DRUGS IN PREGNANCY
DRUGS IN PREGNANCY
Rates of Spontaneous and Therapeutic
Abortions Following Use of Antidepressants in
Pregnancy: Results From a Large Prospective
Adrienne Einarson, RN,1 Jacqueline Choi, BSc1 Thomas R. Einarson, PhD,2 Gideon Koren, MD1
The Motherisk Program, The Hospital for Sick Children, Toronto ON
Leslie Dan Faculty of Pharmacy, University of Toronto, Toronto ON
Objective: The use of antidepressants during pregnancy remains a Objectif : L’utilisation d’antidépresseurs pendant la grossesse
controversial issue, and there is little information on the risk of demeure une question controversée; de plus, nous ne disposons
spontaneous abortions following antidepressant exposure in early que de peu de données sur le risque d’avortement spontané à la
pregnancy. We sought to examine whether use of antidepressants suite de l’exposition aux antidépresseurs aux débuts de la
increases the rates of spontaneous abortion (SA) and therapeutic grossesse. Nous avons cherché à déterminer si l’utilisation
abortion (TA) in women exposed in early pregnancy. d’antidépresseurs accroît les taux d’avortement spontané (AS) et
Methods: In a cohort of women who contacted the Motherisk d’avortement thérapeutique (AT) chez les femmes y ayant été
program during pregnancy, we compared two groups of women, exposées aux débuts de la grossesse.
one exposed and the other not exposed to antidepressants during
Méthodes : Au sein d’une cohorte de femmes qui ont communiqué
pregnancy, and calculated the associated rates of SA and TA.
avec le programme Motherisk pendant la grossesse, nous avons
Results: Among 937 women exposed to antidepressants prior to and comparé deux groupes de femmes (« femmes exposées aux
during early pregnancy, there were 122 SAs (13.0%) including antidépresseurs pendant la grossesse » et « femmes n’y ayant
three ectopic pregnancies, and in the comparison group there pas été exposées ») et avons calculé les taux connexes d’AS et
were 75 SAs (8.0%) and no ectopic pregnancies. The relative risk d’AT.
was 1.63 (95% CI 1.24–2.14). Three-fold more women reported a
TA in the exposed group, 26 (2.4%) compared to 8 (0.7%) in the Résultats : Chez les 937 femmes exposées aux antidépresseurs
non-exposed group (RR 3.25; 95% CI 1.48–7.14). A sub-analysis avant ou pendant les débuts de la grossesse, nous avons
revealed that in both groups, of 338 women with a prior SA, 58 constaté 122 AS (13,0 %), y compris trois grossesses ectopiques;
(17.2%) reported having a SA in the current pregnancy, compared au sein du groupe témoin, nous avons constaté 75 AS (8,0 %) et
with 61/652 (9.4%) with no prior SA (c = 12.09, P < 0.001). In the nous n’avons constaté aucune grossesse ectopique. Le risque
antidepressant group, the incidence was 20.7%, and in the relatif était de 1,63 (IC à 95 %, 1,24–2,14). Trois fois plus de
non-exposed group, it was 13.3%. Logistic regression confirmed femmes ont signalé un AT dans le groupe exposé, 26 (2,4 %) par
that only antidepressant exposure and prior SA were significantly comparaison avec 8 (0,7 %) dans le groupe non exposé (RR,
associated with current SA. 3,25; IC à 95 %, 1,48–7,14). Une sous-analyse a révélé que dans
les deux groupes, chez les 338 femmes ayant déjà connu un AS,
Conclusion: Exposure to antidepressants in the first trimester of 58 (17,2 %) ont signalé un AS dans le cadre de la grossesse en
pregnancy appears to be associated with a small but statistically cours, par comparaison avec 61/652 (9,4 %) femmes n’ayant
significant increased risk of SA and decision to terminate a 2
jamais connu un AS (c = 12,09, P < 0,001). Au sein du groupe
pregnancy. The risk for SA is further elevated with a history of
« antidépresseur », l’incidence était de 20,7 %; au sein du groupe
previous SA. However, any underlying depression must be taken
non exposé, elle était de 13,3 %. Une régression logistique a
into consideration when evaluating these results.
permis de confirmer que seules l’exposition aux antidépresseurs
et la présence d’antécédents d’AS étaient associées de façon
significative à l’AS pendant la grossesse en cours.
Conclusion : L’exposition aux antidépresseurs au cours du premier
trimestre de la grossesse semble être associée à une faible (mais
significative sur le plan statistique) hausse du risque d’AS, ainsi
Key Words: Pregnancy, depression, antidepressants, miscarriages,
qu’à la décision d’interrompre la grossesse. Le risque d’AS se
trouve encore plus exacerbé en présence d’antécédents d’AS.
Competing Interests: See Acknowledgements. Quoi qu’il en soit, la présence de toute dépression sous-jacente
doit être prise en considération au moment d’évaluer ces résultats.
Received on October 30, 2008
Accepted on November 27, 2008 J Obstet Gynaecol Can 2009;31(5):452–456
452 l MAY JOGC MAI 2009
Rates of Spontaneous and Therapeutic Abortions Following Use of Antidepressants in Pregnancy
INTRODUCTION SA in 11 of 117 women (9.6%).11 A prescription monitoring
post-marketing surveillance from England found the inci-
t has been suggested that among the reported adverse
I outcomes of women exposed to antidepressants during
dence was 11%.12
Finally, in a meta-analysis of seven prospective, compara-
pregnancy is an increased risk for spontaneous abortion
(SA).1 However, to date no definitive association can be tive studies from teratogen information services, we found
inferred, as there is a paucity of studies examining this topic. a small but significant increase in risk (RR 1.45) for SA in
There are also few reported studies examining the incidence the group exposed to antidepressants in early pregnancy.1
of SA in the general population. It is difficult to calculate the Since that meta-analysis was published, our group has pub-
incidence of SA accurately, and even more difficult to calcu- lished three additional prospective comparatives studies
late the incidence of therapeutic abortion (TA). reporting on pregnancy outcomes of women who were
exposed to antidepressants in early pregnancy. We docu-
Notwithstanding, there have been attempts from around mented the incidence of SA in each group, and in the anti-
the world to estimate pregnancy loss in various populations, depressant groups the results were bupropion 15%,13
mostly the incidence of SA.2 A group in England mailed mirtazepine 19%,14 and citalopram 11%.15 The incidence of
questionnaires to 60 000 randomly selected women and TA was also higher in all three antidepressant groups than
estimated that the incidence of SA in the population was in the non-exposed. However, the results were not statisti-
11.5% to 12.7%.3 A large epidemiologic study was con- cally significant, probably because of the insufficient sample
ducted in India to estimate the rates of miscarriage in a very sizes in the studies.
large population of women (90 303), and the incidence of
both SA (7%) and TA (1%) were found to be low. These The aim of this study was to examine the incidence of both
authors found that birth order as well as age substantially SA and TA from prospectively collected cases in our data-
influenced SA risk.4 In a longitudinal Swedish study of 887 base. With a much larger sample size, we felt that it may be
pregnancies in 341 women, there were 108 SAs (12%) and possible to obtain a more definitive answer to the question
173 TAs (19.5%).5 A New Zealand group reported the inci- of whether or not antidepressants are associated with an
dence of SA in the general population at 7.8%.6 A Danish increase in the risk of SA in women who take them in early
group7 estimated the rate of SA by analyzing 1 221 546 pregnancy.
pregnancy outcomes in 634 272 women. The overall inci- METHODS
dence of fetal loss was 13.5%, with the risk of 8.9% in
The Motherisk Program at the Hospital for Sick Children in
women aged 20 to 24 years and increasing to 74.7% in those
Toronto is a teratology information service. Our program
aged 45 years or more.7 A study undertaken in Italy found
provides evidence-based information regarding the safety
that in comparison with women under age 20 years, the risk
and risks associated with exposures to drugs, chemicals,
of SA increased for women aged 35–39 (odds ratio 1.45)
radiation, and infectious diseases during pregnancy and lac-
and for women over age 40 (odds ratio 3.10). The odds ratio
tation to pregnant women, lactating mothers, and their
was 1.93 for women who had been pregnant two or more
health care providers. Women call us for information
regarding the safety of a drug, usually early in pregnancy,
Calculating the incidence of TA is even more difficult, as and most often following recognition of the pregnancy at
therapeutic abortions are performed for both medical and approximately six weeks’ gestation. During the initial tele-
social reasons. Some TAs may be reported as an SA, phone contact, demographics, medical and obstetrical his-
because of guilt and embarrassment. Epidemiologic collec- tories, and details of exposure and concurrent exposures are
tion of this information therefore is difficult and can be recorded on a standardized questionnaire. Details about the
fraught with many serious methodological issues. exposure include duration, timing in pregnancy, dose, fre-
quency, and indication for drug use. At the follow-up inter-
There is very little information regarding the incidence of
view, gestational findings, fetal outcomes, and neonatal
SA and TA when the mother was exposed to antidepres-
health, when appropriate, are documented on a structured
sants in early pregnancy. In a manufacturer’s post-
form by telephone interview with each mother, following a
marketing surveillance study of fluoxetine, the incidence of
detailed explanation of the study and with her consent.
SA was 13.5%; however, there was no comparison group.9
A European teratogen information service reporting on We included in the study women who were currently preg-
various antidepressants, also with no comparison group, nant at the time of call, who contacted us for information
documented the incidence of SA at 11.5%.10 Results from a regarding an antidepressant, and who had been taking the
pharmaco-toxicovigilance service in France, with various medication prior to pregnancy, through the first trimester.
antidepressants and with no comparison group, reported We then compared the data from these women with data
MAY JOGC MAI 2009 l 453
DRUGS IN PREGNANCY
Rates of spontaneous abortions by subgroup (unadjusted)
Group Sub-analysis n SA Incidence %
Current Pregnancy Exposed 937 122 13.0*
Non-exposed 937 75 8.0
Previous SA Exposed 188 39 20.7
Non-exposed 150 20 13.3
No previous SA Exposed 337 35 10.4
Non-exposed 315 25 7.9
Smokers Exposed 154 17 11.0
Non-exposed 123 7 5.7
n: number of women; SA: spontaneous abortion
*P < 0.05; all others non-significant
from an equal number of women who were not exposed to trazodone (n = 18), and venlafaxine (n = 182). We did not
antidepressants and who had called Motherisk for informa- analyze each antidepressant separately because the sample
tion regarding non-teratogenic drugs such as acetamino- size was too small in each group, and there was no reason to
phen. The two groups were matched for maternal age, believe that there would be any differences among the
smoking and alcohol use, and gestational age at the time of groups.
the call to Motherisk. This last piece of information is criti- Among the women exposed to antidepressants, there were
cal when calculating the incidence of SA because, as 122 SAs (13.0%) including three ectopic pregnancies, and
described above, the observed proportion of pregnancies 75 SAs (8.0%) with no ectopic pregnancies in the compari-
ending in loss is highly dependent on the gestational age at
son group (c2 = 12.00; P < 0.001). Among those women
which pregnancies are recognized and how the losses are
who experienced SA, 63/122 exposed women (51.6%) and
identified. Obviously, identifying pregnancy even one week
28/75 non-exposed women (37.3%) had experienced a pre-
earlier makes a major difference in the early period of gesta-
tion since loss rates are highest at that point; thus, greater vious SA (c2 = 3.27; P = 0.07). The relative risk was 1.38
awareness and focus on pregnancy alone will inflate the fre- (95% CI 0.98–1.94). There were also significantly more
quency. In our study, we matched for gestational age at the women who reported a TA in the exposed group (26; 2.4%)
time of call to control for this variable, and then calculated than in the unexposed group (8; 0.7%). The RR for TA was
the incidence of SA and TA in each group. We then esti- 3.25 (95% CI 1.48–7.14).
mated a risk ratio and 95% confidence interval between In a sub-analysis, we selected only women (n = 990) who
exposed and non-exposed women. were in their second or subsequent pregnancy and
compared the current incidence of SA between those who
The study protocol was approved by the Hospital for Sick
reported previous SA and those who reported no previous
Children Research Ethics Board.
SA (Table). Among the 338 women with prior SA,
RESULTS 58 (17.2%) had SA in their current pregnancy, compared
For analyses, we were able to use data from 937 women in with 61 of 652 (9.4%) in those with no prior SA (c2 = 12.09;
each group who fulfilled our criteria. We were unable to use P < 0.001). Among those with prior SA, 39/188 (20.7%)
the remainder of the group because they were not taking the took antidepressants, and 20/150 (13.3%) were
medication prior to pregnancy and into the first trimester. non-exposed (c2 = 2.69; P = 0.101). In the group with no
Some started the medication later in the pregnancy and a prior SA, 35 of 338 (10.4%) took antidepressants and 25 of
number had discontinued the medication in the first couple 290 (7.9%) were non-exposed (c2 = 0.89; P = 0.334).
of weeks of pregnancy. There were 277 smokers (14.8%): 154 (16.4%) in the anti-
The two groups were matched for age, alcohol use, and depressant group and 123 (13.1%) in the comparison
smoking status. Women were exposed to 11 different anti- group. Since these factors can influence pregnancy out-
depressants in five different classes: bupropion (n = 104), comes, we explored the possibility of confounders in the
citalopram (n = 191), escitalopram (n = 23), fluvoxamine data. We used a logistic regression to determine the odds
(n = 50), fluoxetine (n = 51), mirtazepine (n = 39), ratio between antidepressant exposure and SA as a preg-
nefazodone (n = 59), paroxetine (n = 160), sertraline (n = 60), nancy outcome, while controlling for smoking status, prior
454 l MAY JOGC MAI 2009
Rates of Spontaneous and Therapeutic Abortions Following Use of Antidepressants in Pregnancy
occurrence of SA, and all interactions. Since women were The main strength of our study model is the personal inter-
already matched for age and alcohol consumption, we did view with the women, which includes a detailed history of
not include these variables. drug use and other maternal demographics. Our pregnancy
registry is designed specifically for recording pregnancy
The regression equation showed good fit (Wald c2 = 25.72, outcomes, so we are able to collect details of alcohol,
df = 3; P < 0.001). Two factors were independently signifi- tobacco, and concurrent drug use, as well as other potential
cant: antidepressant exposure (OR 1.64, 95% CI 1.21–2.23) confounders of pregnancy outcome. Importantly, because
and previous SA (OR 2.25, 95% CI 1.62–3.12). Smoking all of the women contacted Motherisk in early pregnancy,
was not related to outcomes (P = 0.203), nor were any of and the details of their pregnancy and drug exposure were
the interaction terms. recorded at that time, the possibility of a recall bias is largely
To our knowledge, this is the largest prospective cohort For many years, it has been accepted that the incidence of
study of the incidence of SA and TA in pregnant women spontaneous abortion in clinically recognized pregnancies
exposed to an antidepressant prior to becoming pregnant, in the population is between 10% and 20%, which repre-
including a comparison group, that has been reported to sents a wide range.1 In our study, we found that the inci-
date. Overall, women who took antidepressants reported a dence differed depending on whether the woman had expe-
marginally (but significantly) higher incidence of SA than rienced a previous SA, because the incidence of SA in both
women in the comparison group (RR 1.63). However, our the antidepressant group (10.4%) and the comparison
study was unable to separate the potential effect of the anti- group (7.9%) was lower in primigravid women.
depressant from the possible adverse effects of the depres- CONCLUSION
sion itself; this confounding by indication cannot be
answered by this study design because we did not have a It appears that there is a small but significant increase in the
group of women with depression who were untreated. We incidence of SA in women exposed to antidepressants in
also found that women in both groups who experienced SA early pregnancy, as well as a higher incidence of TA. The
in a previous pregnancy reported a higher incidence of SA apparent synergy in adverse effects of antidepressant ther-
in the current pregnancy. This result confirms the findings apy and depression necessitates further research, as it is still
of a group from Paris16 who conducted a case-control study unknown whether depression, either treated with sub thera-
to evaluate risk factors associated with SA. In their study, a peutic doses of antidepressants (a common practice in preg-
total of 279 cases and 279 controls were compared for nancy) or untreated, is associated with an increased risk for
sociodemographic characteristics, reproductive history, and SA. In the meantime, a woman who requires treatment with
other conditions surrounding conception. Prior fetal losses an antidepressant during pregnancy should discuss the risks
(OR 2.30 for ³ 2 losses; 95% CI 1.17–4.61) and maternal and benefits of pharmacological treatment with her physi-
age at pregnancy (> 30 years) appeared to be major and cian, since she must be in good mental health to ensure the
independent risk factors of spontaneous abortion.16 Ethnic best possible outcome for herself and her baby.
origin and psychological problems around the time of con- ACKNOWLEDGEMENTS
ception were also identified as possible risk factors.16
Dr Koren and Ms Einarson have received research support
There was also a three-fold higher incidence of TA in the from Janssen-Ortho and Wyeth. Dr Koren has received
exposed group than in the non-exposed group. This may be research support from Apotex, Duchesnay, Novartis, and
because of fear of teratogenicity, as we have documented in Pfizer. Ms Einarson has received unrestricted research
previous research that women perceive psychotropic drugs grants from GlaxoSmithKline for studying ondansetron in
as more dangerous to the fetus, despite lack of evidence, pregnancy and from Organon for studying mirtazapine in
than other drugs such as antibiotics.17 The increase could pregnancy. Dr Einarson has received research support from
also be the result of poor mental health status at the time, if Bristol-Myers Squibb, Eli Lilly, Janssen-Ortho, Lundbeck,
these women felt that they were not able to cope with a Novo Nordisk, and Organon. Ms Choi reports no compet-
pregnancy. ing interests.
Despite the fact that cigarette smoking has been associated REFERENCES
with an increased risk for SA in many studies,18,19 it was not 1. Hemels ME, Einarson A, Koren G, Lanctot KL, Einarson TR.
associated with a higher incidence in our study. Almost all Antidepressant use during pregnancy and the rates of spontaneous
abortions: a meta-analysis. Ann Pharmacother 2005;39:803–9.
of the smokers had reduced their tobacco consumption
2. Wilcox AJ, Weinberg CR, O’Connor JF, Baird DD, Schlatterer JP, Canfield
during pregnancy, so were considered light smokers (less RE, et al. Incidence of early loss of pregnancy. N Engl J Med
than 10/day). 1988;319:189–94.
MAY JOGC MAI 2009 l 455
DRUGS IN PREGNANCY
3. Savitz DA, Hertz-Picciotto I, Poole C, Olshan AF. Epidemiologic measures 12. Wilton LV, Pearce GL, Martin RM, Mackay FJ, Mann RD. The outcomes of
of the course and outcome of pregnancy. Epidemiol Rev 2002;24:91–101. pregnancy in women exposed to newly marketed drugs in general practice
4. Pallikadavath S, Stones RW. Miscarriage in India: a population-based study. in England. Br J Obstet Gynaecol 1998;105:882–9.
Fertil Steril 2005;84:516–8. 13. Chan BCF, Koren G, Fayez I, Kalra S, Voyer-Lavigne S, Boshier A, et al.
5. Blohm F, Fridén B, Milsom I. A prospective longitudinal population-based Pregnancy outcome of women exposed to bupropion during pregnancy:
study of clinical miscarriage in an urban Swedish population. BJOG a prospective comparative study. Am J Obstet Gynecol 2005;192:932–6.
14. Djulus J, Koren G, Einarson TR, Wilton L, Shakir S, Diav-Citrin O, et al.
6. Yang CJ, Stone P, Stewart AW. The epidemiology of recurrent miscarriage: Exposure to mirtazapine during pregnancy: a prospective, comparative
a descriptive study of 1214 prepregnant women with recurrent miscarriage. study of birth outcomes. J Clin Psychiatry 2006;67:1280–4.
Aust N Z J Obstet Gynaecol 2006;46:316–22.
15. Sivojelezova A, Shuhaiber S, Sarkissian L, Einarson A, Koren G. Citalopram
7. Nybo Andersen A-M, Wohlfahrt J, Christens P, Olsen J, Melbye M.
use in pregnancy: prospective comparative evaluation of pregnancy and fetal
Maternal age and fetal loss: population based register linkage study. BMJ
2000;320:1708–12. outcome. Am J Obstet Gynecol 2005;193:2004–9.
8. Osborn JF, Cattaruzza MS, Spinelli A. Risk of spontaneous abortion in 16. Coste J, Job-Spira N, Fernandez H. Risk factors for spontaneous abortion:
Italy, 1978–1995 and the effect of maternal age, gravidity, marital status, a case-control study in France. Hum Reprod 1991;6:1332–7.
and education. Am J Epidemiol 2000;151:98–105. 17. Bonari L, Koren G, Einarson TR, Jasper JD, Taddio A, Einarson A .Use of
9. Goldstein DJ, Corbin LA, Sundell KL. Effects of first-trimester fluoxetine antidepressants by pregnant women: evaluation of perception of risk,
exposure on the newborn. Obstet Gynecol 1997; 89(5 Pt 1):713–8. efficacy of evidence based counseling and determinants of decision making.
10. McElhatton PR, Garbis HM, Eléfant E, Vial T, Bellemin B, Mastroiacovo P, Arch Womens Ment Health 2005;8:214–20.
et al. The outcome of pregnancy in 689 women exposed to therapeutic 18. Nielsen A, Hannibal CG, Lindekilde BE, Tolstrup J, Frederiksen K, Munk
doses of antidepressants. A collaborative study of the European Network of C, Bergholt T, Buss L, Ottesen B, Gronbaek M, Kjaer SK. Maternal
Teratology Information Services (ENTIS). Reprod Toxicol 1996;10:285–94. smoking predicts the risk of spontaneous abortion. Acta Obstet Gynecol
11. Brunel P, Vial T, Roche I, Bertolotti E, Evreux JC. Follow-up of 151
pregnant women exposed to antidepressant treatment (MAOI excluded) 19. Malik S, Cleves MA, Honein MA, Romitti PA, Botto LD, Yang S, Hobbs
during organogenesis [article in French]. Therapie 1994 CA. Maternal smoking and congenital heart defects. National Birth Defects
Mar-Apr;49(2):117–22. Prevention Study. Pediatrics 2008 Apr;121(4):e810–6.
456 l MAY JOGC MAI 2009