Long Bone Union Accurately Predicted by Cortical Bridging within 4 Months

BACKGROUND: Previous retrospective research found that the presence or absence of bridging callus within 4 months postoperatively discriminated accurately between eventual union and nonunion of fractures of the tibial shaft. However, there remains no consensus regarding early prognostication of long bone nonunion. We prospectively assessed the accuracy and reliability of the presence of any bridging callus within 4 months in a cohort that was expanded to include both tibial and femoral shaft fractures. METHODS: We identified 194 consecutive fractures of the shaft of the tibia (OTA/AO type 42-A, B, or C) and femur (OTA/AO type 32-A, B, or C) that were treated with intramedullary nailing. Exclusions for inadequate follow-up (55), extended delay prior to nailing (10), and skeletal immaturity (3) resulted in a study population of 126 fractures (56 tibiae and 70 femora) in 115 patients. Digital radiographs made between 3 and 4 months postoperatively were independently assessed by 3 orthopaedic traumatologists. The accuracy of assessment of the presence of any bridging callus, bicortical bridging, and tricortical bridging to predict union or nonunion was assessed with chi-square analysis and by interobserver reliability (kappa statistic). RESULTS: The nonunion rate was 4% (5 of 126 fractures). The presence of any bridging callus by 4 months accurately predicted union (121 of 122 fractures) and its absence predicted nonunion (4 of 4 fractures). There was 1 incorrect prediction of union for a fracture that failed to unite (p < 0.001). Bicortical or greater bridging predicted union when present (116 of 116 fractures) and nonunion when absent (5 of 10 fractures), incorrectly predicting that 5 healing fractures would go on to nonunion (p < 0.001). Tricortical or greater bridging predicted union when present (103 of 103 fractures) and nonunion when absent (5 of 23 fractures), incorrectly predicting that 18 healing fractures would go on to nonunion (p < 0.001). Interobserver reliability was calculated for any bridging (kappa value, 0.91), bicortical bridging (kappa value, 0.79), tricortical bridging (kappa value, 0.71), and the exact number of cortices bridged (kappa value, 0.67). CONCLUSIONS: The presence of any bridging callus within 4 months accurately predicts the final healing outcome for tibial and femoral shaft fractures treated with intramedullary nailing. This criterion is simple and reliable, and only standard radiographs are needed to make the determination. Basing the prognosis on the bridging of additional cortices risks overestimation of the nonunion rate and is associated with relatively poor reliability.