Relative validity and reproducibility of food frequency questionnaire for individuals on hemodialysis (NUGE-HD study)

Introduction Adequate assessment of food intake is essential to establish the magnitude and direction of the relationship of food, nutrients, and bioactive compounds with clinical outcomes of individuals in hemodialysis. We evaluated the relative validity and reproducibility of a specific food frequency questionnaire for individuals on hemodialysis (FFQ-HD). Methods Eighty-two participants (57.3% male, 57.5 +/- 14.4 years) from the open cohort Nutrition and Genetics in Hemodialysis Outcomes participated in this study. The relative validity of the FFQ-HD was assessed using the mean of two 24-h food recall (24hR) adjusted for within-subject variability as a reference method. We also performed Pearson correlations, and agreement between tertile, kappa statistics, and Bland-Altman scatter plots were validated. Reproducibility was assessed after 1 year using intraclass correlation coefficient (ICC). Findings Daily energy intake was not different between FFQ-HD and 24hR (mean difference of 50.1 kcal). Intake of protein, linolenic acid, fiber, phosphorus, potassium, sodium, calcium, and sugar showed a moderate correlation (r between 0.4 and 0.5) among instruments, while mean correlation coefficient was r = 0.38 to food group intake. Bland-Altman plots showed good agreement for micronutrients, phosphorus, sodium, and potassium and for the groups "flour, bread, and pasta" and "processed, canned meat, salts, and seasonings". The reproducibility of FFQ-HD for nutrients and food groups was satisfactory, reaching a maximum ICC of 0.72 and 0.59, respectively. Discussion The FFQ-HD showed moderate validity and reproducibility for calories, nutrients, and food groups of clinical and nutritional interest for HD subjects so that it can be a useful tool in epidemiological studies in this population.