Predicting Slow Walking Speed From a Pooled Cohort Analysis: Sarcopenia Definitions, Agreement, and Prevalence in Australia and New Zealand

Abstract

Background Recent operational definitions of sarcopenia have not been replicated and compared in Australian and New Zealand (ANZ) populations. We aimed to identify sarcopenia measures that discriminate ANZ adults with slow walking speed (<0.8 m/s) and determine agreement between the Sarcopenia Definitions and Outcomes Consortium (SDOC) and revised European Working Group for Sarcopenia in Older People (EWGSOP2) operational definitions of sarcopenia. Methods Eight studies comprising 8,100 ANZ community-dwelling adults (mean age±standard deviation 62.0+14.4 years) with walking speed, grip strength (GR) and lean mass data were combined. Replicating the SDOC methodology, fifteen candidate variables were included in sex-stratified classification and regression tree (CART) models and receiver operating characteristic (ROC) curves on a pooled cohort with complete data to identify variables and cut points discriminating slow walking speed (<0.8 m/s). Agreement and prevalence estimates were compared using Cohen’s Kappa (CK). Results ROC curves identified GR as the strongest variable for discriminating slow from normal walking speed in women (GR<20.50 kg, area under curve [AUC]=0.68) and men (GR<31.05 kg, AUC=0.64). Near perfect agreement was found between the derived ANZ cut points and SDOC cut points (CK 0.8-1.0). Sarcopenia prevalence ranged from 1.5% (EWGSOP2) to 37.2% (SDOC) in women and 1.0% (EWGSOP2) to 9.1% (SDOC) in men, with no agreement (CK<0.2) between EWGSOP2 and SDOC. Conclusions GR is the primary discriminating characteristic for slow walking speed in ANZ women and men, consistent with findings from the SDOC. SDOC and EWGSOP2 definitions showed no agreement suggesting these proposed definitions measure different characteristics and identify people with sarcopenia differently.