The effects of influenza-like illness definitions on influenza surveillance in hospitalized children / 中华疾病控制杂志

ABSTRACT

Objective To evaluate clinical factors associated with laboratory-confirmed influenza infection and probe into the effects of different influenza-like illness(ILI) definitions on influenza surveillance in hospitalized children. Methods The influenza surveillance on the hospitalized children with acute respiratory infection was carried out in Soochow university affiliated children’s hospital from October 2017 to May 2018. The definition of influenza-like illness(ILI), which proposed by world health organization (WHO), the European center for disease prevention and control (ECDC) and ministry of public health of China, was analyzed to determine the area under receiver operating characteristic curve (ROC), sensibility, specificity, and the positive and negative predictive values of the ILI definition using the laboratory evidence of influenza virus infection as golden criteria. Logistic regression model was employed to explore the risk factors of the laboratory confirmed influenza infection among the hospitalized children. Results Of the enrolled 1 459 hospitalized children, 204(14.0%) were lab-confirmed influenza cases. The ECDC definition had the highest sensitivity (91.7%, 95% CI 87.9%-95.5%) but the lowest specificity (44.6%, 95% CI 41.9%-47.4%). The WHO definition had the lowest specificity (70.6%, 95% CI 64.3%-76.8%). China’s definition was the most discriminant definition with relatively high sensitivity (91.2%, 95% CI 87.3%-95.1%) and specificity (51.5%, 95% CI 48.8%-54.3%), and its positive area under curve value (71.2%, 95% CI 67.9%-74.5%) was the highest. Multivariate analysis model showed that the detection rate of influenza virus in children with fever (≥38 ℃) was higher than that in children without fever (OR=7.03, 95% CI3.89-12.70). Conclusions It is suggested to adopt China’s ILI definition to get better output during influenza surveillance among hospitalized children.