Subscription Information for:

MAJOR ARTICLE

Evaluation of Laboratory Methods for Diagnosis

of Varicella
Jessica Leung,1 Rafael Harpaz,1 Andrew L. Baughman,1 Karl Heath,2,a Vladimir Loparev,1 Marietta Vázquez,3
Barbara M. Watson,2,a and D. Scott Schmid1
1
Centers for Disease Control and Prevention, Atlanta, Georgia; 2Philadelphia Department of Health, Pennsylvania; and 3Yale University School
of Medicine, New Haven, Connecticut

Background. The incidence of varicella disease is declining as a result of vaccination, making clinical diagnosis
more challenging, particularly for vaccine-modified cases. We conducted a comprehensive evaluation of laboratory
tests and specimen types to assess diagnostic performance and determine what role testing can play after skin
lesions have resolved.
Methods. We enrolled patients with suspected varicella disease in 2 communities. Enrollees were visited at the
time of rash onset and 2 weeks later. Multiple skin lesion, oral, urine, and blood or serum specimens were requested
at each visit and tested for varicella zoster virus (VZV) immunoglobulin (Ig) G, IgM, and IgA antibody by enzyme-
linked immunoassay; for VZV antigen by direct fluorescent antibody; and/or for VZV DNA by polymerase chain
reaction (PCR). Clinical certainty of the diagnosis of varicella disease was scored. PCR results from first-visit
vesicles or scab specimens served as the gold standard in assessing test performance.
Results. Of 93 enrollees, 53 were confirmed to have varicella disease. Among 20 unmodified cases, PCR testing
was 95%–100% sensitive for macular and/or papular lesions and for oral specimens collected at the first visit;
most specimens from the second visit yielded negative results. Among 27 vaccine-modified cases, macular and/or
papular lesions collected at the first visit were also 100% sensitive; yields from other specimens were poorer, and
few specimens from the second visit tested positive. Clinical diagnosis was 100% and 85% sensitive for diagnosing
unmodified and vaccine-modified varicella cases, respectively.
Conclusions. PCR testing of skin lesion specimens remains convenient and accurate for diagnosing varicella
disease in vaccinated and unvaccinated persons. PCR of oral specimens can sometimes aid in diagnosis of varicella
disease, even after rash resolves.

Since 1995, when varicella vaccine was licensed, the predictive value of the clinical diagnosis, especially by
incidence of varicella disease has decreased by 180% new physicians with limited experience with varicella.
[1, 2]. Before vaccine licensure, varicella disease was In addition, varicella disease in previously vaccinated
ubiquitous among children and had a characteristic persons is relatively common and often highly modi-
clinical presentation that allowed for a clinical diagnosis fied, with lesions that are fewer in number, more tran-
[1, 3]. However, the decrease in the rate of disease has sient, and often macular and/or papular rather than
likely been accompanied by a decrease in the positive vesicular. Laboratory testing is therefore increasingly
important for diagnosis of varicella disease and for case
confirmation by public health authorities. There is also
Received 10 December 2009; accepted 12 March 2010; electronically published an increasing need for alternative specimen types for
26 May 2010. diagnosis of varicella disease, especially during outbreak
The findings and conclusions in this article are those of the authors and do
not necessarily represent the views of the Centers for Disease Control and investigations, which often commence after rashes have
Prevention. resolved.
a
Present affiliations: Independent Consultant for the GAVI Alliance, Geneva,
Switzerland (B.M.W.); and Merck, West Point, Pennsylvania (K.H.). We evaluated various laboratory tests and specimen
Reprints or correspondence: Jessica Leung, Centers for Disease Control and types to assess their performance in diagnosis of vari-
Prevention, National Center for Immunization and Respiratory Diseases, 1600
Clifton Rd, Mailstop A-47, Atlanta, GA 30333 ([email protected]). cella disease and to see whether they can serve as ad-
Clinical Infectious Diseases 2010; 51(1):23–32 juncts to polymerase chain reaction (PCR) testing of
 2010 by the Infectious Diseases Society of America. All rights reserved.
1058-4838/2010/5101-0004$15.00
skin lesions for varicella zoster virus (VZV) DNA,
DOI: 10.1086/653113 which is currently regarded as the most sensitive and

Laboratory Diagnosis of Varicella Disease • CID 2010:51 (1 July) • 23

specific method. We also explored whether diagnostic perfor- using an in-house whole-cell enzyme-linked immunoassay
mance varied in persons with a history of varicella disease or (ELISA); samples that yielded negative or equivocal results were
vaccination. retested using the more sensitive glycoprotein ELISA, as previ-
ously described [10]. In-house capture ELISAs were used to de-
METHODS tect VZV IgM and IgA antibody [10]. IgG titration was done for
patients with both acute- and convalescent-phase venipuncture
Study enrollment. During the period from January 2005
blood samples, as previously described [10]; a ⭓4-fold increase
through June 2006, we enrolled persons of any age in Phila-
in titer was considered diagnostic of disease infection. For acute-
delphia, Pennsylvania, and New Haven County, Connecticut,
phase serum samples with detectable VZV IgG, avidity was tested,
who were suspected by physicians and school nurses as having
as previously described, to provide supporting information re-
varicella disease. Both sites have had extensive experience with
garding history of varicella disease [11, 12].
varicella surveillance and have participated in population-based
Direct fluorescent antibody (DFA) testing requires intact in-
studies of varicella disease and vaccine effectiveness [4–9]. This
fected cells and, thus, was performed only on vesicular samples
study was approved by the institutional review boards of the
(Chemicon; Chemicon International). Results were interpret-
Centers for Disease Control and Prevention (CDC) and both
ed on the basis of the manufacturer’s recommendations. Slides
study sites. Enrollees or their parents or guardians provided
with ⭓20 cells were classified as adequate.
informed consent.
DNA samples for PCR testing were extracted and purified
Data and specimen collection. Study participants were vis-
using an automated platform (Magna Pure; Roche Diagnostics)
ited twice by a physician or study nurse experienced with di-
[13–15] and genotyped as wild-type or Oka strain, as described
agnosis of varicella disease and specimen collection. The first
elsewhere [10, 11]. Because VZV is highly cell associated, oral
visit occurred as soon as possible after rash onset; the second
and urine specimens were prepared to concentrate cellular ma-
visit occurred ∼2 weeks later. Standard questionnaires were
terial. Oral fluid specimens were prepared for DNA extraction
used for enrollee interviews. During the first visit, the study
nurses or physicians were asked to rate the clinical certainty of by soaking collection sponges overnight in phosphate-buffered
their diagnosis of varicella disease on a 5-point scale (1, un- saline, centrifuging at 400g for 10 min, and adding 200 mL of
likely; 5, very likely). reconstituted oral fluid to 300 mL of Magna Pure lysing buffer.
During both visits, we obtained skin lesion, buccal and throat Urine samples were prepared for DNA extraction by centri-
swab, oral fluid, urine, and blood specimens. Using established fuging for 10 min at 400g, aspirating 200 mL from the tube
methods, we obtained samples from each skin lesion type (ves- bottom, and mixing with 300 mL of Magna Pure lysing buffer.
icles, macular and/or papular lesions, and scabs) present [10, Actin was used as a PCR control; specimens with undetectable
11]. With the exception of urine and blood samples, we col- actin DNA were considered to be inadequate. Real-time Fröster
lected duplicate specimens whenever possible. resonance energy transfer PCR protocols targeting 4 vaccine-
Samples from vesicles, macular and/or papular lesions, and associated DNA polymorphisms in opening reading frames 38
scabs were collected with a slide and/or polyester swab. All (69349), 54 (95241), and 62 (106262, 107252) were done as
samples from lesions were placed in a sealed container and described elsewhere [13–15].
maintained at ambient temperatures. Classification of cases. PCR results from a vesicle or scab
Buccal and throat swab samples were collected by holding 2 specimen collected at the first visit was defined as our gold
swabs together and gently swabbing the buccal mucosa and the standard used to rule-in or rule-out a diagnosis of varicella
oropharynx (including tonsillar pillars), respectively, for several disease. Enrollees with a vesicle or scab specimen with a positive
seconds. Oral fluid was collected by using 2 OraSure sponges PCR result were classified as confirmed varicella case patients;
(OraSure Technologies) together and gently swabbing the cheeks enrollees with a specimen with a negative result were considered
and gums for several seconds. The sponges were then left in the non–case patients. If vesicle or scab specimens were not avail-
mouth for an additional minute to absorb oral fluid. able or if results were discordant, varicella case status was de-
Urine was collected in a standard urinalysis cup. Blood sam- fined as indeterminate, and the participants were excluded from
ples were obtained by venipuncture and/or finger-stick, de- analysis. Case patients were categorized on the basis of prior
pending on the enrollee’s consent. Blood samples obtained via varicella disease and on vaccination history. Cases in patients
finger-stick were blotted onto 2 filter paper circles (10 mm in with no varicella vaccination or disease history were termed
diameter). “unmodified varicella cases,” and cases in those with prior var-
Laboratory methods. Specimens were tested at the Na- icella vaccination were termed “vaccine-modified cases.” Prior
tional VZV Laboratory, CDC (Atlanta, GA), using multiple varicella disease and vaccination were validated by medical
methods. chart review. If the reported disease history was inconsistent
Serum samples were tested for VZV immunoglobulin (Ig) G with serologic results, a hierarchical approach was used to clas-

24 • CID 2010:51 (1 July) • Leung et al

sify prior varicella disease status, with IgG and/or avidity test of these participants had vesicles, and only 1 had a macular
results as the diagnostic reference standard. and/or papular lesion at this visit. PCR testing of scab specimens
Analysis of sensitivity and specificity. We defined sensitiv- was 90% sensitive; it was considerably less sensitive for oral,
ity for each diagnostic test as the percentage of confirmed var- urine, and blood specimens, although some samples had pos-
icella case patients (ie, on the basis of our gold standard) who itive test results up to 20 days after rash onset.
tested positive by the diagnostic assay. Specificity was defined During the first visit, PCR testing of macular and/or papular
as the percentage of non–case patients (also on the basis of our lesion, throat swab, and oral fluid samples obtained from 27
gold standard) who tested negative by the diagnostic assay. We enrollees with confirmed vaccine-modified varicella disease was
used results from the first adequate specimen of each type 70%–100% sensitive (Table 4). Sensitivity did not change by
collected to calculate sensitivity and specificity. We calculated interval between rash onset and specimen collection during the
95% exact binomial confidence intervals. SAS software, version 5 days after rash onset (data not shown). Sensitivity of PCR
9.1 (SAS Institute) [16], was used to analyze all data. testing of macular and/or papular lesions also did not vary by
lesion number among case patients with !50 lesions (data not
RESULTS shown). There were only 3 patients with vaccine-modified dis-
Demographic data and clinical and epidemiologic results. ease with whole-blood specimens. High clinical suspicion of
Of 299 patients with suspected varicella disease, 93 (66 from varicella disease was 85% sensitive. At the second visit, we
New Haven and 27 from Philadelphia) agreed to participate collected oral and urine specimens from almost all enrollees in
and were eligible for the study. Ages ranged from 0 to 48 years this group and scab specimens from 14 (52%) of them. Only
(median, 10 years), and most patients were white and non- 1 case patient had a vesicle; it tested negative. PCR testing of
Hispanic (Table 1). Because of differences in case ascertainment scab specimens was 69% sensitive. Sensitivity was low for other
by study site, the proportion of suspected varicella disease cases specimens, although some samples had positive results up to
that were vaccine modified varied by age and race. Of note, no 17 days after rash onset.
vaccinated case patients had received 11 dose of varicella vac- Varicella disease was excluded for 15 patients. We were able
cine. Vesicle or scab lesions sampled from 53 of 93 enrollees to use their results to determine that the diagnostic tests were
tested positive for VZV DNA, including specimens obtained specific (Table 5). Low clinical suspicion of varicella by clini-
from 20 of 27 enrollees who had no history of varicella disease cians was 70% specific.
or vaccination, 3 of 11 who had a history of varicella, 27 of 50 We did not have sufficient number of participants to assess
who had been vaccinated, and 3 of 5 whose disease (4 patients) sensitivity of either clinical or laboratory diagnosis by day num-
or vaccination (1 patient) status was unknown (Figure 1). Ves- ber following rash onset. For all enrollees tested positive by
icle and scab PCR results for remaining enrollees were either PCR, VZV was wild-type.
negative (15 of 93) or indeterminate (unavailable or discordant; Discordance in gold standard results. We established PCR
25 of 93).
results from a vesicle or scab specimen collected at the first
Clinical characteristics of the 53 case patients and 15 non–
visit as our gold standard which to classify varicella case-status.
case patients are shown in Table 2. Vaccine-modified cases-
To support this choice, we attempted to collect at least 2 vesicle
tients with vaccine-modified disease had fever, ⭓50 lesions,
or scab specimens at the first visit to see whether they yielded
or vesicles.
concordant results.
Diagnostic performance. Data regarding specimen and test
At the first visit, we obtained 97 specimens (2–8 per patient)
performance are shown in Tables 3–5. Only 3 confirmed case
from 20 enrollees with unmodified varicella disease; PCR results
patients had a history of varicella disease (performance was not
assessable because of the small sample size). for all 97 specimens were concordant-positive. We obtained 13
Among 20 case patients classified as having unmodified var- specimens (2–6 per patient) from 4 enrollees with prior varicella
icella disease (ie, they had no history of varicella disease or disease; PCR results for specimens from 3 of 4 case patients
vaccination), PCR testing of macular and/or papular lesions, were concordant-positive. The patient with disconcordant re-
cheek and throat swab samples, and oral fluid samples collected sults had 2 vesicles or scabs with negative PCR results and 1
during the first visit was 95%–100% sensitive (Table 3). Just 4 scab with positive PCR results. We obtained 106 specimens (1–
of 15 specimens were deemed adequate for DFA testing; 3 of 6 per patient) from the 29 enrollees with vaccine-modified
these 4 tested positive. High clinical suspicion of varicella dis- varicella disease; PCR results from 27 of 29 were concordant-
ease by clinicians was 100% sensitive. During the second visit, positive. The 2 patients with disconcordant results each had 3–
oral specimens were collected from almost all case patients with 5 vesicles or scabs test positive by PCR and 1 vesicle test negative
unmodified varicella disease, and scab, urine, and blood sam- by PCR. Finally, we obtained 38 specimens (1–5 per patient)
ples were obtained from approximately one-half of them. None from 15 confirmed non–case patients; all PCR results were

Laboratory Diagnosis of Varicella Disease • CID 2010:51 (1 July) • 25

 Table 1. Demographic Characteristics and Clinical Diagnosis, by Vaccination and Disease History, in Patients with
Suspected Cases of Varicella Disease

No. (%) of enrollees

Vaccinated,
Unvaccinated, no history Unknown
no history Unvaccinated, (vaccine-modified vaccination or
(unmodified varicella) prior history varicella) disease history
Variable, group (n p 27) (n p 11) (n p 50) (n p 5)
Study site
Philadelphia, PA 14 (52) 7 (64) 3 (6) 3 (60)
New Haven County, CT 13 (48) 4 (36) 47 (94) 2 (40)
Age group, years
0–4 9 (33) 0 (0) 8 (16) 0 (0)
5–19 11 (41) 8 (73) 42 (84) 4 (80)
⭓20 7 (26) 3 (27) 0 (0) 1 (20)
Sex
Male 18 (67) 4 (36) 27 (54) 5 (100)
Female 9 (33) 7 (64) 23 (46) 0 (0)
Ethnicity
Hispanic 7 (26) 0 (0) 5 (10) 1 (20)
Non-Hispanic 20 (74) 11 (100) 45 (90) 4 (80)
Race
Black 12 (44) 4 (36) 3 (6) 1 (20)
White 14 (52) 7 (64) 46 (92) 4 (80)
Other 1 (4) 0 (0) 1 (2) 0 (0)
Chickenpox likely diagnosisa
Yes 22 (88) 9 (90) 38 (83) 4 (80)
No 3 (12) 1 (10) 8 (17) 1 (20)
Unknown 2 1 4 0
Certainty of diagnosis when chickenpox
was likely diagnosis
Uncertain (rated 1–3)b 3 (14)c,d 5 (56)c,e 8 (21)c,f 3 (75)c,g
b c,d c,e c,f
Highly certain (rated 4–5) 19 (86) 4 (44) 30 (79) 1 (25)c,g
a
Percentage of enrollees excludes enrollees categorized as “unknown” and calculated only among those with known information.
b
At the first patient visit, diagnosing study nurse or physician rated the clinical certainty of their diagnosis of varicella disease on a 5-
point scale, with 1 being unlikely and 5 being very likely.
c
No. (%) of suspect case-patients among those the physician or study nurse thought varicella was the most likely diagnosis.
d
Data are for 22 patients.
e
Data are for 9 patients.
f
Data are for 38 patients.
g
Data are for 4 patients.

negative concordant. In total, our gold standard provided dis- of rash onset to inform clinicians seeing patients with acute
cordant results for 3 (4%) of 68 suspect VZV cases. illness, and again 2 weeks after rash onset, to assist public health
workers who are often only able to investigate sporadic or
DISCUSSION outbreak-associated cases after resolution of the illness. We at-
Varicella vaccination rates have reached high levels in the tempted to determine whether alternatives to PCR testing exist
United States [1]. The resulting decrease in the incidence of when skin lesions are no longer available for sampling.
varicella disease, coupled with its modification among vacci- PCR testing of vesicles or scabs sampled during early illness
nated individuals, is likely to lead to increasing uncertainty in provides sensitive and specific evidence of varicella [17–22],
the clinical diagnosis of varicella and greater reliance on the and previous studies have revealed it to be considerably more
laboratory testing. To our knowledge, this is the first study to sensitive than shell viral culture or standard culture [18, 19,
systematically evaluate a wide range of laboratory procedures 23]. This test served as the gold standard for our study. We
for their ability to diagnose varicella in both vaccinated and found that almost all PCR results were concordant, providing
unvaccinated people. We tested specimens collected at the time reassuring evidence about the performance of our gold stan-

26 • CID 2010:51 (1 July) • Leung et al

Figure 1. Study enrollment flowchart—final enrollee varicella classification (positive, negative, or indeterminate) as determined by polymerase chain
reaction (PCR) testing. +, positive; -, negative; +/-, indeterminate. aPCR result from a vesicle or scab specimen collected at the first visit was defined
as the gold standard and used to classify the disease status of suspect case-patients. bOf 93 study participants, 12 had equivocal whole-cell enzyme-
linked immunoassay (ELISA) immunoglobulin (Ig) G results from either the first or second visit. Of the 12 equivocal results, 3 were reclassified as IgG
positive, 5 were re-classified as IgG negative, and the remaining 4 could not be classified on the basis of glycoprotein ELISA (gpELISA) results. Among
the laboratory-confirmed unvaccinated case patients, there was 1 who was reclassified as having a negative IgG as determined on the basis of the
gpELISA result and 1 who could not be reclassified. There were 2 laboratory-confirmed vaccinated case patients with equivocal whole-cell IgG ELISA
results, which were re-tested with gpELISA: one was reclassified as having a positive IgG, and the other’s results remained equivocal after gpELISA
testing. Of the 93 participants, 20 had negative whole-cell ELISA IgG results for specimens obtained during either the first or second visit; all remained
IgG negative after gpELISA testing. cThere were 14 unvaccinated enrollees with both disease history information and avidity test results, which were
concordant with the exception of a 1-month old infant, who had reported no prior varicella disease history, but had a high varicella zoster virus (VZV)
IgG avidity result. The high avidity in this case is likely from maternal antibodies and therefore this enrollee was classified as unvaccinated with no
prior disease history. There were 3 vaccinated enrollees with both disease history information and avidity test results. All had no prior varicella disease
history, but had high VZV IgG avidity results; their high avidity is likely due to their varicella vaccination. dVaricella status for enrollees categorized
as PCR +/⫺ was defined as indeterminate because they were either missing a vesicle or scab sample at the first visit or they had disconcordant
vesicle or scab PCR results in the first visit. All but 3 cases were classified as indeterminate because the patients were missing a vesicle and scab
specimen at the first visit; the 3 exceptions included 1 unvaccinated patient with a prior disease history and 2 vaccinated patients.

dard. We found that PCR testing of macular and/or papular cases in our study were sampled 0–3 days after rash onset,
lesions collected soon after rash onset also yielded sensitive whereas 77% of cases in Weinmann et al [10] were sampled at
results. These findings are particularly important, because vac- 1–7 days. When detectable, VZV-specific IgM antibody is an
cine-modified varicella disease often manifests with macular indication of recent exposure to the virus, but it may not dis-
and/or papular lesions only [3, 24]. Sensitivity did not vary by criminate between primary infection, reinfection, or reactiva-
number of total lesions during rash, suggesting that any avail- tion [10, 17]. Documentation of increases in IgG antibody titers
able lesions should be adequate. PCR testing of oral specimens requires collection of 2 specimens, making it unsuitable for
was also sensitive, particularly among case patients with un- diagnosing varicella disease during the acute illness. Further-
modified varicella disease. Both skin lesions and oral specimens more, increases in IgG titers may be difficult to demonstrate
yielded specific results as well. All tests were less useful 2 weeks in people with preexisting titers due to vaccination or previous
after rash onset; skin lesions were not readily available once infection. During primary infection, IgG responses may not be
the illness resolved, and test results for oral specimens were detectable at the acute time-point and are probably still de-
mostly negative. veloping 2 weeks after rash onset [26]; end point titration did
Other tests were less valuable for diagnosing varicella [10, not improve our test sensitivity. Urine yielded insensitive re-
17, 25]. DFA requires properly collected specimens and yielded sults. Although VZV in urine or oral fluid specimens should
mostly indeterminate results, even in the hands of experienced all be cell-associated, false-negative results could have been
staff. IgM was insensitive, although one study found IgM to possible if some VZV occurred in a cell-free state.
be 75% sensitive among vaccine-modified cases [10]. Although Oral specimens and urine samples do not appear to offer
the reason for this discrepancy is unknown, it may be due to clear advantages for diagnosing varicella disease, particularly
the timing of specimen collection; 78% of vaccine-modified because skin lesions can be sampled readily and with minimal

Laboratory Diagnosis of Varicella Disease • CID 2010:51 (1 July) • 27

 Table 2. Clinical Characteristics of Case Patients and Non–Case Patients

No. (%) of enrollees

Case patients
Vaccinated,
Unvaccinated, no history Unknown
no history Unvaccinated, (vaccine-modified vaccination or Non-case
(unmodified varicella) prior history varicella) disease history patients
Variable, group (n p 20) (n p 3) (n p 27) (n p 3) (n p 15)
a,b
Fever (temperature, ⭓37.25C [⭓99F])
Yes 13 (65) 1 (33) 11 (42) 1 (33) 1 (7)
No 7 (35) 2 (67) 15 (58) 2 (67) 13 (93)
Unknown 0 0 1 0 1
Itchy rashb
Yes 16 (89) 3 (100) 24 (92) 3 (100) 8 (53)
No 2 (11) 0 (0) 2 (8) 0 (0) 7 (47)
Unknown 2 0 1 0 0
No. of lesions
!50 2 (10) 1 (33) 13 (48) 0 (0) 8 (53)
⭓50 18 (90) 2 (67) 14 (52) 3 (100) 7 (47)
Type of lesionsc
Macules 7 (35) 1 (33) 22 (81) 1 (33) 10 (67)
Papules 11 (55) 1 (33) 19 (70) 1 (33) 10 (67)
Pustules 9 (45) 0 (0) 4 (15) 0 (0) 3 (20)
Scabs 14 (70) 3 (100) 24 (89) 2 (67) 7 (47)
Vesicles 15 (75) 1 (33) 11 (41) 3 (100) 4 (27)
b
Chickenpox likely diagnosis
Yes 19 (100) 2 (67) 26 (100) 3 (100) 9 (90)
No 0 (0) 1 (33) 0 (0) 0 (0) 1 (10)
Unknown 1 0 1 0 5
Certainty of diagnosis when chickenpox
likely diagnosis
d e e e e e
Uncertain (rated 1–3) 0 (0) 0 (0) 4 (15) 0 (0) 6 (67)
d e e e e e
Highly certain (rated 4–5) 19 (100) 2 (100) 22 (85) 3 (100) 3 (33)
a
Measured tactilely or by thermometer.
b
Percentage of enrollees excludes enrollees categorized as unknown and calculated only among those with known information.
c
Represents whether type of lesion was present during rash. Categories are not mutually exclusive, so cases may have involved multiple types of lesions.
d
At the first patient visit, diagnosing study nurse or physician rated the clinical certainty of their diagnosis of varicella disease on a 5-point scale, with 1
being unlikely and 5 being very likely.
e
No. (%) of enrollees calculated among those where the physician or study nurse thought varicella disease was the most likely diagnosis.

invasiveness. However, in select circumstances, such uncon- accurate, even among vaccinated persons, particularly when the
ventional specimens can serve as useful alternatives, because clinician is confident about the diagnosis. Scabs, recognized
they are easy to obtain and can sometimes yield positive results exposure to varicella or herpes zoster disease, and school at-
after skin lesions have resolved. This might be particularly rel- tendance have been found to support the diagnosis of vaccine-
evant in the setting of outbreak investigations in which many modified varicella disease [27].
children need to be evaluated after their rashes have cleared. Eleven study participants reported prior episodes of varicella
Better laboratory tools are still needed to fill this public health disease, but only 3 had evidence of prior disease. Second ep-
need. isodes of varicella disease have been reported in the literature
Among our study participants with unmodified or vaccine- [28–31], including one study suggesting that the proportion of
modified varicella disease, clinical diagnosis was 100% and 85% such cases is increasing [31], perhaps in association with re-
sensitive, respectively. We were able to rule out cases of sus- duced VZV-specific immunity as exposures to varicella de-
pected varicella disease through laboratory testing, allowing us creases or with careful case seeking for vaccine-modified dis-
to determine that the specificity of clinical diagnosis was 70%. ease. Avidity testing may provide a useful tool to monitor this
Our results suggest that clinical diagnosis of varicella can be phenomenon more comprehensively.

28 • CID 2010:51 (1 July) • Leung et al

 Table 3. Sensitivity of Clinical Diagnosis and Laboratory Testing among 20 Persons with Unmodified
Varicella Disease

No. of
No. of specimens No. of
specimens found to be VZV- positive Sensitivity,
a
Visit no., diagnostic test collected adequate specimens % (95% CI)
b
Visit 1
c
Clinical diagnosis (rated 4–5) 19 NA 19 100 (82–100)
Vesicle swab DFA 15 4 3 75 (19–99)
Vesicle slide DFA 0 0 0 —
Macular and/or papular swab PCR 15 15 15 100 (78–100)
Macular and/or papular slide PCR 7 7 7 100 (59–100)
Cheek swab PCR 20 20 19 95 (75–100)
Throat swab PCR 18 18 18 100 (81–100)
Oral fluid PCR 19 19 19 100 (82–100)
Oral fluid IgA 19 17 0 0 (0–20)
Urine PCR 15 12 7 58 (28–85)
Whole-blood PCR 12 12 5 42 (15–72)
Whole-blood IgM 12 12 3 25 (5–57)
Whole-blood IgA 12 12 2 17 (2–48)
Fingerstick IgM 13 11 2 18 (2–52)
IgG titer (4-fold increase) 9 9 3 33 (7–70)
Visit 2d
Vesicle swab PCR 0 0 0 —
Vesicle slide PCR 0 0 0 —
Scab PCR 11 10 9 90 (56–100)
Vesicle swab DFA 0 0 0 —
Vesicle slide DFA 0 0 0 —
Macular and/or papular swab PCR 1 1 0 0 (0–98)
Macular and/or papular slide PCR 1 1 0 0 (0–98)
Cheek swab PCR 17 16 1 6 (0–30)
Throat swab PCR 17 16 3 19 (4–46)
Oral fluid PCR 17 16 5 31 (11–59)
Oral fluid IgA 17 16 0 0 (0–21)
Urine PCR 12 9 1 11 (0–48)
Whole-blood PCR 9 8 2 25 (3–65)
Whole-blood IgM 9 8 2 25 (3–65)
Whole-blood IgA 9 8 1 13 (0–53)
Fingerstick IgM 11 9 2 22 (3–60)

NOTE. Polymerase chain reaction (PCR) results from a vesicle or scab specimen collected at the first visit were defined as
the gold standard and used to classify disease status of case-patients. Dashes (—) indicate value was not calculated because
adequate sample was not obtained. CI, confidence interval; DFA, direct fluorescent antibody; Ig, immunoglobulin; NA, not
applicable; PCR, polymerase chain reaction; VZV, varicella zoster virus.
a
No. of VZV-positive specimens/no. of adequate specimens.
b
For 17 (85%) of 20 enrollees, the first visit took place 0–5 days (range, 0–9) after rash onset.
c
At the first patient visit, diagnosing study nurse or physician rated the clinical certainty of their diagnosis of varicella disease
on a 5-point scale, with 1 being unlikely and 5 being very likely.
d
For 15 (88%) of 17 enrollees having 2 visits, the second visit took place 13–17 days (range, 13–20 days) after rash onset.

There were limitations to our study. It was conducted using lection is important. Persons with mild modified varicella dis-
experienced staff, in communities with clinicians that had par- ease may not have sought medical attention or the diagnosis
ticipated in prior varicella surveillance activities. The quality of may have otherwise not been considered.
case finding, clinical recognition, and sample collection may As the epidemiology of varicella disease changes, it has be-
therefore have differed from other settings, which could have come increasingly important to test suspected cases to obtain
affected assay performance, particularly for PCR testing of mac- a clinical diagnosis for case management and outbreak inves-
ular and/or papular lesions, for which proper specimen col- tigation and to monitor the impact of the varicella vaccine

Laboratory Diagnosis of Varicella Disease • CID 2010:51 (1 July) • 29

 Table 4. Sensitivity of Clinical Diagnosis and Laboratory Testing among 27 Persons with Vaccine-Modified
Varicella Disease

No. of
No. of specimens No. of
specimens found to be VZV- positive Sensitivity,
a
Visit no., diagnostic test collected adequate specimens % (95% CI)
b
Visit 1
c
Clinical diagnosis (rated 4–5) 26 NA 22 85 (65–96)
Vesicle swab DFA 6 1 0 0 (0–98)
Vesicle slide DFA 1 0 0 —
Macular and/or papular swab PCR 24 24 24 100 (86–100)
Macular and/or papular slide PCR 20 18 18 100 (81–100)
Cheek swab PCR 27 26 16 62 (41–80)
Throat swab PCR 27 27 19 70 (50–86)
Oral fluid PCR 27 26 22 85 (65–96)
Oral fluid IgA 27 27 0 0 (0–13)
Urine PCR 26 24 8 33 (16–55)
Whole-blood PCR 3 3 0 0 (0–71)
Whole-blood IgM 3 3 1 33 (1–91)
Whole-blood IgA 3 2 0 0 (0–84)
Fingerstick IgM 22 18 3 17 (4–41)
IgG titer (4-fold increase) 1 1 1 100 (3–100)
Visit 2d
Vesicle swab PCR 1 1 0 0 (0–98)
Vesicle slide PCR 0 0 0 —
Scab PCR 14 13 9 69 (39–91)
Vesicle swab DFA 0 0 0 —
Vesicle slide DFA 1 0 0 —
Macular and/or papular swab PCR 5 5 3 60 (15–95)
Macular and/or papular slide PCR 5 4 1 25 (1–81)
Cheek swab PCR 25 24 0 0 (0–14)
Throat swab PCR 24 23 1 4 (0–22)
Oral fluid PCR 25 25 4 16 (5–36)
Oral fluid IgA 25 24 0 0 (0–14)
Urine PCR 22 19 0 0 (0–18)
Whole-blood PCR 1 1 1 100 (3–100)
Whole-blood IgM 1 1 0 0 (0–98)
Whole-blood IgA 1 1 0 0 (0–98)
Fingerstick IgM 19 14 4 29 (8–58)

NOTE. Polymerase chain reaction (PCR) results from a vesicle or scab specimen collected at the first visit were defined as
the gold standard and used to classify disease status of case-patients. Dashes (—) indicate value was not calculated because
adequate sample was not obtained. CI, confidence interval; DFA, direct fluorescent antibody; Ig, immunoglobulin; NA, not
applicable; PCR, polymerase chain reaction; VZV, varicella zoster virus.
a
No. of VZV-positive specimens/no. of adequate specimens.
b
For 21 (78%) of 27 enrollees, the first visit took place 0–3 days (range, 0–5 days) after rash onset.
c
At the first patient visit, diagnosing study nurse or physician rated the clinical certainty of their diagnosis of varicella disease
on a 5-point scale, with 1 being unlikely and 5 being very likely.
d
For 20 (80%) of 25 enrollees having 2 visits, the second visit took place 12–17 days (range, 12–19 days) after rash onset.

program. We have shown that PCR testing of skin lesions is case ascertainment is critical for evaluating the impact and
highly sensitive and specific for detecting VZV, and oral spec- effectiveness of varicella vaccine. For now, clinicians and public
imens can play a supporting diagnostic role in certain settings. health workers should be encouraged to request PCR testing
However, PCR testing is not universally available, and better of suspected cases of varicella disease, and public and com-
tests would be useful for public health workers to diagnose mercial laboratories should be encouraged to conduct such
varicella disease after transient lesions have cleared. This is testing. Lastly, the research community should be encouraged
particularly important in outbreak situations, in which accurate to develop new methods for diagnosing varicella and ascer-

30 • CID 2010:51 (1 July) • Leung et al

Table 5. Specificity of Clinical Diagnosis and Laboratory Testing among 15 Persons with Rash-
Illnesses for Whom Varicella Disease Was Excluded

No. of
No. of specimens No. of
specimens found to be VZV- negative Specificity,
Visit no., diagnostic test collected adequate specimens %a (95% CI)
Visit 1
Clinical diagnosis (not likely vari-
cella or certainty diagnosis
rated 1–3)b 10 NA 7 70 (30–93)
Vesicle swab DFA 7 0 0 —
Vesicle slide DFA 0 0 0 —
Macular and/or papular swab PCR 13 11 11 100 (72–100)
Macular and/or papular slide PCR 11 9 9 100 (66–100)
Cheek swab PCR 15 15 15 100 (78–100)
Throat swab PCR 15 15 15 100 (78–100)
Oral fluid PCR 15 15 15 100 (78–100)
Oral fluid IgA 15 14 14 100 (77–100)
Urine PCR 11 11 11 100 (72–100)
Whole-blood PCR 3 3 3 100 (29–100)
Whole-blood IgM 4 3 3 100 (29–100)
Whole-blood IgA 4 4 4 100 (40–100)
Fingerstick IgM 13 9 9 100 (66–100)
IgG titer (4-fold increase) 3 3 3 100 (29–100)
Visit 2
Vesicle swab PCR 7 5 5 100 (48–100)
Vesicle slide PCR 7 7 7 100 (59–100)
Scab PCR 9 9 9 100 (66–100)
Vesicle swab DFA 0 0 0 —
Vesicle slide DFA 0 0 0 —
Macular and/or papular swab PCR 2 1 1 100 (3–100)
Macular and/or papular slide PCR 2 1 1 100 (3–100)
Cheek swab PCR 12 11 10 91 (59–100)
Throat swab PCR 12 11 11 100 (72–100)
Oral fluid PCR 12 11 10 91 (59–100)
Oral fluid IgA 12 12 12 100 (74–100)
Urine PCR 11 11 11 100 (72–100)
Whole-blood PCR 3 3 3 100 (29–100)
Whole-blood IgM 3 2 2 100 (16–100)
Whole-blood IgA 3 3 3 100 (29–100)
Fingerstick IgM 10 9 9 100 (66–100)

NOTE. Polymerase chain reaction (PCR) results from a vesicle or scab specimen collected at the first visit were
defined as the gold standard and used to classify disease status of case-patients. These 15 enrollees include all
enrollees, regardless of varicella vaccination and disease history, who were laboratory-confirmed as not being a
varicella case. Dashes (—) indicate value was not calculated because adequate sample was not obtained. CI,
confidence interval; DFA, direct fluorescent antibody; Ig, immunoglobulin; NA, not applicable; PCR, polymerase chain
reaction; VZV, varicella zoster virus.
a
No. of VZV-negative specimens/no. of adequate specimens.
b
Represents the number of non-cases with available data on physician diagnosis and a rating for certainty of
diagnosis if the physician thought that varicella might be a likely diagnosis on a 5-point scale, with 1 being unlikely
and 5 being very likely. One case was not diagnosed as likely varicella by the clinician; nine cases were thought
to be likely varicella by the clinician and had available data on the rating of certainty of diagnosis; and 5 cases were
missing clinician diagnosis
taining recent VZV exposure (eg, the analysis of VZV-specif- Lessons from a rash outbreak investigation in the republic of the Congo.
Am J Trop Med Hyg 2009; 80:503–507.
ic T and B cells, memory B cells, and activation markers in 13. Loparev VN, Rubtcova EN, Bostik V, et al. Distribution of varicella-
peripheral blood lymphocytes) in the absence of remaining zoster virus (VZV) wild-type genotypes in northern and southern Eu-
lesions. rope: evidence for high conservation of circulating genotypes. Virology
2009; 383:216–225.
Acknowledgments 14. Loparev VN, McCaustland K, Holloway BP, Krause PR, Takayama M,
Schmid DS. Rapid genotyping of varicella-zoster virus vaccine and
We acknowledge and thank Valarie Parcells and Nancy Holabird for their wild-type strains with fluorophore-labeled hybridization probes. J Clin
extensive work in collecting the specimens and questionnaire information; Microbiol 2000; 38:4315–4319.
Dr Eugene Shapiro for his support in planning and setting up this study; 15. Loparev VN, Martro E, Rubtcova E, et al. Toward universal varicella-
and the reporting sites in West Philadelphia and New Haven who partic- zoster virus (VZV) genotyping: diversity of VZV strains from France
ipated in the study. We express our gratitude to all the members of the and Spain. J Clin Microbiol 2007; 45:559–563.
CDC VZV laboratory, including Denise Brown, Martha Thieme, and Mar- 16. SAS Institute. SAS OnlineDoc 9.1. Cary, NC: SAS Institute, 2004.
lene Deleon-Carnes, for their help conducting the laboratory testing; to http://support.sas.com/91doc/docMainpage.jsp. Accessed 22 Septem-
Tureka Watson, Riduan Joesoef, and Cedric Brown for their assistance with ber 2009.
data-cleaning and statistical advice; to Dr Aisha Jumaan for her advice 17. Arvin AM. Varicella-zoster virus. Clin Microbiol Rev 1996; 9:361–381.
during this project; and to Dr Stephanie Bialek and Claudia Chesley for 18. Epsy MJ, Teo R, Ross TK, et al. Diagnosis of varicella-zoster virus
their thoughtful review of this manuscript. infections in the clinical laboratory by LightCycler PCR. J Clin Mi-
Potential conflicts of interest. B.M.W. served on the Merck advisory crobiol 2000; 38:3187–3189.
boards in 2007–2008. K.H. is currently an employee and stockholder with 19. Stranska R, Schuurman R, de Vos M, van Loon AM. Routine use of
Merck. All other authors: no conflicts. a highly automated and internally controlled real-time PCR for the
diagnosis of herpes simplex virus and varicella-zoster virus infections.
References J Clin Virol 2004; 30:39–44.
20. Koropchak C, Graham G, Palmer J, et al. Investigation of varicella-
1. Marin M, Meissner HC, Seward JF. Varicella prevention in the United zoster virus infection by polymerase chain reaction in the immuno-
States: a review of success and challenges. Pediatrics 2008; 122:e744– competent host with acute varicella. J Infect Dis 1991; 163:1016–1022.
e751. 21. Beards G, Graham C, Pillay D. Investigation of vesicular rashes for
2. Marin M, Güris D, Chaves SS, Schmid S, Seward JF; Advisory Com- HSV and VZV by PCR. J Med Virol 1998; 54:155–157.
mittee on Immumization Practices, Centers for Disease Control and 22. Nahass GT, Mandel MJ, Cook S, Fan W, Leonardi CL. Detection of
Prevention. Prevention of varicella: recommendations of the Advisory herpes simplex virus and varicella-zoster infection from cutaneous le-
Committee on Immumization Practices. MMWR Morb Mortal Wkly
sions in different clinical stages with the polymerase chain reaction. J
Rep 2007; 56(RR-4):1–40.
Am Acad Dermatol 1995; 32:730–733.
3. Vázquez M. Varicella infections and varicella vaccine in the 21st cen-
23. Schmid DS, Jumaan AO. Impact of varicella vaccine on varicella-zoster
tury. Pediatr Infect Dis J 2004; 23:871–872.
virus dynamics. Clin Microbiol Rev 2010; 23:202–217.
4. Guris D, Jumaan AO, Mascola L, et al. Changing varicella epidemiolo-
24. Chaves SS, Zhang J, Civen R, et al. Varicella disease among vaccinated
gy in active surveillance sites—United States, 1995–2005. J Infect Dis
persons: clinical and epidemiological characteristics, 1997–2005. J In-
2008; 197(Suppl 2):S71–S75.
fect Dis 2008; 197(Suppl 2):S127–S131.
5. Watson B, Civen R, Reynolds M, et al. Validity of self-reported varicella
25. Vázquez M, LaRussa PS, Gershon AA, Steinberg SP, Freudigman K,
disease history in pregnant women attending prenatal clinics. Public
Shapiro ED. The effectiveness of the varicella vaccine in clinical prac-
Health Rep 2007; 122:499–506.
tice. N Engl J Med 2001; 344:955–960.
6. Seward JF, Watson BM, Peterson CL, et al. Varicella disease after in-
26. Watson B, Keller PM, Ellis RW, Starr SE. Cell-mediated immune re-
troduction of varicella vaccine in the United States, 1995–2000. JAMA
sponses after immunization of healthy seronegative children with var-
2002; 287:606–611.
icella vaccine: kinetics and specificity. J Infect Dis 1990; 162:794–799.
7. Vázquez M, LaRussa PS, Gershon AA, et al. Effectiveness over time of
27. Perella D, Watson BM, Heath K, Robinson D, Schmid DS, Spain SV.
varicella vaccine. JAMA 2004; 291:851–855.
Laboratory confirmation and evaluation of case definitions for diag-
8. Lin F, Hadler JL. Epidemiology of primary varicella and herpes zoster
hospitalizations: the pre-varicella vaccine era. J Infect Dis 2000; 181: nosing breakthrough varicella infections. In: Program and abstracts of
1897–1905. the 41st annual meeting of the National Immunization Conference
9. Niccolai LM, Ogden LG, Muehlenbein CE, Dziura JD, Vázquez M, (Kansas City). Atlanta, GA: Centers for Disease Control and Preven-
Shapiro ED. Methodological issues in design and analysis of a matched tion, 2007:37.
case-control study of a vaccine’s effectiveness. J Clin Epidemiol 2007;60: 28. Terada K, Kawano S, Shimada Y, Yagi Y, Kataoka N. Recurrent chick-
1127–1131. enpox after natural infection. Pediatr Infect Dis J 1996; 15:179–181.
10. Weinmann S, Chun C, Mullooly JP, et al. Laboratory diagnosis and 29. Junker AK, Angus E, Thomas EE. Recurrent varicella-zoster virus in-
characteristics of breakthrough varicella in children. J Infect Dis 2008; fections in apparently immunocompetent children. Pediatr Infect Dis
197(Suppl 2):S132–S138. J 1991; 10:569–575.
11. Lopez AS, Burnett-Hartman A, Nambiar R, et al. Transmission of a 30. Gershon AA, Steinberg SP, Gelb L. Clinical reinfection with varicella-
newly characterized strain of varicella-zoster virus from a patient with zoster virus. J Infect Dis 1984; 149:137–142.
herpes zoster in a long-term care facility, West Virginia, 2004. J Infect 31. Hall S, Maupin T, Seward J, et al. Second varicella infections: are they
Dis 2008; 197:646–653. more common than previously thought? Pediatrics 2002; 109:1068–
12. MacNeil A, Reynolds MG, Carroll DS, et al. Monkeypox or varicella? 1073.

32 • CID 2010:51 (1 July) • Leung et al