The n e w e ng l a n d j o u r na l of m e dic i n e

original article

Loss of Vaccine-Induced Immunity

to Varicella over Time
Sandra S. Chaves, M.D., M.Sc., Paul Gargiullo, Ph.D., John X. Zhang, Ph.D.,
Rachel Civen, M.D., Dalya Guris, M.D., M.P.H., Laurene Mascola, M.D., M.P.H.,
and Jane F. Seward, M.B., B.S., M.P.H.

A BS T R AC T

Background
The introduction of universal varicella vaccination in 1995 has substantially reduced From the Centers for Disease Control and
varicella-related morbidity and mortality in the United States. However, it remains Prevention, Atlanta (S.S.C., P.G., J.X.Z.,
D.G., J.F.S.); and the Los Angeles County
unclear whether vaccine-induced immunity wanes over time, a condition that may Department of Health Services, Los An-
result in increased susceptibility later in life, when the risk of serious complications geles (R.C., L.M.). Address reprint requests
may be greater than in childhood. to Dr. Chaves at the Centers for Disease
Control and Prevention, 1600 Clifton Rd.,
NE, Mailstop A-47, Atlanta, GA 30333, or at
Methods bev8@cdc.gov.
We examined 10 years (1995 to 2004) of active surveillance data from a sentinel
N Engl J Med 2007;356:1121-9.
population of 350,000 subjects to determine whether the severity and incidence of Copyright 2007 Massachusetts Medical Society.
breakthrough varicella (with an onset of rash >42 days after vaccination) increased
with the time since vaccination. We used multivariate logistic regression to adjust
for the year of disease onset (calendar year) and the subjects age at both disease
onset and vaccination.

Results
A total of 11,356 subjects were reported to have varicella during the surveillance
period, of whom 1080 (9.5%) had breakthrough disease. Children between the ages
of 8 and 12 years who had been vaccinated at least 5 years previously were signifi-
cantly more likely to have moderate or severe disease than were those who had been
vaccinated less than 5 years previously (risk ratio, 2.6; 95% confidence interval [CI],
1.2 to 5.8). The annual rate of breakthrough varicella significantly increased with
the time since vaccination, from 1.6 cases per 1000 person-years (95% CI, 1.2 to 2.0)
within 1 year after vaccination to 9.0 per 1000 person-years (95% CI, 6.9 to 11.7) at
5 years and 58.2 per 1000 person-years (95% CI, 36.0 to 94.0) at 9 years.

Conclusions
A second dose of varicella vaccine, now recommended for all children, could improve
protection from both primary vaccine failure and waning vaccine-induced immunity.

n engl j med 356;11 www.nejm.org march 15, 2007 1121

Downloaded from www.nejm.org on July 2, 2008 . Copyright 2007 Massachusetts Medical Society. All rights reserved.
 The n e w e ng l a n d j o u r na l of m e dic i n e

T
he implementation of a universal Me thods
varicella vaccination program in the United
States in 1995 has resulted in a substantial Study Design
reduction in morbidity, mortality, and health care The Varicella Active Surveillance Project has been
costs associated with the disease.1-5 Despite this described previously.1 Briefly, since January 1995,
success, however, outbreaks of varicella continue to enhanced community-based surveillance for var-
occur, mostly in highly vaccinated school commu- icella has been conducted among a population of
nities.6-8 Several studies of these school outbreaks 350,000 persons in Antelope Valley, California,
have suggested that the time since vaccination a well-defined area 40 miles northeast of Los An-
may be associated with the risk of breakthrough geles. The population is predominantly white
varicella.6-10 It has been hypothesized that exog- (80%); of these persons, about 30% describe them-
enous reexposure to the virus may be needed to selves as Hispanic. The surveillance system com-
boost humoral and cellular immunity to varicella prises 300 reporting sites, which include child
zoster virus (VZV).11 As the incidence of varicella care centers, public and private schools, physicians
has decreased, so have the opportunities for com- in private practice, health maintenance organiza-
munity exposure to varicella needed to boost vac- tions, and public health clinics. Sites report on
cine-induced immunity. However, these investi- varicella every 2 weeks, regardless of whether a
gations of small outbreaks were not sufficiently subject with disease has been identified. Local
powerful to conclude that immunity to varicella personnel conduct a structured telephone inter-
wanes after vaccination. view with all subjects (or their parents or guard-
Serologic studies have indicated that the result ians) to collect demographic, clinical, and epidemi
of an enzyme-linked immunosorbent assay for a ologic data. Vaccination status is determined by
specific level of immunity to varicella (VZV glyco- a parental report of the childs vaccination record.
protein antigen of 5 units per milliliter) is an Since 1997, at least 80% of parental reports have
approximate correlate of protection against vari- been verified with health care providers or school
cella. On the basis of the results of this assay, records. The number of doses of varicella vaccine
studies have shown that about 15% of children that are administered each month has also been
receiving one dose of varicella vaccine do not have collected since 1995. We estimated that from 1995
levels of antibody that protect them from acquir- to 2004, on the basis of capturerecapture tech-
ing disease.12 This finding is compatible with niques, the annual reporting of varicella among
postlicensure studies indicating that one dose of children between the ages of 2 and 18 years was
varicella vaccine is about 80 to 85% effective approximately 70% complete (range, 66 to 84).
against any disease presentation.6-8,13-15 Thus, 15 Our study was evaluated by officials at the Cen-
to 20% of vaccinated children are at risk for vari- ters for Disease Control and Prevention (CDC) and
cella if they are exposed to VZV, either because the collaborating institutions. It was deemed that
they had no immune response or because vacci- we did not need to obtain individual informed
nation provided only partial protection. consent from the subjects.
Waning of immunity after varicella vaccina-
tion in terms of measurable antibodies has been Definitions of Disease
demonstrated to occur in health care workers.16 We defined a case of varicella as an acute onset of
To assess whether vaccine-induced immunity to a diffuse maculopapularvesicular rash without
varicella wanes, we used 10 years of data from a another apparent cause. We defined breakthrough
community-based active surveillance site to look varicella as a rash that developed more than 42
at the independent effect of the time since vac- days after the subject had been vaccinated with
cination on the severity and incidence of break- the live attenuated VZV vaccine Oka/Merck (Vari
through varicella. Waning of immunity is of par- vax, Merck). Since the vaccine itself may cause
ticular public health interest because it may rash, we excluded subjects with varicella who had
result in increased susceptibility later in life, when been vaccinated within the previous 42 days. We
the risk of severe complications may be greater categorized the severity of disease as either mild
than that in childhood. (<50 lesions) or moderate to severe (50 lesions).

1122 n engl j med 356;11 www.nejm.org march 15, 2007

Downloaded from www.nejm.org on July 2, 2008 . Copyright 2007 Massachusetts Medical Society. All rights reserved.
 Loss of Vaccine-induced Immunity to Varicella

Statistical Analysis jects (9.5%) had an onset of rash more than 42

We examined any association between potential days after vaccination (breakthrough varicella).
predictors of increased severity of disease separate Of these subjects, 770 (71.3%) either were evalu-
ly for subjects who were vaccinated and those who ated by a physician or received a laboratory diag-
were not vaccinated, using a two-sided chi-square nosis. Among all vaccinated and unvaccinated
test. We constructed two unconditional logistic- subjects, the proportion of children between the
regression models one for vaccinated subjects ages of 8 and 12 years was higher in the vacci-
and one for unvaccinated subjects to determine nated group than in the unvaccinated group (26%
which variables remained independent predictors vs. 19%, P<0.001). Vaccination status was similar
that subjects would have moderate-to-severe dis- in boys and girls, and racial and ethnic charac-
ease. Variables that had a significant association teristics in both the vaccinated and unvaccinated
with disease severity in the univariate analysis groups reflected the population profile in the
were included in the multivariate regression mod- surveillance area.
els. Variables that were not significantly associat- In 2003 and 2004, the average number of re-
ed with disease severity but that changed the odds ported cases of varicella was 420, representing an
ratio for severity by 10% or more when removed 85% decline from the total number of 2794 cases
from the analysis were also kept in the final reported in 1995. In the last 3 years of the decade
model.17 under study, however, no substantial reduction in
To determine the effect of the time since vac- disease was observed, despite a steady increase
cination on the incidence of breakthrough vari in estimates of vaccination coverage among chil-
cella of any severity, we constructed a Poisson dren 19 to 35 months of age (Fig. 1). The propor-
regression model. We used the data on doses of tion of cases that occurred in vaccinated children
varicella vaccine administered each month to increased from 1% (23 of 2269 subjects) in 1996
children between the ages of 1 and 12 years to 18% (126 of 704) in 2000 to 60% (312 of 521)
from 1995 through 2004 to create a study cohort in 2004.
for calculating person-years at risk for each year In 1995, before the full implementation of the
after vaccination. We excluded children over the varicella vaccination program, approximately 73%
age of 12 years, who may have received two doses. of cases of varicella occurred in children 6 years
Person-years at risk were defined as beginning at of age or younger, with peak disease frequency
the time of vaccination and ending at disease on between the ages of 3 and 6 years (Fig. 2A). This
set. Within each calendar year, however, person- pattern of disease distribution has changed in
years were not calculated for individual subjects recent years. In 2004, children who were 6 years
but were aggregated within groups as defined by
age at vaccination and age at disease onset. The Vaccinated Unvaccinated Coverage
rate of varicella in each year after vaccination was
3000
No. of Subjects with Varicella

100
then adjusted for the age at disease onset and

Vaccination Coverage (%)

90
2500
the year at risk for disease. Both variables were 80
70
considered predictors of the likelihood of expo- 2000
60
sure to VZV and therefore potential confounders 1500 50
for the association of the risk of breakthrough 1000
40
30
disease with the time since vaccination. All data 500 20
were analyzed at the CDC with the use of SAS 10
0 0
software, version 9.01 (SAS Institute). 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004
Year
R e sult s
Figure 1. Number of Subjects with Varicella According to the Calendar Year
and Vaccination Status and Annual Estimates of Vaccination Coverage.
Subjects with Varicella
Estimates of vaccination coverage are for children between the ages of 19 and
During the decade from January 1995 through 35 months in Los Angeles County and are based on the National Immuniza-
December 2004, a total of 11,356 subjects with tion Survey.
varicella were identified. Among them, 1080 sub-

n engl j med 356;11 www.nejm.org march 15, 2007 1123

Downloaded from www.nejm.org on July 2, 2008 . Copyright 2007 Massachusetts Medical Society. All rights reserved.
 The n e w e ng l a n d j o u r na l of m e dic i n e

A
450

400
2004 1995
350
No. of Subjects with Varicella

300

250

200

150

100

50

0
<1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
Age (yr)

B
50
No. of Subjects with Varicella

40 Vaccinated

30

Unvaccinated
20

10

0
<1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
Age (yr)

Figure 2. Age Distribution of Subjects with Varicella in 1995 and 2004 and According to Vaccination Status in 2004.
Panel A shows a shift in age distribution over the decade from 1995 to 2004. Panel B shows the shift in age distribution according to
vaccination status in 2004.
ICM
AUTHOR: Chaves RETAKE 1st
FIGURE: 2 of 3 2nd
REG F
3rd
of age or younger accounted
CASE for only 30% of all to vaccination Revised status. The frequency of moderate-
Line 4-C
cases of varicella in the surveillance
EMail
ARTIST: tv area. Among to-severeSIZEdisease increased with increasing age
H/T H/T 39p6
vaccinated children, disease
Enon frequency peaked Combobe- regardless of vaccination status. Moderate-to-
tween the ages of 6 and 9 years, AUTHOR,
whereasPLEASE
among NOTE:severe disease among vaccinated subjects increased
unvaccinated children, theFigure
peakhas been redrawn and type has been reset.
occurred between in frequency from 22% among children between
Please check carefully.
the ages of 9 and 12 years (Fig. 2B). the ages of 1 and 7 years to 44% among those 13
JOB: 35611 years
ISSUE: of age or older (P<0.001 by the chi-square
03-15-07
Severity of Disease test for trend) (Table 1). Among vaccinees, the
In univariate analysis, we assessed various factors frequency of moderate-to-severe disease increased
associated with the severity of varicella according from 18% in the period from 1995 to 1998 to

1124 n engl j med 356;11 www.nejm.org march 15, 2007

Downloaded from www.nejm.org on July 2, 2008 . Copyright 2007 Massachusetts Medical Society. All rights reserved.
 Loss of Vaccine-induced Immunity to Varicella

Table 1. Factors Associated with the Severity of Varicella According to Vaccination Status.*

Variable Moderate-to-Severe Varicella

Vaccinated Subjects P Value Unvaccinated Subjects P Value
no./total no. (%) no./total no. (%)
Age at disease onset
<1 yr 0 <0.001 334/567 (59) <0.001
17 yr 166/771 (22) 4204/6753 (62)
812 yr 104/281 (37) 1335/1946 (69)
13 yr 12/27 (44) 764/1004 (76)
Period at risk
19951998 28/156 (18) <0.001 5379/8336 (65) 0.008
19992000 71/327 (22) 881/1399 (63)
20012004 183/596 (31) 377/535 (70)
Reporting source
Parents 101/467 (22) 0.01 4694/7310 (64) 0.03
Schools 5/17 (29) 126/215 (59)
Health care providers 176/595 (30) 1815/2742 (66)
Age at vaccination
12 yr 154/667 (23) 0.004 NA
35 yr 88/304 (29)
6 yr 40/108 (37)
Time since vaccination NA
<5 yr 172/744 (23) <0.001
5 yr 110/335 (33)

* NA denotes not applicable.

All P values are for comparisons with the last category in each variable and were calculated by the chi-square test for trend.

31% in the period from 2001 to 2004 (P<0.001). disease among both vaccinated and unvaccinated
Among unvaccinated subjects, the increased per- children was associated with the reporting source.
centage of subjects with moderate-to-severe dis- Health care providers were somewhat more like
ease was noticeable only during the period from ly to report cases of moderate-to-severe disease
2001 to 2004. Vaccinated children with 50 or more than were parents (Table 1).
lesions were twice as likely to have complications The final logistic-regression model for the un
such as pneumonia, ataxia, and skin superinfec- vaccinated group included the subjects age at the
tion as were those with fewer than 50 lesions onset of disease, the year of disease onset (calen-
(P=0.03 by Fishers exact test) (data not shown). dar year), and the reporting source. Among un-
When assessed according to the time since vaccinated subjects, the age at disease onset was
vaccination, the frequency of moderate-to-severe the only independent predictor of disease se
disease among vaccinated children increased 1.4 verity (Table 2). As compared with infants, sub-
times among those who had been vaccinated 5 or jects who were 13 years of age or older were 2.2
more years previously, as compared with those times as likely to have moderate-to-severe disease
who had been vaccinated less than 5 years previ- (P<0.001).
ously (33% vs. 23%, P<0.001) (Table 1). Vaccina- In the vaccinated group, the time since vacci-
tion at the age of 6 years or older was also associ- nation, the age at vaccination, and the age at dis
ated with moderate-to-severe disease, as compared ease onset were identified as collinear predictors
with vaccination during the first or second year for severity and could not be simultaneously in-
(P=0.004). The frequency of moderate-to-severe cluded in the logistic-regression model. There-

n engl j med 356;11 www.nejm.org march 15, 2007 1125

Downloaded from www.nejm.org on July 2, 2008 . Copyright 2007 Massachusetts Medical Society. All rights reserved.
 The n e w e ng l a n d j o u r na l of m e dic i n e

Overall, 71% of breakthrough cases were con-

Table 2. Multivariate Analysis of Factors Associated with the Severity
of Varicella among Unvaccinated Children.* firmed by laboratory analysis or diagnosed by a
physician, although this factor varied according
Variable Adjusted Odds Ratio 95% CI P Value to the age group. When we restricted our analy-
Period of disease onset sis of the severity of breakthrough varicella to the
20012004 1.00 time from vaccination to laboratory confirmation
19992000 0.81 0.651.01 0.06 or a physicians diagnosis, the findings were simi-
lar. However, these findings were no longer sta-
19951998 0.91 0.741.11 0.36
tistically significant owing to smaller numbers.
Age at onset Among 209 vaccinated children between the ages
<1 yr 1.00 of 8 and 12 years, subjects who had been vacci-
17 yr 1.16 0.971.38 0.09 nated 5 years or more years previously were 2.0
812 yr 1.53 1.261.85 <0.001 times as likely to have moderate-to-severe break-
through disease (after adjustment for the age at
13 yr 2.20 1.762.74 <0.001
vaccination and the calendar year) as were chil-
Reporting source dren who had been vaccinated less than 5 years
Health care providers 1.00 previously (95% confidence interval [CI], 0.9 to
School 0.96 0.871.06 0.45 4.7; P=0.09).
Parents 0.78 0.561.04 0.09
Annual Rates of Breakthrough Disease
* Odds ratios are for moderate-to-severe disease; the odds ratio for each vari The Poisson regression model showed that among
able has been adjusted for the other variables. P values and 95% confidence children who had been vaccinated between 12
intervals (CIs) were calculated with the use of the Wald chi-square test.
months and 12 years of age, the annual rates of
breakthrough varicella increased with the time
fore, in order to control for the effect of age at since vaccination, even after adjustment for the
the onset of disease on the severity of disease age at disease onset and the calendar year as po-
among vaccinated children, we included only vac- tential confounders for changes in the likelihood
cinees within the narrow age band of 8 to 12 of exposure. The rate of breakthrough varicella
years. This age group had the greatest variability increased significantly with each year after vac-
in the age at vaccination, which allowed for the cination, from 1.6 cases per 1000 person-years
examination of the independent effect of the time (95% CI, 1.2 to 2.0) within the first year to 9.0
since vaccination. Moreover, among subjects be- per 1000 person-years (95% CI, 6.9 to 11.7) at
tween the ages of 8 and 12 years, no significant 5 years and 20.4 per 1000 person-years (95% CI,
difference in the percentage of subjects with 14.1 to 29.6) at 8 years. The rate of breakthrough
moderate-to-severe disease according to the year varicella 9 years after vaccination was 58.2 per
of age was found (P=0.40). 1000 person-years (95% CI, 36.0 to 94.0), but ow-
The logistic-regression model for vaccinated ing to a small number of subjects, the 95% CI was
children between the ages of 8 and 12 years re- very wide, even though it did not overlap with the
tained the effects of the time since vaccination, CI estimated for previous years (Fig. 3).
the age at vaccination, and the calendar year as
predictors of disease severity. The calendar year Dis cus sion
was used as a continuous variable and represent
ed the variation of background rates of varicella Our analysis provides evidence that the protection
since the introduction of the vaccination program. afforded by one dose of varicella vaccine in chil-
Among vaccinated children between the ages of dren may wane with time. We found that both
8 and 12 years at disease onset, after adjustment the severity and incidence of breakthrough disease
for the age at vaccination and the calendar year, among vaccinees increased with the time since
subjects who had been vaccinated 5 or more years vaccination. Children between the ages of 8 and
previously were 2.6 times as likely to have mod- 12 years who had been vaccinated 5 years or more
erate-to-severe disease as were those who had been previously were two times as likely to have mod-
vaccinated less than 5 years previously (P=0.01) erate-to-severe breakthrough disease as were those
(Table 3). who had been vaccinated less than 5 years previ-

1126 n engl j med 356;11 www.nejm.org march 15, 2007

Downloaded from www.nejm.org on July 2, 2008 . Copyright 2007 Massachusetts Medical Society. All rights reserved.
 Loss of Vaccine-induced Immunity to Varicella

ously, regardless of the age at disease onset, the

Table 3. Multivariate Analysis of Factors Associated with the Severity
age at vaccination, and the calendar year when of Varicella among Vaccinated Children between the Ages of 8 and 12 Years.*
the disease developed. Furthermore, incidence
rates of breakthrough varicella increased more Variable Adjusted Odds Ratio 95% CI P Value
than 12 times from the first year of vaccination Time since vaccination
to year 8 after vaccination, after adjustment for 5 yr 2.60 1.205.80 0.01
age and calendar year, factors that were consid- <5 yr 1.00
ered to be surrogates for changes in exposure. Age at vaccination
This increase in breakthrough disease was ob-
12 yr 0.42 0.151.15 0.13
served in the context of a substantial decline in
35 yr 0.56 0.251.24 0.56
varicella.
The effect of the varicella vaccination program 6 yr 1.00
in the Antelope Valley surveillance area has been Calendar year 0.96 0.811.31 0.61
documented previously.1 From 1995 to 2000, there
* Odds ratios are for moderate-to-severe disease; the odds ratio for each vari-
was an overall decrease in disease incidence of able has been adjusted for the other variables. P values and 95% confidence
71%, with a documented reduction in disease bur intervals (CIs) were calculated with the use of the Wald chi-square test.
den among all age groups findings that are
consistent with herd immunity. Our data show a
decline of 85% in varicella from 1995 to 2003 an efficacy of varicella vaccine in the range of 44
2004. The reduction in exposure to VZV increases to 100%.6-10,13-15,23 These studies suggested that
the risk that the remaining unvaccinated children an increased risk of breakthrough varicella may
and adolescents may be susceptible to varicella and be associated with a history of asthma14 or ec-
its complications when they reach adulthood.11,18 zema,10 vaccination with varicella vaccine within
Implementation of a vaccination policy that re- 28 days after the administration of live attenuated
quires documentation of varicella vaccination or measlesmumpsrubella vaccine,24 prescription
other evidence of immunity for entry to middle of oral corticosteroids 3 months before the onset
school, high school, and college is critical for the of varicella,24 and an early age at vaccination
protection of this susceptible cohort.8,19 (variously defined).9,10,13,24 However, these asso-
The reduced circulation of VZV in the study ciations were not consistently reproduced in all
area or an inadequate initial response to vacci- studies. Furthermore, such factors are unlikely to
nation may have contributed to the waning of have been confounders in our study, since they
vaccine-induced immunity to varicella.11,18,20 An should not bear an association with the time since
assessment of the duration of protection afford vaccination. As compared with previous stud-
ed by one dose of the varicella vaccine adminis- ies,6,9,10,15,22 our study examined the indepen-
tered to children under the age of 13 years be- dent effect of the time since vaccination on the
tween December 1991 and January 1993 showed risk of breakthrough varicella. Our data con-
an increase in the geometric mean titer of vari- tained a sufficient number of subjects who had
cella IgG antibody during a 10-year follow-up been vaccinated at different ages and for whom
period.21 The most likely explanation for this in- the interval between vaccination and the onset of
crease was an anamnestic immune response due disease varied, which allowed us to control for
to exposure to wild-type VZV, since varicella was these factors.
still common during that period. Several limitations of our study should be con-
These data suggest a steady decline over a sidered when interpreting the results. Owing to
period of years in disease protection afforded by constraints with respect to the number of sub-
a single dose of the varicella vaccine in the con- jects in our study, our definition of disease sever-
text of diminished circulation of wild-type virus. ity was limited to fewer than 50 lesions or 50 or
In contrast, a casecontrol study showed a 13% more lesions and did not include complications or
decline in vaccine effectiveness during the first hospitalization. Nonetheless, complications were
year of vaccination, followed by stable levels of more likely to develop in vaccinated children with
protection from 2 to 8 years after vaccination.22 50 or more lesions than in children with fewer
Numerous studies of postlicensure effectiveness than 50 lesions. Since breakthrough disease with
that were performed during the vaccine era showed few lesions may be mild, such cases may have

n engl j med 356;11 www.nejm.org march 15, 2007 1127

Downloaded from www.nejm.org on July 2, 2008 . Copyright 2007 Massachusetts Medical Society. All rights reserved.
 The n e w e ng l a n d j o u r na l of m e dic i n e

nity to varicella. Clinical studies have suggested

105
that a second dose of varicella vaccine could pro-
90 vide increased protection against disease by in-
(cases/1000 person-yr)
Breakthrough Varicella

75 creasing the proportion of children with protec-

60 tive antibody titers and an improved cellular
45
immune response.21,25-27 The findings from our
study and other scientific evidence were taken
30
into account when, in June 2006, the Advisory
15
Committee on Immunization Practices adopted a
0 recommendation that children between the ages
0 1 2 3 4 5 6 7 8 9 10
of 4 and 6 years receive a second dose of varicella
Years since Vaccination
vaccine. The panel also recommended that a sec-
Figure 3. Adjusted Rates of Breakthrough Varicella among Children Vaccinat- ond catch-up dose of varicella vaccine be given to
ed between the Ages of 12 Months and 12 Years, According to the Year after children, adolescents, and adults who previously
Vaccination.
had received one dose.19 No long-term data are
Rates of disease were adjusted for the age at disease onset and the calendar
year with the use of a Poisson regression model. All 95% confidence intervals,
available on the duration of immunity afforded by
which are indicated by I bars, were the second dose of vaccine.
ICM
AUTHOR: Chavescalculated with the use of the
RETAKE 1st Wald
chi-square test. 2nd Supported by a cooperative agreement between the CDC and
FIGURE: 3 of 3
REG F
3rd the Los Angeles Department of Health Services for Active Varicella
CASE Revised Surveillance and Epidemiologic Studies.
EMail Line 4-C SIZE After completion of this study, Dr. Guris became an employee
ARTIST: tv
Enon been underreported,
H/T which
H/T could22p3have led to an of Merck. Dr. Mascola reports receiving lecture fees from Merck
Combo and serving on the companys paid advisory board. No other po-
underestimation of the rates of breakthrough
AUTHOR, PLEASE NOTE: tential conflict of interest relevant to this article was reported.
disease.
Figure A possible
has been redrawn andincrease in reset.
type has been rates of ascertain- The findings and conclusions in this report are those of the
ment of Please
breakthrough
check carefully.varicella in recent years authors and do not necessarily represent the views of the CDC.
due to increased awarenessISSUE: would be unlikely to We thank all the surveillance sites in the Antelope Valley
JOB: 35611 03-15-07 community for their continued reporting and involvement in the
bias our results for either rates or severity of dis- Varicella Active Surveillance Project; Tina Carbajal, Teresa Mau-
ease, since our analysis is based on the time since pin, and Jorge Garcia-Herrera for their thorough investigation of
vaccination and not on the calendar year. varicella cases and coordination of the VZV disease surveillance
activities in Antelope Valley; and Dr. Marc-Alain Widdowson and
In summary, our study provides clinical evi- Dr. Umesh Parashar for their critical reading of the manuscript
dence of the waning of vaccine-acquired immu- and their editorial assistance.

References
1. Seward JF, Watson BM, Peterson CL, EM, Hedberg K, Cieslak PR. Chickenpox vaccine and likelihood of long term break-
et al. Varicella disease after introduction outbreak in a highly vaccinated school through infection. Pediatr Infect Dis J
of varicella vaccine in the United States, population. Pediatrics 2004;113:455-9. 2002;21:337-42.
1995-2000. JAMA 2002;287:606-11. 7. Outbreak of varicella among vaccinat 13. Vzquez M, LaRussa P, Gershon A,
2. Mullooly JP, Maher JE, Drew L, Schuler ed children Michigan, 2003. MMWR Steinberg S, Freudigman K, Shapiro E.
R, Hu W. Evaluation of the impact of an Morb Mortal Wkly Rep 2004;53:389-92. The effectiveness of the varicella vaccine
HMOs varicella vaccination program on 8. Lopez AS, Guris D, Zimmerman L, et in clinical practice. N Engl J Med 2001;
incidence of varicella. Vaccine 2004;22: al. One dose of varicella vaccine does not 344:955-6.
1480-5. prevent school outbreaks: is it time for a 14. Izurieta HS, Strebel PM, Blake PA.
3. Nguyen HQ, Jumaan AO, Seward JF. second dose? Pediatrics 2006;117:e1070- Postlicensure effectiveness of varicella vac-
Decline in mortality due to varicella after e1077. cine during an outbreak in a child care
implementation of varicella vaccination in 9. Galil K, Lee B, Strine T, et al. Out- center. JAMA 1997;278:1495-9.
the United States. N Engl J Med 2005;352: break of varicella at a day-care center de- 15. Clements DA, Moreira SP, Coplan PM,
450-8. spite vaccination. N Engl J Med 2002;347: Bland CL, Walter EB. Postlicensure study
4. Zhou F, Harpaz R, Jumaan AO, Wins 1909-15. of varicella vaccine effectiveness in a day-
ton CA, Shefer A. Impact of varicella vac- 10. Haddad MB, Hill MB, Pavia AT, et al. care setting. Pediatr Infect Dis J 1999;18:
cination on health care utilization. JAMA Vaccine effectiveness during a varicella 1047-50.
2005;294:797-802. outbreak among schoolchildren: Utah, 16. Ampofo K, Saiman L, LaRussa P, Stein
5. Davis MM, Patel MS, Gebremariam A. 2002-2003. Pediatrics 2005;115:1488-93. berg S, Annunziato P, Gershon A. Persis-
Decline in varicella-related hospitalizations 11. Arvin AM. Immune responses to vari- tence of immunity to live attenuated vari-
and expenditures for children and adults cella-zoster virus. Infect Dis Clin North cella vaccine in healthy adults. Clin Infect
after introduction of varicella vaccine in Am 1996;10:529-70. Dis 2002;34:774-9.
the United States. Pediatrics 2004;114: 12. Li S, Chan IS, Matthews H, et al. In- 17. Maldonado G, Greenland S. Simula-
786-92. verse relationship between six week post- tion study of confounder-selection strate-
6. Tugwell BD, Lee LE, Gillette H, Lorber vaccination varicella antibody response to gies. Am J Epidemiol 1993;138:923-36.

1128 n engl j med 356;11 www.nejm.org march 15, 2007

Downloaded from www.nejm.org on July 2, 2008 . Copyright 2007 Massachusetts Medical Society. All rights reserved.
 Loss of Vaccine-induced Immunity to Varicella

18. Brisson M, Edmunds WJ, Gay NJ. Vari cella vaccine. Pediatr Infect Dis J 2004; Safety and immunogenicity of one vs. two
cella vaccination: impact of vaccine effi- 23:132-7. injections of Oka/Merck varicella vaccine
cacy on the epidemiology of VZV. J Med 22. Vazquez M, LaRussa PS, Gershon AA, in healthy children. Pediatr Infect Dis J
Virol 2003;70:Suppl 1:S31-S37. et al. Effectiveness over time of varicella 1996;15:49-54.
19. Advisory Committee on Immunization vaccine. JAMA 2004;291:851-5. 26. Watson B, Rothstein E, Bernstein H,
Practices. ACIP provisional recommenda- 23. Lee BR, Feaver SL, Miller CA, Hedberg et al. Safety and cellular and humoral
tions: last updated on December 20, 2006. CW, Ehresmann KR. An elementary school immune responses of a booster dose of
(Accessed February 16, 2007, at http://www. outbreak of varicella attributed to vaccine varicella vaccine 6 years after primary
cdc.gov/nip/recs/provisional_recs/default. failure: policy implications. J Infect Dis immunization. J Infect Dis 1995;172:217-
htm.) 2004;190:477-83. 9.
20. Krause PR. Evidence for frequent re- 24. Verstraeten T, Jumaan AO, Mullooly JP, 27. Watson B, Boardman C, Laufer D, et al.
activation of the Oka varicella vaccine et al. A retrospective cohort study of the Humoral and cell-mediated immune re-
strain in healthy vaccinees. Arch Virol association of varicella vaccine failure with sponses in healthy children after one or
Suppl 2001;17:7-15. asthma, steroid use, age at vaccination, two doses of varicella vaccine. Clin Infect
21. Kuter B, Matthews H, Shinefield H, et and measles-mumps-rubella vaccination. Dis 1995;20:316-9.
al. Ten year follow-up of healthy children Pediatrics 2003;112:e98-e103. Copyright 2007 Massachusetts Medical Society.
who received one or two injections of vari- 25. Ngai AL, Staehle BO, Kuter BJ, et al.

n engl j med 356;11 www.nejm.org march 15, 2007 1129

Downloaded from www.nejm.org on July 2, 2008 . Copyright 2007 Massachusetts Medical Society. All rights reserved.