Thursday, March 08, 2012

Today's "Women's Health" Study

This is an interesting story about a piece of research looking into the relationship between abortion and mental illness:
A leading psychiatry journal has distanced itself from a controversial study that it published in 2009 which suggested a link between abortion and mental illness, including such severe forms as post-traumatic stress disorder, panic attacks, and drug addiction.
In an unusual commentary, one of the Journal of Psychiatric Research's editors-in-chief and a co-author warned that the 2009 paper, which has been widely cited by legislators and advocates to argue that abortion raises a woman's risk of mental illness and to push for laws requiring providers to tell women that, in fact "does not support assertions that abortions led to psychopathology."
The reason for this unusual step has to do with a fairly major methodological problem in the original study:
Steinberg said that the biggest problem in the original Coleman study was that "many of the incidents of mental illness she included came before the abortion." That cast doubt on whether abortion triggered mental illness. Instead, women with mental illness might have been more likely to have an unwanted pregnancy and terminate it.
In other words, if the observed correlation is about causation, the most likely direction of the cause-and-effect chain is from mental health problems to unwanted pregnancies.

All this makes me wonder why the Journal of Psychiatric Research didn't require the author(s) to fix the problem before being given the green light for publication. But better late than never, right?

The linked Reuters article is interesting in another sense. It begins by pointing out a potential bias in one of the two critics of the Coleman study:
In 2010 Julia Steinberg of the University of California, San Francisco, and Lawrence Finer of the nonprofit Guttmacher Institute published their own analysis of the same data from the comorbidity survey. They identified a number of errors in the Coleman paper, including statistical ones.
The Guttmacher Institute is a non-profit research and education group that advocates for reproductive rights, including access to abortion.
But if you read far enough into the article you find that Coleman may also be biased:
Another concern has been whether Coleman fully disclosed any possible conflicts of interest. In a presentation she gave in 2011 to the American Association of Pro-Life Obstetricians and Gynecologists, she said, "I have a plan to develop a new non-profit organization devoted to understanding and publicizing the real risks of abortion. I would like to bring together many credentialed scientists with a research program pertaining to the physical, psychological, and/or relational effects of abortion on women and their families."
Interesting, right? But note if the journal had initially used the old academic concept of peer review properly (by asking a skilled statistician to read the article, say), then the question of possible bias would have been irrelevant.

It's good to remember that even academic journals make mistakes, and, just like the rest of us, should learn from them. I applaud the Journal of Psychiatric Research for trying to fix its mistake.

I don't really applaud all these attempts to turn opposition to abortion into some sort of an empowering women's health issue. Giving birth can have serious health risks, including the risk of postnatal depression in women who are vulnerable to depression. But no state requires pregnant women to be counseled about that.