Article Text

Download PDFPDF

E-063 Primary results from the predict (pre-hospital emergency LVO detection during initial care transport) study: a prospective, multi-center, non-inferiority comparison between the novel PREDICT composite 4-item scale and the rapid arterial occlusion evaluation (RACE) scale to detect large-vessel occlusion (LVO) strokes when administered by us-based ems personnel to suspected stroke patients at the initial pre-hospital encounter
  1. R James1,
  2. A Cruz1,
  3. E Fortuny1,
  4. B Ugiliweneza1,
  5. D Wang1,
  6. A White2,
  7. N Khattar1,
  8. S Adams1,
  9. B Gallinore1,
  10. D Ding1,
  11. S Wolfe3,
  12. D Heck4
  1. 1Neurosurgery, University of Louisville, Louisville, KY
  2. 2Radiology, University of Louisville, Louisville, KY
  3. 3Neurosurgery, Wake Forest Baptist Health, Lexington, NC
  4. 4Radiology, Forsythe Medical Center, Winston-Salem, NC


Introduction Accurate identification of large vessel occlusion (LVO) strokes in the pre-hospital setting is imperative to reduce time to thrombectomy and improve outcomes. Prior to this study, RACE was the only stroke severity scale designed to detect LVO that had been prospectively validated with Emergency Medical Services (EMS). The goal of our non-inferiority study is to prospectively validate and compare the novel PREDICT-3 item scale, PASS scale, and a composite of these two scales: the PREDICT Composite 4-item (PREDICT-4) scale to the RACE scale in detecting LVO in the prehospital setting.

Methods Adult patients suspected of having a stroke by EMS and transported to a participating comprehensive stroke center had both PREDICT-4 and RACE scales administered prospectively and recorded in a secure web-based database. Admission NIHSS score and final diagnosis were also recorded. Cerebrovascular imaging studies (CTA, MRA or DSA) were reviewed by a blinded, independent neuroradiologist to determine LVO diagnosis. LVO was defined as occlusion of intracranial internal carotid artery (ICA), middle cerebral artery (MCA) M1 portion, or proximal aspect of single or multiple M2 branches of MCA, or basilar artery (BA). We used SAS and c-statistics to create receiver-operating characteristic (ROC) curves to determine the area under the curve (AUC) and optimal score cut point (CP) for each scale. We also calculated the sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV) and overall accuracy for each optimal CP score.

Results 250 subjects were enrolled, 17 were excluded for lack of neurovascular imaging, 1 was excluded for incomplete stroke scale data, leaving 232 subjects who met inclusion criteria for analysis. 31 subjects had LVO and the prevalence of LVO was 13.4%. In table 1, we report the performance of each stroke severity scale at their respective optimal CP. In our study, the optimal CP for RACE was ≥6 while the original RACE study was ≥5.

Abstract E-063 Figure 1 A Receiver Operating Characteristic (ROC) curve illustrating the sensitivity, specificity and area under the curve of the PREDICT-3 scale. 1B Receiver Operating Characteristic (ROC) curve illustrating the sensitivity, specificity and area under the curve of the RACE Scale
Abstract E-063 Table 1 Comparing accuracy of the individual scales performed in this trial

Conclusions This large, US-based, prospective, pre-hospital, EMS-administered study showed greater AUC for RACE. However, all three scales were non-inferior in AUC with respect to RACE. These newer, simpler scales may have acceptable diagnostic ability for LVO and be more easily adopted by EMS.

Disclosures R. James: 1; C; Medtronic, Medtronic. A. Cruz: None. E. Fortuny: None. B. Ugiliweneza: None. D. Wang: None. A. White: None. N. Khattar: None. S. Adams: None. B. Gallinore: None. D. Ding: None. S. Wolfe: None. D. Heck: None.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.