Validation of a multi-frequency bioelectrical impedance analysis device for the assessment of body composition in older adults with type 2 diabetes

The study addresses the need for accurate, accessible assessment of muscle mass in older adults with type 2 diabetes, a population with elevated sarcopenia prevalence and associated morbidity. MRI and CT are accurate but impractical; DXA is a research reference but has limitations (availability, body size constraints, hydration/body thickness effects). DSMF-BIA is portable and affordable but may be influenced by fluid status—relevant in T2DM due to comorbidities and treatments affecting body water. Prior validations of DSMF-BIA exist in general and obese middle-aged populations, but not specifically in older adults with T2DM. The research question: How well does DSMF-BIA (InBody-770) agree with DXA for assessing body composition (especially lean mass and ASMI) in older adults with T2DM, and is this agreement affected by sex or by short-term interventions (diet, resistance training, SGLT-2 inhibitor)?

Background literature indicates rising proportions of older adults and increased T2DM prevalence. Sarcopenia is common and consequential in older adults, with higher prevalence among those with T2DM (meta-analysis: ~31% vs 16.2% in controls), and associated with adverse outcomes (frailty, disability, sleep disorders, albuminuria, diabetic foot disease, cardiovascular disease). DXA is widely used to estimate regional and whole-body composition but is constrained by availability and technical limitations (e.g., very tall/obese individuals, hydration/body thickness). BIA assesses TBW, FFM, and FM via tissue electrical properties, is practical, but assumptions about constant FFM composition can be violated with aging and disease. In T2DM, body fluid alterations due to complications and medications (e.g., thiazolidinediones, insulin, SGLT-2 inhibitors) may impact BIA validity. Prior studies have validated InBody analyzers in middle-aged adults and obese women, but not older adults with T2DM. Reviews have called for validation of BIA for sarcopenia assessment in T2DM populations.

Design and setting: Post hoc analysis within the CEV-65 randomized trial conducted at the Institute of Endocrinology, Metabolism, and Hypertension, Tel-Aviv Sourasky Medical Center (May 2018–February 2021); IRB approved (NCT03560375). Participants: Community-dwelling older adults with T2DM (age ≥65 years), HbA1c 6.5–8.0%, low physical activity, recruited per ADA criteria. Key exclusions: current SGLT-2 use, recent steroids, uncorrected hypothyroidism, eGFR <45 ml/min, advanced neuropathy, active resistance training or nutritional therapy, recent diet change or weight-loss program. Sample: Of 100 randomized participants (60 women), 84 (49 women) had both DSMF-BIA and DXA at baseline and were analyzed. Interventions: After baseline, participants randomized to 10-week intervention: (1) circuit resistance training (two supervised sessions plus three home sessions/week); (2) ad libitum plant-forward Mediterranean ('vegeterranean') diet; (3) empagliflozin 10 mg daily. At 10 weeks, paired DXA and DSMF-BIA were available for 7 (CRT), 10 (diet), and 11 (empagliflozin) participants. Measurements: Conducted mornings before 10 am after ≥8 h fast; participants avoided morning exercise. DSMF-BIA: InBody 770, standing position; 30 measurements using six frequencies (1, 5, 50, 250, 500, 1000 kHz) across five segments (right/left arms, trunk, right/left legs). DXA: Lunar Prodigy (GE Healthcare), enCORE v13.31.016; standard automated scan modes; 5–10 min scans; single blinded technician performed analyses with manual ROI adjustments for arms, legs, trunk. Outcomes: Regional lean mass (kg), fat mass (kg), total body fat (%). ASMI calculated as sum of arm and leg lean mass divided by height squared (kg/m²). Additional measures: Height, weight, waist circumference, fasting glucose, HbA1c; sarcopenia defined by EWGSOP cut-offs (ASMI <7.0 kg/m² men, <5.5 kg/m² women). Statistics: Descriptive means ± SD. Agreement between DSMF-BIA and DXA assessed via Bland-Altman plots with regression for proportional bias (p < 0.05 significant) and intraclass correlation coefficients (ICC; two-way mixed, absolute agreement, average measures). ICC interpretation: >0.90 excellent, 0.75–0.90 good, 0.50–0.75 moderate, <0.50 poor. Paired t-tests compared means. Sex-stratified analyses and assessment post-intervention performed. Significance set at p < 0.05 using SPSS v20.

Participants: Mean age 71.4 ± 5.3 years; BMI 30.0 ± 5.6 kg/m²; HbA1c 7.66% ± 1.32; diabetes duration 15.5 ± 10.0 years. Sarcopenia prevalence by DXA thresholds: 8% (7/84). Agreement and mean differences: • Legs lean mass: DSMF-BIA 14.76 ± 3.62 kg vs DXA 15.19 ± 3.52 kg; paired difference ~−0.43 kg; Bland-Altman showed no proportional bias (p = 0.353). • Arms lean mass: DSMF-BIA 5.61 ± 1.60 kg vs DXA 5.29 ± 1.37 kg; mean difference +0.31 kg; significant bias (p = 0.001). • Trunk lean mass: DSMF-BIA 22.91 ± 4.86 kg vs DXA 24.21 ± 5.01 kg; mean difference −1.29 kg (limits of agreement approx. −4.80 to 2.21 kg); significant. • ASMI: DSMF-BIA 7.43 ± 1.28 vs DXA 7.47 ± 1.29 kg/m²; no difference (p = 0.84); ICC = 0.965 (p < 0.0001); mean difference −0.068 (p = 0.595). • Total body fat %: DSMF-BIA 38.44 ± 6.71% vs DXA 36.98 ± 6.46%; mean difference +1.46% (Bland-Altman bias not significant p = 0.356), with DSMF-BIA tending to overestimate. • Arms fat mass: DSMF-BIA 5.55 ± 2.72 kg vs DXA 3.21 ± 1.58 kg; mean difference +2.32 kg; ICC lower for arms fat (total sample ICC ≈ 0.615; men 0.483; women 0.749). • Legs fat mass: DSMF-BIA 8.54 ± 2.55 kg vs DXA 8.92 ± 3.47 kg; mean difference −0.38 kg (p = 0.043). • Trunk fat mass: DSMF-BIA 16.30 ± 4.88 kg vs DXA 17.48 ± 5.57 kg; mean difference −1.17 kg (p < 0.0001). ICCs: Overall excellent correlations (ICC > 0.90, p < 0.05) for most lean and fat measures, except arm fat mass (moderate). Sex-specific analyses showed good-to-excellent agreement for regional lean mass in both sexes, with arm lean mass in men showing moderate agreement (ICC = 0.706). Post-intervention (10 weeks): Across diet, CRT, and empagliflozin groups, patterns persisted (arm LBM overestimated; legs LBM and ASMI slightly underestimated; fat % overestimated) with high ICCs (generally ≥0.95) and acceptable Bland-Altman limits; interventions did not materially affect agreement. Diagnostic performance for sarcopenia: Compared with DXA, DSMF-BIA showed high specificity (93%) and high negative predictive value (97%).

DSMF-BIA (InBody-770) demonstrated high agreement with DXA for assessing lean mass, especially appendicular measures and ASMI, in older adults with T2DM. Agreement for fat-related measures was generally good, though DSMF-BIA tended to overestimate total fat percentage modestly and arm fat mass substantially, with lower reliability for arm fat. Short-term interventions relevant to T2DM management (plant-forward Mediterranean diet, circuit resistance training, empagliflozin) did not alter the level of agreement, supporting robustness across changes in body composition and potentially in body fluids. In contrast to prior studies in healthy or middle-aged cohorts where DSMF-BIA often underestimated fat and overestimated FFM, this T2DM older cohort showed slight overestimation of BF and % fat, underscoring population-specific validation needs. Given DSMF-BIA’s practicality vs DXA’s limitations (access, size constraints, hydration/body thickness effects), these findings support DSMF-BIA as a clinical screening tool for sarcopenia in older adults with T2DM, with the caveat of cautious interpretation of arm fat metrics.

In older adults with T2DM, DSMF-BIA shows high agreement with DXA for muscle mass assessment, particularly appendicular lean mass and ASMI, and is a reliable, accessible screening method for sarcopenia. Agreement is maintained after diet, resistance training, or empagliflozin interventions. Routine use of DSMF-BIA in clinics caring for older patients with diabetes should be considered to facilitate early recognition and management of low muscle mass. Future work should validate DSMF-BIA across broader T2DM populations (including those with advanced complications and higher sarcopenia prevalence), standardize regional definitions, and establish reference populations and cross-manufacturer calibration.

Limitations include the relatively small number of participants with paired post-intervention measures, short follow-up (10 weeks), and the post hoc nature in a study not originally designed specifically to validate DSMF-BIA versus DXA. The cohort comprised relatively young older adults with T2DM and limited complications and a low prevalence of sarcopenia, which may restrict external validity. DXA itself has limitations related to body size and hydration, which may affect comparisons.