Arteriovenous Fistula Use in the United States and Dialysis Facility-Level Comorbidity Burden

Rationale & objective: Patients with multiple comorbid conditions are less likely to use an arteriovenous fistula (AVF) for hemodialysis vascular access. Some dialysis facilities have high rates of AVF placement despite having patients with many comorbid conditions. This study describes variation in facility-level use of AVFs across the facility-level burden of patient comorbid conditions.

Study design: Retrospective cohort study.

Setting & participants: Medicare patients receiving hemodialysis for 1 year or more in US dialysis facilities.

Predictors: Facility-level burden of patient comorbid conditions; patient characteristics.

Outcomes: Odds of AVFs versus other access types; facility-level use of AVFs.

Analytical approach: Facility-level comorbidity burden was calculated by summing individual comorbid conditions, determining the average per patient, then defining 11 groups based on facility percentile ranking. Generalized estimating equations with a logit link were used to estimate the odds of AVF placement at the patient level. For the facility-level analysis, a generalized estimating equation model with the identity link was fit to characterize the percentage of AVF use at each facility.

Results: Overall, AVF use was 65.8% in 315,919 prevalent hemodialysis patients among 5,813 facilities. After adjustment for patient characteristics, AVF use was 0.27, 0.30, 1.05, and 1.74 percentage points lower than the median among facilities in the 61st to 70th, 71st to 80th, 81st to 90th, and 91st to 99th percentiles of comorbidity, respectively, and 0.42, 0.63, 1.34, and 1.90 percentage points higher than the median among facilities in the 31st to 40th, 21st to 30th, 11th to 20th, and 1st to 10th percentiles of comorbidity, respectively. Facilities in the greater than 99th percentile of comorbidity burden had AVF use that was 3.47 percentage points lower than the median. Facilities in the less than 1st percentile of comorbidity burden had AVF use that was 2.64 percentage points greater than the median.

Limitations: Limited to Medicare dialysis-dependent patients treated for 1 year or more.

Conclusions: After adjustment for patient characteristics, we found small differences in facility rates of AVF use except in the extremes of high or low levels of comorbidity burden. Our study demonstrates that dialysis facilities with a relatively high patient comorbidity burden can achieve similar fistula rates as facilities with healthier patients. Although high comorbidity burden does not explain low facility AVF use, additional study is needed to understand differences in AVF use rates between facilities with similar comorbidity burdens.