PLOS One · Nov 20, 2026
Bioelectrical impedance analysis versus dual-energy X-ray absorptiometry for sarcopenia assessment in type 2 diabetes mellitus: A systematic review and meta-analysis
By Raj S, Shetty S, Suvarna R et al.
Background: Sarcopenia is common in type 2 diabetes mellitus (T2DM). Bioelectrical impedance analysis (BIA) is cheaper and more accessible than dual-energy X-ray absorptiometry (DXA). We assessed whether BIA can replace DXA for measuring muscle mass and classifying low muscle mass in adults with T2DM. Methods: PubMed/MEDLINE, Embase, Scopus, Web of Science and CINAHL were searched to March 2026 (updated 7 October 2026) for studies measuring muscle mass with both methods in adults with T2DM. Risk of bias was assessed with QUADAS-C. The primary outcome was Bland–Altman agreement (bias, DXA − BIA; population limits of agreement); classification, diagnostic accuracy and correlation (association only) were secondary. Random-effects models used restricted maximum likelihood with Hartung–Knapp intervals; certainty was rated with GRADE. Results: Seven studies (646 participants, six countries) were included. For whole-body fat-free mass (three studies), the pooled bias was +0.82 kg (95% CI −0.69 to +2.34; population limits of agreement −6.4 to +8.0 kg). For appendicular lean-mass indices (three studies), the bias was −0.13 kg/m ² (95% CI −1.32 to +1.06; I ² = 98%; limits −1.8 to +1.5 kg/m ² ). In three studies classifying low muscle mass by both methods, the pooled difference in proportions (BIA − DXA) was +3.6 percentage points (95% CI −8.6 to +15.7). With the same thresholds for both methods (two studies), BIA sensitivity was 71% and 91% and specificity 94% and 64%. Estimates were strongly correlated (r = 0.90), reflecting association, not agreement. Certainty of evidence was low or very low. Conclusions: In adults with T2DM, BIA and DXA showed no clear average difference, but individual estimates and classifications differed too much for interchangeable use. Where DXA is unavailable, standardised BIA may support initial assessment, but results that would change management, particularly near diagnostic thresholds, should be confirmed with DXA.
doi.org/10.1371/journal.pone.0353604Open accessPubMed 43063219Biomarkers