#6633 DIURETIC RESISTANCE PREDICTORS IN ACUTE DECOMPENSATED HEART FAILURE ADMISSIONS: A BROADER LOOK

Abstract

Abstract Background and Aims Failing to relieve congestion portends a rise in mortality rate and length of stay in acute decompensated heart failure (ADHF) hospitalizations. That situation corresponds to diuretic resistance (RDIUR). We sought to evaluate clinical and biochemical predictors of RDIUR, including those non-conventional ones we noticed as relevant in prognosis in our clinical practice. Method We conducted an observational, analytical and cross-sectional study, with prospective and consecutive patient (P) data acquisition. We included P over 18 years-old, who were admitted to the Intensive Care Telemetry Unit in our Hospital in the period between July, 2011 and May, 2022 due to ADHF diagnosis. All P received diuretic IV therapy as part of pharmacological treatment. As mentioned in earlier presentations, urine output (UO) was registered in all P, and failure to achieve an UO ≥ 1.5 ml/kg/hour under a pre-established protocol that included and initial 100 mg of IV furosemide bolus followed by a 2 hours of a 5 mg/hour furosemide continuous infusion. Under failure to achieve the goal, furosemide dose was doubled for two additional hours. Failure to respond to this strategy was called “diuretic resistance”. P who received a heart transplant during index admission or those who were on chronic dialysis were excluded from this study. Clinical, epidemiological and biochemical variables were registered. Missing data was imputed by means of multiple regression with chained equations. We constructed a multivariate logistic regression, with selection of relevant predictors completed by resampling and regularization of a Lasso regression model. Statistical significance was considered with a 5% of alpha error. Analysis was conducted using R (R Core Team, 2021), with RStudio (RStudio Team, 2020) as Integrated Development Environment for R, and the tidymodels package (Silge and Kuhn, 2022). Results A total of 1359 P were included. RDIUR was observed in 177 P (13%). These P were younger (69 years old vs 72 p = 0.02), with a higher prevalence of diabetes (39.5% vs 28%; p = 0.002), admission creatinine levels (1.4 vs 1.1 mg/dl; p < 0.001), lower total T3 serum levels (0.56 ng/ml vs 0.7; p < 0.001), with higher incidence of cholestasis (61 vs 40%; p < 0.001), and anasarca as clinical phenotype of systemic congestion (17.5% vs 8.5% p < 0.001). In-hospital mortality was higher in P who developed RDIUR (32.8 vs 5.6%, p < 0.001), in the same way that inotropic support (62 vs 16, p < 0.001), worsening heart failure (48% vs 11; p < 0.001), and furosemide dose at 2nd admission day (175.4 vs 123.4 p < 0.001) were higher in these P. We found no differences in left ventricular ejection fraction (38 vs 41 p = 0.21), first day natriuresis (75.5 vs 78.4, p = 0.093), coronary etiology vs others (20% vs 22; p = 0.26) or one month readmission rate (33% vs 37; p = 0.27). In the adjusted model, inotropic support (OR 5.2; IC95% 3.4-8; p < 0.001), worsening heart failure (OR 3.4; IC95% 2.2-5.3; P < 0.001), furosemide dose at 2nd day (OR 1, IC95%1, P 0.001), admission creatininemia (OR 2.05; IC95% 1.7-2.6; p < 0.001), anasarca phenotype (OR 2.3; IC95 1.4-3.9; p = 0.002) and normal T3 levels (OR 0.05; IC95% 0.02-0.14; p < 0.001) were identified as independent predictors of RDIUR development. Model accuracy was 0.89 and ROC-AUC 0.87. In the same way, RDIUR (OR 2.57; IC95% 1.5-4.3; p < 0.001), worsening heart failure (OR 4.1 IC95% 2.4-7; p < 0.001), inotropic support (OR 3; IC95% 1.7-5.2; p < 0.001) and cholestasis (OR 3.1; IC95% 1.9-5.2; p < 0.001) were found as independent predictors of in-hospital mortality. Conclusion Diuretic resistance is a serious event in patients hospitalized for ADHF, and considerably associated with mortality. A decreased renal reserve, low T3 levels, and anasarca at admission were significantly associated with diuretic resistance development. Considering “new” biochemical predictors as admission T3 levels might add strength to this subset of patients, opposed to information displayed by “classical” event predictors in ADHF patients, as ventricular function or etiology of cardiac disease. Due to clinical potential implications these observational results should be confirmed under evidence based protocols.