Determinants of Chest Radiography Sensitivity for COVID-19: A Multi-Institutional Study in the United States
Stephanie Stephanie, Thomas Shum, Heather Cleveland, Suryanarayana Reddy Challa, Allison Herring, Francine L. Jacobson, Hiroto Hatabu, Suzanne Byrne, Kumar Shashi, Tetsuro Araki, Jose A. Hernandez, Charles S. White, Rydhwana Hossain, Andetta R. Hunsaker, Mark M. Hammer
Abstract
Purpose To evaluate the sensitivity, specificity, and severity of chest radiographs and chest CT scans over time in patients confirmed positive for COVID-19 and those confirmed negative for COVID-19 and to evaluate determinants of false-negative results. Materials and Methods In a retrospective multi-institutional study, 254 patients with reverse-transcription polymerase chain reaction–verified COVID-19, who underwent at least one chest radiography examination or chest CT, were compared with 254 age- and sex-matched controls who were confirmed negative for COVID-19. Chest radiograph severity, sensitivity, and specificity were determined with respect to time after onset of symptoms; sensitivity and specificity for chest CT scans were determined without time stratification. Performance of serial chest radiographs against CT scans was determined by comparing area under the receiver operating characteristic curves (AUC). A multivariable logistic regression analysis was performed to assess factors related to false-negative findings on chest radiographs. Results COVID-19–positive chest radiograph severity and sensitivity increased with time (from sensitivity of 55% at ≤ 2 days to 79% at > 11 days; P < .001 for trends of both severity and sensitivity), whereas chest radiograph specificity decreased over time (from 83% to 70%, P = .02). The findings of serial chest radiographs demonstrated an increase in AUC (first chest radiograph, AUC = 0.79; second chest radiograph, AUC = 0.87; P = .02), and second chest radiographs approached the accuracy of CT (AUC = 0.92, P = .11). COVID-19 sensitivity of first chest radiograph, second chest radiograph, and CT was 73%, 83%, and 88%, whereas specificity was 80%, 73%, and 77%, respectively. Normal and mild severity chest radiograph findings were the largest factor behind false-negative findings on chest radiographs (40% normal and 87% combined normal/mild). Young age and African American ethnicity increased false-negative finding rates. Conclusion Chest radiography sensitivity in COVID-19 detection increases with time, and serial chest radiography of patients confirmed positive for COVID-19 has accuracy approaching that of chest CT. Supplemental material is available for this article. Keywords: Adults, CT, Conventional Radiography, Infection, Lung, Pulmonary © RSNA, 2020