Department of Epidemiology and Biostatistics, College of Public Health, Athens.
Division of Infectious Diseases and Geographic Medicine, School of Medicine, Stanford University, California.
J Infect Dis. 2019 Oct 22;220(11):1780-1789. doi: 10.1093/infdis/jiz328.
Avian influenza A H7N9 (A/H7N9) is characterized by rapid progressive pneumonia and respiratory failure. Mortality among laboratory-confirmed cases is above 30%; however, the clinical course of disease is variable and patients at high risk for death are not well characterized.
We obtained demographic, clinical, and laboratory information on all A/H7N9 patients in Zhejiang province from China Centers for Disease Control and Prevention electronic databases. Risk factors for death were identified using logistic regression and a risk score was created using regression coefficients from multivariable models. We externally validated this score in an independent cohort from Jiangsu province.
Among 305 A/H7N9 patients, 115 (37.7%) died. Four independent predictors of death were identified: older age, diabetes, bilateral lung infection, and neutrophil percentage. We constructed a score with 0-13 points. Mortality rates in low- (0-3), medium- (4-6), and high-risk (7-13) groups were 4.6%, 32.1%, and 62.7% (Ptrend < .0001). In a validation cohort of 111 A/H7N9 patients, 61 (55%) died. Mortality rates in low-, medium-, and high-risk groups were 35.5%, 55.8, and 67.4% (Ptrend = .0063).
We developed and validated a simple-to-use, predictive risk score for clinical use, identifying patients at high mortality risk.
甲型 H7N9 禽流感（A/H7N9）的特征为快速进展性肺炎和呼吸衰竭。经实验室确诊的病例死亡率超过 30%；然而，疾病的临床病程是多变的，且死亡风险高的患者特征并不明显。
我们从中国疾病预防控制中心电子数据库中获取了浙江省所有 A/H7N9 患者的人口统计学、临床和实验室信息。使用逻辑回归确定死亡的危险因素，并使用多变量模型的回归系数创建风险评分。我们在江苏省的一个独立队列中对该评分进行了外部验证。
在 305 例 A/H7N9 患者中，有 115 例（37.7%）死亡。确定了四个独立的死亡预测因素：年龄较大、糖尿病、双肺感染和中性粒细胞百分比。我们构建了一个 0-13 分的评分。低危（0-3 分）、中危（4-6 分）和高危（7-13 分）组的死亡率分别为 4.6%、32.1%和 62.7%（趋势 P<0.0001）。在 111 例 A/H7N9 患者的验证队列中，有 61 例（55%）死亡。低危、中危和高危组的死亡率分别为 35.5%、55.8%和 67.4%（趋势 P=0.0063）。
我们开发并验证了一种简单易用的预测风险评分，用于临床应用，可识别高死亡率风险的患者。
