Outcomes of the First and Second Hemodialysis Fistula: A Cohort Study.

RATIONALE & OBJECTIVE Fistulas are the preferred form of hemodialysis access; however, many fistulas fail to mature into usable accesses after creation. Data for outcomes after placement of a second fistula are limited. STUDY DESIGN Cohort study. SETTING & PARTICIPANTS People who initiated hemodialysis therapy in any of 5 Canadian dialysis programs (2004-2012) and had at least 1 hemodialysis fistula placed. PREDICTOR Second versus initial fistula; receipt of 2 versus 1 fistula; second versus first fistula in recipients of 2 fistulas. OUTCOMES Catheter-free fistula use during 1 year following initiation of hemodialysis therapy or following fistula creation, if created after hemodialysis therapy start; proportion of time with catheter-free use; time to catheter free use; time of functional patency. ANALYTICAL APPROACH Logistic regression; fractional regression. RESULTS Among the 1,091 study participants (mean age, 64±15 [SD] years; 63% men; 59% with diabetes), 901 received 1 and 190 received 2 fistulas. 38% of second fistulas versus 46% of first fistulas were used catheter free at least once. Average percentages of time that second and initial fistulas were used catheter free were 34% and 42%, respectively (OR, 0.72; 95% CI, 0.54-0.94). Compared with people who received 1 fistula, those who received 2 fistulas were less likely to achieve catheter-free use (26% vs 56%) and remain catheter free (23% vs 49% of time; OR, 0.30, 95% CI, 0.24-0.39). Among people who received 2 fistulas, the proportion of time that the second fistula was used catheter free was 11% higher with each 10% greater proportion of time that the first fistula was used catheter free (95% CI, 1%-22%). Model discrimination was modest (C index, 0.69). LIMITATIONS Unknown criteria for patient selection for 1 or 2 fistulas; unknown reasons for prolonged catheter use. CONCLUSIONS Outcomes of a second fistula may be inferior to outcomes of the initial fistula. First and second fistula outcomes are weakly correlated and difficult to predict based on clinical characteristics.