This is a big problem. Published papers are not removed from scientific journals because of minor mistakes. Papers are retracted rarely (around 1 in 500), but the rate is increasing, and the majority of these are due to misconduct. Retractions are decided by the authors (when they realize a major mistake) or by the publisher (for fraud, plagiarism, ethics, etc.).
The lawsuit was filed after anti-abortion doctors alleged that the Food and Drug Administration had ignored safety concerns when it relaxed restrictions on access to mifepristone. They based their case on a scientific study that cited the number of emergency room visits following mifepristone use to argue that the drug is dangerous. However, after the study was published, other scientists raised serious concerns about the statistical methodology and questioned the conclusions.
For example, outside groups found that one of the retracted studies used inaccurate medical codes (numbers used to interpret medical procedures and diagnoses) to count “abortion-related” emergency room visits. The study used a code for an ectopic pregnancy, which occurs naturally and is unrelated to abortion.
The findings were also presented in other deceptive ways, such as charts and graphs containing double Y-axes (two scales on the same chart). Below is a screenshot of the chart as it was originally published:
Emergency department visits within the first 30 days
Abortion
Source: Studnicki et al., 2021
Emergency department visits within the first 30 days
Abortion
Source: Studnicki et al., 2021
Visiting an emergency department within 30 days of the abortion
Source: Studnicki et al., 2021
However, when the y-axis is properly displayed on a single scale, abortions contribute significantly less to emergency room visits. Here’s the graph at the correct scale:
Emergency department visits within the first 30 days
Abortion
Source: Upadhyay and Adkins, 2024
Emergency department visits within the first 30 days
Abortion
Source: Upadhyay and Adkins, 2024
Visiting an emergency department within 30 days of the abortion
Source: Upadhyay and Adkins, 2024
After these concerns were raised, Sage commissioned two subject matter experts and one independent statistical analyst to peer review the studies (which were published in 2019, 2021, and 2022). Sage did not make the results of the expert reviews public, as is standard practice, but given the importance of this case, they may be made public.
Experts identified significant ethical issues and scientific errors, including that reviewers knew at least one of the authors of the three studies, and that several are members of the same pro-life advocacy group, despite the studies declaring no conflicts of interest. The Sage investigation also concluded that there were “unjustified or erroneous factual assumptions,” “significant errors,” and “misleading presentations” of the data that “demonstrate a lack of scientific rigor and invalidate the authors’ conclusions in whole or in part.”
Clinical guidance and policies are (ideally) built on decades of research and consideration of the entire evidence. In the case of mifepristone, over 100 studies have shown it to be safe; in fact, it is safer than Tylenol. There are only a few contradictory studies. However, big mistakes can get through the peer review process, and in rare cases the “mistakes” are intentional and egregious. Even if the studies are retracted, they can do great harm. (See Wakefield’s study on autism and the measles, mumps, rubella vaccine.)
We all deserve a strong data foundation to make smart medical decisions. This data foundation relies heavily on ethical scientists and a strong review process. Retracted studies should never end up in the Supreme Court.
