ABSTRACT
BACKGROUND: Clinical practice guidelines for influenza have been implemented to maximise the appropriate use of empirical oseltamivir; however, good predictive values are required. METHODS: Between October 2011 and September 2013, children aged < 15 years who presented at the Bamrasnaradura Infectious Diseases Institute with an influenza-like illness plus either (i) pneumonia or (ii) being in a higher risk group for influenza complications were prospectively enrolled. Respiratory specimens were taken for real-time polymerase chain reaction testing (RT-PCR). Clinical characteristics, laboratory data and oseltamivir therapy were recorded. RESULTS: 85 cases were enrolled. Of these, the proportions of those with pneumonia, who were aged < 2 years and who had underlying diseases were 74.1%, 56.5% and 38.8%, respectively. RT-PCR detected respiratory syncytial virusamong (35.3%), influenza (22.3by%), adenovirus (14.1%), human metapneumovirus (5.9%), para-influenza (3.5%) and no viruses (25.9 %). Pneumonia (OR 0.16, 95% CI 0.05-0.50) and having two clinical criteria (OR 0.24, 95% CI 0.08-0.76) were significantly negative predictors of influenza. Having cluster transmissions (OR 5.18, 95% CI 1.38-19.37) and a monocyte proportion >7% (OR 3.58, 95% CI 1.15-11.17) were significantly positive predictors of influenza. The mean (SD) percentage of influenza-like illness during the study period was 7.04 (2.02). CONCLUSIONS: Clinical criteria guidelines yielded a low predictive value (22.3%) for influenza in children. Seasonality, cluster transmission, white blood cell and differential counts may be helpful in diagnosing influenza. Nonetheless, empirical oseltamivir should not be delayed for those in need.