Comparison of fracture prediction tools in individuals without and with early chronic kidney disease: A population-based analysis of CARTaGENE.
J Bone Miner Res. 2020 Feb 05;: Authors: Desbiens LC, Sidibé A, Beaudoin C, Jean S, Mac-Way F
Whether fracture prediction tools developed for the management of osteoporosis can be used in CKD is poorly known. We aimed to compare the performance of fracture prediction tools in non-CKD and CKD. We analyzed CARTaGENE, a population-based survey of 40- to 69-year-old individuals recruited between 2009 and 2010. Renal function was assessed using baseline creatinine and categorized according to KDIGO guidelines (non-CKD, stage 2, stage 3). Individuals without creatinine measurements or with advanced CKD (stage 4 or 5; prevalence < 0.25%) were excluded. Predicted 5-year fracture probabilities (using FRAX, QFracture and Garvan) were computed at baseline. Fracture incidence (major fracture [MOF] or any fracture) was evaluated in administrative databases from recruitment to March 2016. Discrimination (hazard ratios per standard deviation increase in Cox models; c-statistics) and calibration (standardized incidence ratios before and after recalibration) were assessed in each CKD strata. We included 19,393 individuals (9,522 non-CKD; 9,114 stage 2; 757 stage 3). 830 patients had any fracture during follow-up, including 352 MOF. FRAX (HR=1.89 [1.63-2.20] non-CKD; 1.64 [1.41-1.91] stage 2; 1.76 [1.10-2.82] stage 3) and QFracture (HR=1.90 [1.62-2.22] non-CKD; 1.57 [1.35-1.82] stage 2; 1.86 [1.19-2.91] stage 3) discriminated MOF similarly in non-CKD and CKD. In contrast, the discrimination of Garvan for any fracture tended to be lower in CKD stage 3 compared to non-CKD and CKD stage 2 (HR=1.36 [1.22-1.52] non-CKD; 1.34 [1.20-1.50] stage 2; 1.11 [0.79-1.55] stage 3). Before recalibration, FRAX globally overestimated fracture risk while QFracture and Garvan globally underestimated fracture risk. After recalibration, FRAX and QFracture were adequately calibrated for MOF in all CKD strata while Garvan tended to underestimate any fracture risk in CKD stage 3 (SIR=1.31 [0.95-1.81]). In conclusion, the discrimination and calibration of FRAX and QFracture is similar in non-CKD and CKD. Garvan may have a lower discrimination in CKD stage 3 and underestimate fracture risk in these patients. This article is protected by copyright. All rights reserved. PMID: 32022942 [PubMed - as supplied by publisher]