| Summary: | BACKGROUND: Influenza is a viral infection that can lead to serious complications and death(s) in vulnerable groups if not diagnosed and managed in a timely manner. This study was conducted to improve the accuracy of predicting influenza through various clinical and statistical models. METHODOLOGY: A retrospective cross sectional analysis was done on demographic and epidemiological data collected from March 2009 to March 2010. Patients were classified as ILI or SARI using WHO case definitions. Respiratory specimens were tested by RT-PCR. Clinical symptoms and co-morbid conditions were analyzed using binary logistic regression models. RESULTS: In the first approach, analysis compared children (≤12) and adults (>12). Of 1,243 cases, 262 (21%) tested positive for A(H1N1)pdm09 and the proportion of children (≤12) and adults (>12) were 27% and 73% respectively. Four symptoms predicted influenza in children: fever (OR 2.849, 95% CI 1.931-8.722), cough (OR 1.99, 95% CI 1.512-3.643), diarrhea (OR 2.100, 95% CI 2.040-3.25) and respiratory disease (OR 3.269, 95% CI 2.128-12.624). In adults, the strongest clinical predictor was fever (OR 2.80, 95% CI 1.025-3.135) followed by cough (OR 1.431, 95% CI 1.032-2.815). In the second instance, patients were separated into two groups: SARI 326 (26%) and ILI 917 (74%) cases. Male to female ratio was 1.41∶1.12 for SARI and 2∶1.5 for ILI cases. Chi-square test showed that fever, cough and sore throat were significant factors for A(H1N1)pdm09 infections (p = 0.008). CONCLUSION: Studies in a primary care setting should be encouraged focused on patients with influenza-like illness to develop sensitive clinical case definition that will help to improve accuracy of detecting influenza infections. Formulation of a standard "one size fits all" case definition that best correlates with influenza infections can help guide decisions for additional diagnostic testing and also discourage unjustified antibiotic prescription and usage in clinical practice.
|
|---|
