Predictive Symptoms and Signs of Laboratory-confirmed Influenza: A Prospective Surveillance Study of Two Metropolitan Areas in Taiwan

Influenza infection poses annual threats and leads to significant morbidity and mortality. Early diagnosis is the key to successful treatment. Laboratory-based diagnosis has various limitations. Diagnosis based on symptoms or signs is still indispensable in clinical practice. We investigated the symptoms or signs associated with laboratory-confirmed influenza. A prospective study across 2 influenza seasons was performed from June 2010 to June 2012 at 2 branches (Taipei and Lin-Kou) of Chang Gung Memorial Hospital. Patients who visited outpatient clinics with suspected acute respiratory tract infection were sampled by throat swab or nasopharyngeal swab. RT-PCR and/or virus culture were used as a reference standard. We used logistic regression to identify the symptoms or signs associated with laboratory-confirmed influenza infection. We also evaluated the performance metrics of different influenza-like illness used in Taiwan, the USA, and WHO. A total of 158 patients were included in the study. The prevalence of influenza infection was 45% (71/158). Fever, cough, rhinorrhea, sneezing, and nasal congestion were significant predictors for influenza infection. Whereas fever + cough had a best sensitivity (86%; confidence interval [CI] 76%–93%), fever + cough and sneezing had a best specificity (77%; CI 62%–88%). Different case definitions of influenza-like illness had comparable accuracy in sensitivity and specificity. Clinical diagnosis based on symptoms and signs is useful for allocating resources, identifying those who may benefit from early antiviral therapy and providing valuable information for surveillance purpose.