Recognition of early risky factors in patients suffering from critical influenza A H1N1

Abstract

To investigate clinical risky factors and improve the ability to discover critically ill patients suffering from influenza A H1N1 in emergency department. A retrospective study of 65 cases with severe influenza A H1N1 was conducted. Univariate analysis and multivariate Logistic regression analysis were used to identify independent predictors of critically ill cases. Goodness of Fit test was performed on an established clinical predicting scoring (prediction of critically ill with influenza score, PCIS) system. Univariate analysis showed that significant difference between critically ill and severe group consisted of eight factors including age distribution, more than one comorbidity, chronic cardiovascular disease, oncologic diseases, sputum, myalgia/arthralgia, dyspnea and oxygen saturation at admission. A multivariate Logistic regression showed an association between development of critical illness and more than one comorbidity [odds ratio (OR)=7.611, 95% confidence interval (95%CI) 1.635-35.429], myalgia/arthralgia (OR=7.523, 95%CI 1.462-38.716), dyspnea (OR=11.090, 95%CI 1.373-89.565), and oxygen saturation<0.95 at admission (OR=8.088, 95%CI 1.019-84.969, all P<0.05). The prognostic criteria had a good discriminative ability [area under receiver operating characteristic curve (AUC) was 0.922, 95%CI 0.860-0.985, P=0.000]. The PCIS scoring system was established according to the level of four high risk factors. Low risk (PCIS 0-1), intermediate risk (PCIS 2), high risk (PCIS 3), and very high risk (PCIS≥4) were categorized for predicting the occurrence of critical illness, and the Goodness of Fit test was good (R(2)=0.940 6, P=0.030 1). Emergency physician can predict the development of critical condition in patients with influenza A H1N1 by using clinical characteristics including comorbidity, myalgia/arthralgia, dyspnea, and oxygen saturation at admission, and it is helpful in making clinical decision.