Prevalence and predictors of asymptomatic vertebral fracture in patients with end-stage renal disease

OBJECTIVE: This study aimed to investigate the prevalence and predictors of asymptomatic vertebral fracture in patients with end-stage renal disease undergoing hemodialysis. METHODS: This cross-sectional study included 80 patients with end-stage renal disease undergoing hemodialysis. Medical history, Fracture Risk Assessment Tool and anteroposterior and lateral radiographs of the thoracolumbar and lumbosacral spine were obtained. Vertebral fractures were identified using the Genant semiquantitative assessment. RESULTS: Radiography demonstrated asymptomatic vertebral fracture in 22 patients (27.5%). FRAX® results for major osteoporotic fracture (area under the curve, 0.64) and hip fracture (area under the curve, 0.62) were able to discriminate patients with prevalent asymptomatic vertebral fracture. A multivariate analysis demonstrated that a 1-year average corrected calcium (odds ratio, 0.38), steroid use (odds ratio, 8.99), and a serum albumin concentration <25 g/dL (odds ratio, 28.82) significantly predicted prevalent asymptomatic vertebral fracture (clinical model; area under the curve, 0.82). Combining the 1-year average corrected calcium and serum albumin concentration <25 g/dL with FRAX® results for major osteoporotic fracture (area under the curve, 0.78) and FRAX® results for hip (area under the curve, 0.75) produced a significantly greater area under the curve value to predict fracture when compared with FRAX® result for major osteoporotic fracture and FRAX® result for hip (P = 0.022). CONCLUSION: Asymptomatic vertebral fracture is prevalent. FRAX® results for major osteoporotic fracture and hip provided lower ability in predicting asymptomatic vertebral facture when compared to the clinical model. Combining a 1-year average corrected calcium and serum albumin concentration <25 g/dL with FRAX® result for major osteoporotic fracture or hip improved the model's performance and provided comparable area under the curve to the clinical model.