Comparative evaluation of the sarcopenia index and 5 related creatinine-cystatin C biomarkers in relation to all-cause mortality in older adults: A cohort study.

Several creatinine- and cystatin C-based biomarkers have been individually associated with mortality, but their comparative associations have rarely been evaluated head-to-head under a common analytic framework. We compared 6 related biomarkers in relation to all-cause mortality in older adults. In the China Health and Retirement Longitudinal Study 2015 to 2020 cohort, we studied 2987 men (428 deaths) and 2995 women (228 deaths) aged ≥60 years. Six related biomarkers (sarcopenia index [SI], predicted skeletal muscle index, total body muscle mass, creatinine-to-cystatin C ratio [CCR], eGFRcys(2012)/eGFRcr(2021), and eGFRcys(2012) - eGFRcr(2021)) were evaluated using sex-stratified Cox models, a prespecified mutually adjusted bootstrap ranking framework, and supportive external evidence from the National Health and Nutrition Examination Survey. In fully adjusted single-biomarker models, SI, predicted skeletal muscle index, and total body muscle mass showed the most consistent inverse associations with all-cause mortality, whereas CCR, eGFRcys(2012)/eGFRcr(2021), and eGFRcys(2012) - eGFRcr(2021) were weaker, particularly in women. Within the joint 6-biomarker framework, SI ranked highest in both sexes, whereas CCR showed a reversed association after mutual adjustment. Discrimination was broadly similar across single biomarkers, and the joint 6-biomarker model provided only modest improvement beyond standard covariates. Supportive analyses, including the National Health and Nutrition Examination Survey, were directionally consistent overall but less precise. Among 6 closely related creatinine-cystatin C biomarkers, SI showed the most consistent comparative association pattern with all-cause mortality. However, incremental discrimination beyond standard covariates was small, supporting SI primarily as the clearest comparative marker within this biomarker family rather than as evidence of major predictive superiority.