Near-infrared reactance spectroscopy (NIRS) has become increasingly popular in personal and professional settings now that it has been adapted to provide comprehensive body composition assessments. However, whether NIRS agrees with criterion methods remains unknown. Thus, this study aimed to determine the agreement between NIRS and DXA-derived body composition estimates. Ninety-seven participants completed body composition assessments using DXA, and first-generation (NIRSG1), second-generation (NIRSG2), and muscle-specific NIRS (NIRSFIT) devices. On a separate day, a subset of participants (n = 63) performed maximal voluntary contractions (MVC) on a handgrip dynamometer, which were used in conjunction with total appendicular lean mass (ALM) estimates to provide ratios (MVC/total ALM or MVC/ALM of the arms only) depicting muscle quality index (MQI). Fat mass, fat-free mass, body fat %, and ALM, from NIRSG2, but not bone mineral content (BMC), and NIRSFIT demonstrated equivalence (using equivalence tests) with DXA with R2 from 0.83 to 0.97; though BMC revealed concordance coefficients of 0.83 and an R2 of 0.88. MQI using total ALM from NIRS was not equivalent to DXA, but demonstrated low root mean squared error (0.08 kg/kg) and 95% limits of agreement (±0.21 kg/kg). Indices of visceral adipose tissue (iVAT) from NIRSG1 and NIRSG2 were significantly different (p < 0.001), but were both significantly associated with DXA VAT (NIRSG1 R2: 0.53; NIRSG2 R2: 0.62; both p < 0.001). NIRS appears to demonstrate acceptable agreement with DXA and continual improvements could make NIRS a viable alternative for comprehensive body composition assessments.
Read full abstract