A study, published in BMJ Evidence Based Medicine, looked at caffeine use and pregnancy.
At Science Media Center, Dr. Adam Jacobs, Associate Director of Biostatistics, Premier Research, said:
“I note the author has published 2 books on the dangers of coffee which in my opinion should have been included in declarations of interest for the journal article.
“One of the factors mentioned as a strength of the paper is that the meta-analyses had consistent results. It seems likely that the meta analyses are summarizing mostly the same primary studies, so that consistency is hardly surprising. If the primary studies are affected by biases and confounding, then those biases and confounding are going to be consistently brought through into the meta-analysis.
“There are some obvious potential confounders here. Given that pregnant women have been advised to avoid excessive caffeine consumption for at least the last 40 years, you might expect that women who drink coffee during pregnancy are generally less likely to follow health advice, and possibly in some ways which are quite hard to measure. Even if obvious confounders such as smoking status are taken into account, more subtle confounders may still have affected the analysis and make it hard to be sure that any observed associations with caffeine use are causal.
“There is also the question of publication bias. In general, null studies are less likely to be published than studies with positive results, so it’s possible that there have been other, unpublished, studies showing that caffeine has no effect on pregnancy outcomes. This is not addressed in the paper at all.
“Finally, I note that the paper is a narrative review, which is generally less reliable than a systematic review.
“Overall, I think if I were a pregnant woman, I don’t think this paper would make me overly concerned about drinking the occasional cup of coffee.”
Webinar 
Perspectives Blog 