The mainstream media has been touting a new study that supposedly shows the free birth control given out under the Obamacare will help reduce the number of abortions.
According to the Washington University team conducting the study, the rates of abortion and teen pregnancy dropped when St. Louis women were introduced to a wide range of “no cost” contraception. The report, which was based on data collected from 2008-2010, found a contrast in the number of births (6.3 per 1,000 teenagers) in this study versus the national average at the time, 34 per 1,000.
Immediately, abortion backers and the media pounced on the research as justification for forcing employers to foot the costs of these drugs.
Jeanne Monahan, Director of the Family Research Council’s Center for Human Dignity says the study is flawed.
“The report may appear persuasive, but it’s in direct conflict with studies from Sweden, the U.K., and Spain, which show that contraceptive use actually leads to higher pregnancy and abortion rates because it encourages riskier sexual behavior,” she says. “In this instance, Washington University chose women who wanted reversible contraceptive methods, but didn’t compare their data with that of a neutral control group.”
“You could even conclude, that the President’s mandate might ultimately cause more unplanned pregnancies since it orders health plans to cover contraceptives that these researchers claim are less effective. Of course, one thing that is effective about Obama’s mandate is how thoroughly it tramples religious freedom. And if that continues, the birth control may be free, but Americans won’t be,” she adds.
Meanwhile, one new analysis indicates the study is flawed and birth control, while reducing abortions in the short term, increases the number of abortions in the long-term:
What kind of study is this?
There are different kinds of medical research. An ecological study looks at two or more variables across a population (e.g. the BMJ study associating OCP use with prostate cancer). Ecological studies provide the weakest evidence for causation, since innumerable factors could confound the link. A cohort study provides a slightly stronger level of evidence. Cohort studies select a population of patients who are exposed to a particular variable of interest and studies them over time (e.g. Sexual Behavior and OC: A Pilot Study).
What kind of research is Dr. Peipert’s study? At first I labeled it a cohort study: it’s watching an exposed group over time for a variable (numbers of unintended pregnancies). But the study itself reads:
the analysis comparing repeat abortion in the St. Louis region with that in Kansas City and nonmetropolitan Missouri is essentially an ecological study. There may be several factors that affect the rates of repeat abortion, such as the economic recession, federal changes in Title X funding for family planning, and Missouri state laws that limit access to abortion.
So, it’s more than likely that the study is is a hybrid of the lowest standard of evidence for causation and a slightly higher one.
Are there confounding variables?
Patient self-selection is may be a factor. The CHOICE project was advertised by ”newspaper reports, study flyers, and word of mouth,” and participants were either self-referred, or ”recruited from the two abortion facilities in the St. Louis region and through provider referral….”
It’s not inconceivable that patients who had a pre-existing desire for LARC joined the study to have these relatively expensive methods provided at no cost. The CHOICE project counseling that patients at WUSTL received heavily emphasized LARCs, even before patients were enrolled in the study. There are no controls for women’s baseline interest in LARCs or their relative motivation to avoid pregnancy in the study.
What did they do?
To learn more about the CHOICE project, I read the study describing how the CHOICE project enrolled and counseled patients. Enrollment as described here did not include a medical records review other than a history; there was no physical exam by the contraceptive prescriber; participants were interviewed by a trained counselor who often did not have a healthcare background; and there were no follow-up face-to-face interviews or exams (participants were followed up by telephone at three and six months after enrollment, and every six months thereafter).
I am also mystified by the hazy quantification of the key outcome (unintended pregnancies). I imagined that participants would be asked “were you pregnant in the last six months” and (if yes) “did you mean to get pregnant?”
But instead, Peipart et al. report proxies of unintended pregnancies like repeat abortions, abortions, and teen births. Repeat abortions were counted by the abortion facilities working with the CHOICE project; teen births were counted by state vital statistics records. The study does not detail how abortions were counted, so I contacted the corresponding author. Were abortions or unintended pregnancies measured directly in any way, such as by questionnaire or medical record review? His reply:
Outcomes were measured by telephone interview and asking each participant if they experienced a pregnancy or if they missed a menstrual period. [Awesome! I thought that would do it!] If they did miss a period (or were late), we would encourage them to come in for a pregnancy test. Some (but not all) pregnancies were validated with a medical chart review, when available. [Oh. That sounds…not very rigorous.]
First: how can the direct assessment of the key outcome variable be so unimportant as to not be included in the paper? An entire paper was published detailing how many women chose IUDs and implants, but I had to contact the author about the key variable in CHOICE’s pinnacle study??
Second: why didn’t the team include all their data in the study? A table quantifying missed periods, true and false positive pregnancy tests, and possible pregnancies lost to follow up would have been completely appropriate. In fact, I expected it and became wary when I didn’t find it.
What are the results?
As I stated above, the abortion rates for the CHOICE project participants were lower than for the surrounding area. These results were measured over three years, which I consider slightly premature given that 75% of the participants chose LARCs, which last 3 to 10 years (3 for an implant; 5 to 10 for an IUD). What happens to these women long-term? There is evidence to show that contraceptives reduce abortion rates and unplanned pregnancies in the short-term and increase them in the long run.