In the News. Digging into the Math of a Study Attacking the Safety of the Abortion Pill, The Washington Post

The Ethics and Public Policy Center report on mifepristone has earned headlines on the right, but what does the data show?

Analysis by Glenn Kessler in https://www.washingtonpost.com...

The Ethics and Public Policy Center, a think tank that says it opposes “the extreme progressive agenda while building consensus of conservatives,” recently issued a report on a key abortion medication, mifepristone, that it says raises questions about its safety. After analyzing insurance claims for more than 865,000 prescribed mifepristone abortions, the group said it had determined that almost 11 percent of women experienced a “serious adverse event,” much higher than an overall 0.5 percent rate found in clinical studies.

That headline number led Republicans like Sen. Josh Hawley (Missouri) to declare that the abortion pill is dangerous. “Those are astounding, jaw-dropping numbers, and they have been largely hidden from the public,” he said in an opinion article in the Federalist. An adverse health event, he noted, is “a major, potentially life-threatening medical disaster. We’re talking about things like sepsis, infection, and hemorrhaging, the kind of things that land you in an ER.”

Citing the report, Hawley on Thursday secured a commitment from Jim O’Neill, the Trump administration nominee for deputy secretary of Health and Human Services, to review the safety of mifepristone.

A medication abortion using a combination of mifepristone and misoprostol in 2023 accounted for 63 percent of abortions, double the percentage in 2014, according to the Guttmacher Institute, which supports abortion rights. Since 1988, about 100 countries have approved the use of mifepristone, according to Gynuity Health Projects, which also favors abortion rights.

We dug into the EPPC report, which calls the use of mifepristone “chemical abortions,” and posed 20 questions to EPPC, which were answered in detail via email by spokesman Hunter Estes after consultation with the researchers. The report lists 94,605 adverse events, for a rate of 10.93 percent, between 2017 and 2023. (Some women had more than one event, so the report says it avoided double-counting when calculating overall rate.) But the headline number seems less solid when the individual data presented in the report is examined.

We should note that, unlike most credible medical studies, the report did not undergo a formal external peer review before publication. Estes said “the study was internally reviewed and adjudicated by a panel of physicians, who carefully evaluated the clinical classifications, coding, and outcome assessments to ensure medical accuracy and consistency.”

Moreover, the report oddly does not reveal the database it used, though similar examinations of insurance data have revealed the source of the information.

Estes cited a confidentiality agreement with the vendor. “Given the high level of politicization and public scrutiny surrounding this topic, there are concerns — shared by both our team and, we believe, the data vendor — about potential backlash or reputational harm to entities associated with this research,” Estes said. He added that because similar data sources are available commercially, such as the Medicaid Analytic eXtract (MAX) files and the Merative MarketScan Research commercial database, “we do not view the confidentiality of our specific dataset as a barrier for others interested in obtaining similar data and replicating or extending this research.”

The report says it follows the Food and Drug Administration’s definition of a serious adverse event (SAE), and that all such events in a 45-day window after the abortion were reviewed and adjudicated by a panel of physicians. The report says a 45-day window was chosen to focus on events related to the abortion and not a possible subsequent pregnancy.

As a reader service, we will dissect the report by category of event, starting with the biggest. The most serious problems associated with abortion, such as sepsis and infection, rank rather low. One interesting finding is that clinical studies show similar rates for some adverse events to the EPPC study — and the inclusion of some adverse events raise questions.

‘Other abortion-specific complications’ (5.68 percent)

This is the biggest single category, accounting for about half of the incidents, but the report offers a vague description: “[diagnostic] codes specifically related to an abortion or miscarriage, as well as life-threatening mental health diagnoses, etc.”

Estes said “other abortion-specific complications include problems like damage to the woman’s internal organs from the abortion, fetal tissue remaining inside the woman’s uterus after the abortion, kidney failure, as well as life-threatening mental health diagnoses occurring after the abortion.”

Regarding mental health diagnoses, Estes said “we limited mental health codes to only those that met the criteria for life-threatening in order to not overestimate that category,” such as R45.850 (Homicidal ideations) and R45.851 (Suicidal ideations). But in response to our questions, he admitted “there were zero instances of patients with those two SAE mental health diagnosis codes in the 45-day window following the chemical abortion.”

Emergency room visit (4.73 percent)

An emergency room visit by itself does not count as a severe adverse event under the FDA definition, and people often go to ERs when they are on Medicaid or do not have a primary care doctor. Estes said the researchers tried to account for that. He said they excluded “72 percent of emergency room visits that took place within 45 days of abortion from our analysis because they were either not serious enough or were unrelated to the chemical abortion or because they were not serious,” such as covid exposure, nicotine dependence, asthma and anxiety disorder.

It’s worth noting that the FDA-approved label of Mifeprex, the brand name for the abortion pill, says that clinical studies in the United States found that as many as 4.6 percent of patients had an emergency room visit. Estes said that the nearly identical rate of 4.7 percent in the study further supported “the validity of the data.”

This is a big part of the study’s overall statistic — and yet it’s already on the abortion-pill label.

Hemorrhage (3.31 percent)

In this instance, there is a large disparity with the drug label, which reports that a non-U.S. study found a hemorrhage rate of 0.1 percent. Este said this difference shows why real-world data matters, as clinical trials and international studies may occur under conditions that are different from American health-care settings. He said the study relied on diagnostic codes with descriptions including the word “hemorrhage” or “abnormal uterine or vaginal bleeding” that were determined to be severe or life-threatening.

However, normal bleeding occurs with a medication abortion.

Just relying on a diagnostic code, rather than a treatment code, might elevate the numbers. We raised the possibility of a patient who has an episode of heavy bleeding with expulsion of the pregnancy, and the clinician may use the imprecise diagnosis code of hemorrhage to justify checking the patient’s blood count. But then the patient feels fine, the bleeding has resolved, and the blood count is normal.

Estes reiterated that each case was assessed as to whether it met the definition of a serious adverse event. He added that “patients who underwent surgical intervention due to incomplete abortion were separately categorized and did not artificially inflate hemorrhage or other complication rates.”

Repeated (surgical) abortion (2.84 percent)

Here again, the results are similar to the label on the drug: a range of 2.6 percent to 3.8 percent. We were puzzled by the inclusion, as it’s clear the abortion pill does not always work, and then an actual abortion must take place. Even when used properly, the two-pill regimen works about 87 to 98 percent of the time, according to Planned Parenthood.

Moreover, if an actual abortion is an adverse event, the pill prevents many more adverse events from happening.

“Surgical intervention was classified as a serious adverse event when it met criteria for an SAE,” Estes said. “Had the chemical abortion worked as intended, no surgical intervention would have been necessary. The label’s documentation that sometimes the drug fails and therefore requires a surgical intervention does not change the fact that it is a serious adverse event under the FDA definition.”

Infection (1.34 percent)

This figure is also much higher than on the label, which cites a rate of 0.2 percent from a non-U.S. study. There are infections that may be unpleasant — such as yeast and urinary tract infections — that result from engaging in unrecommended behavior after an abortion.

Estes said that did not matter.

“We are reporting the experiences of real women who take mifepristone in America today, with whatever preexisting conditions, quality of health care, and degree of compliance with recommendations these women have in practice,” he said. He added that the study excluded cases of vaginitis (which can result from, among other things, a yeast infection), including those treated in an outpatient setting.

Hospitalization (0.66 percent)

This rate is similar to the pill label, which says clinical studies show a rate as high as 0.7 percent.

Ectopic pregnancy (0.35 percent)

While the rate is small, this is another puzzling inclusion. Ectopic pregnancy is supposed to be ruled out before using mifepristone. We noted that the data appeared to indicate when the pill was prescribed. Sometimes a patient is prescribed mifepristone while a workup is ongoing for an ectopic pregnancy, but the patient doesn’t take the medications immediately. If an ectopic pregnancy is subsequently confirmed, the patient never takes the mifepristone and misoprostol.

“We acknowledge that a limitation of our dataset is that it does not capture whether a patient actually ingested the medication — only whether it was dispensed or administered,” Estes said. “This is an inherent constraint of claims and administrative data.”

Other life-threatening events (0.22 percent)

Transfusion (0.15 percent)

Sepsis (0.10 percent)

These are relatively small numbers. On the label, the rates for transfusion are higher — as much as 0.5 percent — as well as for sepsis (0.2 percent).