We thank Dr Powell,1 Dr Garcia-Algar et al.,2 Drs Parker and Brennan,3 and Drs Astley and Grant4 for their interest in our recently published five papers on the potential effects of low to moderate alcohol intake during pregnancy. These papers also looked at the potential effects of alcohol binge drinking in early pregnancy on the neuropsychological development in young children.
First, we agree that the null results from our studies must be interpreted with caution. The published results were limited to the measurement of three neuropsychological effects in children aged 5 years, and did not include other ages or other health outcomes. Although these studies suggest that there are no serious effects on these three functions with low to moderate levels of alcohol consumption, they cannot rule out other harmful effects or negative health outcomes in older children.
We agree that reliable measurement of alcohol consumption, particularly alcohol consumption by pregnant women, is very difficult in both clinical and research settings. It is generally acknowledged that pregnant women are likely to underreport their alcohol consumption, irrespective of the actual level of consumption. This problem has been assessed extensively in the literature and many improvements in questions and the manner of questioning have been developed that have increased the reliability of the information. These improved techniques were applied in the Danish National Birth Cohort (DNBC), and some relevant references are provided in the papers. Based on post-partum meconium analysis, Garcia–Algar et al.2 suggest that pregnant women do not underreport, but misreport, and therefore that self-reports are unreliable and that no conclusion can be drawn. Although self-reporting may be flawed, we believe that it still yields important and useful information. At the time of data collection on alcohol exposure through the DNBC, procedures for obtaining exposure information represented a significant step forward. Questions were asked during pregnancy rather than years later after a child was found to have problems. We agree that, as bio-measures may be developed and refined, future studies may be able to hone in on more precise aspects of exposure.
Interestingly, in their own recent paper,5 the authors state that, ‘mothers from Mediterranean countries tend to underreport their drinking…., probably due to social pressure and guilt feeling.’ This may be correct, and stresses the importance of underreporting. This may also explain why Astley and Grant,4 describe that a small proportion of mothers with children who have fetal alcohol syndrome (FAS) reported a low to moderate weekly intake of alcohol. To our knowledge, in the scientific literature, FAS is generally described for much higher levels of prenatal exposure. As mentioned in our papers, one of the important methodological aspects of the collection of exposure data relates to the fact that, in Denmark, many pregnant women, doctors and midwives consider low to moderate alcohol consumption (such as the levels investigated in this study) to be acceptable.6–8 Therefore, as opposed to many other countries, self-reports by women in Denmark are likely to be less biased and the results therefore more reliable. As mentioned in some of the articles, this was one reason why the study was specifically conducted in Denmark.
Although we agree that objective measures of alcohol intake would be ideal, such markers do not appear to exist for low intake levels or for measurement in early pregnancy. The original cut-offs for meconium analyses were based on comparison with self-reports9 and comparison of women with no alcohol intake, social drinkers (levels not specified) and heavy drinkers. Essentially no differences were detected in the levels of biomarkers between self-reported nondrinkers and social drinkers. This suggests that, even if meconium analyses may distinguish between heavy drinkers and social drinkers/nondrinkers, they cannot be used at this time to measure low weekly alcohol intake in early pregnancy, and they have not been shown to reliably reflect a few binge drinking episodes in early pregnancy, the focus of our study.
When the Lifestyle During Pregnancy Study was planned, we were well aware of the difficulties in the assessment of executive functioning, particularly in preschool children. We were also aware that executive functions are not fully developed in 5-year-old children, and this was one of the reasons why we decided to use a parent and teacher rating scale, such as the Behaviour Rating Inventory of Executive function (BRIEF), rather than a child administered test. However, given the literature on prenatal alcohol exposure (at higher levels) and executive dysfunction, we thought that it was important to include at least some measure of these skills. In the paper, we point out that it is possible that the BRIEF is not sufficiently sensitive to detect small effects of maternal alcohol consumption. However, when evaluating the results for the BRIEF and executive functions, the results for attention and general intelligence should also be considered. These outcomes were assessed with child administered tests, and the results corroborate the findings for BRIEF. Finally, time constraints and the testing fatigue of the child needed to be considered as the IQ and attention measures required child administration. As stated in the articles, we hope future research will build on our findings as they consider potential measures of all areas of neurodevelopment.
In the Discussion section of the paper, we mention the age of the child at assessment as one important limitation of the study. We acknowledge that brain and human cognition is not fully developed at the age of 5 years and, consequently, it is possible that some long-term effects of prenatal alcohol exposure on cognition can be detected only later in life. Further, we understand that the concern about the age of assessment may be most relevant for complex executive functions. Although we recognise the problems related to the age of assessment, we believe that these problems should be seen in a balanced perspective that distinguishes between the assessment of individual children and the assessment of groups of children exposed to different levels of maternal alcohol consumption. For example, it is well known that the development of intelligence during childhood may be relatively unstable in the individual child, but also that intelligence is relatively stable from age 4–5 years on a group level.10 However, we note that at least one study suggests that early effects of prenatal alcohol exposure may be diluted in later childhood and adolescence, rather than becoming more pronounced.11 We hope that we conveyed that we see this study as an initial effort to measure these domains of neurodevelopment using these particular measures, and believe additional future studies will assess children at older ages, perhaps using different, newer assessment measures.
Drs Parker and Brennan3 suggest that ‘the data challenge such a large body of extant evidence’. Generally, our results seem to be well in line with previous findings.12,13
Finally, a few general comments are made by the authors. First, Dr Garcia-Algar et al.2 state that, ‘there is clear evidence from animal studies and from human clinical observation that prenatal exposure to alcohol has deleterious effects…’. We agree and mention in the papers that there is evidence that daily intake of alcohol may be potentially damaging to the developing fetus. This is well documented in the literature. Our aim was to shed light on the potential effects of low, weekly average intakes of alcohol and binge drinking independently of high daily consumption. We have done so in a follow-up design, which has the status of evidence level 2b. Animal studies and clinical observations are evidence level 5.14
Second, Dr Powell1 suggests that our study is irresponsible. Considering that alcohol is a known teratogen, that a large number of women drink small amounts of alcohol during pregnancy and that a large number of women admit to binge drinking in early pregnancy, we believe that it is reasonable to investigate to what extent this may influence the developing fetus and the child in later life. We consider the publication of our findings to be the responsible and ethical course. Again, our intent is that future studies should be conducted on this topic, building on our methods and results.
In conclusion, we believe that the Lifestyle During Pregnancy Study contributes important methodological and statistical approaches to the literature, and that these findings should be considered and incorporated in future studies of low to moderate alcohol consumption and binge drinking during pregnancy. We recognise that this is a single study in one specific population that used a very particular set of exposure and outcome measures, and by no means answers all questions regarding this topic. We reiterate our conclusion from each of the articles that, for pregnant women, small amounts of alcohol consumed occasionally may not present serious concern. However, prenatal alcohol exposure is known to cause adverse reproductive outcomes, birth defects and developmental disability. Thus, we believe that the most conservative advice for women is not to drink alcohol during pregnancy.
We again thank Drs Powell, Garcia-Algar et al., Parker and Brennan, and Astley and Grant1–4 for their thoughtful comments and for continuing the scientific dialogue on this important topic.