Rapid Influenza Antigen Test for Diagnosis of Pandemic (H1N1) 2009

We compared the QuickVue Influenza test with PCR for diagnosing pandemic (H1N1) 2009 in 404 persons with influenza-like illness. Overall sensitivity, specificity, and positive and negative predictive values were 66%, 84%, 84%, and 64%, respectively. Rapid test results should be interpreted cautiously when pandemic (H1N1) 2009 virus is suspected.

S ince its emergence, the pandemic (H1N1) 2009 virus has spread rapidly throughout the world. To diagnose infl uenza at the point of care, many clinicians rely on commercial rapid enzyme immunoassay tests, which are currently unable to differentiate between infl uenza A virus subtypes (1). Compared with PCR and viral culture, the sensitivity of rapid tests for seasonal infl uenza varies from 70% to 90% in children and <40% to 60% in adults (2,3). The positive and negative predictive values (PPVs and NPVs) of rapid tests depend on the prevalence of infl uenza viruses among the population being tested (2,3).
We compare PCR with a rapid infl uenza test to better characterize the diagnostic utility of the rapid test during the current pandemic. The QuickVue Infl uenza test (Quidel Corp., San Diego, CA, USA) detects infl uenza A and B viruses but does not distinguish between them. Clinicians may use the test in their offi ces because it is waived from Clinical Laboratory Improvements Amendment requirements based on documentation that test results by persons without formal laboratory training are in concordance with results by trained laboratorians.

The Study
The California Department of Public Health (CDPH) supplied QuickVue Infl uenza test kits to clinicians participating in the Centers for Disease Control and Prevention (CDC) Sentinel Provider Infl uenza Surveillance Program. Sentinel providers performed the QuickVue Infl uenza test on a fi rst respiratory specimen obtained from outpatients with infl uenza-like illness (fever >100°F and cough and/or sore throat) using the foam swab provided by QuickVue. Clinicians collected a second respiratory specimen using a sterile Dacron swab that was stored in viral transport media at 4°C for <72 hours before shipment to CDPH. Sentinel providers recorded information about patient demographics, symptoms, and QuickVue test results on a standardized specimen collection form.
At CDPH, specimens were tested by an infl uenza A universal real-time reverse transcription-PCR (rRT-PCR) assay with an analytical sensitivity (50% tissue culture infective dose /PCR input) of 0.51 for infl uenza A (4). If infl uenza A virus nucleic acid was detected, subtyping for human infl uenza A (H1 and H3) was performed. Specimens negative for any subtype were tested for pandemic (H1N1) 2009 by using a rRT-PCR detection panel provided by CDC. For all PCR testing, a cycle threshold (Ct, the cycle count at which amplifi ed product yielded a detectable fl uorescent signal) <40 was interpreted as positive. Sensitivity, specifi city, predictive values, likelihood ratios, and posttest probabilities were estimated according to standard defi nitions (5). This activity was reviewed by the California Committee for the Protection of Human Subjects and determined to be a public health response that did not require institutional review board approval.
The median age of patients with infl uenza-like illness was 19 years (range 0-80 years). The median time from illness onset to specimen collection was 2 days (range 0-20 days 57% to 84% and a negative test result decreasing it to 36%. The sensitivity, specifi city, PPV, and NPV of the rapid test compared to PCR for persons <18 years of age were 68%, 80%, 87%, and 56%, respectively, and for persons >18 years were 64%, 86%, 82%, and 69%, respectively. Ct values were available for 389 specimens in which pandemic (H1N1) 2009 virus was detected by PCR; of these, the median infl uenza A PCR Ct value was 26 for 135 specimens with a negative rapid test result and 21 for 254 specimens with a positive rapid test result (p<0.0001); samples with higher viral loads were more likely to be positive by rapid test (Figure). Even so, ≈25% of PCR-positive, rapid test-negative specimens had Ct values <23.
Other smaller studies have found comparable sensitivities, but higher specifi cities, for rapid antigen tests for pandemic (H1N1) 2009. In a CDC study of 45 samples provided by state laboratories, the sensitivity of all rapid tests was 40%-69%, including 69% for QuickVue Infl uenza A+B (6). Others have found the QuickVue rapid tests to have sensitivities of 51%-63% and specifi cities of 99%-100% (7)(8)(9). During a large cluster of school outbreaks in New York, NY, USA, the sensitivity and specifi city of the Binax NOW (Inverness Medical International, Bedford, UK) rapid test were 17.8% and 93.6%, respectively (10). As we found, positive rapid antigen test results in other studies also appear to correlate with higher concentrations of pandemic (H1N1) 2009 virus (6,11,12).

Conclusions
Our fi ndings illustrate the challenges clinicians face during the current pandemic. Because clinical symptoms of pandemic (H1N1) 2009 are nonspecifi c, defi nitive diagnosis requires confi rmatory PCR testing, which, when available, often requires several days between specimen collection and reporting of results. Rapid antigen tests are the only current option for screening and diagnosis at the point of care. Current CDC guidelines recommend that highrisk and hospitalized infected patients be treated promptly with antiviral drugs and managed by using specifi c infection control precautions (13). Given the frequency of error found in this study, pandemic (H1N1) 2009 cannot be excluded solely because of a negative rapid antigen test result. Likewise, false-positive results, which would be expected to increase when the prevalence of infl uenza as a cause of infl uenza-like illness decreases, may result in unwarranted treatment and infection control measures that can be labor and resource intensive. Although rapid antigen tests are reported to have high specifi city for seasonal infl uenza, our fi ndings confl ict with previous assumptions that rapid antigen tests are suffi ciently specifi c to guide decisions about withholding antiviral treatment or chemoprophylaxis for pandemic (H1N1) 2009 (2).
A difference in swab types between rapid and PCR testing might have affected sensitivity of the rapid test re-   sults. Likewise, although infl uenza B virus was detected in only 9 (0.09%) of 10,367 specimens during the 7.5 months of statewide surveillance, some rapid test results may have been interpreted as falsely positive due to infection with infl uenza B.
In conclusion, we found the QuickVue infl uenza test had suboptimal sensitivity and specifi city for the detection of pandemic (H1N1) 2009 during a period of increased prevalence in California. This fi nding suggests that rapid test results that may lead to changes in clinical management or public health intervention should be confi rmed with PCR. A strength of our study is its refl ection of typical testing practices in outpatient settings and the need for reconsideration of the clinical application of rapid test results. The development of more accurate point-of-care tests for seasonal and pandemic (H1N1) 2009 infection is urgently needed.