Objective: To explore a modified CT scoring system, its feasibility for disease severity evaluation and its predictive value in coronavirus disease 2019 (COVID-19) patients. Methods: This study was a multi-center retrospective cohort study. Patients confirmed with COVID-19 were recruited in three medical centers located in Beijing, Wuhan and Nanchang from January 27, 2020 to March 8, 2020. Demographics, clinical data, and CT images were collected. CT were analyzed by two emergency physicians of more than ten years' work experience independently through a modified scoring system. Final score was determined by average score from the two reviewers if consensus was not reached. The lung was divided into 6 zones (upper, middle, and lower on both sides) by the level of trachea carina and the level of lower pulmonary veins. The target lesion types included ground-glass opacity (GGO), consolidation, overall lung involvement, and crazy-paving pattern. Bronchiectasis, cavity, pleural effusion, etc., were not included in CT reading and analysis because of low incidence. The reviewers evaluated the extent of the targeted patterns (GGO, consolidation) and overall affected lung parenchyma for each zone, using Likert scale, ranging from 0-4 (0=absent; 1=1%-25%; 2=26%-50%; 3=51%-75%; 4=76%-100%). Thus, GGO score, consolidation score, and overall lung involvement score were sum of 6 zones ranging from 0-24. For crazy-paving pattern, it was only coded as absent or present (0 or 1) for each zone and therefore ranging from 0-6. Results: A total of 197 patients from 3 medical centers and 522 CT scans entered final analysis. The median age of the patients was 64 years, and 54.8% were male. There were 76(38.8%) patients had hypertension and 30(15.3%) patients had diabetes mellitus. There were 75 of the patients classified as moderate cases, as well as 95 severe cases and 27 critical cases. As initial symptom, dry cough occurred in 170 patients, 134 patients had fever, and 125 patients had dyspnea. Reparatory rate, oxygen saturation, lymphocyte count and CURB 65 score on admission day varied among patients with different disease severity scale. There were 50 of the patients suffered from deterioration during hospital stay. The median time consumed for each CT by clinicians was 86.5 seconds. Cronbach's alpha for GGO, consolidation, crazy-paving pattern, and overall lung involvement between two clinicians were 0.809, 0.712, 0.678, and 0.906, respectively, showing good or excellent inter-rater correlation. There were 193 (98.0%) patients had GGO, 147 (74.6%) had consolidation, and 126(64.0%) had crazy-paving pattern throughout clinical course. Bilateral lung involvement was observed in 183(92.9%) patients. Median time of interval for CT scan in our study was 7 days so that the whole clinical course was divided into stages by week for further analysis. From the second week on, the CT scores of various types of lesions in severe or critically patients were higher than those of moderate cases. After the fifth week, the course of disease entered the recovery period. The CT score of the upper lung zones was lower than that of other zones in moderate and severe cases. Similar distribution was not observed in critical patients. For moderate cases, the ground glass opacity score at the second week had predictive value for the escalation of the severity classification during hospitalization. The area under the receiver operating characteristic curve was 0.849, the best cut-off value was 5 points, with sensitivity of 84.2% and specificity of 75.0%. Conclusions: It is feasible for clinicians to use the modified semi-quantitative CT scoring system to evaluate patients with COVID-19. Severe/critical patients had higher scores for ground glass opacity, consolidation, crazy-paving pattern, and overall lung involvement than moderate cases. The ground glass opacity score in the second week had an optimal predictive value for escalation of disease severity during hospitalization in moderate patients on admission. The frequency of CT scan should be reduced after entering the recovery stage.
目的： 探讨新型冠状病毒肺炎（COVID-19）患者CT评分及其预测价值。 方法： 回顾性分析2020年1月27日至3月8日确诊的3个临床中心的197例COVID-19患者的临床及影像资料，患者中位年龄64岁，54.8%为男性。由两位临床医师进行CT评分。两侧肺野以气管隆凸水平和下肺静脉水平分为共6个区域。目标病变类型为磨玻璃影、肺实变、总体病变范围及是否存在铺路石征。 结果： 197例COVID-19患者共纳入可分析的CT图像522个。从第2周开始，重型及危重型患者各类病变CT评分均高于普通型患者。第5周以后病程进入恢复期。上肺区域CT评分低于其他区域。入院分级为普通型的患者，病程第2周的磨玻璃影评分对于住院期间严重程度分级的升级具有预测价值，其受试者工作特征曲线下面积0.849，最佳截断值为5分，敏感度84.2%，特异度75.0%。 结论： 由临床医师使用CT评分对新型冠状病毒肺炎患者进行评估具有可行性。病程第2周磨玻璃影评分对普通型患者住院期间病情恶化具有预测价值。恢复期可酌情降低复查CT频率。.