Thirty percent of U.S. women will have an abortion by the time they are age 45 years.1 Understanding whether such a common procedure causes mental health problems is important for clinical practice and policy. Conflicting findings currently exist in the literature; these reflect limitations of available data and variations in methodologic rigor allowing for control over confounding factors. Some of the strongest evidence has come from research linking Danish population registries reporting reproductive events and those reporting inpatient and outpatient psychiatric admissions. Although analyses using these registries found higher rates of psychiatric admissions in the 12 months after abortion compared with the 12 months after birth, rates of these disorders were also higher during the 9 months before abortion compared with the 9 months before birth.2
This research addresses shortcomings in existing U.S. studies, including inappropriate comparison groups, inadequate measurement of mental health outcomes, and failure to control for confounding factors.3–5 The study reported here tests whether women's risk of having clinical mental health problems, assessed by structured psychiatric interviews, is higher after a first abortion compared with after a first childbirth, contrasting findings with and without adjustment for possible confounding factors. Moreover, the study here also contrasts findings with another study using the same data set that examined the association between lifetime abortion (compared with no abortion) and lifetime mental health outcomes,6 which means it did not assess when the mental health outcomes occurred relative to the abortion.
The current study extends existing research by using a U.S. national sample on which psychiatric disorders meeting Diagnostic and Statistical Manual of Mental Disorders, 4th Edition (DSM-IV) criteria were assessed. This provides an evaluation of actual incidence of disorders that is not biased by health care use.
PATIENTS AND METHODS
The National Comorbidity Survey-Replication was designed to be representative of English-speaking adults ages 18 years or older living in the noninstitutionalized civilian household population of the continental United States.7 It was administered by trained lay interviewers using laptop computer-assisted personal interview methods between February 2001 and April 2003. National Comorbidity Survey-Replication data assess the prevalence of DSM-IV mental health disorders in the United States and their correlates7,8 and is the most recent cohort of the National Comorbidity Surveys. More specific details on study design and measures may be found elsewhere.7,8
The National Comorbidity Survey-Replication interview had two parts. Part 1 assessed presence and history of mental health disorders and suicidal behaviors. Part 2 assessed potential correlates of these disorders among a subsample of Part 1 respondents, consisting of those who screened positive for a mental health disorder in Part 1 supplemented by a probability subsample of other respondents. Because questions pertaining to pregnancy history and adverse exposures were administered in Part 2, the sample consisted of women who were administered both Parts 1 and 2. Of these, only those who were 13 years of age or younger in 1973, when Roe v Wade 9 legalized having an abortion in the United States (n=1,633), were included to ensure women's entire reproductive lifespans occurred during the time when abortion was legal. Eighty-nine who reported not having had sex, 124 missing on pregnancy history information necessary for this study, and 484 who reported having had no children and no abortions were excluded, leaving 936 women aged 18–42 years for analysis (see Appendix 1, available online at http://links.lww.com/AOG/A463 for flow diagram of sample). This research was exempt from the University of California, San Francisco Committee on Human Research approval because data were deidentified.
Women were asked questions about previous abortions, childbirths, and miscarriages. Women were coded as having had an abortion if they reported having had an abortion, the number they had, and the age at their first abortion. Of the 936 women, 259 reported having had an abortion (191 only one and 68 more than one abortion). Women reporting an abortion were asked their age at first abortion, which was used to code when mental health problems occurred relative to a woman's first abortion.
Women reported the total number of children to which they ever gave birth and the current age of their biological children in three ranges up to 17 years old. Because no ranges went beyond 17 years old, we did not have age ranges for 110 women who had childbirths and no abortions and whose oldest child was 18 years or older. Women were coded as having had a delivery if they reported having had no abortions and had all children aged 17 years or younger. Women's exact age at first childbirth was not elicited. For those whose children were all aged 17 years or younger, age at first birth can be bracketed to within a range based on the age of the respondent at the time of the interview and the age range (0–4 years, 5–12 years, or 13–17 years) within which the oldest biological child falls. This allowed us to compute a possible range for a woman's age at first childbirth to determine when mental health problems occurred relative to her first childbirth. Those who had children 18 years or older (n=110) were excluded from analyses. The analyses presented used the youngest possible age within the range because this attributes the maximum number of mental health disorders to be in the postpregnancy period and the mean age at first childbirth using the youngest possible age was the same as the mean age at first abortion. In Appendixes 2, 4, and 5 (available online at http://links.lww.com/AOG/A463), we explore sensitivity analyses to assess the affect of this choice.
Based on prior research with these data,10–12 20 mental health disorders that met the DSM-IV, Text Revision 13 criteria and three suicidal behaviors were grouped into the following categories: anxiety disorders (panic disorder, agoraphobia, specific phobia, social phobia, generalized anxiety disorder, and posttraumatic stress disorder), mood disorders (major depression, bipolar disorder, and dysthymia), impulse-control disorders (attention deficit disorder, conduct disorder, oppositional defiant disorder, and intermittent explosive disorder), substance use disorders (alcohol abuse without dependence, alcohol dependence, drug abuse without dependence, drug dependence), eating disorders (anorexia, bulimia, and any binge eating disorder), and suicidal behaviors (thinking about suicide, making plans for suicide, attempting suicide).
For each category we computed whether a woman experienced a disorder postpregnancy and, if so, the number of years postpregnancy that the first disorder of that type occurred. The reported age of onset and most recent occurrence were linked to the woman's age at first abortion or age at first childbirth to determine whether mental health disorders occurred before or after her pregnancy outcome. If the disorder was recorded at exactly the age of the woman's first abortion or first childbirth, this was coded as occurring postpregnancy (at 0.5 months so it would be included in analyses). If a disorder occurred after a woman's first abortion or first childbirth, the number of years after this event that the disorder occurred was computed. Findings did not differ if we coded mental health disorders that occurred at the age of the woman's first abortion or childbirth as occurring before the pregnancy. We present findings with mental health outcomes as occurring at the same age as the pregnancy coded as occurring postpregnancy.
We examined factors that have been shown to differ between women who have abortions compared with women who do not14–20: number of prefirst abortion or prefirst childbirth (henceforth prepregnancy) mental health problems, number of prepregnancy adverse exposures, prepregnancy miscarriage history, age at first pregnancy, race or ethnicity, and childhood economic status (low compared with not low). Number of prepregnancy mental health problems was determined by assessing whether each of the mental health disorders occurred before women's first abortion (if having had an abortion) or first childbirth (if having had no abortions). The number occurring beforehand was then coded into none, one, two, and three or more mental health problems prepregnancy. Number of prepregnancy adversities was computed by summing the number of the following experiences that occurred to the women in the sample before their first abortion or childbirth: any parental loss (parental death, parental divorce, or the woman going away for 6 months or longer to foster care or to live with other relatives), any parental mental illness (depression, anxiety, or substance use disorder), parental criminal behavior, parental violent conflict, physical abuse, sexual abuse, childhood neglect, any personal safety threat, and intimate partner violence. Number of prepregnancy adversities was then coded into none, one, or two or more. Sociodemographic factors we included were age at pregnancy event, race or ethnicity (non-Hispanic white, African American, Hispanic, other) and childhood economic status (low compared with not low). Similar to another study with this data set,21 childhood economic status was coded as low if women reported growing up in a family that received government assistance for 6 months or more during childhood. In addition, women who reported living with only one parental figure and that parent had less than a high school education, women who reported living with two parental figures, both of whom had less than a high school education, or women who reported living with no parental figures were coded as growing up in low economic circumstances. For 29 women who did not have information on parental education, we followed Green and colleagues' definition21 in which women who grew up with both a male and female parental figure or in single-parent homes where the parent worked “most or all of the time” were coded as not growing up in low childhood economic status. We did not have information on respondents' marital status or education level at the time of the pregnancy event.
Survival analyses for each of the six categories of mental health problem were conducted using Cox proportional hazard models. The age at first abortion or childbirth was considered the entry point into the study and we analyzed time from then until a first mental health problem in that category or if no mental health problem until their age at the time of interview. Following recommendations,22 we checked that the proportional hazard assumption was met for the abortion compared with childbirth coefficient.
We considered several models with various levels of adjustment for confounding. Model 1 tested the relationship between first abortion compared with first delivery and time until postpregnancy onset of a mental health problem in an unadjusted Cox proportional hazards model. Model 2 adjusted for prepregnancy mental health only. Model 3 adjusted for prepregnancy adversities, prepregnancy miscarriage, age at time of pregnancy, and childhood economic status; and Model 4 controlled for all the factors in Models 2 and 3. The data were collected using a multistage cluster probability area sample, which requires special analytic care. Using Stata 10.1, we incorporated sampling weights for the Part 2 subsample to guarantee national representativeness and clusters and strata to correctly calculate standard errors with this design.
Women in the abortion group were more likely to have never been married (P=.001), to have had more prior adverse experiences (P=.049), and to have had a prepregnancy miscarriage (P=.044) than women in the childbirth group (Table 1). In addition, before their pregnancy, women who had abortions had significantly more mental health disorders overall (P<.001) and were significantly more likely to have had an anxiety disorder (P=.004), mood disorder (P<.001), substance use disorder (P<.001), and suicidal ideation (P<.001) and marginally more likely to have had an impulse-control disorder (P=.059) than women in the childbirth group (Table 2).
The total follow-up time for women was 8,095 years for anxiety disorders, 8,437 years for mood disorders, 8,909 years for impulse-control disorders, 8,923 years for substance use disorders, 9,645 years for eating disorders, and 8,920 years for suicidal ideation. In addition, 50% of women contributed 9 or more years for anxiety disorders, 10 or more years for mood disorders, 12 or more years for impulse-control disorders, 12 or more years for substance use disorders, 12 or more years for eating disorders, and 13 or more years for suicidal ideation.
Women who had abortions were more likely to have more mental health problems postpregnancy than women in the childbirth group (P<.001) and to have each type of disorder or suicidal ideation at some point postpregnancy (all P values <.045; Table 2). This is reflected in hazard ratios presented in Table 3, which show that women who had an abortion had an increased hazard of having a postpregnancy anxiety disorder (hazard ratio [HR] 1.52, 95% confidence interval [CI] 1.08–2.15), mood disorder (HR 1.56, 95% CI 1.23–1.98), impulse-control disorder (HR 1.62, 95% CI 1.02–2.57), substance use disorder (HR 3.05, 95% CI 1.94–4.79), eating disorder (HR 2.53, 95% CI 1.09–5.86), and suicidal ideation (HR 1.62, 95% CI 1.09–2.40) compared with women in the childbirth group, when no factors were considered in analyses (Model 1).
Adjustment for number of prepregnancy mental health disorders markedly attenuated the HRs (Model 2). However, the HRs for all of the psychiatric disorders, with the exceptions of impulse-control and eating disorders, remained significant after adjustment for prepregnancy adversities, prepregnancy miscarriage, age at the time of pregnancy, and childhood economic situation (Model 3). Adjustment for prepregnancy mental health and the factors included in Model 3 (Model 4) revealed HRs similar to Model 2. Once prepregnancy mental health is included, none of the HRs except for substance use disorders remained significant: anxiety disorders HR 1.12 (95% CI 0.87–1.46), mood disorders HR 1.18 (95% CI 0.88–1.56), impulse-control disorders HR 1.10 (95% CI 0.75–1.62), substance use disorders HR 2.30 (95% CI 1.35–3.92), eating disorders HR 1.82 (95% CI 0.63–5.25), and suicidal ideation HR 1.25 (95% CI 0.88–1.78).
Appendix 3 (available online at http://links.lww.com/AOG/A463) presents the HRs of all factors in Model 4. Having three or more mental health problems compared with none before the pregnancy event significantly increased women's hazards of having postpregnancy mental health disorders (all P values <.005). In addition, women who were younger at the time of their pregnancy event were more likely to have postpregnancy mental health problems for all disorders except mood and eating disorders (all P values <.011).
To make the findings more concrete, Table 4 presents descriptive analyses giving the predicted probability of having a first postpregnancy mental health problem by 5 years postpregnancy for each of the six mental health outcomes for various prototypical women. For all outcomes, with the exception of substance use disorders, the likelihood of having a disorder was not statistically different between the abortion and childbirth groups (see Table 3). For instance, for a woman who is 20 years of age at the time of her pregnancy, the predicted probability of having a mood disorder by 5 years postpregnancy for a woman with no prepregnancy mental health problems is .05 for both the abortion and childbirth groups.
Results from the sensitivity analysis, presented in Appendices 4 and 5 (available online at http://links.lww.com/AOG/A463), yielded findings consistent with the findings presented. One difference, however, should be noted. When using the oldest possible age at first childbirth for the childbirth group, there was no statistically significant association between abortion and subsequent substance use disorders in Model 4 (HR 1.53, 95% CI 0.91–2.56).
The current research adds to the literature by using a nationally representative U.S. sample, examining clinical-level mental health disorders, comparing women who abort with women who give birth, and controlling for a range of mental health disorders occurring before the pregnancy. We found that although abortion compared with childbirth was associated with all six categories of disorders in unadjusted models, in models adjusted for prepregnancy mental health, the associations between abortion and subsequent anxiety, mood, impulse-control, and eating disorders and suicidal ideation were markedly attenuated and nonsignificant, whereas substance abuse remained significant. In addition, prepregnancy mental health was a strong predictor of postpregnancy mental health. These results, along with our findings that women having abortions had a higher number of prepregnancy mental health disorders, lend support to the perspective that the link between abortion and subsequent mental health link may be driven by other factors—in this case prior mental health.3,23 They also support other research showing that prepregnancy mental health is a strong predictor of postpregnancy mental health3–5,13–16,19,23,24 and having abortions.15,16,23
Unlike the other disorders, the association between abortion and substance use disorders remained significant, although it was reduced after controlling for prepregnancy mental health problems. This may reflect an actual association between abortion and substance use or merely inadequate control for all factors specifically related to substance use disorders such as women's risk-taking tendencies. Such risk-taking tendencies may be manifested by substance use and sexual behaviors that lead to having unwanted pregnancies (and abortions). Therefore, the association between abortion and substance use may have remained significant because we were not able to control for risk-taking tendencies. Alternatively, subthreshold substance use problems, which were not controlled for in analyses, are likely related to having both unwanted pregnancies and subsequent substance use disorders. It could also be that both the internal motivation and social pressure for women to avoid substance use during pregnancy carries over to the period of motherhood and contributes to a lower risk of substance use disorders after a childbirth relative to after an abortion. It should also be noted that this association was not statistically significant in supplementary analyses when oldest possible age at first childbirth was used (see Appendix 5, http://links.lww.com/AOG/A463) and all factors were included, suggesting it may not be reliable.
These results differ from a prior analysis with the National Comorbidity Survey-Replication data, which reported a significant association between lifetime abortion and lifetime mood disorders, anxiety disorders, substance use disorders, and suicidal ideation. Although the researchers controlled for age at interview, marital status, race, education, income, and violence experience,6 they did not differentiate prepregnancy from postpregnancy mental disorders. By using lifetime mental health as the outcome, they could not determine if the mental health disorder occurred before or after the abortion. The current analysis coded mental health postabortion or postchildbirth, illustrating the importance of considering timing of abortion relative to the mental health problem and controlling for prior mental health.
Although the birth group in the current study is better matched to the abortion group than in other prior research, they are likely to differ in substantial ways, only some of which can be adequately adjusted. In the United States, 95% of abortions and 37% of births are the result of unintended pregnancies.25,26 Limiting the birth group to women who had delivered at least one neonate and reported no abortions meant that many women in the comparison group had successfully avoided unintended pregnancies and were thus less likely to have some of the risk factors for mental health disorders, which are also risk factors for unintended pregnancy. That we found no significant association between abortion and subsequent mental health after accounting for prepregnancy mental health for five outcomes even with this differential selection provides strong evidence against the claim that abortion significantly harms women's mental health.
The use of the National Comorbidity Survey-Replication has both advantages and disadvantages. Although data are representative of English-speaking noninstitutionalized women living in the continental United States, they are not representative of those living in Hawaii or Alaska or who are not fluent in English. In addition, although clinical mental health disorders are assessed according to DSM-IV criteria, data are based on self-report and recall of life events. Even though care was taken to maximize accuracy of reporting and diagnoses, some mental health inaccuracy in diagnoses and abortion underreporting are likely.27,28 For instance, in a similar sample from the National Comorbidity Survey-Replication, we found that only 44% of abortions were reported17; this is similar to the level of underreporting of abortion from other national data sets.28 It is not known how inaccuracy of mental health diagnoses or reporting of abortion influences the findings. Nevertheless, these findings support previous prospectively collected data, which has compared women who abort with women who give birth.2,29,30 For instance, in a reanalysis of a sample from New Zealand, Fergusson and colleagues29 found that, compared with women who gave birth to unwanted pregnancies, women who had abortions were not at a significant increased risk of subsequent anxiety disorders, depressive disorders, substance use disorders, and suicidal ideation when mental health at age 15 years, childhood adversities, and family environment were controlled in analyses. In addition, research using Danish registries data has found support for the notion that mental health problems around the time of abortion are driven by factors other than the abortion.2,30
Groups opposed to abortion have used studies showing associations between abortion and mental disorders to advocate for restrictive policies. The results reported here show that policies that require women be told that abortion increases their risk of anxiety, depression, and suicide31,32 lack an evidence base. In this study, it was not abortion that increased women's risk of these mental health problems, but rather prior mental health problems that increased women's risk for both abortion and subsequent mental health disorders and suicidal behaviors. This does not mean that abortion care settings should ignore mental health. Women seeking abortions may be at higher risk of prior untreated mental health disorders and the abortion care setting may be an important intervention point for mental health screening and referrals.
1. Jones RK, Kavanaugh ML. Changes in abortion rates between 2000 and 2008 and lifetime incidence of abortion. Obstet Gynecol 2011;117:1358–66.
2. Munk-Olsen T, Laursen TM, Pedersen CB, Lidegaard Ø, Mortensen PB. Induced first-trimester abortion and risk of mental disorder. N Engl J Med 2011;364:332–9.
3. Major B, Appelbaum M, Beckman L, Dutton MA, Russo NF, West C. Abortion and mental health: evaluating the evidence. Am Psychol 2009;64:863–90.
4. National Collaborating Centre for Mental Health. Induced abortion and mental health: a systematic review of the mental health outcomes of induced abortion, including their prevalence and associated factors. London (UK): Academy of Medical Royal Colleges; 2011.
5. Charles VE, Polis CB, Sridhara SK, Blum RW. Abortion and long-term mental health outcomes: a systematic review of the evidence. Contraception 2008;78:436–50.
6. Mota NP, Burnett M, Sareen J. Associations between abortion, mental disorders, and suicidal behaviour in a nationally representative sample. Can J Psychiatry 2010;55;239–47.
7. Kessler RC, Berglund P, Chiu WT, Demler O, Heeringa S, Hiripi E, et al.. The US National Comorbidity Survey Replication (NCS-R): design and field procedures. Int J Methods Psychiatr Res 2004;13:69–92.
8. Kessler RC, Ustun TB. The World Mental Health (WMH) Survey Initiative Version of the World Health Organization (WHO) Composite International Diagnostic Interview (CIDI). Int J Methods Psychiatr Res 2004;13:93–121.
9. Roe v Wade, 410 U.S. 113 (1973).
10. Kessler RC, Berglund P, Demler O, Jin R, Merikangas KR, Walters EE. Lifetime prevalence and age-of-onset distributions of DSM-IV disorders in the National Comorbidity Survey Replication. Arch Gen Psychiatry 2005;62:593–602.
11. Hudson JI, Hiripi E, Pope HG Jr, Kessler RC. The prevalence and correlates of eating disorders in the National Comorbidity Survey Replication. Biol Psychiatry 2007;61:348–58.
12. Nock MK, Hwang I, Sampson NA, Kessler RC. Mental disorders, comorbidity and suicidal behavior: results from the National Comorbidity Survey Replication. Mol Psychiatry 2010;15:868–76.
13. American Psychiatric Association. Diagnostic and statistical manual of mental disorders–IV–TR. Washington (DC): American Psychiatric Association; 2000.
14. Fergusson DM, Horwood LJ, Boden JM. Abortion and mental health disorders: evidence from a 30-year longitudinal study. Br J Psychiatry 2008;193:444–51.
15. Steinberg JR, Russo NF. Abortion and anxiety: what’s the relationship? Soc Sci Med 2008;67:238–52.
16. Steinberg JR, Becker D, Henderson JT. Does the outcome of a first pregnancy predict depression, suicidal ideation, or lower self-esteem? Data from the National Comorbidity Survey. Am J Orthop 2011;81:193–201.
17. Steinberg JR, Tschann JM. Childhood adversities and subsequent risk of one or multiple abortions. Soc Sci Med 2013;81:53–9.
18. Gee RE, Mitra N, Wan F, Chavkin DE, Long JA. Power over parity: intimate partner violence and issues of fertility control. Am J Obstet Gynecol 2009;201:148.e1–7.
19. Taft AJ, Watson LF. Depression and termination of pregnancy (induced abortion) in a national cohort of young Australian women: the confounding effect of women’s experience of violence. BMC Public Health 2008;875.
20. Bleil ME, Adler NE, Pasch LA, Sternfeld B, Reijo-Pera RA, Cedars MI. Adverse childhood experiences and repeat induced abortion. Am J Obstet Gynecol 2011;204:122.e1–6.
21. Green JG, McLaughlin KA, Berglund PA, Gruber MJ, Sampson NA, Zaslavsky AM, et al.. Childhood adversities and adult psychiatric disorders in the national comorbidity survey replication I: associations with first onset of DSM-IV disorders. Arch Gen Psychiatry 2010;67:113–23.
22. Vittinghoff E, Glidden DV, Shiboski SC, McCulloch CE. Chapter 6: Survival analysis. In:Vittinghoff E, Glidden D, Shiboski S, McCulloch C, editors. Regression methods in biostatistics: linear, logistic, survival, and repeated measures models. 2nd ed. New York (NY): Springer; 2012. p. 203–59.
23. Steinberg JR, Finer LB. Examining the association of abortion history and current mental health: a reanalysis of the National Comorbidity Survey using a common-risk-factors model. Soc Sci Med 2011;72:72–82.
24. O’Hara MW, McCabe JE. Postpartum depression: current status and future directions. Annu Rev Clin Psychol 2013;9:379–407.
25. Finer LB, Zolna MR. Unintended pregnancy in the United States: incidence and disparities, 2006. Contraception 2011;84:478–85.
26. Mosher WD, Jones J, Abma JC. Intended and unintended births in the United States: 1982–2010. In: National health statistics reports 2012. Hyattsville (MD): National Center for Health Statistics; 2012.
27. Kessler RC, Abelson J, Demler O, Escobar JI, Gibbon M, Guyer ME, et al.. Clinical calibration of DSM-IV diagnoses in the World Mental Health (WMH) version of the World health Organization Composite International Diagnostic Interview (WMH-CIDI). Int J Methods Psychiatr Res 2004;13:122–39.
28. Jones RK, Kost K. Underreporting of induced and spontaneous abortion in the United States: an analysis of the 2002 National Survey of Family Growth. Stud Fam Plann 2007;38:187–97.
29. Fergusson DM, Horwood LJ, Boden JM. Does abortion reduce the mental health risks of unwanted or unintended pregnancy? A reappraisal of the evidence. Aust N Z J Psychiatry 2013;47:817–27.
30. Munk-Olsen T, Laursen TM, Pedersen CB, Lidegaard O, Mortensen PB. First-time first-trimester induced abortion and risk of readmission to a psychiatric hospital in women with a history of treated mental disorder. Arch Gen Psychiatry 2012;69:159–65.
© 2014 by The American College of Obstetricians and Gynecologists. Published by Wolters Kluwer Health, Inc. All rights reserved.
31. Planned Parenthood v Rounds. 686 F. 3d 889 (8th Circuit 2012).