITAL SCALE TO DETECT LARGE VESSEL OCCLUSION STROKES”

In Response

We would like to thank Heldner et al. for taking an interest on our work and applying the FAST-ED scale to their large patient cohort. Triage for stroke centers will undergo major changes given the recently proven benefit of endovascular approaches in the treatment of large vessel occlusion strokes (LVOS). The FAST-ED scale was designed to aid in the decision to triage patients to primary (PSC) versus comprehensive stroke centers (CSC) in the pre-hospital setting. In STOPStroke (n=727; 33% LVOS), FAST-ED had comparable accuracy to predict LVOS to the NIHSS and higher accuracy than RACE and CPSS (AUC: FAST-ED=0.81 as reference; NIHSS=0.80, p=0.28; RACE=0.77, p=0.02; and CPSS=0.75, p=0.002)¹. Heldner et al. independently corroborated these findings in the Bernese cohort (n=1085; 60.5% LVOS; AUC: FAST-ED=0.847, NIHSS=0.846, RACE=0.831, CPSSS=0.802)². In comparison to RACE, FAST-ED scores only one point for facial weakness and no points for leg weakness. This lowers the chances of a pure motor stroke (~85% lacunar in etiology) being diagnosed as a LVOS³. Unlike RACE, FAST-ED scores up to 2 points for gaze deviation, which is one of the most powerful predictors of LVOS^{2, 4}. Unlike CPSSS, FAST-ED evaluates for neglect and thus better detects non-dominant hemisphere LVOS. Unlike RACE and CPSSS, FAST ED separately assesses expressive and receptive aphasia, optimizing topographic cortical representation and consequently LVOS detection.

We fully agree with Heldner et al. that clinical severity scales cannot identify LVOS with 100% accuracy and as such cannot replace vascular neuroimaging. However, we believe that the applicability of these scales resides on using the best available field data to optimize the destination triage of stroke patients. From a pragmatic standpoint, it is important to acknowledge that the mere presence of a LVOS does not equate to thrombectomy. In fact, it remains unproven whether low NIHSS LVOS benefit from endovascular therapy⁵. Furthermore, lower clinical severity LVOS presumably have more robust collateral flow (and consequently better natural history and longer endovascular treatment windows) and/or relatively more distal occlusions (and consequently better response to IV t-PA). Therefore, it seems rationale that most of such patients should be first evaluated and treated at the closest stroke center prior to be considered for transportation to more distant centers, as in this population delaying the administration of IV t-PA would likely cause more harm than benefit. It is also critical to acknowledge that the reasonable amount of time delay to IV t-PA in favor of mechanical thrombectomy remains largely empiric. As such, the development of smartphone apps to compute scales and algorithms to assist decisions in the field and their subsequent prospective validation seem like next logical steps in the optimization of stroke systems of care.

Finally, while we understand Heldner’s comments regarding rural vs. urban differences in triage may very well apply to their region, it is essential to consider the broad spectrum in terms of population density, road traffic, air transport availability, and distances to primary versus comprehensive centers across different regions and countries. Therefore, we believe clinical severity scales will still have an important role in many if not most locations.