Severe acute respiratory syndrome: quantitative assessment from chest radiographs with clinical and prognostic...

This study examined findings of severe acute respiratory syndrome (SARS) on chest radiographs and presented a classification scheme using quantitative radiographic data supported by clinical parameters.

Three radiologists who were blinded to the identity, diagnosis, treatment protocol, and outcome of each patient independently evaluated serial chest radiographs from 67 patients with confirmed SARS. In addition to the chest radiographic abnormalities and percentage of involvement, several quantitative improvement parameters, including the peak to 50% improvement time (PIT(50)), were collected. Correlation between PIT(50) and clinical parameters (duration of fever, cough, dyspnea, oxygen supplementation, intubation, and death) were evaluated using Wilcoxon's rank sum testing and Spearman's correlation.

The most common initial findings were unifocal air-space disease in the periphery of the lower lungs occurring a mean of 3.6 +/-2.4 (SD) days from symptoms onset. Peak abnormalities were seen at 10.4 +/- 2.9 days. PIT(50) was dependent on disease severity, showing a strong linear correlation with the clinical parameter duration of oxygen supplementation (r = 0.44, p = 0.0015). Three patterns of disease were recognized: pattern A (severe, 29.9%) with PIT(50) of more than 10 days, pattern B (typical, 44.8%) with PIT(50) of 10 or fewer days, and pattern C (mild, 25.4%) with minimal findings throughout the course of the disease. This classification was supported by collaborative clinical parameters.

The quantitative radiographic parameter PIT(50) has strong clinical correlation and can be used to differentiate severity of disease into severe, typical, and mild types.