Accuracy of bedside antigen tests in the diagnosis of new influenza A/H1N1v infection

Authors


Corresponding author and reprint requests: Guenter Weiss, MD, Department of Internal Medicine I, Clinical Immunology and Infectious Diseases, Medical University; Anichstrasse 35, A-6020 Innsbruck, Austria
E-mail: guenter.weiss@i-med.ac.at

Abstract

Clin Microbiol Infect 2011; 17: 235–237

Abstract

To evaluate the clinical reliability of two rapid influenza detection tests (RIDTs), we analyzed 107 specimens from patients with clinically suspected pandemic influenza A/H1N1v by these tests as well as by real-time PCR as a standard. Both RIDTs had a moderate sensitivity (28–32%), a high specificity (93–99%) and a negative predictive value of 80%. These results will impact on the clinical management and isolation precautions in patients with suspected infection. Although a positive RITD is mostly confirmatory, a negative result in the presence of high clinical likelihood of infection should be interpreted with caution and be re-evaluated by PCR.

During the outbreak of the pandemic influenza A (H1N1v) in 2009, tests for rapid detection of influenza antigens (RIDT) were widely recommended and widely used as bedside screening for influenza in suspected patients. These tests had a high predicted sensitivity, which made them attractive as a clinical screening test [1,2]. However, the parallel testing of such patients by means of real-time PCR from the same nasal or pharyngeal swabs used for RIDT indicated that the sensitivity of the RIDT might be lower. This is a serious concern because the rapid diagnosis of infection with H1N1v is important for preventing the further spread of the disease. Because the symptoms of the infection with the pandemic influenza virus are variable [3,4], the availability of reliable rapid bedside tests is relevant for guiding the clinical management of patients in terms of transmission of the infection. Thus, as an initial assessment, several test methods were investigated and compared for their ability to detect H1N1 viral antigens in respiratory clinical specimens by the CDC [5–8]. The results obtained indicated that the RIDTs were capable of detecting novel influenza A (H1N1v) virus from respiratory specimens containing high levels of virus with a varying sensitivity (40–69%) [5]. With the epidemiologic spread of novel influenza, confirmatory tests for the diagnosis of H1N1v infection by real-time PCR were introduced [9,10]. We thus studied the accuracy of two available RIDTs in comparison with real-time PCR for H1N1v.

One hundred and seven specimens of nasal or pharyngeal swabs were obtained from patients with influenza-like symptoms presenting at the Department of Internal Medicine, University Hospital of Innsbruck, Austria. Swabs were eluted in 500 μL of transport medium: 100 μL of the elution were used for the rapid antigen tests and the remaining 300 μL were used for performance of real-time PCR for H1N1v. Of these specimens 107 were tested with BinaxNow Influenza A&B Rapid Test (Inverness Medical, Cologne, Germany) [8,11–13] and 96 in parallel with BD Directigen EZ Flu A&B test (Becton& Dickinson, Sparks, MD, USA) [8] in accordance with the manufacturer’s instructions. Results were categorized as either negative, weakly positive or positive. For calculation, weak bands in the test field were defined as positive. In five patients, swabs were taken on two consecutive days because of a strong suspicion of influenza despite an initial negative rapid antigen test result.

TaqMan real-time PCR targeting the haemagglutinin (HA) genes of porcine influenza A/H1N1v virus was established and performed as described previously [14]. The primers and probe used were: FluSw H1 F236 (TgggAAATCCAgAgTgTgAATCACT) as sense primer, FluSwH1 R318 (CgTTCCATTgTCTgAACTAgRTgTT) as antisense primer and Flu Sw H1 TM292 (FAM – CCACAATgTAggACCATgAgCTTgCTgT- BBQ) as probe.

The diagnostic potential of the antigen tests was compared with the results of the real-time PCR, which was set as the standard. Statistics were calculated with spss statistical software, version 11.5, for Windows (SPSS Inc., Chicago, IL, USA).

Among the 107 clinical specimens, 29 were tested positive for influenza A/H1N1v by real-time PCR. Using BinaxNOW Influenza A&B rapid antigen test, nine of the 107 tested specimens were also found positive, eight of which were confirmed by real-time PCR. Accordingly, in comparison with the results of real-time PCR, one specimen in the rapid test was classified as false positive, whereas 21 were found to be false negative.

With the BD Directigen EZ Flu A&B rapid antigen test, 13 specimens out of 96 gave positive results and, according to the results of the real-time PCR, eight were classified as true positives, whereas five were false positives, and, in addition, 17 rapid antigen tests provided false negative results. Thus, both RIDTs had a low sensitivity (32% for BD Directigen EZ Flu A&B vs. 27.6% for BinaxNow Influenza A&B). However, both tests revealed a high specificity (98.7% for BinaxNow Influenza A&B and 93% for BD Directigen EZ Flu A&B), which is in line with the findings of the CDC. Subjects with a positive result in the BD Directigen EZ Flu A&B were more likely to lack confirmation of infection than with the BinaxNow Influenza A&B assay (positive predictive value 88.9% for BinaxNOW Influenza A&B vs. 61.5% for BD Directigen EZ Flu A&B). Both assays displayed similar negative predictive values (78.6% for BinaxNOW Influenza A&B and 79.5% for BD Directigen EZ Flu A&B). Disease prevalence for pandemic influenza was 26% for all samples performed with the BD Directigen EZ Flu A&B and 27% for samples evaluated with BinaxNOW Influenza A&B.

Because there was no background of seasonal influenza A or B at the time of the present study, cross reactions of RIDTs with ‘seasonal’ influenza are unlikely to explain the low rate of false positive results (false positive rate 1% for BinaxNOW Influenza A&B and 5% for BD Directigen EZ Flu A&B). In five patients, a second clinical specimen was obtained, 1 or 2 days after the first test because of primary negative testing in the presence of a high suspicion of H1N1v infection. Two of the consecutive tests were classified as positive by the BinaxNow Influenza A&B assay (one false and one true positive according to real-time PCR), whereas three clinical swab eluents turned positive with the BD Directigen EZ Flu A&B test (one true positive and two false positives).

Our investigations of rapid antigen influenza tests during pandemic influenza A/H1N1v confirm the high specificity of each RIDT but a significant lack of sensitivity [15,16]. However, the results obtained indicate a better sensitivity for the BinaxNow Influenza A&B antigen test than that recently described by Drexler et al. [17] who calculated a sensitivity of 11.1% during the 2009 pandemic season. This low sensitivity refers to a high percentage of children among the investigated patients and a lack of test sensitivity for influenza A/H1N1v [17]. However, in the present study, we only investigated adults and the diagnostic workup for antigen testing and PCR were performed using the same specimen, ruling out that any differences in obtaining nasal or pharyngeal swabs are responsible for the low sensitivity of the rapid test observed. Accordingly, a definitive diagnosis of infection with influenza A/H1N1 virus requires real-time PCR analysis [10] in patients who test negative with RIDT, whereas a positive RIDT is confirmatory for infection with influenza A/H1N1v, and does not require confirmation by real-time PCR analysis. Rapid antigen tests have decreased sensitivity for detecting pandemic influenza A/H1N1v compared to seasonal influenza A subtypes [18]. However, many primary health centres do not have access to real-time PCR to allow the re-evaluation of test results in patients with a high suspicion of infection but a negative RIDT. Thus, point of care strategies that allow practitioners rapid access to laboratories providing continuous reception/processing of samples and a reduction in the time needed to provide consolidated results have both proven helpful for the clinical management of pandemic influenza [2]. However, it appears that the sensitivity of the tests is lower than expected, which could be a reflection of low viral load in an early stage of the infection, comprising an argument in favour of repeated testing.

Moreover, enrichment of the viral concentration in the specimen by reducing the volume of the elution medium may improve the sensitivity of RIDTs. Improvements of the test and/or the specimen sampling procedure are mandatory for increasing the sensitivity of rapid tests that are clinically useful during the onset of a new pandemic wave of influenza infection. The results of the present study will impact on the clinical management and isolation precautions in patients with suspected pandemic influenza infection. On the basis of our observations, patients with a high clinical suspicion for infection with influenza A/H1N1v but a negative rapid antigen test should be considered positive until the results of the real-time PCR become available, whereas a positive result does not require confirmation by real-time PCR. An accurate and timely diagnosis can provide information that is useful for the appropriate treatment and care of the patient, as well as for the implementation of prevention measures against the spread of influenza to other patients [19].

Transparency Declaration

None of the authors has any financial disclosure or conflicts of interest in relation to this study.

Ancillary