Table 6.
Diagnostic performance of ROSIER, FAST, LAPSS and GZSS
| Scale | Optimal cut-off | Sensitivity (95 % CI) | P | Specificity (95 % CI) | P | PLR (95 % CI) | NLR (95 % CI) | PPV (95 % CI) | NPV (95 % CI) | Diagnostic accuracy (%) |
|---|---|---|---|---|---|---|---|---|---|---|
| ROSIER | 0 | 77.7 (73.0–81.9) | 0.031 | 70.7 (57.3–81.9) | 0.754 | 2.65 (1.8–4.0) | 0.32 (0.2–0.4) | 94.2 (90.9–96.6) | 33.9 (25.5–43.1) | 76.68 |
| FAST | 0 | 76.0 (71.2–80.3) | 0.004 | 63.8 (50.1–76.0) | 0.210 | 2.1 (1.5–3.0) | 0.38 (0.3–0.5) | 92.8 (89.2–95.5) | 30.1 (22.1–39.0) | 74.28 |
| LAPSS | 0 | 56.4 (51.1–61.6) | <0.001 | 87.9 (76.7–95.0) | 0.008 | 4.68 (2.3–9.4) | 0.50 (0.4–0.6) | 96.7 (93.2–98.6) | 24.6 (18.9–31.1) | 60.82 |
| GZSS | 1.5 | 83.2 (79.0–87.0) | Reference | 74.1 (61.0–84.7) | Reference | 3.22 (2.1–5.0) | 0.23 (0.2–0.3) | 95.2 (92.2–97.3) | 41.7 (32.1–51.9) | 81.97 |
ROSIER Recognition of Stroke in the Emergency Room scale, FAST Face Arm Speech Test, LAPSS Los Angeles Pre-Hospital Stroke Screen, GZSS Guangzhou Stroke Scale, PLR positive likelihood ratio, NLR negative likelihood ratio, PPV positive predictive value, NPV negative predictive value