Narrative and systematic reviews of abortion and mental health

The American Psychological Association’s narrative review set out to ascertain whether there is an increased risk of mental health problems following an abortion compared with birth, and the factors that may be related to such an increased risk ^{Reference Major, Appelbaum, Beckman, Dutton, Russo and West8} (updated by Major et al ^{Reference Major, Appelbaum, Beckman, Dutton, Russo and West10} ). Reviewing 56 studies comparing abortion and other pregnancy outcomes, and 23 with no comparator, they concluded that the relative risk of developing mental health problems following a single, first-trimester abortion of an unplanned pregnancy was no greater than the risk following giving birth.

Charles and colleagues’ ^{Reference Charles, Polis, Sridhara and Blum9} systematic review of more than 700 studies, including only 21 for analysis, compared the longer-term mental health outcomes of abortion with those of birth. The authors developed five quality criteria to rank the studies, such as appropriateness of the comparison group and whether they controlled for pre-abortion mental health status. Using these criteria, four studies were judged to be ‘very good’, eight ‘fair’, eight ‘poor’ and one ‘very poor’. They concluded that restricting the analysis to studies of very good quality resulted in little or no difference in mental health outcome following abortion compared with birth, and that, as the quality declined, there was an increased possibility that studies would conclude that abortion was more likely than birth to end in poor mental health outcomes.

Meta-analyses of abortion and mental health

The Coleman ^{Reference Coleman2} meta-analysis cannot be regarded as a formal systematic review because search strategies and exclusion criteria were not published. Nevertheless, the 22 pooled studies had to: assess the impact of abortion against a no-abortion group; have a sample size of at least 100; use odds ratios; and have been published in English-language peer-reviewed journals between 1995 and 2009. Although studies were required to control for third variables, they were not required to control for mental health problems before the abortion. In addition, the no-abortion groups included ‘no abortion’, ‘pregnancy delivered’ or ‘unintended pregnancy delivered’. Coleman concluded that there was ‘a moderate to highly increased risk of mental health problems after abortion over the risks of birth’. The volume of highly critical letters that followed publication is a testament, not just to the review’s methodological flaws, but also to the personal, ethical, religious and political importance of abortion to different groups.

Fergusson and colleagues ^{Reference Fergusson, Horwood and Boden7} meta-analysed eight studies identified in the three reviews discussed above, comparing abortion versus unwanted/unintended pregnancy, and suggest that, following an unwanted/unintended pregnancy, abortion is more likely than birth to be associated with anxiety, self-harm and substance misuse (but not depression). Although the main analysis did not control for mental health before abortion, a secondary analysis of four studies, which did control for mental health, found that an increased risk of general psychiatric problems remained.

Last year, we completed a systematic review, narrative synthesis and limited meta-analysis of the mental health outcomes following abortion and birth of an unwanted pregnancy. ¹¹ We concluded that the majority of studies in the review were subject to multiple limitations, such as heterogeneity of mental health outcomes and their methods of assessment, inadequate control of confounding variables, and comparator populations that rarely included unwanted pregnancies continuing to birth. For these reasons, our meta-analysis only included four studies, all of which controlled for prior mental health problems and used unwanted/unintended pregnancies for analysis. We found that there were small effects associated with abortion, increasing the risks of self-harm and decreasing risks for psychosis. However, these four studies were conducted in different countries: two ^{Reference Steinberg and Russo12,Reference Cougle, Reardon and Coleman13} in the USA, where abortion is available on-demand, one ^{Reference Gilchrist, Hannaford, Frank and Kay14} in the UK and one ^{Reference Fergusson, Horwood and Boden15} in New Zealand. In the UK and New Zealand the legal grounds for abortion are to mitigate any physical or mental harm should the pregnancy be delivered. Finding that women in these studies had a higher rate of mental health problems after abortion risks the criticism that it would be like finding out that people who took hangover remedies had an increased risk of headache. ^{Reference Dwyer and Jackson16}

From abortion to unwanted pregnancy: making sense of the data

A recent prospective, population-based cohort study ^{Reference Munk-Olsen, Laursen, Pedersen, Lidegaard and Mortensen17} used three Danish national registers to identify, and follow, females who had no prior history of mental health problems, and who went on either to have a first abortion or to give birth. The authors estimated the rates of psychiatric contact in the 9 months preceding abortion or birth, and in the subsequent 12 months. They showed that females who had an abortion had significantly higher rates of mental health problems in the 9 months before abortion than those in the 9 months before birth; and that for those who had an abortion, their rates of psychiatric contact after the abortion were no greater than before the abortion. In contrast, females who gave birth had significantly higher rates of mental health problems after birth than before. It is a distinct possibility that mental health problems precede unwanted pregnancy and/or unwanted pregnancy may lead to mental health problems.

In our view, all three meta-analyses ^{Reference Coleman2,Reference Major, Appelbaum, Beckman, Dutton, Russo and West10,11} added little to a well-done narrative review, and risked giving the impression that the results were more scientific and reliable. GRADE evaluation of our meta-analysis suggested that the outcomes were of very low quality and therefore at significant risk of not being correct; a critique likely to be true of Fergusson and colleagues’ meta-analysis. We concluded that the American Psychological Association’s ^{Reference Fergusson, Horwood and Boden7} and Charles et al’s ^{Reference Major, Appelbaum, Beckman, Dutton, Russo and West8} reviews were probably accurate: for a woman carrying an unwanted pregnancy, current evidence suggests that her mental health is probably largely unaffected whether she chooses to have an abortion or to continue to birth. In Denmark, where a woman has the right to choose, and in the UK where the right to choose is curtailed by the requirement for professional approval, two high-quality studies, some 16 years apart, appear to support the general conclusion of these and our review: that there is no or little increased risk of mental health problems following abortion. ^{Reference Gilchrist, Hannaford, Frank and Kay14,Reference Munk-Olsen, Laursen, Pedersen, Lidegaard and Mortensen17}

Unwanted pregnancy is both a potential personal catastrophe and a major public health problem, leading to over 46 million abortions each year worldwide, of which over 19 million are illegal abortions. ¹⁸ In the UK before the Abortion Act 1967 there were an estimated 100 000 illegal abortions each year. ¹⁹ The vast majority of unsafe/illegal abortions now occur in low- and middle-income countries, with a mortality rate between 100- and 1000-fold greater than legal abortions; however, the mortality rate for illegal abortions in the USA is still 50 times greater than that for legal abortions. ^{Reference Sedgh, Henshaw, Singh, Ahman and Shah1} Making abortion legal has helped to substantially reduce the physical health risks for women with an unwanted pregnancy; perhaps moving our attention from what all too often appears to be morally and politically influenced concerns about abortion, we could focus on the mental health needs associated with unwanted pregnancy both in clinical practice and research.