A Risk Classification Model To Predict Mortality Among Laboratory-Confirmed Avian Influenza A H7n9 Patients: A Population-Based Observational Cohort Study

Leonardo Martinez,Wei Cheng,Xiaoxiao Wang,Feng Ling,Lan Mu,Changwei Li,Xiang Huo,Mark H Ebell,Haodi Huang,Limei Zhu,Chao Li,Enfu Chen,Andreas Handel,Ye Shen

DOI: https://doi.org/10.1093/infdis/jiz328

2019-01-01

Abstract:Background. 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.Methods. 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.Results. 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% (P-trend < .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% (P-trend = .0063).Conclusions. We developed and validated a simple-to-use, predictive risk score for clinical use, identifying patients at high mortality risk.