Diagnostic performance of the Sofia Analyzer in outpatient clinics

Abstract

BACKGROUND: Rapid influenza diagnostic tests (RIDTs) are frequently used to allow timely indication of influenza virus infections at the point of care, informing clinical management and infection control measures. The objective of our study was to assess the diagnostic accuracy of the Sofia Influenza A+B Fluorescent Immunoassay Analyzer (Quidel, San Diego, CA) compared to RTPCR in outpatient clinics. METHOD: We recruited patients with acute respiratory illness (ARI) from January 2013 through April 2015 in outpatient clinics in Hong Kong. ARI was defined as having ≥2 of the following signs or symptoms: fever ≥37.8°C, cough, sore throat, runny nose, headache, myalgia and phlegm. Two respiratory specimens were collected from each participant for testing by the Sofia Analyzer according to the manufacturer’s instructions and subsequent testing by quantitative RT-PCR for influenza A and B respectively. We estimated the point estimates and the exact binomial 95% confidence intervals for the measures of diagnostic accuracy for the Sofia RIDT; and assessed the associations of the covariates including viral load and age with the result of the RIDT by logistic regression. RESULTS: Data were available from 1230 outpatients. The Sofia RIDT had an overall sensitivity of 77% (95% CI 71-82%) and specificity of 91% (95% CI 89-92%) for influenza virus infection confirmed to RT-PCR; a sensitivity and specificity of 74% (95% CI 68-81%) and 97% (95% CI 96-98%) for influenza A, and 77% (95% CI 63-87%) and 94% (95% CI 92-95%) for influenza B respectively. The Sofia RIDT had a threshold for high sensitivity at viral loads 5 log10 copies/ml or above for both Influenza A and B. Viral load remained significantly associated with a positive RIDT result after adjusting for days since symptom onset, age and sex (adjusted OR: 1.98, 95% CI: 1.83-2.15). A prevalence of influenza of 22% would be required for the Sofia RIDT to correctly detect more than 70% of influenza virus infections while a prevalence of 10% would give the test a positive predictive value of 49%. CONCLUSION: Compared to other RIDTs, the estimated sensitivity and specificity for Sofia RIDT is considered high and moderate respectively, and has good positive predictive value during influenza epidemics.