Table 2.
Sensitivities and specificities of single prognostic methods as recommended by ERC/ESICM and variations thereof
| Method | Sensitivity (95% CI) | Specificity (95% CI) | TP | TN | FP | FN | N= | Poor outcome |
|---|---|---|---|---|---|---|---|---|
| GCS-M ≤ 2 | 71.8 (66.1–76.9) | 95.6 (92.8–97.4) | 191 | 305 | 14 | 75 | 585 | 266 (45.5) |
| GCS-M ≤ 3 | 77.1 (71.7–81.7) | 92.8 (89.4–95.2) | 205 | 296 | 23 | 61 | 585 | 266 (45.5) |
| GCS-M ≤ 4 | 85.7 (81–89.4) | 83.7 (79.3–87.4) | 228 | 267 | 52 | 38 | 585 | 266 (45.5) |
| PR/CR | 20.1 (15.6–25.4) | 100 (92.4–100) | 51 | 47 | 0 | 203 | 301 | 254 (84.4) |
| SSEP | 45.3 (37.9–53.1) | 97.4 (86.8–99.6) | 73 | 38 | 1 | 88 | 200 | 161 (80.5) |
| NSE ≥ 33*/** | 67.3 (61.9–72.3) | 89.9 (86.2–92.7) | 208 | 303 | 34 | 101 | 646 | 309 (47.8) |
| NSE ≥ 48*/≥ 38** | 60.2 (54.6–65.5) | 96.4 (94–98) | 186 | 325 | 12 | 123 | 646 | 309 (47.8) |
| EEG ERC/ESICM | 31.7 (25.9–38.1) | 98.8 (93.6–99.8) | 70 | 83 | 1 | 151 | 305 | 221 (72.5) |
| EEG “highly malignant” | 38 (31.9–44.6) | 98.8 (93.6–99.8) | 84 | 83 | 1 | 137 | 305 | 221 (72.5) |
| S. Myoclonus ≤ 48 h | 6.9 (5–9.5) | 99.8 (98.7–100) | 34 | 439 | 1 | 459 | 933 | 493 (53.8) |
| CT | 32.3 (26.7–38.4) | 98.3 (94.2–99.6) | 76 | 119 | 2 | 159 | 356 | 235 (66) |
| MRI | 13 (4.5–32.1) | 100 (75.8–100) | 3 | 12 | 0 | 20 | 35 | 23 (65.7) |
Prognostic accuracies of methods using all available results in the entire TTM-cohort with 6-month outcome (n = 933). The cohort is described in the right column in Table 1 and in eFig. 1. Results presented in numbers or in percentages with 95% confidence intervals. Only patients with available results were included in the statistical analyses. Poor neurological outcome was defined as Cerebral Performance Category Scale 3–5 at 6 months. GCS-M, Glasgow Coma Scale Motor Score on day 4 after cardiac arrest; PR/CR, bilaterally absent pupillary light reflexes and bilaterally absent corneal reflexes; SSEP; bilaterally absent N20 potentials on short-latency somatosensory evoked potentials; NSE, serum neuron-specific enolase in pg/mL * at 48 h and/or ** at 72 h post-arrest; EEG ERC/ESICM, unreactive burst-suppression or unreactive status epilepticus (abundant rhythmic/periodic discharges); EEG “highly malignant”, suppressed background with or without periodic discharges or burst-suppression with or without discharges; S. Myoclonus, presence of early status myoclonus > 30 min ≤ 48 h after cardiac arrest; CT, visually evaluated generalized oedema seen as a reduced differentiation between grey and white matter by local radiologists; MRI, presence of generalized oedema on magnetic resonance imaging. TP, true positive (predicted and reported poor outcome); TN, true negative (predicted and reported good outcome); FP, false positive (predicted poor outcome, reported good outcome); FN, false negative (predicted good outcome, reported poor outcome)