Differences in clinical predictors of influenza in adults and children with influenza-like illness

Abstract

Influenza contributes significantly to morbidity and mortality in the winter season. The aim of the study was to identify clinical signs and symptoms most predictive of influenza infection in children and adults with influenza-like illness. A prospective systematic sampling analysis of clinical data collected through sentinel surveillance system for influenza in 32 primary care centers and one tertiary care hospital in Slovenia during two consecutive influenza seasons (2004/2005 and 2005/2006) was carried out Children and adults who had influenza-like illness, defined as febrille illness with sudden onset, prostration and weakness, muscle and joint pain and at least (cough, sore throat, coryza) were included and tested for influenza A and B virus, adenovirus, respiratory syncytial virus and enterovirus by RT-PCR. Clinical data were evaluated in statistical models to identify the best predictors for the confirmation of influenza for children (under age of 15) and adults. of 1,286 patients with influenza-like symptoms in both seasons 211 were confirmed to have influenza A or B alone and compared to 780 influenza-negative patients. A fever over 38°C, chills, headache, malaise and sore eyes revealed a significant association with positive RT-PCR test for influenza virus in children. In adults, only three symptoms were significantly related to PCR-confirmed influenza infection: fever, cough and abnormal breath sounds. The stepwise logistic regression analysis showed that four symptoms predicted influenza in children: fever (38°C or more) (p=0.010), headache (p=0.030), cough (p=0.044) and absence of abnormal breathing sounds (p=0.015) with sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) of 5.1%, 98.1%, 57.1% and 80.1%, respectively. For adults, the strongest impact on influenza positivity was found for fever (p=0.008) and cough (p=0.085). The model for adufe had less favorable characteristics, with sensitivity, specificity, PPV and NPV of 0%, 100%, 0% and 76.4%, respectively. Differences in clinicai predictors of influenza in children compared to adults were found. The model for adufe was acceptable but not a good one. The model for children was found to be more reliable than the prediction model for adults. © Versita Warsaw and Springer-Verlag Berlin Heidelberg.