Background
Despite several muscle mass measures being used in the current definitions of sarcopenia, their usefulness is uncertain because of limited data on their association with health outcomes. The aim of the study was to compare the performance of different muscle mass measures for predicting incident osteoporosis in postmenopausal women.
Methods
This study included data from 149 166 participants (aged 60.3 ± 5.5 years) as part of the UK Biobank cohort. Body composition was assessed using bioelectrical impedance. The muscle mass measures included were total body skeletal muscle mass (SMM) and appendicular SMM (aSMM) divided by height squared (ht
2
), derived residuals, SMM, SMM adjusted for body mass (SMM/bm × 100), and aSMM normalized for body mass index (aSMM/BMI). Diagnoses of the events were confirmed by primary care physicians and coded according to the World Health Organization's International Classification of Diseases 10th Revision (ICD‐10: M80‐M82).
Results
Over a median follow‐up of 6.75 (5th to 95th percentile interval, 1.53 to 8.37) years, 394 newly diagnosed cases of osteoporosis occurred, with 40 (10.2%) cases being associated with a pathological fracture. SMM/ht
2
, aSMM/ht
2
residual, and SMM were lower in postmenopausal women with osteoporosis compared with women without (all
P
< 0.0001), while SMM/bm × 100 (
P
= 0.003), but not aSMM/BMI (
P
= 0.59), was higher in the osteoporosis group. The unadjusted rates of osteoporosis increased with decreasing quintiles for SMM/ht
2
, aSMM/ht
2
, residuals, and SMM (all
P
trend <0.0001), while the incidence of osteoporosis increased with increasing SMM/bm × 100 (
P
trend =0.001), but not for aSMM/BMI (
P
= 0.45). After minimally adjusting for age and after full adjustment, SMM/ht
2
, aSMM/ht
2
, and SMM were the only measure that consistently predicted osteoporosis in the total group of postmenopausal women [hazard ratio (HR) 0.65–0.67, all
P
≤ 0.0001], in lean women (HR 0.62–0.68; all
P
≤ 0.001), and women with increased adiposity (HR 0.64–0.68; all
P
≤ 0.01). In fully adjusted models, the changes in the
R
2
statistic were 13.4%, 11.6%, and 15.3% for the SMM/ht
2
(aSMM/ht
2
), residual, and SMM, but only 4.9% and 1.3% for SMM/bm × 100 and aSMM/BMI.
Conclusions
Muscle mass measures adjusted for height only (SMM/ht
2
, aSMM/ht
2
) appear to be better muscle‐re...