Michael Kleiner-Shochat a , b , * , Daniel Kapustin c , Marat Fudim d , Andrew P. Ambrosy e , f , Juliya Glantz a , b , Mark Kazatsker a , b , Ilia Kleiner g , Jean Marc Weinstein g , Gurusher Panjrath h , Ariel Roguin a , b , Simcha R. Meisel a , b
28 October 2020
Background: Prediction of readmission and death after hospitalization for heart failure (HF) is an unmet need. Aim: We evaluated the ability of clinical parameters, NT-proBNP level and noninvasive lung impedance (LI), to predict time to readmission (TTR) and time to death (TTD). Methods and Results: The present study is a post hoc analysis of the IMPEDANCE-HF extended trial comprising 290 patients with LVEF ≤45% and New York Heart Association functional class II–IV, randomized 1:1 to LI-guided or conventional therapy. Of all patients, 206 were admitted 766 times for HF during a follow-up of 57 ± 39 months. The normal LI (NLI), representing the “dry” lung status, was calculated for each patient at study entry. The current degree of pulmonary congestion (PC) compared with its dry status was represented by ΔLIR = ([measured LI/NLI] – 1) × 100%. Twenty-six parameters recorded during HF admission were used to predict TTR and TTD. To determine the parameter which mainly impacted TTR and TTD, variables were standardized, and effect size (ES) was calculated. Multivariate analysis by the Andersen-Gill model demonstrated that ΔLIR<sub>admission</sub> (ES = 0.72), ΔLIR<sub>discharge</sub> (ES = –3.14), group assignment (ES = 0.2), maximal troponin during HF admission (ES = 0.19), LVEF related to admission (ES = –0.22) and arterial hypertension (ES = 0.12) are independent predictors of TTR ( p < 0.01, χ<sup>2</sup> = 1,206). Analysis of ES showed that residual PC assessed by ∆LIR<sub>discharge</sub> was the most prominent predictor of TTR. One percent improvement in predischarge PC, assessed by ∆LIR<sub>discharge</sub>, was associated with a likelihood of TTR increase by 14% (hazard ratio [HR] 1.14, 95% confidence interval [CI] 1.13–1.15, p < 0.01) and TTD increase by 8% (HR 1.08, 95% CI 1.07–1.09, p < 0.01). Conclusion: The degree of predischarge PC assessed by ∆LIR is the most dominant predictor of TTR and TTD.