Development and external validation of a practical diagnostic support tool, ‘ABC2-Screener’, to predict sarcopenia among patients on maintenance haemodialysis: A multicentre cross-sectional study

Abstract

AbstractBackground and hypothesisSarcopenia is common in patients undergoing maintenance haemodialysis (MHD); however, the current diagnostic support tools for sarcopenia are difficult to implement in dialysis clinics. This study aimed to develop a clinically friendly screening tool to predict sarcopenia using ubiquitous clinical data.MethodsThis cross-sectional multicentre study enrolled 373 and 129 patients undergoing MHD in the derivation and external validation cohorts, respectively. The Asian Working Group for Sarcopenia diagnostic criteria were used as a sarcopenia reference standard. Candidate predictors, such as age, sex, body mass index, routine blood tests, and the one-item Clinical Frailty Scale (CFS) version 2.0, were used to develop an original web-based model and a paper-based point score system using backward elimination selection. The two tools were completed using optimism-corrected regression coefficients for each variable, derived by bootstrapping. Their performance was evaluated by examining the discrimination and calibration in the two cohorts.ResultsIn total, 98 (26.3%) and 44 (34.1%) patients in the derivation and validation cohorts were diagnosed with sarcopenia, respectively. For internal validation, the area under the receiver operating characteristic curve (AUROC) for the original model and the point score system were 0.97 (95% CI: 0.96– 0.98) and 0.95 (95% CI: 0.93–0.97), respectively. Calibration plots for the original model showed excellent agreement between the predicted and observed probabilities. In contrast, the point-score-based model underestimated sarcopenia in the moderate-risk range. For external validation, the original model achieved an AUROC of 0.97 (95% CI: 0.95–1.00), while the point score system achieved an AUROC of 0.91 (95% CI: 0.87–0.96). The calibration plots for both models showed similar performances to those of the internal validation.ConclusionIn patients undergoing MHD, our practical diagnostic support tool ‘the ABC2-Screener’ has good discrimination and calibration abilities and can be easily used at any medical facility.