The authors concluded that the highest quality evidence indicated that there was no difference in the mental health of women who had abortions and women who delivered unintended pregnancies. Given the potential for error and bias in the review process, the absence of statistical data and the clinical heterogeneity between studies, the reliability of the authors' conclusions is unclear.

To determine the long-term effects of abortion on women's mental health.

PubMed, EMBASE, Scopus and PsycINFO were searched from January 1989 to August 2007 for articles published in English. Search terms were reported. Reference lists of relevant review articles were searched.

Studies of women who underwent abortion were eligible for inclusion if they included at least 100 participants and had a follow-up period of at least 90 days. Studies were excluded if the abortion was illegal or due to fetal abnormality, if the study focused on sexual functioning after abortion, compared different methods of abortion, compared adult and adolescent processing of abortion or investigated cultural differences in reaction to abortion.

In the included studies abortion was established by self report, record linkage or by recruiting participants at clinics. Most participants had only one abortion. A variety of comparison groups were included in the studies, including both intended and unintended pregnancies, miscarriages and women who were not pregnant. The age of participants ranged from 12 to 54. Some studies were of adolescents only. Various outcomes were measured, most commonly depression, anxiety, self esteem, guilt and suicide; both standardised and non-standardised measures were used. The maximum follow-up time was 25 years. Included studies were conducted in seven different countries.

Two reviewers independently appraised and rated each study. In case of disagreements, a third reviewer appraised the article and the senior author was consulted.

The methodological quality of the included studies was evaluated for use of appropriate comparison group, use of valid mental health measures, control for pre-existing mental health status, control of confounding variables and comprehensive exploration of the research question. Studies were then graded as excellent, very good, fair, poor or very poor. The authors did not state how many reviewers performed the validity assessment.

Studies were categorised as neutral (no significant differences between groups), negative (the abortion group fared worse than the comparison group) or mixed (included both types of findings). The authors did not state how the data was extracted.

The studies were combined in a narrative synthesis.

Twenty one studies (22 comparisons) were included for the review (n=289,976): seven prospective cohort studies (n=15,272; three studies used the same participants); 11 retrospective cohort studies (n=266,554); and three cross-sectional studies (n=8,150). Sample sizes ranged from 120 to 133,950 patients. Four studies were classified as very good, eight as fair, eight poor and one very poor. No statistical data or p values were reported for any of the studies.

Very good studies (four studies, n=23,285): A cross-sectional study (n=3,482) controlled for confounding variables found no difference in anxiety symptoms between women who underwent abortion and those who delivered unintended pregnancies. A prospective cohort study found no difference in the rates of psychiatric disorder between women who had abortions and those who delivered unintended pregnancies (n=13,261). A retrospective study that compared women who had an abortion compared to those who delivered unwanted pregnancies found no difference in self esteem between the groups (n=5,295). A retrospective study found no difference in depression between women who underwent abortion and those who delivered an unintended pregnancy (n=1,247).

Fair Studies (eight studies, n=14,255): Two studies reported worse outcomes for women who underwent abortion in terms of concurrent or subsequent mental health problems (n=520) and depression and anxiety (n=2,846) compared to women who gave birth. Three studies reported no differences between women who had an abortion and women who miscarried or gave birth (n=9,813). Three studies reported mixed outcomes: higher levels of depression among married women who had an abortion compared to married women who gave birth (one study, n=1,076); and higher levels of relief and avoidance and lower levels of grief and loss in women who underwent abortion compared to women who miscarried (two studies, n=120).

Four poor or very poor studies reported negative findings in women who had undergone abortion, four reported mixed findings and one reported no difference between women who had undergone abortion compared to controls.

The highest quality evidence indicated that there was no difference in the mental health outcomes of women who had abortions compared with control groups or respective comparison groups.

The review question was clearly stated, however, the length of follow up was not reported for all studies, which made it unclear to what extent the findings of the review addressed the long-term outcomes of abortion. Inclusion criteria for participants, intervention and outcomes were implicit in the research question, but were not explicitly stated. Inclusion criteria for study design were not stated. Several relevant databases were searched. There appeared to be no steps taken to minimise publication and language biases. The authors reported that they appraised and rated each study independently, but it was unclear whether this refered to study selection, validity assessment or data extraction, or a combination of these. Therefore, the possibility of reviewer error and bias could not be ruled out. The methodological quality of the included studies was assessed and used to inform the results. There was a wide range of methodological heterogeneity between the included studies in the outcomes measured and the tools used, which made comparisons between studies difficult. Given the clinical heterogeneity between the included studies, the decision to combine studies into a narrative synthesis was appropriate. But, the absence of statistical data made it difficult for the reader to determine the significance of the findings for themself. Given the potential for error and bias in the review process and the absence of statistical data, the reliability of the authors' conclusions is unclear.

Practice: The authors stated that the evidence base did not support the enforcement of informed consent laws that inform women of the mental health sequelae of abortion.

Research: The authors stated that any future research conducted on the mental health outcomes of abortion should be longitudinal, prospective, have an appropriate comparison group, use validated mental health measures, control for pre-existing mental health and other confounding variables and comprehensively explore the research question.

Frary Family Fund.

This is a critical abstract of a systematic review that meets the criteria for inclusion on DARE. Each critical abstract contains a brief summary of the review methods, results and conclusions followed by a detailed critical assessment on the reliability of the review and the conclusions drawn.