Background Rheumatoid cachexia (i.e. loss of fat free mass (FFM) and gain in fat mass (FM)) is often observed in chronic rheumatoid arthritis (RA) patients, with deleterious effects on morbidity and functional capacity [1,2]. In addition, the use of glucocorticoids is associated with increase in FM. Therefore, careful assessment of body composition is important in RA patients. Dual-energy X-ray absorptiometry (DXA) is a valid and reliable method to assess body composition, but is also expensive, time-consuming, invasive and thus not feasible for every-day use. Bioelectrical impedance analysis (BIA) is a relatively cheap, rapid, and non-invasive method to assess body composition in clinical practice, however BIA has not been validated in RA patients.
Objectives To validate BIA against two different, widely used DXA devices for the assessment of body composition in patients with RA.
Methods This cross-sectional validation study was performed in RA patients who visited study center A or B in the context of the COBRA-light extension studyl. Body composition was assessed by DXA (study center A: Hologic Delphi; study center B: GE Lunar IDxa) and BIA (Quadscan 4000) on the same day, and recorded as FM, FM%, FFM and FFM%. Agreement between both methods was assessed by paired t-tests, Bland-Altman plots and intra-class correlation coefficients (ICC), in which DXA was used as reference method.
Results Of the 149 extension study patients, 43 patients (63% women, mean age 56 years, mean disease duration 5 years) were assessed by both DXA and BIA at the same day, and were included in this validation study. In study center A (n=25), BIA overestimated FM by 0.2 kg (p=0.71), and significantly underestimated FFM by 2.2 kg (p<0.001). FM% and FFM% were respectively over- and underestimated by 1.1% (both p=0.05). In study center B (n=18), BIA systematically underestimated FM by 3.1 kg, and overestimated FFM by 2.2 kg; FM% and FFM% were respectively under- and overestimated by 3.6% (all: p<0.001). Bland-Altman plots showed wide limits of agreement for all assessments in both study centers (Figure 1). These limits of agreement, indicating measurement error on individual level, were approximately 8 kg for FM and FFM assessments in both study centers. The ICC's, indicating measurement error on group level, showed good agreement in study center A (>0.90) and moderate-to-good agreement in study center B (>0.85).
Conclusions This study showed that BIA is an invalid method for the assessment of body composition in RA patients, both at individual and group level, and that errors depend on the DXA device used as reference method.
Walsmith, Int J Cardiol 2002, 85:89–99
Summers, Nat Rev Rheumatol 2010, 6:445–51
Disclosure of Interest None declared