Covid-19: development and results of a CT puntaje admission chest CT predictive value

Abstract

Purpose: To create a CT Score to objectively and quantitatively predict the severity and evolution   of COVID-19 pneumonia in concordance with unenhanced upon admission chest CT findings.   Material and Methods: We retrospectively evaluated 98 patients with a diagnosis of SARS-CoV-2   confirmed by RT-PCR admitted to the general ward. We developed a CT score to quantify the   imaging involvement of the disease at hospital admission. This score values the type of patterns   and the total burden of the lesion (expressed as a percentage of the parenchymal involvement).   A Receiver Operating Characteristic (ROC) curve analysis was performed as a test of diagnostic   accuracy of the developed Score. Results: 98 patients were analyzed, using as cut-off point a   CT score puntaje ≤ 14. No patients with unfavourable evolution were detected (100 % Negative   Predictive Value, 80 % sensitivity, 100 % specifity to predict favourable development). CT Score   < 22 (91.2 % Negative Predictive Value for the need of oxygen reservoir masks and 94.7 % for   unfavourable outcome). A CT Score ≥ 22 predicted a need for oxygen therapy and unfavourable   development. (92.6 % Positive Predictive Value, 80 % sensitivity and 65 % specificity). The area   under curve (AUC) was 0.8197, which makes it a test with a high diagnostic discriminatory capacity.  Conclusion: CT Score is useful to determine the radiological assessment of pulmonary involvement   in three grades: minor, moderate and severe. The imaging findings are highly correlated with clinical   evolution variables. It can be considered an important tool for prognostic value and to adapt early   and timely therapeutics behaviours in the development of this illness.