Skip to content
Lies, Damned Lies, and Statistics

A Good Headline: Teenage Use of Contraception Leads to Depression. Never Mind if It’s True.

A new study claims that oral contraceptives increase mood-related mental illness in women. It does not.

Birth control pills
A one-month dosage of hormonal birth control pills is shown in Sacramento, California. (Rich Pedroncelli/The Associated Press)

Taking a break from the overwhelming wave of bad COVID-19 research, this week I’d like to mention a new study on contraceptives. The study claims to have discovered brain alterations in women who started using oral contraception as a teenager that would help explain “the increased vulnerability to mood-related mental illness in women.” 

It does no such thing. 

The study involved two groups of college-age women: 75 who had a brain MRI (27 were oral contraceptive users, and 12 of those started as a teen), and 140 who did a stress test (58 were oral contraceptive users, and 26 of those started as a teen). 

Already we can see that this isn’t a large sample of women who use contraceptives at all, let alone those who started as teenagers. 

Worse, this wasn’t a random assignment study or anything close to it: Women weren’t randomly assigned to use contraceptives or when to start. Women who make radically different choices might already be somewhat different. Indeed, if you look at the table of descriptive statistics, the two groups of women (users and non-users) were very different from the outset. For example, 80-85% of the users were white, compared to only 35% of the non-users (who were mostly Black and Asian). We don’t know how different the teenage users were, because the researchers don’t bother to say. 

Even worse, the researchers took too many measurements to list: hormones in saliva, memory tests, stress tests, and activity in literally every area of the brain. They then report a staggering number of correlations that overwhelm the reader. 

For example, while there were no differences in estrogen or cortisol levels, non-users “in the luteal phase of the menstrual cycle showed higher salivary progesterone levels” compared to users. Also, users’ brains had more white matter volume in the left parahippocampal gyrus (whatever that is). The study goes on like this for several pages. 

At least the researchers told the reader about many of these (likely meaningless) correlations, but then they made a fateful choice: to selectively highlight a few of them. 

That is, among all the possible correlations, the researchers “found” that within users, there was a “blunted cortisol response” after the stress test in women who began using contraceptives as a teen, which they take to mean that “there may be long-term effects” that can “be a risk factor for depression.” 

That stark conclusion isn’t justified at all. Among the relative handful of college students who responded to participate in this study in the first place, there’s no telling why a particular subset of them might have different hormonal or brain response to stress tests. 

It makes a good headline, though: Teenage use of contraception leads to depression. Never mind if it’s true.

Grants

Arnold Ventures funds projects to understand problems and identify policy solutions.

Map of the U.S. made of Arnold Ventures icons