"Does advanced imaging improve the reliability of 2018 AO/OTA classification for trochanteric femoral fractures?"

Background: The influence of advanced imaging modalities on the reliability of 2018 AO/OTA classification and detection of features contributing to fracture instability have not been adequately studied in the literature.

Methods: This prospective study was conducted to assess the reliability of 2018 AO/OTA classification for trochanteric femoral fractures, and features of instability in 50 patients using radiographs, multiplanar reformats (MPR), and 3D-reconstruction with fracture segmentation (3DR-FS) by 3 observers on 2 occasions at an interval of 4 weeks.

Results: Mean inter-observer reliabilities on radiographs, MPR, 3DR-FS were 0.36, 0.39, 0.46 after first reading, and 0.35, 0.35, 0.44 after second reading respectively. The mean intra-observer reliabilities for radiographs, MPR, 3DR-FS were 0.36, 0.43, and 0.50 respectively. Inter-observer reliabilities on subgroup analysis were 0.35, 0.27, 0.40 for subgroup A1 and 0.19, 0.22, and 0.30 for subgroup A2 on radiographs, MPR and 3DR-FS respectively. All results were statistically significant (p-value <0.05). Posterior comminution was detected in 125(41 %), 247(82 %), and 255 (85 %) out of total 300 instances on radiographs, MPR and 3DR-FS respectively. Coronal split was detected in 113(37 %), 189(63 %), and 242 (80 %) out of total 300 instances on radiographs, MPR, and 3DR-FS respectively. Flexion of the proximal fragment was detected in 106(35 %), 163(54 %), and 180 (60 %) out of total 300 instances on radiographs, MPR, and 3DR-FS respectively. Lateral wall fracture was detected in 45(15 %), 80(26 %), and 138 (46 %) out of total 300 instances on radiographs, MPR, and 3DR-FS respectively. The banana fragment was detected in 5(1 %), 16(5 %), and 83 (27 %) out of a total of 300 instances on radiographs, MPR, and 3DR-FS respectively.

Conclusions: 2018 AO/OTA classification for trochanteric femoral fractures has 'fair' reliabilities on radiographs and MPR that improve to 'moderate' on adding 3DR-FS. We attribute it mostly to the difficulty in localizing innominate tubercle. Adding advanced imaging helps better detect fracture instabilities.

Level of evidence: Level I, Diagnostic.