Algorithm to assess causality after individual adverse events following immunizations

Article (PDF Available)inVaccine 30(39):5791-8 · April 2012with102 Reads
DOI: 10.1016/j.vaccine.2012.04.005 · Source: PubMed
Abstract
Assessing individual reports of adverse events following immunizations (AEFI) can be challenging. Most published reviews are based on expert opinions, but the methods and logic used to arrive at these opinions are neither well described nor understood by many health care providers and scientists. We developed a standardized algorithm to assist in collecting and interpreting data, and to help assess causality after individual AEFI. Key questions that should be asked during the assessment of AEFI include: Is the diagnosis of the AEFI correct? Does clinical or laboratory evidence exist that supports possible causes for the AEFI other than the vaccine in the affected individual? Is there a known causal association between the AEFI and the vaccine? Is there strong evidence against a causal association? Is there a specific laboratory test implicating the vaccine in the pathogenesis? An algorithm can assist with addressing these questions in a standardized, transparent manner which can be tracked and reassessed if additional information becomes available. Examples in this document illustrate the process of using the algorithm to determine causality. As new epidemiologic and clinical data become available, the algorithm and guidelines will need to be modified. Feedback from users of the algorithm will be invaluable in this process. We hope that this algorithm approach can assist with educational efforts to improve the collection of key information on AEFI and provide a platform for teaching about causality assessment.

Figures

Vaccine
30 (2012) 5791–
5798
Contents
lists
available
at
SciVerse
ScienceDirect
Vaccine
jou
rn
al
h
om
epa
ge:
www.elsevier.com/locate/vaccine
Algorithm
to
assess
causality
after
individual
adverse
events
following
immunizations
Neal
A.
Halsey
a,
,
Kathryn
M.
Edwards
b
,
Cornelia
L.
Dekker
c
,
Nicola
P.
Klein
d
,
Roger
Baxter
d
,
Philip
LaRussa
e
,
Colin
Marchant
f
,
Barbara
Slade
g
,
Claudia
Vellozzi
g
,
the
Causality
Working
Group
of
the
Clinical
Immunization
Safety
Assessment
network
1
a
Institute
for
Vaccine
Safety,
Johns
Hopkins
Bloomberg
School
of
Public
Health,
Baltimore,
MD,
United
States
b
Vanderbilt
Vaccine
Research
Program,
Department
of
Pediatrics,
Vanderbilt
University,
Nashville,
TN,
United
States
c
Department
of
Pediatrics,
Division
of
Pediatric
Infectious
Diseases,
Stanford
University
School
of
Medicine,
Palo
Alto,
CA,
United
States
d
Kaiser
Permanente
Vaccine
Study
Center,
Oakland,
CA,
United
States
e
Department
of
Pediatrics,
Columbia
University,
New
York
City,
NY,
United
States
f
Department
of
Pediatrics,
Boston
Medical
Center,
Boston,
MA,
United
States
g
Immunization
Safety
Office,
Centers
for
Disease
Control
and
Prevention,
United
States
a
r
t
i
c
l
e
i
n
f
o
Article
history:
Received
21
November
2011
Received
in
revised
form
5
March
2012
Accepted
1
April
2012
Available online 14 April 2012
Keywords:
Vaccines
Adverse
events
Vaccine
safety
Causality
Algorithm
a
b
s
t
r
a
c
t
Assessing
individual
reports
of
adverse
events
following
immunizations
(AEFI)
can
be
challenging.
Most
published
reviews
are
based
on
expert
opinions,
but
the
methods
and
logic
used
to
arrive
at
these
opinions
are
neither
well
described
nor
understood
by
many
health
care
providers
and
scientists.
We
developed
a
standardized
algorithm
to
assist
in
collecting
and
interpreting
data,
and
to
help
assess
causality
after
individual
AEFI.
Key
questions
that
should
be
asked
during
the
assessment
of
AEFI
include:
Is
the
diagnosis
of
the
AEFI
correct?
Does
clinical
or
laboratory
evidence
exist
that
supports
possible
causes
for
the
AEFI
other
than
the
vaccine
in
the
affected
individual?
Is
there
a
known
causal
association
between
the
AEFI
and
the
vaccine?
Is
there
strong
evidence
against
a
causal
association?
Is
there
a
specific
laboratory
test
implicating
the
vaccine
in
the
pathogenesis?
An
algorithm
can
assist
with
addressing
these
questions
in
a
standardized,
transparent
manner
which
can
be
tracked
and
reassessed
if
additional
information
becomes
available.
Examples
in
this
document
illustrate
the
process
of
using
the
algorithm
to
determine
causality.
As
new
epidemiologic
and
clinical
data
become
available,
the
algorithm
and
guidelines
will
need
to
be
modified.
Feedback
from
users
of
the
algorithm
will
be
invaluable
in
this
process.
We
hope
that
this
algorithm
approach
can
assist
with
educational
efforts
to
improve
the
collection
of
key
information
on
AEFI
and
provide
a
platform
for
teaching
about
causality
assessment.
© 2012 Elsevier Ltd. All rights reserved.
1.
Introduction
Immunizations
have
led
to
dramatic
declines
in
the
morbidity
and
mortality
associated
with
most
vaccine-preventable
diseases
[1].
However,
because
many
of
these
diseases
have
been
nearly
eliminated,
some
individuals
do
not
fully
appreciate
the
risks
from
The
findings
and
conclusions
in
this
report
are
those
of
the
author(s)
and
do
not
necessarily
represent
the
official
position
of
the
Centers
for
Disease
Control
and
Prevention.
Corresponding
author
at:
Institute
for
Vaccine
Safety,
Johns
Hopkins
Bloomberg
School
of
Public
Health,
615N.
Wolfe
Street,
Baltimore,
MD
21205,
USA.
Tel.:
+1
410
955
6964;
fax:
+1
410
502
6733.
E-mail
addresses:
tproveau@jhsph.edu,
nhalsey@jhsph.edu
(N.A.
Halsey).
1
Karen
Broder,
Jerry
Tokars,
Rosanna
Setse,
Barbara
Pahud,
Howard
Choi,
Robert
Sparks,
Sarah
Elizabeth
Williams,
Melvin
Berger,
Stephen
Dreskin,
Renata
Engler,
Jane
Gidudu.
the
diseases
and
this
contributes
to
increased
focus
on
vaccine
safety.
Adverse
events
following
immunizations
(AEFI)
may
be
coincidental
events
or
the
vaccine
may
have
increased
the
risk
of
the
adverse
event
(AE).
Health
care
providers
and
public
health
officials
should
become
familiar
with
the
process
of
evaluating
indi-
vidual
AEFI,
assessing
them
for
causality
to
improve
management
of
patients,
and
assist
in
monitoring
vaccine
safety
[2,3].
There
is
an
important
distinction
between
determining
that
a
vaccine
caused
an
AEFI
in
a
particular
patient
versus
establishing
that
the
vaccine
can
cause
a
specific
AEFI
in
the
general
population.
Establishing
general
causation
usually
requires
well-designed
epidemiologic
studies
demonstrating
significantly
increased
risk
of
the
AE
following
the
vaccine
compared
with
unvaccinated
individuals,
and/or
mechanistic
studies
such
as
laboratory
inves-
tigations
implicating
the
vaccine
in
affected
individuals
[4].
Local
reactions
at
the
vaccine
injection
site,
immediate
hypersensitivity
reactions
in
the
absence
of
other
exposures,
and
possibly
recurrent
0264-410X/$
see
front
matter ©
2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.vaccine.2012.04.005
5792 N.A.
Halsey
et
al.
/
Vaccine
30 (2012) 5791–
5798
AEs
following
repeated
challenges
with
the
same
vaccine
[5]
are
considered
to
be
evidence
supporting
a
causal
relationship
[4].
Most
experts
prefer
to
see
consistent
evidence
provided
by
different
investigators
working
with
different
populations
before
concluding
that
a
causal
relationship
has
been
established
[3].
Nevertheless,
on
occasion
a
single
case
may
provide
sufficient
evi-
dence
to
establish
a
causal
relationship.
For
example,
identification
of
measles
vaccine
virus
in
the
lung
tissue
of
a
single
HIV-infected
individual
who
developed
progressive
pulmonary
disease
approxi-
mately
nine
months
after
immunization
indicated
that
the
vaccine
was
responsible
for
the
disease
[6].
A
complete
discussion
of
the
science
of
causality
assessment
is
beyond
the
scope
of
this
manuscript,
and
more
detailed
reviews
are
available
[7–9].
After
a
causal
relationship
with
a
vaccine
has
been
established
for
a
specific
AEFI,
subsequent
individuals
who
develop
the
same
AEFI
should
be
carefully
examined
to
determine
if
the
evidence
supports
a
causal
relationship
in
each
individual.
Evidence
for
review
includes
vaccine
receipt
documentation,
confirmation
of
the
diagnosis,
evaluation
of
evidence
for
other
potential
causes,
deter-
mination
if
the
AEFI
developed
within
a
timeframe
known
to
be
associated
with
an
increased
risk,
and
relevant
data
in
the
medical
literature.
The
Clinical
Immunization
Safety
Assessment
(CISA)
[10]
net-
work
has
reviewed
patients
with
AEFI
as
part
of
its
mission
to
improve
the
understanding
of
vaccine
safety.
CISA
developed
an
algorithm
to
illustrate
the
multiple
questions
that
must
be
addressed
in
a
logical
sequence
when
assessing
causality
[10].
The
algorithm
is
intended
to
assist
health
care
providers,
public
health
officials,
and
scientists
understand
the
process
of
causal-
ity
assessment
on
individual
patients
with
AEFI
and
to
encourage
the
collection
of
key
information
needed
to
make
an
appropriate
assessment
of
causality
based
on
the
available
scientific
evidence
for
the
AEFI.
This
algorithm
is
intended
for
use
with
individual
patients
and
is
not
intended
to
establish
or
rule
out
causal
asso-
ciations
in
a
general
manner
as
applied
to
other
patients
with
the
same
AEFI.
Each
report
should
be
reviewed
based
upon
the
evidence
available.
Many
of
the
AEFI
used
to
review
this
algorithm
were
reported
to
the
Vaccine
Adverse
Event
Reporting
System
(VAERS).
VAERS,
established
in
1990,
is
a
national
passive
surveillance
system,
jointly
managed
by
the
Centers
for
Disease
Control
and
Preven-
tion
(CDC)
and
the
Food
and
Drug
Administration
(FDA).
Health
care
providers,
vaccine
manufacturers,
patients,
and
caregivers
are
encouraged
to
report
post-vaccination
AEFI
to
VAERS,
regardless
of
their
cause.
The
role
of
VAERS
is
to
detect
unexpected
or
unusual
patterns
of
AEFI
following
the
licensure
of
a
vaccine,
especially
unusual
or
rare
AEFI
that
are
unlikely
to
be
recognized
in
pre-
licensure
clinical
trials.
2.
Methods
We
used
established
criteria
for
causality
assessment
applied
to
individual
patients
to
develop
a
generalizable
predictive
algo-
rithm
[4,5,7].
The
algorithm
was
developed
over
the
past
four
years
and
was
revised
to
accommodate
nuances
encountered
in
the
pro-
cess
of
assessing
AEFI.
Versions
of
the
algorithm
have
been
used
to
evaluate
over
500
AEFI.
Most
AEFI
were
reported
to
VAERS,
some
were
reported
in
scientific
publications,
and
some
were
identified
by
clinicians
who
contacted
CISA
investigators
for
advice.
2.1.
Definitions
Conclusions
regarding
the
likelihood
of
a
causal
relationship
between
vaccines
and
AEFI
for
individual
patients
often
vary,
depending
on
the
strength
of
the
evidence
and
whether
another
factor
may
be
more
likely
to
have
caused
the
event.
A
variety
of
terms
is
used
by
investigators,
healthcare
providers,
govern-
ment
agencies,
and
the
public
to
describe
the
probability
of
an
evidence-based
causal
relationship
between
a
vaccine
(or
other
medicinal
product)
and
an
AEFI
[11,12].
The
World
Health
Organi-
zation
(WHO)
has
used
the
terms
“very
likely/certain,”
“probable,”
“possible,”
“unlikely,”
“unrelated,”
and
“unclassifiable”
[13]
to
rate
a
vaccine-AEFI
causal
relationship.
These
and
similar
terms
(“defi-
nite,”
“probable,”
or
“possible”)
are
often
used
for
describing
the
certainty
of
a
diagnosis
leading
to
confusion
and
inconsistency
of
interpretation
when
describing
causal
relationships.
For
this
reason,
the
CISA
Causality
Working
Group
has
selected
terms
to
describe
causal
relationships
for
individual
cases
that
reflect
the
uncertainty
that
is
often
present
in
causality
assessment.
“Consistent
with”
a
causal
relationship
means
that
the
available
evidence
supports
a
causal
relationship
between
the
vaccine
and
the
AEFI
in
the
individual
but
it
does
not
rule
out
the
possibility
that
the
AEFI
may
have
been
caused
by
a
factor
other
than
the
vac-
cine.
The
term
is
used
when
the
vaccine
is
known
to
have
caused
the
AEFI
in
others,
but
there
is
some
uncertainty
regarding
the
possibility
that
the
relationship
was
coincidental
in
the
individ-
ual
under
consideration
and
some
other
factor
was
the
primary
cause.
“Inconsistent
with”
a
causal
association
means
that
the
available
evidence
does
not
support
a
causal
relationship
between
vaccine
administration
and
the
reported
AEFI
in
the
individual.
“Indeterminate”
means
that
the
available
evidence
is
insufficient
to
support
or
rule
out
a
causal
relationship
in
the
individual.
In
many
instances
explanatory
notes
or
caveats
will
be
indicated
to
clarify
this
designation
based
on
available
evidence.
The
algorithm
is
intended
as
a
guide,
not
a
rigid
set
of
rules
for
conducting
causality
assessment.
We
propose
evaluating
data
in
a
stepwise
process
and
reevaluating
as
needed
when
new
information
becomes
available.
For
some
decision
points,
expert
subspecialty
advice
may
be
needed.
2.2.
Stepwise
process
for
assessing
causality
Please
refer
to
the
respective
numbered
decision
points
on
the
algorithm
(Fig.
1)
or
to
the
web-based
version
(http://tinyurl.com/VaccineCausality),
which
presents
the
reviewer
with
one
question
at
a
time.
Note
that
a
“no”
answer
to
key
questions
takes
you
to
the
next
key
question.
A
“yes”
answer
usually
takes
you
to
a
branch
question,
which
is
noted
by
adding
a
lower
case
“a,”
“b,”
etc.,
as
needed.
1.
Is
the
diagnosis
of
the
AEFI
correct?
The
diagnosis
should
be
validated
using
information
obtained
from
the
individual’s
medical
history,
physical
examination,
and
laboratory
data.
Criteria
for
determining
the
diagnosis
of
the
many
different
AEFI
(such
as
encephalitis
or
Guillain–Barré
syndrome)
are
beyond
the
scope
of
this
report;
however,
The
Brighton
Collaboration
[14]
has
developed
clinical
case
defini-
tions
for
more
than
20
AEFI.
The
preliminary
diagnosis
may
change
during
the
course
of
evaluating
a
clinical
disorder
as
additional
data
become
available.
The
reviewer
will
also
need
to
make
judgments
regarding
the
adequacy
of
available
data
and/or
seek
additional
information
from
the
medical
records.
Many
AEFI
have
incorrect
diagnoses,
or
their
reports
con-
tain
insufficient
data
to
make
a
correct
diagnosis
with
a
high
degree
of
certainty
[15,16].
If
the
reported
diagnosis
is
found
to
be
incorrect,
it
is
appropriate
to
conclude
“other
diagnosis”;
the
new
diagnosis
can
then
be
reviewed
for
a
possible
causal
association
with
the
vaccine.
When
the
diagnosis
is
uncertain,
N.A.
Halsey
et
al.
/
Vaccine
30 (2012) 5791–
5798 5793
Fig.
1.
Review
of
case
reports
of
adverse
events
following
immunizations.
the
preliminary
assessment
should
be
“indeterminate,
diagno-
sis
uncertain.”
If
the
reviewer
decides
to
continue
the
assessment
based
on
a
presumptive
diagnosis
while
seeking
additional
data,
a
caveat
should
be
added
to
any
subsequent
conclusions
reflect-
ing
the
uncertainty
of
the
diagnosis.
2.
Does
clinical
or
laboratory
evidence
exist
that
supports
possi-
ble
causes
for
the
AEFI
other
than
the
vaccine
in
the
affected
individual?
The
evaluation
should
determine:
whether
there
are
other
known
causes
for
the
AEFI;
if
other
known
causes
exist,
whether
clinical
testing
can
deter-
mine
if
these
other
causes
were
present
in
the
patient;
if
other
causes
were
present,
was
the
patient
given
the
appro-
priate
tests?
and
if
tests
were
administered,
the
results.
There
are
multiple
possible
causes
for
many
AEFI.
Evidence
should
be
evaluated
to
determine
whether
there
is
a
relation-
ship
between
an
infection
or
other
exposure
(i.e.
a
medication
or
food
allergen)
and
the
AEFI.
If
evidence
for
other
causes
is
not
available,
a
decision
should
be
made
regarding
the
feasibility
of
obtaining
it.
The
interim
assessment
should
be
“indeterminate.”
The
reviewer
can
add
an
explanatory
note
such
as
“seek
evidence
for
other
causes”
or
“evidence
for
other
causes
not
obtained,”
if
it
is
unlikely
that
additional
evidence
can
be
obtained.
At
the
time
the
AEFI
is
initially
evaluated
the
results
of
ongoing
studies
may
be
unavailable;
however,
further
follow-up
may
reveal
evi-
dence
of
other
causes.
If
studies
for
other
causes
have
not
been
conducted,
it
is
appropriate
to
advise
the
treating
clinicians
to
seek
such
data
if
possible.
If
the
reviewer
chooses
to
continue
assessing
the
potential
role
of
the
vaccine
in
the
AEFI,
a
state-
ment
should
be
included
in
the
final
assessment
regarding
the
evidence
for
other
causes.
2a.
If
evidence
for
other
causes
exists,
is
the
evidence
considered
definitive?
If
the
evidence
for
other
causes
of
the
AEFI
is
definitive
(e.g.
isolation
of
an
infectious
agent
in
affected
fluid
or
tis-
sue),
a
conclusion
of
“inconsistent
with
a
causal
association”
is
appropriate
and
the
explanatory
note
“other
cause
iden-
tified”
may
be
added.
However,
an
AEFI
may
have
multiple
causes.
Although
an
exposure
may
be
a
known
risk
factor,
such
as
respiratory
infections
preceding
certain
neurolog-
ical
disorders,
clinical
evidence
of
an
exposure
may
not
5794 N.A.
Halsey
et
al.
/
Vaccine
30 (2012) 5791–
5798
constitute
definitive
evidence
that
the
respiratory
infection
caused
the
AEFI.
Where
the
evidence
is
not
definitive,
the
conclusion
should
be
“indeterminate”;
if
the
reviewer
elects
to
continue
with
the
algorithm
assessment,
a
qualifying
statement
should
be
added
to
the
final
conclusion
regard-
ing
the
causality
of
the
vaccine
such
as
“evidence
of
other
possible
causes
exists.”
3.
Is
there
a
known
causal
association
between
the
AEFI
and
the
vaccine?
The
reviewer
should
determine
if
the
reported
AEFI
has
been
shown
to
have
a
causal
association
with
administered
vaccines
based
on
published
scientific
studies
using
established
causal
principles
[4].
Individual
published
case
reports
or
reports
to
VAERS
should
not
be
accepted
as
evidence
for
a
causal
associa-
tion
[15,17].
Although
there
is
no
master
list
of
all
known
causal
relationships
between
vaccines
and
AEFIs,
reviewers
can
con-
sult
reviews
by
the
IOM
Committee
to
Review
Adverse
Effects
of
Vaccines,
which
are
often
accepted
as
conclusive
[18,19].
The
Vaccine
Injury
Compensation
Program
(VICP)
also
maintains
a
table
showing
AEFI
[20,21]
for
which
the
program
com-
pensates
affected
individuals.
However,
this
list
is
incomplete,
excludes
mild
AEFI
and
those
that
do
not
result
in
persistent
injuries.
Also,
compensation
is
sometimes
provided
to
patients
for
some
AEFI
where
the
evidence
for
a
causal
association
is
incomplete.
Examples
of
AEFI
that
are
known
to
have
causal
associations
include
local
reactions
near
the
injection
site
including
ery-
thema,
tenderness,
and
swelling
[22].
These
reactions
usually
develop
within
hours
or
a
few
days
after
vaccine
administration.
Also,
nodules
near
the
injection
site
sometimes
develop
up
to
several
weeks
after
vaccination
and
recurrent
nodules
have
been
described
after
repeat
injections
of
vaccines
containing
similar
components
[23,24].
It
is
reasonable
to
assume
that
an
immu-
nization
at
or
near
this
site
was
the
cause
of
the
reaction
unless
the
site
cannot
be
differentiated
from
other
immunizations
that
were
given
nearby.
However,
coincidental
trauma
to
the
extrem-
ity
may
cause
pain
and
swelling
which
could
be
confused
with
an
injection
site
reaction.
Also,
some
underlying
condition
may
coincidentally
become
apparent
within
a
few
days
after
an
injec-
tion,
such
as
a
tumor
near
the
injection
site,
as
occurred
in
a
clinical
vaccine
trial
where
investigators
determined
that
the
tumor
had
been
growing
for
several
months
prior
to
vaccination
[25].
Swelling
of
the
entire
limb
following
DTaP
[26],
and
large
local
reactions
following
pneumococcal
polysaccharide
vaccines
[27]
are
other
examples
of
local
reactions
that
are
well
described
after
these
respective
vaccines.
If
no
accepted
causal
relationship
is
determined,
go
to
Question
4.
If
limited
data
suggesting
a
possible
association
are
found,
go
to
Question
5.
3a.
Did
the
event
occur
within
the
time
window
of
known
increased
risk?
If
a
causal
relationship
between
a
vaccine
and
the
event
has
been
established,
the
reviewer
should
determine
if
the
onset
of
the
AEFI
was
during
the
post-vaccination
win-
dow
of
time
known
to
be
associated
with
increased
risk.
The
VICP
table
noted
above
includes
time
windows
for
compensation
[21].
If
the
event
occurred
during
the
time
window
defined
for
known
increased
risk,
the
AEFI
can
be
classified
as
“consistent
with”
a
causal
association.
For
exam-
ple,
the
risk
of
febrile
seizures
increases
during
the
7–10
days
following
the
first
dose
of
live
attenuated
MMRV
vac-
cines
[28].
Young
children
frequently
develop
respiratory
and
other
infections
coincidental
with
the
administration
of
vaccines;
these
infections
could
be
the
source
of
fever
as
noted
among
placebo
recipients
in
clinical
trials
[29,30].
Since
there
is
no
definitive
test
to
determine
what
caused
the
fever,
it
may
be
difficult
to
determine
whether
the
vac-
cine
or
some
coincidental
factor
was
the
primary
cause
of
the
fever
that
precipitated
the
febrile
seizure
in
an
individ-
ual.
The
“consistent
with”
assessment
reflects
the
concept
that
the
vaccine
could
have
caused
a
febrile
seizure
dur-
ing
the
time
window
of
increased
risk,
but
other
factors
may
have
contributed
to
the
illness.
Similarly,
hypersensi-
tivity
reactions
such
as
urticaria
(“hives”),
angioedema,
or
anaphylaxis
usually
occur
within
a
few
minutes
after
expo-
sure,
but
onset
up
to
4
h
after
exposure
is
accepted
for
immunoglobulin
E
(IgE)
mediated
[31]
reactions.
For
events
that
occur
within
4
h
after
vaccination,
it
is
reasonable
to
conclude
“consistent
with
a
causal
relationship.”
However,
there
may
have
been
other
exposures
within
the
4
h
time
window
capable
of
causing
the
same
reaction.
For
exam-
ple,
among
the
AEFI
we
reviewed
were
two
individuals
who
developed
urticaria
within
1
h
after
receipt
of
a
vac-
cine;
however,
they
had
consumed
seafood
within
minutes
prior
to
the
onset
of
the
reaction.
Skin
testing
revealed
the
seafood
to
be
the
cause
of
the
reaction,
and
not
the
vaccine.
A
publication
by
CISA
investigators
titled
“An
Algorithm
for
Treatment
of
Patients
with
Hypersensitivity
Reactions
after
Vaccines”
presents
a
more
detailed
description
of
the
steps
involved
in
evaluating
patients
with
possible
immediate
hypersensitivity
reactions
following
the
administration
of
vaccines
[30].
3b.
Are
there
qualifying
factors?
Are
there
circumstances
in
which
an
event
that
occurred
outside
the
recognized
time
window
of
increased
risk
might
still
have
been
caused
by
the
vaccine?
For
example,
vaccine-
associated
paralytic
poliomyelitis
usually
occurs
less
than
30
days
after
vaccine
exposure
in
healthy
children
[29].
How-
ever,
in
immunocompromised
children
vaccine-associated
paralytic
poliomyelitis
could
have
its
onset
more
than
30
days
after
vaccination
[32].
Also,
hypersensitivity
reactions
such
as
hives
can
occur
beyond
the
usual
4
h
time
window
of
accepted
risk
for
IgE-mediated
hypersensitivity
because
hives
can
be
mediated
by
IgG
antibody,
by
immune
com-
plexes
as
part
of
a
serum
sickness
reaction,
or
through
other
mechanisms
[33,34].
If
no
qualifying
factors
exist,
AEFI
occurring
outside
of
recognized
time
windows
of
increased
risk
should
be
classified
as
“inconsistent
with”
a
causal
asso-
ciation
from
the
vaccine(s)
received,
and
an
explanatory
note
may
be
added.
Additional
work
is
needed
to
identify
other
qualifying
factors
that
might
affect
usual
time
windows
of
increased
risk
for
different
AEFI
with
known
causal
associa-
tions.
4.
Is
there
strong
evidence
against
a
causal
association?
In
the
presence
of
strong
scientific
evidence
against
a
causal
association
with
the
vaccine(s)
administered,
the
event
should
be
classified
as
“inconsistent
with”
a
causal
association.
For
example,
individual
case
reports
raised
concerns
about
the
possibility
of
diphtheria,
tetanus,
and
pertussis
(DTP)
or
hep-
atitis
B
vaccines
causing
sudden
infant
death
syndrome
(SIDS)
[35].
However,
controlled
studies
later
demonstrated
that
the
events
were
coincidental
and
the
risk
of
SIDS
was
actually
lower
during
the
several
days
after
DTP
vaccination
than
would
be
expected
by
chance.
Subsequent
evidence
indicated
that
sleep
position
was
a
major
factor
in
the
development
of
SIDS
[36].
5.
Is
the
AEFI
an
infection?
In
most
illnesses,
there
is
clear
evidence
for
or
against
an
infection
as
the
cause,
even
if
the
agent
has
not
been
identified.
Some
disorders,
such
as
central
nervous
system
inflammation
(as
evidenced
from
inflammatory
cells
in
the
cerebrospinal
fluid
or
neuroimaging
studies)
can
be
caused
by
N.A.
Halsey
et
al.
/
Vaccine
30 (2012) 5791–
5798 5795
acute
or
chronic
infections,
or
by
aberrant
immune
responses
affecting
host
tissues.
For
patients
with
evidence
of
infectious
diseases
(e.g.
laboratory
evidence
of
infection
or
having
a
clin-
ical
syndrome
that
may
be
caused
by
infectious
agents),
the
answer
is
yes,
and
reviewers
should
proceed
to
Question
6.
For
patients
without
evidence
of
an
infectious
process,
skip
to
Question
10.
6.
Was
evidence
of
an
infectious
agent
found
in
the
patient?
If
no
agent
was
sought,
efforts
should
be
made
to
locate
clinical
specimens
for
testing.
If
specimens
are
not
avail-
able,
the
designation
should
be
“indeterminate.”
If
specimens
are
available,
they
should
be
sent
for
culture,
antigen
detec-
tion
(e.g.
immunofluorescence)
or
methods
for
detecting
DNA
or
RNA
(e.g.
polymerase
chain
reaction)
to
provide
evidence
of
the
vaccine
agent
(if
a
live
vaccine
was
received)
or
of
other
infectious
agents
that
could
have
caused
the
disorder.
The
designation
should
be:
“indeterminate,
additional
testing
indicated.”
6a.
If
yes,
was
it
the
vaccine
agent?
If
the
infectious
agent
identified
is
the
same
virus
or
bac-
teria
as
the
live
vaccine
agent,
further
testing
is
indicated
to
determine
if
this
is
an
illness
caused
by
a
coincidental
infection
with
the
naturally
occurring
wild
type
agent
or
the
vaccine
strain.
If
no,
proceed
to
Question
7.
6b.
If
yes,
was
the
genetic
sequence
confirmed
to
be
vaccine
origin?
If
the
organism
identified
in
the
patient
has
a
different
sequence
than
the
organism
used
to
produce
the
live
vac-
cine,
the
possibility
of
coincidental
infection
by
the
wild
type
organism
should
be
evaluated.
For
example,
some
chil-
dren
developed
clinical
varicella
with
more
than
100
skin
lesions
within
two
weeks
after
receipt
of
varicella
vaccine
[37].
Although
a
small
number
of
the
children
tested
had
sequences
of
the
vaccine
virus
strain
used
to
produce
the
vaccine,
the
majority
of
children
had
sequences
indicating
wild
type
viruses.
If
the
wild
type
virus
sequence
is
con-
firmed,
the
correct
conclusion
is
“Inconsistent
with
a
causal
association,
other
agent
identified.”
However,
detecting
the
varicella
vaccine
strain
virus
in
the
cerebrospinal
fluid
from
a
patient
with
clinical
meningoencephalitis
following
vac-
cination
is
evidence
that
the
vaccine
agent
was
a
likely
cause
of
meningoencephalitis
[38].
If
yes
(vaccine
origin
is
confirmed),
proceed
to
Question
6c.
If
sequencing
was
not
performed,
“indeterminate”
is
the
correct
assessment,
with
an
explanatory
note
indicating
that
wild-type
infection
has
not
been
ruled
out.
6c.
If
yes,
is
the
agent
expected
in
tissues
or
body
fluids
within
this
time
window?
Live
vaccine
agents
replicate
in
the
body
and
may
appear
transiently
in
tissues
or
body
fluids;
detecting
these
agents
during
the
time
of
expected
replication
after
vaccina-
tion
could
result
in
the
false
assumption
that
the
AEFI
is
attributable
to
the
vaccine
agent.
The
presence
of
the
vac-
cine
agent
may
be
coincidental
with
to
the
illness,
which
is
caused
by
some
other
factor.
For
example,
live
rotavirus
vac-
cine
viruses
are
detectable
in
the
stool
from
three
to
nine
days
after
the
first
dose
of
vaccine
[39].
A
child
may
develop
a
coin-
cidental
illness
that
was
caused
by
another
infection
or
by
an
underlying
disease
during
this
time
of
vaccine
virus
repli-
cation.
In
this
situation,
the
detection
of
the
vaccine
agent
should
not
be
considered
conclusive
evidence
of
a
causal
association.
In
such
situations,
it
is
appropriate
to
classify
the
association
as
“indeterminate”
and
to
conduct
studies
to
identify
other
causes.
However,
if
the
vaccine
agent
is
found
in
affected
tissue
when
vaccine
virus
replication
is
not
expected
(e.g.,
6c.
answer
is
“no”)
and
no
evidence
for
other
causes
exists,
this
evidence
may
be
designated
as
“consistent
with
a
causal
association.”
2.3.
Testing
the
vaccine
for
contamination
If
the
infectious
agent
identified
in
the
patient
is
not
the
live
vaccine
agent,
the
infectious
disease
may
be
caused
by
a
coinci-
dental
community-acquired
infection
or
possible
contamination
of
the
vaccine
with
other
adventitious
agents
during
vaccine
pro-
duction
or
handling.
Vaccine
contamination
after
release
by
the
manufacturer
can
occur
when
multidose
vials
are
not
handled
properly.
This
risk
can
be
reduced
or
eliminated
by
using
single
dose
vials
or
prefilled
syringes.
The
risk
window
for
time
from
vaccination
to
onset
of
disease
from
a
contaminated
vaccine
will
vary
by
the
infectious
agent
and
the
dose
in
the
vaccine.
Health
care
providers
assessing
possible
contamination
of
vaccines
should
seek
the
advice
of
public
health
officials
and/or
infectious
disease
experts.
Guidance
on
injection
safety
and
investigations
can
be
found
at
www.who.int/injection safety.
7. Was
the
same
agent
detected
in
the
vial?
If
contamination
is
suspected,
the
vaccine
vial
(if
available)
should
be
tested
for
the
same
agent.
Other
vials
from
the
same
lot
should
also
be
tested
in
collaboration
with
public
health
officials
if
there
is
concern
about
contamination
during
produc-
tion.
Examples
of
AEFI
caused
by
contamination
of
multi-dose
vials
include
several
instances
of
group
A
Streptococcal
celluli-
tis,
abscess,
and
sepsis
in
children
who
received
injections
from
multi-dose
DTP
vials;
and
bacterial
contamination
of
multi-dose
measles
vaccine
in
developing
country
settings
where
resid-
ual
vaccine
was
inappropriately
stored
overnight
rather
than
discarded
as
per
recommendations
[40–42].
Inadvertent
con-
tamination
can
rarely
occur
during
vaccine
manufacturing
as
reported
in
2004
when
influenza
vaccine
was
suspected
to
be
contaminated
with
Serratia
marcescens
[43].
If
the
organism
found
in
the
vial
is
the
same
as
that
identified
in
the
patient,
then
the
assessment
should
be
“consistent
with
a
causal
associ-
ation,
vial
contamination.”
If
evidence
of
bacterial
contamination
of
a
vial
is
found
then
an
investigation
should
take
place
to
try
to
determine
the
time,
place,
and
procedures
that
allowed
con-
tamination
to
occur
and
institute
corrective
action
to
prevent
recurrences.
8.
Do
other
affected
patients
exist?
If
more
than
one
individual
who
received
vaccine
from
the
same
vial
or
lot
develops
a
similar
infectious
process
caused
by
the
same
organism,
it
is
reasonable
to
assume
that
the
vial
or
lot
may
have
been
contaminated,
even
if
the
vial
or
syringe
is
no
longer
available
for
testing.
In
such
instances,
it
is
reasonable
to
conclude
“indeterminate”
and
the
phrase
“additional
stud-
ies
needed”
can
be
added.
If
the
infectious
agents
are
available
from
the
patients,
genetic
testing
can
be
performed
to
deter-
mine
if
the
agents
are
identical;
this
would
be
unlikely
to
occur
by
chance
unless
the
patients
had
some
other
common
exposure.
Also,
affected
individuals
could
be
tested
for
evidence
of
recent
infection
by
the
suspect
agent
by
looking
for
serum
antibody
responses
to
the
suspected
agent.
9.
Is
there
evidence
of
infection
at
the
injection
site?
If
no
other
affected
patients
exist,
reviewers
should
determine
whether
or
not
there
was
clinical
evidence
of
infection
(such
as
an
abscess
or
cellulitis)
at
or
near
the
injection
site.
Care
should
be
taken
to
distinguish
actual
cellulitis
from
an
extensive
local
reaction
without
infection
such
as
have
been
observed
following
DTaP,
pneumococcal
polysaccharide,
and
other
vaccines
[26,27].
If
there
is
evidence
of
a
local
as
well
as
systemic
infection,
the
conclusion
should
be
“indeterminate”
pending
the
receipt
of
5796 N.A.
Halsey
et
al.
/
Vaccine
30 (2012) 5791–
5798
additional
data
about
infection
by
a
specific
agent
and/or
vial
contamination.
The
specific
organism
identified
provides
valu-
able
clues.
Many
viruses
and
some
bacteria
commonly
cause
coincidental
infections.
In
the
absence
of
other
affected
individ-
uals
who
received
the
same
vaccine,
such
infections
are
most
likely
coincidental
infections.
However,
infections
from
unusual
organisms
that
are
not
typically
associated
with
spontaneous
infections
may
raise
suspicions
that
the
vial
has
been
contami-
nated.
If
there
is
no
evidence
of
infection
at
the
injection
site
and
the
evidence
supports
a
coincidental
infection,
the
causal
assess-
ment
should
indicate
“inconsistent
with
a
causal
association,
coincidental
infection.”
2.4.
Other
AEFI
10. Is
there
a
specific
laboratory
test
implicating
the
vaccine
in
the
pathogenesis?
At
the
present
time
we
are
not
aware
of
specific
laboratory
tests
other
than
PCR
for
infectious
agents
(discussed
in
6b)
that
provide
definitive
evidence.
However,
the
potential
exists
for
development
of
new
immunologic
assays
that
could
identify
vaccine
components
contributing
to
disease,
such
as
antigen-
antibody
immune
complexes.
10a. If
yes,
what
is
the
possibility
of
a
false
positive?
The
possibility
of
a
false
positive
test
result
should
be
evaluated.
If
the
test
has
been
validated
and
found
to
have
a
low
probability
of
false
positives,
“consistent
with
causal
association”
is
the
appropriate
assessment,
although
false
positive
tests
do
occur
even
with
vali-
dated
tests.
For
experimental
assays
that
have
not
been
validated,
“indeterminate”
is
the
most
appropriate
assign-
ment
for
causal
association
with
the
vaccine.
For
example,
as
mentioned
above,
patients
with
gastrointestinal
dis-
ease
and/or
autism
were
reported
(by
intestinal
biopsy
or
peripheral
blood
cells)
to
have
positive
PCR
tests
for
measles
virus,
but
it
was
subsequently
shown
that
these
tests
were
amplifying
host
tissue
and
not
the
vaccine
virus
[44].
11.
If
the
answer
to
Question
10
is
no,
was
the
AEFI
a
nerve
or
joint
injury
near
the
site
of
injection
that
occurred
shortly
after
the
injection?
11a.
If
yes,
is
there
evidence
of
an
incorrect
site
of
administra-
tion
that
could
explain
the
injury
or
inflammation?
Examples
include
direct
injury
to
the
radial
nerve
after
injection
of
anthrax
(or
other)
vaccines
below
the
del-
toid
muscle
[45]
or
inflammation
of
the
subacromial
bursa
caused
by
injections
in
the
proximal
1/3
of
the
deltoid
muscle
[46].
The
IOM
has
determined
that
the
evidence
supporting
a
causal
relationship
was
convincing
[9].
Imag-
ing
or
nerve
conduction
studies
have
been
helpful
in
some
of
these
cases.
There
are
multiple
causes
of
persistent
pain
syndromes
and
clinical
investigation
by
an
expert
is
often
indicated.
If
there
is
evidence
of
damage
to
a
nerve
at
the
site
of
injection
or
imaging
and
clinical
data
indicat-
ing
that
the
injection
contributed
to
the
inflammation,
“consistent
with
causal
association”
is
the
appropriate
assessment,
with
an
explanatory
note
summarizing
the
evidence.
If
the
clinical
information
and
possible
imaging
data
are
inconclusive,
the
correct
classification
is
“indeterminate.”
If
the
AEFI
is
not
a
local
nerve
or
joint
injury,
the
correct
classification
is
“indeterminate”,
because
causality
can-
not
be
determined
for
an
individual
AEFI
when
general
causality
has
not
been
established
or
ruled
out
by
earlier
investigations.
Larger
population-based
studies
or
inves-
tigations
based
on
specific
tests
linking
the
vaccine
to
the
AEFI
are
needed
to
assess
causality.
Even
if
large
numbers
of
AEFIs
are
identified,
this
information
may
only
provide
a
“signal”
indicating
the
need
for
further
investigation.
Common
situations
that
have
led
to
misunderstandings
about
causal
associations
include
repeated
episodes
of
the
same
disorder
following
vaccines,
or
exacerbation
of
a
dis-
ease
after
receipt
of
a
vaccine.
Two
episodes
of
the
same
illness
following
the
same
vaccine
during
a
biologically
plausible
time
window
may
suggest
a
causal
relationship
in
individual
cases,
but
coincidental
associations
may
also
occur,
especially
with
disorders
where
there
is
no
specific
diagnostic
test
to
determine
the
factors
responsible
for
causing
the
disease.
There
are
no
clear
criteria
for
accep-
tance
of
rechallenge
data
to
determine
causal
associations
with
vaccines.
In
1994,
the
IOM
Vaccine
Safety
Commit-
tee
used
rechallenge
data
from
several
affected
patients
to
conclude
that
tetanus
toxoid-containing
vaccines
can
cause
GBS
[47].
However,
in
2011,
after
reviewing
addi-
tional
studies
that
yielded
no
evidence
of
increased
risk,
the
IOM
Committee
to
Review
Adverse
Effects
of
Vaccines
published
a
report
concluding
that
there
was
inade-
quate
evidence
to
either
establish
or
dismiss
a
causal
relationship
between
diphtheria
toxoid-,
tetanus
toxoid-
,
or
acellular
pertussis-containing
vaccines
and
GBS
[19].
Many
patients
with
chronic
or
relapsing
diseases
receive
vaccines,
and
some
relapses
occur
that
are
temporally
associated
with
immunizations.
It
is
usually
impossible
to
use
individual
case
reviews
to
determine
the
possibility
of
disease
relapse
being
caused
by
a
vaccine;
controlled
stud-
ies
are
needed
to
determine
if
there
is
an
increased
risk
of
relapse.
2.5.
Case
Studies:
Applying
the
Algorithm
2.5.1.
Case
1:
Encephalitis
A
39-year-old
nurse
received
intranasal
live
attenuated
influenza
vaccine.
She
was
well
on
the
day
of
vaccination
but
devel-
oped
a
sore
throat,
body
aches,
and
a
low
grade
fever
two
days
after
vaccination.
The
next
day
she
awoke
feeling
weak
with
diaphore-
sis
and
pleuritic
chest
pain.
Her
primary
care
physician
made
a
clinical
diagnosis
of
possible
pneumonia
and
she
was
treated
with
clarithromycin;
however,
her
chest
radiograph
was
normal.
Three
days
later
she
developed
worsening
sore
throat,
anterior
lymph
node
swelling,
generalized
weakness,
and
a
severe
headache.
On
day
11
after
vaccination
she
presented
to
an
emergency
room
with
vomiting
and
confusion.
Cerebrospinal
fluid
(CSF)
revealed
a
white
blood
cell
count
of
480/mm
3
,
protein
97
g/l,
glucose
2
mmol/l,
and
PCR
for
herpes
sim-
plex
virus
was
negative.
She
was
admitted
with
a
presumptive
diagnosis
of
meningitis/encephalitis
caused
by
the
live
influenza
vaccine.
She
was
treated
with
Acyclovir,
Vancomycin,
and
Cefurox-
ime.
Her
illness
resolved
quickly
and
she
was
discharged
after
three
days
of
hospitalization
with
a
diagnosis
of
“post-vaccination
encephalitis.”
Following
the
algorithm,
the
initial
diagnosis
of
encephalitis
was
verified
by
the
clinical
presentation
and
CSF
ele-
vated
white
blood
cell
count
and
protein
(Question
1).
The
presence
of
respiratory
and
other
symptoms
could
have
been
interpreted
as
evidence
for
other
viral
infections;
however,
patients
receiving
live
attenuated
influenza
vaccine
can
develop
respiratory
symptoms
after
vaccination.
This
illness
should
have
been
initially
classified
as
“indeterminate,
additional
testing
indicated”
(after
Question
2).
After
hearing
about
this
case,
a
CISA
investigator
obtained
per-
mission
from
the
treating
physicians
and
the
patient
to
retrieve
residual
CSF;
a
subsequent
viral
culture
of
the
CSF
revealed
an
enterovirus.
The
answers
to
Questions
2
and
2a
are
changed
to
“yes”
N.A.
Halsey
et
al.
/
Vaccine
30 (2012) 5791–
5798 5797
as
there
was
definitive
evidence
of
another
cause
for
the
AEFI.
The
final
classification
was
“inconsistent
with
causal
association,
other
cause
identified.”
2.5.2.
Case
2:
Acute
disseminated
encephalomyelitis
(ADEM)
A
previously
healthy
11-year-old
boy
developed
headache,
excessive
sleeping,
and
severe
neck
and
back
pain
five
weeks
after
receiving
meningococcal
conjugate
and
Tdap
vaccines.
CSF
analyses
revealed
109
wbc/mm
3
,
protein
67
mg/dl
and
glucose
66
mg/dl.
A
CT
scan
of
the
head
was
normal
and
the
initial
diagno-
sis
was
aseptic
meningitis.
CSF,
urine
and
blood
bacterial
cultures
revealed
no
growth.
Two
days
later
an
MRI
of
the
brain
and
spine
revealed
abnormalities
consistent
with
ADEM.
No
viral
cultures
were
obtained
and
there
was
no
evidence
of
any
other
acute
infec-
tions.
Applying
the
algorithm,
the
diagnosis
was
correct
(Question
1).
There
was
no
apparent
evidence
for
other
causes
(Question
2),
how-
ever,
many
viral
infections
have
been
reported
that
precede
the
onset
of
ADEM,
and
diagnostic
studies
(viral
cultures
and
serologic
tests)
were
not
conducted
to
rule
out
those
infections.
Vaccines
are
not
a
proven
cause
of
ADEM
(Question
3),
even
though
there
are
case
reports
of
associations
and
some
reviews
include
vaccines
as
possible
sources
of
immunologic
stimuli
that
could
result
in
ADEM
[48].
There
is
no
strong
evidence
against
a
causal
associ-
ation
from
epidemiologic
and
other
studies
(Question
4),
so
the
algorithm
took
us
to
Question
5.
ADEM
is
not
an
infection;
the
pathogenesis
appeared
to
be
an
aberrant
immune
response,
which
led
us
to
Question
10.
Note
that
the
CSF
inflammation
might
cause
a
reviewer
to
choose
“yes”
to
the
infection
question;
however,
fol-
lowing
this
branch
would
result
in
the
same
classification.
There
was
no
specific
laboratory
test
implicating
either
vaccine
in
the
pathogenic
process
(Question
10),
and
because
this
was
the
first
episode
of
this
disorder
for
this
patient
(Question
10a),
the
correct
assessment
was
“indeterminate”
for
the
causal
relationship.
2.5.3.
Case
3:
Encephalitis
in
an
infant
following
maternal
yellow
fever
vaccination
A
23-day-old,
breast-fed
infant
living
in
Brazil
presented
with
fever,
irritability,
and
seizures.
His
mother
had
received
yellow
fever
vaccine
15
days
earlier
and
there
were
no
other
known
exposures.
A
lumbar
puncture
revealed
a
cerebrospinal
fluid
(CSF)
with
white
blood
cells
128/mm
3
,
and
protein
concentration
of
106
mg/dl.
Tests
for
bacterial
and
other
viral
causes
were
nega-
tive
but
PCR
with
genetic
sequencing
revealed
yellow
fever
virus
with
a
genetic
sequence
that
matched
the
vaccine
virus.
Applying
the
algorithm,
the
diagnosis
was
correct
(Question
1),
there
was
no
evidence
for
other
causes
(Question
2),
there
was
no
known
causal
association
at
the
time
of
the
investigation
(Question
3),
there
was
no
strong
evidence
against
a
causal
association
(Question
4),
the
AEFI
was
an
infectious
process
(Question
5),
the
vaccine
agent
was
found
in
the
CSF
(Question
6a),
and
the
genetic
sequencing
matched
the
vaccine
virus
(Question
6b),
leading
to
the
conclusion
that
this
was
“consistent
with”
a
causal
association.
This
is
an
illustrative
case
from
the
published
literature
[49].
Subsequently
two
other
infants
have
acquired
yellow
fever
vaccine
virus
via
breast-feeding
and
developed
encephalitis
[50,51].
3.
Conclusions
Assessing
causality
in
an
individual
patient
is
challenging
because
of
the
multiple
factors
that
must
be
taken
into
considera-
tion
and
the
need
for
careful
clinical
evaluation
and
judgment.
In
many
reports
of
AEFI,
conclusions
about
causality
cannot
be
deter-
mined
because
of
insufficient
information.
The
CISA
network,
with
input
from
clinical
experts
from
multiple
disciplines,
does
assess
causality
for
selected
AEFI.
We
hope
that
this
systematic
approach
to
the
assessment
of
AEFI
will
be
of
assistance
to
others.
The
sci-
ence
of
causality
assessment
is
undergoing
constant
review,
and
adjustments
to
this
approach
will
likely
be
necessary.
There
are
several
limitations
in
the
information
that
has
gone
into
the
cur-
rent
version
of
the
algorithm
including
the
need
for
clarification
of
time
windows
of
known
increased
risk
for
AEFI
where
causal
assessments
have
been
determined
and
possible
qualifying
factors
affecting
time
windows
in
individual
patients.
We
have
reviewed
more
than
500
reports
of
AEFI
using
the
algorithm.
We
will
con-
tinue
review
additional
reports
and
update
the
algorithm
as
needed
periodically
in
the
same
manner
that
guidelines
for
vaccine
use
are
updated
as
new
scientific
information
becomes
available.
A
web-based
version
of
the
algorithm
is
available
(http://tinyurl.
com/VaccineCausality)
and
it
will
be
updated
frequently.
Comments
and
suggestions
for
improvement
are
welcome
and
can
be
sent
to
VaccineCausality@gmail.com.
Acknowledgments
Tina
Proveaux
provided
editorial
and
technical
support.
Ketan
Jumani
provided
technical
support.
This
study
was
funded
through
a
subcontract
with
America’s
Health
Insurance
Plans
(AHIP)
under
contract
200-2002-00732
from
the
Centers
for
Disease
Control
and
Prevention
(CDC).
References
[1]
National
Immunization
Program
CDC.
Achievements
in
public
health,
1900–1999:
impact
of
vaccines
universally
recommended
for
children
United
States,
1900–1998.
MMWR
Morbidity
and
Mortality
Weekly
Report
1999;48:243–8.
[2]
Iannelli
V.
Understanding
vaccine
safety.
Updated
December
3,
2003.
Avail-
able
at:
www.pediatrics.about.com/cs/immunizations/a/vaccine
safety.htm
(accessed
July
20,
2012).
[3]
Ellenberg
SS,
Braun
MM.
Monitoring
the
safety
of
vaccines:
assessing
the
risks.
Drug
Safety
2002;25(3):145–52.
[4]
Halsey
NA.
The
science
of
evaluation
of
AEs
associated
with
vaccination.
Sem-
inars
in
Pediatric
Infectious
Diseases
2002;13(3):205–14.
[5]
Wise
RP,
Kiminyo
KP,
Salive
ME.
Hair
loss
after
routine
immunizations.
JAMA
1997;278(14):1176–8.
[6]
Angel
JB,
Walpita
P,
Lerch
RA,
et
al.
Vaccine-associated
measles
pneumonitis
in
an
adult
with
AIDS.
Annals
of
Internal
Medicine
1998;129(2):104–6.
[7]
Rothman
KJ,
Causation
Greenland
S.
Causal
Inference
in
Epidemiology.
Ameri-
can
Journal
of
Public
Health
2005;95:S144–50.
[8]
Coggon
DI,
Martyn
CN.
Time
and
chance:
the
stochastic
nature
of
disease
cau-
sation.
Lancet
2005;365(9468):1434–7.
[9]
Institute
of
Medicine.
In:
Stratton
KR,
Ford
A,
Rusch
E,
Clayton
EW,
editors.
Adverse
effects
of
vaccines:
evidence
and
causality.
Washington,
DC:
National
Academy
Press;
2011.
[10]
LaRussa
PS,
Edwards
KM,
Dekker
CL,
et
al.
Understanding
the
role
of
human
variation
in
vaccine
adverse
events:
the
Clinical
Immunization
Safety
Assess-
ment
Network.
Pediatrics
2011;127(Suppl.
1):S65–73.
[11]
Arimone
Y,
Begaud
B,
Miremont-Salame
G<et-al>.
Agreement
of
expert
judg-
ment
in
causality
assessment
of
adverse
drug
reactions.
European
Journal
of
Clinical
Pharmacology
2005;61:169–73.
[12]
WHO.
Causality
assessment
of
AEs
following
immunization.
Weekly
Epidemi-
ological
Record
2001;76(12):85–9.
[13]
WHO.
Immunization
safety.
Last
updated
April
6,
2009.
Available
at:
www.who.int/immunization
safety/en/
(accessed
July
20,
2012).
[14]
Brighton
Collaboration.
Available
definitions
available
at:
https://
brightoncollaboration.org/public/what-we-do/standards/case-
definitions/available-definitions.html
(accessed
July
20,
2012).
[15]
Slade
BA,
Leidel
L,
Vellozzi
C,
et
al.
Postlicensure
safety
surveillance
for
quadri-
valent
human
papillomavirus
recombinant
vaccine.
JAMA
2009;302(7):750–7.
[16]
Vellozzi
C,
Broder
KR,
Haber
P,
et
al.
AEs
following
influenza
A
(H1N1)
2009
monovalent
vaccines
reported
to
the
Vaccine
AE
Reporting
System,
United
States,
October
1,
2009–January
31,
2010.
Vaccine
2010;28(45):7248–55.
[17]
Varricchio
F,
Iskander
J,
Destefano
F,
et
al.
Understanding
vaccine
safety
infor-
mation
from
the
Vaccine
AE
Reporting
System.
Pediatric
Infectious
Disease
Journal
2004;23(4):287–94.
[18]
Institute
of
Medicine.
In:
Stratton
KR,
Ford
A,
Rusch
E,
Clayton
EW,
editors.
Adverse
effects
of
vaccines:
evidence
and
causality.
Washington,
DC:
National
Academy
Press;
2011.
pp.
19–34.
[19]
Institute
of
Medicine.
In:
Stratton
KR,
Howe
CJ,
Johnston
Jr
RB,
editors.
Adverse
events
associated
with
childhood
vaccines:
evidence
bearing
on
causality.
Washington,
DC:
National
Academy
Press;
1994.
[20] National
Vaccine
Injury
Compensation
Program.
About
VICP.
Available
at:
www.hrsa.gov/vaccinecompensation
(accessed
July
20,
2012).
5798 N.A.
Halsey
et
al.
/
Vaccine
30 (2012) 5791–
5798
[21]
Cook
KM,
Evans
E.
The
National
Vaccine
Injury
Compensation
Program.
Pedi-
atrics
2011;127:S74–7.
[22]
Deloria
MA,
Blackwelder
WC,
Decker
MD,
et
al.
Association
of
reactions
after
consecutive
acellular
or
whole-cell
pertussis
vaccine
immunizations.
Pediatrics
1995;96(3
Pt
2):592–4.
[23]
Klein
NP,
Edwards
KM,
Sparks
RC,
Dekker
CL.
Recurrent
sterile
abscesses
fol-
lowing
aluminum
adjuvant-containing
vaccines.
BMJ
Case
Reports
2009;200:9,
http://dx.doi.org/10.1136/bcr.09.2008.0951
.
[24]
Bergfors
E,
Björkelund
C,
Trollfors
B.
Nineteen
cases
of
persistent
pru-
ritic
nodules
and
contact
allergy
to
aluminium
after
injection
of
com-
monly
used
aluminium-
adsorbed
vaccines.
European
Journal
of
Pediatrics
2005;164:691–7.
[25]
Sikand
VK,
Halsey
N,
Krause
PJ,
et
al.
Safety
and
immunogenicity
of
a
recombi-
nant
Borrelia
burgdorferi
outer
surface
protein
A
vaccine
against
Lyme
disease
in
healthy
children
and
adolescents:
a
randomized
controlled
trial.
Pediatrics
2001;108(1):123–8.
[26] Rennels