Recent headlines from a number of major news outlets are stirring up fears about the dangers of acetaminophen (sold as the branded drug Tylenol) during pregnancy:
But the news from this study isn’t nearly as nearly as dire as you might think based on this messaging. In fact, these headlines are either wrong or highly misleading for two simple reasons.
Some headlines had a more cautious tone
These misleading headlines weren’t the rule, however; some news outlets used more appropriately cautious language to describe the relationship between acetaminophen and behavioral symptoms:
Others were even more strongly skeptical and led with doubts about the quality of the evidence:
A 42% increase in risk? Why that figure is misleading
Another problem with some reporting was an over-reliance on relative risk figures provided in the study. The New York Times, for example, said that acetaminophen use was associated with “a 42 percent increased relative risk of conduct problems and a 31 percent higher risk of hyperactivity symptoms in children.”
Forty-two percent is technically accurate but 42% of what? The Times never tells us. The answer was accessible to reporters who drilled down to a supplemental e-table in the study. There it was revealed that the actual rates of conduct problems were 8.8% in children whose mothers didn’t use acetaminophen and 12.3% in children whose mothers did. The corresponding rates for hyperactivity symptoms were 9.5% and 12.5%.
An absolute difference of 3 percentage points could certainly encompass a whole lot of children given the popularity of acetaminophen. But it’s likely not what most consumers are thinking about when they hear the phrase “42 percent increase.”
‘Just a two percentage point difference’
Writing at Forbes, Sy Mukherjee made the same point using a slightly different set of statistics from the study.
Just under 5% of all the children in the study showed signs of increased behavioral problems, with just a two percentage point difference between those whose mothers hadn’t taken acetaminophen and those who had.
As did David Harris writing on the NPR Shots blog:
The 20 percent to 45 percent increase is actually a small change. To pick one representative endpoint: Among women who had not taken the drug, 4.3 percent of their children registered an elevated score on the “SDQ total difficulties” test. Compare that with 6.3 percent of children born to women who did take the drug.
So, the risk went from a small percentage to a slightly less small percentage of these children.
Several stories fortunately acknowledged limitations, confounding factors
Then there’s the issue of confounding factors and other study limitations. Interestingly, even the stories who whiffed on the headlines seemed to a pretty good job in this department. CNN, for example, cited an independent source — Dr. Hal Lawrence of the American College of Obstetrics and Gynecology — who explained that “Behavioral disorders are multifactorial and very difficult to associate with a singular cause. The brain does not stop developing until at least 15 months of age, which leaves room for children to be exposed to a number of factors that could potentially lead to behavioral issues.” (Side note: While Dr. Lawrence’s point remains valid, doesn’t the brain continue developing far beyond age 15 months?)
The story also quoted another source who pointed out the lack of dosage information and the fact that symptoms were reported by the parents, not professional observers, which could tend to skew the ratings.
CBS Denver went even further in a piece by medical editor Dr. Dave Hnida who says “it’s important to recognize that the study doesn’t prove the drug causes problems. In fact, far from it.”
Researchers had no idea how much and how often the women took the drug. One tab. Two? Twenty over two weeks?
There was no direct cause and effect: in other words—take this much acetaminophen and it will cause a behavioral problem later–didn’t see that here.
It did not measure the degree of behavioral problems or always use official diagnoses.
All in all, the study proved nothing solid.
Alarming headlines followed by more measured reporting
But if Hnida’s right about the flimsiness of the study, why then headline the piece with the fear-provoking question: “Will Tylenol cause your child to have ADD or ADHD?” and lead with the statement: “…a new study suggests that women who took acetaminophen at certain times during pregnancy may be putting their child at risk for later problems with behavior and hyperactivity.”
HealthDay took a similar tack, headlining its coverage: “Could Tylenol during pregnancy up risk of ADHD in kids?” An expert quote further down in the piece tells readers what they need to know about that question. “There are many degrees of separation and many bridges of evidence that would have to be crossed before you can establish a causal relationship. This needs a lot more science behind it.”
For today’s short-attention-span headline skimmers, I’d much prefer a headline and lead paragraph that speaks to the shaky nature of the evidence right up front.
A step in the right direction?
With that being said, I’m not one to let the perfect be the enemy of the good. With most stories I looked at — at least on this study — important caveats and nuance were communicated at some point in the coverage. And because of that, I think most readers will come away with a basically sound sense of what this study can and cannot tell us about acetaminophen and pregnancy.
That’s more than I can say about a lot of parenting and pregnancy stories. Not too long ago, for example, many of the same sites mentioned above sounded the alarm on seafood and mercury concerns in pregnancy as well as swaddling and SIDS risk without enough scrutiny of what the research couldn’t tell us.
Now, though, we see signs of more balanced reporting, and that’s a welcome trend to not just us, but parents and pregnant women.
Hat tip to Jennifer Brokaw, MD (@JenniferBrokaw) for pointing us to the study and related coverage.