Abstract
The emergence of the COVID-19 pandemic caused a significant shortage of medical personnel and the prioritization of life-saving procedures on internal medicine and cardiology wards. Thus, the cost- and time-effectiveness of each procedure proved vital. Implementing elements of imaging diagnostics into the physical examination of COVID-19 patients could prove beneficial to the treatment process, providing important clinical data at the moment of admission. Sixty-three patients with positive COVID-19 test results were enrolled into our study and underwent physical examination expanded with a handheld ultrasound device (HUD)-performed bedside assessment included: right ventricle measurement, visual and automated LVEF assessment, four-point compression ultrasound test (CUS) of lower extremities and lung ultrasound. Routine testing consisting of computed-tomography chest scanning, CT-pulmonary angiogram and full echocardiography performed on a high-end stationary device was completed in the following 24 h. Lung abnormalities characteristic for COVID-19 were detected in CT in 53 (84%) patients. The sensitivity and specificity of bedside HUD examination for detecting lung pathologies was 0.92 and 0.90, respectively. Increased number of B-lines had a sensitivity of 0.81, specificity 0.83 for the ground glass symptom in CT examination (AUC 0.82; p < 0.0001); pleural thickening sensitivity 0.95, specificity 0.88 (AUC 0.91, p < 0.0001); lung consolidations sensitivity 0.71, specificity 0.86 (AUC 0.79, p < 0.0001). In 20 patients (32%), pulmonary embolism was confirmed. RV was dilated in HUD examination in 27 patients (43%), CUS was positive in two patients. During HUD examination, software-derived LV function analysis failed to measure LVEF in 29 (46%) cases. HUD proved its potential as the first-line modality for the collection of heart-lung-vein imaging information among patients with severe COVID-19. HUD-derived diagnosis was especially effective for the initial assessment of lung involvement. Expectedly, in this group of patients with high prevalence of severe pneumonia, HUD-diagnosed RV enlargement had moderate predictive value and the option to simultaneously detect lower limb venous thrombosis was clinically attractive. Although most of the LV images were suitable for the visual assessment of LVEF, an AI-enhanced software algorithm failed in almost 50% of the study population.