P1432 Heart failure subgroups according to left ventricular ejection fraction A latent class analysis. (17th January 2020)

• Record Type: Journal Article
• Authors: Morbach, C; Henneges, C; Sahiti, F; Breunig, M; Cejka, V; Ertl, G; Frantz, S; Angermann, C E; Stoerk, S
• Is Part Of: European heart journal. Volume 21(2020)Supplement 1
• Publication Date: 2020-01-17
• Subjects: Cardiovascular system -- Imaging -- Periodicals; Heart -- Imaging -- Periodicals; 616.10754
• DOI: 10.1093/ehjci/jez319.861
• Languages: English
• ISSNs: 2047-2404

Abstract

Abstract: Funding Acknowledgements: German Research Foundation (BMBF 01EO1004 and 01EO1504) OnBehalf: AHF Background & Aims Heart failure (HF) is classified according to left ventricular (LV) ejection fraction (EF) into heart failure with reduced (HFrEF) and heart failure with preserved EF (HFpEF). In 2016, a third subgroup, heart failure with mid-range EF (HFmrEF), has been introduced by the ESC. We aimed to identify the number of naturally occurring heart failure subgroups according to LVEF using latent class analysis. Methods The AHF registry is a monocentric prospective follow-up study that comprehensively phenotypes consecutive patients hospitalized for acute heart failure (AHF). Echocardiography was performed within 72 hours prior to discharge. We first estimated the distribution of LVEF using histogram and kernel density estimation methods (bandwidth was selected by biased cross-validation). We then fitted Gaussian Mixture Models with increasing number of components to the data. To select the optimal number of components we calculated the Akaike Information Criterion (AIC) and the Bayesian Information Criterion (BIC). The minimum of each criterion suggests the optimal number of components for the final model. The BIC requires more data to select more components than the AIC and hence is more conservative. Finally, for each criterion the optimal model was determined. Results Out of 629 patients, 585 (93%) patients received echocardiography and in 498 (79.2%) the LVEF could be calculated using Simpson´s biplane or monoplane method. The BIC suggested two (panel B), the AIC three components (panel A). In the two-component model, mean ± SD LVEF values were 60.2 ± 8.7% and 30.8 ± 9.6%, thus covering 56% and 44% of patients, respectively (panel D). In the three-component model, respective LVEF values were 64.9 ± 6.2%, 50.2 ± 6.9%, and 28.4 ± 8.1%, thus covering 35%, 27%, and 38% of patients (panel C). Conclusions Our analysis suggests that LVEF in patients with AHF is not a continuum, but clusters in two or three subgroups. In line with the HFrEF and HFpEF classification, the more conservative model suggested two subgroups of LVEF. The less restrictive model allowed for a third subgroup, compatible with HFmrEF. Future analyses will better characterize the identified subgroups.