Feasibility, Reproducibility, and Clinical Validity of a Quantitative Chest X-Ray Assessment for COVID-19

Abstract

Chest X-ray (CXR) is an essential first-line tool in COVID-19 pneumonia diagnosis and management. Our study aimed at assessing 1) CXR manifestations, frequency, and distribution; 2) the feasibility and repeatability of a CXR severity score; and 3) the correlation between the CXR severity score and clinical and laboratory parameters. We reviewed baseline CXRs and clinical data of consecutive patients who presented to our emergency department and resulted positive at SARS-CoV-2 reverse transcriptase-PCR oropharyngeal swab test from March 1, 2020 to April 6, 2020. Lung abnormalities and their distribution were analyzed. A score of CXR severity was assigned by two radiologists, independently, according to the extent of lung involvement, with a maximum score of 8 for CXR. Correlations between the CXR score and the clinical data were assessed. One hundred fifty-five patients were included; 143/155 (92%) were positive at baseline CXR. Ground-glass opacity was the most common finding (141/143, 99%). Involvement was mainly bilateral (96/143, 67%), with peripheral distribution (79/143, 55%). The mean CXR severity score was 3.3 (±2); interobserver agreement was excellent, with a Cohen's K correlation coefficient of 0.901. The CXR score showed a significant positive correlation with C-reactive protein, lactate dehydrogenase, and fever duration, and a negative correlation with oxygen saturation. Chest X-ray findings are in line with those reported by computed tomography studies. The use of a visual CXR score, easy to assess and highly reproducible, can reflect the clinical severity and help the patients' management.

Figure 1.

(A–C) Examples of chest X-ray abnormalities. (A) A 37-year-old woman with a 3-day fever, cough, and conjunctivitis. Chest X-ray shows a focal ground-glass opacity involving the lower field of the left lung (white rectangle). (B) An 86-year-old woman presented with dyspnea (SpO₂ 88%), fever, and cough. Chest X-ray shows an area of consolidation in the middle-lower left fields (black rectangle). Ground-glass opacification (GGO) is recognizable in the right lung. (C) A 77-year-old man with a history of diabetes and arterial hypertension presented to the emergency department with a 7-day fever (38°C), dry cough, and dyspnea with low SpO₂ (84%). Chest X-ray shows bilateral consolidations in middle lung fields (black rectangles). Bilateral areas of GGOs are also recognizable, particularly evident in the lower field of the right lung (white rectangle).

Figure 2.

(A–D) Examples of chest X-ray (CXR) score assignment. (A) Chest X-ray showing a focal area of ground-glass opacifications (GGOs) in the upper field of the right lung (black arrow). The involvement of the right lung was < 25%; therefore, the CXR severity score assigned was 1. (B) Chest X-ray showing bilateral areas of GGOs involving the lower lung zones (black arrows). On both the left and right lungs, the involvement was < 50%; therefore, the score was 2 for each lung, with a global score of 4. (C) Chest X-ray showing huge areas of GGOs with bilateral involvement (black arrows), and saving of the upper field of the right lung; the extension on the left side was > 75% (score 4), whereas the involvement on the right side was < 75% (score 3); therefore, the overall score was 4 + 3 = 7. (D) Chest X-ray showing bilateral involvement, with areas of GGOs and consolidation (black arrows) involving all the lung fields. On both the left and right lungs, the involvement was > 75% (score 4); therefore, the global score was 4 + 4 = 8.

Figure 3.

Chest X-ray severity score showed a significant positive correlation with C-reactive protein blood levels (P < 0.001; r-value 0.545). This figure appears in color at www.ajtmh.org.

Figure 4.

Chest X-ray severity score showed a significant positive correlation with lactate dehydrogenase blood levels (P < 0.001; r-value 0.770). This figure appears in color at www.ajtmh.org.

Figure 5.

Chest X-ray severity score showed a significant negative correlation with oxygen saturation (P < 0.001; r-value −0.547).This figure appears in color at www.ajtmh.org.