Caffeine in pregnancy and birth weight: randomised double blind trial

Kirsten Patrick looks at a recent randomised controlled trial that sought to find out whether caffeine intake in pregnancy affects length of gestation and birth weight

In some countries women are advised against drinking coffee during pregnancy because some studies have shown that babies whose mothers consumed more than 300 mg of caffeine a day during pregnancy had lower birth weights and shorter gestations than babies who had been exposed to little or no caffeine during fetal life. Not all studies have found this, however, with some studies showing no association at all. To add to the confusion, researchers have also noted that women who consume larger amounts of caffeine while they are pregnant also tend to smoke more, to drink more alcohol, and to be less well educated than women whose caffeine intake is low. Smoking, alcohol, and a lower level of maternal education are factors well known to be linked to low birth weight and prematurity, which indicates that caffeine may not be the villain it seems to be. In an attempt to answer the question of whether or not drinking coffee in pregnancy is safe, and to rule out other factors, these authors designed a randomised double blind trial.

Profimedia International sro/alamy

The authors recruited pregnant Danish women who admitted to drinking more than three cups of coffee a day. One cup of instant coffee contains about 75 mg of caffeine, and one cup of brewed coffee contains almost 140 mg, which means that most of the women in the study would have consumed 300 mg or more of caffeine daily by choice. Women completed a questionnaire about their caffeine intake after their antenatal booking visit and participants were recruited before 20 weeks gestation. Women were eligible for inclusion in the study only if they did not have a history of having a low birth weight or preterm baby and if they had no medical illness known to affect fetal growth; they were included in the final analyses only if they gave birth to a single, live baby. After recruitment the women were randomised to receive either caffeinated or decaffeinated coffee to take home and use in place of their usual brand. They were not asked to change their coffee drinking habits in any other way and were not discouraged from drinking other caffeinated drinks or from accepting coffee offered to them in social situations.

A computer generated a randomisation schedule and assigned serial numbers in balanced blocks of six. Staff, who were never in contact with the study participants, then applied a sticker with the serial number to a bag of either decaffeinated or caffeinated coffee, according to the randomisation schedule. The coffee manufacturer supplied both kinds of coffee in identical bags without labels for the purposes of this study. The researcher who interviewed the women registered each one with a serial number and then sent six bags of coffee with their serial number to every participant. This meant that neither the interviewing researcher nor the participant knew which kind of coffee she was using, which is called "double blinding." Double blinding ensures that the knowledge of which group a participant is in does not influence her behaviour or the way that the researcher interacts with her, which could affect the outcome of the study. At the end of the study the researchers asked each woman to guess which kind of coffee she had been given or to say that she didn't know.

In addition to information regarding their coffee drinking habits, study participants supplied details of their age, weight before pregnancy, height, tobacco use, educational level, and previous obstetric history. By collecting this data the researchers can double check that women in one study group are similar to women in the other, and that demographic factors are not swaying the study results. In this study the demographics of the women in the two groups were quite similar. This is almost always the case when a large number of people are randomly separated into two groups, and is the aim of careful randomisation.

Throughout their pregnancies the study participants regularly gave information on how much study coffee they drank, as well as stating how often they consumed other drinks containing caffeine (cola, tea, coffee, and cocoa) and how much they smoked. They were asked to keep a daily diary.

The researchers also collected information about pregnancy outcomes. Birth weight and length of gestation were the main outcome measures of interest. Babies' length, head circumference, abdominal circumference, placental weight, and Apgar scores were of secondary interest.

The researchers calculated that they needed a sample of at least 800 women for the study to show significant differences in birth weight. This statistical calculation is called a power calculation. You can't really do a study that hopes to show differences between two groups without first estimating how many subjects you will need in your study, otherwise you wouldn't know when to stop recruiting.

In this study 568 women were randomised to receive caffeinated instant coffee and 629 women to receive the decaffeinated variety. A small percentage of women from each group dropped out of the study before the end, which is quite usual.

Predictably, women who received caffeinated instant coffee consumed more caffeine on average during the study period than women who received decaffeinated coffee. Information obtained from interviews throughout the study period allowed researchers to estimate each woman's daily caffeine intake. Women in the caffeinated coffee group consumed an average of 182 mg/day more caffeine than those in the decaffeinated coffee group.

Babies' birth weights hardly differed at all between the two groups. The researchers found that the mean difference in birth weight between the caffeinated and the decaffeinated coffee groups amounted to only 16 g after they had taken into account factors such as maternal height and weight and babies' gestational ages at the time that their mothers were recruited into the study. Similarly, there were only small, non-significant differences between the groups for babies' head and abdominal circumferences, lengths, Apgar scores, and placental weights. Length of gestation also seemed to be unaffected by caffeine intake.

Maternal smoking, however, did seem to affect birth weight regardless of the mother's caffeine consumption. Women who smoked more than 10 cigarettes a day at the start of the study had babies with lower birth weights. The average difference between babies whose mothers smoked more than 10 cigarettes a day and those whose mothers smoked less was 263 g, which was a significant difference.

These researchers certainly tackled their research question in a uniquely clever way. Previous studies of the effects of caffeine intake during pregnancy on baby outcome have not been able to adequately take into account other factors such as maternal smoking, alcohol consumption, body mass index, and level of education. This study was able to minimise the impact of those factors through careful randomisation. Double blinding also helped to minimise differences in behaviour between groups. More women in the decaffeinated coffee group guessed correctly which group they had been in, which might have been because they experienced withdrawal symptoms, but blinding was achieved about as well as it could have been. By choosing to recruit women who already consumed a large amount of caffeine and artificially reducing the intake in one group, the researchers also avoided the ethical issues that would arise if the intervention was to increase caffeine consumption in one group.

Even though the study shows that a substantial decrease in caffeine consumption during pregnancy does not lead to higher birth weights or fewer preterm deliveries, it can not tell us that caffeine is safe for the developing fetus. We still know that caffeine takes longer to be eliminated from the pregnant woman's blood stream than from the non-pregnant woman's, and that caffeine increases circulating catecholamine concentrations, which can lead to uteroplacental vasoconstriction and fetal hypoxia. Caffeine also passes freely across the placenta and studies have shown that fetuses do not metabolise caffeine very easily.

This trial also only examined the effects of different levels of caffeine consumption in the second half of pregnancy. If caffeine's greatest impact on fetal growth is in the first half of pregnancy, at which time all these women were consuming high levels of caffeine, then this study would simply have failed to show differences where they do in fact exist.