Prognostic role of renal replacement therapy among hospitalized patients with heart failure in the Brazilian national public health system

Figures

• 

  Figure 1. Study design. Created with BioRender.com.

• 

  Figure 2. Map displays of (A) in-hospital mortality, (B) renal replacement therapy, (C) median cost per day of hospitalization, and (D) median duration of hospitalization for each federal state of Brazil.

• 

  Figure 3. Kaplan-Meier survival curve for patients stratified by the need for renal replacement therapy (RRT). The median survival time in patients receiving renal replacement therapy was 30 days (95% CI, 29–31), compared to 50 days (95% CI, 49–51) in patients not undergoing dialysis.

• 

  Figure 4. Hazard ratios with their 95% confidence intervals for the risk of in-hospital death for any cause from a multi-variable Cox proportional hazards model obtained by stepwise selection of predictors. Renal replacement therapy was associated with an approximately 56% increase of the risk of all-cause in-hospital death. The model included a variable termed “ongoing COVID-19 pandemic”, which accounted for HF admissions occurring after the official announcement of the COVID-19 pandemic by the World Health Organization (WHO) and the Brazilian Government.

• 

  Figure 5. (A) Histograms for the duration of hospitalization in patients not receiving (top) and receiving (bottom) renal replacement therapy. (B) Estimated effects on the duration of hospitalization estimated with a multi-variable generalized linear model using a gamma-distribution and log-link function. The graph shows exponentiated coefficient estimates, which represent multipliers for the duration of hospitalization of 6 days for a white, male patient with age of 67 years equal to the average age for the cohort. Renal replacement therapy was associated with a more than 3-fold increase in the duration of hospitalization.

• 

  Figure 6. The hospitalization cost per day for patients admitted with heart failure was analyzed with a generalized linear model (GLM) with a logarithmic link in the gamma distribution function and with age, sex, ethnicity, heart-transplantation, and renal replacement therapy type included as independent predictors. The coefficients in the GLM represent cost multipliers and are shown here as bars. The estimate for the average cost of hospitalization of white female patients with a mean age of 67 years (intercept term in the model) was 603 Br$. Heart transplantation during hospitalization was by far the largest factor for cost-per-day increases, while less (cost factor <1) was spent on patients of non-white ethnicity. For the effect of geographic region, the central-western region of Brazil, which includes Mato Grosso, Mato Grosso do Sul e Goiás, was the reference region.