A study purporting to show a causal link between abortion and subsequent mental health problems has fundamental analytical errors that render its conclusions invalid, according to researchers at the University of California, San Francisco (UCSF) and the Guttmacher Institute. This conclusion has been confirmed by the editor of the journal in which the study appeared. Most egregiously, the study, by Priscilla Coleman and colleagues, did not distinguish between mental health outcomes that occurred before abortions and those that occurred afterward, but still claimed to show a causal link between abortion and mental disorders.
The study by Coleman and colleagues was published in the Journal of Psychiatric Research in 2009. A letter to the editor by UCSF’s Julia Steinberg and Guttmacher’s Lawrence Finer in the March 2012 issue of the same journal details the study’s serious methodological errors. Significantly, the journal’s editor and the director of the data set used in the study conclude in an accompanying commentary that “the Steinberg-Finer critique has considerable merit,” that the Coleman paper utilized a “flawed” methodology and that “the Coleman et al. (2009) analysis does not support [the authors’] assertions.”
Steinberg and Finer initially published an analysis in 2010 in the peer-reviewed journal Social Science and Medicine showing that the findings of the 2009 Coleman study were not replicable. The JPR editor’s commentary now supports that conclusion. (The full sequence of events is detailed below.)
“This is not a scholarly difference of opinion; their facts were flatly wrong. This was an abuse of the scientific process to reach conclusions that are not supported by the data,” says Julia Steinberg, an assistant professor in UCSF’s Department of Psychiatry. “The shifting explanations and misleading statements that they offered over the past two years served to mask their serious methodological errors.”
The errors are especially problematic because Coleman later cited her own study in a meta-analysis of studies looking at abortion and mental health. The meta-analysis, which was populated primarily by Coleman’s own work, has been sharply criticized by the scientific community for not evaluating the quality of the included studies and for violating well-established guidelines for conducting such analyses.
“Studies claiming to find a causal association between abortion and subsequent mental health problems often suffer from serious methodological limitations that invalidate their conclusions,” says Lawrence Finer, director of domestic research at the Guttmacher Institute. “In thorough reviews, the highest-quality studies have found no causal link between abortion and subsequent mental health problems.”
Even when identified, spurious research can have far-reaching consequences. Mandatory counseling laws in a number of states require women seeking an abortion to receive information, purportedly medically accurate, that has no basis in fact. Among other things, mandatory counseling can require that a woman be told that having an abortion increases her risk of breast cancer, infertility and mental illness. In reality, none of these claims are medically accurate. These laws not only represent a gross intrusion into the doctor-patient relationship, they serve to propagate misinformation, intentionally misinforming the patient on important medical matters.
Full Sequence of Events:
May 2009: Priscilla Coleman, Catherine Coyle, Martha Shuping and Vincent Rue publish “Induced abortion and anxiety, mood, and substance abuse disorders: Isolating the effects of abortion in the national comorbidity survey” in the Journal of Psychiatric Research (JPR). In it, they purport to show a relationship between past abortions and mental disorders that were present “at the time of data collection, providing assurance that in most cases, the abortion preceded the diagnosis,” thus ostensibly supporting a causal relationship between abortion and subsequent mental health.
October 2010: Julia Steinberg and Lawrence Finer publish a reanalysis of the National Comorbidity Survey dataset in the journal Social Science & Medicine, demonstrating that Coleman et al.’s percentages of women with mental disorders are much too high for events occurring in the past 30 days (the measure most similar to “present or absent at the time of data collection”), and also identifying a number of other errors in Coleman and colleagues’ analytical approach.
December 2010: Washington Post health reporter Rob Stein and CBS News health reporter Aina Hunter write pieces on their respective publications’ blogs on Steinberg and Finer’s paper. Coleman responds on the Post blog indicating that the discrepancy exists because their analysis includes mental disorders that occurred in the past 12 months, not 30 days.
December 2010, February 2011 and May 2011: Steinberg and Finer contact the editors of JPR pointing out the errors in Coleman et al.’s paper.
March 2011: In a presentation to the American Association of Pro-Life Obstetricians and Gynecologists, Coleman reiterates her claim that her paper includes mental disorders over the past 12 months.
July 2011: Coleman et al. publish a corrigendum (a statement of error and correction) in JPR, indicating that they used incorrect weights in their analysis. The corrigendum does not clarify whether they used 30-day or 12-month measures of mental disorders.
August 2011: Steinberg and Finer submit a formal letter to the editor of JPR showing conclusively that rather than using 30-day or 12-month measures, Coleman et al.’s analysis included all mental health disorders occurring over the respondents’ entire lifetime, many of which occurred before any abortions, thus negating the causality argument.
September 2011: Coleman publishes a meta-analysis in the British Journal of Psychiatry (BJP) in which she includes this study and other studies conducted by her.
January 2012: Coleman’s meta-analysis is strongly criticized in BJP (here, here, here, here, here, here, and here) for violating several established guidelines for conducting meta-analyses, failing to evaluate the quality of included studies, not adhering to her own exclusion and inclusion criteria, and including studies that did not adjust for prepregnancy mental health.
February 2012: Steinberg and Finer’s letter on the JPR study is published in JPR along with a short response from Coleman. Coleman’s response acknowledges that she and her colleagues did in fact use lifetime mental health diagnoses (rather than 12-month or 30-day diagnoses, as she previously stated or implied) because they wanted to “capture as many cases of mental health problems as possible.” The letter and response are accompanied by a detailed critique coauthored by Ronald Kessler, principal investigator of the National Comorbidity Survey, and Alan Schatzberg, editor-in-chief of JPR. The commentary states that “the Steinberg-Finer critique has considerable merit and that the Coleman analysis does not support their assertions that abortions led to psychopathology in the NCS data.” It describes Coleman’s justification for using lifetime diagnoses as “unpersuasive” and calls the approach of using lifetime diagnoses a “flawed method.” It concludes that the analysis “should consider only mental disorders subsequent to the pregnancy, again arguing for the greater relevance of the Steinberg-Finer results (which focused on recent mental disorders) than the Coleman et al. (2009) results (which considered lifetime disorders whether they occurred before or after the abortion).”