Study Design
The study design was a non-randomized, prospective cohort study. Subjects were followed until the time when one of the following occurred: transfer to a non-participating dialysis unit, kidney transplantation, death or end of study follow-up. Vascular access events were recorded in a prospective manner. Subjects were censored at the time of transfer, transplant or end of follow-up. The study protocol was approved by the institutional review board of the University of Texas Health Science Center San Antonio and written informed consent was obtained from all study participants.
Dialysis treatments in both groups utilized polysulfone, non-reuse, high flux dialyzers (Optiflux 2000, Fresenius); blood flows were 400 ml per minute and dialysate flows were 800 ml per minute. Subjects in the daily hemodialysis group received six dialysis treatments per week, 3 hours each treatment; those in the conventional dialysis group received 3 treatments per week of 4 hours each treatment. Adjustments were made to dialysate bath, dry weight and medications including erythropoietin, phosphate binders, vitamin D sterols, and intravenous iron by the treating nephrologists in order to achieve K/DOQI guidelines for control of blood pressure, anemia and secondary hyperparathyroidism [17]. Aside from the increased frequency of dialysis visits in the daily hemodialysis group, there was no difference in the frequency of clinic visits or monitoring of laboratory values in either group.
Statistical Methods
Baseline characteristics were summarized by mean and standard deviation for continuous variables. For categorical variables, frequencies and percentages were shown. Wilcoxon’s Rank Sum test was used for continuous variables and chi-square or Fisher exact test was used to compare categorical variables between patients with conventional dialysis or with daily dialysis. The unadjusted relationship between the treatment group and time to first access procedure was examined using Kaplan-Meier survival analysis. Cox proportional hazard regression was used to assess the independent relationship and to adjust for potential confounding by age, gender, serum phosphorus, diabetes status, hemoglobin level and erythropoietin dose. To compare the rate of access procedure during follow-up time between the two groups, Poisson regression was used with Huber-White sandwich variance estimator accounting for over-dispersion. By recognizing the observed imbalance between the groups due to a small sample size, propensity score analysis was conducted to control for a potential bias due to imbalance between two treatment groups. Propensity score adjustment preserved statistical power by reducing confounders into a single variable. Propensity scores were estimated as the logit of a binary logistic regression predicting probability of being assigned into daily hemodialysis group as a function of other risk factors including age, gender, serum phosphorus, diabetes status, hemoglobin level and erythropoietin dose. Then the propensity score was added as a covariate in the multivariable models to further evaluate the adjusted effect of treatment group on outcomes (propensity adjusted model). A sensitivity analysis was performed in which subjects with catheters were excluded from analysis and there was no significant change in the results. All the analysis and calculations were done using R version 2.10.0 (www.r-project.org). All statistical inferences were assessed at a 2-sided 5% significant level.