 |
|
 |
|
 |
 |
|
 |
 |
 |
 |
 |
 |
 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
ARTICLE |
Department of Community-Based Medicine, University of Bristol, Bristol, United Kingdom
 |
ABSTRACT |
|---|
 TOP ABSTRACT PATIENTS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
METHODS. This prospective, population-based study used data from the Avon Longitudinal Study of Parents and Children. We investigated the relationship between self-reports of the amount and frequency of alcohol use in the first trimester and the presence of clinically significant mental health (behavioral and emotional) problems at 47 and 81 months (parental report: n = 9086 and 8046, respectively) and at 93 to 108 months (teacher report: n = 5648).
RESULTS. After controlling for a range of prenatal and postnatal factors, the consumption of <1 drink per week during the first trimester was independently associated with clinically significant mental health problems in girls at 47 months. This gender-specific association persisted at 81 months and was confirmed by later teacher ratings.
CONCLUSIONS. Very low levels of alcohol consumption during early pregnancy may have a negative and persistent effect on mental health outcomes. Given the lack of a clear dose-response relationship and unexpected gender effects, these findings should be considered preliminary and need additional investigation.
Key Words: pregnancy • prenatal alcohol exposure • alcohol drinking • mental health problems • fetal alcohol spectrum disorder • ALSPAC
Abbreviations: ALSPAC—Avon Longitudinal Study of Parents and Children • SDQ—Strengths and Difficulties Questionnaire • EPDS—Edinburgh Postnatal Depression Scale • OR—odds ratio • CI—confidence interval
Alcohol consumption during pregnancy is a major public health concern because of potential adverse long-term physical, neurodevelopmental, and behavioral consequences for children.1 There is robust evidence of an etiologic role involving high levels of alcohol consumption, characterized at the extreme end as fetal alcohol syndrome.2 Moderate levels (1–2 drinks per day) of alcohol consumption during pregnancy are also associated with childhood cognitive, learning, attentional, and behavioral problems.3–10 There are conflicting reports as to whether lower levels of alcohol consumption are independently associated with adverse mental health (behavioral and emotional) or cognitive outcomes in exposed children, which raises the question of whether there is a safe level of drinking during pregnancy.11–13 Policy recommendations in many countries have become more cautious in recent years, suggesting abstinence throughout pregnancy. However, the extent to which these recommendations are based on evidence remains uncertain.
As the majority of women of childbearing age in industrialized countries drink alcohol, the fetus is most likely to be exposed during the first trimester, especially before pregnancy recognition.14 Moderate alcohol consumption during this early stage, either reflecting a critical period for neurodevelopment or acting as a marker of the duration of drinking across pregnancy, is a risk factor for behavioral and learning problems.3,4 We have used prospective data from a population-based cohort of mothers and children to investigate whether there is any independent effect of very low levels of alcohol consumption (<1 drink per week) during early pregnancy on childhood mental health outcomes. Because behavioral problems are more common in boys, the large sample size also provided a unique opportunity to investigate the possibility of gender differences in vulnerability to prenatal alcohol exposure. We expected that only higher levels of alcohol consumption would be associated with later mental health problems and that any associations might be more readily detectable in boys.
|
PATIENTS AND METHODS |
|---|
|
TOPABSTRACTPATIENTS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Detailed self-report information has been obtained from questionnaires completed by the mothers at regular intervals during the pregnancy and since the birth (additional details are available at www.alspac.bris.ac.uk). From the cohort, a 10% randomly selected subsample of parents whose infants were born within the last 6 months of the survey were invited to bring their children to a research clinic (the Children in Focus group) at regular intervals, where a number of more detailed assessments were conducted.
Ethical permission for the ALSPAC was granted by the research ethics committees of the local health care trusts (United Bristol Healthcare and Frenchay and Southmead National Health Service Trusts), and the study was also monitored by the ALSPAC Ethics and Law Advisory Committee.
Measures
Exposure Variables
Information on the mother's alcohol consumption before the pregnancy and during the first trimester was collected by postal questionnaire at 18 weeks' gestation. The mother was asked about her frequency of drinking alcoholic drinks. Response categories were never, <1 glass per week, 
1 glass per week, 1 to 2 glasses a day, 3 to 9 glasses a day, or >10 glasses a day. Examples were provided to specify that 1 glass was equivalent to 1 unit (8 g) of alcohol. For the analyses, the groups consuming 1 glass per week were combined.
Outcome Variables
Child mental health outcomes were measured using the Strengths and Difficulties Questionnaire (SDQ), a well-validated measure of childhood mental health.16,17 It has 4 symptom scales relating to hyperactivity/inattention, conduct problems, emotional symptoms, and peer relationships (which are summed to provide a total score), as well as a prosocial behavior scale. In keeping with other ALSPAC investigations of childhood psychopathology, we used validated cutoffs for the total score that identify the top 10% of the population.16,18 These cutoffs have been compared against a diagnostic interview in a large, nationally representative community sample and are highly associated with an independently assessed psychiatric diagnosis.17 Parent SDQs were obtained at 47 and 81 months and teacher SDQs at 93 to 108 months.
Confounder Variables
Potential confounding factors associated with alcohol consumption and mental health problems that were measured in the ALSPAC were included in this study. Maternal and sociodemographic factors collected during pregnancy were categorized for analysis as follows: maternal age (
20, 21–34, or 35 years); parity (none or 1); maternal smoking and use of cannabis and other illicit drugs in the first trimester (all yes or no); highest level of maternal education (dichotomized to university degree or not); housing tenure (home ownership or not); and whether the mother was currently married. The dichotomized smoking data described above were based on maternal report of the number of times per day she smoked during the first trimester (response categories were 0, 1–4, 5–9, 10–14, 15–19, 20–24, 25–29, and 30 times). Because prenatal smoking is a known risk factor for childhood behavioral problems, the multivariable analyses (see "Analysis" below) were repeated using this ordinal scale of smoking as a confounder variable.
Maternal mental health was measured at 18 weeks' gestation and when the child was 33 months old using the well-validated Edinburgh Postnatal Depression Scale (EPDS).19 High scores (>12) are highly associated with a diagnosis of a depressive disorder.20 Child factors included gestational age (36 or 37 weeks), birth weight, gender, and ethnicity. Information on the mother's postnatal alcohol consumption (during the previous week) was obtained when the child was aged 47 months.
Although cognitive outcomes were not the focus of this study, full scale IQ data at 49 months were available from the Children in Focus group, measured using the Wechsler Preschool and Primary Scale of Intelligence-Revised United Kingdom Edition.21
Analysis
The main focus of the analyses was the relationship between alcohol consumption (exposure) in the first trimester and child outcomes at 47 months. The analyses were repeated to explore longer-term outcomes (81 months) and replication of findings using ratings from another informant (teacher). Hence, all of the available SDQ data at each time point were used.
The analyses were based on children from singleton births (to minimize clustering effects) alive at 1 year of age (n = 13617) and consisted of several stages. First, using 
2 tests, we investigated whether SDQ response status was associated with prenatal alcohol consumption (exposure), other maternal factors, and SDQ scores at the earlier time points. The associations between the exposure and possible confounding factors were then examined. Second, the relationship between prepregnancy alcohol use (less versus 1 glass per week) and high total SDQ scores at 47 months was investigated. Third, the relationship between prenatal alcohol consumption (comparing, in turn, less than and 1 glass per week against none [baseline]) and high total SDQ scores was assessed for the whole sample and then by gender, using logistic regression analyses to provide odds ratio (OR) estimates. Likelihood ratio tests were used to examine possible interactions between gender and alcohol exposure. Fourth, multivariable logistic regression analyses, controlling for the confounder variables outlined above, provided adjusted OR estimates. Fifth, sensitivity analyses were performed for the subsample with complete data at 47 months to investigate the effects of missing data on the findings. For 47-month outcomes, the multivariable logistic regression analyses were repeated with the addition of the maternal EPDS at 33 months and alcohol consumption (yes or no) at 47 months. Hence, this controlled for the effects of key risk factors in the postnatal environment.
Finally, 3 sets of exploratory analyses were conducted to assess whether our choice of outcome or confounder variables might have influenced the findings. First, because higher levels of alcohol consumption during pregnancy have been reported to be more strongly associated with behavioral than emotional problems and the use of global total SDQ scores might obscure the existence of any dose-response association, we examined the relationships between levels of alcohol use and the 2 SDQ behavioral problem subscales (conduct problems and hyperactivity) at all 3 of the time points. As with the total score, this involved validated cutoffs to identify the top 10% of the population.16,18 Second, in the Children in Focus subsample, we explored whether there was any association between levels of alcohol use and full-scale IQ scores (using 1-way analysis of variance). Third, because there is recent evidence suggesting that the risk for fetal alcohol spectrum disorders is greater in older mothers,8,22 we also examined whether there was any interaction between alcohol consumption and older maternal age in relation to child SDQ scores (this was done twice, using cutoffs set at 30 and 35 years of age).
|
RESULTS |
|---|
|
TOPABSTRACTPATIENTS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
2 = 18.28; P < .001) and at 81 months (66% vs 62%; 2 = 26.28; P < .001). Teacher SDQ response rates (45%; n = 5648) were not associated with maternal alcohol consumption. Maternal correlates of nonresponse at 47 months included lower age, higher parity, smoking, use of cannabis and other illicit drugs, depression, being unmarried, rented tenure, and lower level of education. Nonresponse at each later time point was associated with high total SDQ scores at the previous point. However, the SDQ scale score distributions were in keeping with the expected population distributions (www.sdqinfo.com).
Correlates of Alcohol Consumption in the First Trimester
Among children on whom SDQs were available at 47 months, 44% (4038 of 9086) of mothers had not consumed any alcohol in the first trimester, 40% (3626 of 9086) consumed <1 glass per week, and 16% (1422 of 9086) consumed 1 glass per week. Table 1 shows that maternal correlates of alcohol consumption included higher age, parity, and level of education; smoking; use of cannabis and other illicit drugs; depression; and being unmarried.
|
1 glass per week before the pregnancy. This prepregnancy alcohol use was not associated with high total SDQ scores at 47 months. Similarly, after controlling for potential confounders, as well as gender, alcohol consumption in the first trimester was not associated with high total parent SDQ scores. In contrast, in the multivariable logistic regression analysis, <1 glass per week in the first trimester was associated with high total teacher SDQ scores (adjusted OR: 1.36; 95% confidence interval [CI]: 1.04–1.78; P = .025). There was an interaction (P = .02) between gender and alcohol exposure in relation to 81-month outcomes, and findings are reported by gender in Table 2. After controlling for confounders, <1 glass per week in the first trimester was associated with high total SDQ scores in girls at all 3 of the time points.
|
Sensitivity Analyses
Sensitivity analyses exploring the effects of missing data at 47 months found a slight increase in the strength of the bivariate association between <1 glass per week and high total SDQ scores in girls (OR: 1.45; 95% CI: 1.03–2.04; P = .035). After controlling for postnatal factors (maternal EPDS at 33 months and alcohol consumption at 47 months) in the full multivariable analyses, the association with higher scores in girls remained (OR: 1.45; 95% CI: 1.01–2.10; P = .045).
To assess whether the gender-specific findings reflected a severity effect because of a lower proportion of girls than boys having high total SDQ scores, the analyses were repeated using a stricter SDQ cutoff that identified the top 5% of each gender. This involved raising the cutoffs for boys and lowering these for girls. In the whole sample, after controlling for potential confounders, as well as gender, 1 glass per week in the first trimester was now associated with high total parent SDQ scores at 47 months (OR: 1.45; 95% CI: 1.08–1.94; P = .013). This association was only apparent in boys, suggesting some support for a severity effect influencing the findings with regard to higher levels of drinking (dose-response association). The strength of the association involving <1 glass per week only increased with regard to high total teacher SDQ scores (OR: 1.64; 95% CI: 1.18–2.29; P = .003). Even with this stricter cutoff, the association with teacher ratings was only statistically significant in girls. Collectively, these findings suggest that the effects of low levels of alcohol consumption in pregnancy are more apparent in children with severe mental health problems. The interaction between gender and alcohol exposure was also greater at both 47 (2 = 5.29; P = .07) and 81 (2 = 13.09; P = .001) months, indicating that associations between alcohol exposure and high SDQ scores were more prominent in girls.
Relationship Between Alcohol Consumption and High Behavioral Problem Scores
To explore whether the use of total SDQ scores obscured the existence of any dose-response association, we examined the relationships between levels of alcohol use and high conduct problem and hyperactivity scores at all 3 of the time points (Table 3). These data relate to the whole sample, because there was no interaction between gender and alcohol exposure in relation to these subscales. The findings involving conduct problems suggest the presence of some dose-response relationship on the univariable analyses, although most associations disappear after controlling for confounders.
|
1 glass per week (1-way analysis of variance, F = 1.89; P = .15). This suggests that the association between low-level alcohol intake in pregnancy and mental health effects in the child was not contingent on lower IQ.
Role of Other Possible Effect Modifiers
Finally, we examined whether there was any interaction between alcohol exposure and older maternal age in relation to child SDQ scores. There was no statistically significant evidence of an interaction when the cutoff was set at either 30 or 35 years of age.
|
DISCUSSION |
|---|
|
TOPABSTRACTPATIENTS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Gender Effects
The association between prenatal alcohol exposure and mental health problems was more readily demonstrable in girls. The later teacher ratings also suggested that girls seem to be more vulnerable than boys to the effects of low levels of alcohol in pregnancy. Notably, there was also evidence of some effect in boys but only at higher cutoffs for severity and with higher levels of drinking. This dose-response effect is more expected.13,22 Previous work involving clinically referred children with attention-deficit/hyperactivity disorder has not found evidence of any gender modification.23 However, other studies have found that maternal smoking in pregnancy is only associated with conduct problems in boys.24,25 Gender-specific differences have also been reported for the patterns by which prenatal alcohol exposure moderates the effect of prenatal cocaine use on child behavioral outcomes.26
Animal studies have demonstrated gender differences in the effects of prenatal alcohol exposure on responsiveness to stress, suggesting greater vulnerability among female offspring.27 Nevertheless, the gender differences found in the present study are unexpected and may be a chance finding. Further replication is required before any firm conclusions can be drawn. Mental health (especially behavioral and neurodevelopmental) disorders are more common in boys, suggesting that they might be developmentally more vulnerable to the range of risk factors for childhood psychopathology. There is also less variance in girls' total scores, and it is possible that the small additional risk related to low levels of alcohol exposure in pregnancy is more readily demonstrable in girls.
Methodologic Issues
The ALSPAC data set is very well placed to overcome the methodologic limitations affecting many previous studies,12 such as the use of high-risk samples that do not readily generalize to the wider population and limited information on antecedent and contemporaneous confounder variables (eg, maternal mental health). This is the largest study investigating the effects of prenatal alcohol use on child mental health outcomes, enabling gender-specific analyses and the detection of possible effects involving <1 glass per week. The outcome measure was collected from 2 different raters over 3 time points. The multivariable analysis approach attempted to control for a large number of confounding factors to provide an indication of the independent risk from low-level alcohol consumption. Although the findings are broadly generalizable to United Kingdom populations in geographical regions with both urban and rural areas, they may not apply as closely to deprived inner-city populations that are vulnerable to multiple and overlapping risk factors.
By using prospectively collected data, the risk of recall bias relating to the exposure is minimized. Although we relied on self-reported alcohol consumption, the validity of such self-reports with regard to neurobehavioral outcomes has been demonstrated.7 These exposure data were collected 15 years ago when there might have been a higher prevalence of alcohol consumption during pregnancy and less stigma about reporting it. The reported rates are similar to US data from the same period.14 However, it is possible that the desire to give socially acceptable responses might have contributed to underreporting. Although systematic misclassification of low levels of alcohol use to "none" might have underestimated the strength of the findings, it is more likely that possible systematic misclassification of high levels of alcohol use as low levels may have increased the power to demonstrate a difference between self-reported "low" levels of use and abstinence. Hence, the reported association might reflect the effects of higher levels or greater duration of alcohol use.
The relative absence of dose-response findings may reflect several other methodologic factors. There was no dose-response relationship when using the conventional SDQ cutoffs as the outcome measure, but a dose-response effect was demonstrated when the SDQ severity cutoff was raised so as to include the top 5% of each gender. Similarly, on univariable analyses, there were dose-response associations between levels of alcohol consumption and conduct problems at all 3 of the time points. Hence, the findings may reflect the way in which the SDQ was used as the outcome measure. It is also possible that the SDQ may only be acting as a marker of the key adverse neurodevelopmental outcome. Similarly, the exposure was assessed during early pregnancy, and this may also only be a marker of the key risk, for example, consumption later in pregnancy, duration and amount of drinking across the whole pregnancy, or intermittent binge drinking. Finally, as described in relation to the gender findings, these might be chance findings.
There are 3 other limitations of this study. First, we relied on exposure data from the first trimester only. We could not distinguish between drinking before and after pregnancy recognition, and these may have different correlates. As well as considering the frequency and amount of alcohol consumption, future work should also consider the pattern and duration of intake, because these are also associated with adverse outcomes.3,11,22 Second, in keeping with large longitudinal studies, there was sample attrition over time, and nonresponse bias may have influenced the findings. For example, SDQ nonresponse was associated with higher levels of alcohol use, as well as other risk factors for child mental health problems. However, alcohol use in pregnancy was not associated with reduced SDQ availability. Third, although we controlled for a large number of confounders (including other prenatal exposures, indicators of socioeconomic status, and child and postnatal factors), it is possible that the findings reflect residual confounding. For example, the association may disappear when controlling for familial risk for psychiatric disorders.28 Because our findings only apply to children with high levels of mental health problems, they may reflect a gene-environment correlation involving maternal propensity to drink during pregnancy and the child's predisposition to developing a psychiatric disorder. It is also possible that genetic factors moderate the effects of exposure to alcohol. For example, recent work has suggested that the association between a haplotype of the dopamine transporter gene and attention-deficit/hyperactivity disorder only occurred in clinically referred children whose mothers had drunk during the pregnancy.29
Clinical Implications
At a population level, public health educational approaches are required to reduce the proportion of pregnant women consuming alcohol. Currently, there are international differences in advice given to pregnant women. For example, guidance in the United Kingdom suggests that 4 drinks per week is acceptable, whereas in the United States, there is a public health warning advising abstinence from alcohol, and universal screening of alcohol use during pregnancy has been recommended.1,30 Because this is a modifiable risk factor with ongoing uncertainty about a safe threshold, there is an opportunity for all clinicians in routine contact with pregnant women to assess alcohol consumption and recommend abstinence. Similarly, clinicians involved in assessing clinically significant childhood mental health problems should routinely enquire about exposure to alcohol during the pregnancy. With higher levels of alcohol exposure in utero, there is now increasing evidence of the persistence of severe mental health problems into adulthood,31–32 which highlights the importance of improved identification and secondary prevention efforts during childhood.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTPATIENTS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
We are extremely grateful to all of the families who took part, the midwives for help in recruiting them, and the whole ALSPAC team, which includes interviewers, computer and laboratory technicians, clerical workers, research scientists, volunteers, managers, receptionists, and nurses.
|
FOOTNOTES |
|---|
Address correspondence to Kapil Sayal, PhD, Centre for Child and Adolescent Health, Department of Community-Based Medicine, University of Bristol, Hampton House, Cotham Hill, Bristol BS6 6JS, United Kingdom. E-mail: kapil.sayal{at}bristol.ac.uk
The authors have indicated they have no financial relationships relevant to this article to disclose.
This article is the work of the authors, and Dr Sayal will serve as the guarantor for its contents.
|
REFERENCES |
|---|
|
TOPABSTRACTPATIENTS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  H. A. Ricciotti State of the Art Reviews: Nutrition and Lifestyle for a Healthy Pregnancy American Journal of Lifestyle Medicine, April 1, 2008; 2(2): 151 - 158. [Abstract] [PDF]
|
|
|
|
 |
|
 K. Sayal Alcohol consumption in pregnancy as a risk factor for later mental health problems Evid. Based Ment. Health, November 1, 2007; 10(4): 98 - 100. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
||
|
||
Home | My Pediatrics | Journal Information | Current Issue | Past Issues | Subscriptions & Services | Contact Us Click here for faster international access © 2007 American Academy of Pediatrics. All rights reserved. |
||
|
||
