Muscle failure and risk of adverse outcomes in older adults: a derivation and multicohort validation study

Background: Sarcopenia is inconsistently defined, and its definition might not adequately capture muscle failure, underlying disability, and other adverse outcomes in older adults. We aimed to develop and validate an evidence-based, outcome-driven model of muscle failure. Methods: In this derivation and multicohort validation study, we analysed data from the SPRINTT randomised controlled trial (1519 participants; mean age 78·9 years [SD 5·8]; 431 [28·4%] men, 1088 (71·6%) women) and validated findings in five independent cohorts (ilSIRENTE, NHANES, HRS, ELSA, and CHARLS). Neuromuscular domains including appendicular lean mass, muscle strength and power, mobility, and physical activity were assessed. Principal component analysis identified clusters of measures, which were tested for associations with disability in mobility and disability in activities of daily living, hospitalisation, and mortality using adjusted regression models. Predictive performance was evaluated with receiver operating characteristic curves. Area under the curve (AUC) values were defined as acceptable (0·7–0·8), excellent (0·8–0·9), or outstanding (>0·9). Findings: We identified two major muscle failure indexes: mobility (short-distance walking speed, 400 m walk) and physical activity (step counts, time spent standing, stepping, and sitting). In SPRINTT, the mobility model strongly predicted disability (AUC 0·721, 95% CI 0·698–0·795; p<0·0001), whereas the physical activity model was more closely associated with hospitalisation (0·646, 0·570–0·731; p=0·00040) and mortality (0·746, 0·612–0·926; p=0·0020). Combined models including mobility, handgrip strength, appendicular lean mass, and absolute muscle power improved prediction of mortality (0·746, 0·549–0·955; p=0·0076). External validation confirmed acceptable-to-strong discrimination for disability outcomes across cohorts (AUCs 0·7–0·9), while associations with hospitalisation and mortality were weaker and inconsistent (0·4–0·9). Compared with consensus-based sarcopenia definitions, the evidence-based model seemed to show better predictive ability for disability. Interpretation: Distinct mobility measures provide a robust framework for identifying muscle failure and predicting disability in older adults. Continuous activity monitoring might improve prediction of hospitalisation and mortality. This outcome-driven approach supports refinement of sarcopenia assessment and its implementation in clinical practice.