You are on page 1of 9

Original Research

Associations Between Abortion, Mental Disorders,


and Suicidal Behaviour in a Nationally
Representative Sample
Natalie P Mota, BA1; Margaret Burnett, MD, FRCPC2; Jitender Sareen, MD, FRCPC3

Objective: Most previous studies that have investigated the relation between abortion and
mental illness have presented mixed findings. We examined the relation between abortion,
mental disorders, and suicidality using a US nationally representative sample.
Methods: Data came from the National Comorbidity Survey Replication (n = 3310 women, aged
18 years and older). The World Health OrganizationComposite International Diagnostic
Interview was used to assess mental disorders based on the Diagnostic and Statistical Manual
of Mental Disorders, Fourth Edition, criteria and lifetime abortion in women. Multiple logistic
regression analyses were employed to examine associations between abortion and lifetime
mood, anxiety, substance use, eating, and disruptive behaviour disorders, as well as suicidal
ideation and suicide attempts. We calculated the percentage of respondents whose mental
disorder came after the first abortion. The role of violence was also explored. Population
attributable fractions were calculated for significant associations between abortion and mental
disorders.
Results: After adjusting for sociodemographics, abortion was associated with an increased
likelihood of several mental disordersmood disorders (adjusted odds ratio [AOR] ranging from
1.75 to 1.91), anxiety disorders (AOR ranging from 1.87 to 1.91), substance use disorders (AOR
ranging from 3.14 to 4.99), as well as suicidal ideation and suicide attempts (AOR ranging from
1.97 to 2.18). Adjusting for violence weakened some of these associations. For all disorders
examined, less than one-half of women reported that their mental disorder had begun after the
first abortion. Population attributable fractions ranged from 5.8% (suicidal ideation) to 24.7%
(drug abuse).
Conclusions: Our study confirms a strong association between abortion and mental disorders.
Possible mechanisms of this relation are discussed.
Can J Psychiatry. 2010;55(4):239247.

Clinical Implications

Some women with a history of abortion develop emotional problems, and clinicians should
assess for mental disorders, particularly SUDs, in these women.
Clinicians should screen for a history of abortion in women presenting with mood, anxiety,
or SUDs as a potential contributing factor.
Women presenting for abortion should be screened for a history of violence.

Limitations

Mental disorders were assessed by lay interviewers.


The study was cross-sectional and caution must be exerted when considering the direction
of the associations found.
We could not identify the gestation at which the abortion occurred and whether the
pregnancy was wanted or unwanted.

Key Words: abortion, mental disorders, suicidal behaviour, epidemiology

The Canadian Journal of Psychiatry, Vol 55, No 4, April 2010 W 239


Original Research

I t remains to be decided whether a relation exists between


induced abortion and mental illness. Several, but not all,
studies have found an association between a history of abor-
on Mental Health and Abortion reached a similar conclusion
in their recent report, at least among women who have had
only one abortion.27 Both depression symptoms2830 and gen-
tion and subsequent distress and mental disorders.14 Even eral anxiety30 have been found to be lower postabortion than
with an increase of studies using larger, community-based
before the abortion occurred. Further, these rates were
samples and interview-based assessment of mental disorders
deemed comparable with those in general population sam-
(as opposed to screening measures), debate continues over
this important research question. ples.2830 Null findings between abortion, depression,12,31 and
anxiety23 have also been described. In line with these find-
A positive association has been found between having an ings, Major et al29 found in their longitudinal study that only a
abortion and subsequent symptoms of anxiety511 and depres- very small percentage of women met DSM-III-R criteria for
sion. 68,10,12,13 Substance use, 8,10,11,1417 suicidal behav- PTSD 2 years after the abortion. Similarly, Steinberg and
iour,8,1821 and prior behavioural difficulties (symptoms of Russo32 found no relation between abortion and PTSD, social
conduct and oppositional defiant disorder) in childhood and anxiety, and generalized anxiety disorder in 2 nationally rep-
adolescence8,11,14 have also been found to be positively related resentative datasets when several covariates were taken into
with having had an abortion. A recent study22 in a nationally account. One of these important covariates was violence,
representative US sample collected between 1990 and 1992 which was also found to render the relation between abortion
supported these findings by identifying a relation between and depression nonsignificant in a study by Taft and
abortion and DSM-III-R mood, anxiety, and substance use Watson.31 Finally, in a large sample of women experiencing
disorders. Further, overall outpatient mental health services an unplanned pregnancy, no association was found between
use23 and hospital admission rates for psychiatric reasons24 are abortion and nonpsychotic mental illness based on diagnoses
higher in women who have undergone abortion, compared made with the ICD.18
with different samples of women who have not had abortions.
However, the directionality of the relation between abortion The mixed findings in previous research may have numerous
and mental illness remains unclear. Several studies have explanations. First, many studies have used small, restricted,
found preabortion symptomatology to be significantly associ- or treatment-seeking samples. These studies are therefore
ated with having an abortion and also with postabortion men- limited by selection bias and the findings may not be
tal problems.7,8,10,14 To our knowledge, no study to date has generalizable to the population at large. Second, variable
examined the relation between abortion and eating disorders, methodology and study design have been used across studies.
even though a much higher prevalence of unplanned pregnan- For example, several studies have used retrospective cohort
cies has been found in women with bulimia, compared with studies or cross-sectional designs where recall bias is an
women without the disorder.25 One study that looked at preg- issue, or have used different comparison groups of women,
nancy outcome in a sample of 82 women with anorexia or which adds to the difficulty of interpreting findings across
bulimia found that 30% of reported pregnancies resulted in studies.14,15,22 Previous research has also used a range of
induced abortion.26 However, the study was limited by the assessment tools assessing only symptoms and probable dis-
lack of a control group. order rather than using a standardized interview.6,7,13,15,31,33 In
the few studies that have examined ICD- or DSM-based men-
Other recent studies have found no association, or a negative tal disorders in representative samples, most have either
association, between having an abortion and future symptoms emphasized specific disorders (that is, depression and anxi-
of psychopathology. A recent review by Charles et al1 noted ety), have grouped them into summary categories, or have
that the best studies in the area have failed to identify a strong assessed disorders according to DSM-III-R criteria.
relation between abortion and subsequent poorer mental
health. The American Psychological Association Task Force Our study aims to address numerous methodological limita-
tions of previous research by using a large, recently collected,
nationally representative sample. We have 2 specific objec-
Abbreviations used in this article tives. First, we will examine the relation between abortion
DSM Diagnostic and Statistical Manual of Mental and a wide range of DSM-IVbased lifetime mental disor-
Disorders d e r s a n d su ic id a l b e h a v io u r a f ter ad j u s tin g f o r
ICD International Classification of Diseases sociodemographics and violence exposure. Second, we will
NCS-R National Comorbidity Survey Replication examine the temporal relation between age of onset of mental
PAF population attributable fraction illness and age of first abortion.
PTSD posttraumatic stress disorder
Methods
SUD substance use disorder
WMH-CIDI World Mental HealthComposite International
Sample
Diagnostic Interview The NCS-R (methodological details34) was funded by the
National Institute of Mental Health and the National Institute

240 W La Revue canadienne de psychiatrie, vol 55, no 4, avril 2010


Associations Between Abortion, Mental Disorders, and Suicidal Behaviour in a Nationally Representative Sample

of Drug Abuse, and took place between 2001 and 2003 in Sociodemographic Variables
48 states in the US (Alaska and Hawaii were not included). Sociodemographic covariates included age, education, mari-
People who were institutionalized were not included in the tal status, household income, and ethnoracial background
sampling for this survey. The survey sample consisted of and (or) ancestry. In the NCS-R dataset, the marital status
9282 respondents, aged 18 years or older. A subsample of variable was coded into married or cohabitating; separated,
these people (n = 5692), including all respondents who met divorced, or widowed; and never married, and the education
criteria for a lifetime mental disorder plus a probability variable into 0 to 11 years, 12 years, 13 to 15 years, and
subsample of the remaining respondents, took part in Part 2 of 16 years or more. Several studies have been published using
the survey. Part 2 assessed additional mental disorders and these subcategory divisions.37,38 Age was grouped into 18 to
also included questions relating to abortion. Thus the Part 2 29 years, 30 to 44 years, 45 to 59 years, and 60 years and
sample (n = 3310 women) was used in our study. The survey older, approximate quartiles. We also collapsed the
was administered by trained lay interviews in the households ethnoracial background or ancestry variable into 5 categories
of respondents, with an overall response rate of 70.9%. (Asian, Mexican or Other Hispanic, Afro- Caribbean or
African American, Non-Latino White, and Others) from the
Lifetime Abortion 7 category variable that was available in the NCS-R. Finally,
Interviewers inquired about abortion from female participants the continuous variable of household income was divided
who had had sexual intercourse in their lifetimes with the fol- into $0 to $24 999, $25 000 to $44 999, $45 000 to $74 999,
lowing questions: Have you ever had an abortion? and and $75 000 and more.
How old were you (the first time)? Respondents who
refused to answer or who answered dont know were Violence Exposure
removed from all analyses (n = 18). One woman who An any violence variable was created based on 6 traumatic
answered age 0 to age of first abortion was also removed events endorsed or not endorsed by respondents: physical
from analyses. There were also 2 cases that answered yes to abuse by parents and (or) guardians; physical abuse by part-
a history of abortion but dont know or refuse for age of ner; physical abuse by anybody else; rape; other sexual
abortion. Responses to the first question were dichotomized assault; and mugged, held up, or threatened with a weapon.
as no, compared with yes, and responses to the second These variables were previously used to represent violence in
question were made into a continuous variable. a study examining abortion and anxiety disorders in the
NCS.32
Mental Disorder Diagnoses
Version 3.0 of the WMH-CIDI35 was used to assess DSM-IV Statistical Analyses
mental disorders. Major depression, bipolar I disorder, To ensure the representativeness of the sample and the
dysthymia, agoraphobia without panic, generalized anxiety generalizability of results, weighting and stratification vari-
disorder, panic attacks, PTSD, social phobia, specific phobia, ables were used for all statistical analyses. Additionally, the
oppositional defiant disorder, conduct disorder, attention def- SUDAAN program39 was used to apply a variance estimation
icit hyperactive disorder, alcohol abuse, alcohol dependence, technique to all analyses known as Taylor Series
drug abuse, and drug dependence were investigated in our Linearization. This technique adjusts for the complex sam-
study. Summary variables were also created to form any pling designs of surveys such as the NCS-R.
mood, anxiety, disruptive behaviour, substance use, eating
disorder, and mental disorder variables. Owing to the rela- Multiple logistic regression analyses were used to examine
tively small number of cases of individual eating disorders, the relations between abortion status and sociodemographic
we examined only a summary variable consisting of anorexia, variables. Subsequent statistical analyses were adjusted for
bulimia, and binge eating disorder. Agreement between the all sociodemographics (age, education, household income,
WMH-CIDI and diagnoses made by clinicians using the marital status, and ethnoracial background and [or] ances-
Structured Clinical Interview for DSM-IV ranged from mod- try). Multiple logistic regression analyses were also used to
erate to good for most mental disorders.36 investigate the relations between abortion status and lifetime
Axis I mental disorders and suicidal behaviour. Unadjusted
Suicidal Behaviour odds ratios were calculated to assess whether lifetime expo-
Variables representing lifetime suicidal ideation and suicide sure to violence was a significant confounding variable in the
attempts were generated using the following 2 statements, relation between abortion and mental disorders in the present
which were given in written form to participants during the sample. A more stringent logistic regression model examin-
interview: You seriously thought about committing suicide ing associations between abortion and mental disorders was
and You attempted suicide. Interviewers asked respondents then employed, with exposure to violence entered as a
Did [the experience] ever happen to you? to receive a yes covariate along with sociodemographic variables. Owing to
or no response. People were also asked how old they were the large number of comparisons made, a Bonferroni cor-
the first time that these experiences happened. rected P value was applied to these analyses (25 outcome

The Canadian Journal of Psychiatry, Vol 55, No 4, April 2010 W 241


Original Research

Table 1 Abortion status and sociodemographic variables


Have you ever had an abortion? (n = 3291)

No, n = 2839 Yes, n = 452


% (95% CI) % (95% CI)
89.5 (88.290.7) 10.5 (9.311.8)

Variable n (%) n (%) OR (95% CI)

Age, years
18 to 29 674 (22.6) 106 (21.9) 1.00
30 to 44 844 (26.6) 193 (43.1) 1.68 (1.242.27)a
45 to 59 747 (26.0) 121 (26.6) 1.06 (0.741.52)
60 574 (24.9) 32 (8.5) 0.35 (0.200.61)b
Education, years
11 397 (15.6) 54 (16.7) 1.00
12 884 (34.0) 122 (29.5) 0.81 (0.491.33)
13 to 15 868 (22.2) 155 (29.7) 0.99 (0.611.61)
16 690 (22.2) 121 (24.0) 1.01 (0.621.64)
Marital status
Married or cohabitating 1523 (51.7) 231 (53.5) 1.00
Separated, divorced, or widowed 755 (27.0) 127 (25.8) 0.92 (0.611.40)
Never married 561 (21.4) 94 (20.6) 0.93 (0.631.37)
Ethnoracial background
Asian 29 (1.2) 17 (4.8) 1.00
Mexican or Other Hispanic 262 (10.0) 51 (15.0) 0.38 (0.141.05)
Afro-Caribbean or African American 363 (13.0) 90 (16.7) 0.33 (0.120.90)c
Non-Latino White 2090 (73.4) 277 (60.5) 0.21 (0.080.53)a
Others 95 (2.5) 17 (3.0) 0.31 (0.110.86)c
Household income
$24 999 898 (34.41) 134 (28.2) 1.00
$25 000 to $44 999 629 (19.9) 87 (19.5) 1.20 (0.741.96)
$45 000 to $74 999 665 (22.0) 93 (19.8) 1.11 (0.721.70)
$75 000 713 (23.7) 138 (32.5) 1.71 (1.092.68)c

All ns were unweighted and percents were weighted.


a
P < 0.01; b P < 0.001; c P = < 0.05

measures: P = 0.05 divided by 25 outcome measures = 0.002). outcome (mental disorders) that would not have occurred if
For each significant association between abortion and indi- the exposure (abortion) were removed.
vidual mental disorders after adjusting for sociodemographic Finally, to gain an indication of what proportion of respon-
variables and exposure to violence, PAFs were calculated dents had the onset of their mental disorder(s) after their first
using the following formula: abortion, a set of new variables was created by subtracting the
age of onset of each mental disorder from the age of first
P( OR 1) abortion for each disorder that emerged as having a signifi-
P(OR 1) + 1 cant relation with abortion. A dichotomous variable was cre-
ated for each disorder (people whose mental disorder onset
where P was the weighted prevalence of a lifetime history of came after their abortion onset, compared with remaining
abortion in the NCS-R sample and OR was the adjusted odds people with mental disorderabortion comorbidity) to deter-
ratio for abortion and each mental disorder.40 A PAF is mine the percentage frequency of people whose mental disor-
intended to approximate the percentage of a particular der had occurred after their first abortion.

242 W La Revue canadienne de psychiatrie, vol 55, no 4, avril 2010


Associations Between Abortion, Mental Disorders, and Suicidal Behaviour in a Nationally Representative Sample

Table 2 Relation between abortion and lifetime mental disorders


No abortion, n = 2839 Abortion, n = 452
% (95% CI) % (95% CI)
89.5 (88.290.7) 10.5 (9.311.8)

Psychiatric diagnostic
category n (%) n (%) AOR-1 (95% CI) AOR-2 (95% CI) PAF % (95% CI)
a
Major depression 865 (18.7) 169 (29.3) 1.75 (1.292.37) 1.51 (1.142.01)
Dysthymia 209 (4.7) 50 (8.5) 1.79 (1.212.63) 1.42 (0.982.04)
Bipolar I disorder 46 (0.9) 17 (2.8) 2.80 (1.385.70) 1.97 (0.974.01)
Any mood disorder 958 (20.7) 196 (34.0) 1.91 (1.432.54)a 1.61 (1.242.11)a 6.0 (3.010.0)
Agoraphobia (without panic) 126 (2.7) 27 (4.2) 1.51 (0.942.42) 1.26 (0.821.95)
a
Generalized anxiety disorder 426 (9.1) 97 (16.0) 1.91 (1.372.65) 1.67 (1.172.22)
Panic disorder 260 (5.8) 54 (9.5) 1.53 (0.962.43) 1.31 (0.871.97)
Panic attack 1182 (30.8) 221 (41.8) 1.45 (1.081.95) 1.25 (0.951.65)
PTSD 375 (8.9) 88 (16.8) 1.91 (1.252.94) 1.46 (0.942.26)
a a
Social phobia 547 (12.0) 129 (21.3) 1.89 (1.392.57) 1.61 (1.232.12) 6.0 (2.410.5)
Specific phobia 669 (15.0) 122 (21.4) 1.44 (1.101.89) 1.27 (0.991.64)
Any anxiety disorder 1767 (44.5) 335 (61.8) 1.87 (1.332.63)a 1.58 (1.152.17)
Oppositional defiant disorder 179 (4.6) 50 (9.0) 1.67 (0.992.83) 1.33 (0.792.22)
Conduct disorder 141 (3.3) 40 (7.1) 1.72 (1.182.49) 1.36 (0.941.98)
Attention-deficit hyperactivity 152 (3.5) 26 (4.4) 0.96 (0.571.62) 0.76 (0.471.24)
disorder
Any disruptive behaviour 338 (8.3) 80 (13.9) 1.34 (0.882.02) 1.05 (0.701.58)
disorder
Alcohol abuse 259 (6.0) 108 (20.9) 4.23 (3.175.64)a 3.61 (2.674.88)a 21.5 (14.928.9)
a a
Alcohol dependence 121 (2.6) 46 (8.0) 3.14 (2.284.32) 2.42 (1.773.31) 13.0 (7.519.5)
Drug abuse 155 (3.5) 87 (16.1) 4.99 (3.547.02)a 4.13 (2.856.00)a 24.7 (16.334.4)
Drug dependence 69 (1.5) 42 (7.5) 4.88 (3.237.36)a 3.87 (2.505.98)a 23.2 (13.634.3)
Any SUD 298 (6.9) 129 (24.6) 4.46 (3.395.85)a 3.80 (2.865.04)a 22.7 (16.329.8)
Any eating disorder 101 (2.4) 23 (4.2) 1.70 (0.863.36) 1.46 (0.742.88)
Any mental disorder 2077 (52.1) 387 (73.5) 2.37 (1.533.67)a 1.98 (1.273.08)
Suicidal ideation 687 (16.1) 161 (28.4) 1.97 (1.462.66)a 1.59 (1.202.11)a 5.8 (2.110.4)
Suicide attempt 254 (5.6) 69 (12.1) 2.18 (1.662.86)a 1.51 (1.151.97)

All ns were unweighted. All percents were weighted.


AOR-1 = Adjusted for age, marital status, race, education, and household income.
AOR-2 = Adjusted for age, marital status, race, education, household income, and any violence.
a
P = 0.002
= PAF were not calculated for these mental disorders as they did not reach significance when adjusting for sociodemographic variables and violence
exposure.

Results women with a household income of more than $75 000 were
more likely to report an abortion than those in the $0 to
Table 1 shows the analysis of abortion status and
$24 999 category.
sociodemographic characteristics. A lifetime history of abor-
tion was reported by 10.5% (95% CI 9.29 to 11.83) of women. Results of unadjusted logistic regression analyses examining
Women aged 30 to 44 years were more likely than women exposure to violence as a potential confounding variable in
aged 18 to 29 years to have a lifetime history of abortion, and the abortionmental disorder relation showed significant
women 60 years of age and older were less likely than the ref- positive associations between abortion status and exposure to
erence group to report ever having had an abortion. Further, violence (OR 2.56, 95% CI 1.73 to 3.80) and between vio-
Afro-Caribbean or African American, Non-Latino White, and lence and all mental disorder summary categories (OR range
women of Other descent were less likely than women of Asian 2.87 to 5.88results not shown but available from the
descent to endorse a lifetime history of abortion. Finally, authors).

The Canadian Journal of Psychiatry, Vol 55, No 4, April 2010 W 243


Original Research

Table 3 Percentage of people with age of onset of WMH-CIDI, an internationally recognized standardized
mental disorder after age of first abortion interview, to diagnose mental disorders.35 Previously, much
of the literature has relied on womens self-reporting of
Lifetime psychiatric diagnostic Mental disorder after abortion
category % (95% CI)
symptomatology.
The findings of our study should be considered in the context
Major depression 48.8 (40.157.5)
of the following limitations. Mental disorders were assessed
Generalized anxiety disorder 45.3 (34.556.7)
by lay interviewers. Nonetheless, agreement was moderate to
Social phobia 9.7 (5.916.1) good between interviewer, and clinician-made diagnoses for
Alcohol abuse 44.7 (33.956.1) most of the mental disorders examined.36 The data used in our
Alcohol dependence 45.9 (30.662.1) study were from a cross-sectional survey, and therefore cau-
Drug abuse 41.9 (29.955.0) tion must be exerted when considering the direction of the
associations found. Our age of onset analyses did allow for a
Drug dependence 49.5 (31.068.1)
proximal assessment of a temporal relation between abortion
Suicidal ideation 29.9 (21.739.6)
and mental disorders. These analyses examined only age of
Suicide attempt 23.1 (13.636.2) first abortion, and we were unable to distinguish whether
mental disorder preceded or followed abortion if both events
occurred in the same year. Further, recall of age of abortion or
onset of mental illness may have been unreliable, particularly
Table 2 displays the results of the relations between abortion if they occurred in the distant past. We were also unable to
and mental disorders and suicidality, as well as the PAFs that identify elective abortions for genetic abnormalities, com-
were calculated for each significant relation. Women report- pared with those occurring for other reasons, whether the
ing a lifetime history of abortion were more likely to have a pregnancy was originally wanted or unwanted, or when in the
mood disorder. Suicidal ideation and suicide attempts were pregnancy the abortion occurred. Unintended pregnancy
also more likely to be reported by women who had had an itself may be a stressful event that can be a confounding fac-
abortion. Several of these significant relations remained sig- tor in the relation between abortion and mental illness,29,32,41
nificant even after adjusting for violence. However, in the and it is important to study mental disorder risk in women
case of major depression and suicide attempts, the associa- who had an abortion, compared with those who had an
tions were no longer significant. Similarly, associations were unplanned pregnancy but carried the baby to term.27 Unfortu-
demonstrated between some anxiety disorders and abortion. nately, the NCS-R could not be used to examine these associ-
However, with the exception of social phobia, these relations ations. We were also unable to adjust for some potential
were weakened when adjusting for violence. SUDs were confounding variables in our analyses, such as personality
strongly associated with a history of abortion, irrespective of variables, because they were not assessed in the survey.8
violence exposure. No significant relation was found between Finally, it is possible that abortion may have been
abortion and the any eating disorder category. PAFs ranged underreported,42,43 owing to the stigma that surrounds it.
from 5.8% for suicidal ideation to a high 24.7% for drug
abuse. A higher likelihood of any lifetime mood disorder was found
to exist in women who had experienced an abortion, com-
Table 3 displays the percentage frequency of women whose pared with those who had never had an abortion. It is likely
onset of mental disorder and suicidal behaviour came after that a woman experiencing a mood disorder would be less
their age of first abortion. To ascertain the temporality of inclined to carry through with a pregnancy, instead choosing
events, this analysis was performed for the mental disorders abortion. Conversely, the circumstances of an unplanned
that were shown to be independently associated with abortion pregnancy and abortion could precipitate a mood disorder in
in our preliminary logistic regression analyses. The percent- a susceptible person. This relation is likely bidirectional, as
age of women whose mental disorder followed their abortion the percentage frequency of major depression that followed
was under 50% for all mental disorders and for suicidal abortion was about equal to the proportion of cases occurring
ideation and suicide attempts. Less than 10% of people devel- before or during the same year as the first abortion.
oped social phobia after their abortion. For some disorders,
particularly major depression and drug dependence, about Women who had had an abortion were also more likely to
one-half of women had an age of onset of mental disorder have an anxiety disorder, although the relation was weakened
occurring after their first abortion. when adjusting for exposure to violence and only social pho-
bia was able to withstand the conservative P value that was
used. The significant positive relation between social phobia
Discussion and abortion is contrary to the null findings in the recent
To our knowledge, our study was one of the first to examine study by Steinberg and Russo.32 However, that study used
associations between abortion and a wide range of individual women who delivered their babies, as opposed to women
mental disorders using DSM-IV criteria in a large, nationally who had never had an abortion, as their comparison group. It
representative sample of adult women. In addition, it used the is possible that having an anxiety disorder may contribute to a

244 W La Revue canadienne de psychiatrie, vol 55, no 4, avril 2010


Associations Between Abortion, Mental Disorders, and Suicidal Behaviour in a Nationally Representative Sample

decision to have an abortion by undermining self-confidence and a decision to have an abortion. Finally, there could be
and the ability to parent successfully. However, the apparent shared vulnerability factors (for example, environmental fac-
association between mental disorders and abortion may also tors and personality pathology) that might be associated both
be driven by confounding factors that influence both phenom- with abortion and with mental illness.8 Future work needs to
ena. For example, exposure to violence22,31,32,44,45 may be one explore these and other possible mechanisms.
of these factors, as well as poor social or spousal The implications of these findings and areas for future direc-
support.14,46,47 tion are several. First, clinicians should assess for mental dis-
The strongest associations in our study were between abortion orders in women requesting an abortion, as the mental
and SUDs. This association was independent of a history of disorder often precedes, rather than succeeds, abortion.
exposure to violence. Again, the proportion of cases of SUDs These findings are consistent with those studies that have
occurring after the age of first abortion was less than, or about found mental health before the abortion to play an important
equal to, the proportion of cases occurring before or during role in the overall relation between abortion and mental
the same year as the procedure. SUDs have been found to be health.8,14,18,29 Future research should further investigate the
associated with sexual behaviours such as having a greater bidirectional relations found in our study by continuing to use
number of partners and infrequent condom use.4850 Clearly, valid assessments of mental disorders and representative
these behaviours can place people at an increased risk for an samples of women, employing a longitudinal design, and
unintended pregnancy, and consequently, for an abortion. In improving the assessment of abortion by corroborating
fact, many studies have identified a relation between early and self-reported histories with other reliable sources.
risky sexual behaviours and abortion.8,11,14,51,52 Conversely,
however, some researchers have discussed the possibility that Conclusion
women may self-medicate their pain with alcohol or drugs SUDs appear to be more prevalent in women who report hav-
after an abortion has occurred.15,16 A bidirectional relation is ing had at least one abortion. Mood and anxiety disorders, as
likely to exist. What should be noted is that the PAFs calcu- well as suicidality, are also associated with a history of abor-
lated for all SUDs were substantial, suggesting that a large tion but the relation is somewhat weaker and less consistent.
proportion of the onset of these disorders may be attributable Further, exposure to violence is a confounding factor in sev-
to abortion. eral of the associations between mental disorders and abor-
To our knowledge, this is the first study to examine the associ- tion. Our study does not support a unidirectional relation
ation between abortion and eating disorders, and the relation between abortion and mental disorders. Women undergoing
was not found to be significant. However, the null finding abortion are just as likely, if not more likely, to have a
may be the result of the low prevalence of these disorders in pre-existing mental disorder than to develop a new mental
our sample. Future studies should examine this association disorder subsequent to the abortion.
with a directed sample.
Acknowledgements
Finally, both suicidal ideation and suicide attempts were posi-
The current study was supported by a Canadian Institutes of
tively associated with abortion in our study after adjusting for
Health Research New Investigator Award (#152348) for
sociodemographic variables, and the association with suicidal Dr Sareen and by a Social Sciences and Humanities Research
ideation remained significant after adjusting for violence as Council Canada Graduate ScholarshipMasters Level Award
well. Only about 30% of ideators and 23% of attempters and Manitoba Graduate Scholarship for Ms Mota.
reported their age of suicidal behaviour subsequent to the first
abortion. This result is at odds with a previous study showing References
increased suicidality in women after abortion, compared with
women who had delivered their babies.20 However, Morgan et 1. Charles VE, Polis CB, Sridhara SK, et al. Abortion and long-term mental health
outcomes: a systematic review of the evidence. Contraception.
als20 study was limited by examining only patients seeking 2008;78:436450.
admission for suicide attempts and not a general population 2. Thorp JM, Hartmann KE, Shadigan E. Long-term physical and psychological
health consequences of induced abortion: review of the evidence. Obstet
sample. Some researchers have postulated that women who Gynecol Surv. 2003;58:6779.
have endured, or are enduring, such psychological distress do 3. Bradshaw Z, Slade P. The effects of induced abortion on emotional experiences
not feel they are capable to care for a baby when they find and relationships: a critical review of the literature. Clin Psychol Rev.
2003;23:929958.
themselves pregnant and therefore terminate the pregnancy.53 4. Coleman PK, Reardon DC, Strahan T, et al. The psychology of abortion:
a review and suggestions for future research. Psychol Health. 2005;20:237271.
There are several possible mechanisms that may explain the 5. Rue VM, Coleman PK, Rue JJ, et al. Induced abortion and traumatic stress:
association between abortion and mental illness. First, there a preliminary comparison of American and Russian women. Med Sci Monit.
2004;10:SR5SR16.
could be a direct causal relation where the abortion increases 6. Broen AN, Moum T, Bdtker AS, et al. The course of mental health after
the likelihood of a mental disorder, or a mental disorder miscarriage and induced abortion: a longitudinal, five-year follow-up study.
BMC Med. 2005;3:18.
increases the likelihood of abortion. Second, there could be an 7. Broen AN, Moum T, Bdtker AS, et al. Predictors of anxiety and depression
indirect mechanism. For example, mental disorder may be following pregnancy termination: a longitudinal five-year follow-up study.
Acta Obstet Gynecol Scand. 2006;85:317323.
associated with poor social support or impulsivity54,55 that 8. Fergusson DM, Horwood LJ, Ridder EM. Abortion in young women and
might lead to an increased likelihood of unplanned pregnancy subsequent mental health. J Child Psychol Psychiatry. 2006;47:1624.

The Canadian Journal of Psychiatry, Vol 55, No 4, April 2010 W 245


Original Research

9. Cougle JR, Reardon DC, Coleman PK. Generalized anxiety following standardized clinical assessments in the WHO World Mental Health surveys. Int
unintended pregnancies resolved through childbirth and abortion: a cohort study J Methods Psychiatr Res. 2006;15:167180.
of the 1995 National Survey of Family Growth. J Anxiety Disord. 37. Kessler RC, Berglund P, Demler O, et al. Lifetime prevalence and age-of-onset
2005;19:137142. distributions of DSM-IV disorders in the National Comorbidity Survey
10. Fergusson DM, Horwood LJ, Boden JM. Abortion and mental health disorders: Replication. Arch Gen Psychiatry. 2005;62:593602.
evidence from a 30-year longitudinal study. Br J Psychiatry. 2008;193:444451. 38. Kessler RC, Chiu WT, Demler O, et al. Prevalence, severity, and comorbidity of
11. Dingle K, Alati R, Clavarino A, et al. Pregnancy loss and psychiatric disorders in 12-month DSM-IV disorders in the National Comorbidity Survey Replication.
young women: an Australian birth cohort study. Br J Psychiatry. Arch Gen Psychiatry. 2005;62:617627.
2008;193:455460. 39. Research Triangle Institute (RTI). Software for Survey Data Analyses
12. Rees DI, Sabia JJ. The relationship between abortion and depression: new (SUDAAN) Version 9.01 [software]. Research Triangle Park (NC): RTI; 2004.
evidence from the Fragile Families and Child Wellbeing Study. Med Sci Monit.
40. Rockhill B, Newman B, Weinberg C. Use and misuse of population attributable
2007;13:CR430CR436.
fractions. Am J Public Health. 1998;88:1519.
13. Pedersen W. Abortion and depression: a population-based longitudinal of young
41. Dwyer JM, Jackson T. Unwanted pregnancy, mental health and abortion:
women. Scand J Public Health. 2008;36:424428.
untangling the evidence. Aust New Zealand Health Policy. 2008;5:2.
14. Pedersen W. Childbirth, abortion, and subsequent substance abuse in young
women: a population-based longitudinal study. Addiction. 2007;102:19711978. 42. Jagannathan R. Relying on surveys to understand abortion behavior: some
15. Reardon DC, Coleman PK, Cougle JR. Substance use associated with cautionary evidence. Am J Public Health. 2001;91:18251831.
unintended pregnancy outcomes in the National Longitudinal Survey of Youth. 43. Jones RK, Kost K. Underreporting of induced and spontaneous abortion in the
Am J Drug Alcohol Abuse. 2004;30:369383. United States: an analysis of the 2002 National Survey of Family Growth. Stud
16. Coleman PK, Reardon DC, Rue VM, et al. A history of induced abortion in Fam Plann. 2007;38:187197.
relation to substance use during subsequent pregnancies carried to term. Am J 44. Bourassa D, Brub J. The prevalence of intimate partner violence among
Obstet Gynecol. 2002;187:16731678. women and teenagers seeking abortion compared with those continuing
17. Reardon DC, Ney PG. Abortion and subsequent substance abuse. Am J Drug pregnancy. J Obstet Gynaecol Can. 2007;29:415423.
Alcohol Abuse. 2000;26:6175. 45. Taft AJ, Watson LF. Termination of pregnancy: associations with partner
18. Gilchrist AC, Hannaford PC, Frank P, et al. Termination of pregnancy and violence and other factors in a national cohort of young Australian women.
psychiatric morbidity. Br J Psychiatry. 1995;67:243248. Aust N Z J Public Health. 2007;31:135142.
19. Gissler M, Hemminki E, Lnnqvist J. Suicides after pregnancy in Finland, 46. Broen AN, Moum T, Bdtker AS, et al. Reasons for induced abortion and their
198794: register linkage study. BMJ. 1996;313:14311434. relation to womens emotional distress: a prospective, two-year follow-up
20. Morgan CL, Evans M, Peters JR. Suicides after pregnancy. Mental health may study. Gen Hosp Psychiatry. 2005;27:3643.
deteriorate as a direct effect of induced abortion. BMJ. 1997;314:902. 47. Congleton GK, Calhoun LG. Post-abortion perceptions: a comparison of
21. Reardon DC, Ney PG, Scheuren F, et al. Deaths associated with pregnancy self-identified distressed and nondistressed populations. Int J Soc Psychiatry.
outcome: a record linkage study of low income women. South Med J. 1993;39:255265.
2002;95:834841. 48. Carey MP, Carey KB, Maisto SA, et al. HIV risk behavior among psychiatric
22. Coleman PK, Coyle CT, Shuping M, et al. Induced abortion and anxiety, mood outpatients: association with psychiatric disorder, substance use disorder, and
and substance abuse disorders: isolating the effects of abortion in the national gender. J Nerv Ment Dis. 2004;192:289295.
comorbidity survey. J Psychiatr Res. 2009;43:770776. 49. Ramrakha S, Caspi A, Dickson N, et al. Psychiatric disorders and risky sexual
23. Coleman PK, Reardon DC, Rue VM, et al. State-funded abortions versus behavior in young adulthood: cross sectional study in birth cohort. BMJ.
deliveries: a comparison of outpatient mental health claims over 4 years. Am J 2000;321:263266.
Orthopsychiatry. 2002;72:141152.
50. Tubman JG, Gil AG, Wagner EF, et al. Patterns of sexual risk behaviors and
24. Reardon DC, Cougle JR, Rue VM, et al. Psychiatric admissions of low-income
psychiatric disorders in a community sample of young adults. J Behav Med.
women following abortion and childbirth. CMAJ. 2003;168:12531256.
2003;26:473500.
25. Morgan JF, Lacey H, Chung E. Risk of postnatal depression, miscarriage, and
preterm birth in bulimia nervosa: retrospective controlled study. Psychosom 51. Larsson M, Aneblom G, Odlind V, et al. Reasons for pregnancy termination,
Med. 2006;68:487492. contraceptive habits and contraceptive failure among Swedish women
26. Blais MA, Becker AE, Burwell RA, et al. Pregnancy: outcome and impact on requesting an early pregnancy termination. Acta Obstet Gynecol Scand.
symptomatology in a cohort of eating-disordered women. Int J Eat Disord. 2002;81:6471.
2000;27:140149. 52. Svare E, Kjaer SK, Thomsen BL, et al. Determinants for non-use of
27. American Psychological Association, Task Force on Mental Health and contraception at first intercourse; a study of 10841 young Danish women from
Abortion. Report of the Task Force on Mental Health and Abortion. 2008 the general population. Contraception. 2002;66:345350.
[Internet]. Washington (DC): APA; 2008 [cited 2009 Mar 20]. Available from: 53. Suri R, Altshuler LA, Mintz J. Depression and the decision to abort. Am J
http://www.apa.org/pi/wpo/mental-health-abortion-report.pdf. Psychiatry. 2004;161:1502.
28. Sit D, Rothschild AJ, Creinin MD, et al. Psychiatric outcomes following medical 54. Moeller FG, Barratt ES, Dougherty DM, et al. Psychiatric aspects of
and surgical abortion. Hum Reprod. 2007;22:878884. impulsivity. Am J Psychiatry. 2001;158:17831793.
29. Major B, Cozzarelli C, Cooper ML, et al. Psychological responses of women 55. Callaghan P, Morrissey J. Social support and health: a review. J Adv Nurs.
after first-trimester abortion. Arch Gen Psychiatry. 2000;57:777784. 1993;18:203210.
30. Pope LM, Adler NE, Tschann JM. Postabortion psychological adjustment: are
minors at increased risk? J Adolesc Health. 2001;29:211.
31. Taft AJ, Watson LF. Depression and termination of pregnancy (induced
abortion) in a national cohort of young Australian women: the confounding Manuscript received January 2009, revised, and accepted May 2009.
effect of womens experience of violence. BMC Public Health. 2008;8:75. This paper was previously presented at the 58th Annual Conference for
32. Steinberg JR, Russo NF. Abortion and anxiety: whats the relationship? Soc Sci the Canadian Psychiatric Association, Vancouver, BC, September 57,
Med. 2008;67:238252. 2008.
33. Cougle JR, Reardon DC, Coleman PK. Depression associated with abortion and 1
Student, Department of Psychology, University of Manitoba, Winnipeg,
childbirth: a long-term analysis of the NLSY cohort. Med Sci Monit.
Manitoba; Research Associate, Department of Psychiatry, University of
2003;9:CR157CR164.
34. Kessler RC, Berglund P, Chiu WT, et al. The US National Comorbidity Survey Manitoba, Winnipeg, Manitoba.
2
Replication (NCS-R): design and field procedures. Int J Methods Psychiatr Res. Professor, Department of Obstetrics, Gynecology, and Reproductive
2004;13:6992. Sciences, University of Manitoba, Winnipeg, Manitoba.
35. Kessler RC, Ustun TB. The World Mental Health (WMH) Survey Initiative 3
Version of the World Health Organization (WHO) Composite International Associate Professor, Departments of Psychiatry, Community Health
Diagnostic Interview (CIDI). Int J Methods Psychiatr Res. 2004;13:93121. Sciences, and Psychology, University of Manitoba, Winnipeg, Manitoba.
36. Haro JM, Arbabzadeh-Bouchez S, Brugha TS, et al. Concordance of the Address for correspondence: Dr J Sareen, PZ-430 PsycHealth Centre, 771
Composite International Diagnostic Interview Version 3.0 (CIDI 3.0) with Bannatyne Avenue, Winnipeg, MB R3E 3N4; sareen@cc.umanitoba.ca.

246 W La Revue canadienne de psychiatrie, vol 55, no 4, avril 2010


Associations Between Abortion, Mental Disorders, and Suicidal Behaviour in a Nationally Representative Sample

Rsum : Associations entre avortement, troubles mentaux, et comportement


suicidaire dans un chantillon nationalement reprsentatif
Objectif : La plupart des tudes antrieures qui ont recherch la relation entre lavortement et la
maladie mentale ont prsent des rsultats partags. Nous avons examin la relation entre
lavortement, les troubles mentaux et la suicidabilit laide dun chantillon nationalement
reprsentatif des .-U.
Mthodes : Les donnes provenaient de la National Comorbidity Survey Replication (n = 3310
femmes de 18 ans et plus). Lentrevue diagnostique composite internationale de lOrganisation
mondiale de la sant a servi valuer les troubles mentaux selon le Manuel diagnostique et
statistique des troubles mentaux, 4e dition, lavortement au cours de la vie. Des analyses de
rgression logistique multiple ont t employes pour examiner les associations entre lavortement et
les troubles de lhumeur, danxit, dutilisation de substances, lalimentation, et les troubles de
comportement perturbateur, ainsi que lidation suicidaire et les tentatives de suicide au cours de la
vie. Nous avons calcul le pourcentage de rpondantes dont les troubles mentaux sont apparus
aprs le premier avortement. Le rle de la violence a aussi t explor. Les fractions attribuables
dans la population ont t calcules pour les associations significatives entre lavortement et les
troubles mentaux.
Rsultats : Aprs ajustement pour les donnes sociodmographiques, lavortement tait associ
une probabilit accrue de troubles mentaux graves troubles de lhumeur (rapport de cotes ajust
[RCA] allant de 1,75 1,91), troubles anxieux (RCA allant de 1,87 1,91), troubles dutilisation de
substances (RCA allant de 3,14 4,99), ainsi quidation suicidaire et tentatives de suicide (RCA
allant de 1,97 2,18). Lajustement pour la violence affaiblissait certaines de ces associations. Pour
tous les troubles examins, moins de la moiti des femmes dclaraient que leur trouble mental avait
dbut aprs le premier avortement. Les fractions attribuables dans la population allaient de 5,8 %
(idation suicidaire) 24,7 % (toxicomanie).
Conclusions : Notre tude confirme une forte association entre lavortement et les troubles
mentaux. Les mcanismes possibles de cette association sont discuts.

The Canadian Journal of Psychiatry, Vol 55, No 4, April 2010 W 247