Agreement and systematic difference between bioelectrical impedance analysis and dual-energy X-ray absorptiometry for...

Accuracy and agreement of bioelectrical impedance analysis (BIA) are essential for diagnosing sarcopenia. Previous studies have examined the validity of appendicular lean mass (ALM) estimated by BIA devices against dual-energy X-ray absorptiometry (DXA); however, the results are inconsistent. Furthermore, to our knowledge, no study has directly compared the prevalence of low muscle mass (LMM) assessed by DXA and BIA in the same cohort. This study aimed to validate a segmental multi-frequency bioelectrical impedance analysis (MF-BIA) device (InBody 770) against a DXA device (Hologic) and compare the prevalence of LMM in community-dwelling older adults aged 65-84 years.

The study population included 1617 community-dwelling older adults (680 men and 937 women). After overnight fasting, participants underwent MF-BIA measurement using InBody 770, followed by DXA (Hologic Discovery) in the morning. ALM was obtained from device outputs; additionally, ALM was recalculated from impedance variables (Z5, Z50, Z250; Ht²/Z50, Z250/Z5, and 1/Z50) using Yamada's equation. LMM was defined using the Asian Working Group for Sarcopenia (AWGS) 2019 criteria. We also applied previously proposed cut-offs for original InBody values (<6.6 kg/m² for men, and <5.0 kg/m² for women).

The ALM obtained by InBody software exhibited a strong correlation with the ALM obtained by DXA (r = 0.95). On average, InBody ALM was ∼9.2 % lower than DXA (∼7.0 % in men; ∼10.7 % in women). Bland-Altman (InBody - DXA) showed a mean difference of -1.64 kg with 95 % limits of agreement (LoA) from -4.14 to 0.87 kg, and proportional difference (slope -0.03, p < 0.001). The LMM prevalence was 38.1 % by InBody (original) vs 9.1 % by DXA. After recalibration (InBody × 1.10) or Yamada's equation, LMM prevalence approximated DXA-based estimates (e.g., 9.7 % and similar by Yamada). Using InBody-specific cut-offs also yielded a similar prevalence (9.5 %) to DXA.

In community-dwelling older adults, InBody 770 and Hologic DXA provide highly correlated ALM estimates but show a systematic method/device difference (∼10 % on average), with non-negligible LoA and proportional difference, which materially impacts LMM screening prevalence. Recalculation from impedance variables (e.g., Yamada's equation), a simple ∼1.10 scaling, or InBody-specific cut-offs can improve agreement with DXA-referenced estimates. Findings are device-/method-specific and should be externally validated.