Robert F. Kennedy Jr. has had a long love affair with junk science, and as secretary of Health and Human Services, he has embraced it once more, most brazenly to justify his false claims that vaccines cause autism. Last week, he brought yet another shoddily designed study to a different fight. In a Senate Committee hearing, he cited a report that few scientists would recognize as science in order to justify an FDA safety review of the drug mifepristone, which is used in the majority of abortions in the United States.
President Donald Trump had previously asked HHS to study the drug’s safety, and Kennedy emphasized at the hearing that a review of the drug would be a top FDA priority. The unusually high rate of adverse events identified in the report, he noted, “indicates that at very least, the label should be changed.” In other words, the top U.S. health official is prepared to rework—based at least in part on a poorly designed report that has not undergone scientific review—the government’s official guidance on a widely used drug.
The report that Kennedy cited was posted late last month to the website of the Ethics and Public Policy Center, a Washington, D.C.–based think tank focused on “pushing back against the extreme progressive agenda while building a consensus for conservatives,” according to its website. The study’s authors, Jamie Bryan Hall, EPPC’s director of data analysis, and Ryan Anderson, the organization’s president, are not health experts, and neither seems to have a record of publishing scientific research through peer review. Their methods deviated wildly from what is standard in the world of health research, and so, predictably, did their conclusions: In sharp contrast to dozens of trials conducted around the globe over decades, the EPPC report determined that mifepristone is a danger to women.
The EPPC has written that its report “presents a careful and conservative assessment of abortion pill safety.” However, the study lacks basic transparency about how that assessment was made. The authors relied on data from an insurance database that, according to the report, included more than 800,000 mifepristone abortions from 2017 to 2023. But the authors don’t actually say which database they used, so “there’s no way for anybody to try to re-create their analysis to see if they receive the same results,” Sara Redd, of the Center for Reproductive Health Research in the Southeast at Emory University’s Rollins School of Public Health, told me. (In an email, Hunter Estes, EPPC’s communications director, told me that the center’s contract with their data vendor prevents EPPC from sharing the name of the database or even of the vendor. But, he added, “this insurance data is available from approximately a dozen data brokers and is widely used by researchers and health professionals.”)
[Read: The other abortion pill]
The report also took some peculiar methodological steps to arrive at its conclusions. One of its key findings is that more than 10 percent of people who take mifepristone experience what the study refers to as “serious adverse events.” (A variety of studies put the rate of significant adverse events from medical abortions involving mifepristone at less than 0.3 percent, which makes the drug safer than Tylenol and Viagra.) But the EPPC study’s unusually wide-ranging criteria for defining those events raise a lot of questions. The researchers counted ectopic pregnancy as an adverse event, arguing that doctors should have ruled it out before prescribing mifepristone. (The American College of Obstetricians and Gynecologists acknowledges that mifepristone can be dangerous in cases of ectopic pregnancy but says that ruling out the rare condition—a process that involves an ultrasound—is unnecessary for most women taking the drug.) The authors counted episodes in which a surgical procedure was required to complete the abortion after mifepristone—patients require additional treatment in about one in 20 cases, so the FDA considers this a recognized outcome rather than an adverse effect. They counted “other life-threatening adverse events,” including heart problems and mental-health concerns, that women in the study experienced in the weeks after the abortion—which may have had nothing to do with mifepristone.
They also counted “serious” events documented during emergency-room visits made within 45 days of a patient taking mifepristone. However, the report doesn’t fully explain how they knew that those events were connected with mifepristone, and to judge which ones counted as “serious,” they used a scale designed for cancer research, which has not been validated for use in studies of abortion care. Loosely counting emergency-room visits could artificially inflate the estimate of risk associated with getting an abortion, Ushma Upadhyay, an epidemiologist and a reproductive-health researcher at UC San Francisco, told me: In a study she led of abortion-related emergency-room visits from 2009 to 2013, half of patients had such mild symptoms that they did not need any treatment. She also said that the authors did not effectively distinguish between the outcomes of abortions and of miscarriages treated with mifepristone, or between normal amounts of post-abortion bleeding and severe hemorrhage.
In the weeks following the report’s publication, EPPC published two follow-up documents with more details about the study’s methodology, which experts told me are still not convincing. As the documents explained, the authors relied on diagnostic codes to separate miscarriages, which are often also treated with mifepristone, from abortions—a practice that may yield imprecise results. The report included only suicidal and homicidal ideation among mental-health diagnoses categorized as serious adverse events—but that still does not prove that those diagnoses were connected to an abortion, Redd told me. It used “only codes related to hemorrhage or serious bleeding (according to the FDA definition)”—which would still not be enough to distinguish between the normal amount of post-mifepristone bleeding and something more serious, Upadhyay said.
[Read: A possible substitute for mifepristone is already on pharmacy shelves]
According to EPPC, peer review of the report was not possible due to “extensive pro-abortion bias in the peer-review process,” but a group of data scientists, analysts, and engineers “conducted and validated” the project, with assistance from doctors. None of their names appears on the report. When I asked about that decision, the EPPC representative wrote, “It is routine for individuals with controversial opinions to be subjected to a range of personal and professional attacks, including threats of violence in their own homes.”
So far, the most prevalent attacks on the study have been about its substance. Alice Mark, an ob-gyn and the medical director of the National Abortion Federation, told me that “to call it a study dignifies it too much.” Some anti-abortion advocates, too, have cautioned against overstating the study’s rigor: Earlier this month, Politico reported that Christina Francis, the CEO of the American Association of Pro-Life OBGYNs, said on a private Zoom call with anti-abortion leaders that although the report contains credible data and should inspire further research, it is “not a study in the traditional sense” and “not conclusive proof of anything.”
Anti-abortion activists have long seen mifepristone as a problem. In the years since the Supreme Court’s 2022 decision to overturn the national right to abortion, abortions have increased in part due to a 2021 FDA decision that allowed mifepristone and misoprostol (a drug often used in parallel for abortion) to be prescribed via telehealth and mailed. According to reporting by Politico, questioning mifepristone’s safety is part of a larger strategy called “Rolling Thunder” that aims to cut off that access. High-quality data have failed to validate those questions, so second-rate research has often been used to make the case against mifepristone. In 2023, for example, a federal judge ruled that mifepristone should be taken off the market by citing low-quality studies that reported adverse effects from mifepristone. (The Supreme Court later threw out the lawsuit on procedural grounds.) Due to their “lack of scientific rigor,” two of the studies cited were ultimately retracted by the journal that had published them.
[Read: Anti-abortion conservatives’ first target if Trump returns]
When, in the past, the FDA has evaluated mifepristone’s safety—which it’s done several times since mifepristone’s initial approval, in 2000—it has expanded access to mifepristone rather than curtailed it. If the agency evaluates mifepristone again, and its staff are allowed to independently assess the science, the FDA could loosen its rules for mifepristone even more, Elizabeth Raymond, an ob-gyn and a researcher who specializes in mifepristone safety, told me. Plenty of data support using mifepristone later in pregnancy than is currently approved, for instance.
But Upadhyay told me she worries that FDA Chief Marty Makary—who has previously claimed that fetuses can “resist” the tools of abortion by 20 weeks of gestation—or Kennedy could put their thumb on the scale to restrict mifepristone access, regardless of what FDA staff recommend. “I don’t want them to do a review, because I don’t trust them to base any decisions they make on science,” Upadhyay said. (HHS and the FDA did not answer my questions about the FDA’s plans to review mifepristone safety on the basis of the EPPC report. In an email, an HHS spokesperson told me of the FDA, “The agency rigorously evaluates the latest scientific data, leveraging gold standard science to make informed decisions.”)
Although Kennedy has said that he reads scientific papers critically for a living, his approach to the medical literature most resembles “an extreme version of what lawyers do to defend a client: create a narrative and then find supporting evidence,” Robert Califf, who led the FDA under Presidents Joe Biden and Barack Obama, told me an email. The scientific method involves the opposite: constructing a hypothesis and trying to disprove it with an open mind. When different people conducting the same experiment come to the same conclusion, it’s not a sign of a shared ideology; it’s a sign of a shared reality.