Fracture Risk Prediction in Patients Receiving Dialysis.

Abstract

Patients receiving dialysis face a 5-fold higher risk of skeletal fractures compared to the general population, along with increased mortality and a longer hospital stay after a fracture. However, fracture risk varies among patients, and reliable methods to assess risk are limited. Using linked healthcare databases, we conducted a population-based cohort study of adults aged 40-90 years receiving maintenance dialysis with an available parathyroid hormone (PTH) value in Ontario, Canada (2010-2017). Patients were followed for 3 years or until death, first major fracture, or provincial emigration. Using Fine and Gray subdistribution hazards models accounting for the competing risk of death, we developed a fracture prediction tool (Dial-Frac) that included demographic, comorbidity, and laboratory information. Models were compared using measures of discrimination, calibration, and model fit and were internally validated using 10-fold cross-validation. The cohort included 11,599 patients receiving dialysis. Mean age was 66 years; 39% were female, and 12% had a previous fracture. Over 3 years, 839 patients (7.2%) experienced a fracture (44% of which occurred in the first year); hip fracture was most common (299/839). The final prediction model included age, sex, previous fracture, previous kidney transplant, proton pump inhibitor use, and PTH and serum albumin concentrations. The time dependent area under the receiver operating curve for fracture at 1 and 3 years were 0.79 and 0.71, respectively, indicating good discrimination. We developed Dial-Frac, an easy-to-implement equation that predicts fracture risk in patients receiving maintenance dialysis. It predicts risk at 1 and 3 years without requiring bone density measurements or additional tests, helping to identify high-risk patients for further screening, preventive interventions and clinical trial participation. Next steps include external validation.