Aims
Sarcopenia definitions include different measures of muscle strength, physical performance and body size/composition. We aimed to identify which baseline measures best predict incident mortality and falls, and prevalent slow walking speed, in older women and men.
Methods
Data for 899 women (mean age±standard deviation, 68.7±4.3 years) and 497 men (69.4±3.9 years) from the Dubbo Osteoporosis Epidemiology Study 2, comprising sixty variables/composite variables for muscle strength (quadriceps strength), physical performance (walking speed, timed up and go (TUG) and sit to stand (STS) tests), body size (weight, height) and body composition (lean mass, body fat) were included. Sex-stratified Classification and Regression Tree (CART) analyses calculated baseline variable importance and accuracy for predicting incident mortality and falls, and prevalent slow walking speed (<0.8m/s).
Results
Over 14.5 years follow-up, 103/899 (11.5%) women and 96/497 (19.3%) men died, 345/899 (38.4%) women and 172/497 (34.6%) men had 1 or more fall(s), and 304/860 (35.3%) women and 172/461 (31.7%) had baseline slow walking speed (<0.8m/s). CART models identified age and walking speed adjusted for height as the most important predictors for mortality in women, and quadriceps strength with body size adjustments as the most important predictor for mortality in men. In both sexes, STS (with adjustments) was the most important predictor for incident falls, and TUG test was the most important predictor for prevalent slow walking speed (<0.8m/s). Body composition measures were not important predictors for any outcome.
Conclusions
Muscle strength and physical performance variables and cut points predict falls and mortality differently in women and men, suggesting targeted sex-specific application of selected measures may improve outcome prediction in older adults.