Three hundred sixteen NPA were received. Sixteen specimens were excluded because of insufficient quantity, leaving 300 specimens for analysis. Fifty-three influenza virus strains were isolated in cell culture. Of those, 48 (90.5%) were influenza B virus strains. One specimen had both RSV and parainfluenza virus.
Thirty-five specimens were only tested for RSV and parainfluenza virus by DFA and were positive for one of the viruses. Table shows the performance of the QuickVue test and DFA compared to that of viral cell culture. There were 85 positive results detected by QuickVue, for a sensitivity of 79.2% (42 of 53) and a specificity of 82.6% (204 of 247), with a positive predictive value (PPV) of 49.4% (42 of 85) and a negative predictive value (NPV) of 94.9% (204 of 215). When reading QuickVue, 41 of 85 positive results were difficult to interpret (very fade pink line) and 36 of those were false-positive results. In accordance with the manufacturer's recommendations, we labeled them as positive. When these dubious results were considered negative, the sensitivity decreased to 69.8% (37 of 53) but the specificity improved to 97.1% (240 of 247), which was comparable to the specificity stated in the package insert for nasal washes: 99% (95% confidence interval [CI], 93 to 99%). The PPV and NPV were 84.1% (37 of 44) and 93.8% (240 of 256), respectively.
Performance of rapid tests for the detection of influenza virus infection in NPA compared with that of viral culture
One hundred thirty-one specimens were assayed by QuickVue after being thawed once. Table shows the performance of the test comparing specimens that were thawed and those that were assayed fresh.
Performance of Quick Vue with thawed and fresh specimens
Of the 265 specimens on which DFA was performed, 43 were positive. The sensitivity and specificity of DFA were 73.6% (39 of 53) and 98.1% (208 of 212), respectively, with a PPV of 90.7% (39 of 43) and an NPV of 93.7% (208 of 222). Only 11 specimens (3.6%) had noninterpretable results by DFA. Excluding these results did not change the performance of the test.
Immunoassays for rapid detection of influenza virus have been studied in the past. Directigen Flu-A (Becton Dickinson, Cockeysville, Md.), an enzyme immunoassay membrane test, has reported sensitivities ranging from 64.2 to 84.7% and reported specificities ranging from 90 to 100%, respectively (2
). ZstatFlu (ZymeTx), a neuraminidase detection assay for both influenza viruses, has a reported sensitivity of 70.1% and a reported specificity of 92.4% for nasal wash specimens from children with respiratory infections (8
The sensitivity and specificity of QuickVue are similar to those of Directigen and ZstatFlu. The strains that were isolated this year were mainly influenza B viruses. When the sensitivity of QuickVue was calculated for influenza A virus (only five strains), it rose to 80%, compared to 68.8% for influenza B virus. Noyola et al. found a greater sensitivity for ZstatFlu when testing for influenza A virus (76.4%) than when testing for influenza B virus (40.9%) (8
DFA of NPA had a sensitivity similar to that of QuickVue, with good specificity. Specimens of poor quality (<25 cells/well) that were not interpretable by DFA were also negative by QuickVue, even if the culture was positive. In the literature, the reported sensitivities of DFA range from 59.3 to 84% and the reported specificities range from 87.7 to 100%, which are compatible with our results (1
The cost of QuickVue was comparable to that of cell culture in our laboratory. However, the former allowed detection of influenza virus only and distinction between types A and B was impossible. QuickVue was rapid and extremely easy to perform, but specimens containing VTM are likely not optimal specimens for the assay. QuickVue would still make a useful test for emergency rooms, where decisions about influenza treatment need to be timely. DFA of the specimen, however, had the lowest cost and allowed us to detect the presence of other respiratory viruses with good sensitivity and high specificity.