Estimating of Residual Kidney Function by Multi-Frequency Bioelectrical Impedance Analysis in Hemodialysis Patients Without Urine Collection

Abstract

Background/Aims: Residual kidney function (RKF) is a pivotal predictor of better clinical outcomes in maintenance hemodialysis (MHD) patients. So far there has been no attempt to use bioimpedance analysis (BIA) measurements to calculate residual glomerular filtration rate (GFR) in dialysis population. We hypothesized that performing of multi-frequency BIA at the beginning and end of hemodialysis session can enable us to predict the measured residual GFR in MHD patients. Thus our aim was to develop and validate a new RKF prediction equation using multi-frequency BIA in MHD patients. Methods: It was diagnostic test evaluation study in a prospective cohort. Participants (n=88; mean age, 66.3±13.2 years, 59.1% males) were recruited from a single hemodialysis center. A new equation (eGFRBIA) to predict RKF, utilizing BIA measurements performed pre- and post-dialysis, was generated and cross-validated by the leave-one-out procedure. GFR estimated as the mean of urea and creatinine clearance (mGFR) using urine collections during entire interdialytic period. Results: A prediction equation for mGFR that includes both pre- and post-dialysis BIA measurements provided a better estimate than either pre- or post-dialysis measurements alone. Mean bias between predicted and measured GFR was -0.12 ml/min. Passing and Bablok regression showed no bias and no significant deviation in linearity. Concordance correlation coefficient indicated good agreement between the eGFRBIA and mGFR (0.75, P<0.001). Using cut-off predicted mGFR levels >2 ml/min/1.73 m2 yielded an area under curve of 0.96, sensitivity 85%, and specificity 89% in predicting mGFR. The κ scores for intraobserver reproducibility were consistent with substantial agreement between first and second estimation of RKF according to eGFRBIA (weighted κ was 0.60 [0.37-0.83]). Conclusion: We present a valid and clinically obtainable method to predict RKF in MHD patients. This method, which uses BIA, may prove as accurate, convenient and easily reproducible while it is operator independent.