Original Article

Risk of Congenital Anomalies in Pregnant

Users of Non-Steroidal Anti-Inflammatory Drugs:
A Nested Case-Control Study
Benjamin Ofori,
1
Driss Oraichi,
1
Lucie Blais,
2
Evelyne Rey,
3
and Anick Berard
1
1
Research Center, Sainte-Justine Hospital, Montreal, Quebec, Canada
2
Faculty of Pharmacy, University of Montreal, Pavillon Jean-Coutu, Montreal, Quebec, Canada
3
Faculty of Medicine, University of Montreal, Pavillon Roger-Gaudry, Montreal, Quebec, Canada
BACKGROUND: Many women take non-steroidal anti-inflammatory drugs (NSAIDs) during pregnancy but the risks
for the infant remain controversial. We carried out a study to quantify the association between those women prescribed
NSAIDs in early pregnancy and congenital anomalies. METHODS: A population-based pregnancy registry was built by
linking data from three administrative databases in Quebec between 19972003. The inclusion criteria were mothers of
live singleton infants, between 1545 years of age, covered by the RAMQ drug plan Z12 months before and during
pregnancy, and prescribed an NSAID or other medications during pregnancy. We selected as cases infants with any
congenital anomaly (ICD-9; 740-759) diagnosed in the first year of life. Up to 10 controls, defined as infants with no
congenital anomalies detected were selected for each case. Adjusted odds ratios (OR) and 95% confidence intervals (CI)
were estimated. RESULTS: Within the registry, 36,387 pregnant women met the inclusion criteria. We identified 93 births
with congenital anomalies in 1056 women (8.8%) who filled prescriptions for NSAIDs in the first trimester of pregnancy,
compared to 2478 in 35,331 (7%) women who did not. The adjusted OR for any congenital anomalies for women who
filled a prescription for NSAIDs in the first trimester was 2.21 (95% CI 51.722.85). The adjusted OR for the anomalies
related to cardiac septal closure was 3.34 (95% CI 51.875.98). There were no significant associations with anomalies
of other major organ systems. CONCLUSIONS: Our study suggests that women prescribed NSAIDs during
early pregnancy may be at a greater risk of having children with congenital anomalies, specifically cardiac septal
defects. Birth Defects Research (Part B) 77:268279, 2006. r 2006 Wiley-Liss, Inc.
Key words: non-steroidal anti-inflammatory drugs (NSAIDs); pregnancy;
congenital anomalies; cardiac septal closure anomalies; pregnancy
registry
INTRODUCTION
Non-steroidal anti-inflammatory drugs (NSAIDs) are
among the drugs prescribed most commonly in the
first trimester of pregnancy (Olesen et al., 1999). Over-
the-counter (OTC) NSAID use is also prevalent in this
population with nearly 15% using ibuprofen (Glover
et al., 2003). With the use of NSAIDs being so common,
many women may not associate it with potentially
deleterious effects either for them or their unborn child.
Furthermore, because 50% of pregnancies are unplanned
(Henshaw, 1998), many women may not be aware of their
status until well into the critical weeks of organogenesis
by which time inadvertent exposure may already have
occurred.
The effects of NSAID exposure on the fetus toward
the end of pregnancy are well documented, causing
premature closure of the ductus arteriosus (Moise, 1993),
and patent ductus arteriosus (Suarez et al., 2002). There
is however much controversy over the risk between
maternal exposure and other congenital anomalies.
Animal studies have shown that aspirin may increase
the risk of congenital anomalies (Cook et al., 2003).
Data from human studies are conflicting. A meta-
analysis of studies of first trimester exposure to aspirin
showed that the pooled results of case-control studies
Published online in Wiley InterScience (www.interscience.wiley.com)
DOI: 10.1002/bdrb.20085
Contract Grant sponsors: Les Fonds de la Recherche en Sante du Quebec
(FRSQ); the Reseau Quebecois de recherche sur lusage des medicaments
(RQRUM); Reseau FRSQ for the Wellbeing of Children.
*Correspondence to: Dr. Anick Berard, Sainte-Justine Hospital, Research
Center 3175, chemin de la Cote-Ste-Catherine, Montreal, Quebec H3T
1C5, Canada. E-mail: anick.berard@umontreal.ca
Received 29 May 2006; Accepted 8 July 2006
The design and conduct of the study; collection, management, analysis,
and interpretation of the data; and preparation, review, or approval of the
manuscript was completely independent of the sponsor.
Birth Defects Research (Part B) 77:268279 (2006) & 2006 Wiley-Liss, Inc.
indicated a higher risk of anomalies in exposed infants
than in the non-exposed ones, whereas analysis of
the results of the cohort and randomized controlled
studies indicated no such effect (Kozer et al., 2003).
Increases in the risk of specific types of anomalies have
also been suggested. A case-control study (Martinez-
Frias et al., 1997) of 45 cases of gastroschisis and 690
controls found a three-fold increase in the risk of
gastroschisis from prenatal exposure to salicylates, after
adjustment for maternal age and smoking. A Swedish
birth registry study (Ericson and Kallen, 2001) comparing
women who used NSAIDs in early pregnancy to those
who did not, found a significantly increased risk of
cardiac defects; most were ventricular and atrial septal
defects. Cardiac defects were also the subject of the
Baltimore-Washington study (Ferencz et al., 1997) in
which 3377 cases with cardiovascular defects were
compared to 3572 controls from the general population.
They found ibuprofen to be a risk factor for cardio-
vascular defects. However, a case-control study examin-
ing muscular ventricular septal defects (Cleves et al.,
2004) and a cohort study of all congenital anomalies
(Nielsen et al., 2001) did not detect any significant
associations with NSAID use. Methodologic choices such
as different underlying cohorts, validity of comparison
groups, and lack of power could partly explain the
different study results.
As the average age of first-time mothers increases,
more women are likely to take NSAIDs for conditions
such as joint and musculoskeletal pain. For this reason,
even small increases in risk may have a significant
impact on public health. Well-powered population-based
studies are necessary to evaluate such risks. We
conducted a nested case-control study on the risk of
congenital anomalies among women who took NSAIDs
during early pregnancy.
MATERIALS AND METHODS
Data Sources
We used three administrative databases of the Pro-
vince of Quebec; La Regie de lAssurance Maladie du
Quebec (RAMQ), Med-E

cho, and Le fichier des evene-

ments demographiques du Quebec (birth and death
registries) of lInstitut de la Statistique du Quebec (ISQ)
(Fig. 1). The RAMQ database contains information on
medical services (diagnoses and procedures) received by
all Quebec residents. All diagnoses are classified accord-
ing to the International Classification of Diseases, 9th
revision, (ICD-9). Although RAMQ covers all Quebec
residents for the cost of physician visits, hospitalizations,
and procedures, it only covers a portion of residents for
the cost of prescribed medications. The RAMQ drug plan
covers individuals 65 years and older, recipients of social
assistance (welfare beneficiaries), and workers and their
families (adherents) who do not have access to a private
insurance program, accounting for approximately 43%
of the overall Quebec population (Regie de lassurance
maladie du Quebec., 1997). It is also estimated that 30%
of women between 1545 years of age in Quebec are
covered by the RAMQ drug plan for their medication.
The Med-E

cho database is a provincial database that

records acute care hospitalization data for all Quebec
residents; it also records gestational age for planned
abortions, miscarriages and deliveries. Le fichier des
evenements demographiques du Quebec (ISQ) provides
demographic variables on the mother, father and baby as
RAMQ
Medications dispensed
Date, drug, form, duration
Medical services:
Diagnostic codes (ICD-9)
Procedure codes
MedEcho
Hospitalisations &ED
visits:
1
ry
& 2
ry
diagnostic codes
(ICD-9)
ISQ
Mother: Demographicdata
Baby: DoB, gender, birth
weight,gestationalage
Medication & Pregnancy
Registry
Babys name
Mothers name
Family name
NAM
Linked by:
Fig. 1. Linkage between administrative databases used in study. Abbreviations: DoB, date of birth; ED, emergency department; ICD-9,
International Classification of Diseases, 9th revision; NAM, Numero dassurance maladie (unique personal identification number).
269 NSAID USE DURING PREGNANCY AND BIRTH DEFECTS
Birth Defects Research (Part B) 77:268279, 2006 DOI 10.1002/bdrb
well as birth weight and gestational age for live births
and stillbirths. Linkage between RAMQ and Med-E

cho
data was done using patients Numero dassurance
maladie (NAM) that is a unique personal identifier for
each person living in Quebec. Linkage between RAMQ
and ISQ data was done using the mothers and babies
names, family names, and dates of birth.
The RAMQ and Med-E

cho databases have often been

used in the past for epidemiologic research (Garbe et al.,
1997; Avorn et al., 1998; Blais et al., 2000). The ISQ
database has also been used in epidemiologic studies
(Roy et al., 2004). Data recorded in the medication
database of the RAMQ have been suitably evaluated and
found to be comprehensive and valid (Tamblyn et al.,
1995). The same has been found for medical diagnoses
recorded in the Med-E

cho database (Levy et al., 1995).

Population
The RAMQ, Med-E

cho and ISQ databases were linked

together to create the Medication and Pregnancy
registry that contains data on all pregnancies that
occurred in Quebec between 01/01/1997 and 06/30/
2003. This population-based pregnancy registry is com-
posed of women with a diagnosis or procedure code
related to pregnancy.
Within the Medication and Pregnancy registry,
women meeting the following eligibility criteria were
included in this present study. The women had to: 1)
be between 1545 years of age at the date of entry in
the registry defined as the first day of gestation (GD 1); 2)
be continuously insured by the RAMQ drug plan for at
least 12 months before GD 1, and during their pregnancy;
3) have filled at least one prescribed medication other
than NSAIDs during their pregnancy; and 4) have had
a singleton live birth. By only including women who
had filled at least one non-NSAID prescription, all the
women in our registry would have had to attend a
physician for medication during pregnancy, and thus
would have had the opportunity to request a prescription
for an NSAID. This would have reduced the likelihood of
significant OTC purchase of NSAIDs within this popula-
tion. Furthermore, a cohort made up of women who had
filled at least one prescription would minimize any
potential health differences that may arise where there
are some women who take prescribed medications and
some who dont. If a woman had more than one
pregnancy between 19972003, the first pregnancy meet-
ing the eligibility criteria was included for analysis.
Exclusions were applied to users of aspirin, indometha-
cin and diclofenac/misoprostol (Arthrotec
s
, Pharmacia
Canada, Inc, Ontario, Canada). Aspirin and indometha-
cin have specific indications in high-risk pregnancies
[e.g., prevention of intrauterine growth retardation
(Leitich et al., 1997), prevention of pregnancy induced
hypertension and associated morbidities (Imperiale and
Petrulis, 1991), polyhydramnios (Cabrol et al., 1996) or to
prevent premature birth (Gerson et al., 1990)] so their
inclusion may introduce a confounding effect. Aspirin
also differs from other NSAIDs by irreversibly inhibi-
ting cyclooxygenase (COX) 1 and 2 (Cronstein, 2002).
Arthrotec
s
is contra-indicated in pregnancy as it may
induce abortions (LAssociation des pharmaciens du
Canada, 2003).
Exposure to NSAIDs
Because we were interested in major congenital
anomalies, the principal variable modeled in our analysis
was a filled prescription for an NSAID in the first
trimester or a filled prescription before the first trimester,
but where the duration of treatment lasted into the first
trimester period; women with such prescriptions are
referred to in this text as having filled a prescription for
NSAIDs in the first trimester. All other women were
grouped under no filled prescriptions for NSAIDs in the
first trimester. NSAIDs considered were those reim-
bursed by the RAMQ at the time of the study, which
included naproxen, ibuprofen, rofecoxib, diclofenac,
celecoxib, flurbiprofen, nabumetone, etodolac, ketopro-
fen, piroxicam, diflunisal, meloxicam, tiaprofenic acid,
and sulindac.
Case Selection
Within our study population, we identified all cases of
congenital anomalies by searching the RAMQ and Med-
E

cho databases for ICD-9 codes 740.0-759.9 recorded

within the first 12 months of each infants life, to allow
for delayed detections or registrations. Our primary
outcome of interest, which was guided by studies
published previously (Ericson and Kallen, 2001; Kallen
and Otterblad, 2003; Bateman et al., 2004), was cardiac
septal closure and related anomalies (ICD-9: 745.0-745.9,
including bulbus cordis anomalies, anomalies of cardiac
septal closure, transposition of great vessels, and
endocardial cushion defects). We also investigated other
specific congenital anomalies related to major organ
systems that are common in the population (New York
State Department of Health, 2005) including: congenital
anomalies of the heart, ICD-9 code 746.0-746.9; anomalies
of the central cardiovascular system including: patent
ductus arteriosus, co-arctation of the aorta, anomalies of
the pulmonary artery, anomalies of the great veins and
primary pulmonary hypertension of the newborn, ICD-9
code 747.0-747.4 and 747.8; anomalies of the respiratory
system, ICD-9 code 748.0-748.9; anomalies of the central
nervous system, ICD-9 code 742.0-742.9; and anomalies
of the limbs and musculoskeletal system, ICD-9 code
754.0-756.9. Anomalies of the respiratory system, though
less common, was selected since certain lung anomalies
such as cystic adenomatoid malformations and pulmon-
ary sequestration are associated with cardiac defects
(Sohaey and Zwiebel, 1996). The cardiac defects were
considered according to their diagnostic code subgroups
to allow the separate examination of anomalies related to
cardiac septal closure such as ventricular septal defects
(Bateman et al., 2004), and other cardiac defects (Ericson
and Kallen, 2001). Anomalies of the central cardiovas-
cular system were considered in order to examine
outcomes known to be associated with maternal NSAID
use such as patent ductus arteriosus and primary
pulmonary hypertension (Zenker et al., 1998; Suarez
et al., 2002).
Control Selection
Up to 10 controls (infants without congenital anoma-
lies) per case were randomly selected within our study
population. Controls were matched to cases on maternal
age (72 years), region of residence (urban vs. rural) and
270 OFORI ET AL.
Birth Defects Research (Part B) 77:268279, 2006 DOI 10.1002/bdrb
diabetes status. Diabetes status was defined as a
diagnosis of diabetes in the year before pregnancy
(ICD-9; 250.0-250.9, 271.4, 790.2) or gestational diabetes
diagnosed at Z26 weeks of pregnancy (ICD-9; 648.0,
648.8), or the dispensing of medications for diabetes in
the 12 months before and during pregnancy, including
insulins and oral hypoglycemics (American Hospital
Formulary System [AHFS] classes; 68:20.08, 68:20.20,
68:20.92). Diabetes is a known risk factor for congenital
anomalies, and gestational diabetes in a primigravida
may be a sign of undiagnosed diabetes. We also selected
controls who had conceived in the same calendar period
as cases by matching on GD 1 (722 weeks) to adjust
for environmental risks that may change over time and
yearly trends in birth defects detection. A 22-week period
was selected to ensure that the pregnancies of cases and
controls overlapped, maximizing the pool of potential
controls.
Covariates
Healthcare utilization variables (RAMQ/Med-E

cho
databases) such as hospitalization and emergency de-
partment (ED) visits, medical visits and number of
prescribers consulted were selected as markers of health
status; these were measured in the 12 months before and
during pregnancy. We determined the presence of the
following chronic co-morbidities (using the RAMQ/
Med-E

cho databases) and the use of specific medications:

rheumatoid arthritis (ICD-9; 714.0) diagnosed at any time
before pregnancy or the filling of prescriptions for any
drugs of the disease modifying anti-rheumatic drugs
group (DMARDs) in the year before pregnancy; the
filling of prescriptions for DMARDs during pregnancy;
the diagnosis of hypothyroidism (ICD-9; 244.0) or the
filling of prescriptions for thyroid medications (AHFS
68:36.04) at any time before or during pregnancy; the
diagnosis of chronic hypertension in the year before and
during pregnancy or of gestational hypertension. The
DMARD group included hydroxychloroquine, sulfasala-
zine, auranofin, aurothiomalate, aurothioglucose, peni-
cillamine, methotrexate, azathioprine, and cyclosporine.
Chronic hypertension was identified by the presence
of ICD-9 codes 401.0405.9, 362.1, 416.0, 437.2, 796.2 or
the filling of prescriptions for any antihypertensive
drugs under the AHFS class 24:08. Gestational hyper-
tension was defined as a diagnosis made at Z20 weeks
of pregnancy and was identified with ICD-9 codes 642.0-
642.9. Other variables were considered during pregnancy
such as the number of prenatal visits, filled prescriptions
for co-medications during pregnancy, and filled pre-
scriptions for oral corticosteroids (including cortisone,
hydrocortisone, fludrocortisone, prednisone, predniso-
lone, dexamethasone) during pregnancy. We also con-
sidered prescriptions filled for drugs in the first trimester
that are known to be teratogenic such as carbamazepine,
phenytoin, valproic acid, lithium, acitretin, isotretinoin,
HMG CoA reductase inhibitors, antineoplastic agents
(AHFS class 10:00), leflunomide, and the androgens
(including danazol, testosterone and methyltestosterone).
Where prescribed medications were considered, we
accounted for items dispensed outside the period of
interest (12 months before or during pregnancy) but
whose duration overlapped into that period. We also
determined the following socio-economic variables from
the RAMQ/ISQ databases: maternal marital status
(living alone vs. co-habiting), number of years education,
and RAMQ insurance status (welfare beneficiary vs.
adherent).
Statistical Analysis
Descriptive statistics were used to compare those
women who filled prescriptions for NSAIDs in the first
trimester to those who did not. To illustrate the
anomalies detected in our population, all congenital
anomalies were also listed according to these two groups.
Univariate and multivariate conditional logistic regres-
sion models were built to quantify the independent
association between the filling of prescriptions for
NSAIDs in the first trimester and congenital anomalies,
adjusting for clinically important confounders and socio-
economic variables. Where a socioeconomic variable
such as years of education was missing, this was dealt
with by mean imputation (Lynch, 2003). Where the
variable missing was marital status (living alone or
cohabiting) we assumed the status of least risk (i.e.,
cohabiting). First, analyses were carried out using all
congenital anomalies combined. Second, analyses were
done stratifying on specific anomalies.
Sensitivity analyses were done excluding women who
gave birth to an infant with an anomaly recorded as
unspecified (Appendix 1). The analyses were repeated
for all congenital anomalies, anomalies related to cardiac
septal closure, and anomalies of the respiratory system.
Given that some of the anomalies included may be
considered as minor (New York State Department of
Health, 2005) we examined the strength of the association
after excluding those codes for minor anomalies
(Appendix 2), in addition to those for unspecified
anomalies (Appendix 1). Finally, we also considered a
sub-analysis by altering two assumptions. First, those
women who filled a prescription before the conception
date were assumed not to have taken sufficient doses of
NSAID to last into the pregnancy period and were
excluded. Second, cases and controls were matched on a
stricter time window for GD 1 (712 weeks) and so
increasing the overlap of the first trimester period.
Odds ratios (OR) with 95% confidence intervals (CI)
were calculated using conditional logistic regression
analysis by applying the SAS Command proc phreg
with specification for discrete ties. All analyses were
conducted in SAS System for Windows Version 8.02 (SAS
Institute Inc, North Carolina, USA).
RESULTS
Within the Medication and Pregnancy registry, 36,387
women met the inclusion criteria and were considered
for analysis: 1056 filled a prescription for at least one
NSAID during the first trimester of pregnancy (2.9%).
The five most commonly filled prescriptions were for
naproxen (35%), ibuprofen (26%), rofecoxib (15%),
diclofenac (9%), and celecoxib (9%), representing 95%
of all NSAID prescriptions. Half of the women filled
prescriptions with an estimated total duration of r10
days. For 95% of the women the estimated total duration
was of r40 days. Most prescriptions were filled within
the first 8 weeks after GD 1; 65.4% of prescriptions were
filled between GD 160 and 11.3% between days 6198.
271 NSAID USE DURING PREGNANCY AND BIRTH DEFECTS
Birth Defects Research (Part B) 77:268279, 2006 DOI 10.1002/bdrb
There were 23.3% of prescriptions that were filled
between 33 days before and the day before GD 1. The
characteristics of those women who filled a prescription
for an NSAID in the first trimester versus those who did
not are presented in Table 1. There is evidence of greater
co-morbidity among the former group of women.
Women who filled prescriptions for NSAIDs in the first
trimester were more likely to be welfare beneficiaries and
living alone (Po0.01). With the exception of prenatal
visits, it is also evident that these women had a greater
use of the health services in the year before and during
pregnancy. Table 2 lists the 12 prescriptions filled most
frequently for medications other than NSAIDs for those
women who also filled NSAID prescriptions in the first
trimester, and those who did not. The prevalence of other
medications prescribed during pregnancy was compar-
able between study groups with penicillins (11.5% vs.
12.8% for those with first trimester NSAID prescriptions
and those without, respectively) and antiemetics (9.9%
vs. 12.7% for those with first trimester NSAID prescrip-
tions and those without, respectively) being the most
frequently prescribed.
We identified 93 births with congenital anomalies in
1056 women (8.8%) who filled a prescription for NSAIDs
Table 1
Characteristics of the Study Population
Filled Prescriptions for
NSAIDs in First Trimester
h
(n51056)
No NSAID
Prescriptions in
First Trimester (n535,331)
Maternal age at conception, mean (7SD) 27.3 (76.29) 27.1 (75.70)
Socioeconomic information
Urban dwellers, n (%) 807 (76.42) 26,701 (75.57)
Welfare beneficiaries, n (%)
a
467 (44.22) 12,266 (34.72)
Living alone, n (%)
a
257 (24.34) 6880 (19.48)
Years of education
a
06, n (%) 33 (3.36) 606 (1.84)
711, n (%) 486 (49.44) 13,774 (41.76)
1214, n (%) 308 (31.33) 11,682 (35.41)
Z15, n (%) 156 (15.87) 6925 (20.99)
Maternal co-morbidities
Diabetes status,
c
n (%)
a
124 (11.74) 3287 (9.30)
Rheumatoid arthritis or DMARD use in year before pregnancy,
d
n (%)
a
12 (1.14) 121 (0.34)
Chronic or gestational hypertension or use of anti-hypertensives,
e
n (%)
b
94 (8.90) 2519 (7.13)
Hypothyroidism before or during pregnancy,
f
n (%) 60 (5.68) 1792 (5.07)
Prescription for DMARDs during pregnancy, n (%)
a
5 (0.47) 24 (0.06)
No. of different medications (excluding NSAID), mean (7SD)
a
3.72 (7 3.16) 2.46 (71.96)
Prescription for oral corticosteroids during pregnancy, n (%)
a
108 (10.23) 2774 (7.85)
Prescription for teratogenic drugs
g
during first trimester, n (%)
a
31 (2.94) 471 (1.33)
Health services utilization
Prenatal visits, mean (7SD) 9.08 (73.47) 9.10 (73.55)
Medical visits in year before pregnancy, mean (7SD)
a
8.92 (77.20) 6.36 (76.07)
No. of prescribers in year before pregnancy, mean (7SD)
a
3.32 (72.31) 2.21 (71.90)
No. of prescribers during pregnancy, mean (7SD)
a
3.14 (71.77) 1.90 (71.26)
ED visit or hospitalization in year before pregnancy, n (%)
a
196 (18.56) 5453 (15.43)
ED visit or hospitalization during pregnancy, n (%)
a
926 (87.69) 32,048 (90.71)
All congenital anomalies in babies n (%)
a
93 (8.81) 2478 (7.01)
Gestational age at delivery
o37 weeks gestation, n (%)
b
85 (8.05) 2248 (6.36)
Birth weight of baby
o2500 g, n (%)
a
85 (8.06) 1897 (5.38)
Gender of baby, male n (%) 555 (52.56) 18,036 (51.05)
Abbreviations: DMARDs, disease modifying anti-rheumatic drugs; ED, emergency department; NSAID, non-steroidal anti-
inflammatory drug; SD, standard deviation.
a
Po0.01.
b
Po0.05.
c
Diabetes status defined as a diagnosis of chronic diabetes in the year before pregnancy (ICD-9; 250.0-250.9, 271.4, 790.2) or gestational
diabetes diagnosed at Z26 weeks of pregnancy (ICD-9; 648.0, 648.8), or the dispensing of medications for diabetes in the 12 months
before and during pregnancy.
d
Rheumatoid arthritis (ICD-9 714.0) diagnosed at any time before pregnancy or DMARDs dispensed in the year before pregnancy.
e
Chronic hypertension identified by; ICD-9 codes 401.0405.9, 362.1, 416.0, 437.2, 796.2 or the filling of prescriptions for any
antihypertensive drugs under the AHFS class 24:08. Gestational hypertension defined as a diagnosis made at Z20 weeks of pregnancy
and identified with ICD-9 codes 642.0-642.9.
f
Hypothyroidism defined as ICD-9; 244.0 or the use of thyroid medication (AHFS 68:36.04) at any time before or during pregnancy.
g
Teratogenic drugs considered: carbamazepine, phenytoin, valproic acid, lithium, acitretin, isotretinoin, HMG CoA reductase inhibitors,
antineoplastic agents, leflunomide, and androgens.
h
Includes women who filled a prescription before the first trimester, but where the duration of treatment over-lapped into the first
trimester.
272 OFORI ET AL.
Birth Defects Research (Part B) 77:268279, 2006 DOI 10.1002/bdrb
in the first trimester compared to 2478 births in the 35,331
women (7.0%) who did not. The proportion of infants
with multiple congenital anomalies amongst those filling
an NSAID prescription in the first trimester and those
who did not was 16.1% versus 14.2%, respectively. A
description of all the congenital anomalies detected in the
study population is shown in Table 3. As mentioned
previously, our primary outcome of interest were the
anomalies related to cardiac septal closure (ICD-9 745).
We also chose to specifically examine common birth
defects grouped by organ system: anomalies of the
musculoskeletal system, anomalies of the central nervous
system; and anomalies that have been associated with
NSAID use: other congenital anomalies of the heart,
anomalies of the central cardiovascular system; and also
anomalies of the respiratory system that have been
associated with congenital anomalies of the heart.
The adjusted OR for any congenital anomaly asso-
ciated with first trimester NSAID exposure was 2.21
(95% CI 51.722.85) (Table 4). The adjusted OR for the
anomalies related to cardiac septal closure was 3.34 (95%
CI 51.875.98), and for anomalies of the respiratory
system 9.55 (95% CI 53.0829.63) (Table 5). Among the
anomalies related to cardiac septal closure, 61% were
atrial septal defects, 31% ventricular septal defects, 2%
endocardial cushion defects, and the remaining 6% being
either transposition of great vessels, common ventricle,
teratology of Fallot or anomalies of the common truncus.
Twenty-two infants (5%) were recorded as having both
an anomaly related to cardiac septal closure and a
chromosomal anomaly, only two of whom were among
those whose mother filled an NSAID prescription in the
first trimester. We found that the majority of respiratory
anomalies were coded as unspecified anomaly of the
respiratory system (ICD-9 748.9), n564 (83%), followed
by anomalies of the larynx, trachea, and bronchus (8%),
atresia/stenosis of nares (5%), anomalies of the nose
(3%), and agenesis/hypoplasia/dysplasia of the lung
(1%). We examined associations of first trimester NSAID
exposure with other important anomalies (Table 5). There
were positive but nonsignificant associations with con-
genital anomalies of the heart, central nervous system,
and musculoskeletal system, and a negative but non-
significant association with anomalies of the central
cardiovascular system. Finally, we detected five cases of
gastroschisis, a rare anomaly associated with maternal
NSAID use but all cases were among those who did not
fill any NSAID prescription in the first trimester.
Comparisons of specific types of NSAID prescribed in
the first trimester of pregnancy was done for all
congenital anomalies, anomalies related to cardiac septal
closure, and anomalies of the respiratory system, all of
which were significant in the combined NSAID prescrip-
tion analyses (Table 6). We detected a significant
association between prescriptions for ibuprofen in the
first trimester and any congenital anomalies (Po0.01).
No other prescriptions for specific NSAIDs in the first
trimester was associated with either anomalies related
to cardiac septal closure or anomalies of the respiratory
system.
Sensitivity Analysis
When we excluded diagnostic codes for unspecified
anomalies (Appendix 1), there remained 62 anomalies
(5.87%) among those who filled a prescription for an
NSAID in the first trimester compared to 1578 (4.47%)
among those who did not. In this analysis the adjusted
OR for any congenital anomaly was 1.99 (95% CI 5
1.452.74). When we considered the anomalies related
to cardiac septal closure, the adjusted OR was 3.42
(95% CI 51.906.14). However, once we restricted cases
of respiratory anomalies only to those that were speci-
fied, we detected just one event (0.09%) among those
who filled an NSAID prescription in the first trimester,
and 12 in those who did not (0.03%) (P40.05). After
Table 2
Top Medications Prescribed Most Frequently During Pregnancy
a
Frequency (%)
Drug Group
b
Filled Prescriptions for NSAIDs
in First Trimester
c
(n51056)
No NSAID Prescriptions in
First Trimester (n535,331)
Acetaminophen products 5.86 3.04
Antacids 3.43
Antiemetics 9.87 12.73
Corticosteroids: inhaled, oral, parenteral 4.04 3.49
Diagnostic reagents for diabetes 3.26 3.72
Iron supplements 2.62 3.99
Macrolide antibiotics 3.57 3.52
Oral contraceptives 3.15 3.55
Penicillins 11.55 12.85
Sympathomimetics, long- and short-acting beta-agonists 4.80 5.26
Topical antifungals 5.10 6.27
Topical steroids 4.74 5.98
Eye, ear, nose corticosteroids 2.45
a
Excluding NSAIDs.
b
Drug groups are defined according to the AHFS classes. In calculating frequencies, exposure to each class is counted only once per
woman.
c
Includes women who filled a prescription before the first trimester, but where the duration of treatment over-lapped into the first
trimester.
273 NSAID USE DURING PREGNANCY AND BIRTH DEFECTS
Birth Defects Research (Part B) 77:268279, 2006 DOI 10.1002/bdrb
the exclusion of diagnostic codes for minor anomalies
(Appendix 2) in addition to those for unspecified
anomalies, there were 55 anomalies (5.21%) in those
filling a prescription in the first trimester and 1385
(3.92%) in those who did not. The adjusted OR for any
major specified congenital anomalies was 2.23 (95%
CI 51.613.09); the OR for anomalies related to cardiac
septal closure remained unchanged. The adjusted OR
after excluding women with prescriptions for NSAIDs
dispensed before GD 1, and by matching GD 1 on a
narrower time window of 712 weeks, was 1.94 (95%
CI 51.492.56) for any congenital anomalies and 2.55
(95% CI 51.374.73) for anomalies related to cardiac
septal closure.
DISCUSSION
Our analysis of data from the Medication and
Pregnancy registry suggests that women who fill
prescriptions for NSAIDs in the first trimester of
pregnancy may be at greater risk of having children
with congenital anomalies than those who do not. The
strongest and most consistent findings were seen with
the anomalies related to cardiac septal closure. Although
there are several diagnoses falling under the ICD-9 code
745, nearly all outcomes recorded were for ventricular
and atrial septal defects.
Very few toxicologic studies on non-aspirin NSAIDs
have been carried out. One such study (Cappon et al.,
2003) investigated the relative contribution of COX1 and
COX2 on NSAID-induced developmental toxicity by
dosing rats and rabbits during sensitive windows for
cardiovascular development and midline closure. The
induction of developmental defects was associated with
compounds that selectively inhibited COX1, suggesting
that NSAIDs with a high COX1/COX2 ratio may be
involved in the disruption of heart development. To
determine whether the expression pattern for COX within
embryos was consistent with the model that teratogenicity
was mediated through direct interactions of this enzyme
with NSAIDs, mRNAexpression of COX1 and COX2 were
analyzed in the rat embryo (Streck et al., 2003). Although
the expression of COX2 mRNA was undetectable during
the critical period for organogenesis, COX1 mRNA was
expressed in embryos throughout the sensitive portion of
organogenesis, indicating that COX1 was a viable candi-
date for a molecular mediator of NSAID teratogenic
activity within embryonic tissue. We were able to detect a
significant association between first trimester exposure
to ibuprofen and any congenital anomalies. Ibuprofen is
known to have high COX1 inhibitory activity (Van Hecken
et al., 2000) and has been associated with congenital
anomalies in another study (Ferencz et al., 1997).
For many years prostaglandins, the products of
cyclooxygenase enzyme activity, have been recognized
as key molecules in reproductive biology (Jabbour et al.,
2006). They have also been implicated in modulating
angiogenesis and vascular function, allowing the deliv-
ery of oxygen and nutrients to tissues, and are involved
in endothelial cell sprouting (Namkoong et al., 2005). It is
conceivable that NSAID mediated teratogenicity may
be a result of vascular disruptions, given the relationship
between COX, prostaglandins, and their vascular and
endothelial effects. One proposal (Clark, 1996) for the
Table 3
Description of All Congenital Anomalies Detected in the Study Population According to Prescription Group
Congenital Anomalies
Congenital Anomaly (ICD-9 code)
Filled Prescriptions for NSAIDs
in First Trimester
b
(n5115)
a
n (%)
No NSAID Prescriptions in First
Trimester (n52946)
a
n (%)
Anencephalus (ICD-9; 740) 0 0
Anomaly of ear, face, neck (ICD-9; 744) 1 (0.87) 46 (1.56)
Anomaly of eye (ICD-9; 743) 6 (5.22) 170 (5.77)
Anomaly of genital organs (ICD-9; 752) 6 (5.22) 234 (7.94)
Anomaly of respiratory system (ICD-9; 748) 9 (7.83) 67 (2.27)
Anomaly of urinary system (ICD-9; 753) 2 (1.74) 53 (1.80)
Anomaly of the integument (ICD-9; 757) 0 31 (1.05)
Bulbus cordis anomalies & anomalies of cardiac septal
closure (ICD-9; 745)
21 (18.26) 414 (14.05)
Chromosomal anomalies (ICD-9; 758) 4 (3.48) 60 (2.04)
Cleft palate and lip (ICD-9; 749) 2 (1.74) 46 (1.56)
Musculoskeletal deformity (ICD-9; 754) 13 (11.30) 490 (16.63)
Spina bifida (ICD-9; 741) 0 9 (0.30)
Other anomaly of circulatory system (ICD-9; 747) 9 (7.83) 120 (4.07)
Other anomaly of heart (ICD-9; 746) 3 (2.61) 122 (4.14)
Other anomaly of nervous system (ICD-9; 742) 3 (2.61) 118 (4.00)
Other anomaly of digestive system (ICD-9; 751) 4 (3.48) 63 (2.14)
Other anomalies of limbs (ICD-9; 755) 3 (2.61) 96 (3.26)
Other anomaly of upper alimentary tract (ICD-9; 750) 5 (4.35) 155 (5.26)
Other musculoskeletal anomalies (ICD-9; 756) 6 (5.22) 317 (10.76)
Other & unspecified anomalies (ICD-9; 759) 18 (15.65) 335 (11.37)
a
Because an infant may be diagnosed with multiple anomalies, the total number of anomalies does not equal the number of infants with
anomalies (n59312478).
b
Includes women who filled a prescription before the first trimester but where the duration of treatment over-lapped into the first
trimester.
274 OFORI ET AL.
Birth Defects Research (Part B) 77:268279, 2006 DOI 10.1002/bdrb
Table 4
Case-Control Analyses of the Association Between Women Who Filled Prescriptions for NSAIDs in First Trimester and
the Risk of Any Congenital Anomaly
All Congenital Anomalies (ICD-9
740.0-759.9) Cases (n52571)
Variable
Crude OR
(95% CI)
f
Adjusted OR
(95% CI)
f
Filled NSAID prescriptions during first trimester (y/n) 2.58 (2.043.26) 2.21 (1.722.85)
Filled NSAID prescriptions in second/third trimester (but not in first) (y/n) 0.78 (0.421.45) 0.61 (0.291.26)
Welfare beneficiary (y/n) 1.06 (0.971.15) 0.87 (0.780.96)
Education (years) 1.01 (0.991.02) 1.02 (1.011.04)
Living alone (y/n) 1.82 (1.632.03) 1.64 (1.451.86)
Gestational age at delivery (weeks) 0.70 (0.680.72) 0.70 (0.680.72)
Rheumatoid arthritis
a
or DMARD
b
use before pregnancy (y/n) 0.42 (0.200.90) 0.20 (0.050.74)
Prescription for DMARDs
b
during pregnancy (y/n) 1.57 (0.663.72) 2.18 (0.499.71)
Prescription for oral corticosteroids during pregnancy (y/n) 1.19 (1.031.37) 1.06 (0.911.24)
Prescription for teratogenic drugs
c
during first trimester (y/n) 1.55 (1.142.11) 1.55 (1.112.17)
Hypothyroidism
d
before or during pregnancy (y/n) 0.87 (0.721.04) 0.79 (0.650.95)
Chronic or gestational hypertension or prescription for antihypertensives
e
(y/n) 1.71 (1.481.98) 1.32 (1.131.55)
No. medical visits in year before pregnancy 1.00 (0.991.01) 0.98 (0.970.99)
No. prescribers in year before pregnancy 1.09 (1.071.11) 1.13 (1.101.61)
No. prescribers during pregnancy 1.09 (1.051.12) 1.00 (0.941.06)
No. prenatal visits 0.98 (0.970.99) 1.01 (1.001.02)
No. co-medications other than NSAIDs during pregnancy 1.05 (1.031.07) 1.02 (0.981.06)
ED visit or hospitalization in year before pregnancy (y/n) 0.78 (0.700.87) 0.76 (0.670.86)
ED visit or hospitalization during pregnancy (y/n) 0.97 (0.841.13) 0.80 (0.680.93)
Abbreviations: CI, confidence interval; ED, emergency department; OR, odds ratio; y/n, yes/no.
a
Rheumatoid arthritis (ICD-9 714.0) diagnosed at any time before pregnancy or DMARDs dispensed in year before pregnancy.
b
DMARD, disease modifying antirheumatic drug, including: hydroxychloroquine, sulfasalazine, auranofin, aurothiomalate,
aurothioglucose, penicillamine, methotrexate, azathioprine, and cyclosporine.
c
Teratogenic drugs considered: carbamazepine, phenytoin, valproic acid, lithium, acitretin, isotretinoin, HMG CoA reductase inhibitors,
antineoplastic agents, leflunomide, and androgens.
d
Hypothyroidism defined as; ICD-9; 244.0 or the use of thyroid medication (AHFS 68:36.04) at any time before or during pregnancy.
e
Chronic hypertension identified by; ICD-9 codes 401.0405.9, 362.1, 416.0, 437.2, 796.2 or prescriptions for any antihypertensive drugs
under the AHFS class 24:08. Gestational hypertension defined as a diagnosis made at Z20 weeks of pregnancy and identified with
ICD-9 codes 642.0-642.9.
f
Crude and adjusted OR are matched for maternal age, region of residence, diabetes status, and calendar year. OR are adjusted for all
variables listed.
Table 5
Case-Control Analyses of the Association Between Women Who Filled Prescriptions for NSAIDs in First Trimester and
the Risk of Other Specific Congenital Anomalies
NSAID Prescription During First
Trimester (Yes vs. No)
Congenital Anomaly (ICD-9 code)
Crude OR
(95% CI)
Adjusted
a
OR
(95% CI)
Anomalies related to cardiac septal closure including TGV, ECD (ICD-9 745) (435 cases) 3.77 (2.236.36) 3.34 (1.875.98)
Anomalies of the respiratory system, (ICD-9 748) (76 cases) 9.75 (3.7125.64) 9.55 (3.0829.63)
Congenital anomaly of the heart, (ICD-9: 746) (125 cases) 1.98 (0.557.14) 1.80 (0.427.70)
Anomalies of the central cardiovascular system, (ICD-9: 7470-747.4 and 747.8) (69 cases) 4.77 (1.3117.33) 0.91 (0.145.86)
Anomalies of the central nervous system, (ICD-9: 742), (121 cases) 1.18 (0.354.02) 1.05 (0.283.90)
Anomalies of the musculoskeletal system, (ICD-9: 754.0-756.9) (884 cases) 1.61 (1.002.58) 1.53 (0.942.49)
Abbreviations: CI, confidence interval; ECD, endocardial cushion defects; ED, emergency department; OR, odds ratio; TGV,
transposition of great vessels.
Crude and adjusted OR are matched for maternal age, region of residence, diabetes status, and first gestational day.
a
Filled NSAID prescriptions in second or third trimester (but not in first), welfare beneficiary, education (yr), living alone, gestational
age at delivery (weeks), rheumatoid arthritis or DMARD use before pregnancy, DMARD use during pregnancy, oral corticosteroid use
during pregnancy, filled prescriptions for a teratogenic drug during first trimester, hypothyroidism before or during pregnancy, chronic,
or gestational hypertension or prescriptions for antihypertensives, no. medical visits in year before pregnancy, no. prescribers in year
before pregnancy, no. prescribers during pregnancy, no. prenatal visits, no. co-medications other than NSAIDs during pregnancy,
ED visit or hospitalization in year before pregnancy, ED visit or hospitalization during pregnancy.
275 NSAID USE DURING PREGNANCY AND BIRTH DEFECTS
Birth Defects Research (Part B) 77:268279, 2006 DOI 10.1002/bdrb
pathogenic mechanism of congenital cardiovascular
anomalies is that forces generated within the developing
heart tube with the initiation of cardiac contractions,
combined with the secondary hemodynamic forces
produced by blood flowing through cardiovascular
structures, influence cell behavior, vessel and chamber
size, and myocardial mass. Two blood flow streams
coursing through the developing fetal heart are respon-
sible for modeling the septal process. After septation, the
fetal heart is a parallel circuit in which resistance is
modulated by prostaglandin agonists and antagonists.
Such alterations in resistance and blood flow are thought
to be responsible for intracardiac changes. It is con-
ceivable that disruptions in the normal resistance of
intracardiac blood flow by inhibitors of prostaglandin
synthesis could result in congenital anomalies such as
ventricular septal defects.
Our study findings regarding the increased risks
associated with specific cardiac anomalies are in
accordance with other epidemiologic studies that have
been published thus far (Ericson and Kallen, 2001; Kallen
and Otterblad, 2003; Bateman et al., 2004). Indeed, the
Swedish birth registry study (Ericson and Kallen, 2001),
using data on first trimester NSAID medication use
collected by interviews during the first prenatal visit
at 1012 weeks of gestation, reported an adjusted OR
for NSAID use and cardiac defects of 1.86 (95% CI 5
1.322.62) (2557 infants whose mothers reported use of
NSAIDs in early pregnancy vs. all 279,734 remaining
births in the registry whose mothers did not report first
trimester use of NSAIDs). The proportion of cardiac
defects in babies of mothers with early NSAID exposure
was 1.6%, most of which were for ventricular or
atrial septal defects, as was found in our study (1.9%).
Our results differ from a recent cohort study (Nielsen
et al., 2001) based on the Danish Birth Registry. In
this study, 1106 women who had taken up prescriptions
for NSAIDs in early pregnancy were compared to a
reference cohort of 17,259 women who were not
prescribed any drugs during pregnancy. Data used
were collected by midwives, physicians attending deliv-
eries, and from hospital discharge records, and was
assembled over a 25-year period. Adjusting for maternal
age, smoking, and birth order, NSAID use was not
associated with congenital anomalies (OR51.27, 95%
CI 50.93.75). The difference between the findings of
our study and theirs can be partly explained by the fact
that the size of our cohort was nearly twice as large as
theirs, giving us more power to detect significant
associations.
We are not aware of associations known previously
between NSAIDs and anomalies of the respiratory
system. Once we excluded anomalies that were defined
as unspecified, there was no longer an association
between NSAID exposure and anomalies of the respira-
tory system. The data available does not permit us to
ascertain the precise nature of the unspecified anomalies
but it is likely that these infants were born with some
clinically important defect. However, this requires further
investigation. As the sensitivity analysis demonstrated,
even after the exclusion of anomalies defined as
unspecified, the overall effect of an increased risk
associated with NSAID prescribing held for the combined
congenital anomalies, and for the anomalies related to
cardiac septal closure. Our results also remained largely
unchanged after a sub analysis excluding women who
had filled their prescriptions before GD1, and using a
tighter time frame on which to match on GD 1. The first
modification was important because of the possibility
that despite the prescribed quantity, women may take
just a few doses of the medication until they feel better.
The latter modification ensured a much better overlap of
the first trimester period between cases and controls.
Proven teratogens do not have the ability to produce
every birth defect (Brent and Beckman, 1994) and indeed
we did not find significant associations with other
congenital anomaly outcomes we examined, irrespective
of the sample size of cases although the associations were
positive. For anomalies of the central cardiovascular
system the association was negative but still non
significant. The direction of the association with anoma-
lies of the central cardiovascular system could be
explained by the adjustment for gestational age in the
model. It would be expected that babies with severe
cardiovascular anomalies such as co-arctation of the
aorta would be born a lot sooner than babies with
musculoskeletal anomalies since cardiovascular function
is critical in maintaining the pregnancy in its latter stages.
Furthermore, certain of these anomalies (e.g., patent
ductus arteriosus) are associated with NSAID use in
the latter stages of pregnancy rather than in the first
trimester (Suarez et al., 2002). One study has found that
Table 6
Congenital Anomalies
a
According to Prescriptions for Specific NSAID Filled During First Trimester
All Congenital Anomalies
(ICD-9 740.0-759.9)
Anomalies Related to Cardiac Septal
Closure Including TGV, ECD (ICD-9 745)
Anomalies of the Respiratory
System (ICD-9 748)
NSAID
b
Yes (n593) No (n5963) Yes (n521) No (n5963) Yes (n59) No (n5963)
Naproxen, n (%) 31 (33.33) 357 (37.07) 8 (38.10) 357 (37.07) 3 (33.33) 357 (37.07)
Ibuprofen, (n, %) 36 (38.71) 246 (25.55)
c
7 (33.33) 246 (25.55) 3 (33.33) 246 (25.55)
Rofecoxib, (n, %) 13 (13.98) 158 (16.41) 3 (14.29) 158 (16.41) 2 (22.22) 158 (16.41)
Diclofenac, (n, %) 6 (6.45) 94 (9.76) 0 94 (9.76) 0 94 (9.96)
Celecoxib, (n, %) 8 (8.60) 94 (9.76) 1 (4.76) 94 (9.76) 0 94 (9.76)
Abbreviations: ECD, endocardial cushion defects; TGV, transposition of great vessels.
a
Sum of the numbers displayed in columns is less than the total number of malformations associated with prescriptions filled for NSAID
in first trimester because not all NSAID drugs are displayed.
b
In the group of women who filled prescriptions for NSAIDs in the first trimester, 41 women filled prescriptions for two different
NSAIDs during the first trimester.
c
Po0.01.
276 OFORI ET AL.
Birth Defects Research (Part B) 77:268279, 2006 DOI 10.1002/bdrb
16% of infants with cardiovascular malformations are
preterm (Tanner et al., 2005), however, there is no
evidence that a shorter gestation causes malformations.
Although we were unable to establish the indications
for NSAID use, we adjusted for prescriptions for
DMARDs during pregnancy, some of which are known
teratogens, and also for prescribing of other common
teratogenic drugs in the first trimester. It is possible that
despite this adjustment residual confounding may still
have been present. However, given the relatively small
number of women (o1.5% of study cohort) on such
medications the impact of any such residual confounding
would be negligible. We were also unable to comprehen-
sively adjust for all co-morbidities. It is evident that
women who filled prescriptions for NSAIDs in the first
trimester also had a greater past and current use of the
health services and this could be related to a range of
pathologies that we were unable to capture. It is possible
that residual confounding related to co-morbidity re-
mained, nevertheless, the co-morbidity adjustments we
were able to make, including hypertension and gesta-
tional hypertension, health service utilization and hy-
pothyroidism are important and a strength of the study.
Hypothyroidism is one of the most common endocrine
disorders found in pregnant women and has been
associated with pregnancy complications such as hyper-
tension, preterm birth, low birth weight, and fetal death
(Casey et al., 2005).
To the extent that measured (and adjusted) confoun-
ders are correlated with unmeasured characteristics,
residual confounding caused by the unmeasured factors
will be reduced or partially adjusted (Schneeweiss, 2006).
It is conceivable that some of the indications for the
prescriptions for NSAIDs may have been responsible for
the observed effects. Viral infections, for example, are
known to be associated with congenital abnormalities
of the heart (March of Dimes, 2006). A fever caused by
viral infection could be an indication for the prescribing
of NSAIDs, and this may result in confounding bias
(confounding by indication). It is evident from Table 2
that the frequency of prescribing of acetaminophen
products was higher among women also prescribed an
NSAID in the first trimester compared to women who
were not, 5.86% versus 3.04%, respectively.
It has long been recognized that the incidences of
various congenital anomalies show geographical, secular,
and seasonal fluctuations; it also varies based on socio-
economic status and maternal age differences (Richards,
1972; Drongowski et al., 1991). We accounted for this by
matching cases to controls on maternal age, region of
residence, and GD 1, as well as on diabetes status that is a
major risk factor for congenital anomalies (Ray et al.,
2001; Clausen et al., 2005).
A recent study (Williams et al., 2004) on the risk factors
for ventricular septal defects revealed that lifestyle
factors such as alcohol and marijuana use significantly
increased the risk for this outcome. Our study lacked
information on alcohol consumption, marijuana/cigar-
ette smoking, and folic acid intake. Although cigarette
smoking may signal risk taking behaviors during preg-
nancy it is not, on its own, associated with an overall
increase in the risk of congenital anomalies, except for
oral cleft (Little et al., 2004) and digital anomalies (Man
and Chang, 2006). In our study, the percentage of cleft lip
and palate among those who filled prescriptions for
NSAIDs in the first trimester and those who did not,
were similar, and there was no association detected
between filling prescriptions for NSAIDs in the first
trimester and anomalies of the musculoskeletal system.
There were also no neural tube defects among those who
filled prescriptions for NSAIDs in the first trimester. This
suggests that smoking and folic acid intake were not
major confounders in this population. We did, however,
adjust for socioeconomic differences that might impact
the above by measuring RAMQ medication insurance
status, education level, and marital status. We also
adjusted for measures of health services utilization that
is an indicator of co-morbidity.
It is possible that there were women whom we
considered as having no filled prescriptions for NSAIDs
in the first trimester but who purchased an NSAID
directly from the pharmacy, resulting in a misclassifica-
tion bias. However, the extent of this bias is likely to be
minimal. First, the number of NSAIDs that are available
without prescription are few, and include only ibuprofen
and aspirin. Second, all the subjects in our study were
users of other prescribed medications and had insurance
cover under the RAMQ medication insurance plan. All
the women would have had the opportunity to receive
any NSAIDs on prescription by having attended a
physician to obtain at least one other drug during
pregnancy, irrespective of whether they eventually
became a case or a control. Similarly, cases and controls
would have been equally likely to purchase an NSAID
OTC, so whilst we acknowledge the possibility of
misclassification, any such misclassification would prob-
ably have been non-differential.
Another possible source of misclassification concerned
the variables for number of years education and marital
status where we found some missing information. The
proportion of missing information, however, was small,
6% for number of years education and o0.1% for marital
status. The possible implication of our method of dealing
with this missing information would be a reduction in
the variance of the variable being imputed with the
consequent shift the effect estimate toward the null.
Among the women who filled prescriptions for
NSAIDs in the first trimester, the majority (65.4%) filled
their prescription within the first 60 days of the beginn-
ing of gestation. The embryonic period, from 1860 days
after conception, is the period when the basic steps in
organogenesis occur (Diav-Citrin and Ornoy, 2000). This
is the period of maximum sensitivity to teratogenicity
since not only are tissues differentiating rapidly, but also
damage to them becomes irreparable. Exposure to
teratogenic agents during this period has the greatest
likelihood of causing a structural anomaly, including
an effect on cardiac organogenesis which takes place up
to the GD 60 (Brent and Beckman, 1994; Diav-Citrin and
Ornoy, 2000).
Despite their importance, pharmaco-epidemiolgic stu-
dies based on administrative databases have been
criticized for the incompleteness of their information on
potential confounders and on certain clinical variables of
interest. As was evident with the outcome for anomalies
of the respiratory system, diagnoses may be grouped
under certain general codes. Nevertheless, in our
sensitivity analysis even after excluding these general
codes, our results remained largely unchanged. Our
study only included women who had live births. With no
277 NSAID USE DURING PREGNANCY AND BIRTH DEFECTS
Birth Defects Research (Part B) 77:268279, 2006 DOI 10.1002/bdrb
access to autopsy and pathology reports, by including
women who had abortions, miscarriages, and fetal
deaths, we could not have been certain that the absence
of a recorded diagnosis truly reflected the absence of a
congenital anomaly. Consideration of only live births
guaranteed that congenital anomalies would be detected
and recorded in the same manner for all study subjects,
avoiding problems of ascertainment bias.
In our study, the confounding effect of specific high-
risk prenatal conditions was reduced by excluding
indomethacin and aspirin users, though not entirely
eliminated since in most cases these drugs would be
prescribed in the latter stages of pregnancy. We acknowl-
edge that an important limitation of our study was the
assumption that those who filled a prescription also had
sufficient exposure to the medicines to induce the effect.
Some doses would have certainly been taken. Absolute
non-compliance with medications obtained by pregnant
women is known to be low, at about 8% (Glover et al.,
2003). However, agents that cause malformations are
known to exhibit threshold phenomena, levels of
exposure above which malformations are produced
(Brent and Beckman, 1994). It was not possible to
establish whether the amount of medicine taken was
above the critical threshold to actually cause the effect.
One important advantage of databases that include the
routine collection of information on dispensed drugs,
including name, dose, and amount dispensed is that they
avoid the limitations associated with the need for long-
term recall (that may result in re-call bias) and allows the
rapid assembly of large cohorts of sufficient power that
would otherwise be costly and time-consuming if done
prospectively. Moreover, such database study in which the
exact timing of the beginning of pregnancy (GD 1, derived
from the MedE

cho and ISQ databases) is known, allows

the effective study of the consequences of medication
use at a time when women may still be unaware of
their pregnancy status. Such database research provides
valuable information in the investigation of associations
that might have an important public health impact.
The potential hazards of taking NSAIDs during
pregnancy are not well publicized. In summary, women
who fill prescriptions for NSAIDs in the first trimester of
pregnancy may be at a greater risk of having children
with congenital anomalies, particularly those related to
cardiac septal closure. This is in accordance with
previous findings but needs to be replicated in other
study populations.
ACKNOWLEDGMENTS
This study was supported by Les Fonds de la Recherche
en Sante du Quebec (FRSQ), the Reseau Quebecois de
recherche sur lusage des medicaments (RQRUM), and
the Reseau FRSQ for the well-being of children. Dr. Anick
Berard is the recipient of a career award from the
Canadian Institutes of Health Research (CIHR)/Health
Research Foundation, and is on the endowment research
Chair of the Famille Louis-Boivin on Medications,
Pregnancy and Lactation at the Faculty of Pharmacy of
the University of Montreal. Dr. Lucie Blais is the recipient
of a career award from the CIHR, and is on the
AstraZeneca endowment research Chair on respiratory
health at the Faculty of Pharmacy of the University of
Montreal.
REFERENCES
Avorn J, Monette J, Lacour A, Bohn RL, Monane M, Mogun H, LeLorier J.
1998. Persistence of use of lipid-lowering medications: a cross-
national study. JAMA 279:14581462.
Bateman DN, McElhatton PR, Dickinson D, Wren C, Matthews JN,
OKeeffe M, Thomas SH. 2004. A case control study to examine the
pharmacological factors underlying ventricular septal defects in the
North of England. Eur J Clin Pharmacol 60:635641.
Blais L, Desgagne A, LeLorier J. 2000. 3-Hydroxy-3-methylglutaryl
coenzyme A reductase inhibitors and the risk of cancer: a nested
case-control study. Arch Intern Med 160:23632368.
Brent RL, Beckman DA. 1994. The contribution of environmental
teratogens to embryonic and fetal loss. Clin Obstet Gynecol 37:
646670.
Cabrol D, Jannet D, Pannier E. 1996. Treatment of symptomatic
polyhydramnios with indomethacin. Eur J Obstet Gynecol Reprod
Biol 66:1115.
Cappon GD, Cook JC, Hurtt ME. 2003. Relationship between cycloox-
ygenase 1 and 2 selective inhibitors and fetal development when
administered to rats and rabbits during the sensitive periods for
heart development and midline closure. Birth Defects Res B Dev
Reprod Toxicol 68:4756.
Casey BM, Dashe JS, Wells CE, McIntire D, Byrd W, Leveno KJ,
Cunningham FG. 2005. Subclinical hypothyroidism and pregnancy
outcomes. Obstet Gynecol 105:239245.
Clark EB. 1996. Pathogenetic mechanisms of congenital cardiovascular
malformations revisited. Semin Perinatol 20:465472.
Clausen TD, Mathiesen E, Ekbom P, Hellmuth E, Mandrup-Poulsen T,
Damm P. 2005. Poor pregnancy outcome in women with type 2
diabetes. Diabetes Care 28:323328.
Cleves MA, Savell VH Jr, Raj S, Zhao W, Correa A, Werler MM,
Hobbs CA. 2004. Maternal use of acetaminophen and non-
steroidal anti-inflammatory drugs (NSAIDs), and muscular
ventricular septal defects. Birth Defects Res A Clin Mol Teratol 70:
107113.
Cook JC, Jacobson CF, Gao F, Tassinari MS, Hurtt ME, DeSesso JM. 2003.
Analysis of the nonsteroidal anti-inflammatory drug literature for
potential developmental toxicity in rats and rabbits. Birth Defects Res
B Dev Reprod Toxicol 68:526.
Cronstein BN. 2002. Cyclooxygenase-2-selective inhibitors: translating
pharmacology into clinical utility. Cleve Clin J Med 69(Suppl):
SI13SI19.
Diav-Citrin O, Ornoy A. 2000. Adverse environment and prevention of
early pregnancy disorders. Early Pregnancy 4:518.
Drongowski RA, Smith RK Jr, Coran AG, Klein MD. 1991. Contribution of
demographic and environmental factors to the etiology of gastro-
schisis: a hypothesis. Fetal Diagn Ther 6:1427.
Ericson A, Kallen BA. 2001. Nonsteroidal anti-inflammatory drugs in
early pregnancy. Reprod Toxicol 15:371375.
Ferencz C, Correa-Villasen or A, Loffredo CA, Wilson PD. 1997. Genetics
and environmental risk factors of major cardiovascular malforma-
tions: the Baltimore-Washington Infant Study: 1981-1989. Armonk,
NY: Futura Publishing.
Garbe E, LeLorier J, Boivin JF, Suissa S. 1997. Risk of ocular hypertension
or open-angle glaucoma in elderly patients on oral glucocorticoids.
Lancet 350:979982.
Gerson A, Abbasi S, Johnson A, Kalchbrenner M, Ashmead G, Bolognese
R. 1990. Safety and efficacy of long-term tocolysis with indometha-
cin. Am J Perinatol 7:7174.
Glover DD, Amonkar M, Rybeck BF, Tracy TS. 2003. Prescription, over-
the-counter, and herbal medicine use in a rural, obstetric population.
Am J Obstet Gynecol 188:10391045.
Henshaw SK. 1998. Unintended pregnancy in the United States. Fam
Plann Perspect 30:2429, 46.
Imperiale TF, Petrulis AS. 1991. A meta-analysis of low-dose aspirin for
the prevention of pregnancy-induced hypertensive disease. JAMA
266:260264.
Jabbour HN, Sales KJ, Smith OP, Battersby S, Boddy SC. 2006.
Prostaglandin receptors are mediators of vascular function in
endometrial pathologies. Mol Cell Endocrinol 252:191200.
Kallen BA, Otterblad OP. 2003. Maternal drug use in early pregnancy and
infant cardiovascular defect. Reprod Toxicol 17:255261.
Kozer E, Costei AM, Boskovic R, Nulman I, Nikfar S, Koren G. 2003.
Effects of aspirin consumption during pregnancy on pregnancy
outcomes: meta-analysis. Birth Defects Res B Dev Reprod Toxicol 68:
7084.
LAssociation des pharmaciens du Canada. 2003. Compendium des
produits et specialites pharmaceutiques. Ottawa, Canada: Associa-
tion des pharmaciens du Canada.
Leitich H, Egarter C, Husslein P, Kaider A, Schemper M. 1997. A meta-
analysis of low dose aspirin for the prevention of intrauterine growth
retardation. Br J Obstet Gynaecol 104:450459.
278 OFORI ET AL.
Birth Defects Research (Part B) 77:268279, 2006 DOI 10.1002/bdrb
Levy AR, Mayo NE, Grimard G. 1995. Rates of transcervical and
pertrochanteric hip fractures in the province of Quebec, Canada,
1981-1992. Am J Epidemiol 142:428436.
Little J, Cardy A, Munger RG. 2004. Tobacco smoking and oral clefts: a
meta-analysis. Bull World Health Organ 82:213218.
Lynch S. Missing data. Accessed March 10 2006. Available at: http://
www.princeton.edu/Bslynch/SOC_504/.
Man LX, Chang B. 2006. Maternal cigarette smoking during pregnancy
increases the risk of having a child with a congenital digital anomaly.
Plast Reconstr Surg 117:301308.
March of Dimes. Miscarriage. Accessed June 14 2006. Available at: http://
www.marchofdimes.com/professionals/681_1192.asp.
Martinez-Frias ML, Rodriguez-Pinilla E, Prieto L. 1997. Prenatal exposure
to salicylates and gastroschisis: a case-control study. Teratology
56:241243.
Moise KJ Jr. 1993. Effect of advancing gestational age on the frequency of
fetal ductal constriction in association with maternal indomethacin
use. Am J Obstet Gynecol 168:13501353.
Namkoong S, Lee SJ, Kim CK, Kim YM, Chung HT, Lee H, Han JA, Ha
KS, Kwon YG, Kim YM. 2005. Prostaglandin E2 stimulates
angiogenesis by activating the nitric oxide/cGMP pathway in human
umbilical vein endothelial cells. Exp Mol Med 37:588600.
New York State Department of Health. 2005. Congenital Malformations
RegistrySummary Report. Statistical Summary of Children Born in
1998-2001 and Diagnoses Through 2003. New York: New York State
Department of Health.
Nielsen GL, Sorensen HT, Larsen H, Pedersen L. 2001. Risk of adverse
birth outcome and miscarriage in pregnant users of non-steroidal
anti-inflammatory drugs: population based observational study and
case-control study. BMJ 322:266270.
Olesen C, Steffensen FH, Nielsen GL, de Jong-van den Berg, Olsen J,
Sorensen HT. 1999. Drug use in first pregnancy and lactation: a
population-based survey among Danish women. The EUROMAP
group. Eur J Clin Pharmacol 55:139144.
Ray JG, OBrien TE, Chan WS. 2001. Preconception care and the risk of
congenital anomalies in the offspring of women with diabetes
mellitus: a meta-analysis. QJM 94:435444.
Regie de lassurance maladie du Quebec. 1997. Statistiques annuelles
Quebec: Government du Quebec.
Richards ID. 1972. A retrospective enquiry into possible teratogenic
effects of drugs in pregnancy. Adv Exp Med Biol 27:441455.
Roy E, Haley N, Leclerc P, Sochanski B, Boudreau JF, Boivin JF. 2004.
Mortality in a cohort of street youth in Montreal. JAMA 292:569574.
Schneeweiss S. 2006. Sensitivity analysis and external adjustment for
unmeasured confounders in epidemiologic database studies of
therapeutics. Pharmacoepidemiol Drug Saf 15:291303.
Sohaey R, Zwiebel WJ. 1996. The fetal thorax: noncardiac chest anomalies.
Semin Ultrasound CT MR 17:3450.
Streck RD, Kumpf SW, Ozolins TR, Stedman DB. 2003. Rat embryos
express transcripts for cyclooxygenase-1 and carbonic anhydrase-4,
but not for cyclooxygenase-2, during organogenesis. Birth Defects
Res B Dev Reprod Toxicol 68:5769.
Suarez VR, Thompson LL, Jain V, Olson GL, Hankins GD, Belfort MA,
Saade GR. 2002. The effect of in utero exposure to indomethacin on
the need for surgical closure of a patent ductus arteriosus in the
neonate. Am J Obstet Gynecol 187:886888.
Tamblyn R, Lavoie G, Petrella L, Monette J. 1995. The use of prescription
claims databases in pharmacoepidemiological research: the accuracy
and comprehensiveness of the prescription claims database in
Quebec. J Clin Epidemiol 48:9991009.
Tanner K, Sabrine N, Wren C. 2005. Cardiovascular malformations among
preterm infants. Pediatrics 116:833838.
Van Hecken A, Schwartz JI, Depre M, De Lepeleire I, Dallob A, Tanaka W,
Wynants K, Buntinx A, Arnout J, Wong PH, Ebel DL, Gertz BJ,
De Schepper PJ. 2000. Comparative inhibitory activity of rofecoxib,
meloxicam, diclofenac, ibuprofen, and naproxen on COX-2 versus
COX-1 in healthy volunteers. J Clin Pharmacol 40:11091120.
Williams LJ, Correa A, Rasmussen S. 2004. Maternal lifestyle factors
and risk for ventricular septal defects. Birth Defects Res A Clin Mol
Teratol 70:5964.
Zenker M, Klinge J, Kruger C, Singer H, Scharf J. 1998. Severe pulmonary
hypertension in a neonate caused by premature closure of the ductus
arteriosus following maternal treatment with diclofenac: a case
report. J Perinat Med 26:231234.
APPENDIX 1. ICD-9 CODES FOR
UNSPECIFIED ANOMALIES EXCLUDED
IN SENSITIVITY ANALYSIS
ICD-9 code Description
759.0759.9 Other and unspecified congenital
anomalies
742.9 Unspecified anomalies of nervous
system
743.9 Unspecified anomaly of eye
744.3 Unspecified anomaly of ear
744.9 Unspecified anomaly of face and neck
745.8 Other endocardial cushion defect
745.9 Unspecified defect of septal closure
746.9 Unspecified anomaly of heart
747.9 Unspecified anomaly of circulatory
system
748.9 Unspecified anomaly of respiratory
system
750.9 Unspecified anomaly of upper
alimentary tract
751.9 Unspecified anomaly of digestive system
752.9 Unspecified anomaly of genital organs
753.9 Unspecified anomaly of urinary system
754.8 Other specified non-teratogenic
anomalies
755.9 Unspecified anomaly of unspecified
limb
756.9 Unspecified anomaly of musculoskeletal
system
757.9 Unspecified anomaly of integument
APPENDIX 2. ICD-9 CODES FOR
MINOR ANOMALIES EXCLUDED
IN SENSITIVITY ANALYSIS
ICD-9 code Description
743.6 Congenital anomalies of eyelids, lacrimal
system, and orbit
744.1 Accessory auricle
744.2 Other specified anomalies of ear
744.4 Branchial cleft cyst
744.8 Other specified anomalies of face and
neck
747.0 Patent ductus arteriosus, if birth weight
o1500 g
747.5 Single umbilical artery
750.0 Tongue tie
752.4 Anomalies of cervix, vagina, and external
female genitalia
752.5 Undescended testicle, if birth weight
o2500 g
754.6 Valgus deformities of feet
755.0 Polydactyly
755.1 Syndactyly without fusion of bone
757.2 Dermatoglyphic anomalies
757.3 Other specified anomalies of skin
757.4 Specified anomalies of hair
757.5 Specified anomalies of nails
757.6 Specified anomalies of breast
757.8 Other specified anomalies of integument
758.4 Balanced autosomal translocation in
normal individual
279 NSAID USE DURING PREGNANCY AND BIRTH DEFECTS
Birth Defects Research (Part B) 77:268279, 2006 DOI 10.1002/bdrb