Unenhanced computed tomography (CT) utility for triage at the emergency department during COVID-19 pandemic

Abstract
Background: Unenhanced chest computed tomography (CT) can assist in the diagnosis and classification of coronavirus disease 2019 (COVID-19), complementing to the reverse-transcription polymerase chain reaction (RT-PCR) tests; the performance of which has yet to be validated in emergency department (ED) setting. The study sought to evaluate the diagnostic performance of chest CT in the diagnosis and management of COVID-19 in ED.
Methods: This retrospective single-center study included 155 patients in ED who underwent both RT-PCR and chest CT for suspected COVID-19 from March 1st to April 1st, 2020. The clinical information, CT images and laboratory reports were reviewed and the performance of CT was assessed, using the RT-PCR as standard reference. Moreover, an adjudication committee retrospectively rated the probability of COVID-19 before and after the CT calculating the net reclassification improvement (NRI). Their final diagnosis was considered as reference. The proportion of patients with negative RT-PCR test that was directed to the referent hospital based on positive CT findings was also assessed.
Results: Among 155 patients, 42% had positive RT-PCR results, and 46% had positive CT findings. Chest CT showed a sensitivity of 84.6%, a specificity of 80.0% and a diagnostic accuracy of 81.9% in suggesting COVID-19 with RT-PCR as reference. Concurrently, corresponding values of 89.4%, 84.3% and 86.5% were retrieved with the adjudication committee diagnosis as reference. For the subgroup of patients with age > 65, specificity and sensitivity were 50% and 80.8%, respectively. In patients with negative RT-PCR results, 20% (18/90) had positive chest CT finding and 22% (4/18) of those were eventually considered as COVID-19 positive according to the adjudication committee. After CT, the estimated probability of COVID-19 changed in 10/104 (11%) patients with available data: 4 (4%) were downgraded, 6 (6%) upgraded. The NRI was 1.92% (NRI event -2.08% + NRI non-event 5.36%). No patient with negative RT-PCR but positive CT was eventually directed to hospital.
Conclusion: Chest CT showed promising sensitivity for diagnosing COVID-19 across all patients’ subgroups. However, CT did not modify the estimated probability of COVID-19 infection in a substantial proportion of patients and its utility as an emergency department triage tool warrants further analyses.
Keywords: COVID-19; CT; Computed tomography; Coronavirus; Pandemic; RT-PCR; Screening; Triage.