GIÁ TRỊ DỰ BÁO TỬ VONG CỦA ĐỘ NẶNG TỔN THƯƠNG PHỔI TRÊN X-QUANG TẠI THỜI ĐIỂM NHẬP VIỆN Ở BỆNH NHÂN COVID-19

SUMMARY

Background: We aimed to investigate the performance of a chest X-ray (CXR) scoring scale of lung injury in prediction of death among patients with COVID-19 admitted at Vinmec Central Park hospital (HCM City, VN) during the peak epidemic in 2021.

Method: Retrospective design; X-ray images and clinical data were collected from all hospitalized patients with SARS-CoV-2 PCR positive from July to September 2021. Three radiologists independently assessed the CXR score at admission which consists of measuring both severity and extent of lung injuries on four lung quadrants (scale o to 24). Association between CXR and mortality risk was estimated using a survival regression with log-log distribution.

Result: The study included 219 patients (mortality rate = 28). There was a high consensus for CXR scoring among 3 radiologists (κ = 0.90; CI95%: 0.89-0.92). PCA analysis revealed that CXR has a similar role as CRP score for predicting risk of mortality. CXR score was the strongest predictor of mortality (tdAUC 0.85 CI95% 0.69–1) within the first 3 weeks after admission, compared with other conventional clinical features. After adjusting for the patient’s age, there was a significant effect of increased CXR score on mortality risk (HR = 1.15, CI95%:1.04-1.27, p=0.009). At a critical threshold of 16 points, the CXR score allows for predicting inhospital mortality with good sensitivity (0.82; CI95%: 0.78 to 0.87) and specificity (0.89; CI95%: 0.88 to 0.90).

Conclusion: The day-one CXR score is an independent and effective predictor of the risk of death in COVID-19 and could be used to identify high-risk patients in countries like Vietnam where CXR is more readily available than CT scans.

Keywords: COVID-19, Chest X-ray, Lung injury score, Mortality prediction