Diuretic Response in Acute Heart Failure: Prediction vs Direct Measurement

BACKGROUND: Limited data are available informing the variability and ability to predict diuretic response (DR) longitudinally during decongestion in acute heart failure (AHF). METHODS: The FASTR pilot trial compared the Reprieve System (n=52) to optimal diuretic therapy control (ODT) (n=44). The ability to longitudinally predict DR, quantified as urine output over the 72-hour intervention, was evaluated in the overall population using baseline characteristics. In the Reprieve arm, directly measured DR during the initial dose finding algorithm was also used to predict longitudinal DR. RESULTS: Baseline characteristics predicted only 24.7% (p<0.001) of the variance in DR over the course of IV diuretic therapy. As the duration of diuretic therapy increased, DR prediction worsened with no relationship to DR after 18 hours (p≥0.097). Directly measured DR strongly predicted subsequent urine output early (e.g., predicting 84.5% in the variance of the subsequent hour DR, p<0.001). However, this also decreased over time with modest correlation with DR at 24 hours (variance 12.7%; p=0.008) and no correlation after 48 hours (p=0.79). In the Reprieve group, serial DR measurement and diuretic titration occurred in 89% of patients eligible for further diuretic titration resulting in stable therapeutic DR over the study period despite the changes in DR over time. CONCLUSIONS: Accurately predicting longitudinal DR using baseline characteristics or a single direct measurement of DR is challenging as DR changes rapidly during decongestion. Due to the dynamic nature of DR over therapy, serial assessment and diuretic titration is required to longitudinally achieve therapeutic DR in most AHF patients.