Results of Total Hip Arthroplasty Differ in Subtypes of High Dislocation

Abstract

In a previous study, we described two subtypes of high dislocation of the hip depending on the presence (C1) or absence (C2) of a false acetabulum, yet we have already presented the concise followup of total hip arthroplasty (THA) in these patients as a group at a minimum of 15 years. In this retrospective study, we investigated differences in the results of THA in the C1 and C2 subtypes of high dislocation such as (1) survivorship of the reconstructions; (2) Merle d'Aubigné-Postel clinical scores; (3) leg lengthening and femoral shortening; and (4) site of reattachment and union rate of the greater trochanter. We included 49 hips of the C1 subtype and 30 hips of the C2 subtype operated on from 1976 to 1994. We evaluated survivorship (using reoperation for any reason as the end point) and performed chart and radiographic reviews. The 15-year survival was 84% (± 10% [95% CI]) for the C1 subtype and 60% (± 17% [95% CI]) for the C2 subtype (p = 0.001). Cox regression analysis, after adjustment for confounding factors, showed also statistically significantly worse survivorship in the C2 subtype (p = 0.021) and, after adjustment for possible predictive factors, found a statistically significant relationship of high dislocation subtype (p = 0.018) and trochanteric union (p = 0.005) with survival of THAs. Pain, function, and mobility scores improved from preoperative to last followup in C1 and C2 groups but they did not differ between C1 and C2 hips. C2 hips were lengthened more (p < 0.001) despite greater amounts of femoral shortening (p = 0.006). Site of reattachment and the risk of greater trochanter nonunion were not different between the groups. We found important differences in fundamental parameters after THA in the high-dislocation subtypes, including the risk of revision, which was higher in patients whose hips did not have a false acetabulum. These findings indicate that while reporting THA results in patients with high dislocation, mixing results of the two subtypes may lead to statistical bias.