Arteriovenous Access Type and Risk of Mortality, Hospitalization, and Sepsis Among Elderly Hemodialysis Patients: A Target Trial Emulation Approach.

ABSTRACT

RATIONALE & OBJECTIVE: Evidence is mixed regarding the optimal choice of the first permanent vascular access for elderly patients receiving hemodialysis (HD). Lacking data from randomized controlled trials, we used a target trial emulation approach to compare arteriovenous fistula (AVF) versus arteriovenous graft (AVG) creation among elderly patients receiving HD. STUDY DESIGN: Retrospective cohort study. SETTING & PARTICIPANTS: Elderly patients included in the US Renal Data System who initiated HD with a catheter and had an AVF or AVG created within 6 months of starting HD. EXPOSURE Creation of an AVF versus an AVG as the incident arteriovenous access. OUTCOMES: All-cause mortality, all-cause and cause-specific hospitalization, and sepsis. ANALYTICAL APPROACH: Target trial emulation approach, high-dimensional propensity score and inverse probability of treatment weighting, and instrumental variable analysis using the proclivity of the operating physician to create a fistula as the instrumental variable. RESULTS: A total of 19,867 patients were included, with 80.1% receiving an AVF and 19.9% an AVG. In unweighted analysis, AVF creation was associated with significantly lower risks of mortality and hospitalization, especially within 6 months after vascular access creation. In inverse probability of treatment weighting analysis, AVF creation was associated with lower incidences of mortality and hospitalization within 6 months after creation (hazard ratios of 0.82 [95% CI, 0.75-0.91] and 0.82 [95% CI, 0.78-0.87] for mortality and all-cause hospitalization, respectively), but not between 6 months and 3 years after access creation. No association between AVF creation and mortality, sepsis, or all-cause, cardiovascular disease-related, or infection-related hospitalization was found in instrumental variable analyses. However, AVF creation was associated with a lower risk of access-related hospitalization not due to infection. LIMITATIONS: Potential for unmeasured confounding, analyses limited to elderly patients, and absence of data on actual access use during follow-up. CONCLUSIONS: Using observational data to emulate a target randomized controlled trial, the type of initial arteriovenous access created was not associated with the risks of mortality, sepsis, or all-cause, cardiovascular disease-related, or infection-related hospitalization among elderly patients who initiated HD with a catheter.