TEST CHARACTERISTICS OF COMMERCIAL INFLUENZA ASSAYS FOR DETECTING PANDEMIC INFLUENZA A (H1N1) IN CHILDREN

Abstract

We assessed the test characteristics of 2 influenza antigen tests, a rapid immunoassay and a direct fluorescence antibody (DFA) assay, in detecting pandemic influenza A (H1N1) in children up to 18 years of age, using polymerase chain reaction as the standard. The sensitivities of BinaxNOW (59.6%) and direct fluorescence antibody (57.3%) were similar. The specificity of each test was >99%.

In April 2009, a novel strain of swine-origin influenza A (H1N1), recently designated by the World Health Organization as “pandemic (H1N1) 2009 virus,” was detected for the first time in the United States.¹ During the succeeding months, this strain infected individuals around the globe, and on June 11, 2009, the World Health Organization raised the level of pandemic influenza alert to phase 6, indicating that a global pandemic was underway.² A diagnostic polymerase chain reaction (PCR) test was quickly developed and disseminated by the Centers for Disease Control and Prevention (CDC) to public health agencies for use in testing clinical samples.³ Providers in other healthcare settings (such as office practices and hospitals) have traditional influenza diagnostic tests available for use, but the test characteristics of these commercial diagnostic tests in detecting pandemic (H1N1) 2009 virus are unknown. Our objective was to determine the test characteristics of a rapid test, the BinaxNOW Influenza A&B (BinaxNOW), and of a direct fluorescence antibody assay (DFA) in detecting pandemic (H1N1) 2009 virus in children.

MATERIALS AND METHODS

All patients up to 18 years of age who were tested for influenza A at Children's Hospital Boston between April 29 and June 12, 2009 were included. After June 12, the Hinton State Laboratory Institute discontinued routine testing for pandemic (H1N1) 2009 virus and limited testing to selected patients preapproved by public health epidemiologists and sentinel surveillance sites.

Nasopharyngeal specimens were collected using Flocked Swabs (Copan, Murrieta, CA) and placed into Universal Transport Media or M4-RT (Remel, Lenexa, KS) before transport. Two swabs were collected from each patient and placed in separate tubes of transport media. One specimen was tested with the BinaxNOW (Inverness Medical, Waltham, MA) or Simulfluor Flu A/B DFA (Millipore, Billerica, MA) or both. Assays were performed following the manufacturer's instructions and laboratory validated procedures. If supplies were available, both BinaxNOW and DFA were done, but some specimens were tested by only 1 method if supplies were not available. The other specimen was tested at the Hinton State Laboratory Institute for pandemic (H1N1) 2009 virus, using real-time reverse-transcriptase polymerase chain reaction (RT-PCR) reagents provided by the Influenza Branch of the CDC.³ Validation of the assay, staff training, and assay proficiency were required before adopting the emergency use authorization for this new test.³

The sensitivities, specificities, positive predictive values, and negative predictive values (NPV) of BinaxNOW and of DFA were calculated using real-time RT-PCR for pandemic (H1N1) 2009 virus as the reference (standard) test. If an inconclusive result was obtained by RT-PCR, the performance of BinaxNOW influenza A&B and DFA was not determined. Test performance was determined for the age groups indicated in Table 1, and for all patients up to and including 18 years of age; 95% confidence intervals were calculated for proportions, using Stata 10.0 (StataCorp, College Station, TX). The Children's Hospital Boston Committee on Clinical Investigation approved this study.

TABLE 1: Test Characteristics for Detection of Pandemic (H1N1) 2009 Virus

RESULTS

A total of 614 specimens were tested for pandemic (H1N1) 2009 virus by RT-PCR. Specimens were submitted from the Emergency Department (483 specimens, 78.7%), outpatient locations (69 specimens, 11.2%), and inpatient locations (62 specimens, 10.1%). RT-PCR results for pandemic (H1N1) 2009 virus were positive for 230 (37.5%) specimens, negative for 380 (61.9%) specimens, and inconclusive for 4 specimens (0.7%). BinaxNOW was used to test 542 specimens, DFA was used to test 534 specimens, and 476 specimens were tested using both assays. Inconclusive results were obtained with 1 specimen tested by BinaxNOW influenza, and 5 tested by DFA.

Table 1 displays the test characteristics for BinaxNOW and DFA for the detection of pandemic (H1N1) 2009 virus, both overall and stratified by age. The sensitivities, specificities, positive predictive values, and NPV of the BinaxNOW and DFA for detecting pandemic (H1N1) 2009 virus were not substantially different. The NPV of each test was significantly lower in patients >5 to 10 years of age than the total NPV for that test. There were no other differences in the performance characteristics within either test for the different age groups.

DISCUSSION

In the United States, infection with novel influenza A of swine origin (H1N1) has occurred predominantly in children.⁴ In Massachusetts, 64% of the confirmed cases were in individuals ≤18 years of age (with a median age of 14 years) for specimens tested from April through June 2009. Our study is the first to assess the performance of BinaxNOW and DFA tests for pandemic (H1N1) 2009 virus in children as compared with the standard of RT-PCR. We found that the BinaxNOW rapid test was 59.6% sensitive in detecting pandemic (H1N1) 2009 virus, which is similar to the sensitivity of DFA (57.3%). Both tests had high specificities and positive predictive values during our study period, which did not include the typical influenza season. The BinaxNOW and DFA had NPV for pandemic (H1N1) 2009 virus of 79.8% and 84.7%, respectively.

A few published studies have assessed the performance of commercially available influenza diagnostic tests in detecting pandemic (H1N1) 2009 virus in adults. In a study involving patients ranging in age from 4 days to 98 years, the sensitivities of rapid antigen tests (BinaxNOW and 3M A + B combined) and DFA were 21.2% and 47.2%, respectively for pandemic (H1N1) 2009 virus.⁵ However, the standard used in that study was not PCR for all samples but rather a composite of assays, and the test characteristics within various pediatric age ranges were not reported. The Naval Health Research Center found that among 39 adults with pandemic (H1N1) 2009 virus confirmed by PCR, the sensitivity of the QuickVue Influenza A + B (Quidel) rapid test was 51%, with a specificity of 99%.⁶ Previous literature has shown that rapid immunoassays for respiratory viral infections have higher sensitivity among children than adults.⁷

It is critical to understand the performance of commercial influenza tests used for pandemic (H1N1) 2009 virus in pediatric age groups, particularly since clinical decisions about infection control measures, treatment, and prophylaxis need to be made before PCR results are often available. Many pediatric providers also may not have PCR testing readily available in their offices, where most children are evaluated. A negative result from commercial antigen assays should not be used as a reason to discontinue isolation precautions for hospitalized patients. Infection control decisions are further complicated by the fact that the CDC recommends the use of fit-tested N95 respirators for routine care of patients with pandemic (H1N1) 2009 virus, whereas droplet precautions are considered appropriate for seasonal influenza.^8,9 Tests that determine the subtype of influenza A virus would be required to determine the correct isolation precautions under current recommendations.

Understanding the ability of commercial tests to detect pandemic (H1N1) 2009 virus can also provide clinicians with valuable information for decisions about antiviral usage, particularly since treatment should be initiated within 48 hours of onset of symptoms and offering prophylaxis may help to prevent the spread of infection to vulnerable individuals. Positive results for influenza A from antigen assays are likely correct but do not distinguish between different influenza A subtypes. The various circulating subtypes of influenza A virus have differing antiviral resistance patterns. Decisions regarding antiviral use for treatment or prophylaxis when antigen tests are positive must be made with knowledge of currently circulating subtypes and their associated antiviral susceptibility patterns. When antigen test results are negative, infection with pandemic (H1N1) 2009 virus cannot be excluded and decisions about treatment and prophylaxis must be based on clinical assessment and the potential benefits or costs of antiviral administration.

There are several limitations to our data. First, we assessed only 1 brand of rapid influenza test and 1 DFA test, preventing us from extrapolating the results to other commercial tests, and we collected data from a single institution. Second, the positive and negative predictive values of any test are dependent on the prevalence of disease; the values would be different when seasonal influenza and/or pandemic (H1N1) 2009 virus are circulating at different levels. Finally, it is assumed that the CDC PCR platform is the standard for detection of pandemic (H1N1) 2009 virus, and therefore we did not calculate test performance in reference to any alternative standard; it is possible that other PCR platforms could have different sensitivities.

Because the test characteristics of BinaxNOW and DFA are similar, it is reasonable for providers to use either of these commercial tests in attempting to diagnose pandemic (H1N1) 2009 virus, and the choice may be influenced by other factors (such as test turnaround time, cost, and available resources). Clinicians must remember that a negative result with these tests is not sufficient to rule out infection with pandemic (H1N1) 2009 virus.