HOSPITALIZATION BURDEN BY VASCULAR ACCESS IN MAINTENANCE HEMODIALYSIS: A MULTICENTER ANALYSIS ACROSS DAVITA CLINICS IN BRAZIL

Vascular access is a critical determinant of morbidity in hemodialysis (HD). Native arteriovenous fistula/graft (AVF/AVG) is the preferred access, whereas central venous catheters (CVC) – both short-term (non-tunneled) and long-term (tunneled) – carry much higher risks of infection, hospitalization and death. For example, CDC data show that HD patients have ~100-fold higher rates of bloodstream infection than the general population (Figure) – largely due to catheter use. In Brazil only ~10% of patients initiate HD with an AVF, and national census data report that in prevalent HD patients ~65% use AVF/AVG, while ~31% use CVC (23% long-term, 8% short-term). This analysis quantifies all-cause hospitalizations by access type in a large multicenter HD population.

We performed a retrospective multicenter analysis (January 2024 to August 2025 – 20 months) using de-identified, aggregated operational data from 103 DaVita Brazil clinics (in-center conventional HD). The mean monthly population was 17,802 patients, with 573 hospital admissions per month on average. Vascular access was categorized as short-term CVC (temporary catheter), long-term CVC (tunneled catheter), or AVF/AVG (arteriovenous fistula or graft). The average monthly counts by access type were: 777 patients (71 admissions) with short-term CVC, 4,536 patients (224 admissions) with long-term CVC, and 12,461 patients (276 admissions) with AVF/AVG. Person-time was calculated as patients × months, and incidence rates (admissions per patient-year) were derived by multiplying the monthly ratios by 12. Ninety-five percent confidence intervals (95%CI) for rates were estimated by Poisson methods. Incidence rate ratios (IRRs) comparing each catheter group to AVF/AVG were computed with 95%CI and p-values (using approximate Poisson regression). Absolute risk differences (RDs) per patient-year were also calculated with 95%CI. Finally, the population attributable fraction (PAF) of hospitalizations “attributable” to catheter use was estimated by comparing observed events to a counterfactual scenario where all patients had the AVF/AVG rate.

Table 1 summarizes the hospitalization rates, IRRs, and risk differences by access type. The hospitalization rate per 100 patient-years was markedly higher for catheter users than for AVF/AVG users: 109.6 (95%CI 103.9–115.4) for short-term CVC, 59.2 (57.5–61.0) for long-term CVC, and 26.6 (25.9–27.3) for AVF/AVG. Using AVF/AVG as the reference (rate=26.6), the IRR was 4.12 (95%CI 3.88–4.36, p<0.001) for short-term CVC and 2.23 (2.14–2.32, p<0.001) for long-term CVC. The absolute risk differences were 0.830 admissions per patient-year (95%CI 0.773–0.888, p<0.001) higher for short-term CVC and 0.326 (0.308–0.345, p<0.001) higher for long-term CVC, compared to AVF/AVG. In other words, every year of dialysis with a short-term catheter was associated with ~83 more admissions per 100 patients than with AVF, and tunneled catheters with ~33 more per 100 patients. These differences were highly statistically significant. In the study population overall, an estimated 31% of hospitalizations were attributable to use of catheters instead of AVF/AVG (PAF ≈0.31), assuming the AVF rate as the counterfactual. No reference group corrections were needed for covariates, as this was an aggregated descriptive analysis. These findings are illustrated in Figure 1, which plots the hospitalization rate per 100 patient-years by access type. The short-term catheter group has markedly higher admissions than long-term catheters, which in turn exceed AVF/AVG. This pattern is consistent with published literature: AVFs achieve lower infection and hospitalization rates than catheters.

Table 1. Hospitalization rates per 100 patient-years by vascular access, with incidence rate ratios (IRRs) and absolute risk differences (RD) comparing catheter access to AVF/AVG. IRRs are from Poisson regression; RDs are differences in rates (catheter – AVF/AVG). All comparisons vs AVF/AVG were highly significant (p<0.001).

In this large real-world analysis, vascular access type was a strong predictor of hospitalization burden in Brazilian hemodialysis patients. Patients dialyzing via short-term CVCs had the highest admission rate (>4-fold that of AVF/AVG), and tunneled catheters also doubled the rate seen with fistulas/grafts. These results quantitatively support the well-known clinical guidance that AVFs and AV grafts should be prioritized to reduce complicationspmc.ncbi.nlm.nih.gov. Approximately one-third of hospitalizations in this population could theoretically be avoided if all patients had the lower AVF/AVG rate. Our findings underscore the need to expand AVF/AVG creation (and maintenance) among chronic HD patients, in line with vascular access guidelines. Future efforts should focus on barriers to fistula maturation and timely referral for AVF creation, especially for patients currently managed with catheters.