Questionable Research Conclusions/Reply

Dear Editor:

I was disappointed to see an article 1 in the April 2010 issue of The Canadian Journal of Psychiatry that I believe is not well founded.

It is hard to see how the associations drawn by the authors could follow from the data that they use. In that case, it is unfortunate that the authors have approached a newspaper to distribute their conclusions, suggesting abortions cause mental problems.

The main problems with the paper are as follows: the research is a re-analysis of data from the National Institute of Mental Health and the National Institute of Drug Abuse in the United States. The authors have no control or knowledge of the quality of the data. In addition, the organizations in the United States were following their own special interests.

Second, the American women in the study sample were chosen ". . . including all respondents who met criteria for a lifetime mental disorder . . .'"�?3 Hence this was very far from a representative sample of women, and many of the subjects were already predisposed to mental problems. The authors presented their findings as being generally applicable to all women.

Third, the study was retrospective, therefore the mental state of each woman was assessed by interviewers at the time of the interview but not before or soon after her abortion. This means the interviewers presumably relied on each woman's memory and self-assessment of her mental state of an abortion that may have occurred years or decades previously. Such data are likely to be extremely inaccurate.

Fourth, to have credibility, this study should have compared the abortion group with a control group of women in similar circumstances. A possible control group could be women who had an unintended pregnancy but did not have an abortion. However, the authors appear to compare the abortion group with all other women in the survey, whether or not they went through any pregnancy, unwanted or otherwise.

Fifth, the authors completely neglect all other factors, not connected with the abortion itself, that might result in a poor mental state of a woman who has an abortion (probably the dataset did not contain such information). Such factors include that before the abortion she: was pressured into it by a partner, had been raped, had other illnesses or disabilities, had very difficult economic circumstances (perhaps a single mother), was a sex trade worker, and so on.

Finally, the authors apply statistical analysis to the data that makes it look very scholarly or scientific, but the lack of high-quality relevant data surely ensures that the analysis is of little value.

References

1. Mota NP, Burnett M, Sareen J. Associations between abortion, mental disorders, and suicidal behaviour in a nationally representative sample. Can J Psychiatry. 2010;55(4):239-247.

Peter Aitchison, BSc, MA, PhD

Winnipeg, Manitoba

Reply

Re: Questionable Research Conclusions

Dear Editor:

We would like to thank Dr Aitchison for his comments regarding our recent study. However, a few clarifications need to be made regarding certain statements made in his letter.

The author suggests that we approached a newspaper for the purposes of claiming that "abortions cause mental problems." First, it should be noted that we did not approach any media. The Canadian Journal of Psychiatry selects a few studies in each issue to include in a media advisory, and our study was one paper highlighted in the April 2010 issue. Further, in talking to the media, we took great care to emphasize that only associations, and not causal links, had been identified between abortion and mental disorders in our study.1

Dr Aitchison states correctly that the sample of women we used included all respondents from Part I of the National Comorbidity Survey Replication (NCS-R) who had been assessed as having a lifetime mental disorder. However, he inaccurately draws the conclusion that the sample was not representative of women in the US population. Part II of the NCS-R also included a probability subsample of the remaining respondents and specific weights were applied to the entire sample to render it representative of individuals in the United States.2

Dr Aitchison mentions that the "authors completely neglect all other factors, not connected with the abortion itself, that might result in a poor mental state of a woman who has an abortion." We completely agree that the NCS-R did not assess for many of the factors mentioned by Dr Aitchison, and that many of these factors, as well as others, can contribute to the relation between mental illness and abortion. However, we feel that we appropriately adjusted for several of the potential confounding variables highlighted by Dr Aitchison in our statistical models, including rape, household income, and marital status.

Please see our response to another "Letter to the Editor"3 for further details.

Natalie P Mota, MA

Margaret Burnett, MD, FRCPC

Jitender Sareen, MD, FRCPC

Winnipeg, Manitoba

References

1. Cosh C. Science-reporting smell test of the week [Internet]. Toronto (ON): Maclean's; May 4, 2010 [cited 2010 May 21]. Available from: http://www2.macleans.ca/tag/natalie-mota/.

2. Kessler RC, Berglund P, Chi� WT, et al. The US National Comorbidity Survey Replication (NCS-R): design and field procedures. Int J Methods Psychiatr Res. 2004;13:69-92.

3. Mota NP, Burnett M, Sareen J. Re: yet another flawed study. Can J Psychiatry. 2010;56(1):72.