1. Application of survival classification and regression tree analysis for identification of subgroups of risk in patients with heart failure and reduced left ventricular ejection fraction
- Author
-
Radu Sascau, Vlad Vasiliu, Adrian Covic, Mihaela Mihaila, Florentina Rusu, Cristian Stătescu, Raluca Popa, Dimitrie Siriopol, Andreea Bucur, Ianis Siriopol, Zahariuc Cătălina, Petru Cianga, Andreea Neamtu, and Mehmet Kanbay
- Subjects
Male ,Cart ,medicine.medical_specialty ,Supine position ,Population ,030204 cardiovascular system & hematology ,Ventricular Function, Left ,03 medical and health sciences ,0302 clinical medicine ,Predictive Value of Tests ,Internal medicine ,medicine ,Humans ,Radiology, Nuclear Medicine and imaging ,030212 general & internal medicine ,education ,Cardiac imaging ,Aged ,Heart Failure ,education.field_of_study ,Ejection fraction ,business.industry ,Proportional hazards model ,Stroke Volume ,Prognosis ,medicine.disease ,Blood pressure ,Echocardiography ,Heart failure ,Cardiology ,Female ,Cardiology and Cardiovascular Medicine ,business - Abstract
The aim of this study was to identify by classification and regression tree (CART) analysis groups of patients with different survival patterns in a population of patients with heart failure and reduced left ventricular ejection fraction (HFrEF) by using standard methods of heart function assessment, as well as well as utilizing non-traditional approaches for determining hydration and nutritional status in HF patients—lung ultrasonography (LUS) and bioimpedance spectroscopy (BIS) analysis. Eligible patients with a left ventricular ejection fraction (LVEF) below 45% were identified via the daily echocardiography assessments. LUS was performed with patients in the supine position, for a total of 28 sites per complete examination. The hydration state and the body composition were assessed using a portable whole-body BIS device. Our study included 151 patients (69.2% males) with a mean age of 67.1 years. During the follow-up 53 (35.1%) patients died. Using the CART algorithm, we identified five groups based on serum sodium, the severity of NYHA class, serum urea and systolic blood pressure. When comparing the two models, the model derived from the CART analysis showed better predictive power than the conventional Cox model (c-index 0.790, 95% CI 0.723–0.857 vs. 0.736, 95%CI 0.664–0.807, p
- Published
- 2021