True incidence of dislocation and associated risk factors in patients with a femoral neck fracture operated with an uncemented hemiarthroplasty

Background: Several factors might be associated with risk of dislocating following uncemented hemiarthroplasty (HA) due to femoral neck fracture (FNF). Current evidence is limited with great variance in reported incidence of dislocation (1-15%). Aim of this study was to identify the true cumulative incidence of first-time dislocation following HA and to identify the associated risk factors. Methods: We performed a retrospective cohort study of patients receiving an HA (BFX Biomet stem, posterior approach) at Copenhagen University Hospital, Bispebjerg, in 2010-2016. Patients were followed until death or end of study (dec 2018). Dislocation was identified by code extraction from the Danish National Patient Registry. Variables included in the multivariate model was defined pre-analysis to include: age, sex and variables with a p-value <0.1 in univariate analysis. A regression model was fitted for 90 days dislocation as the assumption of proportional hazard rate (HR) was not met here after. Results: We identified 772 stems and 58 stems suffered 90 dislocations during the observation period resulting in a 7% (CI 5-9) incidence of dislocation 90 days after index surgery. 55 of the 58 stems (95%) experienced the first dislocation within 90 days after surgery. Only absence of dementia was identified as an independent protective factor in the cause-specific model (HR 0.46 (CI 0.23-0.89)) resulting in a 2.4-fold cumulative risk of experiencing a dislocation in case of dementia. Several other variables such as age, sex, various medical conditions, surgery delay and surgical experience were eliminated as statistical risk factors. We found a decrease in survival probability for patients who experienced a dislocation during follow-up. Conclusions: The incidence of first-time dislocation of HA (BFX Biomet stem, posterior approach) in patients with a hip fracture is found to be 7% 90 days after surgery. Due to the non-existing attribution bias, we claim it to be the true incidence. Dementia was among several variables identified as the only risk factor for dislocation. In perspective, we may consider treating patients with cognitive impairment by other methods than HA.