Which Method Should Be Used to Assess Protein Intake in Peritoneal Dialysis Patients? Assessment of Agreement Between Protein Equivalent of Total Nitrogen Appearance and 24-Hour Dietary Recall

Abstract

After dialysis initiation, a high protein diet is recommended due to significant nutrient losses through dialysate and increased risk of protein energy wasting. In peritoneal dialysis (PD) patients, protein intake can be assessed through different methods that have some advantages and limitations, which affect its use on routine care. The aim of this study is to evaluate the agreement between 2 different methods (24-hour dietary recall and PNA-protein equivalent of total nitrogen appearance) on estimating protein intake in PD patients. Patients on PD for at least 3months, aged 18years old or more, were enrolled. To estimate protein intake, 24-hour dietary recall and PNA was used. PNA was calculated from 24-hour urine on the same day of the 24-hour dietary recall. Fifty individuals on PD were included, mean age 55.7±16.2years, and body mass index 26.0±4.5kg/m2. The average energy consumption was 1788.79±504.40kcal/day, which corresponds to 26.81±9.11kcal/kg current body weight (BW)/day and 29.82±8.39kcal/kg ideal body weight (IBW)/day. The median of total daily and normalized protein intake estimated using dietary recall was 61.43 (45.28-87.40) g/day, 0.90 (0.58-1.22) g/kg current BW/day, and 1.04 (0.77-1.32) g/kg IBW/day, respectively. Daily protein intake estimated by PNA was 55.75 (48.27-67.74) g/day, protein intake normalized by current BW was 0.81 (0.72-0.99) g/kg and 0.92 (0.83-1.06) g/kg IBW/day. Bland-Altman analysis indicates no systematic bias for the assessment of total protein intake and normalized protein intake for current and ideal BW. Significant proportionality bias was observed for both evaluations, showing there is a dispersion of the values. Despite the absence of systematic bias in the Bland-Altman analysis, there is no agreement in the assessment of protein intake by dietary recall and PNA, due to the existence of proportionality bias. Thus, values can be influenced biased by the magnitude of the measures.