Performance of the RACE Score for the Prehospital Identification of Large Vessel Occlusion Stroke in a Suburban/Rural EMS Service
Authors: Dickson RL, Crowe RP, Patrick C, et al.
Published in: Prehospital Emergency Care, 2019 Jan 22.
It has been estimated that large-vessel occlusion strokes account for 24% of all strokes, and it’s well known that treating LVO strokes with endovascular therapy leads to reduced disability and improved functional independence. Therefore, it’s important to appropriately screen potential stroke patients, identify LVO early, and transport to an appropriate stroke treatment center that can perform endovascular therapy.
However, few stroke screening tools have been tested in the prehospital environment. The Rapid Arterial Occlusion Evaluation (RACE) scale was created to identify LVO stroke patients and validated in prehospital trials in Spain.
We’ve discussed previously that one study alone does not typically provide enough evidence to “prove” anything. All studies should be replicated. While the RACE scale had been validated elsewhere, it had not been tested in the United States. The authors of this month’s study recently published a manuscript assessing the prehospital performance of the RACE scale for identification of patients experiencing LVO stroke following implementation at a large suburban/rural agency in Texas.
This was a retrospective study that included data for all patients 16 and older who presented with a provider’s primary or secondary impression of stroke from June 1, 2016 to November 1, 2017. All ALS personnel received four hours of mandatory RACE scale training, and in April 2016 the RACE scale was incorporated into their protocols for suspected stroke patients. Protocols specified that after an initial positive stroke screening, providers performed the RACE scale, and if the RACE scale score was 5 or more, the patient was transported directly to a comprehensive stroke center. The initial stroke scale was the Southeast Texas Assessment and Transport Stroke (STATS) tool, which was historically used in the region.
The study linked ePCR data with hospital data from six receiving facilities in the coverage area. Data obtained from the ePCR included sex, age, race, the EMS provider’s primary and secondary impressions, time of last known normal, and the documented RACE score. Hospital outcome data included whether the hospital physician believed the patient experienced an LVO stroke. Records without hospital outcome data were excluded from the analysis.
Before we get to the study results, we’ll need a brief discussion about predictive values. To evaluate the performance of the RACE scale, the authors calculated a sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV). These measures are not as complicated as they sound.
Sensitivity, specificity, PPV, and NPV are all reported as percentages. Sensitivity is the ability of a test to correctly identify those with the disease. So in this case the sensitivity would be the ability of the RACE scale to correctly predict LVO. Specificity is the converse: the ability of a test to correctly identify those without the disease, or here to rule out those without LVO. PPV is the percentage of patients with a positive test who actually have the disease, and NPV is the percentage of patients with a negative test who actually do not have the disease.
There is some nuance needed to understand these differences. For example, we all know that just because a test result comes out positive doesn’t necessarily mean the patient has the disease. While sensitivity tells us the probability a test result is positive when the patient has the disease, the PPV tells us the probability that the patient has the disease when the test is positive. In the same way, specificity tells us the probability a test result will be negative when the patient does not have the disease, and NPV tells us the probability the patient does not have the disease when the test is negative.
Now let’s get to the results. During the study period there were 505 patients with a documented primary or secondary impression of stroke. Of these 440 were included in the analysis. The 65 who were excluded were omitted because their record did not have a RACE score documented, did not have hospital outcome data, or the patient was under 16. Of the 440 included in the analysis, 50.7% were female, and the median age was 70 years.
More than three-quarters (76%) of patients had a last known normal time of 4.5 hours or less, and 21% had a last known normal time that exceeded 6 hours. There were 61% (269/440) who had a hospital diagnosis that indicated stroke and 15% (64/440) who had a diagnosis indicating LVO. LVO was present in 40% (64/160) of patients diagnosed with ischemic stroke.
There were 33% (146/440) of patients with a RACE score of 5 or more. Of those, 73% (107/146) were transported directly to a comprehensive stroke center. The strongest predictive value for LVO was a RACE score of 5 or more, with a sensitivity of 66%, specificity of 72%, PPV of 29%, and NPV of 93%. The results of this study were similar to those from the initial validation study performed in Spain, but the PPV (29% vs. 42%) and sensitivity (66% vs. 85%) were lower.
Because this study was performed in a single, large county-based EMS system, the results may not be generalizable to other types of agencies or settings. One specific difference the authors noted was the racial/ethnic composition of patients, with most in this study being white/non-Hispanic. The authors also noted they were unable to obtain data on the performance of vessel imaging, results of vessel imaging, and administration of tPA.
This a well-written and interesting study that adds to the relevant literature on stroke, LVO, and stroke screening tools. Read the entire manuscript; there are many additional analyses that will enhance your understanding.
Antonio R. Fernandez, PhD, NRP, FAHA, is research director at the EMS Performance Improvement Center and an assistant professor in the Department of Emergency Medicine at the University of North Carolina–Chapel Hill. He is on the board of advisors of the Prehospital Care Research Forum at UCLA.