Table 1.
Characteristics of prospective validation studies published for the highest-quality clinical prediction rules
| Setting | Populationa | Outcome prevalence | Performance | Limitations/Comments | |
|---|---|---|---|---|---|
| PECARN [3] | |||||
| Kuppermann 2009 [3] | 25 PED in the US | 8627 patients (2216 < 2y; 6411 > 2y) Same as original rule |
ciTBI = 88 (25 < 2y; 63 > 2y) |
< 2y SENS 100 (95% CI 86.3–100) SPEC 53.7 (95% CI 51.6–55.7) NPV 100 (95% CI 99.7–100) PPV 2.4 (95% CI 1.6–3.5) > 2y SENS 96.8 (95% CI 89.0–99.6) SPEC 58.2 (95% CI 57.0–59.4) NPV 99.95 (95% CI 99.80–99.99) PPV 2.2 (95% CI 1.7–2.9) |
Internal validation |
| Schonfeld 2014 [95] | 1 PED in the US; 1 PED in Italy |
2428 patients (956 < 2y; 1472 > 2y) Same as original rule |
ciTBI = 19 (6 < 2y; 13 > 2y) Positive CT = 69 |
< 2y SENS 100 (95% CI 64.3–100) SPEC 43.2 (95% CI 40–46.3) PPV 1.7 (95% CI 0.6–3.2) NPV 100 (95% CI 99.4–100) > 2y SENS 100 (95% CI 79.4–100) SPEC 48.3 (95% CI 45.8–50.9) PPV 2.0 (95% CI 1.1–3.2) NPV 100 (95% CI 99.8–100) |
- Low number of study outcome (n = 19; NS = 0) - 17.5% of record retrospectively collected following prospective implementation of the rule - Incomplete clinical follow up |
| Easter 2014 [96] | 1 PED in the US | 1009 patients Included GCS = 13 (0.4%) |
ciTBI = 21 | SENS 100 (95% CI 84–100) SPEC 62 (95% CI 59–66) LR+ 2.7 (95% CI 2.5–2.9) LR- 0 (95% CI 0-undefined) |
Included patients with GCS = 13 (but only 4 patients) Only overall rule performance reported |
| CATCH [93] | |||||
| Osmond 2012 [98] | 9 PED in Canada | 4060 patients Same as original rule |
Injury requiring NS = 23 Acute brain injury on CT = 197 |
For the four high risk factors - for injuries needing NS SENS 87 (95% CI 68–98) SPEC 87 (95% CI 86–88) PPV 3.6 (95% CI 2.3–5.5) NPV 99.9 (95% CI 99.8–100) For the 7 CATCH predictors - for acute brain injury on CT SENS 98 (95% CI 95–99) SPEC 65 (95% CI 64–67) PPV 12.7 (95% CI 11.1–14.4) NPV 99.8 (95% CI 99.6–99.9) |
Abstract form only. (Complete accuracy measures calculated based on the numbers provided to the authors to the NICE head injury guideline working group – see Table 15 of the NICE guideline appendices) Predicted CT rate: 14% for identifying injuries that require neurological intervention; 38% for acute brain injury on CT |
| Easter 2014 [96] | 1 PED in the US | 1009 patients not selected based on CATCH symptoms; presenting within 24 h |
Injury requiring NS = 4 Any injury on CT = 52 |
For injury requiring NS SENS 75 (95% CI 19–99) SPEC 43 (95% CI 40–46) LR+ 1.3 (95% CI 0.7–2.3) LR- 0.6 (95% CI 0.1–3.2) For any injury on CT SENS 90 (95% CI 79–97) SPEC 45 (95% CI 42–48) LR+ 1.6 (95% CI 1.5–1.8) LR- 0.4 (95% CI 0.3–0.6) |
Different inclusion criteria compared with original rule Outcome not exactly as in original rule |
| CHALICE [92] | |||||
| Easter 2014 [96] | 1 PED in the US | 1009 patients up to 18 years of age; presenting within 24 h | Injury requiring NS = 4 Any injury on CT = 52 |
For injury requiring NS SENS 75 (95% CI 19–99) SPEC 84 (95% CI 81–86) LR+ 4.5 (95% CI 2.5–8.2) LR- 0.3 (95% CI 0.1–1.6) For any injury on CT SENS 64 (95% CI 47–79) SPEC 86 (95% CI 83–88) LR+ 4.4 (95% CI 3.3–5.9) LR- 0.4 (95% CI 0.3–0.6) |
Different age limit compared with orginal rule Outcomes not exactly as in original rule |
ciTBI = clinically important traumatic brain injury (as per PECARN definition); LR + = positive likelihood ratio; LR- = negative likelihood ratio; NS = neurosurgery; PED = Pediatric Emergency Department; SENS = sensitivity; SPEC = specificity; US = United States
adifference with the original CPR population are reported where appropriate