Major B, Cozzarelli C, Cooper ML, Zubek J, Richards C, Wilhite M, Gramzow RH. Psychological Responses of Women After First-Trimester Abortion. Arch Gen Psychiatry. 2000;57(8):777-784. doi:10.1001/archpsyc.57.8.777
Controversy exists over psychological risks associated with abortion. The objectives of this study were to examine women's emotions, evaluations, and mental health after an abortion, as well as changes over time in these responses and their predictors.
Women arriving at 1 of 3 sites for an abortion of a first-trimester unintended pregnancy were randomly approached to participate in a longitudinal study with 4 assessments—1 hour before the abortion, and 1 hour, 1 month, and 2 years after the abortion. Eight hundred eighty-two (85%) of 1043 eligible women approached agreed; 442 (50%) of 882 were followed for 2 years. Preabortion and postabortion depression and self-esteem, postabortion emotions, decision satisfaction, perceived harm and benefit, and posttraumatic stress disorder were assessed. Demographic variables and prior mental health were examined as predictors of postabortion psychological responses.
Two years postabortion, 301 (72%) of 418 women were satisfied with their decision; 306 (69%) of 441 said they would have the abortion again; 315 (72%) of 440 reported more benefit than harm from their abortion; and 308 (80%) of 386 were not depressed. Six (1%) of 442 reported posttraumatic stress disorder. Depression decreased and self-esteem increased from preabortion to postabortion, but negative emotions increased and decision satisfaction decreased over time. Prepregnancy history of depression was a risk factor for depression, lower self-esteem, and more negative abortion-specific outcomes 2 years postabortion. Younger age and having more children preabortion also predicted more negative abortion evaluations.
Most women do not experience psychological problems or regret their abortion 2 years postabortion, but some do. Those who do tend to be women with a prior history of depression.
UNWANTED pregnancy and abortion are important public health concerns. Since the Supreme Court's landmark 1973 decision in Roe v Wade,1 approximately 1.5 million legal abortions have been performed each year in the United States. Approximately 1 (21%) of 5 American women of childbearing age has had a legal abortion.2- 4 Despite the prevalence of elective abortion, controversy exists about the mental health risks associated with this procedure.5- 7 Some claim that severe psychological distress following abortion is common, and that women who have abortions are prone to experience postabortion syndrome—posttraumatic stress disorder (PTSD) similar to that experienced by some combat veterans and victims of natural disasters, rape, and child abuse.8- 11 Most reviews of empirical research, however, conclude that freely chosen legal abortion, particularly in the first trimester of pregnancy, does not pose a substantial mental health risk.7,12- 16
Drawing firm conclusions about postabortion responses is hampered by methodological limitations in the literature.5,7,16,17 The claim that postabortion problems are common is based primarily on clinical case studies of women who have sought professional help for psychological problems after their abortions or on studies of women who identified themselves in advance as having suffered psychological trauma after an abortion.9,11 These studies are likely to be biased in the direction of overestimating the prevalence of postabortion psychological problems. The claim that postabortion problems are rare is based primarily on studies of random samples of women who arrive at clinics, physicians' offices, or hospitals to have an abortion.18- 21 These studies, although more empirically sound, usually include only short-term assessments of women's postabortion adjustment—a few hours to a few months postabortion. They may be biased in the direction of underestimating longer-term postabortion problems. Most studies also fail to distinguish between clinically significant mental health outcomes (such as depression or psychosis) and feelings of sadness, loss, or regret, which, although unpleasant, do not necessarily signify a psychiatric disorder. Studies also have not addressed whether feelings about an abortion change over time.
The goals of this study were to examine women's general mental health and abortion-related emotions and evaluations following a first-trimester abortion of an unintended pregnancy, changes over time in these responses, and predictors of these responses. Preabortion and postabortion distress (depression) and well-being (self-esteem), and postabortion emotions, appraisals, decision satisfaction, and PTSD were assessed among women followed for 2 years from the day of their abortion.
The sample consisted of 442 women obtaining a vacuum-aspiration abortion between February 1993 and September 1993 at 1 of 3 abortion providers (2 freestanding clinics and 1 physician's office) in Buffalo, NY. The study was restricted to women obtaining a first-trimester abortion of a pregnancy that they indicated was unintended and was not the result of rape. This ensured that the sample reflected the prototypical situation of most women who currently obtain abortions in America.22 During the data collection period, 1749 women arrived at the sites to obtain an abortion. Of these, 1177 were randomly approached to participate and were screened for eligibility according to the criteria outlined above. A total of 134 women were deemed ineligible to participate. This included those who were in their second trimester of pregnancy (n = 70), were not pregnant (n = 2), reported that they intended to become pregnant (n = 7), reported that their pregnancy was the result of rape (n = 1), or were not given time by the clinic to participate prior to their abortion (n = 54). Eighty-five percent (882/1043) of eligible women who were asked to participate agreed. Seventy percent (615/882) were reinterviewed 1 month after their abortion and 442 (50%) of 882 were interviewed 2 years after their abortion. The 442 women interviewed at all 4 time points constitute the final sample. Participants gave informed consent prior to data collection before their abortion and 1 month and 2 years after their abortion.
Demographic characteristics of the final sample and of patients who have abortions nationwide are shown in Table 1. The sample was highly similar demographically to the national profile of patients who have abortions.23 The only notable difference was that Hispanic women were underrepresented, reflecting the small Hispanic population of Buffalo. The mean age at the time of the abortion was 24 years (range, 14-40); the majority were single, white, and raising at least 1 child.
Comparisons of the demographic characteristics of the 85% of women who agreed to participate with the 15% who declined indicated that the 2 groups did not differ significantly on any variable except age. Women who agreed to participate were younger (mean, 23.68 years) than those who declined (mean, 25.92 years) (F1,1042 = 21.16; P<.001). The 442 women in our final sample also were compared with the 440 women who participated initially (on the day of the procedure) but were lost to follow-up subsequently. The 2 groups were compared on all of the demographic (Table 1) and psychological (Table 2) variables assessed prior to the 2-year measurement. No significant differences emerged between the final sample and women lost to attrition on any variable. Thus, our final sample showed no evidence of selection or retention bias.
After their initial screening and counseling sessions with clinic staff, a researcher randomly approached women individually to solicit their participation in a study about women's reactions to having an abortion. Researchers were all working toward or had obtained degrees (PhD or BA) in psychology. Women were assured that the study was being conducted with the clinic's endorsement, that their responses would be confidential and anonymous, and that refusal to participate would in no way affect their treatment at the clinic. Women who consented completed a preabortion questionnaire (the T1 assessment). Approximately 1 to 2 hours later, women underwent a vacuum aspiration abortion. Follow-up questionnaires were completed in the recovery room approximately 1 hour after the abortion (T2). Women were paid $20 for their participation at the 1-month follow-up session (T3) and $50 for their participation in a 2-year follow-up session (T4). T3 questionnaires were completed in person at the clinic, a neutral site (28%), or by mail (72%). T4 questionnaires were completed in person at a neutral site (58%) or by mail (42%). Comparisons between women who completed the follow-up questionnaires in person vs by mail revealed that, at both T3 and T4, women who completed the questionnaires in person were more likely to be members of ethnic minority groups and to receive Medicaid. Method of assessment was not related to any other demographic or outcome measure.
Both abortion-specific and general mental health outcomes were assessed, typically 2 times after the abortion. The time at which each outcome was assessed is indicated in parentheses after the measure is described.
Emotional reactions to the abortion (assessed at T2 and T4) included 6 negative emotions ("sad," "disappointed," "guilty," "blue," "low," and "feelings of loss"), 3 positive emotions ("happy," "pleased," and "satisfied"), and the single emotion of relief. Emotions were assessed on a scale of 1 (not at all) to 5 (a great deal). Negative and positive emotions were separately averaged to form reliable scales of negative (T2, α = .92; T4, α = .92) and positive (T2, α = .88; T4, α = .81) emotion.
Satisfaction with the abortion decision was measured (T3 and T4) with 2 items created for this study ("All in all, how satisfied are you with your decision to have your recent abortion?" and "All in all, how do you feel about your decision to have an abortion?"). Responses were assessed on scales ranging from 1 (very dissatisfied, definitely the wrong decision) to 5 (very satisfied, definitely the right decision) and were averaged (T3, α = .79; T4, α = .80). Higher scores indicate greater satisfaction.
Women's appraisals of abortion-related harm (eg, "I think the abortion has had a negative effect on me) and benefit (eg, "I think the abortion has had a positive [good] effect on me") were each assessed with 3 items (T3 and T4). Responses were assessed on a scale of 1 (strongly disagree) to 5 (strongly agree) and were averaged to form measures of harm (T3, α = .87; T4, α = .84) and benefit (T3, α = .78; T4, α = .70) appraisals.
At T4, women were asked, "If you had the decision to make over again under the same circumstances that you were in 2 years ago, would you make the same decision to have the abortion?" They responded on a scale from 1 (definitely no) to 5 (definitely yes).
Depression was assessed at all 4 time points using the Brief Symptom Inventory,24 a widely used, standardized, and normed questionnaire measure of depression. Respondents indicated on 5-point scales (from 0 = "not at all" to 4 = "a great deal") the extent to which each symptom (eg, feeling lonely, hopeless, or worthless) had bothered them in the month prior to the abortion (T1, α = .80), right now (T2, α = .79), in the month after the abortion (T3, α = .89), or in the past 2 weeks (T4, α = .89).
An adapted version of the Diagnostic Interview Schedule25 was used to diagnose a history of depression of at least 2 weeks' duration (0 = no, 1 = yes) prior to the pregnancy (assessed at T3) and in the 2 years after the abortion (assessed at T4). Normally, the Diagnostic Interview Schedule is administered by an interviewer. Recognizing that some participants would complete measures by mail, we created a questionnaire version of the Diagnostic Interview Schedule that closely resembled the original interview version. A random sample of 35 women completed both forms of the measure. Thirty-two (91%) of these women received the same diagnosis using both methods, a very high agreement rate.
Positive mental health was assessed using a shortened, 4-item version of the Rosenberg Self-Esteem Inventory26 (T1, T3, and T4), a well-validated, widely used measure of self-esteem. Self-esteem is a key component of mental health.27 Women indicated how they usually felt on scales of 1 (strongly disagree) to 5 (strongly agree). Items selected for use had exhibited the highest item-total correlations in a comparable sample.28 The abbreviated scale demonstrated adequate reliability (α at T1 = .76, T3 = .83, and T4 = .76).
The presence of postabortion syndrome was assessed (T4) with a published measure of PTSD created for use with Vietnam War veterans29 that was adapted to make it specific to responses to the abortion. This measure assessed PTSD using diagnostic criteria set forth in the diagnostic manual of the DSM-III-R.30 Women were asked whether the abortion was persistently reexperienced (in dreams or flashbacks, for example); whether there was persistent avoidance of stimuli associated with the abortion (such as efforts to avoid feelings or thoughts associated with abortion); whether there was a numbing of general responsiveness that had not been present before the abortion; and whether there were persistent symptoms of increased arousal (such as difficulty falling asleep). If these symptoms occurred, women were asked whether they lasted more than 1 month. If so, women were classified as meeting the criteria for PTSD; otherwise, they were classified as not showing evidence of this syndrome.
Analyses are presented in several steps. First, we provide descriptive statistics for the outcome variables (Table 3). Second, for outcomes (emotions, appraisals) that are comparable within a specific period, we compare responses within that period using repeated-measures analysis of variance. Third, for outcomes measured across time, we examine mean changes across time using repeated-measures analysis of variance (Table 3). For outcomes assessed at more than 2 periods (depression, self-esteem), we include contrasts between each pair of means. Fourth, we examine correlations among the different outcomes at T4 (Table 4). Fifth, we examine the influence of demographic characteristics, medical complications following abortion, and prior mental health on postabortion adjustment by entering these variables simultaneously into separate multiple regression equations predicting each of the 2-year outcome measures (Table 5). In all cases, statistical significance was considered to be P<.05. All tests were 2-tailed.
At T2, women reported feeling more relief than positive emotions, more relief than negative emotions, and more positive than negative emotions (Table 3). At T4, women continued to feel more relief than either positive or negative emotions. Positive and negative emotions did not differ. Across time, relief and positive emotions declined and negative emotions increased.
At both 1 month and 2 years postabortion, most women felt they had benefited from their abortion more than they had been harmed by it (Table 3). These appraisals did not change over time.
Decision satisfaction was high both at 1 month (T3) and 2 years (T4) postabortion, but it decreased over time (Table 3). At T3, 329 (78.7%) of 418 women reported that they had made the right decision and that they were satisfied with their decision (ie, their mean rating for the 2 decision satisfaction items was above the midpoint of the 1-5 scale); 45 (10.8%) of 418 were dissatisfied and felt they had made the wrong decision; and 44 (10.5%) of 418 were neutral. At T4, 301 (72%) of 418 were satisfied and 68 (16.3%) of 418 were dissatisfied.
Three hundred six (69%) of 441 women said they would definitely or probably have the abortion again if they had to make the decision over; 84 (19%) of 441 said that they would definitely not or probably not; and 51 (12%) of 441 were undecided.
Pairwise comparisons indicated that depression levels decreased from T1 to T2, and increased from T2 to T3 and from T3 to T4 (Table 3). Depression scores were lower at all times postabortion than preabortion. Direct comparisons across time are hampered, however, by the different time frames used to assess depressive symptoms across the 4 time points.
Diagnosis of clinical depression on the basis of the Diagnostic Interview Schedule revealed that 99 (26%) of 386 of the women had experienced an episode of clinical depression at some time prior to the pregnancy, whereas 78 (20%) of 386 had experienced an episode of clinical depression in the 2 years after their abortion (Table 3). Self-esteem increased over time and was higher postabortion than preabortion (Table 3). Six (1%) of 442 women met the diagnostic criteria for PTSD based on their responses to the abortion-specific measure (Table 3).
Pearson correlation coefficients revealed considerable covariation among the 2-year postabortion outcomes (Table 4). Women who had better mental health (eg, less depression and higher self-esteem) also reported more positive abortion emotions and evaluations. Based on their Brief Symptom Inventory scores and decision satisfaction, 27 (6.2%) of 438 were both dissatisfied with their decision (had scores below the scale midpoint for satisfaction) and clinically depressed (had scores above the published cutoff for depression) 2 years postabortion. Two hundred eighty-five (65%) of 438 were neither depressed nor dissatisfied with their decision.
Multiple regression analyses simultaneously examined demographic characteristics, prior mental health, and women's reports of physical complications from the abortion as predictors of 2-year postabortion outcomes. Physical complications were assessed 1 month after abortion with a single dichotomous (yes/no) item asking women whether they "had experienced physical complications (eg, abnormal bleeding or pelvic infection) since their abortion." Seventy-three (17%) of 431 indicated yes. Regression analyses revealed that a prepregnancy history of depression consistently predicted poorer postabortion mental health and more negative abortion-related emotions and evaluations (Table 5). In addition, younger women evaluated their abortion more negatively, as did women who had more children at the time of the abortion. African American women had higher self-esteem than did women of other ethnic groups. The positive association observed between Hispanic ethnicity and postabortion depression is suspect due to the small number of Hispanic women sampled (n = 11). No other variable in the model was associated significantly with any outcome measure 2 years postabortion, including whether women reported physical complications after an abortion.
Results support prior conclusions that severe psychological distress after an abortion is rare.7,12- 21,31- 35 The percent of women experiencing clinical depression within 2 years after abortion (20%) equals the rate of depression nationally among all women 15 to 35 years of age (20%).36 Mental health did not decline postabortion. The rate of PTSD associated with abortion (1%) was substantially lower than the rate of PTSD in the general population of women in this age group (10.75%) and than the rate following traumas such as childhood physical abuse (48.5%) or rape (46%).37 Most women were satisfied with their decision, believed they had benefited more than had been harmed by their abortion, and would have the abortion again. These findings refute claims that women typically regret an abortion.8- 11 Nonetheless, 16.3% were dissatisfied and 19% would not make the same decision again. Over time, negative emotions increased and decision satisfaction decreased. Although sadness and regret are not psychological disorders, these feelings should not be dismissed.
As in prior research,20,27,28 preabortion mental health emerged as the best predictor of postabortion mental health and feelings about an abortion.38- 45 Women with a prior history of depression may be predisposed to subsequent depression and regret, regardless of whether or not they have an unintended pregnancy and how they choose to resolve that pregnancy. Younger women and those who had more children preabortion also were more likely to evaluate their abortion negatively.
The method of the current study improves on prior studies of adjustment to abortion in several ways, thereby permitting stronger conclusions about the prevalence of postabortion psychological problems. Nonetheless, some limitations deserve comment. First, as a result of the stringent anonymity and confidentiality requirements involved in following an abortion sample over time, approximately half of our original sample was lost to attrition prior to the 2-year follow-up. This high attrition rate raises concerns about whether the final sample was representative of the initial group. Confidence that it was, and that postabortion problems were neither overestimated nor underestimated, is increased, however, by the lack of evidence of retention bias in the final sample. Second, all outcomes were measured with self-report instruments. To the extent that women are unaware of their true feelings, responses on self-report instruments might not accurately reflect those feelings. True feelings of postabortion regret may be overestimated or underestimated. Third, the design of this study does not permit determination of whether psychological distress reported by our participants after abortion was caused by the abortion or by other events (eg, divorce or job loss) that intervened between the abortion and subsequent assessments of distress. Ethical considerations preclude the randomized, controlled experiments necessary to show definitively the effects of abortion on mental health. A fourth limitation was the lack of a good baseline measure of mental health prior to the discovery of the pregnancy. To have such a baseline, a massive longitudinal study of women's mental health over time is needed, in which pregnancies and resolutions of those pregnancies are carefully tracked. Fifth, Hispanic women were underrepresented in our sample; thus, our conclusions may not be generalizable to this group. Finally, our data do not address truly long-term adjustment to abortion. At 5 or 10 years postabortion, more women may have experienced intervening events (eg, birth of children, fertility problems, or marriage) that lead them to reappraise a prior abortion, either in a more positive or more negative light. Additional postabortion follow-up is needed to address this issue.
Ultimately, the psychological risks of abortion must be compared with the psychological risks of its alternatives. When women become pregnant unintentionally, they have few alternatives, any of which could be a source of regret or distress. Studies of women who give up a child for adoption suggest that feelings of loss and sadness are common,46 although no well-controlled studies have compared the reactions of these women with reactions of women who have an abortion. In contrast, studies comparing the mental health of women who have an abortion and women who carry an unintended pregnancy to term and keep the child are more common. These studies consistently find that the former are at no greater risk for psychological problems than the latter.21,31- 35,47 Thus, for most women, elective abortion of an unintended pregnancy does not pose a risk to mental health.
Accepted for publication October 8, 1999.
This study was supported by research grant 5R01MH47989 from the National Institute of Mental Health, Rockville, Md (Dr Major), and by an award from the California Wellness Foundation/University of California Wellness Lecture Program, Woodland Hills (Dr Major).
We thank the clinic staff who assisted in this project and the women who participated, without whom this study could not have been carried out. We also thank Wendy Quinton, MA, for her assistance with statistical analyses.
Corresponding author: Brenda Major, PhD, Department of Psychology, University of California, Santa Barbara, CA 93106 (e-mail: firstname.lastname@example.org).