Low sensitivity of rapid diagnostic test for influenza.

Abstract

The QuickVue Influenza A+B Test (Quidel) was used to test nasal swab specimens obtained from persons with influenza-like illness in 3 different populations. Compared with reverse-transcriptase polymerase chain reaction, the test sensitivity was low for all populations (median, 27%; range, 19%-32%), whereas the specificity was high (median, 97%; range, 96%-99.6%).

Full text

Page 1

BRIEF REPORT • CID 2009:48 (1 May) • e89
B R I E F R E P O R T
Low Sensitivity of Rapid Diagnostic
Test for Influenza
Timothy M. Uyeki,1Ramakrishna Prasad,2Charles Vukotich,2
Samuel Stebbins,2Charles R. Rinaldo,2Yu-hui Ferng,3
Stephen S. Morse,3Elaine L. Larson,3Allison E. Aiello,4Brian Davis,4
and Arnold S. Monto4
1Influenza Division, Centers for Disease Control and Prevention, Atlanta,
Georgia;
New York; and
2University of Pittsburgh, Pittsburgh;
4University of Michigan, Ann Arbor
3Columbia University, New York,
The QuickVue Influenza A+B Test (Quidel) was used to test
nasal swab specimens obtained from persons with influenza-
like illness in 3 different populations. Compared with re-
verse-transcriptase polymerase chain reaction, the test sen-
sitivity was low for all populations (median, 27%; range,
19%–32%), whereas the specificity was high (median, 97%;
range, 96%–99.6%).
Rapid influenza diagnostic tests that detect influenza viral an-
tigens are used to screen patients with suspected influenza and
offer the advantage of providing a timely result, compared with
other tests for influenza, that can influence clinical decision
making [1]. Use of rapid influenza tests can help to reduce
unnecessary diagnostic testing, to facilitate antiviral treatment,
to decrease inappropriate antibiotic use, and to reduce the du-
ration of treatment in the emergency department or hospital-
ization [2–8]. Rapid influenza tests have also been used for
public health purposes to investigate suspected influenza out-
breaks. Identification of influenza virus infection by rapid tests
can facilitate prompt implementation of control measures be-
fore confirmatory test results are available from shell vial cul-
ture, tissue cell viral culture, or RT-PCR.
As diagnostic tests, the accuracy of rapid influenza tests is
less than that of “gold standard” reference tests, such as viral
culture or RT-PCR [1]. In particular, a wide range of sensitiv-
Received 30 October 2008; accepted 22 December 2008; electronically published 26 March
2009.
The views expressed herein are those of the authors and do not represent the policies of
Columbia University, University of Pittsburgh, University of Michigan, orthe CentersforDisease
Control and Prevention.
Reprints or correspondence: Dr. Timothy M. Uyeki, Epidemiology and Prevention Branch,
Influenza Div., MS A-20, Centers for Disease Control and Prevention, 1600 Clifton Rd. NE,
Atlanta, GA 30333 (tuyeki@cdc.gov).
Clinical Infectious Diseases 2009;48:e89–92
? 2009 by the Infectious Diseases Society of America. All rights reserved.
1058-4838/2009/4809-00E2$15.00
DOI: 10.1086/597828
ities of the rapid influenza tests have been reported, whereas
specificities have been reported to be high, compared with ref-
erence tests [1, 8–12]. In the context of enrolling participants
in studies to better understand transmission of influenza virus,
a rapid influenza diagnostic test was used to screen persons
with influenza-like illness at 3 separate study sites in the United
States during the 2007–2008 influenza seasons. Here, we report
our findings that compared the accuracy of the rapid test with
that of RT-PCR or viral culture and an investigation of these
results.
Methods.
During 2007–2008, studies at 3 different sites
were conducted to assess influenza transmission and non-
pharmaceutical interventions among different populations. In
each study, the QuickVue Influenza A+B Test (Quidel)wasused
to test respiratory specimens from ill participants. The study
target populations and the case definitions used at each site to
recruit study participants were as follows: site 1 (Ann Arbor,
MI), university students who presented to an outpatient clinic
with influenza-like illness “ILI1,” defined as cough plus at least
1 of the following: fever or feverishness, chills, or body aches;
site 2 (New York, NY), children and adults at Headstart pro-
grams, preschools, day care centers, pediatric clinics, and
Women, Infants, and Children program centers who were iden-
tified by study personnel as having influenza-like illness “ILI2,”
defined as temperature of ?37.8?C and cough and/or sore
throat in the absence of a known cause other than influenza
virus infection; and site 3 (Pittsburgh, PA), elementary school
students with ILI2 identified by study personnel at school visits.
At all sites, a foam swab supplied with the test kit was used to
collect a nasal swab sample from participants who met the case
definitions, and the specimen was tested immediately or within
a few hours by study personnel (sites 1 and 3) or was placed
in standard sterile viral transport media on ice and then re-
frigerated and tested within 4–8 h (site 2). Testing was done
using the QuickVue Influenza A+B Test in accordance with the
manufacturer’s instructions.
Additional respiratory specimens for confirmatory assays
were collected at the same time that nasal swabs were obtained
from participants. At site 1, a throat swab specimen was col-
lected from participants who met criteria for ILI1 by using a
sterile polyester swab and was placed in sterile, veal-infused
viral transport media on cold packs and then transported to a
laboratory and refrigerated for up to 72 h before testing for
influenza A and B viruses by real-time RT-PCR and by tissue
cell viral culture with standard laboratory methods (2007–
2008). At site 2, a deep nasal swab specimen was collected by

Page 2

e90 • CID 2009:48 (1 May) • BRIEF REPORT
Table 1.Results of diagnostic testing for influenza A and B viruses at 3 US sites, 2007–2008.
QuickVuea
results
Site 1
PCR results
(n p 303)
Site 1
culture results
(n p 303)
Site 2
PCR results
(n p 67)
Site 2
culture results
(n p 71)
Site 3
PCR results
(n p 287)
PositiveNegativePositiveNegative PositiveNegativePositiveNegative Positive Negative
Positive
Negative
14
60
2 12
37
4 10
21
111
21
0 28
77
7
227 25235 39175
NOTE.
were as follows: site 1,
aQuickVue Influenza A+B Test (Quidel).
Data are no. of specimens. Results are for influenza A or B virus. The number of specimens positive for influenza B virus by RT-PCR
; site 2, ; and site 3,.np 3np5np 50
a sterile polyester swab and either was placed on a cold pack,
refrigerated up to 1 day, and transported to a commercial lab-
oratory for tissue cell viral culture (November 2007–February
2008) or was shipped on cold packs to a public health labo-
ratory, where the specimen was aliquotted and frozen at?70?C
before being thawed and tested for influenza A and B by mul-
tiplex PCR (March–May 2008). At site 3, an additional nasal
swab specimen was collected from participants who met ILI2
criteria by using a sterile dacron swab and was transported to
the laboratory for influenza A and B virus testing by multiplex
PCR assay within a few hours after specimen collection (2007–
2008).
We calculated the sensitivity, specificity, and positive and
negative predictive values, as well as 95% CIs calculated using
binomial exact methods, for the QuickVue Influenza A+B Test
compared with confirmatory influenza testing by RT-PCR (all
sites) and viral culture (sites 1 and 2) for all specimens tested
at each site throughout the study period.Astandardinstrument
was used to collect data on the study populations from each
site and to compare methods of collecting, processing, and
testing respiratory specimens. Descriptive statistics were used
to analyze the characteristics of the study populations. Testing
of study participants was approved by the institutional review
boards of the institutions involved.
Results.
The majority of the study participants were chil-
dren and young adults; the median age was 19 years (range,
18–22 years) at site 1, 4 years (range, 3 months–71 years) at
site 2, and 8 years (range, 5–12 years) at site 3. At the time of
study enrollment, the median time from illness onset to spec-
imen collection was 3 days (range, 0–7 days) at site 1, 2 days
(range, 1 to 13 days; the time from onset to enrollment was
13 days for 6 of the 138 participants) at site 2, and 3 days
(range, !1 to 10 days) at site 3. The test results and test pa-
rameters for each site are presented in tables 1 and 2. Overall,
the calculated test parameters for the 3 sites in testing for either
influenza A or influenza B virus by QuickVue Influenza A+B
Test, compared with RT-PCR, were as follows: median sensi-
tivity, 26.7% (range, 18.9%–32.3%); median specificity, 97.2%
(range, 96.2%–99.6%); median positive predictive value,87.5%
(range, 80.0%–90.9%); and median negative predictive value,
69.4% (range, 62.5%–79.1%). Although the number of speci-
mens positive for influenza B virus was very small at sites 1
and 3, the median sensitivity was substantially higher for in-
fluenza A virus than for influenza B virus at site 2 but was still
suboptimal.
An investigation of possible explanations for the low sensi-
tivities was initiated. All specimens obtained were nasal swab
specimens (the preferred specimen for the QuickVue Influenza
A+B Test) collected using the foam swab supplied in the test
kit. All study sites used a small number of trained study per-
sonnel to collect clinical specimens and to perform the rapid
influenza test, and all specimens were tested either immediately
or, for a small number of participants, within 4–8 h after spec-
imen collection, in accordance with the manufacturer’s instruc-
tions. The test kit lot numbers for all QuickVue Influenza A+B
Tests used were different for each site, none of the kits were
expired, and all kits were stored under proper conditions as
recommended by the manufacturer.
Discussion.
We found very low sensitivity of the QuickVue
Influenza A+B Test, compared with RT-PCR, for detection of
either influenza A or influenza B virus among ill children and
young adults in 3 geographically distinct study populations.
Although most other reported sensitivities of rapid influenza
diagnostic tests were moderate in comparison with confirma-
tory assays such as RT-PCR or viral culture, we observed sen-
sitivities that were substantially lower than those reported pre-
viously for this test or for other commercially available rapid
influenza antigen tests [1, 8, 10–12]. Only 1 other published
study has reported such low sensitivity for this test in com-
parison with RT-PCR [11]. The specificity of the QuickVue
Influenza A+B Test was very high at all sites (196%), compared
with RT-PCR, which is consistent with results of previous stud-
ies [1, 8–12].
We were unable to identify common factors across all sites
that might account for the consistently suboptimal test sensi-
tivities. There were no deviations from the manufacturer’s rec-
ommended specimen type, sample storage, or test procedures,
and most specimens were tested immediately at 2 sites; at the
third site, only a small number of specimens were tested within
4–8 h after collection. It is unlikely thatthenasalswabspecimen

Page 3

BRIEF REPORT • CID 2009:48 (1 May) • e91
Table 2.Test parameters of the QuickVue Influenza A+B Test at 3 US sites, 2007–2008.
Parameter
Site 1
Site 2a
Site 3
PCR-tested specimens
(np 287)
PCR-tested
specimens
(np 303)
Culture-tested
specimens
(np 303)
PCR-tested
specimens
(np 67)
Culture-tested
specimens
(np 71)
PCR- and culture-tested
specimens
(np 138)
Sensitivity, % (95% CI)
Influenza A or B
Influenza A
Influenza B
Specificity, % (95% CI)
Influenza A or B
Influenza A
Influenza B
PPV, % (95% CI)
Influenza A or B
Influenza A
Influenza B
NPV, % (95% CI)
Influenza A or B
Influenza A
Influenza B
18.9 (10.8–29.7)
19.7 (11.2–30.9)
0 (0–70.8)
24.4 (13.3–38.9)
25.5 (14.0–40.4)
0 (0–84.2)
32.3 (16.7–51.4)
55.6 (21.2–86.3)
22.7 (7.8–45.4)
34.4 (18.6–53.2)
44.4 (21.5–69.2)
21.4 (4.7–50.8)
33.3 (22.0–46.3)
48.2 (28.7–68.1)
22.2 (10.1–39.2)
26.7 (18.5–36.2)
30.9 (19.2–44.8)
22.0 (11.5–36.0)
99.1 (96.9–99.9)
99.1 (96.9–99.9)
100 (98.8–100)
98.4 (96.0–99.6)
98.4 (96.1–99.6)
100 (98.8–100)
97.2 (85.5–99.9)
100 (93.6–100)
97.8 (88.2–99.9)
100 (91.0–100)
100 (93.3–100)
100 (93.7–100)
98.7 (92.8–100)
100 (96.7–100)
99.0 (94.7–100)
96.2 (92.2–98.4)
92.2 (88.0–95.3)
89.9 (85.3–93.4)
87.5 (61.7–98.5)
87.5 (61.7–98.5)
NA
75 (47.6–92.7)
75 (47.6–92.7)
NA
90.9 (58.7–99.8)
100 (47.8–100)
83.3 (35.9–99.6)
100 (71.5–100)
100 (63.1–100)
100 (29.2–100)
95.5 (77.2–99.9)
100 (75.3–100)
88.9 (51.8–99.7)
80.0 (63.1–91.6)
48.6 (31.4–66.0)
31.4 (16.9–49.3)
79.1 (73.9–83.7)
80.1 (75.1–84.6)
99.0 (97.1–99.8)
87.1 (82.7–90.8)
87.8 (83.5–91.4)
99.3 (97.6–99.9)
62.5 (54.4–72.5)
93.6 (84.3–98.2)
72.1 (59.2–82.9)
65.0 (51.6–76.9)
84.1 (72.7–92.1)
83.8 (72.9–91.6)
64.0 (54.4–72.5)
88.8 (81.9–93.7)
78.3 (70.2–85.1)
69.4 (63.4–75.1)
84.9 (79.9–89.1)
84.5 (79.5–88.8)
NOTE.
aAt site 2, viral culture was used as the reference standard comparison test from 14 November 2007 through 29 February 2008; RT-PCR was used as the
reference standard from 1 March 2008 through 19 May 2008.
NA, could not be calculated; NPV, negative predictive value; PPV, positive predictive value.
obtained for rapid testing had substantially less viral antigen
than did the specimen collected at the same time for the com-
parison assay for participants consistently across all 3 sites. The
choice of a reference standard test can affect the test sensitivity.
For example, sensitivitieshavegenerallybeenhigherwhenrapid
influenza tests were compared with viral culture [9, 10] than
when they were compared with RT-PCR [10–12]. However,
this does not account for the low sensitivities that we observed
using either method. At site 1, the QuickVue Influenza A+B
Test (using nasal swab specimens) in comparison with RT-PCR
(using throat swab specimens) resulted in a slightly lower sen-
sitivity (19.7%) for either influenza A or influenza B virus than
was found in the comparison with viral culture (24.4%),
whereas at site 2, the sensitivities for the 2 comparisons were
nearly identical (34.4% for comparison with culture and 32.3%
for comparison with RT-PCR). These sensitivities were much
less than those previously reported in other studies of the
QuickVue Influenza A+B Test [9, 10].
Our findings have clinical and public health implications
because this rapid influenza diagnostic test is used to promptly
detect influenza virus infections to prescribe antiviraltreatment
or to implement outbreak-control interventions. Becauseofthe
very low sensitivity of the QuickVue Influenza A+B Test, the
negative predictive value was moderate, indicating that there
were false-negative results. Even though the specificity was very
high, the positive predictive value was moderately high across
all sites, which indicates that although a positive result was
reasonably predictive of a diagnosis of influenza for the patients
tested, there were also false-positive results. For rapid influenza
diagnostic tests with moderate sensitivity and high specificity
compared with RT-PCR, positive and negative predictivevalues
vary by the prevalence of circulating influenza viruses among
the population being tested [8]. Other factors that might de-
crease the sensitivity and accuracy of a rapid influenza antigen
test are improper specimen collection; not testing the recom-
mended clinical specimen, because results may vary by the kind
of respiratory sample tested [12]; use of a swab that is not
recommended (e.g., using an unapproved tip or shaft material
or not using the foam swab supplied with the test); prolonged
time from illness onset to specimen collection, because viral
shedding may have decreased to undetectable levels; and im-
proper handling or storage procedures before testing of spec-
imens. Some possibilities to increase the test sensitivity include
use of flocked swabs or testing of combined specimens (e.g.,
pooled nasal and throat swab specimens). Although we eval-
uated only 1 of a number of commercially available rapid in-
fluenza tests, our findings highlight the need to develop more-
sensitive rapid influenza diagnostic tests to detect influenza A
and B viruses in respiratory specimens from patients of all ages
and to inform patient management.
Acknowledgments
We thank Donald Burke (principal investigator, Pittsburgh Influenza
Prevention Project); University of Michigan Health Services and Housing;
David Shay, William Thompson, and Steve Watermen (Centers for Disease
Control and Prevention); study personnel at all sites; participants and their
families; and the Division of Global Migration and Quarantine, Centers
for Disease Control and Prevention.
Financial support.
Centers for Disease Control and Prevention

Page 4

e92 • CID 2009:48 (1 May) • BRIEF REPORT
(through cooperative agreements to the University of Michigan, University
of Pittsburgh, and Columbia University in support of studies of non-
pharmaceutical interventions to control influenza transmission).
Potential conflicts of interest.
All authors: no conflicts.
References
1. Uyeki TM. Influenza diagnosis and treatment in children: a review of
studies on clinically useful tests and antiviral treatment for influenza.
Pediatr Infect Dis J 2003;22:164–77.
2. Sharma V, Dowd MD, SlaughterAJ, SimonSD.Effectofrapiddiagnosis
of influenza virus type A on the emergency department management
of febrile infants and toddlers. Arch Pediatr Adolesc Med 2002;156:
41–3.
3. Abanses JC, Dowd MD, SimonSD,SharmaV.Impactofrapidinfluenza
testing at triage on management of febrile infants and young children.
Pediatr Emerg Care 2006;22:145–9.
4. Poehling KA, Zhu Y, Tang YW, Edwards K. Accuracy and impact of a
point-of-care rapid influenza test in young children with respiratory
illnesses. Arch Pediatr Adolesc Med 2006;160:713–8.
5. Bonner AB, Monroe KW, Talley LI, Klasner AE, Kimberlin DW. Impact
of the rapid diagnosis of influenza on physician decision-making and
patient management in the pediatric emergency department: results of
a randomized, prospective, controlled trial. Pediatrics 2003;112:363–7.
6. D’Heilly SJ, Janoff EN, Nichol P, Nichol KL. Rapid diagnosis of influ-
enza infection in older adults: influence on clinical care in a routine
clinical setting. J Clin Virol 2008;42:124–8.
7. Falsey AR, Murata Y, Walsh EE. Impact of rapid diagnosis on man-
agement of adults hospitalized with influenza. Arch Intern Med
2007;167:354–60.
8. Grijalva CG, Poehling KA, Edwards KM, et al. Accuracy and inter-
pretation of rapid influenza tests in children. Pediatrics 2007;119:
e6–11.
9. Hurt AC, Alexander R, Hibbert J, Deed N, Barr IG. Performance of
six influenza rapid tests in detecting human influenza in clinical spec-
imens. J Clin Virol 2007;39:132–5.
10. Mehlmann M, Bonner AB, Williams JV, et al. Comparison of the
MChip to viral culture, reverse transcription-PCR, and the QuickVue
Influenza A+B test for rapid diagnosis of influenza. J Clin Microbiol
2007;45:1234–7.
11. Rashid H, Shafi S, Haworth E, et al. Value of rapid testing for influenza
among Hajj pilgrims. Travel Med Infect Dis 2007;5:310–3.
12. Agoritsas K, Mack K, Bonsu BK, Goodman D, Salamon D, Marcon
MJ. Evaluation of the Quidel QuickVue test for detection of influenza
A and B viruses in the pediatric emergency medicine setting by use of
three specimen collection methods. J Clin Microbiol 2006;44:2638–41.