Definition and retrospective application of a clinical scoring system for COVID-19 triage at presentation

Background: A simple scoring system for triage of suspected patients with COVID-19 is lacking. Methods: A multi-disciplinary team developed a screening score taking into account epidemiology history, clinical feature, radiographic feature, and routine blood test. At fever clinics, the screening score was used to identify the patients with moderate to high probability of COVID-19 among all the suspected patients. The patients with moderate to high probability of COVID-19 were allocated to a single room in an isolation ward with level-3 protection. And those with low probability were allocated to a single room in a general ward with level-2 protection. At the isolation ward, the screening score was used to identify the confirmed and probable cases after two consecutive real-time reverse transcription polymerase chain reaction (RT-PCR) tests. The data in the People’s Hospital of Changshou District were used for internal validation and those in the People’s Hospital of Yubei District for external validation. Results: We enrolled 76 and 40 patients for internal and external validation, respectively. In the internal validation cohort, the area under the curve of receiver operating characteristics (AUC) was 0.96 [95% confidence interval (CI): 0.89–0.99] for the diagnosis of moderate to high probability of cases among all the suspected patients. Using 60 as cut-off value, the sensitivity and specificity were 88% and 93%, respectively. In the isolation ward, the AUC was 0.94 (95% CI: 0.83–0.99) for the diagnosis of confirmed and probable cases. Using 90 as cut-off value, the sensitivity and specificity were 78% and 100%, respectively. These results were confirmed in the validation cohort. Conclusion: The scoring system provides a reference on COVID-19 triage in fever clinics to reduce misdiagnosis and consumption of protective supplies. The reviews of this paper are available via the supplemental material section.