Ninety-two percent of the 361 specimens tested were nasopharyngeal aspirates, and 8% were nasopharyngeal swabs. Of the 38 nasopharyngeal swabs, 5 were positive for RSV and another 10 were positive for influenza virus. The Proflu-1 real-time RT-PCR assay had a specificity of 100% and sensitivities of 94.7% (95% confidence interval [CI], 84.5 to 98.6) and 98.2% (95% CI, 93.0 to 99.7) for the detection of influenza virus and RSV, respectively (Table ). The accuracy of these results agrees with a previous study of the Proflu-1 assay by LeGoff et al. of a severely diseased pediatric population (28
). For our 54 RSV-positive specimens tested by all three methods (119 overall), the RSV positivity rates were 94.7% for PCR, 81.7% for antigen immunoassay, and 56.9% for viral isolation. Similarly, for the 32 influenza virus-positive specimens tested by all three methods (59 overall), the detection rates were 94.7% for PCR, 58.8% for antigen immunoassay, and 53.5% for viral isolation. Other viruses that were isolated in cell culture from individual specimens included 3 isolates of adenovirus, 19 isolates of rhinovirus-like virus, 17 isolates of parainfluenza virus type 3, and 1 isolate each of parainfluenza virus types 1 and 2. One specimen that was positive for RSV by antigen immunoassay and PCR testing was considered a false negative by viral isolation, but a rhinovirus-like virus also grew in cell culture. A second specimen was identified by cell culture as a dual infection with RSV and rhinovirus-like virus. A dual infection with RSV and influenza A virus was detected in only one specimen.
Accuracy of the Prodesse Proflu-1, Binax NOW RSV, and BD Directigen Flu A+B assays and conventional virus culturea
Seventy-one percent of patients who were tested for RSV and 65.6% of those tested for influenza virus were less than 18 years old. Overall, the highest incidence of RSV positivity was found in pediatric patients, of whom 98 (82.4%) were between 1 month and 2 years old and of whom 10 (8.4%) were <1 month old. The mean/median patient ages associated with all of the specimens tested for RSV and the ages of the patients who tested positive for RSV were 19.7 years/1.5 years and 38.1 months/6.3 months, respectively. In contrast, 38 (64.4%) of the influenza virus-positive samples were from individuals >18 years old, with 20 of those being >65 years old. Overall, with the Directigen A+B assay, the mean/median patient ages associated with the specimens tested and the ages of those patients who tested positive for influenza virus were 22.9 years/1.8 years and 42.0 years/40.5 years, respectively. As a consequence, since viral shedding is generally higher in children, the accuracy of RSV testing in this study should not be generalized to adults and, conversely, the results of influenza virus testing should be generalized to a pediatric population with caution.
An additional 12 influenza virus (22.2% of the total PCR positives)- and 8 RSV (7.3% of total PCR positives)-positive results with the Proflu-1 assay were reclassified from potential false positive to true positive after agreement with either the Cepheid influenza virus A/B or the Cepheid RSV PCR assay. There was highly significant agreement for the correlation of PCR CT measurements between sequential PCR tests for influenza virus (r2 = 0.80, P < 0.0001) and RSV (r2 = 0.84, P < 0.0001) with the Proflu-1 and Cepheid analyte-specific reagent-derived assays. These CT values ranged from 26.47 to 35.98 for RSV and from 20.38 to 35.95 for influenza virus. This correlation of sequential PCR CT values supports the robustness of the determined PCR specificity since CT values are inversely proportional to the amount of target nucleic acid detected. This type of supporting evidence for the use of PCR for influenza virus and RSV testing has not been previously demonstrated. For these positive results obtained only by sequential PCR testing, the CT levels were indicative of an abundant nucleic acid target, defined here as a CT of <29, in 58% and 38% of the nucleic acid extractions for influenza virus and RSV, respectively.
The sensitivity of the Proflu-1 RT-PCR was substantially higher in comparison to viral culture or the Binax NOW RSV and BD Directigen A+B antigen immunoassays (Table ). The improved sensitivity of PCR over antigen testing and virus isolation for RSV (14
) and influenza virus (17
) has been previously demonstrated by uniplex, as well as multiplex, testing (4
). Studies have also demonstrated improved sensitivity of PCR in combination with culture as a composite “gold standard” or as the alternate reference test for RSV or influenza virus (1
). A very small minority of previous studies evaluating PCR detection of influenza virus and RSV have included controls to monitor inhibition due to inadequate extraction and purification of nucleic acids (16
). Estimates of sensitivity are incomplete or compromised without the inclusion of an internal control to monitor these potential false-negative results. The extraction of nucleic acids did not include an initial centrifugation of the respiratory specimens. Less inhibition was seen when respiratory specimens were first centrifuged to remove the inhibitors in cellular debris. However, this cellular material contains respiratory virus and the centrifugation step was found to decrease detection sensitivity by removing it (data not shown). In this study, 13% of the specimens tested by PCR had a failed internal control upon initial testing, which was reduced to 9% on repeat testing of the extracts after a freeze-thaw cycle and reduced further to 4% after repetition of the nucleic acid extraction and purification steps. In this study, 25% of the specimens that were initially negative for RSV or influenza virus by PCR in combination with a failed internal control were subsequently determined to be positive on repeat PCR testing with or without repeated extraction.
As might be expected for a test with higher sensitivity, the mean PCR CT
values were lower for the specimens that also tested positive for RSV or influenza virus by antigen testing or cell culture than for those specimens that tested negative (Fig. ). Significant differences were observed between the mean PCR CT
values of specimens antigen positive and negative by the NOW RSV (P
< 0.0001) and Directigen A+B (P
= 0.0235) assays. The simplest explanation for these results is found in the larger number of pediatric patients who tested positive for RSV than for influenza virus and the well-described association of high viral shedding in pediatric patients with RSV infections (13
FIG. 1. Comparison of PCR CT values obtained with the Proflu-1 multiplex assay for specimens that tested positive or negative for RSV (A) or influenza virus (B) by virus isolation or antigen immunoassay. The differences between the mean PCR CT values of antigen (more ...)
The performance of both the NOW RSV and Directigen A+B assays is at least equal to that of other commercially available antigen immunoassays (2
). In this study, the NOW RSV assay had a sensitivity of 81.7% (95% CI, 73.2 to 88.1), which agrees well with the results obtained previously by other laboratories when testing nasopharyngeal aspirates from pediatric patients (Table ). The Directigen A+B assay had an overall sensitivity of 58.8% (95% CI, 44.2 to 72.1) for the detection of influenza virus. Separate sensitivities of 61.4% (95% CI, 45.5 to 75.3) for the detection of influenza A virus and 42.9% (95% CI, 11.8 to 79.8) for the detection of influenza B virus were obtained, although the number of influenza B virus-positive specimens tested was very small (Table ). While these results compare well with recent studies (7
), there nevertheless exists substantial variability in the reported sensitivity of the Directigen A+B assay for the detection of influenza virus (Table ). Differences due to both the specimen type tested (24
) and patient age (8
) have been postulated to affect the sensitivity reported for influenza virus antigen immunoassays with the Directigen A+B and other, similar, commercial products. Neither the association nor a definitive explanation for the observed decrease in antigen immunoassay sensitivity for influenza virus detection in adult versus pediatric patients has been conclusively demonstrated to date. Young children have been reported to have higher attack rates and more prolonged viral shedding (19
). Similarly, it has been empirically shown that the sensitivity of influenza virus antigen immunoassays is highest in patients ≤5 years old (38
). Patterns of virus shedding and differences in the quality of specimens between age groups are plausible explanations (41
). This would help explain the low Directigen A+B assay sensitivity reported here, since despite the testing of samples from patients with a wide age range, the mean age of the influenza virus-positive patients was 42 years. Nevertheless, it is important to recognize that greater than 90% of the deaths due to influenza and its complications occur among elderly persons and nursing home residents are at higher risk of serious influenza-related complications than are elderly persons living in the community (18
). Oseltamivir postexposure prophylaxis during nursing home outbreaks is used to reduce serious complications and death but must commence within 48 h of the onset of symptoms (18
). In this study, PCR was the only method which provided results both rapid and accurate enough for the effective initiation of antiviral therapy.
Comparison of reports of the accuracy of the Binax NOW RSV antigen immunoassay
Comparison of reports of the accuracy the BD Directigen A+B antigen immunoassay
The Binax NOW RSV and Directigen Flu A+B antigen immunoassays had sensitivities that were 25% and 5% higher than that of culture, respectively (Table ). The thermolability of RSV is well described, and samples must be kept cold during transport without freeze-thawing and be inoculated onto a cell culture as quickly as possible (2
). This fact may account for the lower sensitivity of RSV isolation in this study since the inoculation of our specimens onto a cell culture was delayed. The difference between the mean PCR CT
values of culture-positive and -negative specimens was not statistically significant for RSV or influenza virus. When this observation is considered together with the similar wide distribution of the interquartile range of PCR CT
values (Fig. ), it suggests that viral nucleic acid was present in a range of quantities in both culture-positive and culture-negative specimens. The simplest explanation for this observation is that the loss of virus culturability occurred with minimal RNA and antigen degradation and that this was an important contributing cause of the lower sensitivities observed with culture. A range of different sensitivities for virus isolation have been reported in other studies. Importantly, a small number of studies have described the immediate culture of pediatric patient specimens as having a very high sensitivity for RSV and influenza virus detection when cell culture was performed on site (20
). In addition to the importance of immediate virus isolation for accurate cell culture results, the methodology is also heterogeneous and the sensitivity for the detection of these viruses differs when different cell culture methods are used (12
Isolation of RSV and influenza virus with R-mix cells can provide positive culture results in as little as 1 to 2 days, whereas conventional tube culture methods typically provide a 5- to 6-day turnaround time for positive specimens (12
). The average turnaround times in this study for specimens positive by tube cell culture for RSV and influenza virus were 10.6 and 8.8 days, respectively. This prolonged turnaround time for viral isolation reduced its usefulness in patient management. Moreover, in this study, the sensitivity of viral isolation performed off site was shown to be 53.5% (95% CI, 37.8 to 68.5) and 56.9% (95% CI,44.1 to 68.9) for influenza virus and RSV, respectively, greatly diminishing the use of viral culture as a confirmatory method. In contrast, with real-time RT-PCR, eight specimens required an average of 45 min for the extraction and purification of nucleic acids by a semiautomated method, which was followed by 45 min for the preparation of the Proflu-1 assay and 97 min of cycling run time (a total of 3 h).
The definitive diagnosis of RSV and influenza virus infections depends on the microbiology laboratory. A rapid diagnosis of infections with these viruses is required to implement effective infection control measures to limit nosocomial transmission but is also associated with a reduced length of hospitalization and other hospital-related costs (3
). In the case of influenza virus testing in particular, the annual impact of this virus is expected to intensify since people aged ≥80 years are the fastest growing segment of the U.S. population (18
). PCR testing represents an alternative to the unacceptably low sensitivity of rapid antigen immunoassays for influenza virus in this high-risk group of people. In clinical practice, antigen immunoassays and direct immunofluorescence antigen (DFA) testing with fluorescent antibody have often been relied upon to make a rapid diagnosis of influenza virus and RSV infections. DFA testing with fluorescent antibody remains an excellent screening test, although the technical time required to obtain results and the subjective nature of the results can be serious limitations. Rapid antigen testing by immunoassay methods remains a very attractive option for laboratories because they are convenient and rapid and possess a high positive predictive value. Although rapid results can be obtained in 30 min by this method, the sensitivity can be low and negative results require confirmation by more sensitive testing. Viral isolation by cell culture has traditionally been the method used for this purpose; however, these results can be delayed, thereby negating the potential impact of confirmatory testing on patient care. In a clinical setting where the sensitivity of viral isolation is less than optimal, real-time reverse transcriptase PCR testing is a more accurate and timely confirmatory test for influenza virus and RSV antigen testing.