Influenza causes a broad range of illnesses, and its diagnosis based solely on clinical symptoms is difficult [1]. There are several diagnostic methods currently available including viral culture, molecular assays, serologic tests, and antigen detection tests. The sensitivity and specificity of the assays used for the detection of influenza virus vary depending on the type of method used [2, 3]. Molecular methods including rapid molecular assay and reverse transcription polymerase chain reaction (RT-PCR) are widely used as the standard methods for diagnosing influenza in clinical labo ratories. Molecular assays have high sensitivity and specificity for the detection of influenza virus; however, usually, appropriate laboratory settings and well-trained technicians are required to perform these tests. In contrast, rapid influenza diagnostic tests (RIDTs) are easier to handle and can detect influenza viral anti-gens in 10-15 minutes, making them valuable in the outpatient setting. Accordingly, RIDTs are widely used for the diagnosis of influenza as an initial step in clinical fields [2, 4]. Since results from RIDT methods are available before those from PCR or immunoassays, clinicians can make decisions on whether to start antiviral treatment based on these results [4]. However, the sensitivity of RIDTs is highly variable—ranging from 40-98% [2, 5, 6]—which can frequently lead to false negative results, thereby making it important to recognize the limitation of these methods while interpreting the results. In this study, we evaluated the diagnostic performance of a new RIDT, SD Standard F influenza A/B FIA (Standard F; SD Biosensor, Inc., Suwon, Korea), by comparing its performance with that of the commercially available RIDT, BD Veritor Flu A+B (Veritor; Becton Dickinson and Company Diagnostic, Sparks, MD, USA), relative to RT-PCR.

Both RIDTs involve immunochromatographic detection of nucleoprotein antigens of influenza A or B. The influenza viral antigens bind to the antiinfluenza antibodies conjugated to detector particles in the test strip [7]. The difference between the two RIDTs is that the Veritor uses colloidal-gold detector particles, whereas the Standard F uses fluorescent dyes as the detector. The results of both the assays are interpreted by a digital reading instrument. In case of the Veritor, only qualitative results such as positive or negative are displayed, whereas in Standard F, the qualitative results along with the fluorescence intensities as cutoff index (COI) values are displayed. According to the Standard F manufacturer’s instructions, test results of COI ≥1.00 are presented as positive and below 1.00 as negative for influenza.

We used 117 left-over, non-duplicated, nasopharyngeal samples from patients who visited the Severance hospital with a flu-like illness (fever >37.8°C and/or clinical symptoms of headache, cough, sore throat, and myalgia) from January 2018 to April 2018; the samples were from 12 adults and 105 children ranging from less than one to 17 years of age (median age, 2 years old). All the samples were analyzed by the AdvanSure RV real-time RT-PCR assay (rRT-PCR; LG Life Sciences, Seoul, Korea) and immediately preserved at -70°C after rRT-PCR. The thawing process was perfomed at room temperature (15-30°C) before analyzing the samples using the two RIDTs. The two RIDTs were performed simultaneously and all the results obtained were compared with those from the rRT-PCR analysis. We interpreted samples with a cycle threshold (Ct) value <25.00 as a positive result according to the manufacturer’s instructions. This study has been approved by the Institutional Review Board (IRB) of Yonsei University College of Medicine (IRB 1-2017-0093).

Sensitivity, specificity, positive predictive value, and negative predictive values with 95% confidence intervals (CIs) were calculated using standard formulas. Statistical analysis was performed using Microsoft Excel (Microsoft Corporation, WA, USA) and Analyse-it software, Method Evaluation edition version 5.11 (Analyze-it Software LTD, City West Business Park, Leeds, UK). Pearson correlation coefficient was calculated to evaluate the correlation between the Ct values from rRT-PCR and COI values from the Stan-dard F. P values <0.05 were considered to indicate statistical sig-nificance.

Sensitivity and specificity of the two RIDTs for the detection of influenza are shown in Table 1. After rRT-PCR analysis for influ-enza A and B (A/B), 31 of the 117 specimens were found to be A positive/B negative, 17 were A negative/B positive, only one was A positive/B positive, and the rest 68 were A negative/B negative. Using rRT-PCR as the testing standard, the Veritor and Standard F assays were found to demonstrate a sensitivity of 65.6% (21/32) and 71.9% (23/32), respectively, for the detection of influenza A with a detection specificity of 100.0% (68/68). Additionally, both the assays demonstrated a sensitivity of 66.7% (12/18) and speci city of 100.0% (68/68) for the detection of influenza B. Fig. 1 shows comparison of Ct values of the influenza-positive specimens. A total of 49 samples were found to be positive by rRT-PCR, and almost all the samples with low Ct values, below 14.4 for influenza A and 13.3 for influenza B, were detected by both the assays.

In our study, Standard F showed a diagnostic performance com-parable with that of Veritor in detecting influenza A infection with a higher sensitivity. Two out of the 31 rRT-PCR influenza A positive samples (Ct value of 9.79 and 20.31) showed discordant results between the two RIDTs, positive by Standard F and negative by Veritor. Those cases were clinically diagnosed with influenza A infection and treated with medication, which was in line with the results obtained by the Standard F method. Our result is compara-ble with a previous report that reveals improved sensitivity and a lowered limit of detection (LOD) of fluorescent immunochromatography-based assays compared to that of colloidal gold-based rapid diagnostic kits [8]. However, the sensitivity of both the RIDT as-says was lower than the value suggested by the manufacturer. The differences in sensitivity could be due to the nature of the samples that we used. The study was conducted using samples that were preserved at -70°C after RT-PCR analysis. It is possible that degradation of the antigens during the thawing process could have resulted in the unexpectedly lower sensitivity values [9]. It is possible that the RIDT assays could not detect certain influenza-positive samples confirmed by PCR due to the presence of an altered epitope due to minor changes in the protein structure that was not detectable by the kit, and this could have led to differences between the sensitivity reported in our study and that suggested by the manufacturer [10].

The Standard F test displayed not only qualitative results but also quantitative COI values. The COI values displayed a moderate negative correlation with Ct values of the influenza-positive specimens (influenza A: R=-0.77, P <0.001; influenza B: R=-0.88, P <0.001; Fig. 2). Kim et al. [11] and Ryu et al. [12] studied dilution tests for fluorescent lateral flow immunoassays and described a correlation between the COI values and concentration of the target epitope such as rotavirus and antibody to hepatitis B surface antigen. Our results also demonstrate that the COI values might reflect the influenza viral load and can help estimate the relative amount of influenza antigen. Therefore, the COI values can provide additional information and guidance to a clinician analyzing the negative results. If the result of a symptomatic patient is negative with a COI value just below the cut-off level, a clinician can consider the possibility of early infection with low viral load; in such circumstances, a repeat test can be conducted within a short time-frame, or other viral assays such as RT-PCR can be given immediate consideration.

Consistent with previous studies, the two RIDTs showed a slightly higher overall sensitivity for influenza A (65.6%-71.9 %) than for in-fluenza B (66.7%) [5, 13-15] and the sensitivity and specificity of both the RIDTs were higher than the pooled values of sensitivity and specificity of 53.2%-62.3% and 98.2%-99.8%, respectively [2, 5]. An earlier study of the Veritor test revealed a sensitivity of 73.0%-93.8% for influenza A and 55.6%-94.2% for influenza B, and an overall specificity greater than 95% [2, 7, 14, 16-19]. These values are slightly higher than those reported in our current study. These wide ranges of sensitivity can be explained by factors including specimen type, assay kit lot, the phase of infection (related to viral load and shedding), and differences in the targeted study population; the sensitivity of the tests has been found to be higher for samples obtained from children [9, 18, 20].

Our study has some limitations. First, we conducted this study with a small number of samples. In addition, we did not use fresh samples, and this might have had a negative influence on the sensitivity of the assays.

In conclusion, the Standard F method shows a comparable re-sult for diagnosing influenza infection, is as sensitive as other RIDTs, and provides additional semi-quantitative information that can aid in diagnosis.