My pregnant friend has studiously avoided cats (toxoplasmosis), alcohol (fetal alcohol syndrome), tuna (mercury), soft cheese (listeriosis), and sushi (oy gevalt). Recently she sent me this, wondering if she should also give up her daily macchiato habit:
New York Times, January 21, 2008, “Pregnancy Problems Tied to Caffeine”
This finding is surprising because no serious clinician has any idea how this could possibly work. Most miscarriages are caused by chromosomal abnormalities. Caffeine does many things, but nobody has ever shown that it harms chromosomes. It’s difficult to imagine how caffeine could possibly cause miscarriage. Is this harm real?
Anytime the press reports a possible peril to a developing fetus, it’s easy for the report to add an item to the list of pleasures that are prohibited during pregnancy. Here’s how:
Pretend that we’re concerned that watching television might harm a fetus. (It sounds plausible, right? As a society we believe that some things are likely to cause harm that we do not yet know, and these things include caffeine and television.) Let’s do a thought experiment that could make a medical researcher famous.
1) Give questionnaires to some women. Ask how much TV they watched during pregnancy.
2) Ask about lots of bad things that could possibly happen to a developing child: Obesity, prematurity, attention-deficit-hyperactivity disorder, and more.
3) Send the questionnaires to statisticians; instruct them to analyze the results in hundreds of ways.
If you analyze imperfect data in enough ways, you can always find some meaningless, coincidental correlation. Biostatisticians call this practice “data dredging.” This technique frequently yields misleading results, but some researchers do it anyway. This is how the caffeine/pregnancy study was done. Our health journalists could have ignored this exercise in data dredging, but, instead, they broadcasted it on TV.
Back to the thought experiment. Imagine that the statisticians dredge up a correlation between watching television in pregnancy and having a child with Down syndrome.
4) Publish the result.
Down syndrome, just like miscarriage, is caused by a chromosomal problem. It’s completely unknown how television—or caffeine—could possibly harm chromosomes. However, now the American Pregnancy Association states that women should minimize caffeine intake, even though they acknowledge that studies on this topic have been “inconclusive.” How do they explain their recommendation? “It is still better to play it safe when it comes to inconclusive studies.”
Our television study was also inconclusive. I suppose it’s always better to “play it safe,” after every study result, no matter how flimsy. No more coffee (miscarriage) or TV (Down syndrome) for pregnant mothers!
Yes, this thought experiment is inane. However, it’s difficult to show how it is different from the caffeine study. There are other, worse flaws with the methods of the caffeine study, making it difficult to conclude anything from it. Good science may someday show that tiny amounts of caffeine in pregnancy cause harm. However, seeing how little we know, it may turn out that caffeine is actually beneficial in pregnancy. We just don’t know. This study didn’t tell us any more about caffeine than our thought experiment told us about television.
This is not trivial. Thousands of women will suffer headaches from caffeine withdrawal as a result of this study. In exchange for their headaches, these women receive protection from a condition that caffeine probably cannot cause.
Let them drink coffee.
*The proper study design to answer this question is a randomized, placebo-controlled, double blind clinical trial. This sort of data dredging study—the one reported in the Times—is, as described above, largely useless for answering a clinical question.
This blog is not intended to diagnose, treat, cure or prevent any condition or disease. Please consult with your own physician or health care specialist regarding the suggestions and recommendations made within.