Early chest x-ray severity assessed by the MBrixia score is associated with mortality in hospitalized patients with COVID-19 pneumonia.

Chest x-ray (CXR) severity scores, such as the MBrixia score, are associated with COVID-19 severity, but most have not been validated after the introduction of antivirals and dexamethasone as standard of care. The MBrixia score's association with mortality and the interobserver variability among radiologists remain unclear. Adults hospitalized with COVID-19 at a tertiary center in Denmark between June 15, 2020 and February 14, 2021, had their CXR severity quantified from 0 to 36 using the MBrixia score. Association between MBrixia score and received respiratory support was assessed by Poisson regression. Correlation between MBrixia score and blood biomarkers was assessed using Kendall rank correlation. Interobserver variability between three blinded radiologists was assessed by Kappa values and Bland-Altman plots. Associations between MBrixia score at admission and 90-day mortality were examined by adjusted Cox Proportional Hazards models. Among 279 patients with 606 CXRs, higher levels of respiratory support were associated with a higher estimated mean MBrixia score. Nine of 15 investigated blood biomarkers were significantly correlated with MBrixia score. Moderate agreement between radiologists was observed, with higher agreement among senior radiologists. Among 251 patients with CXRs at time of admission, all patients had an MBrixia score < 27. Those falling in the ranges, 10-18 and 19-27, compared to 0-9, had an adjusted hazard ratio for 90-day mortality of 1.22 (CI 95% 0.66-2.26) and 3.13 (CI 95% 1.26-7.76), respectively. The MBrixia score was validated as a marker of disease severity, and scores > 18 at admission were associated with a threefold increase in 90-day mortality.