Operative Versus Nonoperative Treatment of Geriatric Ankle Fractures: A Medicare Part A Claims Database Analysis

Background: The incidence of ankle fractures is increasing in the geriatric population, and several studies suggest them to be the third most common extremity fracture in this age group. Previous work has reflected relatively low complication rates during operative treatment. Little is known, however, about the association between these injuries and overall mortality, nor whether operative intervention has any effect on mortality. We hypothesized that geriatric ankle fractures would be correlated with an elevated mortality rate and that operative intervention would be associated with a reduced mortality when compared to nonoperative management. Methods: Following Institutional Review Board approval we retrospectively assessed all relevant 2008 part A inpatient claims from the Medicare database. We queried diagnosis codes for ankle fractures, and then excluded any patients whose age was less then 65 or had an admission related to an ankle fracture during the previous year. Operative patients were then identified by their ICD-9 procedure codes occurring within 30 days of their initial diagnosis code; all other patients were presumed to be treated without operative intervention, thereby creating 2 groups for comparison. We then analyzed this database for specific variables including overall mortality, length of stay, age distribution, and other demographical characteristics. Groups were compared with Elixhauser and Deyo–Charlson scores to determine the level of comorbidities in each group. Multivariate logistic regression analysis was used to determine if operative intervention had a protective effect. Results: In all, 19 648 patients with an ankle fracture were identified. Of those, 15 193 underwent operative intervention (77.3% ) and 4455 were treated nonoperatively (22.7% ). The mean ages for nonoperative and operative intervention were 80.9 and 76.5, respectively (P < .0001). The average length of stay for nonoperative management was 4.5 days, while operative intervention resulted in a length of stay of 4.6 days (P = .43). One-year mortality was 21.5% for the nonoperative group and 9.1% for the operative group (P < .0001). The mean Elixhauser score for the nonoperative group was 2.5 and 2.2 for the operative group (P < .0001). The mean Deyo–Charlson score was 1.3 and 1.0 for the nonoperative and operative groups, respectively (P < .0001). Multivariate logistic regression analysis demonstrated an odds ratio of 0.534 of death within 1 year for patients unde