Gilchrist

Gilchrist AC, Hannaford PC, Frank P, Kay CR. Termination of pregnancy and psychiatric morbidity. Br J Psychiatry. 1995;167:243-248.

Please register and contribute to the development of these notes into a narrative by editing the sections or adding sections.

Abstract

Gilchrist AC, Hannaford PC, Frank P, Kay CR. Termination of pregnancy and psychiatric morbidity.Br J Psychiatry. 1995;167:243-248.

BACKGROUND. We investigated whether reported psychiatric morbidity was increased after termination of pregnancy compared with other outcomes of an unplanned pregnancy.

METHOD. This was a prospective cohort study of 13,261 women with an unplanned pregnancy. Psychiatric morbidity reported by GPs after the conclusion of the pregnancy was compared in four groups: women who had a termination of pregnancy (6410), women who did not request a termination (6151), women who were refused a termination (379), and women who changed their minds before the termination was performed (321).

RESULTS. Rates of total reported psychiatric disorder were no higher after termination of pregnancy than after childbirth. Women with a previous history of psychiatric illness were most at risk of disorder after the end of their pregnancy, whatever its outcome. Women without a previous history of psychosis had an apparently lower risk of psychosis after termination than postpartum (relative risk RR = 0.4, 95% confidence interval CI = 0.3-0.7), but rates of psychosis leading to hospital admission were similar. In women with no previous history of psychiatric illness, deliberate self-harm (DSH) was more common in those who had a termination (RR 1.7, 95% CI 1.1-2.6), or who were refused a termination (RR 2.9, 95% CI 1.3-6.3).

CONCLUSIONS. The findings on DSH are probably explicable by confounding variables, such as adverse social factors, associated both with the request for termination and with subsequent self-harm. No overall increase in reported psychiatric morbidity was found.

Additional Key Findings

1. The findings confirmed that women with prior psychiatric problems are worse off postabortion
2. Women with the most fragile mental health prior to an abortion, i.e., psychosis, were worse off postabortion
3. The findings indicated that among women with no prior psychiatric history, significantly higher risks of deliberate self harm were observed both after an abortion and after a refused abortion.

Additional Notes Regarding Population Sample and Methodology

1. Following screening and risk-benefit analyses, attending physicians refused to peform abortions on 379 women.
2. An additional 321 women changed their minds after screening and consultation with their attending physicians.
3. British women who do not have abortions were underrepresented in the study. In the study sample 48.3% of the women had abortions, a percentage which is much higher than the abortion rate in the UK. One source reports that only 22.8% of pregnancies in the UK end in abortion.[1]
4. All general practitioners reporting were volunteers and were not blind to condition when making their counts. The authors do not disclose the conditions under which volunteers were selected, nor the rate of volunteers among those invited to volunteer, nor any measure or attempts to grade or screen the volunteer physicians relative to age, gender, practice or attitudes regarding abortion, or any other factors which might influence the observer's judgments and reports. This self-selected group of participating physicians may have been biased. Surveys of GP's in Britain find that about 80% report a "pro-choice" perspective which may influence their recommendations for abortion and their subjective interpreation of post-abortion reactions. Clearly, those who recommend for abortion would be disinclined to believe that their recommendations were in error. See additional notes below regarding the reluctance of women to return to physicians for followup care following an abortion.
5. GP's reported details every 6 months.
6. Data was reported without any actual followup interviews on the part of the GP. A GP who had not seen a patient in the last six months might therefore simply report that there were no observed psychological problems.
7. Information was obtained only from women who volunteered and "agreed to their family doctor supplying anonymous data to the study center." (Research shows that women who expect to deal poorly with an abortion do in fact have more post-abortion problems. Such women might prefer not to be excluded from a followup study for fear of being exposed to additional stress.)
8. Selection bias may have occured among women volunteers.
9. According to the authors, "Had follow-up interviews been required, it is likely that participation would have been greatly reduced; in a pilot survey nearly half of the women who had a termination said that they would refuse to participate if they could not remain anonymous."
10. The findings are inconsistent with record based research in Canada which found that 24% of women who had abortions subsequently made visits to psychiatrists compared to 3% in the general population. and record based research in the United States (Reardon, CMAJ).

Strengths

1. It was prospective with a large sample size

1. The study used four comparison groups

1. those who never requested abortion, including the combination of both those who delivered healthy babies and those who miscarried or had other adverse results;
2. those who had an induced abortion;
3. those who originally requested abortion but changed their minds after consulting with physician; and
4. those who requested termination but for whome physicians refused to perform the abortion after screening and a risk/benefit analysis.

Weaknesses

1. This study is not applicable to American experience because British abortion law is much more protective of women's health and requires a level of screening, counseling, and risk benefit analysis not normally found in the United States. In Britain, before an abortion is performed two medical doctors have to evaluate the patient and both agree that the risks of abortion are less than the risk associated with childbirth. In the sample used for this study, 700 women (approximately 10% of all those requesting an abortion) did not have an abortion after a risk-benefit screening and consultation with their physicians. It seems apparent that these women were likely at greatest risk of adverse outcomes. Such screening and risk benefit analysis is not typically found in the American context, where instead abortion is generally provided simply on request. As this process of screening by two physicians in Britain may better serve to identify and protect women who are being pressured into unwanted abortions, and would therefore reduce the risk of severe negative psychological reactions among this group of women for whom an unwanted abortion, it is highly likely that British women may be exposed to less psychological trauma associated with unwanted, unsafe, or unnecessary abortions as compared to American women. The potential protective effects of such screening are indicated by research among women who had abortions in the United States in which it was found that 64% reported feeling pressured into the abortion by other people (Rue). In addition to reducing the risk of women being pressured into unwanted abortions by third parties, two physician screening in the UK may also reduce the risk that women will have abortions in violation of their moral views, or their maternal desires, which are two of many statistically validated risk factors for subsequent psychiatric disorders.
2. No standardized measures for mental health diagnoses were employed.
3. Only the first reported episode of illness was recorded. Though the authors had the data to report on average number of contacts for each illness (a proxy for the duration and degree of the psychological episodes), they did not disclose any measure for duration or severity. The only exception is that they did report psychotic episodes within the first 12 months after delivery or termination...but did not identify prior history of abortion in thise cases. Given the eight year span of the study, the lack of information about when treatments occurred relative to the pregnancy outcome may also have a diluting effect in regard to recency to the stressor.
4. The failure to report timing of the first incident of psychiatric illness is underscored by the admission in the discussion that there were indeed "Difference in the timing of admission and the past psychiatric history of women admitted postpartum or post-termination...suggest different underlying mechanisms." If there are indeed "different mechanisms" underlying the difference in timing of psychological illness following pregnancies carried to term versus those aborted, isn't that exactly what should be studied. Instead, they note a difference in timing but don't provide the details. Since proximity to the event supports a casual connection, this is a very serious omission.
5. The study spanned, potentially from 1979 thru 1987, with women being introduced into the data set throughout that period. The authors received information about deaths, but they chose not to report deaths . . . which is especially concerning given the elevated rates of suicide attempts and completed suicides among women who abort.
6. The study groups are not clearly delineated. Women with a prior history of abortion were mixed into each group. The comparison of women who did not have abortions during the study period, therefore, actually included women with a history of abortion. This is especially important since there is strong evidence that women with a history of abortion have more mental health problems and substance use during and after subsequent pregnancies. It is also unclear what adjustment, if any, was made if women carried to term but subsequently had an abortion.
7. By the end of the study, the attrition rate was 65.6% for those had abortions and 57.5% for those who did not (p. 247). Such attrition rates are high and problematic. The fact that they were higher for women who had abortions, which may indicate greater psychological distress, is especially problematic. Those women who are having mental health problems that are trauma-related are precisely the most likely to be in the drop-out pool as they do not wish to go back to a doctor who might bring the incident back to mind. The authors report that "Most loss to follow-up occurred because patients left the practice of the recruiting doctor. Women no longer under observation were slightly younger, of lower parity and higher educational status, and more likely to be single than the original cohort."
8. Evaluation of the psychological state of patients was reported by general practitioners, not psychiatrists. The report of the study itself states: “The major disadvantages of using general practitioners’ reports were the likelihood of under-recognition and an imprecise diagnosis of psychiatric disorder” (p. 247). The authors even remark that the GP's assessments of 'puerperal psychosis' were almost certainly inaccurate.
9. The GP's who participated may have also been the same doctors who recommended the abortion to their patients. This involvement may have biased these GP's toward underestimating the negative effects on their patients and overestimating the pre-existing psychological illnesses, which is typically the legal justification for recommending an abortion for social reasons.
10. The GPs who participated in this catchment study were volunteers and no attempt was made to control for selection bias. It is possible that many, most, or all volunteered to participate in the study because of a special interest in the issue, and/or because they regularly referred for or performed abortions. The study had no blind or double blind controls and all contributing volunteers were aware of the implications of every judgement they made in preparing their reports. This study therefore falls far short of the objective quality of the record based studies done in Canada, Finland, and the United States, all of which found significantly higher rates of mental health treatments or suicide following abortion. Notably, the authors acknowledge that the risk of errors in diagnostic assessments by recourse to a strong standard of treatment via analysis of "episodes of psychiatric illness leading to hospital admission." In this regard, however, record bases studies are clearly a superior methodology and have clearly shown significantly higher rates of psychiatric hospitalization following abortion compared to delivery and miscarriage.(Reardon, CMAJ)
11. Research has indicated that women who have negative abortion reactions are less likely to return to the physician who referred or performed the abortion. For example, a survey of 2,215 abortion patients in 12 abortion clinics in the US found that two out of three women do not return for follow-up appointments at the abortion clinic. (see 'From the Patient’s Perspective - Quality of Abortion Care', Picker Institute. (1999). Boston, MA.) Women embarrassed a past abortion may change providers to avoid facing the stress of seeing the doctor who approved the abortion. In addition, poor followup may result in underestimation of the problem of significant adjustment problems post-abortion. Data in Gilchrist confirms this finding in that by the end of the study, significantly fewer women who aborted. 34.4%, were still under the care of the physician reporting on them comared to 4.4$ of those who did not request an abortion.
12. Data regarding prior psychiatric history in this study was reported by a local GP whose may not have had the complete patients’ health records due to lack of comprehensive record linkage in the UK.
13. This study had insufficient power to detect significant differences between those women who requested a termination and changed their minds, and those who were refused abortion.
14. Only extreme outcomes were measured – drug overdoses rather than substance abuse in general; only diagnosed PTSD but not the more prevalent sub-clinical levels of PTSD or the common practice of PTSD going undiagnosed; psychotic episodes which are rare in the population under either condition.
15. There are thousands of case studies of adult women who attribute post-trauma symptoms to their first-trimester abortions, narratives of which are being included in court cases and otherwise publicized. The vast majority of these case studies would not fit into the criteria of extreme problems counted in the Gilchrist 1995 study. Case studies may be inadequate for establishing prevalence or for comparison to the aftermath of other options for dealing with an unplanned pregnancy, but can a statistical study that would exclude those case studies be adequate?
16. Women who have miscarriages are known to have higher rates of subsequent psychological distress compared to women who deliver health children. By including women who miscarry with women who carried to term, the study fails to provide a comparison between rates of psychological illness for women who carry to term--which is of course their intent. While miscarriage is an unavoidable risk, the choice women face is between trying to carry to term and having an induced abortion. Therefore, it seems that the comparison between psychological risks of abortion and carrying to term would be relevant to both women and physicians--excluding the risks of psychiatric distress that may follow a miscarriage. While all measures are relevant, the failure to distinguish between successful delivery and miscarriages in this study may have obscured a relative risk of abortion compared to delivery.

Notes that may require further investigation

1. The study indicates that some dropouts occurred due to death (p 244 col 1), but the authors fail to report the distribution or cause of deaths. Were there for example, an excess number of suicides or accidents among women who had abortions, as has been found in numerous other studies? If so, it appears from the methodology employed that cases of abortion associated suicide would not been included in any of the measure of psychiatric distress. In other words, women who experienced this most sever psychiatric distress would simply have been counted as having no ill effects and as having "dropped out" of the study.
2. Ronsmans C, et al. "Mortality in pregnant and nonpregnant women in England and Wales 1997–2002: are pregnant women healthier?" in Lewis G, editor. Why Mothers Die 2000-2002. The Sixth Report of the Confidential Enquiries into Maternal Death in the United Kingdom. London: RCOG Press;2004

Following the studies of Gissler and Reardon showing lower mortality rates associated with childbirth, the Ronsmans study in Britain confirmed that there is a lower risk of mortality during pregnancy and until one year after birth compared to women without a recent pregnancy. Specifically reporting that:

"All-cause mortality in women aged 15–44 years was 58.4 deaths per 100,000 women per year.... Surprisingly, however, mortality during pregnancy or within 1 year after birth was between four and five times lower than mortality in women without a recent pregnancy. The rate ratios comparing the pregnancy–42 day and the 43–365 postpartum periods with nonpregnant women were 0.21 and 0.22, respectively."

Surprisingly, however this government funded inquiry failed to report any data on mortality rates assocaited with abortion. Given the fact that the authors were aware of the findings of Gissler and Reardon, the failure to report an analysis of death rates assocaited with abortion appears to be a deliberate attempt to suppress findings which would confirm previous research.

While this study fails to report mortality rates relative to pregnancy outcomes, it does report the following citations:

"In the USA, women who had delivered a live or stillborn infant in the previous year were half as likely to die as women who had not recently delivered." citing Jocums SB, Berg CJ, Entman SS, Mitchell EF. Postdelivery mortality in Tennessee, 1989–1991. Obstet Gynecol 1998; 91: 766–70.

"In Canada, mortality rates during pregnancy or within 42 days of its termination and between 43 and 225 days postpartum were about half those of nonpregnant women."citing Turner LA, Kramer MS, Liu S. Cause-specific mortality during and after pregnancy and the definition of maternal death. Chronic Dis Can 2002; 23: 1–8.

"In Finland, the age-adjusted risk of a natural death within a year after birth or a miscarriage was half that of women without a pregnancy." citing Gissler M, Berg C, Bouvier-Colle MH, Buekens P. Pregnancy-associated mortality after birth, spontaneous abortion or induced abortion in Finland, 1987-2000. Am J Ob Gyn 2004; 190:422-427.

NOT MENTIONED was the following findings from the Gissler 2004 study:

The age-adjusted mortality rate for women during pregnancy and within one year of pregnancy termination was 36.7 deaths per 100,000 pregnancies, which was significantly lower than the mortality rate among nonpregnant women 57.0 per 100,000 person-years (RR=0.64, 95% CI 0.58-0.71). The mortality was lower after a birth (28.2/100,000) than after a spontaneous (51.9/100,000) or induced abortion (83.1/100,000). We observed a significant increase in the risk of death from cerebrovascular diseases after delivery among women aged 15-24 years (RR=4.08, 95% CI 1.58-10.55).

This three fold higher death rate following abortion is certainly noteworthy and deserving additional investigation. Therefore it is hard to avoid the conclusion that this failure to examine and report on abortion associated deaths in this official British study may reflect a bias in the British research community which may also be reflected in studies regarding the negative pscyhological effects associated with abortion.

Criticisms by Dr. Philip Ney

The study by Gilchrist et al. is based on the concept of an unplanned pregnancy, but the authors make little attempt to define what this is and how itjavascript:void(0) was determined. As every physician knows, people are ambivalent about the inception and conception of almost every pregnancy. There are very few people who actually put much effort into planning a pregnancy, and those are mostly people who use natural family planning methods. Most "plan" only by withdrawing contraception. A recent report of the Alan Guttmacher Institute states that "the proportion of women wanting to become pregnant is extremely low, less than 1 in 5 in industrialised countries." If contracepting or not contracepting means whether the pregnancy is planned or not, then there is no basis for making statements about psychiatric sequlae of any pregnancy outcome. Many people change their mind almost in the middle of intercourse about whether they want or plan to have a baby.

The review of the literature is very biased. There are many relevant studies not cited. Gilchrist et al. do not summarize the references of Doane & Quigley and David et al. correctly.

Since the authors were only using major psychiatric illness classifications, it appears that they did not expect to find or look for the constellation of symptoms and signs now known as the Post_Abortion Syndrome. Post_Abortion Syndrome is now reasonably well recognised and defined, but not included in ICD _ 8.

Although the authors state this study examined a variety of pregnancy outcomes, they did not compare a live birth to a miscarriage or to a stillbirth or to an abortion. They found that the rates of miscarriage were different in the different groups. Miscarriages in the non_abortion group would tend to increase the morbidity because miscarriages do result in higher rates of both physical and psychiatric morbidity. Miscarriages in the abortion group would tend to decrease the apparent morbidity because the effects of the miscarriages are less than the effects of the abortion.

This study relied on general practitioners' assessment of psychiatric morbidity and used the not too precise catagorizations of ICD 8. They diagnosed 225 puerperal psychosis; much higher than the estimated prevalence. The authors found that only 13 of these puerperal psychosis were admitted for treatment, yet almost every case of a puerperal psychosis should be admitted. It seems family physicians were wrong in their diagnosis of puerperal psychosis by a factor of 17. It is likely they were equally out on the other psychiatric diagnosis. The authors did admit that the estimation of puerperal psychosis was too high. The authors found that there is a significantly higher rate of deliberate self_harm (DSH) following an abortion. Eighty_nine (89) % of these were overdoses, which are not difficult to diagnose. If the family physicians were better able to diagnose psychiatric morbidity of other kinds, it is likely that they might have found higher rates in the TOP group.

The authors state that the general practitioners would not have a systematic bias in diagnosing. However, since these general practitioners were referring their patients for TOP, they are less likely to see any adverse effects of a procedure they recommended. Why did the authors not include family physicians who do not make abortion referrals? Physicians of the Christian Medical and Dental Society (CMDS) Canada have a significantly lower rate of abortions and miscarriages in their practices compared to other general practitioners.

The general practitioners' follow up in this study was poor. They lost 65.6% to follow up by the end of the study from the abortion group, and 57.6% from the non_abortion group. The authors state that most of those who were lost to follow up were single, highly educated women. Other studies have shown these women are more likely to have an abortion.

Since those in the refused abortion group were probably refused because of psychiatric problems, psychiatric morbidity in the TOP group should be lower. The authors state that although the DSH was higher in the TOP group, the rates fell more rapidly than in the non_abortion group. They failed to note that the rate the TOP group fell to, i.e. 3.8 was still higher than the baseline group of the non_TOP group, 3.0.

Gilchrist et al. did not show the demographic variables in each group, but state that the data "were indirectly standardised for age, marital status, smoking habit, age at leaving full_time education, gravidity, and previous history of induced abortion at recruitment, since the comparison groups differed on these characteristics." At the end of this article they also state that "the lack of more detailed social information was, however, an important limitation, given the evidence that poor social support increases the risk of psychological morbidity after abortion." They then, to try and explain why DSH is higher in the abortion group, state, "the most likely explanation is that they were at risk because of coexisting social or psychological difficulties associated with both their decision to seek a termination and their subsequent risk of deliberate self_harm." This confusing obfuscation seems to be an attempt to deny the findings that psychiatric morbidity, apart from DSH, was not higher in the group who were refused TOP. The authors state that "risk ratios (RR) were calculated with reference to the group of those who did not request a termination." "The 95% confidence intervals (CI) were calculated using the assumption that the standard deviation of the log of relative risk is equal to the sum of the reciprocals of the observed number of cases in the two groups being compared." This is a questionable assumption, especially in view of the fact that the crude rates for psychosis are; TOP group .1 per 1000, non_TOP group .05 per 1000.

The fact that the psychiatric morbidity of the termination group was not lower than a comparison group of women who requested abortion and changed their minds, effectively demonstrates that abortion is not an effective treatment for psychiatric illness. This study also demonstrates that abortion makes psychiatric conditions of all kinds worse. Yet, without scientific or clinical support, these general practitioners used "previous or anticipated psychiatric illness" as a justification for abortion. This is a practice that the Canadian Psychiatric Association has officially deplored.

Criticisms by Priscilla Coleman

"Incredulously, the Gilchrist et al. (1995) study received a rating of “Good”, when very few controls for confounding 3rd variables were employed, meaning the comparison groups may very well have differed systematically with regard to income, relationship quality including exposure to domestic violence, social support, and other potentially critical factors. Further Gilchrist et al. reported retaining only 34.4% of the termination group and only 43.4% of the group that did not request a termination at the end of the study. No standardized measures for mental health diagnoses were employed and evaluation of the psychological state of patients was reported by general practitioners, not psychiatrists. The GPs were volunteers and no attempt was made to control for selection bias."