Background Various human and animal studies suggest that peak alcohol exposure during a binge episode, rather than total alcohol exposure, may determine fetal development. Research about the impact of binge drinking on birth outcomes is sparse and inconclusive. Data from the Born in Bradford cohort study were used to explore the impact of binge drinking on birth outcomes.
Methods Interview-administered questionnaire data about the lifestyle and social characteristics of 10 851 pregnancies were linked to maternity and birth data. The impact of self-reported binge drinking (5 units: 40 g of pure alcohol) on two birth outcomes (small for gestational age (SGA) and preterm birth (<37 weeks)) was assessed using multivariate logistic regression models, while adjusting for confounders.
Results The percentage of women classified as binge drinkers fell from 24.5% before pregnancy to 9% during the first trimester and 3.1% during the second trimester. There was a significant association between SGA birth and binge drinking (all categories combined; OR 1.68, 95% CI 1.15 to 2.47, p=0.01). No association was observed between moderate drinking and either birth outcome, or between binge drinking and preterm birth.
Conclusions Binge drinking during the second trimester of pregnancy was associated with an increased risk of SGA birth. No association was found between any level of alcohol consumption and premature birth. This work supports previous research showing no association between SGA and low-alcohol exposure but adds to evidence of a dose–response relationship with significant risks observed at binge drinking levels.
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Alcohol is a teratogen that can cross the placenta during pregnancy to enter the baby's blood, potentially altering the development of the fetus. Frequent alcohol use early in pregnancy has been linked to congenital malformations of the heart, brain and kidney1 and fetal death.2 Heavy alcohol consumption throughout pregnancy causes fetal alcohol syndrome3 and has been associated with preterm birth and growth restriction,4 as well as an increase in the risk of postnatal infection, neonatal mortality and physical and cognitive development.5 ,6 Growth-restricted babies have a greater risk of various neonatal complications including breathing problems, respiratory infections and hypothermia and impaired neurodevelopment.7 There is little evidence that low-to-moderate alcohol consumption is associated with adverse birth outcomes,4 ,8 although studies have shown that light-to-moderate consumption (1–4 drinks/week) may be associated with behavioural and mental-health problems in children aged 4–8 years.9 ,10 Consequently, drinking in moderation during pregnancy is considered acceptable by some national health bodies (UK, Switzerland), although in most countries in Europe and North America, total abstinence is supported.11 Despite this guidance, the prevalence of alcohol consumption during pregnancy varies from 30% to 70% internationally.12–15
Binge drinking is defined as drinking heavily in a short space of time or heavy episodic drinking.16 A recent increase in binge drinking in Western cultures, particularly in young people, has been well documented,17 leading to a view that alcohol may now be the most harmful drug to society.18 During pregnancy, the prevalence of binge drinking has been reported to be between 3% and 26% internationally.12–14 ,19 This type of alcohol consumption may pose a particular danger as animal studies suggest that peak alcohol exposure (ie, during binge episodes), rather than total-alcohol exposure, may determine fetal development.20 The few epidemiological studies looking at the relationship between prenatal binge drinking and birth outcomes have produced inconsistent results. A systematic review of the effects of prenatal binge drinking on birth outcomes reported no consistently significant effects on birth weight or preterm birth.21 Further studies have both supported19 and rejected22 an association with preterm birth. With respect to small for gestational age (SGA) births, studies report no relationship with binge drinking19 ,22 ,23 while further works have demonstrated an increased risk of SGA associated with pre-pregnancy,20 early and mid-pregnancy24 ,25 and late pregnancy binge drinking.26 A meta-analysis of the effects of alcohol during pregnancy,4 although not addressing binge drinking directly, reported increasing risks of SGA above 1.5 drinks/day and preterm birth above 1 drink/day, with higher drink averages likely to include women who binge drink.
Previous research has also reported that the effect of alcohol on birth outcomes may be modified by smoking.27 Various biochemical interactions of alcohol and tobacco with vitamins, folate and other antioxidants may affect growth restriction during pregnancy.28 Studies have described greater reductions in birth weight associated with drinking and smoking than drinking alone,29 as well as the additive effects of smoking and drinking on the risk of SGA.30
Research into the risk of binge drinking during pregnancy is required to inform public-health guidance for expectant mothers and support risk assessment during pregnancy. We used data from the Born in Bradford (BiB) cohort study to explore the impact of binge drinking, before and during pregnancy, on birth outcomes.
Participant and recruitment
Bradford is a city in the North of England with high levels of socioeconomic deprivation and ethnic diversity. The BiB study is a longitudinal multiethnic birth cohort study aiming to examine the impact of environmental, psychological and genetic factors on maternal and child health and well-being.31 Women were recruited between 2007 and 2010 at the Bradford Royal Infirmary while attending an oral glucose tolerance test (OTT) that all women are invited to at 26–28 weeks’ gestation. During OTT, women were provided with study information, full consent was discussed and obtained, and women were invited to complete an interviewer-administered baseline questionnaire while waiting to take a blood test. Over 80% of those attending were recruited. The cohort is broadly characteristic of the city's maternal population, with a slight under-representation of young women.31 Ethical approval for data collection was granted by the Bradford Research Ethics Committee (Ref 07/H1302/112).
Baseline questionnaire data about lifestyle and social factors were linked to maternity data for 11 378 pregnancies. We limited our analysis to singleton births due to potential differences in the aetiology of preterm and SGA in twins (327 twins and triplets were excluded). Stillbirths (n=74) and births where maternity data were missing (n=343) were excluded from the analysis, giving a final sample size of 10 851.
Preterm birth was defined as less than 37 weeks’ gestational age, with gestational age being based on the actual delivery date and the estimated date of delivery calculated by the physician or midwife from the dating scan (if available) or last menstrual period. SGA was defined as a birth weight less than the 10th percentile for the infant's gestational age, using gender-adjusted standard UK growth charts.32 SGA thresholds were adjusted for nulliparous/multiparous differences in birth weight using previous estimates.33 This was carried out by taking the percentage range in birth weight between nulliparous and multiparous women for each gestation week and imposing this range on the standard growth chart figures. Gravida minus one was used to estimate parity where parity information was missing (n=384) and unadjusted growth chart figures were used where both gravida and parity were absent (n=34).
As part of the baseline questionnaire, women were asked to estimate their alcohol consumption for three time periods:
3 months before pregnancy (before pregnancy);
The first 3 months of pregnancy (first trimester);
From the beginning of the fourth month until recruitment (second trimester).
For each time period, there was a general drinking question “Did you drink any alcohol (in the 3 months before pregnancy) (in the first 3 months of pregnancy) (from the beginning of the 4th month until now of your pregnancy?)” with possible answers; ‘alcohol once per week or more’, ‘alcohol occasionally’, ‘not at all’ or ‘don't remember’. Those who drank alcohol more than once per week were asked to estimate the weekly number of alcohol units they consumed. Women were also asked how often they had consumed five or more units of alcohol on one occasion, defined as a binge, with possible answers; every day, nearly every day, 1–4 times/week, 1–3 times/month, rarely and never. Five categories of alcohol consumption were defined from these questions: abstinent, low/moderate (<5 units/week with no binge drinking), binge (occasional (rarely or 1–3 times/month)), binge (frequent (1–4 times/week)), heavy (binge drinking nearly every day or every day)) following a previous method of using total alcohol consumption and frequency of binge drinking19 (table 1). The ‘Binge (frequent)’ and ‘heavy’ categories were combined into a single category, ‘binge (frequent)’, for the purposes of the statistical models due to the very low number in the heavy category.
A literature search was conducted to derive maternal (age at delivery, maternal height, ethnicity, pre-pregnancy hypertension, pre-eclampsia, pre-pregnancy diabetes, gestational diabetes, cohabiting status, body mass index (BMI)), birth (parity, route of birth, presentation), sociodemographic (mothers’ education, fathers’ education, mothers’ employment status, fathers’ employment status, financial status, social security benefits, neighbourhood deprivation) and behavioural (smoker, use of other tobacco products (eg, Paan), exposure to environmental tobacco smoke, caffeine intake, illegal drug use) risk factors that may have masked or confounded the relationship between binge drinking and birth outcomes (see table 2 and online supplementary table S1 for groupings and referent grouping). All variables were categorical apart from three mean centred continuous variables (height: mean 161.6 cm, age: mean 27.3 years, BMI: mean 26).
All covariables were screened using univariate logistic regression with separate SGA and preterm birth outcome variables. If significant (p<0.2), they were entered into fully adjusted logistic regression models for SGA and preterm birth. Backwards stepwise regression was used to simplify the models by sequentially removing non-significant variables that did not reduce how well the data fitted the models (using Akaike information criterion34).
The three smoking variables were multicollinear (r>0.79 in all cases), with 82% of first trimester smokers continuing to smoke during the second trimester. Consequently, we created a single smoking variable by combining the three time periods into the following categories; ‘didn't smoke before or during pregnancy’ (referent group), ‘smoked before pregnancy but quit during pregnancy’, ‘smoked before and during pregnancy (light smoker: 1–10 cigarettes per day’), ‘smoked before and during pregnancy (heavy smoker: >10 cigarettes per day’). Although the alcohol exposure variables showed some collinearity (r ranged from 0.24 to 0.46), all three variables were left in the fully adjusted models to examine the relative impact of alcohol exposure before and during pregnancy. The referent group for the alcohol variables was the abstinent group. We performed sensitivity analysis on these ORs by running the fully adjusted models with each alcohol variable included separately. Model specification was assessed through inspection of model residuals and goodness-of-fit tests.
We generated alcohol and smoking interaction terms for the first and second trimesters of pregnancy. Categories within the interaction variables were restricted to ‘abstinent’, ‘low-moderate’ and ‘all binge categories’ against ‘non-smoker during pregnancy’ or ‘smoker during pregnancy’.
This study had 80% power at a 95% level of confidence to detect a 30% increase in the odds of SGA associated with binge drinking (all categories combined) during the first trimester, and 54% increase in odds of SGA associated with frequent binge drinking. Equivalent power values for preterm birth were 43% and 82%, respectively.
Nearly a third of the women drank alcohol during the 3 months before pregnancy (29.3%), dropping to 15.1% during the first trimester and 11.6% during the second trimester (figure 1). The percentage of women classified as binge or heavy drinkers fell from 24.5% before pregnancy to 9% during the first trimester and 3.1% during the second trimester. In other words, 36.6% of women binge drinking pre-pregnancy continued to binge drink during the first trimester and 12.5% during the second trimester. The percentage of women who smoked before but not during pregnancy was 3.6%. This contrasts with 12.3% of women who smoked before pregnancy and then 1–10 cigarettes/day during pregnancy, and a further 3.9% who smoked before and then more than 10 cigarettes/day during pregnancy (table 2). Large ethnic differences in the prevalence of smoking during pregnancy (White: 33.3% (n=1429), Pakistani: 3.1% (n=155) and binge drinking during the first trimester (White: 20.7% (n=884), Pakistani: 0.04% (n=2) were observed.
Within the cohort, 13.3% of births were classified as SGA and 5.4% as preterm. The prevalence of SGA was highest among women who were abstinent (15.1%) or binge drinkers (9.3%) before pregnancy. This U-shaped risk pattern was repeated during pregnancy (table 3), although these figures are unadjusted and based on small numbers. High SGA prevalence in the abstinent group is largely explained by Pakistani women who are predominantly abstinent and have smaller babies (SGA=17.8%). There was a similar prevalence of preterm birth across all pre-pregnancy alcohol consumption categories. A higher prevalence of preterm birth was observed in women who frequently binged during the first and second trimesters, although these estimates are based on low numbers.
Within our fully adjusted models, SGA birth was not associated with any level of alcohol exposure before or during the first trimester of pregnancy (table 4). However, there was a significantly elevated odds ratio of SGA births associated with occasional binge drinking during the second trimester (OR 1.58, 95% CI 1.05 to 2.38, p=0.03) and all binge categories combined during the second trimester (OR=1.68, 95% CI 1.15 to 2.47, p=0.01). The population attributable risk associated with binge drinking was calculated as 9.7%. An apparent dose–response relationship between the odds of an SGA birth and binge drinking during the second trimester did not reach statistical significance for the frequent bingeing category (OR 2.30, 95% CI 0.92 to 5.77, p=0.08), an imprecise estimate based on only 36 women. There was no statistically significant association between any level of alcohol exposure before or during pregnancy and preterm birth (table 5).
Smoking during pregnancy was significantly associated with SGA (1–10 cigarettes/day during pregnancy OR=2.29, 95% CI 1.87 to 2.8, p<0.001; >10 cigarettes/day OR 3.17, 95% CI 2.34 to 4.31, p<0.001) and preterm birth (1–10 cigarettes/day during pregnancy OR=1.81; 95% CI 1.4 to 2.35, p<0.001, >10 cigarettes/day OR 2.06, 95% CI 1.35 to 3.04, p<0.001). ORs for smoking along with other variables within the fully adjusted models are contained within online supplementary tables S2 and S3.
Our interaction terms showed no synergistic effect of smoking and binge drinking on birth outcomes. Smoking rather than binge drinking conferred significantly raised odds of SGA (alcohol abstinence smokers (first and second trimesters; p<0.001), low-to-moderate drinkers who smoked (first trimester; p=0.04). Alcohol abstinent smokers had significantly raised odds of preterm birth during the first (p<0.001) and second (p=0.01) trimesters. Stratified analysis confirmed the importance of smoking as binge drinking smokers had greater odds of SGA than binge drinking non-smokers (see online supplementary tables S4 and S5).
Sensitivity analysis indicated that by adding the alcohol variables one at a time, the ORs did not alter by more than 20% and their statistical significance (above or below p=0.05) did not change. However, the significance of the frequent binge category during the second trimester category changed from p=0.08 to 0.03 (with OR increasing by 13%) when added individually to the SGA model, reflecting its marginal significance. The fully adjusted models provided reasonable explanation of the observed data (area under the curve for the SGA model=0.71, preterm model=0.66; Hosmer and Lemeshow’s goodness-of-fit tests were non-significant, SGA=0.36, preterm=0.50). Model residuals exhibited an approximate normal distribution around the mean with a slight positive skew for both SGA and preterm models, as would be expected with relatively rare outcomes.
Within our fully adjusted model, there was a 68% increase in the odds of SGA associated with binge drinking during the second trimester of pregnancy. Our findings demonstrate that 1 in 10 SGA births may be associated with binge drinking during pregnancy. This study found no association between alcohol exposure before pregnancy or during the first trimester and SGA birth. No association was found between any type of alcohol exposure and premature birth.
Seven of 10 pre-pregnancy smokers continued to smoke during pregnancy. Women who smoked during pregnancy had double the odds of a preterm or SGA birth. In contrast to smoking, there was a sharp decrease in binge drinking during pregnancy. Two-thirds of the women binge drinking in pre-pregnancy stopped binge drinking during the first trimester and seven of eight stopped during the second trimester.
How this compares with previous works
A previous systematic review21 found no consistent evidence of adverse birth outcomes associated with binge drinking but noted methodological weaknesses that we have addressed via measuring binge drinking prenatally and during different stages of pregnancy, and encompassing two measures of binge frequency. Previous work looking specifically at the association between binge drinking during pregnancy and SGA, while adjusting for confounders, has reported a significant increase in risk associated with heavy drinking (>14 drinks/week in a US population) during the last trimester,26 and with binge drinking during the first, second and third trimesters (≥3 drinks/day in an Italian population).24 Other work has demonstrated reduced birth weight associated with binge drinking (≥4 drinks) in early and late pregnancy.25 No association was found between binge drinking during pregnancy and SGA in studies from Switzerland22 and Australia19 (binge=≥3 and ≥5 drinks, respectively). Our work appears to support the evidence demonstrating an increase in risk of SGA associated with binge drinking during pregnancy. Disagreement between study results may be due to differences in the statistical power, and variation in the definitions of a binge episode and what constitutes an alcoholic drink. Variation in study effect sizes around the mean dose–response relationship between alcohol consumption and SGA4 would also be expected.
Our finding of an association between SGA and binge drinking during the second but not first trimester of pregnancy is difficult to explain. Fetal organ development occurs during the first trimester with an acceleration in fetal growth during the second trimester. It may be that in our study population, cumulative binge exposure during the first and second trimester exposures was required to increase the risk of SGA.
The relationship between preconceptional or periconceptional binge drinking and SGA is less clear. No consistent evidence of a link between prenatal binge drinking and either SGA or low birth weight has emerged from our or other works,19 ,35 although a single study has demonstrated an increased risk of SGA associated with moderate-to-heavy drinkers who binge drink pre-pregnancy.26
Previous studies have found an association between preterm birth and moderate-to-high alcohol consumption19 and >10 drinks/week36 supporting evidence of a dose–response relationship with increasing risk of preterm birth above 1–1.5 drinks/day during pregnancy.4 Our work did not replicate these findings, in common with systematic reviews8 ,21 and research37 reporting no link between either moderate drinking or binge drinking and prematurity.
Smoking was associated with increased risk of SGA and preterm birth, supporting a large body of evidence of its harmful effects.38 In our study, 32% of smokers quit during pregnancy, in line with estimates of 30–50% from other large cohorts.39 In contrast, 87% of pre-pregnancy binge drinkers had quit by the second trimester, equivalent to the quit rates of 86% and 96% in other studies.19 ,26 Our research did not demonstrate any additive effects of alcohol and smoking either because our study was not powered to detect these associations, or our groupings were too broad and a synergistic impact can only be demonstrated at high levels of smoking and drinking.40
Our results should be placed in the context of longer term child outcomes. Here, research has shown little impact of moderate prenatal alcohol consumption on cognitive development of children,21 ,41 or of binge drinking on intelligence in early childhood,42 although an association between >9 drinks/week or >4 drinks on any one occasion during pregnancy and impaired attention in early childhood has been reported.14 ,43 Other risks associated with binge drinking are unplanned pregnancies and an increase in the risk of acquiring sexually transmitted diseases.44
Strengths and limitations
The relative strengths of this study compared with previous works is that we have been able to differentiate between occasional and frequent binge drinking before and during different periods of pregnancy, while adjusting for multiple birth confounders. In addition, our sample is drawn from a population where prenatal binge drinking is not recommended but remains common compared with other locations,19 ,22 ,24 ,25 allowing us to study a relatively large number of second trimester binge drinkers (n=333). We address previous methodological concerns45 by using SGA rather than birth weight to remove the influence of prematurity on growth restriction, adjusting for parity and sex within the SGA calculation, and controlling for age, ethnicity and delivery mode within logistic regression models. This work is generalisable to deprived and predominantly white populations where it is socially acceptable for women to drink alcohol, with a significant proportion continuing during pregnancy. The results are more applicable to North European populations where binge drinking is a common way of consuming alcohol, as opposed to North American and Mediterranean cultures where underreporting may be higher and the acceptability of bingeing lower. Although half of the birth cohort was from Pakistani or Bangladeshi backgrounds where prenatal alcohol consumption is extremely low, we have adjusted for this potential bias within our models.
Some residual confounding is likely to be caused by mental health issues and other life stresses in deprived populations. The BiB study is currently developing a measure of non-psychotic psychological distress in non-clinical populations using the general health question data from women's baseline questionnaires. We hope to use this to control for severity of mental distress in future studies of the BiB cohort. Obstetric history and an unknown aetiology in one of four preterm births46 are further sources of residual confounding. Residual confounding has been suggested as a possible cause of the reported protective effect of alcohol against low birth weight,19 ,21 known as the ‘healthy drinker’ effect where women with poor obstetric history abstain from drinking. Our work supports this explanation as a protective effect of drinking in our unadjusted SGA model disappeared when maternal and other potential confounders were added.
Although we have used an internationally recognised standard of <10% as an SGA threshold, this will result in misclassification because some growth-restricted babies will not be <10% and some <10% will merely be constitutionally small (although the inclusion of maternal height as a covariable should minimise this bias). This effect is difficult to quantify and increases the likelihood of a type II error.
Public health guidance generally recommends abstinence from alcohol during pregnancy, so women may under-report antenatal binge drinking. Reporting bias has been minimised within our study by the use of interview-administered questionnaires during pregnancy rather than postnatal or self-completed questionnaires. This is considered to improve the validity47 and accuracy48 of alcohol exposure measurement. Our alcohol estimates during the pre-pregnancy period may actually include some early pregnancy drinking as 60% of women discover they are pregnant after 6 weeks of gestation.49 Consequently, the real risk of binge drinking prior to pregnancy is likely to be lower than stated.
Within the social science literature, the definition of a binge episode for women varies from four or more drinks to half a bottle of spirits.17 Within obstetric outcome research, definitions are reasonably consistent with typical values of five UK units or 40 g of alcohol (this study), six Australian units or 50 g,19 with a definition of four or more drinks in about 2 hours commonly used for US research of this nature.50
Implications for practice
Alcohol consumption during pregnancy is a contentious topic. There is a case to support a total abstinence public health message, even though there is little evidence of harm associated with moderate drinking. Women may find it hard to assess how much alcohol they have consumed due to varying drink sizes and alcohol concentrations by type of beverage. The message that women should abstain completely from drinking during pregnancy is the safest policy and one that is adopted by most countries.11 Some countries add that if women do drink, they should drink no more than 1–2 units of alcohol once or twice a week and to not binge drink, which appears appropriate, given our evidence and previous research.4 ,8 For the public, this avoids creating alarm over consuming small amounts of alcohol while pregnant,51 supports women's autonomy and allows them to make informed evidence-based choices. Pregnant women require culturally specific guidance on what constitutes a binge episode, supported by clearly defined standard drink thresholds and concordant alcohol labelling.52 ,53
A greater understanding of the clustering of risk factors has been recommended54 in order to support strategies addressing multiple unhealthy behaviours. For example, a quarter of the UK's-adult population have three of four major risk factors for premature mortality (smoking, excess alcohol, poor diet and low-physical activity).54 In our study, of the women who reported binge drinking in early pregnancy, 45% also smoked, and 11% also smoked and were overweight or obese. Targeting interventions at alcohol or smoking in isolation, therefore, ignores the clustering of risk factors within pregnant women. Evidence about how women give up multiple unhealthy behaviours in pregnancy is scarce,55 and further research is required to guide professionals in taking a holistic rather than siloed approach to maternal health.
These findings provide evidence of an increase in risk of SGA births associated with binge drinking (5 units: 40 g of pure alcohol). We found no increased risk of SGA in women who drank low or moderate levels of alcohol during pregnancy. No association was found between any level of alcohol consumption and premature birth. This work supports previous research showing no association between SGA and low-alcohol exposure but adds to evidence of a dose–response relationship with significant risks associated with binge drinking. Clear definitions of what constitutes a binge drinking episode are required for pregnant women.
What is already known on this subject
Estimates of binge drinking during pregnancy vary from 3% to 26% internationally. Animal studies suggest that peak alcohol exposure during a binge episode, rather than total alcohol exposure, may determine fetal development. Research about the impact of binge drinking on the risk of premature or small babies is sparse and inconclusive.
What this study adds
Our study has quantified the risk of binge drinking during different stages of pregnancy on birth outcomes.
It demonstrates a 68% increase in the risk of a small baby for women who binge drink during the second trimester of pregnancy.
This supports evidence of a dose–response relationship between alcohol consumption and the risk of fetal growth restriction.
The authors would like to thank the women and families who took part in this study, the midwives who recruited them and the Born in Bradford team who interviewed them. The authors also acknowledge the help of Professor Gerry Shaper for his comments on earlier drafts of this research output, and Gillian Santorelli and Lesley Fairley of the Born in Bradford team for statistical support.
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Files in this Data Supplement:
- Data supplement 1 - Online appendices
Contributors DLC conceived and led on the design of the study, the analysis of data, interpretation of results, and also drafted and revised the paper. ESP cleaned and extracted the data, monitored the data analysis, and provided input to the design of the study, the statistical analysis plan, statistical software support, interpretation of results and drafts of the paper. JW conceived and initiated the cohort study, monitored all aspects of data collection and cleaning, provided major input to the design of the study, interpretation of results, and redrafting of the paper.
Funding Funded by the National Institute for Health Research Collaboration for Applied Health Research and Care (NIHR CLAHRC).
Competing interests None.
Ethics approval Ethical approval for the data collection was granted by the Bradford Research Ethics Committee (Ref 07/H1302/112). Mothers and fathers who agreed to be in the study filled in a baseline questionnaire and provided written consent for this information to be used for research purposes. Patient consent forms are not required as no personal details or information appear in the manuscript.
Provenance and peer review Not commissioned; externally peer reviewed.