Prognostic value of the sliding length of cephalocervical screws to predict the risk of non-union after osteosynthesis: a retrospective analysis of 86 patients with intracapsular femoral neck fractures

Between December 2008 and December 2013, 96 patients with femoral neck fracture were treated surgically with Dual SC Screws (DSCS) [10] at our hospital (Munakata Suikokai General Hospital, Japan). The inclusion criteria were as follows: (1) weight bearing initiated within 2 weeks after the surgery; (2) availability of postoperative radiographs at three time-points (immediate postoperative period; 2 weeks after surgery; final follow-up); and (3) minimum duration of follow-up of 6 months.

The exclusion criteria were as follows: (1) patients with pathological fractures and open fractures and (2) those with postoperative infections or severe complications that did not allow initiation of weight bearing within 2 weeks after the surgery.

In total, 53 fractures were classified as undisplaced (Garden stages I and II) and 33 fractures as displaced (Garden stages III and IV). All fractures were treated using a traction table and fluoroscopy without any open reduction. At first, we longitudinally pulled the lower limbs and gradually performed abduction, and a hip joint was internally rotated gently for good reduction of the displaced fractures.

All operations were performed using a radiolucent traction table under fluoroscopic guidance using standard procedures outlined below [7]. A 4-cm-long lateral incision was made directly through the skin to the bone; a guide wire was inserted from the lateral cortex (at the level for lesseor trochanter) to the femoral head [inferior 1/2 area in anteroposterior (AP) view] at an approximate angle of 135°. As the first guide wire was inserted, fluoroscopy was used to confirm satisfactory alignment, close to the calcar on the AP view and centrally on the lateral view. A second guide wire was inserted to the superior 1/2 area of the femoral head in a parallel direction to the first guide wire, temporarily before reaming to avoid proximal fragment rotation. The second guide wire should rest on the dorsal cortex on lateral view, and the first and second wires spread as far as possible on the AP view. After reaming, the lag screws were introduced. The screws were parallel in both planes, and the tip of each screw was driven to within 5 mm of the subchondral bone plate. All patients were permitted weight bearing with crutches/walker assistance within at least 2 weeks after surgery to prevent long-term bed rest.

All patients were retrospectively evaluated using medical records. Data pertaining to the following variables were extracted: age, sex, waiting time for surgery (days), operation time (min), blood loss (g), duration of follow-up (months), and pre-/postoperative walking ability.

The AP view of the hip was obtained with the patients in the supine position and the lower limbs in a medial rotation of 15°. The radiographic magnification was determined by dividing the known diameter of the lag screw with the diameter measured on the radiographs [7]. The Garden alignment index was calculated from the immediate postoperative radiograph [11]. The length of the proximal/distal screws was measured at each time point. The early sliding length (ESL) after weight bearing was measured by comparing the AP radiographs obtained at the first and second time-points; the final sliding length (FSL) was measured by comparing the AP radiographs obtained at the first and third time-points. Non-union was defined at 6 months postoperatively as re-displacement/an absence of radiographically visible trabeculae across the fracture line [9]. Radiographic evaluation was performed by a senior orthopedician who was blinded to the study.

All statistical processing was performed with JMP11 (SAS, Cary, NC). The Mann–Whitney U test and χ ² test were used to compare continuous and nominal variables between union and non-union groups, respectively. The effect of ESL/FSL of proximal/distal screw on non-union was analyzed by logistic regression. Receiver operating characteristic (ROC) curves were generated for the ESL/FSL as the predictor of non-union. The cut-off point was determined by the area under the ROC curve. P < 0.05 indicated statistical significance for all analyses.