Abstract: Background　The COVID-19 has become a global epidemic.How to effectively treat it and reduce the fatality rate is still at exploratory stage.Identification of risk factors for death is critical to optimize treatment strategies for COVID-19.Objective　To investigate risk factors for COVID-19 death，and based on this，to develop a COVID-19 death predictive scoring system.Methods　Two hundred and seventy patients，who discharged or deceased with confirmed COVID-19 in Zhongnan Hospital of Wuhan University during January 1 to February 29，2020，were reviewed.Baseline data were obtained，including demographics，admission symptoms and vital signs，complications and laboratory test results.According to the clinical outcome，i.e.discharged and deceased，the patients were divided into recovery group and death group.Risk factors of death were explored by using multivariate logistic regression analysis，and used for the development of a predictive scoring system for death.Results　Two hundred and forty-five cases were finally included，including 212 discharged and 33 deceased during hospitalization.Factors independently associated with death were age ≥ 65 years〔OR=7.177，95%CI（1.715，30.038），P<0.05〕，SpO2≤93%〔OR=15.456，95%CI（3.343，71.450），P<0.05〕，BUN≥7 mmol/L〔OR=7.115，95%CI（1.550，32.652），P<0.05〕，PCT≥0.1 μg/L〔OR=23.895，95%CI（4.209，135.639），P<0.05〕.In predicting death due to COVID-19，the AUC of ASBP（A:age，S:SpO2，B:BUN，P:PCT） scoring system was 0.967〔95%CI（0.931，0.987）〕，and that of CURB-65 scoring system was 0.885〔95%CI（0.831，0.926）〕，showing a significant difference（Z=2.816，P<0.01）.5 was the maximal Youden Index for ASBP scoring system and was chosen as the cut-off value，with 0.871 sensitivity and 0.957 specificity.And 2 was the maximal Youden Index for CURB-65 scoring system and was chosen as the cut-off value，with 0.903 sensitivity and 0.735 specificity.Conclusion　Older age，lower SpO2，higher levels of BUN and PCT were independent risk factors for COVID-19 death，and the ASBP scoring system developed based these factors might be available for the assessment of death risk of COVID-19.

Key words: COVID-19, SARS-CoV-2, Mortality, Prognosis, Risk factors, Wuhan

摘要： 背景　新型冠状病毒肺炎（COVID-19）已全球流行，如何有效治疗、降低病死率尚处于摸索阶段。寻找死亡危险因素将有利于优化COVID-19治疗策略。目的　探讨COVID-19患者死亡的危险因素并构建预测评分系统。方法　选取2020-01-01至02-29在武汉大学中南医院住院的COVID-19患者270例，收集患者基线资料，包括一般资料、入院症状、合并症、入院生命体征、实验室检查。根据临床结局即治愈出院和死亡分为痊愈组和死亡组。采用多因素Logistic回归分析探讨COVID-19患者死亡的危险因素，基于此构建COVID-19患者死亡预测评分系统。结果　最终245例患者纳入研究，其中治愈出院212例（痊愈组），死亡33例（死亡组）。多因素Logistic回归分析结果显示，年龄≥65岁〔OR=7.177，95%CI（1.715，30.038），P<0.05〕、氧饱和度（SpO2）≤93%〔OR=15.456，95%CI（3.343，71.450），P<0.05〕、血尿素氮（BUN）≥7 mmol/L〔OR=7.115，95%CI（1.550，32.652），P<0.05〕、血降钙素原（PCT）≥0.1 μg/L〔OR=23.895，95%CI（4.209，135.639），P<0.05〕是COVID-19患者死亡的独立危险因素。ASBP（A：年龄，S：SpO2，B：BUN，P：PCT）评分系统对COVID-19患者死亡预测价值的ROC曲线下面积（AUC）95%CI为0.967（0.931，0.987），CURB-65评分系统对COVID-19患者死亡预测价值的AUC 95%CI为0.885（0.831，0.926）。两种评分系统AUC比较，差异有统计学意义（Z=2.816，P<0.01）。选择约登指数最大者作为预测截断值，即ASBP评分系统为5分、CURB-65评分系统2分，相应的灵敏度分别为0.871、0.903，特异度分别为0.957、0.735。结论　年龄≥65岁、SpO2≤93%、BUN≥7 mmol/L和PCT≥0.1 μg/L是COVID-19死亡的独立危险因素，以此构建的ASBP评分系统可用于COVID-19死亡预警。

关键词: 新型冠状病毒肺炎, 新型冠状病毒, 死亡率, 预后, 危险因素, 武汉