Abstract P1010: Prediction Algorithms of Sarcopenia in US Older Adults

Sarcopenia is a significant risk factor for falls, disability, and mortality in US older adults. Currently, simple screening tools identifying sarcopenia in US older adults remain less clear. PURPOSE: The study developed simple prediction algorithms for sarcopenia using several anthropometric variables from the National and Nutrition Health Examination Survey (NHANES). Methods: The study participants included 3,377 men and women aged 60 to 80. All participants completed baseline anthropometric measurements, clinical evaluations, gait speed, 5 times sit-to-stand test (SST), and body composition assessment. Sarcopenia was defined as the combination of low muscle mass and SST or slow gait speed following the guidelines of the European Working Group on Sarcopenia in Older People (EWGSOP) and the Foundation for the National Institutes of Health (FNIH). Low muscle mass was classified using the appendicular skeletal muscle mass (ASM) divided by height in meters squared as <7.0 kg/m 2 in men and <5.5 kg/m 2 in women or ASM divided by body mass index (BMI) as <0.789 in men and <0.512 in women. Slow gait speed was classified as <0.8 m/s in both men and women. Low muscle strength was classified as greater than 15 seconds of SST. Multivariate stepwise logistic regression models were used to develop prediction algorithms for sarcopenia. Harrel’s c-statistics were used to measure the discrimination, and the Hosmer-Lemeshow Chi-square statistics were used for calibration measures. The bootstrap procedures were used to cross-validate the developed prediction model’s accuracy. Results: The prediction algorithms for sarcopenia included age, sex, BMI, and waist girth using EWGSOP criteria (cases = 74), indicating the c-statistics of 0.94 with bootstrap cross-validation c-statistics of 0.9416. The P-value of the Hosmer-Lemeshow statistics was 0.92. Using FNIH criteria to define sarcopenia (cases = 224), age, sex, height, BMI, and waist girth maximized the discrimination index (c-statistics = 0.95) with the bootstrap cross-validation of 0.949. There was a valid calibration measure (P-value for the Hosmer-Lemeshow statistics = 0.18). Conclusion: There was good discrimination and calibration measures to predict sarcopenia using EWGSOP criteria (age, sex, BMI, and waist girth) and FNIH criteria (age, sex, height, BMI, and waist girth). Public health agencies should consider recommending these simple prediction algorithms for screening sarcopenia across community-based US older adults.