Urinary Creatinine Based Equations for Estimation of Fat Free Mass in Patients with Intestinal Insufficiency or Intestinal Failure

Background & aims: Assessment of body composition is an important aspect of disease management in patients with intestinal insufficiency (INS) or intestinal failure (IF). However, in daily clinical settings most body composition methods are too expensive or impractical, leaving body composition to be assessed by less reliable methods such as skin fold thickness. The aim of this study was to investigate and validate the use of an equation for the estimation of fat-free mass (FFM) with bioelectrical impedance analysis (BIA) as reference method. Methods: A literature search for identification of urinary creatinine-based FFM-prediction equations was carried out a long side the creation of an equation by multiple linear regression. The correlation of each equation with FFM (measured by BIA in 277 patients with either INS or IF) was done by Pearson's correlation. Further investigation and validation of performance was done for the equations with the strongest correlation by Bland–Altman analysis, determination of root mean square error (RMSE), and intraclass correlation (ICC). The validation was carried out in a new group of 37 patients with either INS or IF. Results: A total of 11 prediction equations were correlated with FFM measured by BIA. The equation called FFMmultiple and FFM-5 had the strongest correlation (r = 0.969, p < 0.01 and r = 0.950, p < 0.01, respectively). FFMmultiple was superior to FFM-5 regarding Bland–Altman analysis, RMSE, and ICC in the study group (Mean bias ± Standard Deviation = 0.042 ± 2.352 versus 0.309 ± 3.196; 95% limits of agreement = [-4.568; 4.651] versus [-5.955; 6.578]; RMSE = 0.158 versus 0.236; ICC = 0.969 versus 0.948). Cross-validation resulted in a Bland–Altman analysis with a statistically significant difference between FFMmultiple and FFM by BIA. FFM-5 showed wide 95% limits of agreement ([-6.977; 6.421]). Conclusions: Two urinary creatinine-based equations (FFMmultiple and FFM-5) showed promising results as possible substitutes to BIA, however further investigation and cross validation revealed inauspicious results. Thus, the present study cannot recommend the use of a prediction equation instead of BIA for the assessment of FFM in patients with INS and IF.