Professional Documents
Culture Documents
ORIGINAL ARTICLE
A B S T R AC T
propensity-score adjustment to
control for depression severity
BACKGROUND
and
other
potential
Whether the use of selective
confounders.
serotonin-reuptake inhibitors
(SSRIs) and other anti-RESULTS
depressants
during
pregnancy is associated with
an
increased
risk
of
congenital cardiac defects is
uncertain. In particular, there
are concerns about a possible
asso-ciation
between
paroxetine use and right
ventricular outflow tract
obstruction and between
sertraline use and ventricular
septal defects.
METHODS
From
the
Division
of
Pharmacoepide
mi-ology
and
Pharmacoecono
mics,
Department
of
Medicine,
Brigham
and
Womens
Hospital
and
Harvard Medical
School (K.F.H.,
J.A.,
J.M.F.,
H.M.,
R.L.,
M.K., S.S.), the
Department of
Epidemiology,
Harvard School
of Public Health
(K.P., S.H.-D.),
the Center for
Womens
Mental Health,
Massa-chusetts
General
Hospital
(L.S.C.),
and
the
Medical
Genetics Unit,
MassGeneral
Hos-pital
for
Children
(L.B.H.) all in
Boston;
and
Duke University
School
of
Medicine,
Durham,
NC
(S.S.). Address
reprint re-quests
to
Dr.
Huybrechts at
the Division of
Pharmacoepide
miology
and
Pharmacoeconomics,
Department of
Medicine, Brig
ham
and
Womens
Hospital, 1620
Tremont
St.,
Suite
3030,
Boston,
MA
A
t
o
t
a
l
o
f
6
4
,
3
8
9
w
o
m
e
n
(
6
.
8
%
)
u
s
e
d
a
n
t
i
d
e
p
r
e
s
s
a
n
t
s
d
u
r
i
n
g
t
h
e
f
i
r
s
t
t
r
i
m
e
s
t
e
r
.
O
v
e
r
a
l
l
,
6
4
0
3
i
n
f
a
n
t
s
w
h
o
w
e
r
e
o
s
e
d
t
o
a
n
t
i
d
e
p
r
e
s
s
a
n
t
s
w
e
r
e
b
o
r
n
w
i
t
h
a
c
a
r
d
i
a
c
n
o
t
d
e
f
e
c
t
e
x
p
(
7
2
.
3
i
n
f
a
n
t
s
w
i
t
h
a
c
a
r
d
i
a
c
d
e
f
e
c
t
p
e
r
1
0
,
0
0
0
i
n
f
a
n
t
s
)
,
a
s
c
o
Copyright 2014
Massachusetts
Medical Society.
m
p
a
r
e
d
w
i
t
h
5
8
0
i
n
f
a
n
t
s
w
i
t
h
e
x
p
o
s
u
r
e
(
9
0
.
1
p
e
r
1
0
,
0
0
0
i
n
f
a
n
t
s
)
.
c
t
s
A
s
s
o
c
i
a
t
i
o
n
s
w
e
r
e
b
e
t
w
e
e
n
a
n
t
i
d
e
p
r
e
s
s
a
n
t
u
s
e
a
n
d
c
a
r
d
i
a
c
d
e
f
e
a
t
t
e
n
u
a
t
e
d
w
i
t
h
i
n
c
r
e
a
s
i
n
g
l
e
v
e
l
s
o
f
a
d
j
u
s
t
m
e
n
t
f
o
r
c
o
n
f
o
u
n
d
i
n
g
.
T
h
e
r
e
l
a
t
i
v
e
r
i
s
k
s
o
f
a
n
y
c
a
r
d
i
a
c
d
e
f
e
c
t
w
i
t
h
t
h
e
u
s
e
o
f
S
S
R
I
s
w
e
r
e
1
.
2
5
(
9
5
%
c
o
n
f
i
d
e
n
c
e
i
n
t
e
r
v
a
l
[
C
I
]
,
1
.
1
3
t
o
1
.
3
8
)
i
n
t
h
e
u
n
a
d
j
u
s
t
e
d
a
n
a
l
y
s
i
s
,
1
.
1
2
(
9
5
%
C
I
,
1
.
0
0
t
o
1
.
2
6
)
i
n
t
h
e
a
n
a
l
y
s
i
s
r
e
s
t
r
i
c
t
e
d
t
o
w
o
m
e
n
w
i
t
h
d
e
p
r
e
s
s
i
o
n
,
a
n
d
1
.
0
6
(
9
5
%
C
I
,
0
.
9
3
t
o
1
.
2
2
)
i
n
t
h
e
f
u
l
l
y
a
d
j
u
s
t
e
d
a
n
a
l
y
s
i
s
r
e
s
t
r
i
c
t
e
d
t
o
w
o
m
e
n
w
i
t
h
d
e
p
r
e
s
s
i
o
n
.
W
e
f
o
u
n
d
n
o
s
i
g
n
i
f
i
c
a
n
t
a
s
s
o
c
i
a
t
i
o
n
b
e
t
w
e
e
n
t
h
e
u
s
e
o
f
p
a
r
o
x
e
t
i
n
e
a
n
d
r
i
g
h
t
v
e
n
t
r
i
c
u
l
a
r
5
9
o
u
t
f
l
o
w
t
o
t
r
a
c
t
o
b
s
t
r
u
c
t
i
o
n
(
r
e
l
a
t
i
v
e
r
i
s
k
,
1
.
0
7
;
9
5
%
C
I
,
0
.
1
.
9
3
)
o
r
b
e
t
w
e
e
n
t
h
e
u
s
e
o
f
s
e
r
t
r
a
l
i
n
e
a
n
d
v
e
n
t
r
i
c
u
l
a
r
CONCLUSIONS
s
e
p
t
a
l
d
e
f
e
c
t
s
(
r
e
l
a
t
i
v
e
r
i
s
k
,
1
.
0
4
;
9
5
%
C
I
,
0
.
7
6
t
o
1
.
4
1
)
.
T
h
e
r
e
s
u
l
t
s
o
f
t
h
i
s
l
a
r
g
e
,
p
o
p
u
l
a
t
i
o
n
b
a
s
e
d
c
o
h
o
r
t
s
t
u
d
y
s
u
g
g
e
s
t
e
d
n
o
s
u
b
s
t
a
n
t
i
a
l
i
n
c
r
e
a
s
e
i
n
t
h
e
r
i
s
k
o
f
c
a
r
d
i
a
c
m
a
l
f
o
r
m
a
t
i
o
n
s
a
t
t
r
i
b
u
t
a
b
l
e
t
o
a
n
t
i
d
e
p
r
e
s
s
a
n
t
u
s
e
d
u
r
i
n
g
t
h
e
f
i
r
s
t
t
r
i
m
e
s
t
e
r
.
(
F
u
n
d
e
d
b
y
t
h
e
A
g
e
n
c
y
f
o
r
H
e
a
l
t
h
c
a
r
e
R
e
s
e
a
r
c
h
a
n
d
Q
u
a
l
i
t
y
a
n
d
t
h
e
N
a
t
i
o
n
a
l
I
n
s
t
i
t
u
t
e
s
o
f
H
e
a
l
t
h
.
)
D
o
w
n
l
o
a
d
e
d
f
r
o
m
ejm.org
on
August
21,2015.
For
personal
useonly.
Noother
uses
without
permissio
n.
C
o
p
yr
ig
ht
2
0
1
4
M
as
sa
c
h
us
et
ts
M
e
di
TheNewEnglandJournalofMedicine
ca
l
S
o
ci
et
y.
A
ll
ri
g
ht
s
re
se
rv
e
d.
antidepressants,
-19
but
LINICAL
DEPRESSION
considerable
OCCURS IN 10
contro-versy
TO
15% ofremains
pregnantregarding
1
women. Thewhether this is a
use of antiserious concern
depressantor much ado
medicationsabout little, as
during
pregnancy hasnoted in an
increasededitorial
steadily overpublished with
time, withtwo of the
reportedreports.13,14,20
prevalences of 8
to 13% in the
of
publicly
insured pregnant
women
and adolescents
in the United
States to assess
the risk of
congenital
cardiac defects
after the use of
specific
antidepressants,
with attention
to the potential
for
confounding by
the underlying
depression and
associated
factors.
Studies have
diverse
United States. associations,
Selective
often in the
serotonincontext
of
reuptake
multiple
inhibitors
comparisons.
(SSRIs) are the
M
Yet, at least two
most commonly
E
studies
have
prescribed
T
shown
a
antidepressants
H
doubled
or
during
tripled
risk
of
O
4
pregnancy.
Inright
D
2005, on theventricular
S
basis of earlyoutflow
tract
results of twoobstruction
DATA
epidemiologic asso-ciated
SOURCE AND
studies, the Food
with paroxetine STUDY
2-4shown
and
Drug 13,14
and of
Administration use
(FDA) warnedventricular
health care pro-septal defects
fessionals
thatassociated with
early
prenatalsertraline
exposure
touse.13,19
A
paroxetine may
meta-analysis
increase the risk
estimated
a
of
congenital
50%
increase
in
cardiac malformations, and thethe prevalence
cardiac
FDA reclassifiedof
defects
overall
the drug to pregparoxnancy categorywith
D (evidence ofetine use during
first
human fetal risk,the
21
but benefits maytrimester.
It
5
warrant
use). has re-mained
unclear,
Most
malformations however,
cited in the earlywhether these
reports leading toassocia-tions
the FDA warn-are causal or
ing were septaldue
to
defects.
Sincesystematic error
then,
severalor chance. We
studies
haveconducted
a
evaluated
thecohort
study
teratogenicity ofusing a large
SSRIs and othernational
COHORT
The
study
cohort
was
drawn from the
Medicaid
Analytic
eXtract for 46
U.S. states and
Washington,
D.C., for the
period of 2000
through 2007.
Mon-tana and
Connecticut
were excluded
because
of
difficulty
in
linking data for
mothers
and
infants,
Michigan was
excluded
because
of
incomplete
data, and data
from Arizona
were
not
available. The
Medicaid
individualAnalytic eXtractlevel
data set contains demographic
and Medicaid
enroll-ment
information, as
well as data on
all physician
services and
hospitalization
s
and
the
accompanying
diagnoses and
procedures and
all
filled
outpatient
medication
prescriptions.
The
development
of our study
cohort
has
been described
22
previously.
Briefly,
we
identified all
completed
pregnancies in
women
and
adolescents 12
to 55 years of
age (hereafter,
women) and
linked
these
pregnancies to
liveborn
infants. Us-ing
a
validated
23
algorithm,
we estimated
the date of the
last menstrual
period on the
basis of the
delivery date
and diagnostic
codes
indicative of
preterm
delivery.
Finally,
we
required all the
women to be
eligible
for
Medicaid,
without
supplementary
private
insurance or
restricted
benefits, from
3
months
The
etiologically
of
sup-ply.
relevant
Women were
window
forconsidered to
exposure
ex-have had expotended from thesure if the days
date of the lastof
supply
menstrual
overlapped
period throughwith the first
day
90
oftrimester. We
pregnancy (firstdefined
the
trimester). Wefollowing expodetermined
sure categories:
maternal use ofany
SSRI,
antidepressants paroxetine,
by a review ofsertraline,
pharmacy
fluoxetine,
dispensing
tricyclic
records, usingantidepressants,
the dispensingserotonin
date and thenorepinephrine
number of daysreuptake
inhibitors
(SNRIs),
bupropion,
and
other
antidepressants
(Table S1 in the
Supplementary
Appendix,
available with
the full text of
this article at
NEJM.org).
The refer-ence
group consisted
of
women
without exposure
to
antidepressants
during the first
trimester.
permission.
TheNewEnglandJournalofMedicine
Copyright
2398
Downloadedfromnejm.orgon 2014
August21, Massachusetts
N ENGL J
2015.For Medical
MED 370;25
personaluse Society.All
NEJM.ORG
rights
only.Noother
JUNE 19,
useswithout reserved.
2014
26
COVARIATES
conditions.
address
27
To
by
the
underlying
indication, we considered
prox-ies for depression
severity
(number
of
depression
diagnoses
received as an outpatient
and as an in-patient) and
other indications for
antidepressant use (other
mental health disorders,
pain-related
diagnoses,
sleep
disorders,
the
premenstrual
ten-sion
syndrome, smoking, and
the
chronic
fatigue
syndrome).
STATISTICAL ANALYSIS
We
compared
the
distributions
of
sociodemo-graphic,
clinical, and health care
utilization char-acteristics
among
the
various
exposure groups, and we
calculated the absolute
risks
of
cardiac
malformations. Logisticregression analysis was
used to estimate odds
ratios for cardiac malformations
and
their
corresponding 95% confidence intervals. Because
the odds ratio is an excellent approximation of
the risk ratio in the case
of rare outcomes, the
results are referred to as confounders.29
28
relative risks. Use of the Propensity scores
robust variance esti-mator were derived from
predicted
to account for correlations the
of
within
women
with probability
treatment
multiple pregnancies did
estimated
in
logistic-regression
model
that
contained all the
covariates
listed
above
without
additional variable
selection.
We
created 100 equally
sized propensityscore
strata,
dropped
uninformative
strata (i.e., all
strata that did not
contain at least one
treated woman and
one
untreated
woman),
and
stratified
the
outcome
models
according to these
propensity-score
strata. In all cases,
less than 0.5% of
treated women and
less than 0.1% of
untreated women
were in-cluded in
the dropped strata.
In confirmatory
analyses, we used
a
highdimensional
propensity-score
algorithm, which
evaluates
thousands
of
diagnoses,
procedures,
and
pharmacy-claim
codes to identify
and
prioritize
covariates
that
serve as proxies
for unmeasured
ew
T
Eng
h
land
e
Jour
N
nal
Downloadedfromnejm.org
of
onAugust21,2015. Massachusetts
Me
Forpersonaluseonly. MedicalSociety.
dici
Nootheruseswithout Allrights
ne
permission.Copyright reserved.
2014
2399
31
Appendix).
confounders. AIn the analysis
total of 200of SSRIs, we
stratified the
empirically
analysis
identi-fied
confounders according to
were selectedthe use of just
and were com-that drug class
bined
withversus the use
mul-tiple
investigator- of
antidepressant
identified
covariates
toclasses.
improve
To evaluate
adjustment forthe effect of
30
confounding. potential
No
ad-misclassifijustments werecation
of
made
forexposure, we
multiple
redefined
comparisons. exposure status
We
as having had
performed
prespecified one or more
subgroup andprescriptions
filled during the
sen-sitivity
analyses
tofirst trimester
evaluate
the(as compared
robustness ofwith days of
that
the
primarysupply
results (for anyoverlap
with
cardiac
the
first
malformation). trimester); we
Because
theredefined the
cohort
wereference group
studied
wasas women with
younger
andno
more raciallyantidepressant
diverse
than
exposure
cohorts
in
throughout
previous
pregnancy. To
13,14
studies,
weevaluate
the
tested for effecteffect
of
modification potential
ac-cording to
misclassificatio
age (<30 years
vs. 30 years)n of outcome,
and race orwe restricted
ethnic
groupthe outcome to
(white
vs.inpatient
nonwhite). Wediagnoses only
and extended
con-ducted
doseresponse infant followanalyses
forup to 1 year.
low, medium,We corrected
and high dosesodds ratios for
of
outcome misantidepressants classification
using the firstusing
and
highestsensitivities and
doses
specificities
dispensed
consistent with
(Table S4 in
the
positive
the
predictive
Supplementary
values
es-
timated in the
internal
validation
26,32
study.
To
further assess
whether
outcomes were
well cap-tured,
we evaluated
some
wellknown associations in our
data set in
particular,
associations
between
cardiac
malformations
and
maternal
dia-betes, use
of
an
anticonvulsant
agent,
or
multifetal
33
pregnancy.
RE
SU
LT
S
CHARACTERI
STICS OF
THE STUDY
COHORT
We
identified
949,504 eligible
pregnancies.
Women
could
have contributed
more than 1
pregnancy to the
cohort. During
the
first
trimester, 64,389
wom-en (6.8%)
used
an
antidepressant:
46,144 women
(4.9%)
were
exposed to an
SSRI,
5954
(0.6%) to a
tricyclic
antidepressant,
6904 (0.7%) to
an SNRI, 8856
(0.9%)
to
bupropion, and
7055 (0.7%) to
other
antidepressants.
Among
the
SSRIs, sertraline
was used most
frequently
(in
14,040 women),
followed
by
paroxetine (in
11,126)
and
fluoxetine
(in
11,048).
As
compared with
women who
took no antidepressant,
women who
filled
a
prescription
for
an
antidepressant
were
older,
had
greater
health
care
utilization, and
were
more
likely to be
white, to use
other
psychotropic
medications, to
have a chronic
illness
(in
particular,
hypertension
or diabetes),
and to use
suspected
teratogenic
medications
(Table
1).
Baseline
characteristics
were
more
homogeneous
in
analyses
comparing
users
of
various
antidepressant
classes than in
analyses
comparing
users
of
antidepressants
with nonusers
(Tables
S5
through S13 in
the
Supplementary
Appendix).
RISK OF
CARDIAC
MALFORMATI
ON
Overall,
cardiac
malformations
were
diagnosed in
6403 infants
who were not
exposed to an
anti-depressant
during the first
trimester (72.3
cardiac
malformations
per
10,000
infants),
as
compared with
580
infants
who
were
exposed (90.1
cardiac
malformations
per
10,000
infants). This
higher
unadjusted risk
among
exposed
infants
was
observed for
each of the
specific types
of
malformations
considered
(Table 2).
In
unadjusted
analyses, the
relative risk of
any
cardiac
malformation
was 1.25 with
SSRIs (95%
confidence
interval [CI],
1.13 to 1.38),
0.98
with
tricyclic
antidepressants
(95% CI, 0.72
to 1.32), 1.51
with
SNRIs
(95% CI, 1.20
to 1.90), 1.19
with
bupropion
that
comparisons
were
made
between
groups
with
nearly
identical characteristics
(Table 1, and
Tables
S5
through S13 in
the
Supplementary
Appendix).
Adjustment
for
the
propensity
score further
attenuated the
remaining
positive
associations.
permission.
Copyright
TheNewEnglandJournalofMedicine
2014
Downloadedfromnejm.orgon Massachusetts
2400
August21, Medical
2015.For Society.All
N ENGL J
personaluse rights
MED 370;25
NEJM.ORG
only.Noother
reserved.
JUNE 19,
useswithout
2014
Table 1. Selected Cohort Characteristics of Women with Exposure to an SSRI during the First
Trimester and Women without Exposure to an Antidepressant.*
Characteristic
Overall Cohort
Depression-Restricted Cohort
SSRI
(N=46,144)
No Exposure
(N=885,115)
SSRI
(N=36,778)
No Exposure
(N=180,564)
25.65.9
23.95.8
25.56.0
25.353.1
White
34,098 (73.9)
339,144 (38.3)
27,299 (74.2)
136,506 (75.6)
Black
5,438 (11.8)
313,369 (35.4)
4,193 (11.4)
19,380 (10.7)
Hispanic
4,145 (9.0)
164,317 (18.6)
3,261 (8.9)
15,017 (8.3)
Other or unknown
2,463 (5.3)
68,285 (7.7)
2,025 (5.5)
9,661 (5.4)
6,470 (14.0)
98,886 (11.2)
5,250 (14.3)
25,299 (14.0)
1,288 (2.8)
10,628 (1.2)
997 (2.7)
4,957 (2.7)
Age yr
Race or ethnic group no. (%)
1,682 (3.6)
15,364 (1.7)
1,303 (3.5)
6,449 (3.6)
36,783 (79.7)
180,564 (20.4)
36,778 (100.0)
180,564 (100.0)
2.86.5
0.22.0
3.57.1
3.320.3
As an inpatient
0.10.3
0.00.1
0.10.3
0.11.0
7,353 (15.9)
31,681 (3.6)
6,654 (18.1)
33,599 (18.6)
Antipsychotic agent
9,534 (20.7)
48,657 (5.5)
8,621 (23.4)
42,987 (23.8)
Anxiolytic agent
3,148 (6.8)
8,189 (0.9)
2,895 (7.9)
14,320 (7.9)
Benzodiazepine
14,560 (31.6)
49,063 (5.5)
12,856 (35.0)
63,100 (34.9)
13,277 (28.8)
115,608 (13.1)
11,540 (31.4)
56,323 (31.2)
3,764 (8.2)
26,030 (2.9)
3,097 (8.4)
15,756 (8.7)
3,508 (7.6)
26,967 (3.0)
2,806 (7.6)
14,345 (7.9)
Barbiturate
Use of suspected teratogen no. (%)
1*
Plusminus values are means SD. Data are from the Medicaid Analytic eXtract for the period 2000
through 2007. SSRI denotes selective serotonin-reuptake inhibitor.
To account for propensity-score strata, the observations of untreated women were weighted according
to the distribu-tion of the treated women among the propensity-score strata. One uninformative stratum
(containing five women with exposure to an SSRI) was dropped in the propensity-scorestratified analysis.
Race or ethnic group was determined on the basis of information submitted to the Centers for
Medicare and Medicaid Services by individual states, which was based on information that had been collected
and coded from Medicaid appli-cations.
Data on preterm birth were related to the current pregnancy.
The numbers of outpatient and inpatient diagnoses of depression were used as proxies for the severity of depression.
Pregnant women with exposure to known teratogens were excluded from the cohort, although those
with exposure to suspected teratogens were included. Suspected teratogens included angiotensin-converting
enzyme inhibitors, flucon-azole, aminoglycosides, folic acid antagonists, methimazole, potassium iodide,
tetracycline, danazol, misoprostol, statins, warfarin, and propylthiouracil.
2401
Downloadedfromnejm.orgon
TheNewEngland
August21,
Journalof
2015.For
Medicine
personaluse permission.Copyright
only.Noother 2014MassachusettsMedical
useswithout Society.Allrightsreserved.
Table 2. Absolute Risk of Congenital Cardiac Malformations among Infants Born to Mothers with Antidepressant
Exposure and Infants Born to Mothers without Exposure, According to Antidepressant Category in the Overall Cohort.*
Exposure Group
No exposure
Total
No. of
Women
Any Cardiac
Malformation
Right Ventricular
Outflow Tract
Obstruction
Ventricular Septal
Defect
Other Cardiac
Malformation
Events
Risk
Events
Risk
Events
Risk
Events
Risk
no. of
affected
infants
no./10,000
infants
no. of
affected
infants
no./10,000
infants
no. of
affected
infants
no./10,000
infants
no. of
affected
infants
no./10,000
infants
885,115
6403
72.3
1045
11.8
3212
36.3
3232
36.5
Any antidepressant
64,389
580
90.1
84
13.0
286
44.4
318
49.4
SSRI
46,144
416
90.2
61
13.2
201
43.6
226
49.0
Paroxetine
11,126
93
83.6
16
14.4
44
39.5
48
43.1
Sertraline
14,040
129
91.9
17
12.1
63
44.9
71
50.6
Fluoxetine
11,048
99
89.6
16
14.5
48
43.4
55
49.8
Tricyclic antidepressant
5,954
42
70.5
13.4
24
40.3
18
30.2
SNRI
6,904
75
108.6
12
17.4
39
56.5
38
55.0
Bupropion
8,856
76
85.8
11
12.4
39
44.0
49
55.3
Other antidepressant
7,055
74
104.9
11.3
31
43.9
46
65.2
1*
Data are from the Medicaid Analytic eXtract for the period 2000 through 2007. Infants could have had more than one
cardiac malformation. SNRI denotes serotoninnorepinephrine reuptake inhibitor.
anticonvulsant
(1.04; 95% CI, 0.76 to 1.41). Analyses that used high- (relative risk,
dimensional propensity scores yielded es-sentially the same CI, 1.3 to
results (Table S15 in the Sup plementary Appendix).
multifetal
(relative risk,
SUBGROUP AND SENSITIVITY ANALYSES
CI, 2.8 to 3.1).
maternal
There was no evidence of effect modification and
diabetes
(relative
ac-cording to age or race (Table S16 in the
risk,
3.7;
95%
CI,
Supple-mentary Appendix). We did not
3.4
to
4.0),
use
of
observe a dose response relationship either
an
with respect to the first dose or with respect
to the highest dose dispensed (Table S17 in
the Supplementary Ap-pendix). The relative
risk of cardiac malforma-tions associated
with the use of SSRIs was simi-lar among
users of antidepressant monotherapy (relative
risk, 1.04; 95% CI, 0.90 to 1.21) and users of
polytherapy (relative risk, 1.17; 95% CI, 0.90
to 1.53). The overall findings were not qualitatively affected when we varied the
exposure and outcome definitions (Tables
S18 and S19 in the Supplementary
Appendix). The shift in effect estimates
resulting from correction for predicted
outcome misclassification ranged from 1.3 to
9.6% (Table S20 in the Supplementary
Appendix). We replicated the well-known
associations between cardiac malformations
2402
agent
1.6; 95%
1.8), and
pregnancy
2.9; 95%
August
useonly.
TheNewEnglandJournalofMedicine
21,2015.
Noother
For
uses
Downloadedfromnejm.org
personal
without
on
permissio
n.
Copyrigh
t2014
DIS C
US
SION
association
between
antidepressant use
and
cardiac
anomalies,
in
particular findings
with re-spect to
the
use
of
paroxetine
and
13,14,19
sertraline.
In contrast to
analyses in earlier
studies, our adjusted
analyses
restricted
the
cohort to women
with a recorded
diagnosis
of
depression
in
order to mitigate
potential
confounding by
the underMassachus
etts
Medical
Society.
Allrights
reserved.
cardiac
malformation might
lying psychiatric illness andpotential
risk be detected in an
for
associated conditions andfactors
congenital cardiac inbehaviors, factors that mightanomalies.34
increase
the
risk
of In
addition,
structural
cardiacwomen
with
malformations by means ofdepression
and
utilize
several
mechanisms.anxi-ety
Smoking, alcohol and drugmore health care
use, poor maternal diet,resources, including
obesity,
and
chronicultrasonography,
conditions such as diabetesamniocentesis, and
echocardiog-raphy
and hypertension are all
of the infant, than
more common in patients
do their healthy
with depres-sion than in
35,36
those without depressioncounter-parts.
Hence, there is a
and are
higher chance that a
Exposure
to
Antidepr
essants.
A Unadjusted Analysis
p
r
e
s
e
n
t
e
d
ation
among
infants
born to
Exposure GroupO
women
d
with
Any antidepressant
d
exposure
SSRI
s
to
Paroxetine
antidepr
Sertraline
r
essants
Fluoxetine
a
t
during
Tricyclic antidepressant
t
o
the first
SNRI
i
trimester,
Bupropion
o
s
as
Other
s
h
compare
o
d with
a
w
the risk
B Depression-Restricted
Analysis
n
among
Exposure Groupd
t
infants
Any antidepressant
h
born to
SSRI
9
e
women
Paroxetine
5
without
Sertraline
%
r
such
Fluoxetine
i
exposure
Tricyclic antidepressant
c
s
. Panel A
SNRI
o
k
shows
Bupropion
n
the
Other
f
o
unadjust
i
f
ed
d
analysis,
C Depression-Restricted
Analysis awith Propensity-Score
e
Panel B Stratification
Exposure Groupn
n
the
c
y
analysis
Any antidepressant
e
restricted
SSRI
c
to
Paroxetine
i
Sertraline
a
women
n
Fluoxetine
r
with
t
Tricyclic antidepressant
d
depressi
SNRI
e
i
on, and
Bupropion
r
a
Panel C
v
c
the
Other
a
analysis,
l
m
restricted
s
a
to
Figure 1. Risk
l
women
of Cardiac
a
f
with
Malformation
r
o
depressi
in Infants,
e
r
on, that
According to
Maternal
m
used
p
r
o
p
e
n
s
i
t
y
s
c
o
r
e
s
t
r
a
t
i
f
i
c
a
t
i
o
n
i
n
o
r
d
e
r
t
o
a
d
j
u
s
t
f
o
r
a
a
r
e
f
r
o
m
t
h
e
M
e
d
i
c
a
i
d
A
n
a
l
y
t
i
c
e
X
t
r
a
c
t
f
o
r
t
h
e
p
e
r
i
o
d
c
o
n
f
o
u
n
d
e
r
s
.
2
0
0
0
D
a
t
2
0
0
t
h
r
o
u
g
h
7
.
S
N
R
I
d
e
n
o
t
e
s
s
e
r
o
t
o
n
i
n
n
o
r
e
p
i
n
e
p
h
r
i
n
e
r
e
u
p
t
a
k
e
i
n
h
i
b
i
t
o
r
,
a
n
d
S
S
R
I
s
e
lective
serotonin
reuptake
inhibitor.
T
h
e
N
e
w
E
n
g
l
a
n
d
J
o
u
r
n
a
l
o
f
M
e
d
i
c
i
n
e
Downloadedfromnejm.orgon
August21,
2015.For
personaluse
only.Noother
useswithout
permission.
Copyright
2014
Massachusetts
Medical
Society.All
rightsreserved.
2404
org.nejmfromDownloaded
Copyright
2014,19JUNE ORG.NEJM 370;25MEDJENGLN
withoutusesotherNo.onlyusepersonalFor.201521,AugustonMedicineofJournalEnglandNe
wThe .permission
.reservedrightsAll.SocietyMedicalMassachusetts2014
GroupExposure
OutcometoAccording
AnalysisUnadjusted
Depression-RestrictedAnalysis
1.11 (0.891.38)
1.02 (0.781.34)
SSRI
Paroxetine
1.12 (0.871.45)
1.22 (0.742.00)
1.06 (0.791.42)
1.09 (0.621.90)
Sertraline
Fluoxetine
1.03 (0.641.66)
1.23 (0.752.01)
1.13 (0.691.84)
1.02 (0.571.81)
Tricyclic antidepressant
SNRI
1.14 (0.572.28)
1.47 (0.832.60)
1.10 (0.462.68)
1.47 (0.822.62)
Bupropion
Other
1.05 (0.581.91)
0.96 (0.481.93)
1.10 (0.582.06)
0.61 (0.251.47)
1.23 (1.091.38)
1.02 (0.881.19)
1.20 (1.041.39)
1.09 (0.811.47)
1.01 (0.851.21)
0.77 (0.531.12)
Sertraline
Fluoxetine
1.24 (0.961.59)
1.20 (0.901.59)
1.09 (0.821.45)
1.14 (0.831.56)
Tricyclic antidepressant
SNRI
1.11 (0.741.66)
1.56 (1.142.14)
1.08 (0.651.81)
1.36 (0.971.92)
1.22 (0.891.67)
1.21 (0.851.73)
0.93 (0.631.38)
1.04 (0.701.53)
1.35 (1.211.52)
1.27 (1.101.47)
SSRI
Paroxetine
Bupropion
Other
Other cardiac defect
Any antidepressant
SSRI
Paroxetine
1.34 (1.171.54)
1.18 (0.891.57)
1.25 (1.071.47)
1.11 (0.811.53)
Sertraline
Fluoxetine
1.39 (1.101.76)
1.37 (1.051.79)
1.25 (0.961.64)
1.26 (0.931.71)
Tricyclic antidepressant
SNRI
0.83 (0.521.32)
1.51 (1.102.08)
0.95 (0.551.65)
1.50 (1.082.09)
Bupropion
Other
1.52 (1.142.02)
1.79 (1.342.40)
1.34 (0.971.87)
1.61 (1.172.22)
0.2
0.7
1.0
1.5 2.0
OddsRatio(95%CI)
0.7
1.0
CI)(95%RatioOdds
1.5 2.0
0.2
the
positive termination of the
predictive val-ues pregnancy
fant that might otherwiseyielded
similar
have
been
undetectedresults. Finally, the
clinically,
particularlystrength of the
milder defects such as mus-associations
cular ventricular septalbetween
some
defects, which often closewell-known risk
during early childhood. Infactors (diabetes,
addition to restricting theuse
of
an
analyses to women with aanticonvulsant
diagnosis of depres-sion, weagent,
and
adjusted for a large set of multifetal
prespecified and empiricalpregnancy)
and
potential
confoundingcardiac
variables through the use of malformations that
propensity scores. Althoughwere estimated in
this
ap-proach
cannotour data set were
eliminate
all
potentialconsistent
with
confound-ing, it resulted inprior
reports,
exposure
groups
withsupport-ing
the
virtually identical measuredpremise that the
characteristics and tended tooutcomes
of
move the risk estimates
interest were well
further downward.
captured in our
Our crude associationsstudy.
were weaker than those A strength of
that have been reported in
our study was our
some prior studies. A
use
of
the
potential concern is the
Medicaid Analytic
misclassifica-tion of the
eXtract,
which
exposure or the outcome,
provided a very
since
non-differential
misclassification will tendlarge populationcohort,
to bias results toward thebased
objective
28
null. Documentation thatassessment of drug
a prescription was filledexposure, linkable
does not guarantee that theclinical
medication was actuallyinformation,
taken
as
prescribed.access to medical
However,
secondaryrecords,
and
analyses in which we reavailability
of
quired women to have
information
on
filled or refilled a premultiple
pregscription during the first
nancy outcomes
trimester did not substantially
alter
theand on a wide
findings, although therange of poten-tial
estimates were less preciseconfounders.
our
owing to the reducedHowever,
study
also
had
cohort size.
important
We used validatedsome
limitations.
First,
definitions for outcomes
the
cohort
that were based on ICD-9
live
coding, but a non-trivialincluded
births
only.
Severe
proportion of cases were
mal
not confirmed on recordcardiac
formations
that
review.
However,
an
in
analysis that cor-rected theresulted
spontaneous
aborrelative risk with the use of
conser-vative estimates fortion, stillbirth, or
confounding
by
would therefore have beensuch factors would
missed. Although thisbe unlikely to
restriction could result in aexplain the null
bias
that
wouldfindings; data from
National
underestimate the strengththe
and
of the associations, ourHealth
study shares this limitationNutrition
with the stud-ies thatExamination
identified the potentialSurvey indicate,
associations, so this factorfor example, that
of
cannot
explain
thewomen
discrepant
find-ings.childbearing age
use
Moreover, differences inwho
antidepressants
are
the
proportion
of
more
likely
to
terminations
among
smoke
and
to
be
women with depression
treated
withobese than those
antidepressants
versuswho do not use
these medications,
those who were not treated
with
similar
within the levels of
distributions
for
covariates used in the
different
adjustment would have to
37
be great-er than seemsantidepressants.
plausible in order to fully Medicaid
ac-count for our findingscovers the medical
(see the Supplementaryexpenses for more
than 40% of births
Appendix).
in the United
Second, there was the
38
potential
for
misStates.
classification. InformationMedicaid-eligible
pregnant women
on
lifestyle
factors
are
a
young,
contained
in
theracially
diverse,
administrative data was in-vulnerable
complete (e.g., smoking,population that is
obesity, and alcohol andtraditionally
drug abuse or dependence)understudied.
or absent (e.g., body-massHowever,
we
index), as was informationfound no evidence
effect
on the severity of theof
underlying condition (wemodification
to
used
only
proxies).according
sociodemographic
However,
residual
characteristics.
Therefore, unless
there are other
factors
distinguishing our
study cohort from
other populations
of
pregnant
women that affect
the biologic relations under study,
our results should
be gener-alizable
to
other
28
populations.
In
making
decisions
about
whether to continue
or
discontinue
treatment
with
antidepres-sants
during pregnancy,
clinicians
and
women
must
balance
the
potential risks of
treatment with the
risks
of
not
treating
severe
39
depres-sion.
In
conclusion,
our
results suggest that
the
use
of
antidepressants
during the first trimester does not
substantively
increase the risk of
specific
cardiac
defects.
The
accumulated
evidence implies
low absolute risks
and argues
ew
T
Eng
h
land
e
Jour
N
nal
Downloadedfromnejm.org
of
onAugust21,2015. MedicalSociety.
Med
Forpersonaluseonly. Allrights
icin
Nootheruseswithout reserved.
epermission.Copyright
2014Massachusetts
2405
T h e NE W E NGL A ND JOU R NA L o f
Health of the National Institutes of
Health
(NIH)
against
(K01MH099141, to
important
Dr. Huy-
cardiac
teratogenic
effects
associated with
the
most
commonly
used
antidepressant
medications.
Supported by an
award
from
the
Agency
for
Healthcare Research
and Quality (R01
HSO18533), a career
development
grant
from the National
Institute of Mental
M E DICINE
brechts), and a
training grant in
reproductive,
perinatal,
and
pediatric
epidemiology from
the
National
Institute of Child
Health and Human
Development of the
NIH
(T32HD060454, to
Dr. Palmsten).
Disclosure forms
provided by the
authors
are
available with the
full text of this
article at NEJM.org.
We thank Robert
J. Glynn, Sc.D.,
Ph.D., for helpful
comments on an
earlier version of
the manuscript.
REFERENCES
Gynecol
2011;205(1
J, Heron J,):51.e1Francomb 51.e8.
1. Evans
4.
H, Oke S,
Huybr
Golding J.echts KF,
Cohort
Palmsten
study
ofK, Mogun
depressed H, et al.
mood
National
during
trends in
pregnancy antidepres
and
aftersant
child-birth. medicatio
BMJ
n
2001;323:2 treatment
57-60.
among
2. Andradepublicly
SE, Raebelin-sured
MA, Brownpregnant
J, et al. Usewomen.
Gen Hosp
of
antidepressa Psychiatry
nt
medications 2013;35:2
65-71.
during
pregnancy: a
FDA
multisite
advising
study. Am J
of risk of
Obstet
birth
Gynecol
defects
2008;198(2)
with Paxil.
:194.e1News
194.e5.
5.
release of
6.
7.
10.
8.
9
.
Epid
emio
logy
2006
;17:
7014.
11.
Cole
JA,
Ephr
oss
SA,
Cos
mato
s IS,
Walk
er
AM.
Paro
xetin
e in
the
first
trime
s-ter
and
the
preva
lence
of
cong
enital
malform
ation
s.
Phar
maco
epide
miol
Drug
Saf
2007
;16:1
07585.
12.
Brar
d A,
Ram
os E,
Rey E,
Blais
L, StAndr
M,
Oraich
i D.
First
trimes
ter exposure
to
paroxe
tine
and
risk of
cardia
c
malfor
matio
ns in
infants
: the
import
ance
of
dosag
e.
Birth
Defect
s Res
B Dev
Repro
d
Toxic
ol
2007;
80:1827.
reuptake
L
inhibitor
ouik C, Linuse during
AE, Werlerpregnancy.
MM,
Mayo Clin
Hernn
Proc 2009;
dez-Daz S,84:23-7.
Mitchell
AA. Firsttrimester Nordeng
use
ofH,
van
selective Gelder
serotonin- MM,
reuptake Spigset O,
inhibi-tors Koren G,
and the riskEinarson
of
birthA,
defects. NEberhardEngl J MedGran M.
2007;356:2 Pregnancy
675-83.
outcome
after
A
exposure
lwan
S,to
Reefhuis J,antidepres
Rasmussen sants and
SA, Olneythe role of
RS,
maternal
Friedman depression
JM. Use of:
results
selective from the
serotonin- Norwegia
reuptake n Mother
inhibitors and Child
in pregnan-Cohort
cy and theStudy.
J
risk of birthClin
defects. NPsychopha
Engl J Medrmacol
2007;356:2 2012;32:1
684-92.
86-94.
E
inarson A,Oberlande
Pistelli A,r
TF,
DeSantis Warburton
M, et al.W, Misri
Evaluation S, Riggs
of the riskW,
of
Aghajania
congenital n
J,
cardiovascu Hertzman
lar defectsC. Ma-jor
associated congenital
with use ofmalformat
paroxetine ions
during
following
pregnancy. prenatal
Am
Jexposure
Psychiatry to
2008;165:7 serotonin
49-52.
reuptake
[Errata, Aminhibitors
J
and
Psychiatry benzodiaz
2008;165:7 epines
77, 1208.] using
13.
17.
14.
15.
18.
16.
population
W
ichman CL,-based
Moore KM,health
Lang TR,data. Birth
St SauverDe-fects
JL, HeiseRes B Dev
RH
Jr,Reprod
Watson WJ.Toxicol
Con-genital 2008;
83:68-76.
heart
disease
associated
Pedersen
with
seLH,
lective
Henriksen
serotonin
TB,
19.
Vesterm Daz
gaard M,ed S.
Olsen J,ic Algori
Bech BH.atithms
Selective onto
sero-tonin s estima
reuptake in te the
inhibitors pr beginin
eg ning
pregnancy na of
and
nc pregn
congenital y: ancy
malformat de in
ions:
si admin
popula- gnistrati
tion basedco ve
cohort
ns datastudy.
id bases.
BMJ
er Pharm
2009;339: atiacoepi
b3569.
ondemio
s. l Drug
P Saf
Greene
L 2013;
MF.
o 22:16Teratogeni S 24.
city
ofO
SSRIs ne
serious
20Center
concern or13s for
much ado;8 Medic
about
(6 are &
little? N): Medic
Engl
Je6 aid
Med
74Ser2007;356: 05vices.
2732-3. . Medic
aid
Analy
Wurst KE, tic
Poole C, eXtrac
Ephross
t
SA,
Ol- (MAX
shan AF.M )
ar
First
genera
trimester gul
paroxetine lis infor
use
andA matio
the
V, n:
prevalence SeMAX
of
to 1999congenital,gu2005
specifi- ch state
cally
i claims
cardiac, S, anoma
defects: a lies
M
metaitt (http:/
analysis of /www.
le
epidemiol cms.g
m
ogical
ov/
studies. an Resea
M
Birth
rchDefects A, Statist
Gl
Res A Clin icsyn
Mol
Datan
Teratol
and2010;88:1 R Syste
59-70.
J, ms/
D Comp
or uterPalmsten
m DataK,
Huybrecht ut ands
KF,h Syste
Mogun H,C ms/
et
al.R, Medic
Harnessin H aidDat
g
theer aSour
Medicaid n cesGe
Analytic ndnInfo/
eXtract ez MAX
(MAX) to- Gener
alInfo
evaluate
20.
24.
21.
22.
2
3
.
rmati
on.ht
ml).
25.
Hol
mes
LB,
West
gate
MN.
Usin
g
ICD9
code
s to
estab
lish
preva
lence
of
malf
orma
tions
in
newb
orn
infan
ts.
Birth
Defe
cts
Res
A
Clin
Mol
Terat
ol
2012
;94:
20814.
26.
Palm
sten
K,
Huyb
recht
s K,
Kow
al M,
Hern
nde
zDaz
S.
Valid
ity of
mate
rnal
and
infan
t
outco
mes
withi
n
natio
nwid
e
Medicaid
data.
Phar
maco
epide
miol
Drug
Saf
2014
April
16
(Epub
ahead
of
print).
27.
Schne
eweiss
S,
Seeger
JD,
Maclu
re M,
Wang
PS,
Avorn
J,
Glynn
RJ.
Perfor
mance
of
comor
bidity
scores
to
contro
l for
confoundi
ng in
epide
miolo
gic
studie
s
using
claims
data.
Am J
Epide
miol
2001;
154:
85464.
28.
Rothm
an KJ,
Greenl
and S,
Lash
TL.
Moder
n
epidem
iology.
3rd ed.
Philad
elphia:
Lippin
cott
Willia
ms &
Wilkin
s,
2008.
29.
Brait
man
LE,
Rosen
baum
PR.
Rare
outco
mes,
comm
on
treatm
ents:
of
analytic
treatment
strategies effects
using
using
propensity health care
scores. Annclaims
Intern Meddata.
2002;137:6 Epidemiol
93-5.
ogy
2009;20:
S
512-22.
chneeweiss
S, Rassen
JA, GlynnODonnell
RJ, AvornJ, Shelton
J, MogunR.
Drug
H,
therapy of
Brookhart depression
MA. High-and anxiety
dimensiona disorders.
l propensityIn: Brunton
score
L, Chabner
adjustment B,
in studiesKnollmann
30.
31.
2406
N ENGL
J MED
370;25
NEJM.O
RG
JUNE
B,
eds.se s. Int J
Goodman ns Epide
& Gilmansiti miol
The
vi 2005;
pharmacoloty 34:13
gical basisan 70-6.
of
al
therapeutic ys
s 12th ed.es Heart
New York:of defect
McGraw- m s. In:
Hill,
is Holm
2011:397- cl es LB.
416.
as Com-
33.
sifmon
ie malfor
Fox MP,d matio
Lash TL,bi ns.
Greenland na New
S. A meth-ry York:
od
tova Oxfor
automate ri d
probabilist ab Unive
ic
le rsity
32.
Press
,
2012
:11626.
34.
Jenki
ns
KJ,
Corre
a A,
Feins
tein
JA, et
al.
Noni
nherit
ed
risk
factor
s and
cong
enital
19,
2014
TheNewEnglandJournalofMedicine
Downloadedfrom
cardio
vascul
ar
defects
:
current
knowl
edge: a
scientif
ic
statem
ent
from
the
Ameri
can
Heart
Associ
ation
Counci
l
on
Cardio
vas-
chusetts
Medical
Society.
All
rights
reserve
d.
37.
Ana
cular Disease in the Young:
lytic and
endorsed by the American Academy
reporting
of
Pediatrics.
Cir-culation
guidelines: the
2007;115:2995-3014.
National Health
Bar-Oz B, Einarson T,and Nutrition
Einarson A, et al. Paroxetine andExamina-tion
congenital malformations: meta-Survey
analysis and consideration of poten-(NHANES).
tial confounding factors. Clin Ther Hyattsville,
MD: National
2007; 29:918-26.
Center for
Koren G. The effect ofHealth
ascertainment bias in evaluatingStatistics,
gestational
antidepres-santCenters for
exposure. J Popul Ther Clin Disease Control
Pharma-col 2011;18:e174-e175.
and Prevention, 2005
(http://www.cdc
.gov/nchs/data/
nhanes/nhanes_
03_04/nhanes_a
nalytic
_guidelines_dec
_2005.pdf).
35.
36.
38.
Gar
cia G. Maternal
and
Child
Health (MCH)
update: states
increase
eligibility for
childrens
health in 2007.
Washington,
DC: National
Governors
Association,
No-
Altshuler LL,
Harlow BL, et
al. Relapse of
major
depression
during
pregnancy
in
women
who
maintain or discontinue
antidepressant
treatment.
JAMA
2006;295:499507. [Erratum,
JAMA
2006;
296:170.]
Copyright 2014
Massachusetts
Medical Society.
New Medicine
orgon Forpersonaluseonly. MedicalSociety.
T
England AugustNootheruseswithout Allrights
Downloadedfrom
h
Journal
nejm. 21, permission.Copyright reserved.
e
of
2015. 2014Massachusetts
2407