Muscle mass assessment by BIA is highly dependent on device choice, potentially leading to clinically relevant discrepancies in classification when rigid cut-offs are applied.
Key Findings
Results
Significant device differences were found for all body composition parameters between the seca mBCA 515 and InBody 970 devices.
BIA data from 224 adults (85 with cancer, 139 with obesity) were analyzed.
Both devices were segmental multi-frequency BIA devices.
All parameters showed statistically significant differences (all p ≤ 0.005).
Discrepancies were largest for skeletal muscle mass (both in kg and %).
Results
Agreement between the two BIA devices for skeletal muscle mass was poor, with large effect sizes.
Effect sizes for skeletal muscle mass discrepancies were r > 0.8.
Lin's Concordance Correlation Coefficient (CCC) was below 0.90 for skeletal muscle mass, indicating poor agreement.
The Wilcoxon signed-rank test and agreement analyses were used to assess device differences.
Results
Device choice had a significant impact on classification of low fat-free mass in both cancer and obesity patients.
McNemar's test was used to evaluate differences in classification (p < 0.001).
The seca device classified 50% of patients as having low fat-free mass compared to only 20% with the InBody device.
Cut-offs used were those cited in the GLIM criteria for malnutrition.
Results
Device choice significantly affected classification of sarcopenic obesity, with seca classifying substantially more patients as sarcopenic obese than InBody.
90% of patients were classified as having body composition consistent with sarcopenic obesity by seca versus 50% by InBody.
Classification was based on ESPEN and EASO criteria for sarcopenic obesity.
The difference was statistically significant (p < 0.001).
Results
Device discrepancies were more pronounced in cancer patients and in females.
The impact of disease type, sex, and age on device differences was examined through multivariable models.
Cancer patients showed larger discrepancies between devices compared to obesity patients.
Female sex was associated with greater inter-device variability.
Age was also examined as a potential modifier of device differences.
Conclusions
The authors concluded that individualized interpretation of BIA data and further validation of prediction equations in disease-specific subpopulations is warranted.
Applying rigid cut-offs to BIA-derived muscle mass estimates may lead to clinically relevant misclassification.
Different BIA devices use different proprietary prediction equations, which may account for systematic discrepancies.
The study population included vulnerable clinical populations: patients with cancer and patients with obesity.
The authors call for further validation of BIA prediction equations in disease-specific subpopulations.