Femoral cortical thickness influences the pattern of proximal femoral periprosthetic fractures with a cemented stem.

Abstract

Periprosthetic fractures of the proximal femur place a significant burden on the patients who endure them, as well as the medical health system that supports them. The purpose of this study was to determine whether femoral cortical thickness, as an absolute measurement, is a predictor of periprosthetic fracture pattern. A cohort of 102 patients who had sustained a periprosthetic hip fracture were retrospectively identified. This included 58 males and 44 females with a mean age of 79.8years. The femoral periprosthetic fracture pattern was classified based on the Vancouver classification system. Stem fixation was recorded and femoral cortical thickness measured. Patients were grouped into cemented and cementless stems. The relationship between cortical thickness and periprosthetic fracture pattern was assessed using the primary stem fixation method. Receiver operating characteristic (ROC) curve analysis was used to identify a threshold in the cortical thickness that predicted fracture pattern. Multinomial logistic regression analysis was used to adjust for confounding variables to assess the independent influence of cortical thickness on the risk of sustaining a Vancouver type A, B or C. There were 65 (63.7%) patients in the cemented group and 37 (36.3%) in the cementless group. The pattern of periprosthetic fractures around cemented stems was significantly (p<0.001) influenced by the femoral cortical thickness, with a thinner cortical thickness associated with a type A fracture pattern. In contrast, no association between femoral cortical thickness and fracture pattern assessment was demonstrated in the cementless group (p=0.82 Chi square). Comparing the rate of type A fracture patterns between the groups illustrated a significantly decreased risk in the cemented group with a cortical thickness of >7mm (odds ratio 0.03, p<0.001). ROC curve analysis of the cemented group demonstrated a threshold value of 6.3mm, offering a sensitivity of 83.3% and a specificity of 78.9% in predicting an A type fracture. Using this threshold, patients with a cortical thickness of 6.3mm or less were significantly more likely to sustain a Vancouver type A fracture (OR 18.9, 95% CI 2.0-166.7, p<0.001) when compared to patients with a cortical thickness of >6.3mm. In contrast, the ROC curve analysis did not find cortical thickness to be a predictor of fracture pattern in the cementless group. When adjusting for confounding variables, multinomial logistic regression demonstrated a cortical thickness of 6.3mm or less was a significant predictor of a type A fracture (OR 3.28, 95% CI 1.06-10.16, p=0.04) relative to those sustaining a type B fracture. Cortical thickness was found to influence the periprosthetic fracture pattern around cemented femoral stems, but this was not observed with cementless stems. Type A fracture patterns were significantly more likely to occur with a cortical thickness of 6.3mm or less around cemented stems.