Arizona will soon require flawed medical advice as a part of abortion counseling. Last week, Gov. Doug Ducey signed SB 1318 into law, which mandates that providers use the following language with patients: “It may be possible to reverse the effects of a medication abortion if the woman changes her mind but that time is of the essence.” Moreover, the bill continues, “Information on and assistance with reversing the effects of a medication abortion is [sic] available on the Department of Health Services website.” Public policy and medical advice should reflect the best available evidence. What is the scientific evidence supporting “abortion reversal”?

The World’s Published Literature: Six Patients

As of now, no information is available on Arizona’s Department of Health Services website suggesting what medical techniques the government recommends for “abortion reversal.” However, the doctors testifying in support of SB 1318 have claimed that the hormone progesterone can be used to stop medication abortions. A search through the 20 million citations in the National Library of Medicine’s online database, PubMed, turned up only one published report on “abortion reversal.” Rather than a formal study, the article is an incompletely documented collection of anecdotes from Catholic physicians who tried to counter the effects of mifepristone, used in medication abortion, by administering progesterone.

The report, written by Drs. George Delgado and Mary Davenport, makes numerous scientific errors. First, the article recommends a progesterone regimen developed by Dr. T.M. Hilgers of the Pope Paul VI Institute for the Study of Human Reproduction that has not been formally studied and vetted in the peer-reviewed medical literature.

Second, no evidence supports the statement in SB 1318 that “time is of the essence.” In Delgado and Davenport’s article, the interval between mifepristone and progesterone administration ranged from seven to 72 hours for five patients and was unknown for the sixth. Thus, no inference is possible regarding any potential effect of timing. Again, the language of the bill strays far beyond any evidence.

Third, the authors inferred a causal association between the progesterone treatment and continuation of the pregnancy. This error in logic is known as post hoc ergo propter hoc (after the thing, therefore on account of the thing). In other words, the fact that a pregnancy continued after this treatment in no way implies a causal association, only a temporal one. The patients in this report received only mifepristone before their progesterone treatment. By the 1980s, the abortion success rate with mifepristone alone was recognized to be too low for general use; 7 to 40 percent of pregnancies continue after mifepristone alone. The addition of a prostaglandin (uterine stimulant) like misoprostol after mifepristone improved success rates to greater than 95 percent and began the era of medication abortion.

Fourth, no control or comparison group was used, violating the essence of the scientific method. A suitable control group here would have been women who received mifepristone, changed their minds, and did not receive progesterone. Since the article lacked a control group, one cannot say whether the progesterone treatment had any effect on the pregnancy. Even so, the article reports, “The experience of these patients suggests that medical abortion can be arrested by progesterone.” This statement is untenable.

Fifth, the tiny sample size limits any usefulness. A total of seven women underwent this therapy, but the authors were unable to follow up with one of them. This is unexplained, since all births are reportable events: All births and deaths are required by law to be reported to state governments and then to the National Center for Health Statistics. Four of six treated women (67 percent) continued their pregnancies. Any study provides only an estimate of the truth in the larger, general population; confidence intervals should be used by researchers to describe the precision of the estimate. Narrow confidence intervals indicate good precision, and vice versa. The 95 percent confidence interval around the 67 percent of continued pregnancies in this report ranges from 26 to 94 percent. If this natural experiment were repeated 100 times, the true figure for how often pregnancies would continue would fall within this 26-to-94 percent range 95 times out of 100. With this much imprecision, the article provides almost no information.

Finally, the article misuses epidemiologic terms. The report refers to two possible “confounding factors,” or potential causes for biased results: the lack of feticidal effect of mifepristone and lack of documentation of a viable pregnancy before receiving mifepristone. A confounding factor must be related to both the exposure (abortion reversal) and the outcome (pregnancy continuation), but not involved in the causal pathway. Neither “possible confounding factor” was related to the exposure and thus could not cause confounding bias. This suggests a lack of understanding of research methods.

The Alternative: No Treatment

A recent prospective study in France followed women given mifepristone for abortion and who subsequently changed their minds before receiving a prostaglandin. These women did not receive “abortion reversal” progesterone. Among the 46 women exposed to mifepristone at a mean of nine weeks’ gestation, 37 (80 percent) had a live birth. Only 17 percent had a miscarriage, and one chose an induced abortion when Down syndrome was diagnosed in her fetus. Thus, the continuation rate in France without progesterone treatment was similar to that in the Delgado and Davenport article with progesterone treatment.  

Risks of Pregnancy Continuation

Furthermore, SB 1318 does not reflect the potential dangers of interrupting an abortion in progress. Although mifepristone is not known to be associated with birth defects, methotrexate, a drug that is sometimes used instead of mifepristone, and misoprostol have been linked with fetal anomalies. Because misoprostol is used routinely with medication abortion regimens, the American Congress of Obstetricians and Gynecologists’ Practice Bulletin recommends that health-care providers counsel all women who wish to continue their pregnancies after beginning an abortion regarding potential birth defects. SB 1318 is silent about a physician’s ethical obligation to warn women about these potential harms.

Legislating Junk Science

The State of Arizona has effectively based public health policy on six informal clinical anecdotes. A descriptive study without a control group allows no conclusions about potential benefit or harm of treatment. Moreover, the experience of similar women in France suggests that using progesterone for “abortion reversal” has no benefit. Junk science has no place in courts of law, but it continues to prevail in many state legislatures, including the one in Phoenix. When politicians play doctor, everyone suffers. Citizens deserve better.