Risk factors for heart failure in patients with COVID-19.

ABSTRACT

Aim. To establish risk factors for heart failure (HF) in patients with coronavirus disease 2019 (COVID-19). Material and methods. Medical records of 151 patients treated in an infectious disease hospital from November 3, 2020 to February 2, 2021 with a confirmed diagnosis of COVID-19 were retrospectively selected. The collection of clinical, history and laboratory data were carried out by analyzing electronic medical records. We analyzed information on age, sex, body mass index, smoking, and comorbidities. Following laboratory studies were analyzed complete blood count, biochemical blood tests, coagulation profile, acute phase proteins (C-reactive protein (CRP), ferritin, lactate dehydrogenase (LDH)), procalcitonin. The diagnosis of HF was confirmed by clinical performance, echocardiography, and elevated levels of the N-terminal pro-brain natriuretic peptide (NT-proBNP). The risk of HF was taken as the endpoint of the study. Results. The studied sample of patients was divided into two groups depending on HF the 1st group included 46 patients with HF, the 2nd group - 105 patients without HF. The median age was 66,2 (50-92) years (women, 91 (60,3%)). Laboratory indicators, such as the levels of CRP, LDH, procalcitonin, creatinine, bilirubin, differed significantly from each other, and the median values were higher in patients with HF. The neutrophil-to-lymphocyte ratio (NLR) showed significant intergroup differences in the group of patients with HF, the median was 4,97% vs 3,62% (p=0,011) in the group of patients without HF. There were following most significant predictors increasing the HF risk age >=66 years (odds ratio, 8,038, p<0,001), procalcitonin level, which increases the HF risk in patients by 3,8 times (p<0,001), NLR >=4,11% (p=0,010), thrombocytopenia <=220x109/l (p=0,010), history of chronic kidney disease (CKD) (p=0,018). Conclusion. The following predictors of HF were established age >=66 years, procalcitonin >=0,09 ng/ml, NLR >=4,11%, thrombocytopenia <=220x109/l, history of CKD, LDH >=685 U/l and creatinine >=102 micromol/l, international normalized ratio >=1,19, QTc interval >=407,5 ms, bilirubin <=10,7 micromol/l. It is worth noting that the best accuracy values are demonstrated by the Random Forest algorithm (88,5% on the validation set), but the mathematical model of the neural network turned out to be the most sensitive (90,0% on the validation set).Copyright © 2023, Silicea-Poligraf. All rights reserved.