ABSTRACT
BACKGROUND: Use of prehospital stroke scales may enhance stroke detection and improve treatment rates and delays. Current scales, however, may lack detection accuracy. As such, we examined whether adding coordination (Balance) and diplopia (Eyes) assessments increase the accuracy of the Face-Arms-Speech-Time (FAST) scale in a multisite prospective study of emergency response activations for presumed stroke. METHODS: This was a prospective study of emergency response activations for presumed stroke in Santa Clara County, California. Emergency medical responders were trained in the Balance-Eyes-Face-Arms-Speech-Time (BEFAST) scale and administered the scale on scene to all patients who were within 6 hours of onset of neurological symptoms. Each patient's final diagnosis (stroke vs. no stroke) was based on review of hospital records. We compared the performance of the BEFAST and FAST scales for stroke detection. RESULTS: Three hundred fifty-nine patients were included in our analysis. Compared to nonstroke patients (n = 200), stroke patients (n = 159) more often scored positive on each of the five elements of the BEFAST scale (p < 0.05 for each). In multivariable analysis, only facial droop and arm weakness were independent predictors of stroke (p < 0.05). BEFAST and FAST scale accuracy for stroke identification was comparable (area under the curve [AUC] = 0.70 vs. AUC = 0.69, p = 0.36). Optimal cutoff for stroke detection was ≥1 for both scales. At this threshold, the positive predictive value (PPV) was 0.49 for the BEFAST and 0.53 for the FAST scale, and the negative predictive value (NPV) was 0.93 for BEFAST and 0.86 for FAST. CONCLUSION: Adding coordination and diplopia assessments to face, arm, and speech assessment does not improve stroke detection in the prehospital setting.