Cardiac involvement in hospitalized patients with COVID-19 and its incremental value in outcomes prediction

Area of Science:

• Cardiology
• Infectious Diseases
• Medical Imaging

Background:

• Acute COVID-19 infection has been linked to cardiac abnormalities in hospitalized patients.
• Limited data exists on cardiac structure and function abnormalities in COVID-19 patients before diagnostic imaging.
• Previous cardiovascular disease status is crucial for understanding COVID-19's cardiac impact.

Purpose of the Study:

• To investigate cardiac abnormalities in acute COVID-19 patients using cardiovascular imaging.
• To assess the contribution of imaging findings to clinical outcomes, including mortality and ECMO use.
• To identify predictors of in-hospital mortality and ECMO support using machine learning.

Main Methods:

• Retrospective analysis of 724 hospitalized COVID-19 patients, excluding those with prior left ventricular systolic dysfunction.
• Review of health records and cardiovascular imaging studies (echocardiography).
• Application of machine learning models to predict in-hospital mortality and ECMO requirement.

Main Results:

• In patients without prior cardiovascular disease, significant prevalence of left ventricular systolic dysfunction (3.4%), abnormal longitudinal strain (24%), diastolic dysfunction (20%), and right ventricular systolic dysfunction (18%) was observed.
• Moderate to large pericardial effusion was rare (0.4% each).
• Echocardiographic measurements significantly improved the predictive accuracy (AUC 0.91) for mortality and ECMO support when added to clinical and laboratory data.

Conclusions:

• Acute COVID-19 infection is associated with a notable incidence of cardiac dysfunction, even in individuals without a history of heart disease.
• Echocardiographic findings are critical independent predictors of adverse clinical outcomes in COVID-19 patients.
• Integrating echocardiographic data into predictive models enhances their performance for identifying high-risk patients.