variables. The statistical significance of differences between groups was tested using 2-sample t-tests for continuous variables and the Chi-square test for categorical v a r i a b l e s . The degree of correlation between continuous variables was assessed using Pearson’s correlation coefficient. Predictors of nonadherence were identified using logistic regression. Odds ratio (OR) was for the occurrence of nonadherence as it correlated with a one SD decrease in scale score. Variables entered into the multivariate logistic models included age, gender, diuretic use, education, duration of diet therapy, BMI, Kt/V, and diabetes mellitus. The results are expressed as adjusted odds ratios. The patient’s residual urine volume was taken into account by adjusting for diuretic use on the basis of a previous study showing that diuretic users had almost twice the odds of retaining residual urine