ABSTRACT
Introduction: The spread of COVID-19 imposed a re-organization of healthcare services. The increased circulation of the virus among the population and the need for a quick diagnosis suggest the identification of new decisionmaking actions. Aim(s): We aimed to determine whether clinical, laboratory and imaging characteristics can reliably differentiate suspected COVID from positives tested with nasal-pharyngeal(N-P)swab. Method(s): This is a retrospective, cross-sectional study on data collected(March-May 2020)at the Lung Unit, University of Pavia(confirmed cases=50)and at the Emergency Unit, University of Palermo(suspected cases=103). COVID-19 was confirmed by positive N-P swab. Suspected were tested negative. We recorded clinical, laboratory and radiological variables and compared them by Chi-squared or t-test. Logistic regression estimated which variables increased the risk of COVID-19 in both groups. A prediction model and derived ROC-AUCcurve, assessed the accuracy. P<0.05 was considered significant. Result(s): High ReactiveC-Protein, suggestive CTimages, dyspnea and respiratory failure increased the risk of COVID-19, whereas pleural effusion decreased the risk(table). ROC confirmed the discriminative role of these variables between suspected and confirmed COVID-19(AUC 0.91). Conclusion(s): Clinical, laboratory and imaging features predict the diagnosis of COVID-19, independently of the N-P swab result. We suggest the inclusion of these elements in the decision-making to facilitate the management of suspected COVID.