Intended for healthcare professionals


Caffeine intake during pregnancy

BMJ 2008; 337 doi: (Published 03 November 2008) Cite this as: BMJ 2008;337:a2316
  1. Jørn Olsen, professor1,
  2. Bodil Hammer Bech, assistant professor2
  1. 1UCLA School of Public Health, Department of Epidemiology, Box 951772, Los Angeles, CA 90095-1772, USA
  2. 2Institute of Public Health, Department of Epidemiology, University of Aarhus, DK-8000 Aarhus C, Denmark
  1. jo{at}

    Should be minimised, but not replaced with unhealthy alternatives

    In the linked cohort study (doi:10.1136/bmj.a2332), the CARE Study Group reports that consuming caffeine during pregnancy is associated with an increased risk of fetal growth restriction.1 For 100-199 mg caffeine a day the odds ratio was 1.2 (95% confidence interval, 0.9 to 1.6), for 200-299 mg a day it was 1.4 (1.0 to 2.0), and for over 300 mg a day it was 1.5 (1.1 to 2.1).

    Coffee and tea contain a variety of chemical compounds, but most of the health concerns relate to caffeine. One cup of coffee contains about 100 mg of caffeine and a cup of tea about half of this amount; the exact amount varies according to cup size, brewing methods, and brands of coffee or tea.2 Caffeine is also present in cola, chocolate, cocoa, and some drugs. Most of the caffeine that adults consume comes from coffee,3 but in the present study 60% of the caffeine that pregnant women consumed came from tea.1

    Some studies have shown no negative association between drinking tea during pregnancy and fetal growth,4 5 which may indicate that caffeine is not the culprit or that the results for coffee were confounded. However, the present study’s finding of an adverse effect in tea drinkers will reinforce the concern that caffeine is a potential fetotoxic substance.

    Caffeine easily crosses the feto-placental unit, and in large doses it may cause harm to the unborn child. Observational studies have shown associations with reduced fetal growth and fetal death,6 7 8 but not with preterm delivery.9 All of these studies are subject to confounding by unmeasured or partially measured factors.

    Drinking coffee and tea correlate with other lifestyle factors like smoking, work load, and perhaps also dietary habits. Only one study on drinking coffee has used a randomised design, and this study found no overall reduction in birth weight in women who regularly drank coffee containing caffeine compared with those who drank decaffeinated coffee, except in those who smoked.10 Unfortunately, Boylan and colleagues did not stratify their results on smoking habits or cotinine concentrations.

    The metabolism of caffeine depends on genetic and environmental factors.11 Caffeine is metabolised in the liver primarily by CYP1A2 and NAT2, and people can be classified as slow or fast metabolisers. In observational studies, therefore, the fetuses of women who are slow metabolisers will be exposed to more caffeine than the fetuses of fast metabolisers with an equivalent caffeine intake.

    The use of longitudinal sampled biomarkers for caffeine exposure is needed to bypass this problem, but such data have not yet been collected for large samples of pregnant women. Boylan and colleagues rely on self reported intake of products containing caffeine, so their data on caffeine exposure will have large measurement errors. It could be argued that these errors are probably non-differential and may therefore lead to underestimated effect sizes.

    The women in Boylan and colleagues’ study did, however, perform a caffeine half-life challenge test, and although this test was not validated, the results show that faster metabolisers have a higher risk than women with a slower clearance rate who have the same caffeine intake. This indicates that it is not caffeine but one of its metabolites that causes harm—this notion is supported by another study based on biomarkers for caffeine exposure.12 A possible effect modification by clearance rates is important, but as always confounding is a concern. For example, the comparison may be confounded by stress and sleeping patterns or by some of the many other compounds that are metabolised by the same enzyme system. Slow metabolisers with a high intake of caffeine may not be comparable with fast metabolisers with a similar intake.

    Boylan and colleagues found less variation in tea consumption during pregnancy than is normally seen for coffee intake. Many women report an aversion for coffee during pregnancy but not for tea. The authors found no indication for a threshold effect or for a larger effect later in pregnancy when the fetal demand for energy is highest.

    Their results will probably reopen the debate on what advice we should give to women who are or intend to become pregnant. The authors propose that every effort should be made to stop pregnant women consuming caffeine or to reduce intake greatly, because the association between fetal growth restriction and caffeine is reduced for those consuming less than 100 mg a day.1 But we think that this advice is not justified by the current body of evidence, and that such advice may unnecessarily frighten women who have consumed caffeine while pregnant. We do, however, think that pregnant women should be advised to reduce their intake of caffeine products during pregnancy, but that they should not replace caffeine containing beverages with drinks containing alcohol or soft drinks loaded with sugar.


    Cite this as: BMJ 2008;337:a2316



    View Abstract

    Log in

    Log in through your institution


    * For online subscription