Pooled estimates of diagnostic accuracy of makers for diagnosis of appendicitis in children
| Marker | Studies | Threshold | Pooled sensitivity* (95% CI) | Pooled specificity* (95% CI) | Hypothetical cohort of 1000 children with appendicitis prevalence of 40%† | Hypothetical cohort of 1000 children with appendicitis prevalence of 5% | ||
| Missed appendicitis (FN) | Unnecessary treatment (FP) | Missed appendicitis (FN) | Unnecessary treatment (FP) | |||||
| WCC‡ (cells/μL) | 53 | 10 000 | 0.85 (0.80 to 0.89) | 0.58 (0.51 to 0.64) | 60 | 252 | 8 | 399 |
| 12 000 | 0.73 (0.66 to 0.79) | 0.71 (0.65 to 0.76) | 108 | 174 | 14 | 276 | ||
| 15 000 | 0.53 (0.45 to 0.62) | 0.83 (0.78 to 0.87) | 188 | 102 | 24 | 162 | ||
| CRP§ (mg/L) | 38 | 5 | 0.77 (0.69 to 0.84) | 0.54 (0.45 to 0.63) | 92 | 276 | 12 | 437 |
| 10 | 0.68 (0.59 to 0.76) | 0.66 (0.58 to 0.73) | 128 | 204 | 16 | 323 | ||
| 30 | 0.50 (0.39 to 0.60) | 0.81 (0.75 to 0.87) | 200 | 114 | 25 | 181 | ||
| 50 | 0.41 (0.30 to 0.53) | 0.87 (0.80 to 0.91) | 276 | 78 | 35 | 124 | ||
| Neutrophils¶ (% of lymphocytes) | 12 | 75% | 0.76 (0.73 to 0.80) | 0.63 (0.57 to 0.69) | 96 | 222 | 12 | 352 |
| ANC (cells/μL) | 5 | <7500 | 0.90 (0.85 to 0.94) | 0.54 (0.49 to 0.58) | 40 | 276 | 5 | 437 |
| 13 | 7500 | 0.86 (0.82 to 0.89) | 0.60 (0.54 to 0.66) | 56 | 240 | 7 | 380 | |
| 6 | >7500 | 0.70 (0.65 to 0.75) | 0.78 (0.71 to 0.85) | 120 | 132 | 15 | 209 | |
| Procalcitonin (ng/mL) | 6 | 0.1–0.5 | 0.30 (0.26 to 0.35) | 0.83 (0.73 to 0.89) | 280 | 102 | 35 | 161 |
| WCC and CRP** | 6 IPD studies | 10 000 and 10 | 0.59 (0.49 to 0.68) | 0.79 (0.74 to 0.84) | 164 | 126 | 21 | 200 |
| WCC and CRP | 6 IPD studies | 10 000 and 50 | 0.27 (0.19 to 0.37) | 0.93 (0.91 to 0.95) | 292 | 46 | 37 | 67 |
| WCC or CRP | 6 IPD studies | 10 000 and 10 | 0.97 (0.93 to 0.99) | 0.40 (0.30 to 0.51) | 12 | 360 | 2 | 570 |
| WCC or CRP | 6 IPD studies | 10 000 and 50 | 0.91 (0.86 to 0.95) | 0.50 (0.40 to 0.60) | 36 | 300 | 5 | 475 |
Six studies could not be used in the pooled analysis, because 5 studies reported WCC provided age-dependent cut offs and 1 study reported pain duration dependent cut offs.
*All pooled sensitivity and specificity were obtained from the bivariate model, except for WCC and CRP for which we calculated sensitivity and specificity for selected cut offs using the Steinhauser method, comparison with bivariate mode including likelihood ratios for all tests is demonstrated in online supplemental table 5.
†Median prevalence of appendicitis in the included 62 cohort studies is 40%.
‡WCC was evaluated in 59 studies whereof 53 were included in the meta-analysis whereof 12 provided multiple cut offs.
§CRP was evaluated in 39 studies whereof 11 provided multiple cut offs.
¶Neutrophils evaluated in 16 studies of which 12 provided cut off at 75%.
**WCC and CRP calculated from IPD from 6 studies (n=1541 participants) and analysed using the bivariate model).
ANC, absolute neutrophil count; CRP, C reactive protein; FN, False Negatives; FP, False Positives; IPD, individual patient data; WCC, white cell count.