Abstract: Objective To construct prediction models for muscle mass based on single or multiple indexes of muscle strength in the community-dwelling elderly. Methods A total of 140 elderly residents were recruited from the communities between September and December 2019. Simple muscle strengths, including hand-grip strength, biceps curl test, 30 s chair stand test and maximal isometric strength of the quadriceps and hamstring muscles based on the Micro FET3 handheld ergometer were tested. Muscle mass was measured with body composition analyzer (TANITA, Japanese). The correlation between muscle strength and muscle mass were analyzed, and the regression equation was established. Results All indexes of muscle strength (hand-grip strength, biceps curl test, 30 s chair stand test and maximal isometric strength of the quadriceps and hamstring muscles) were positively correlated with muscle mass (r=0. 207-0. 720, P＜0. 001). The multiple regression analysis showed that the muscle mass prediction models based on each index of muscle strength were statistically significant (r²=0. 702-0. 761,P＜0. 001), except the result of 30 s chair stand test. The prediction model of muscle mass based on multiple indexes including sex, body mass index (BMI), non-dominant hand grip strength, result of 30 s chair stand test, maximal isometric strength of the quadriceps was statistically significant (r²=0. 786, P<0. 001). Conclusions All indexes of the simple muscle strengths are positively correlated with muscle mass. Each single muscle strength index and combined multiple muscle strengths can effectively predict muscle mass and the prediction effect is slightly better in the latter. These models provide simple screening methods for assessing muscle mass in the community-dwelling elderly, which is beneficial for subsequent examination and treatment.

Key words: community-dwelling elderly, muscle strength, muscle mass, prediction model