House of Commons - Science and Technology

Select Committee on Science and Technology Twelfth Report

5 Impact of abortion on women's health

125. The debate on health risks associated with abortion is fierce. The most comprehensive and rigorous review of the evidence on health risks for women has been produced by the RCOG.[161] The results of this review were condensed to form guidelines that are followed by nurses and physicians in obtaining consent. The guidelines that are relevant to the impact of abortion on women's health (chapter 5 of the RCOG report) are discussed below and reproduced in full in the Appendix. The work underpinning these guidelines was done in the way identified as good practice in medical and scientific circles, with an expert group, and external reference group and peer review. In addition, the evidence base underpinning recommendations is fully referenced and graded according to its strength.

126. Before we approach the evidence, it is worth noting two issues relating to how the conclusions are drawn. First, a correlation is not the same as a causal link. Abortion is a difficult subject to study since there are so many factors that influence the outcome a pregnancy and/or abortion and it is exceedingly difficult to control for all these variables. This is an issue that permeates the entire body of evidence relating to health risks and abortion.

127. The second is a problem regarding which comparison groups are most appropriate. Dr James Trussel, Professor Katherine Guthrie and Dr Sam Rowlands outlined a range of design studies, which we have paraphrased here:[162]

Ideal design—Women with unwanted pregnancies would be randomly assigned to receive an abortion or to have their request denied without the possibility of their having a termination elsewhere. Of course, research ethics prohibit this type of study from being carried out.
Next best—Women with unwanted pregnancies who have abortions are compared with women who have unwanted pregnancies but whose request for an abortion is denied.
Farthest from ideal—Women with unwanted pregnancies who have abortions are compared with women who got pregnant because they wanted to become mothers and went on to have a child. These are not comparable groups.
Next farthest from ideal—Women with unwanted pregnancies who have abortions are compared with all women giving birth, some of whose births would be unwanted.

128. To which we would add:

Next farthest from ideal 2—Women with unwanted pregnancies who have abortions are compared with women who have not had a pregnancy.
Next farthest from ideal 3—Women with unwanted pregnancies who have abortions are compared with women whose pregnancies miscarried naturally; some of these pregnancies would have been unwanted.

129. A range of comparison groups were used in the studies brought to our attention and we will draw attention to some of these in the following discussion.

Mental health risks

130. In relation to mental health risks associated with abortion, the RCOG guidelines summary states that "some studies suggest that rates of psychiatric illness or self-harm are higher among women who have had an abortion compared with women who give birth and to nonpregnant women or similar age. It must be borne in mind that these findings do not imply a causal association and may reflect continuation of pre-existing conditions."[163] They assign it an evidence-base of class B strength.[164]

131. There are many papers on the mental health risks of abortion. Some conclude that there is a health risk; others conclude that there is not. Professor Patricia Casey points out that opponents pick on different flaws in the research.[165] She says that:

Criticisms of papers that argue that there are mental health risks associated with abortion include:

failure to control for confounders such as previous psychiatric history;
using data obtained from women seeking psychological treatment post-abortion;
comparing inappropriate groups, for example, women who have had abortions and women who have given birth; and
using limited outcome measures, for example, psychiatric hospitalisation or receiving out-patient treatment.

Criticisms of papers that argue that abortion does not increase mental health problems include:

the absence of long-term data spanning years/decades;
high attrition rates in follow-up studies reducing the potential for identifying psychological problems; and
small sample sizes.

132. In submissions to this inquiry, the most commonly quoted paper on mental health following abortion, and one of the most robust, is Fergusson et al 2007.[166] It found that those who had an abortion had elevated rates of depression, anxiety, suicidal behaviours and substance misuse disorders. The authors note a number of strengths and weaknesses of the study.

Strengths: (a) the use of longitudinal design; (b) assessment of mental disorders using standardised diagnostic criteria; (c) the availability of a range of concurrent and prospectively assessed covariate factors; and (d) adjusted contrasts between those having abortion and equivalent groups of those becoming pregnant and those not pregnant.
Weaknesses: (a) omitted covariates, although the study did include an impressive list of covariates; (b) comparison of the rates of abortion reported by the study's cohort differ from the rates in official record data, suggesting an underreporting of abortion rates; and (c) the lack of information on contextual factors associated with the decision to seek an abortion, e.g., the results may reflect the effects of unwanted pregnancy, rather than abortion, on mental health. (See paragraph 134 below.)

133. It is noteworthy that references to the Fergusson et al 2007 study from some pro-life groups make no mention of the weaknesses (for example, the ProLife Alliance[167] or CARE[168]) and those from some pro-choice groups make no mention of the strengths (for example, fpa[169]). The BMA make balanced reference to it.[170]

134. Our concern with this study, and others like it, is that it compares women who have had an abortion with women who have had children or who did not become pregnant. None of these groups are comparable. Furthermore, the extent to which the pregnancies were wanted was not controlled for. The Committee has seen a communication from Dr Fergusson—with his permission—where he expresses regret that the study has been "talked-up" by those who argue that it proves a causative link between induced abortion and subsequent psychiatric morbidity, pointing to another study he has carried out suggesting a causative link between abortion and better outcomes in young women compared to match controls who continued the pregnancy.[171]

135. The most frequently cited paper with the comparison groups we have identified as preferable is Gilchrist et al 1995.[172] Its strength is its range of comparison groups: women who had an abortion, women who did not request an abortion but whose pregnancy was unplanned, women who were refused an abortion, and women who changed their minds before the abortion was performed. It calculated risk ratios with reference to the group of those who did not request an abortion and reported that rates of psychiatric disorder were no higher after an abortion than after child birth; that women with a previous history of psychiatric illness were most at risk of disorder after the end of their pregnancy, whatever its outcome.

136. Gilchrist et al 1995 does, however, have weaknesses. For example, the authors accept that the 'women who were refused an abortion' group was much smaller than the other groups, significantly reducing the power to detect important effects.[173] Also, there was uncertainty about whether this group did in fact include women who, although they had been refused an abortion in one place, had successful sought an abortion elsewhere.[174] The authors became concerned when they noticed that the reported rate of miscarriage was much higher among those women who had been refused an abortion. To be safe, the data were reanalysed without the women who had reported miscarriages and this "did not materially alter our findings".[175]

137. We received little information on other studies that compared women who had abortions with those who were refused abortions. Dr Sam Rowlands introduced a study that examined the effect on children who are the result of pregnancies where an abortion was refused.[176] These children were breast-fed less, suffered more childhood illness, displayed behavioural problems and achieved poorer school performance, and when they reached their 20s they had more social problems, lower job dissatisfaction, fewer friends, more criminality and more drug and alcohol problems. These results are consistent with the conclusions of Joyce et al 2000[177], who conclude that "unwanted pregnancy is associated with prenatal and postpartum maternal behaviors that adversely affect infant child health, but that unwanted pregnancy has little association with birth weight and child cognitive outcomes". However, they gave a word of warning that "Estimates of the association between unwanted pregnancy and maternal behaviors were greatly reduced after controls for unmeasured family background were included in the model".

138. Several submissions noted research showing that all causes death rates are higher amongst women who have had abortion compared with those who had given birth.[178] Dr Chris Richards and Dr Mark Houghton, for example, refer to a Finnish study that reports age-adjusted odds ratios[179] of 1.63 for deaths from natural causes, 4.24 for deaths from accidents, 6.46 for deaths from suicide, and 13.97 for deaths from homicide. However, even though this study and others like it are controlled for demographic characteristics, socioeconomic status, health status and medical disorders, the comparison groups are inappropriate for answering a question about the causal link between abortion and all-cause morbidity.

139. In view of the controversy in this area, we recommend that the Royal College of Psychiatrists update their 1994 report on this issue.

140. We conclude that there is no strong evidence which contradicts the wording of the current RCOG guidelines on the risk to mental health of induced abortion.

Physical health risks

FUTURE REPRODUCTIVE OUTCOMES

141. In relation to future reproductive outcomes, the RCOG guidelines state that "there are no proven associations between induced abortion and subsequent ectopic pregnancy, placenta praevia or infertility. Abortion may be associated with a small increase in the risk of subsequent miscarriage or preterm delivery."[180] They assign it an evidence-base strength of class B.

142. The evidence we received indicates that abortions carry a small increased risk of subsequent premature births[181] and may carry a small increased risk of miscarriage[182]. A review by Thope et al (2002) concludes that:

The population-based studies we reviewed suggest that induced abortion increases the risk of preterm birth in subsequent pregnancies. Moreover, these reports suggest that a dose response effect is present with increasing numbers of abortions associated with increasing risk.[183]

143. Dr Sam Rowlands points out that some studies, including a large well-designed 2006 study, show no links.[184] Dr Rowlands' view is that the evidence on pre-term delivery is still contradictory but commends the precautionary approach taken by the RCOG guidelines.[185]

144. Professor Jane Norman, from RCOG, however, explained that the risk of subsequent preterm deliveries could be reduced:

We know that if women have abortions earlier they are less likely to have cervical damage which may lead to preterm birth. If they have their abortions done by people who are expert, again that reduces that risk.[186]

145. The evidence on miscarriage is less strong. The Thorpe et al (2002) review found no relationship between induced abortion and miscarriages in subsequent pregnancies, but Dr Sam Rowlands points out that the literature on this issue is conflicting: "Two cohort and three case-control studies published prior to 1999 found no association. However two more recent studies have shown a positive association between abortion and subsequent miscarriage."[187] The reason that one might expect there to be a link is that "It is recognised that during abortion the cervix (neck of the womb) may be damaged. It has been hypothesised that such injury could make it less competent in subsequent pregnancies and so less able to 'hold a pregnancy in'."[188]

146. The evidence is not strong when it comes to establishing a link between abortion and:

ectopic pregnancy (when the embryo implants outside the womb, for example, in the fallopian tube);
placenta praevia (when the placenta is low in the uterus and covers all or part of the cervix); and
infertility.

147. For example, of the risk of infertility, the SPUC comments that:

Infection can result from abortion, leading to an increased risk of infertility. This risk is particularly relevant where there is a pre-existing genital infection. This is often dismissed as being unrelated to the abortion procedure, but clearly the procedure can facilitate the spread of infection in the reproductive system.[189]

148. Dr Sam Rowlands argues that measures to prevent infection are now routine practice and that "there is no proven increased risk of subsequent infertility when an abortion is carried out in proper, safe, medical conditions and is not complicated by pelvic inflammatory disease".[190]

149. Similarly, on placenta praevia, the SPUC argues that previous abortion is a risk factor, although not when the method used is vacuum aspiration.[191] However, Dr Sam Rowlands argues that the best study of its kind[192] does not show a link and that earlier, less well designed studies, show variable results.[193]

150. We found no strong evidence of links between abortion and negative future reproductive outcomes with the exception of a possible link with future pre-term deliveries and miscarriages. We conclude that there is no strong evidence which contradicts the wording of the current RCOG guideline on the risk to future reproductive health of induced abortion.

BREAST CANCER

151. In relation to breast cancer, the RCOG guidelines state that "induced abortion is not associated with an increase in breast cancer risk."[194] They assign it an evidence base strength of class B.

152. Dr Joel Brind submitted evidence to this inquiry that claims there is a link between breast cancer and abortion.[195] He is critical of the Lancet-published 2004 meta-analysis by Valerie Beral and colleagues from Oxford University—which found no link between breast cancer and abortion—because it omitted some studies which he considered valid and included others that he considered invalid. Dr Sam Rowlands made a similar accusation of Dr Brind's submission, however, pointing out that several key papers were missing.[196]

153. Dr Richards told us that "if you compare women who keep their pregnancy with those who have an induced abortion, those who have an induced abortion are more likely to get breast cancer later on".[197] This is the comparative group that Dr Brind favours and the result is expected, since carrying a first pregnancy to birth is protective against breast cancer.[198] However, if you look at the rates of cancer between women who have had an abortion and those who have not had children, the effect disappears. Dr Rowlands comments that:

at least nine prospective cohort studies which are more likely to give reliable results: these show no association or a negative association. Recall bias does not occur in record-linkage studies in which study subject data are present in databases; there are now seven such studies published, all of which show no association. Two recent cohort studies of high quality also show no association.[199]

154. We found no evidence which contradicts the wording of the RCOG guidelines on the risk of breast cancer.

POST ABORTION INFECTION

155. In relation to post abortion infection, the RCOG guidelines state that "genital tract infection, including pelvic inflammatory disease of varying degrees of severity, occurs in up to 10% of cases. The risk is reduced when prophylactic antibiotics are given or when lower genital tract infection has been excluded by bacteriological screening."[200] They assigned an evidence base strength of class B.

156. We did not receive any evidence which undermined the RCOG guidelines on post abortion infection.

RESTRICTION OF ACCESS TO ABORTION

157. We did not receive a great deal of information on the effect of the restriction of access to abortion. However, in countries where abortion is illegal, the health impact on women is well documented: for example, the WHO estimates that 68,000 women worldwide die each year due to complications of unsafe abortion; and Nepal recently liberalised its abortion laws, cutting maternal mortality rate from 539 maternal deaths per 100,000 live births in 2001 to 281 per 100,000 live births in 2006.[201] The fpa informed us that:

In Romania, policies restricting access to abortion led to a significant increase in maternal mortality from 20 maternal deaths per 100,000 live births in 1966 to over 100 per 100,000 in 1974 and to 150 per 100,000 in 1983. After the restrictive laws were revoked, the rate of maternal deaths fell rapidly to 40 per 100,000 live births in 1989. It is estimated that around 200,000 Romanian women died between 1966 and 1988 as a result of unsafe abortion.[202]

Informed consent

158. We consider the health risks of abortion to be relevant to the abortion issue mainly as they relate to obtaining informed consent, especially as this would be the basis for proceeding with many abortions if the requirement for two doctors' signatures is lifted. It is important that the guidelines for nurses and physicians gaining consent are accurate and up to date so that the risks are accurately communicated to the patient. The RCOG has said that it will "maintain a watching brief on the need to review recommendations in the light of new research evidence".[203] We note that the last review took place in 2004, and the one before that in 2000. Both of these were funded by the Department of Health.

159. Abortion and possible related health risks is a matter of public health. The Department of Health should take responsibility for a rigorous assessment of the evidence and/or request updated consensus statements from the appropriate professional bodies on the level of risk or absence of risk for the conditions discussed above.

160. We therefore recommend to the Government and the National Institute for Health and Clinical Excellence (NICE) that the clinical guidelines on abortion provision, including health risks associated with abortion, should ultimately be taken over by NICE.

161. We further recommend that the Government fund the RCOG or NICE to review the RCOG guidelines.

162. We believe that most, if not all, providers of abortion services currently make available the content of the latest guidelines prior to obtaining consent.[204]

163. While we recognise that obtaining informed consent is a process that is not always best carried out through leaflets and checklists alone, we recommend that abortion providers are required to ensure this information is given to patients as part of the process of informed consent.

164. To ensure that no patients are misled, we further recommend that the Government consider ways of ensuring that all those claiming to offer pregnancy counselling services make the guidelines available or indicate clearly in their advertising that they do not support referral for abortion.

165. We recommend that Members of Parliament, when considering the issue of health risks in the context of clinical guidance and informed consent, consider also our report and conclusions.

Conclusion

166. This has been a difficult inquiry to conduct in the light of the extremely controversial and sensitive subject matter. Parliament will ultimately take the decision on the reform of abortion law. We have tried to examine the scientific issues as objectively as possible and we offer our conclusions to the House to aid the debate.

161 The Care of Women Requesting Induced Abortion: Evidence-based Clinical Guideline Number 7, RCOG, September 2004 Back

162 SDA 52 Back

163 The Care of Women Requesting Induced Abortion: Evidence-based Clinical Guideline Number 7, RCOG, September 2004 p 35 Back

164 See Glossary for definition Back

165 SDA 31. Please see also SDA 26A, SDA 52, SDA 2A Back

166 Fergusson M, JM Boden & LJ Horwood, "Abortion among young women and subsequent life outcomes", Journal of Child Psychology and Psychiatry, vol 47 (2007) pp 16-24 Back

167 SDA 07 para 7.1 Back

168 SDA 34 section 3(i) Back

169 SDA 10 para 3.10 Back

170 SDA 13 para 17 Back

171 SDA 2A; Ferguson M, J M Boden & L J Horwood, "Abortion among young women and subsequent life outcomes", Perspectives on Sexual and Reproductive Health vol 39 no 1 (2007) pp 6-12 Back

172 Gilchrist AC, PC Hannaford, P Frank & CR Kay, "Termination of Pregnancy and Psychiatric Morbidity", British Journal of Psychiatry, vol 167 (1995) pp 243-248 Back