Abortion Reversal Misinformation in the Guardian
Researchers gratuitously endangered pregnant women for no possible benefit to them -- and the reporter tasked with calling it out, spread misinformation instead
In these troubled times, a lot of people turn to predictable stories for comfort. Whether your comforting myths of choice are fairy tales or bad romcoms, the pay-off is that they follow a familiar structure. So even when it’s a new story, you know what to expect.
Hyperpolarized discourses are great for this. There’s always a pure evil adversary (the other side). A pure good intent (your side). And no pesky information gets in the way. This is not to deny the evil in the world against which good people must struggle. It just makes my ears tingle when I hear that form of story, because I know it from the echo chambers of dark propaganda.
This is a post about yet more abortion myths. As I wrote previously, abortion carries substantial possible risks — most notably a 2x+ increased suicide risk. Prominent abortion providers make common methodological mistakes, systematically misinforming women about these risks. But both sides are guilty of claiming their science is the only true science, which is silly. No one wants to admit we’re stuck in this infinite regress of uncertainty — not knowing a lot, trusting sources that could be wrong, making human errors, and generally being puny mortals. The problem is not sui generis, but reflects complex modern societies grappling (or mostly failing to sentiently grapple) with their profound limitations. This contrasts with the high ideals of what Greenland terms romantic heroic-fantasy science as secular religion.
From that top-level interest in this case study, I’ve been meaning to write about the discourse on abortion pill reversal. It’s a great case study in inescapable misinformation in the misinformation discourse — misinformation misinformation damnation. It’s inescapable, because calling something “misinformation” is a political act — an act of spin. But there is no perfect neutrality possible from which position to correct bias and mistakes. And I have to write about this case now, because it just got better when The Guardian got worse. (But there are so many of these misinformation misinformation damnation cases studies; I’ve recently done two of these on Covid, this one in the abortion series, and this one on a recent EU campaign promoting mass surveillance with misinformation.)
For general background, abortion with pills involves a first drug (mifepristone) administered about a day before a second drug (misoprostol). The idea is that the first drug blocks progesterone, the hormonal star of the pregnancy show. Then, the second causes uterine contractions to expel the fetus. So if a woman takes the first drug and changes her mind, maybe taking a bunch of progesterone to override having blocked it will help her keep the pregnancy. Natural progesterone is super safe as far as we know* (*although we’re pretty ignorant primates). Pregnant women use it all the time in cases of threatened or previous miscarriages. (Doctors used to give them synthetic forms of it too, and this was a very bad idea; but that’s a different story.)
Some preliminary research supports this possibility, that following mifepristone with a progesterone flood can preserve wanted pregnancies after partly attempted medical abortions. For instance, in “Progesterone for preventing pregnancy termination after initiation of medical abortion with mifepristone,” The European Journal of Contraceptive & Reproductive Health Care, 2017, p. 472-475, Deborah Garratt and Joseph V. Turner report three cases where women who took the first drug changed their minds and tried to reverse the abortion. Progesterone use worked for two of the three to keep their pregnancies, and they delivered healthy babies at near-term or term. The dose for these women who changed their minds <48 hours after taking the first drug and having not taken the second drug was: 400 mg 2x/day vaginal progesterone for 3 days, then 400 mg at night for the next 6 days, then 200 mg at night for the next 6 days. So it seems to possibly work pretty well. What’s the problem?
Opposition to Abortion Reversal Research and Practice
There is a huge amount of opposition to abortion reversal research and practice from people who are supposed to be proponents of women’s rights — and specifically, women’s rights to make their own reproductive choices. This opposition comes in part in the form of misinformation saying that the reversal won’t work. The most an opponent can credibly say is that we don’t really know, but that’s not the same thing.
Yet, if you don’t read the underlying scientific evidence and just go on heuristics, it really looks like a bunch of neutral, credible sources pan the technique as useless. The UK Royal College of Obstetricians and Gynaecologists says “ ‘No reputable evidence’ for progesterone use in ‘abortion reversal’, say medical organisations.” Healthline.com says “There Isn’t a Way to 'Reverse' Emergency Contraception Pills or Medical Abortion.” And Self.com says “No, It’s Not Medically Possible to ‘Reverse’ Abortions.” This is all demonstrably, empirically wrong.
All these sources are making the same nullism error as the leading abortion providers who are systematically misinforming women that abortion is risk-mitigating or risk-neutral, when the best available evidence suggests that it actually may more than double their suicide risk. By contrast, it looks like conservative, biased sources say that abortion pill reversal could work — like “When It’s Not Too Late: Can the Abortion Pill Be Reversed?” from Focus on the Family, which Wikipedia describes as a fundamentalist Protestant organization. But, in this case, the fundamentalists are giving out better information than the doctors.
So if you just went on heuristics, like some decision scientists suggest people should do to make better decisions, you would get an inaccurate picture of the evidence here. It may be tempting to say this is a sad reflection on the state of bias in science today. But really it’s just an expected feature of a hyperpolarized discourse that happens to dress up in the secular religious language of science.
Abortion Reversal Attempt Access As Reproductive Choice
Anyway, the neutral-looking sources are all biased and demonstrably, empirically wrong on this. So, as a matter of informed consent, you might say, women should get to know that they don’t have to proceed with the abortion if they change their minds, given that maybe it can be reversed. They could stop the pill protocol and try the progesterone flood, instead. That’s exactly what Garratt and Turner argue in “Providing women with full reproductive choice options: the point of researching medical abortion reversal,” The European Journal of Contraceptive & Reproductive Health Care, 2018.
The point of the research presented is exploring how best to provide women with their full and rightful reproductive choice options. Although far from ideal, if she changes her mind once, twice or thrice she should still be entitled to all available options. A spokesperson for the Royal Australian and New Zealand College of Obstetricians and Gynaecologists (RANZCOG), and prominent pro-abortion activist, commenting on abortion reversal stated that ‘We don’t have women suddenly changing their minds – this is basically not a problem this is a furphy’ [6]. The incongruity brought to light by this spokesperson as well as the respond- ent is that it ‘Seems like RANZCOG is pro-choice ... as long as the choice is only abortion’ [6] (Comments section).
…Disparaging the case report as an anti-abortion effort to dress up political aims as science [2] is without basis and a professional discourtesy.
These arguments stand, but I have a somewhat different view here. Scientists don’t have to pretend to be apolitical, amoral beings (none of us are). It would be fine if Garratt and Turner opposed abortion, and developed a technique to help women who regretted their abortions to attempt reversal. That would still be valid science. The women who wanted to reverse their abortions would still be people who should get the chance to try that if it might work. It should not be for abortion providers to decide that abortion is available and reversal attempts are not — but it is, and they do. And that is a social and political decision that serves abortion provider and proponent interests. Just as all of this science and all of these decisions are embedded in social, political, and moral contexts and consequences from which there is no escape. We should be honest about that.
It’s also worth noting that denying women the opportunity to try to reverse abortion if desired may also contribute to preventable harm to women who otherwise may go on to experience substantial post-abortion distress. It may help some women resist coercion to terminate a wanted pregnancy. It may prevent some risks of medical abortion, like profuse bleeding (see below). The point is that there all all kinds of ways to envision abortion reversal attempts as potentially benefiting women’s health and agency.
Abortion Reversal Misinformation Case Study: Outline
But this is a hyperpolarized discourse, so abortion proponents toe the party line instead — even when that means endangering women and mangling the evidence. Here’s the story of:
How incompetent researchers subjected pregnant women to predictable, preventable, and pointless severe pain and risk in service of panning the possible effectiveness of the progesterone flood abortion pill reversal technique. A technique which their findings predictably do not end up saying much about, although their own evidence fits the overall picture suggesting that it may work.
And how The Guardian then mangled the evidence to support the preferred story about how abortion is safe, and reversing it is dangerous and impossible. The truth is that the evidence suggests abortion — and specifically medical abortion — poses serious possible risks. And more research is needed about the promising potential of progesterone treatment to reverse initiated medical abortion among women who want to do that.
Misinformation in Abortion Reversal Misinformation Discourse
On October 23, 2023 in The Guardian, Carter Sherman published “US judge blocks law that would have been first to ban ‘abortion pill reversal’: Abortion foes claim practice can be used to halt medication abortion while studies don’t support potentially harmful theory.” Sherman’s Guardian page says “Carter Sherman is a reproductive health and justice reporter at Guardian US,” and lists six articles in this issue space, all from October. I am afraid to read these other articles, because this one is so bad.
The title is wrong. As the evidence above suggests, researchers claim some women desire abortion reversal, and progesterone may work for this purpose. It’s obvious spin to label the researchers “abortion foes,” reduce their published research to a claim, and misrepresent the best available evidence as showing abortion reversal is bunk.
The article gets wronger. Sherman writes:
The first randomized, controlled clinical study to attempt to study the “reversal” protocol’s effectiveness, however, was suddenly halted in 2019, after three of its participants ended up in the hospital hemorrhaging blood.
Let’s unpack this a bit. The study at issue is “Mifepristone Antagonization With Progesterone to Prevent Medical Abortion: A Randomized Controlled Trial,” published by Mitchell D Creinin, Melody Y Hou, Laura Dalton, Rachel Steward, and Melissa J Chen in Obstet Gynecol, 2020 Jan;135(1):158-165.
Creinin et al’s Research Misconduct
This was a crazily unethical study. (Spoiler alert.) It concludes “For now, [progesterone abortion reversal] treatment is experimental and should be offered only in institutional review board [IRB]–approved human clinical trials to ensure proper oversight.” This is hilarious, because a functioning IRB would have forbidden this study itself from a mile away. It actively endangered women for no possible benefit to them. This is forbidden by widely accepted research ethics laws and norms. Human subjects research must balance benefits and risks to subjects, particularly vulnerable subjects.
Here are some of the study’s problems:
It enrolled 12 patients, setting it up to be underpowered to detect a difference between treatment and control groups according to their calculations (p. 3). Then researchers downplayed that they did detect potentially meaningful differences, anyway: 1 woman in the progesterone group reported experiencing bleeding that increased to severe, versus 3 in the placebo group (p. 7, Table 2). And “four of six patients in the progesterone group and two of six patients in the placebo group had continuing pregnancies at 2 weeks” — or 4/5 and 2/5 if you exclude the placebo group participant who requested surgical abortion because the bleeding scared her and the progesterone group participant who requested it because she had horrendous nausea (p. 3). Those exclusions are questionable, and the evidence still goes in the suggestive direction that the abortion reversal may work.
This study used oral instead of vaginal progesterone, when there’s a lot of literature suggesting that the former route tends to be wasteful for pretty much everything (hormones, medications, recreational substances). Then it found variability in progesterone level that “may be explained by differential oral absorption of progesterone” (p.6). Yes, Virginia, people use their stomachs for things other than pills. To be fair, there was a basis in the literature for using the oral route instead; but it comes from Delgado et al’s analysis of 325 different medical professionals’ differing attempts to help women who called a helpline wanting to reverse their abortions. They used several different forms and doses of progesterone for this purpose; only 31 involved high-dose oral. There was comparatively way more data on intramuscular injection from that analysis. But ok, injections are labor-intensive and yucky. Maybe at least telling women to not eat or drink for two hours around the drug would have helped, but then they’d probably need to be invested in keeping the pregnancies they intended to abort to actually do it.
Creinin et al enrolled women who were at gestational age 44-63 days (mean 52.5 days). Gestational age is calculated from the first day of the last menstrual period, whereas usually fetal age is dated from conception. So normally we would say these pregnancies were at around 5.5 weeks (38.5 days). That may sound early enough for an abortion. But medical abortion is worse than surgical abortion in terms of severe pain and risks like prolonged bleeding, heavy bleeding, and hemorrhage. These risks are worse the later the abortion. Henshaw et al reported that women randomized to get medical versus surgical abortions were much more likely to say they would choose the other method in the future (22% versus 2%, p.001) — and most of the difference (95%) came from women who got medical abortions who were over 50 days gestational age.
Meanwhile, the researchers in this study made women who were “too early” wait and come back to enroll. They should have known based on previous literature that they had a problem with gratuitously causing pain and bleeding risks. Instead of addressing the problem, they actively structured this study’s enrollment in a way that would be expected to made it worse. At the same time, they also subjected women to unknown additional risks. Previous research on mifepristone-only abortion was, in their words, “small and limited primarily to pregnancies of 49 days or less” (p. 6). It has to be said that a particular set of political views would characterize a team of researchers who think it’s a-ok to wait and keep a pregnancy alive longer before terminating it for research. This is not neutral science (which does not exist).
There was a systematic difference between gestational age between progesterone and placebo groups: 49.5 (47-56) versus 55 (48-61). This often happens in the process of randomization. Blocking on gestational age to prevent this from happening by chance would have been a good idea; this technique is standard in medical research to prevent this kind of problem. As Delgado et al and others note, mifepristone works better for earlier abortions. So this likely systematically biased Creinin et al’s results against showing mifepristone alone worked (possibly heightening the bleeding risks of failed medical abortion), while also biasing the results against showing progesterone’s benefit. It also would have been good to show results by gestational age, since we know this variable matters. These are common design and analysis flaws.
Creinin et al emphasize the scaremongering potential of their findings rather than the positive potential, saying “Patients who use mifepristone for a medical abortion should be advised that not using misoprostol could result in severe hemorrhage, even with progesterone treatment” (p. 5). Yeah, but progesterone seems like it may lessen the bleeding and sometimes save the pregnancy. So someone with different political views might reasonably draw the exact opposite conclusion from the same evidence — namely, that this study supports offering progesterone treatment early and often to women who take mifepristone for abortion, but who may not wish to continue with taking misoprostol to finish it.
Creinin et al emphasize the editorializing potential of their findings rather than recognizing the political nature of the discourse and trying to keep their policy views out of their science at least a little bit, concluding:
some states now require physicians who provide medical abortion to counsel patients that the actions of mifepristone can be reversed if they change their mind… laws should not mandate counseling or provision of any treatment when we do not fully understand treatment efficacy (including best route of administration, dose, and duration) and safety (p. 7).
But telling women what we know, including what we don’t know, is better science and better ethics than letting abortion providers with conflicts of interest opt to not make information on abortion reversal available (much less provide the service) — which is what they otherwise appear to do. By contrast, Delgado et al conclude “The use of progesterone to reverse the effects of the competitive progesterone receptor blocker, mifepristone, appears to be both safe and effective” (p. 11). These teams are writing about the same body of evidence (except Creinin et al’s inconclusive-but-suggestive RCT results post-date Delgado et al). The point is that the evidence is exegetical, and it’s political to say that abortion providers should or shouldn’t at least try to honestly tell women what it says.
It’s similarly political for abortion proponents to deny that some women regret their decisions and want to reverse them. And it’s more than political to embed your own views about this in your scientific article. It doesn’t actually legitimate those views as science — but it does read that way to people who can’t critique the scientific discourse. This is a power play and an abuse of scientific authority.
But enough about Creinin et al’s crazy unethical study that should never have been done, proves only that current research methods and ethics education is a shanda, and actually provides additional suggestive evidence that progesterone may work to reverse the abortion pill, while medical abortion may cause a lot of bleeding — a risk that progesterone treatment may mitigate. Let’s talk about Sherman’s irresponsible, predictably biased misinformation about this study.
Sherman’s Misinformation
Recall Sherman’s representation of Creinin et al’s study:
The first randomized, controlled clinical study to attempt to study the “reversal” protocol’s effectiveness, however, was suddenly halted in 2019, after three of its participants ended up in the hospital hemorrhaging blood.
This misrepresents reversal, not medical abortion, as the risky thing. But in the study, the researchers themselves even recognize that it was the abortion pill, how late it was given, and how it was given without the second drug in follow-up — not the progesterone — that caused the three hospitalizations for heavy bleeding (p. 3-4).
You couldn’t get this from reading Sherman. You have to read the study, which the article doesn’t link to (a bad standard practice driven by perverse incentives: websites don’t want to lose their viewers to external sites because your attention/data is their money), and/or know something about the subject area already, to smell the mistake. But you only had to read the study abstract on the first page to get it.
There, Creinin et al say “Severe hemorrhage requiring ambulance transport to hospital occurred in three patients; one received progesterone (complete expulsion, no aspiration) and two received placebo (aspiration for both, one required transfusion). We halted enrollment after the third hemorrhage” (p. 1). Two-thirds of the hemorrhages resulted from the abortion pill and not the progesterone, because those patients didn’t get the progesterone. The progesterone-group hospitalization for bleeding didn’t actually require any treatment; the woman just got herself observed in case there was a problem.
So the pattern was really two hospitalizations requiring medical intervention for heavy bleeding, one of them getting a hospital diagnosis of hemorrhage. This reflects a design flaw. The lack of adequate safety data on mifepristone-only medical abortion after 49 days of gestational age (among other factors) should have keyed the researchers and their IRB into the fact that this study was unethical to run. It put the subjects at risk of serious harm for no benefit to them. The researchers should have caught this flaw before running the study, and not run it. Unless they did not care about the women or the science, and just wanted to publish something on abortion pill reversal saying it doesn’t work and looks risky.
A safer, more ethical version of this study that may have produced more information about abortion pill reversal efficacy was (and remains) possible. Researchers could have recruited women who actively wanted to keep their pregnancies, thus offering participants a possible benefit. They could have informed them fully about the risks — but the thing is, it’s not clear from existing data that progesterone to attempt abortion reversal is actually riskier than medical abortion itself. Rather, it looks possibly risk-mitigating. Similarly, researchers could even have capped gestational age for enrollment at 52 days, but the ethics of that would be questionable.
Basically, this alternative study design just looks like pro-choice instead of pro-life doctors and researchers talking to women who want to access information and progesterone to attempt abortion reversal, and giving it to them. Why aren’t abortion providers already doing this? I think that’s a good question that providers need to answer.
Similarly, Sherman could have reported on how terrible this study was, or how risky medical abortion actually is in terms of severe pain, and bleeding long and hard. (It’s much riskier than surgical abortion.) But that didn’t fit the preferred narrative. It was easier to scare people into thinking that attempting abortion reversal is what risks women’s lives, when there’s no evidence that it does anything other than save some wanted pregnancies — possibly preventing some severe side effects like severely heavy bleeding and (rarely) suicides in the process.
Creinin et al endangered women for the cause. Sherman just lied to them.
Pessimism and Zen
We don’t live in a world where informed consent meaningfully exists. Or research ethics, with some sort of social and professional penalty for violations thereof. Or even science with a capital S — that heroic-fantasy narrative we all know and love, that contrasts with the real human enterprise which is built on quicksand and needs reconstruction from the foundation up based on the causal revolution in a great many areas.
But at least we live in a predictably inaccurate world in which it probably doesn’t matter how publicly misinformation like Sherman’s gets corrected. Due to perverse incentives, erroneous scientific and journalistic records tend to stand, constructing an aura of validity and acceptance around the myths they spin. Publication is power, and power tells the stories it wants to tell.
Often, powerful social and political networks spin science in ways that require calling the rogue methodologist ninja help line to read between the lines. But in this case, all I did was read. If you’re going to write a scientific journalism article about science, maybe start with reading the first page of the actual science.
At the same time, it’s a problem for modern societies that we can’t systematize critical thought — and we don’t want to recognize that. Critical thought is a dynamic, context-dependent process (Feyerabend). So a major challenge in science — both individual decision-making science, and science as a huge swath of disciplines using diverse methods — is to help improve this process without using checklists or protocols. Rote procedures are wrong tools for dynamic processes. We need science infrastructure that performs “show not tell” critical thinking assistance to support the process, not dictate the product. We can’t build tools that make people read, but maybe we can build tools that help them read more critically. That’s what I’m still working on and will write more about in future posts.