journal of medicine
The
established in 1812
february 6, 2014
A BS T R AC T
Background
International postlicensure studies have identified an increased risk of intussusception after vaccination with the second-generation rotavirus vaccines RotaTeq (RV5,
a pentavalent vaccine) and Rotarix (RV1, a monovalent vaccine). We studied this
association among infants in the United States.
Methods
The study included data from infants 5.0 to 36.9 weeks of age who were enrolled in
three U.S. health plans that participate in the Mini-Sentinel program sponsored by
the Food and Drug Administration. Potential cases of intussusception and vaccine
exposures from 2004 through mid-2011 were identified through procedural and
diagnostic codes. Medical records were reviewed to confirm the occurrence of intussusception and the status with respect to rotavirus vaccination. The primary analysis used a self-controlled risk-interval design that included only vaccinated children.
The secondary analysis used a cohort design that included exposed and unexposed
person-time.
Results
The analyses included 507,874 first doses and 1,277,556 total doses of RV5 and
53,638 first doses and 103,098 total doses of RV1. The statistical power for the
analysis of RV1 was lower than that for the analysis of RV5. The number of excess
cases of intussusception per 100,000 recipients of the first dose of RV5 was significantly elevated, both in the primary analysis (attributable risk, 1.1 [95% confidence
interval, 0.3 to 2.7] for the 7-day risk window and 1.5 [95% CI, 0.2 to 3.2] for the
21-day risk window) and in the secondary analysis (attributable risk, 1.2 [95% CI,
0.2 to 3.2] for the 21-day risk window). No significant increase in risk was seen
after dose 2 or 3. The results with respect to the primary analysis of RV1 were not
significant, but the secondary analysis showed a significant risk after dose 2.
Conclusions
RV5 was associated with approximately 1.5 (95% CI, 0.2 to 3.2) excess cases of intussusception per 100,000 recipients of the first dose. The secondary analysis of
RV1 suggested a potential risk, although the study of RV1 was underpowered. These
risks must be considered in light of the demonstrated benefits of rotavirus vaccination. (Funded by the Food and Drug Administration.)
503
The
n e w e ng l a n d j o u r na l
of
m e dic i n e
mended ages for vaccination plus adequate follow-up time) who were members of an Aetna,
HealthCore, or Humana health plan between
January 2004 and September 2011. Using a distributed database system,19,20 each of these data
partners provided at least 3 consecutive years of
claims and other administrative data during this
period, resulting in approximately 613,000 infant-years observed.
Study Design
504
Outcomes
Potential cases of intussusception during all person-time from 5.0 to 36.9 weeks of age, irrespective of immunization status, were identified in
administrative data on the basis of any of three
codes in either the inpatient or emergency department setting: International Classification of Diseases,
Ninth Revision (ICD-9) code 560.0 (intussusception) or 543.9 (other and unspecified diseases of
the appendix, including intussusception) or CPT
code 74283 (therapeutic enema, contrast or air).
Only first-ever diagnoses were included.
Case status was determined by adjudication
that was based on a review of deidentified fulltext medical records. Cases were excluded if no
intussusception had been seen or an alternative
diagnosis had been made after surgery or air or
liquid-contrast enema. Each remaining potential
case was independently reviewed by one or more
adjudicators; the two main adjudicators were
pediatricians, and the third was an internist.
Adjudicators were unaware of the infants vaccination history and were instructed to classify
intussusception cases with the use of Brighton
Collaboration criteria.26 Level 1 cases were cases
of intussusception confirmed on the basis of
surgical, radiologic, or autopsy criteria and were
used in the primary analyses. Classification
rules were refined for Brighton level 2 cases,
which are defined on the basis of criteria representing less direct evidence of intussusception,
with further differentiation into level 2A cases
(those considered to be possible intussusception
on the basis of positive, equivocal, or discordant
results on abdominal radiography [ultrasonography, plain radiography, or computed tomography])
and level 2B cases (those that met level 2 criteria
but were clearly not intussusception as evidenced
by normal radiologic results). Level 2A cases
combined with inconclusive cases, for which
the record stated a diagnosis of intussusception
but contained insufficient evidence to allow case
classification, were classified as possible in
tussusception and were included in sensitivity
analyses. Level 3 is the lowest level of diagnostic
Statistical Analysis
505
The
n e w e ng l a n d j o u r na l
Since the risk of intussusception varies greatly by age in weeks, the attributable risk may vary
according to the age of the child at the time of
vaccination. We present the average attributable
risk on the basis of the observed age distribution
of the vaccinated children. The attributable risk
was calculated as the number of excess cases of
intussusception per 100,000 doses administered,
according to the formula 100,000no. of cases
in the risk window[1(1relative risk)][no.
of vaccine dosesC], where C is the proportion of
potential cases for which we were able to conduct
a chart review. By including C in the equation,
we adjusted the attributable risk for the missing
charts. We calculated the 95% confidence intervals using the methods of Krishnamoorthy
and Lee27 (see the Supplementary Appendix, available with the full text of this article at NEJM.org).
We emphasize attributable risk over relativerisk estimates in the results, because attributable risks are more relevant from clinical and
public health perspectives and are less sensitive
to differences in the lengths of risk intervals. In
comparing our risk estimates with those of
other studies, we sometimes use relative risks,
either because a study with which we are comparing our results reported only relative risks
or because comparing attributable risks across
countries with different background rates of intussusception can be misleading.
To ensure that our findings were robust, we
used alternative methods for age adjustment in
post hoc analyses. For the SCRI design, we used
the quadratic risk function from the unexposed
cohort person-time as the alternative, and for
the cohort design, we used the rates from Tate
et al.25 In addition, we conducted a series of
sensitivity analyses, which are described in the
Supplementary Appendix.
To identify clusters of intussusception onsets
within the 1-to-42-day period after rotavirus vaccination, we used the temporal scan statistic,28
a self-controlled design, with only vaccinated
children who had intussusception 1 to 42 days
after exposure included in the analysis. We evaluated all potential risk windows starting 1 to 14
days after vaccination and ending 1 to 21 days
after vaccination, with adjustment for the multiple testing inherent in the 203 intervals considered. The test statistic is the maximum likelihood
obtained among these intervals. To adjust for
age, we used the HCUP rates from Tate et al.25
to randomize the day of age at the onset of in506
of
m e dic i n e
R e sult s
Vaccine Doses Administered
RV5
In the SCRI analysis, the attributable risk of intussusception after dose 1 was significantly elevated
for both risk windows (1.1 [95% confidence interval {CI}, 0.3 to 2.7] for the 7-day risk window,
and 1.5 [95% CI, 0.2 to 3.2] for the 21-day risk
window). No significant increase in risk was seen
after dose 2 or dose 3. In the cohort analysis
(with a 21-day risk interval), there was a significant attributable risk after dose 1 (1.2 [95% CI,
0.2 to 3.2]) but not after the other doses (Table 1).
RV1
After dose 1, there was just one case of intussusception in the risk interval, and there were no
Table 1. Case Counts and Risk Estimates for Confirmed Intussusception after RV5.*
Dose, Type
AgeDays after No. of Cases No. of Cases
of Analysis,
Adjustment Vaccination
in Risk
in Control
and Design
Method in Risk Window Window
Window
Dose 1
Prespecified
SCRI
Tate
1 to 7
5
3
SCRI
Tate
1 to 21
8
3
Cohort
PRISM
1 to 21
8
97
Post hoc
SCRI
PRISM
1 to 7
5
3
SCRI
PRISM
1 to 21
8
3
Cohort
Tate
1 to 21
8
97
Dose 2
Prespecified
SCRI
Tate
1 to 7
3
6
SCRI
Tate
1 to 21
5
6
Cohort
PRISM
1 to 21
5
97
Post hoc
SCRI
PRISM
1 to 7
3
6
SCRI
PRISM
1 to 21
5
6
Cohort
Tate
1 to 21
5
97
Dose 3
Prespecified
SCRI
Tate
1 to 7
3
4
SCRI
Tate
1 to 21
4
4
Cohort
PRISM
1 to 21
5
97
Post hoc
SCRI
PRISM
1 to 7
3
4
SCRI
PRISM
1 to 21
4
4
Cohort
Tate
1 to 21
5
97
All doses**
Prespecified
SCRI
Tate
1 to 7
11
13
SCRI
Tate
1 to 21
17
13
Cohort
PRISM
1 to 21
18
97
Post hoc
SCRI
PRISM
1 to 7
11
13
SCRI
PRISM
1 to 21
17
13
Cohort
Tate
1 to 21
18
97
Relative Risk
(95% CI)
Attributable Risk/
100,000 Doses
(95% CI)
256,000 (52,000 to )
(57,000 to )
(57,000 to )
258,000 (52,000 to )
(57,000 to )
(62,000 to )
159,000 (38,000 to )
(47,000 to )
(43,000 to )
152,000 (38,000 to )
10,402,000 (46,000 to )
(42,000 to )
* Cases of intussusception were adjudicated with the use of Brighton Collaboration criteria,26 with level 1 cases considered as confirmed
cases. We used two study designs: a self-controlled risk-interval (SCRI) design and a cohort design.
In prespecified analyses, we adjusted for age in the SCRI design using age-specific background rates extracted by Tate et al.25 from the U.S.
hospital-discharge data of the Healthcare Cost and Utilization Project for 11 years during which no rotavirus vaccine was used, and we adjusted for age in the cohort design using a quadratic risk function drawn from the unexposed person-time. In addition to the prespecified analyses, we performed post hoc analyses in which we used alternative methods for age adjustment to ensure that the findings were robust; for
the SCRI design, we used the quadratic risk function as the alternative, and for the cohort design, we used the rates from Tate et al.25
The control window for the SCRI design was 22 to 42 days after vaccination; the control period for the cohort design was all person-time
except for 0 to 21 days after any rotavirus vaccination (194,520,053 person-days).
A correction factor was incorporated for cases for which medical charts were missing (which accounted for 22% of the total potential cases ascertained). The number of doses resulting in one excess case (last column) is obtained by taking the reciprocal of the attributable
risk expressed in terms of excess cases per 100,000 doses (penultimate column). Dashes are substituted for negative numbers of doses,
since a negative number of doses resulting in an excess case would not be interpretable. In the confidence intervals for number of doses,
the first (lower) number represents the highest risk, and the second (higher or blank) number represents the lowest risk.
The numbers of exposed person-days were 10,931,848 for dose 1; 9,263,327 for dose 2; 6,889,428 for dose 3; and 27,094,157 for all doses.
One of these cases was excluded from SCRI analysis because the age at vaccination plus the required 42-day follow-up period exceeded
the cutoff age for chart review.
** Relative risk estimates for the analyses of all doses represent a blend of the risks of the component doses. Attributable risk estimates for
the analyses of all doses are per 100,000 doses, so the total attributable risk for 100,000 fully vaccinated infants is larger.
507
The
n e w e ng l a n d j o u r na l
of
m e dic i n e
Table 2. Case Counts and Risk Estimates for Confirmed Intussusception after RV1.*
Dose, Type
of Analysis,
and Design
AgeDays after
No. of Cases No. of Cases
Adjustment Vaccination
in Risk
in Control
Method in Risk Window Window
Window
Relative Risk
(95% CI)
Attributable Risk/
100,000 Doses
(95% CI)
2.4
2.4
Dose 1
Prespecified
SCRI
SCRI
Cohort
Tate
1 to 7
Tate
1 to 21
PRISM
1 to 21
97
42,000
42,000
63,000 (10,000 to )
Post hoc
SCRI
PRISM
1 to 7
2.4
SCRI
PRISM
1 to 21
2.4
Tate
1 to 21
97
61,000 (6000 to )
Tate
1 to 7
23,000 (6000 to )
Tate
1 to 21
PRISM
1 to 21
97
Cohort
42,000
42,000
Dose 2
Prespecified
SCRI
SCRI
Cohort
27,000 (5000 to )
Post hoc
SCRI
PRISM
1 to 7
SCRI
PRISM
1 to 21
27,000 (5000 to )
Tate
1 to 21
97
Tate
1 to 7
Tate
1 to 21
35,000 (10,000 to )
PRISM
1 to 21
97
Cohort
All doses
Prespecified
SCRI
SCRI
Cohort
Post hoc
SCRI
PRISM
1 to 7
SCRI
PRISM
1 to 21
35,000 (10,000 to )
Tate
1 to 21
97
Cohort
* The criteria for adjudication of cases, the methods of adjustment for age, and the control windows for the SCRI design and the cohort design were the same as for RV5.
A correction factor was incorporated for cases for which medical charts were missing (which accounted for 22% of the total potential cases
ascertained). The number of doses resulting in one excess case (last column) is obtained by taking the reciprocal of the attributable risk expressed in terms of excess cases per 100,000 doses (penultimate column). Dashes are substituted for negative numbers of doses, since a
negative number of doses resulting in an excess case would not be interpretable. In the confidence intervals for number of doses, the first
(lower) number represents the highest risk, and the second (higher or blank) number represents the lowest risk.
The numbers of exposed person-days were 1,178,772 for dose 1; 917,754 for dose 2; and 2,242,833 for all doses.
Relative risk estimates for the analyses of all doses represent a blend of the risks of the component doses. Attributable risk estimates for the
analyses of all doses are per 100,000 doses, so the total attributable risk for 100,000 fully vaccinated infants is larger.
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0.0
SCRI,
original age
adjustment
using Tate et al.
risk curve
SCRI,
age adjustment
using study
population
risk function
SCRI,
original age
adjustment
using Tate et al.
risk curve
SCRI,
age adjustment
using study
population
risk function
Cohort,
original age
adjustment
using study
population
Cohort,
age adjustment
using Tate et al.
risk curve
Figure 1. Attributable Risk of Intussusception after the First Dose of RotaTeq (RV5) Rotavirus Vaccine.
The attributable risk of intussusception after dose 1 of the RV5 vaccine, shown as the number of excess cases of intussusception per 100,000 recipients, was calculated for two study designs a self-controlled risk-interval (SCRI)
design and a cohort design with the original age-adjustment method (based on the rates from Tate et al.25 in the
SCRI design and the quadratic risk function from the unexposed person-time in the cohort design) and an alternative age-adjustment method (based on the quadratic risk function from the unexposed cohort person-time in the
SCRI design and the rates from Tate et al.25 in the cohort design). For dose 1 of RV5, age adjustment with the use of
the quadratic risk function obtained from the study population results in only slightly lower attributable risks than
age adjustment with the use of hospital-discharge data from Tate et al.25
vaccine, regardless of study design, risk window, vaccinees. Subsequent doses of RV5 were not associated with a significant increase in the risk of
or age-adjustment method.
intussusception. However, an increased risk associated with those doses cannot be ruled out,
Clusters of Intussusception Onset
In the analyses of dose 1 and of all doses of RV5, given the overlapping confidence intervals of the
the temporal scan statistic showed a significant risk estimates for doses 1, 2, and 3. These risks
cluster of onset of intussusception 3 to 7 days must be considered in the context of the benefits
after vaccination (P=0.008 for dose 1; P=0.004 of vaccination, which include the prevention of a
for all doses). There was only a single case of in- projected 53,444 hospitalizations (95% CI, 37,622
tussusception after dose 1 of RV1; therefore, to 72,882) in a U.S. birth cohort of 4.3 million
there were insufficient data for the analysis of children.30
Our results with respect to RV5 are similar to
clusters of onset after dose 1. For all doses of
RV1, there was a significant cluster on day 4 after those of the Vaccine Adverse Event Reporting
System (VAERS), which used an SCRI design to
vaccination (P<0.001) (Fig. 2).
compare the numbers of spontaneously reported
cases in the 3-to-6-day period after vaccination
Discussion
with those in the 0-to-2-day period after vaccinaAfter the first dose of RV5, with a risk window of tion. That study showed an attributable risk of
21 days after vaccination, we found a significant 0.74 excess cases (95% CI, 0.24 to 1.71) per
increase in the risk of intussusception, with ap- 100,000 recipients of the first dose of vaccine
proximately 1.5 excess cases per 100,000 recipi- and no significant results for the other doses.31
ents of the vaccine, which was approximately one In contrast, the population-based Vaccine Safety
tenth the risk with Rotashield.1 The lower bound- Datalink (VSD) has not shown a significant inary of the 95% confidence interval (representing crease in the risk of intussusception after
the higher-risk boundary) was 1 case per 31,000 RV5.14,15 The VSD, which used a cohort design
n engl j med 370;6nejm.orgfebruary 6, 2014
509
The
n e w e ng l a n d j o u r na l
A RV5, Dose 1
No. of Intussusception Cases
Dose 1
5/11 cases
Relative risk, 9.7
P=0.008
2
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41
Dose 3
10/30 cases
Relative risk, 4.5
P=0.004
Dose 2
Dose 1
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41
Dose 2
3/6 cases
Relative risk, 48
P<0.001
Dose 1
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41
Figure 2. Distribution of Intussusception Cases According to Day of Symptom Onset after Vaccination.
The age-adjusted temporal scan statistic showed significant clustering on
days 3 to 7 after the first dose (Panel A) and after all doses (Panel B) of
RV5 and on day 4 after all doses of Rotarix (RV1) (Panel C).
and ICD-9coded visits without chart confirmation, recently reported a relative risk of 2.63
(95% CI, 0.72 to 6.74) for intussusception after
dose 1, for a 7-day risk window32; in contrast,
510
of
m e dic i n e
Wang FT, Wong J, Seeger JD. Postmarketing evaluation of the short-term safety of
the pentavalent rotavirus vaccine. Pediatr
Infect Dis J 2012;31:292-6.
17. Nguyen M, Ball R, Midthun K, Lieu
TA. The Food and Drug Administrations
Post-Licensure Rapid Immunization Safety
Monitoring program: strengthening the
federal vaccine safety enterprise. Pharmacoepidemiol Drug Saf 2012;21:Suppl 1:
291-7.
18. Platt R, Carnahan RM, Brown JS, et al.
The U.S. Food and Drug Administrations
Mini-Sentinel program: status and direction. Pharmacoepidemiol Drug Saf 2012;
21:Suppl 1:1-8.
19. Maro JC, Platt R, Holmes JH, et al. Design of a national distributed health data
network. Ann Intern Med 2009;151:341-4.
20. Curtis LH, Weiner MG, Boudreau DM,
et al. Design considerations, architecture,
and use of the Mini-Sentinel distributed
data system. Pharmacoepidemiol Drug Saf
2012;21:Suppl 1:23-31.
21. Glanz JM, McClure DL, Xu S, et al.
Four different study designs to evaluate
vaccine safety were equally validated with
contrasting limitations. J Clin Epidemiol
2006;59:808-18.
22. McClure DL, Glanz JM, Xu S, Hambidge SJ, Mullooly JP, Baggs J. Comparison of epidemiologic methods for active
surveillance of vaccine safety. Vaccine
2008;26:3341-5.
23. Kramarz P, DeStefano F, Gargiullo
PM, et al. Does influenza vaccination exacerbate asthma? Analysis of a large cohort of children with asthma. Arch Fam
Med 2000;9:617-23.
511
512