Table 2.
Overall diagnostic performance of chest radiography and CT for rib fracture detection
| True positive/false positive | False negative/true negative | Sensitivity (%, 95% CI) | Specificity (%, 95% CI) | Positive predictive value (%, 95% CI) | Negative predictive value (%, 95% CI) | |
|---|---|---|---|---|---|---|
| Per fracture location | ||||||
| Chest radiography (n=68 400) | 912/2391 | 4256/60 841 | 13·5% (8·1 to 21·5) | 97·9% (96·8 to 98·7) | 7·2% (1·9 to 23·6) | 99·4% (97·0 to 99·9) |
| CT (n=63 000) | 2089/3129 | 2671/55 111 | 44·9% (31·7 to 58·9) | 97·0% (95·3 to 98·0) | 12·0% (3·3 to 35·1) | 99·6% (98·8 to 99·9) |
| Difference | .. | .. | 31·4% (23·3 to 37·8; p<0·001) | −0·9% (−1·4 to −0·6; p<0·001) | 4·8% (1·2 to 11·9 p<0·05) | 0·2% (0·0 to 1·0; p≥0·05) |
| Per rib | ||||||
| Chest radiography (n=22 800) | 1579/1410 | 3323/16 488 | 23·1% (12·9 to 37·8) | 96·4% (94·1 to 97·8) | 15·9% (3·2 to 52·1) | 98·1% (91·3 to 99·6) |
| CT (n=21 000) | 2713/1886 | 1801/14 599 | 62·4% (44·9 to 77·1) | 94·1% (90·5 to 96·3) | 18·8% (3·9 to 56·9) | 98·9% (94·8 to 99·8) |
| Difference | .. | .. | 39·3% (31·9 to 42·2; p<0·001) | −2·3% (−3·7 to −1·4; p<0·001) | 2·9% (0·1 to 7·4 p<0·05) | 1·5% (0·1 to 3·3p<0·05) |
| Per case | ||||||
| Chest radiography (n=950) | 405/155 | 241/149 | 64·7% (57·3 to 71·4) | 48·9% (41·6 to 56·2) | 72·3% (68·5 to 75·9) | 38·2% (33·5 to 43·1) |
| CT (n=875) | 465/141 | 130/139 | 81·5% (75·8 to 86·0) | 49·3% (41·8 to 56·9) | 76·7% (76·7 to 73·2) | 51·7% (45·7 to 57·6) |
| Difference | .. | .. | 16·7% (11·5 to 22·2; p<0·001) | 0·4% (−7·0 to 9·0; p≥0·05) | 4·4% (−0·4 to 9·3; p≥0·05) | 13·5% (6·0 to 21·4; p<0·001) |
Data are shown with 95% CI and p values when available. Difference in diagnostic statistics calculated as CT minus chest radiography. The positive and negative results are the total frequencies across all reporters, cases, and locations—eg, 68 400 rib locations (anterior, lateral, and posterior) were included across all reporters on chest radiographs (three locations per rib × 24 ribs × 25 cases × 38 reporters) and 63 000 rib locations across 35 reporters on CT. 136 locations had fractures on autopsy and so a total of 136 × 38 reporters=5168 fractures could be detected on the radiographs; and 136 locations × 35 reporters=4760 fractures on CT images. Estimates are derived from multilevel models and therefore differ from the raw estimates that could be calculated from number of true positive, false positive, false negative, and true negative events.