Eighty-eight patients (54%) or their relatives were contactable. A total of 14 patients (16%) had received a secondary arthroplasty, while 18 patients (21%) had died.
Of the included 40 patients, 31 (77.5%) were treated with a DHS and 9 patients (22.5%) with cannulated screws. There were 19 female (47.5%) and 21 male patients (52.5%). The mean age at the day of surgery was 52 ± 9 years and 59 ± 9 years at the day of follow-up. The median follow-up time in months was 65.5 (5.5 years; IQR: 49.5–91.0). The most common cause of injury was fall (n = 22, 53.7%), followed by sport injuries (13, 31.7%) and traffic accidents (5, 12.2%). The fracture type was 31-B1 in 15 cases (41.7%), 31-B2 in 15 patients (41.7%), and 31-B3 in 6 patients (16.7%). According to the Garden classification, 12 patients (30%) suffered from a Garden I fracture, 14 patients (35%) sustained a Garden II fracture, while in 10 patients (25%) a Garden III fracture and in 4 patients (10%) a Garden IV fracture was seen. Displaced fractures (Garden III and IV fractures) were more often treated with DHS (n = 13 vs. 1). The median interval between injury and surgery was 5.5 h (min.-max.: 2.0–312.0 h). Average surgery time was significantly longer in the DHS group with 64 ± 25 min compared to 39 ± 13 min in patients treated with cannulated screws (p < 0.01; overall: 58 ± 25 min). Surgical time correlated directly with fracture type according to the Garden classification (r = 0.474, p < 0.01), but not with AO/OTA fracture type (r = 0.258, p = 0.064). The median hospitalisation time was 11 days (min.-max.: 5–86 days), comparable in both treatment groups (10 vs. 13 days, p = 0.517) and among patients with the three different fracture types according to AO/OTA (12 vs. 14 vs. 9 days, p = 0.077). Interestingly, time of hospitalisation was significantly longer in females compared to male patients (12 vs. 10 days, p < 0.05), although male patients sustained more dislocated fractures (Table 1).
On appointed follow-up, we observed avascular femoral head necrosis in one female patient (2.5%) who had sustained a B1 fracture which was initially treated with a DHS.
Femoral shortening and leg length discrepancy
Short-term radiographic evaluation
Overall, in 37 patients (92.5%) femoral neck shortening and in 32 patients (80%) femoral shortening was observed 3 months after surgery. At 3 months postoperatively, the median femoral neck shortening was 7 mm (IQR: 3–11 mm, min–max.: 0–25 mm), while femoral shortening of the affected side was 5 mm (IQR: 2–8 mm, min–max.: 0–16 mm). Extent of shortening did not differ between gender, fracture type, trauma mechanism or treatment groups. Interestingly, fractures with higher Pauwels grade lead to a significantly larger extent of femoral shortening (3 vs. 4 vs. 7 mm, p < 0.05). Time of surgery and hospitalisation time did not correlate with neither extent of femoral neck shortening nor with femoral shortening.
Radiological follow-up at 5 years
At invited follow-up (median: 65.5 months), extent of shortening in general did not further increase. The median shortening of the femoral neck was 5 mm, while the observed median femoral shortening was 6 mm. In nine patients (22.5%) femoral neck shortening of more than 10 mm was seen, while femoral shortening of more than 10 mm was measured in ten patients (25%). All measurements did not differ between gender and treatment groups.
In obtained whole-leg standing X-rays, LLD was noticed in 34 patients (85%) with a median LLD of 7 mm. Twenty-two patients (65%) experienced LLD of less than 10 mm, and 12 patients (35%) of more than 10 mm (Fig. 3). No differences in extent of LLD between both genders and surgical techniques were observed. Again, fractures with Pauwels grade III led to significantly increased LLD (5 vs. 5 vs. 10 mm, p < 0.05). Type B2 fractures according to the AO/OTA classification caused significantly larger dimension of LLD compared to B1 and B3 fractures (B1: 5 mm, B2: 10 mm, B3: 5 mm, p < 0.05). No differences regarding Garden classification were found. Extent of LLD correlated with patient weight (r = 0.535, p < 0.05). We did not find any differences in surgery time, hospitalisation time, age, gender, trauma mechanism, and type of surgery. As expected, shortening of the femoral neck (r = 0.662, p < 0.001) as well as femoral shortening (r = 0.443, p > 0.01) at 3 months postoperatively correlated with LLD at invited follow-up. We did not observe significant increase of femoral shortening between 3 months and 5 years postoperative (Fig. 4).
Functional outcome at 5 years
Overall, the functional outcome was excellent with a median HHS of 96 (IQR: 90–100). 30 patients (75%) scored excellent results in the HHS, in 4 patients (10%) good results were found, in 2 patients (5%) fair results and in 4 patients (10%) poor results were documented. We did not see significant differences in functional outcome between gender and treatment.
We did not observe differences in outcome parameters associated with shortening of the femoral neck, femoral shortening or LLD. We found only a statistical trend towards lower HHS in patients with femoral shortening of 10 mm or more (93 vs. 99, p = 0.081). Other tested variables including, time of surgery, surgical technique, trauma mechanism, age, gender, fracture type, and hospitalisation time did not show any correlation with shortening of the femoral neck, femoral shortening, nor with occurrence or extent of LLD.