Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Causal Inferences Regarding Prenatal Alcohol Exposure and Childhood Externalizing Problems

Causal Inferences Regarding Prenatal Alcohol Exposure and Childhood Externalizing Problems Abstract Context Existing research on the neurobehavioral consequences of prenatal alcohol exposure (PAE) has not adequately accounted for genetic and environmental confounds. Objective To examine the association between PAE and offspring externalizing problems in a large representative sample of families in the United States using measured covariates and a quasi-experimental design to account for unmeasured genetic and environmental confounds. Design This study combines information from the National Longitudinal Survey of Youth and the Children of the National Longitudinal Survey of Youth. The analyses statistically controlled for measured characteristics of the mothers and families and exposure to other prenatal psychoactive substances. In the primary analyses, siblings differentially exposed to prenatal alcohol were compared. Setting and Participants Women were recruited from the community using a stratified and clustered probability sample and were followed longitudinally. The sample included 8621 offspring of 4912 mothers. Main Outcome Measures Maternal report of conduct problems (CPs) and attention/impulsivity problems (AIPs) during childhood (ages 4-11 years) using standardized assessments related to psychiatric diagnoses. Results There was an association between PAE and offspring CPs that was independent of confounded genetic and fixed environmental effects and the measured covariates. The CPs in children of mothers who drank daily during pregnancy were 0.35 SD greater than those in children whose mothers never drank during pregnancy. Although AIPs were associated with PAE when comparing unrelated offspring, children whose mothers drank more frequently during pregnancy did not have more AIPs than siblings who were less exposed to alcohol in utero. Additional subsample analyses suggested that maternal polysubstance use during pregnancy may account for the associations between PAE and AIPs. Conclusion These results are consistent with PAE exerting an environmentally mediated causal effect on childhood CPs, but the relation between PAE and AIPs is more likely to be caused by other factors correlated with maternal drinking during pregnancy. Prenatal alcohol exposure (PAE) has been associated with offspring externalizing problems, such as conduct problems (CPs),1 criminality,2 attention/impulsivity problems (AIPs),3,4 and alcohol disorders.5-8 A recent review of the literature9 and a meta-analysis,10 however, have raised questions about the strength of some of the associations. Given the current state of research on prenatal drug exposure, Huizink and Mulder11 recently highlighted important challenges for the research community. In particular, they noted the need to document deleterious outcomes associated with moderate levels of alcohol exposure, explore the adjustment of offspring across time, and delineate the effects of exposure to multiple drugs. Research also needs to adequately control for covariates that may confound the effects of PAE, such as family processes (eg, problematic parenting or family conflict) and parental characteristics, especially maternal substance use.1,12 Researchers also need to account for genetic liabilities that are shared by parents and offspring.13,14 A woman's genetic risk of substance use could be passed down to her children and subsequently affect their behavior.15 Research on the consequences of PAE, therefore, needs studies with large samples, with sufficient statistical power to detect small effects, using analytical methods and designs that can account for potential confounds, including factors that are not measured. Such studies could complement smaller, more intensive studies6,7,13 and provide the opportunity to find convergent evidence from multiple research approaches.16 Because of the myriad of variables associated with prenatal substance exposure, quasi-experimental designs are useful because they can test alternative causal inferences.16 Quasi-experimental designs rely on natural experiments that can pull apart processes that are typically confounded. Genetically informed approaches, one class of quasi-experimental designs, are important because they can separate the co-occurring genetic and environmental processes. The use of quasi-experimental designs, therefore, can help the study of putative environments, such as PAE, move from merely identifying risk factors to highlighting the underlying causal mechanisms of externalizing problems.17 Recently, a variety of quasi-experimental and genetically informed studies related to prenatal substance exposure have been reported. Two studies used the children-of-twins design18-20 to explore smoking during pregnancy (SDP), finding an independent association between SDP and birth weight21 and attention-deficit/hyperactivity disorder (ADHD)14 when shared genetic and environmental liability were taken into account. Sibling comparison studies have also been used to study SDP. The approach contrasts children exposed to more prenatal nicotine to their siblings exposed to less, a design that controls for characteristics of the mothers, including genetic risk associated with SDP. The results suggest that independent associations between SDP and psychosocial adjustment are either greatly reduced or absent.22,23 Only 1 quasi-experimental study, however, has been conducted on the consequences of PAE. Knopik et al14 found that PAE was not independently associated with offspring ADHD in a children-of-twins study. Given the somewhat contradictory findings of the association between PAE and offspring attention problems9 and the results of the 1 genetically informed study,14 a quasi-experimental study of the consequences of PAE on offspring externalizing problems is of critical importance. Inferences about PAE affect public health campaigns and prevention efforts24 in addition to informing theories about the etiology of CPs.17 The need for more rigorous studies of PAE is further supported by recent prevalence estimates that indicate that alcohol use during pregnancy is still relatively common.25 The overall aim of the present article is to explore whether PAE is related to offspring externalizing behaviors and, if so, whether the exposure is consistent with a causal association. The present study has 6 goals. First, these analyses explore the relation between PAE and characteristics of the mothers and families. Second, the study documents the statistical relation between PAE and child externalizing problems. Third, the analyses explore whether measured covariates account for associations with PAE. Fourth, the analyses compare siblings and cousins differentially exposed to prenatal alcohol to account for unmeasured confounds, including genetic and environmental factors. If PAE causes externalizing problems, the association will be found at all levels, especially within mothers (ie, siblings exposed to higher levels of prenatal alcohol would have more problems than their siblings exposed to lower levels).26 The comparison of siblings explicitly controls for maternal genetic confounds because PAE is not correlated with genetic risk within mothers. Fifth, the analyses investigate whether the relation was robust to the quasi-experimental approach and statistical controls. Finally, an analysis of a subsample of the children that includes measures of prenatal exposure to marijuana and cocaine further explores maternal polysubstance use. Ultimately, an environmentally mediated causal effect of PAE on childhood externalizing behaviors—a causal inference—would be supported if the relation remained robust to the measured covariates and quasi-experimental methods because these approaches account for many of the alternative explanations for the association between PAE and externalizing problems. This research was approved by the institutional review board at Indiana University. Methods Sample Mother Generation Sample The National Longitudinal Survey of Youth (NLSY79) is a longitudinal study of adolescents and young adults funded by the Bureau of Labor Statistics. The initial NLSY79 sample included a nationally representative sample of 6111 youths and an oversample of 3652 African American and Hispanic youth. The initial sample of youth used a stratified and clustered probability sample, which included all siblings in each household in the age range. The present study was conducted using 4912 females from the combined sample who had a child who was at least 4 years old by the 2004 survey. The initial NLSY79 assessment had a response rate of 90%. Participants were assessed annually between 1979 and 1994 and biennially since then. Retention rates were 90% or better during the first 16 waves of data collection. Probability weights for the sample are available for the sample of mothers, so the analyses could be weighted to produce population-based estimates. More details concerning the NLSY79 are available elsewhere.27 The characteristics of the females with children are given in Table 1. The sample was racially diverse; based on mothers' self-reports, 17.1% were Hispanic, 25.9% were African American, and 57.1% were non-Hispanic white. Offspring Generation Sample Starting in 1986, biennial assessments of the offspring of the women in the NLSY79 sample were performed, referred to as the Children of the National Longitudinal Survey of Youth (CNLSY) sample.28 The initial response rate was 95%, with subsequent assessments averaging 90%. Although 3 types of interviews are available on the offspring, the present project uses maternal report of the children's characteristics assessed between ages 4 and 11 years (n = 8621 children with reports of PAE). Studies using the offspring of the NLSY79 mothers have been plagued by selection bias because until all childbearing of the mothers is complete, offspring available for assessment were necessarily born to the younger mothers. Given that the overwhelming majority of the childbearing was completed by the 2004 survey and we use assessments of young children, the selection bias for the present analyses is relatively small (B.M.D., J. A. Goodnight, MA, C.A.V.H., I.D.W., J.L.R., P.J.R., and B.B.L., unpublished data, August 2007). Measures Prenatal Exposure to Psychoactive Substances Beginning in 1983, the NLSY79 females reported their prenatal substance use, including alcohol consumption and cigarette use, during each of their pregnancies. Most of the assessments of prenatal substance use reports were completed when the children were young, mostly within 2 years of birth.22Table 2 includes the response categories for alcohol and conversions to days per week for the sample of offspring with information on externalizing problems. A previous article22 analyzed the consequences of SDP. Owing to the small number of pregnancies during which the mothers reported drinking every day, the last 2 categories were collapsed. There are missing data for the prenatal substance use section. Most of the missing PAE data (92%) were due to an incorrect skip in the assessment protocol. Sensitivity analyses (replacing the missing values with zeros, the mean PAE for the mothers, or estimates based on multiple imputation) revealed that the missing values for PAE did not alter the results presented herein, which are based only on offspring with reports of PAE. The maternal assessment started to measure marijuana and cocaine use during pregnancy in the 1986 assessment. Therefore, the variables were available only for 3977 offspring with data on externalizing behaviors. Because of the low prevalence, exposure was coded as exposed or not for each drug. In the subsample, 1.6% of children were exposed to prenatal marijuana and 0.6% to cocaine. Maternal and Familial Characteristics Available characteristics of the mothers and families that could confound the effects of PAE were included in the analyses. Table 1 provides the distribution of the items. When the eventual mothers of the CNLSY sample were aged 15 to 22 years they were asked about their participation in 12 delinquent behaviors using the Self-Reported Delinquency Interview,29 a reliable and valid measure used in contemporary delinquency research.30-32 Symptom counts were regressed on the mother's age at which she completed the survey and were converted to a z score for ease of interpretability. Maternal age at the birth of her first child was also obtained, a variable that predicts offspring delinquency in the CNLSY (B.M.D., J. A. Goodnight, MA, C.A.V.H., I.D.W., J.L.R., P.J.R., and B.B.L., unpublished data, August 2007).33 Net family income was reported for mothers at age 30 years in 1986 inflation-adjusted dollars, a measure that includes income from all adults in the household at that time. Total number of years of schooling was also assessed for each mother. A composite score of the Armed Services Vocational Aptitude Battery was used as a measure of the mother's intellectual ability, which was converted to a z score. The 1994 assessment included an extensive assessment of the mother's alcohol use and alcohol problems (during her lifetime and not limited to her pregnancies). These questions assessed alcohol frequency (days per month), quantity (drinks per day), and the number of binge episodes in the past month, and 25 items assessed lifetime symptoms of alcohol abuse and dependence. A total of 32% of the women also reported having children by more than 1 man. Offspring Childhood Externalizing Problems Mothers rated their 4- to 11-year-old children on the Behavior Problem Index at each wave of assessment. The Behavior Problem Index included 13 items from the Child Behavior Checklist34 that had the strongest associations with Child Behavior Checklist factor scores.35 Mothers rated each child using a 3-point scale. Standard diagnostic approaches36 and previous factor analyses22 revealed that the items load on 3 factors: CPs, oppositional problems, and AIPs. The association between PAE and CPs and AIPs is explored herein. Trajectory analyses on the CNLSY sample36 indicated that the average number of CPs during ages 4 to 11 years was stable and represents a good measure of severity of externalizing problems reported by adolescents. Therefore, mean CP scores across ages 4 to 11 years were calculated by taking the mean of z scores for CPs at each assessed year of age, standardized within each age. The CPs at each age were highly correlated with the mean across ages 4 to 11 years (r = 0.74-0.80). The items assessing CPs used in the CNLSY also overlap substantially with those used in previous population-based longitudinal studies.32,37 The mean of z scores for AIPs was also stable and exhibited good external validity. The AIP scores at each age were highly correlated with the mean score across ages 4 to 11 years (r > 0.80 for each). Mean AIPs were also strongly related to standardized assessments of academic achievement performed at age 10 years, even controlling for CPs (βmath = −.17, P < .001; βreading = −.17, P < .001; and βword recognition = −.19, P < .001). Moreover, the AIP items are similar to those used in several studies that yielded results similar to studies that used full measures of ADHD consistent with the DSM-III-R.38 Statistical analyses Correlations were calculated among the maternal covariates to estimate the associations among PAE and maternal and familial characteristics to explore what factors predicted or were associated with PAE. All of the analyses in the present article were weighted to produce estimates from a representative sample of women in the United States. Hierarchical linear models (HLMs) were used to explore and characterize the associations between PAE and offspring externalizing behaviors because of the nested nature of the data.39 When the NLSY79 and CNLSY samples are combined, there are 3 levels of analysis: the offspring, maternal, and NLSY79 household levels.22 Five primary HLMs were fit to the associations between PAE and each measure of offspring externalizing problems. Model 1 calculated the unadjusted relation between alcohol exposure, measured in mean number of days exposed per week, and each dimension of externalizing behavior, controlling for offspring sex. The analyses included 8621 offspring of 4912 mothers. Although previous analyses suggested that offspring sex,6,40 maternal SDP,41 and family race/ethnicity42 may moderate the associations, the factors did not affect the magnitude of the relations between PAE and CPs or AIPs in the present sample. The analyses also revealed that linear models provided the best fit for CPs and AIPs because the parameters associated with a quadratic effect were small and statistically nonsignificant. Model 2 was fit to the relation between PAE and offspring externalizing behaviors controlling for measured covariates that could confound the association (listed in the “Measures” subsection of the “Methods” section). Model 3 compared siblings differentially exposed to prenatal alcohol (ie, the mother drank alcohol during one pregnancy and not another or drank more during one pregnancy than another). The within-mother association (ie, comparisons of the different offspring of each mother) provides a more stringent test of causality because the comparison controls for unmeasured genetic and environmental confounds that vary between unrelated mothers. The analyses were based on a subset of the families that included children with differing amounts of exposure (3447 offspring from 1258 mothers). Siblings were compared using contrast codes. Each offspring was assigned a value equal to the deviation of the individual measure of PAE from the mean PAE of all the children from the same mother. The contrast codes, therefore, reflect the amount of PAE relative (more or less) to the mean PAE of all the children in the family. The model also included comparisons of cousins born to NLSY79 mothers who are sisters. Although the comparison of cousins is not as strong as a within-mother approach, it can help provide converging evidence concerning causal processes.43 Cousins were also compared using contrast codes; cousins with more PAE, on average, were compared with their cousins with less PAE, on average. The model also compared unrelated individuals (mean PAE in all cousins from one extended family compared with mean PAE in cousins in a separate extended family). More details about how the contrast codes were calculated,22,43 and the statistical justification for their use,44 can be found elsewhere. Model 4 combined the use of measured covariates and the sibling and cousin comparisons, thus using statistical and methodologic controls for potential confounds. Model 5 tested whether PAE was associated with externalizing problems in the subsample of children that included information on prenatal exposure to marijuana and cocaine. The HLMs included prenatal exposure to alcohol, nicotine, marijuana, and cocaine and the sex of the offspring. Because the assessment of all the substances was available only on a subset of the sample, sibling comparisons could not be completed. Multiple imputation was used for all of the HLMs that included maternal and familial covariates to account for missing values. The analyses, therefore, use all available data rather than relying on families with complete maternal data. Given the assumption that the missing values are missing at random or completely at random, multiple imputation yields standard errors that account for uncertainty due to missing values and avoids bias arising from complete-record analysis.45 Unstandardized regression parameters were used because of the difficulty comparing standardized coefficients when exploring causal processes,46 and measurement of PAE is in meaningful units (days per week). Results Associations among mean alcohol use during pregnancy and maternal and familial characteristics Table 3 provides the associations among mean alcohol use during pregnancy and maternal and familial characteristics. Mean PAE across all pregnancies was slightly correlated with maternal delinquency and age at first childbirth. Mean PAE was more strongly related to substance use measures, including mean levels of SDP, alcohol frequency and quantity, and alcohol abuse and dependence items. Binge drinking Δ(Mbinge − Mno binge) = 0.20, where M represents the mean; P < .001) and having children with different biological fathers Δ(Mnot blended − Mblended) = 0.08; P < .001) were also associated with PAE. ASSOCIATION BETWEEN PAE AND OFFSPRING CPs The results of the HLMs exploring the association between PAE and offspring CPs are given in Table 4. Model 1 indicates that for each additional day of PAE per week, offspring had an increase of 0.09 SD in CPs (P < .001). Model 2 explored the association of PAE and CPs controlling for the main effects of race/ethnicity, prenatal nicotine exposure, and maternal covariates. Including the measured covariates slightly reduced the association between PAE and offspring CPs (b = 0.06, where b represents unstandardized regression weight; P < .001) compared with model 1, but the model still revealed a positive and statistically significant relation. The Figure presents the unadjusted and adjusted means for CPs at each level of PAE. The adjusted means are based on least squares estimates after controlling for the measured covariates. The Figure illustrates a positive association between PAE and CPs for the entire sample; children exposed to more prenatal alcohol had more CPs than unrelated children with less PAE, even after controlling for the measured covariates. Model 3 provided a statistical comparison of siblings differentially exposed to prenatal alcohol. The within-mother association with PAE was still significant (b = 0.05; P = .03) and was commensurate with findings from the comparison of cousins and unrelated individuals in the model. The last model included the measured covariates, and the statistical relation with PAE remained within mothers (b = 0.05; P = .04). The cousin comparisons in models 3 and 4 also suggests that PAE is independently associated with offspring CPs. The Figure also provides an estimate of the association between PAE and CPs based on the parameters from model 4. Children with more PAE have more CPs than their siblings with less PAE while controlling for the measured factors. The results of all the models are consequently consistent with a causal association between PAE and offspring CPs. ASSOCIATION BETWEEN PAE AND AIPs The results of the HLMs for AIPs are given in Table 5. Model 1 found an association between PAE and AIPs (b = 0.09 SD per days exposed per week). Model 2 controlled for measured covariates, and the statistical association with PAE remained (b = 0.06; P = .001). The results of model 3, the comparison of siblings (b = 0.03; P = .18), and model 4, the comparison of siblings controlling for measured covariates (b = 0.02; P = .24), revealed greatly reduced and statistically nonsignificant associations. The comparison of cousins in model 4 also revealed no association, further suggesting that the statistical association between PAE and AIPs is not causal. The association between PAE and AIPs, therefore, seems to be due primarily to characteristics related to PAE that were not included in the previous analytical models. Analysis of pae controlling for maternal polysubstance use during pregnancy Additional HLMs of the sample of children with data on exposure to prenatal alcohol, nicotine, cocaine, and marijuana were conducted. The results suggested that PAE was independently associated with CPs; the magnitude of the estimate (b ± SE = 0.04 ± 0.03; P = .16) was commensurate with the results of the full sample, but the parameter could not be precisely estimated in the smaller subsample. Consistent with the sibling comparisons, statistically controlling for all of the measures of prenatal substance exposure resulted in no association between PAE and AIPs (b ± SE = −0.01 ± 0.15; P = .93). Comment The structure of relationships within families in this sample provides a powerful natural laboratory for quasi-experimental designs of environmental risk factors.16,26 The results of the present study indicate that there is an association between PAE and CPs that was not due to genetic factors related to maternal alcohol consumption during pregnancy, to environmental factors correlated with PAE that vary between unrelated mothers, or to a variety of measured covariates, including maternal and familial characteristics and exposure to multiple drugs in utero. The findings thus support a strong inference that PAE causes an increased risk of offspring CPs through environmental processes. Compared with children who were not exposed to any alcohol in utero, the results suggest that offspring exposed to prenatal alcohol every day of the week had an increase of 0.35 SD in CPs (the parameter estimate in model 4 reflecting each additional day of exposure multiplied by 7 days). This finding is consistent with the previous research that found robust associations between PAE and offspring externalizing problems, including studies that controlled for characteristics of both parents7,8,13 and included more precise measures of PAE.6 In addition, the present causal inference regarding PAE and child CPs is consistent with experimental animal studies,11 including research in monkeys47 and mice,48 that found teratogenic effects of PAE. Therefore, prevention efforts should continue to target alcohol consumption during pregnancy.24 Although AIPs were statistically associated with PAE and the association remained robust to the use of maternal covariates, PAE does not seem to be a causal risk factor for AIPs. The comparison of siblings and cousins differentially exposed to prenatal alcohol while controlling for measured characteristics revealed no (or a very limited) association. If the present analyses relied only on the use of maternal covariates, we would have wrongly inferred that PAE was a causal risk factor for AIPs. The analysis of the subsample of children with information on prenatal exposure to multiple drugs further provides converging evidence in the study and suggests that maternal polysubstance use during pregnancy may account for the unadjusted associations between PAE and AIPs. The finding that PAE does not cause AIPs is also consistent with a recent review9 and 1 extant behavior genetic study on PAE.14 The contrasting findings for CPs and AIPs underscore the findings from imaging studies that different brain regions may be more vulnerable to PAE. Structures such as the basal ganglia, corpus callosum, and parietal cortex seem to be particularly susceptible to the effects of PAE.49,50 The brain imaging results and studies comparing children with high levels of PAE with those with ADHD51 highlight the importance of exploring the specific neuroanotomical processes that mediate or account for associations between each neurobehavioral problem associated with PAE. The present study has many advantages. Most important, the study used a quasi-experimental design that controlled for genetic and environmental liability shared by mothers and their offspring. The results also accounted for measured maternal and familial characteristics and maternal polysubstance use during pregnancy, a major limitation of many existing studies.11 The results of the study are also more highly generalizable than those of many previous studies because the present results are based on a sample of offspring of women in a large national study, which was weighted to be representative of all women in the United States. The present study has limitations that must be considered. First, quasi-experimental designs cannot prove causality.16 The analyses do not account for every possible variable that could confound the relation. The only plausible alternative explanation for the present findings is that an unmeasured factor that varied between siblings and was highly correlated with variations in PAE is the actual environmental causal factor. This might include variations in maternal depression at the time of each pregnancy. It could also include characteristics of different fathers when they differed across siblings (I.D.W., B.M.D., C.A.V.H., J.L.R., P.J.R., and B.B.L., unpublished data, August 2007),7 but the analyses at least partially ruled out that explanation by controlling for being in a blended family. A full exploration of the variation in alcohol use during pregnancy within mothers is beyond the scope of the present analyses, although some of the variation is related to the historical reduction that has occurred during the longitudinal NLSY79 study.52 Future studies will need to explore additional risks that covary with PAE within mothers. Second, the frequency measure of drinking during each pregnancy may not precisely measure PAE. The assessment of maternal alcohol use during pregnancy was based on reported frequency of alcohol consumption, which did not include the amount of consumption per day or the timing.6 The assessment was also completed retrospectively. However, the magnitude of the unadjusted associations is consistent with previous research,11 and there is evidence that retrospective reports of prenatal substance use are reliable53,54 and are perhaps even more valid than concurrent assessments.55 Third, the present design is based on variations in alcohol consumption across pregnancies.22 The present findings are relevant to the general population only if women who vary their consumption of alcohol across pregnancies are not different from mothers who consistently drank the same amount during each pregnancy. This assumption may be warranted in the present analyses, as most women (88%) who (1) ever drank during 1 pregnancy and (2) had more than 1 child reported variation in their alcohol use among their pregnancies. The difficulties in accurately reporting alcohol consumption for different children in a family by a mother could have affected the sibling comparison analyses. The fact that the sibling and cousin comparisons were comparable, however, suggests that inaccurate measurement of PAE within mothers cannot fully account for the findings. Fourth, the present analyses explored moderators (offspring sex, maternal SDP, and family race/ethnicity) that have been found to influence the effects of prenatal substance exposure,40-42 but other characteristics, including genetic factors56 and stressful events,47 may also moderate the effects of PAE. Fifth, the present analyses could not determine whether the confounding factors for AIPs were genetic or environmental in origin. Use of the NLSY79 and the CNLSY provides the opportunity to nest the sibling comparison approach within a children-of-twins/siblings design, but not all kinship relations are unambiguous.22 The most recent wave of assessments in the NLSY79 (2006) obtained more precise measures of genetic relatedness, so future analyses will explore the nature of the confounding factors with the full adult and offspring sample. Sixth, the results are limited to externalizing problems during childhood reported by the mothers. Future analyses will need to explore substance use outcomes or other characteristics of adolescents.5,8 Seventh, the study was conducted on a nationally representative sample in which the prevalence of children with extremely high levels of PAE was low. The finding of a linear association between PAE and offspring externalizing problems is consistent with previous research,49 but the sample may not have included enough children exposed to high levels of prenatal alcohol to explore whether there is a critical threshold. Likewise, the results exploring the effects of maternal polysubstance use during pregnancy must be viewed cautiously because of the low prevalence of exposure to cocaine and marijuana. Finally, smaller studies of PAE can conduct more extensive assessments of PAE, offspring mental health problems, and potential confounds. Thus, the present large-scale study complements but does not replace more focused and intensive studies. The present study, which was based on a representative sample, a rigorous design, and a sophisticated analytical approach to identify possible confounds, strengthens the inference that PAE has a causal effect on the risk of CPs in offspring. In contrast, the finding that unmeasured confounds related to PAE may be responsible for greater AIPs in offspring suggests that the research community needs to focus on putative environmental risk factors that are correlated with maternal substance use during pregnancy, particularly ones that vary across siblings. The results are consistent with research using different methods but further emphasize the need for additional quasi-experimental studies of prenatal substance exposure. Correspondence: Brian M. D'Onofrio, PhD, Department of Psychological and Brain Sciences, Indiana University, 1101 E 10th St, Bloomington, IN 47405 (bmdonofr@indiana.edu). Submitted for Publication: September 25, 2006; final revision received April 30, 2007; accepted May 1, 2007. Financial Disclosure: None reported. Funding/Support: This research was supported by grant R01-MH070025 from the National Institute of Mental Health (Dr Lahey). References 1. Sood BDelaney-Black VCovington CNordstrom-Klee BAger JTemplin TJanisse JMartier SSokok RJ Prenatal alcohol exposure and childhood behavior at age 6 to 7 years, I: dose-response effect. Pediatrics 2001;108 (2) e3410.1542/peds.108.2.e34PubMedGoogle Scholar 2. Streissguth APBarr HMKogan JBookstein FL Understanding the Occurrence of Secondary Disabilities in Clients With Fetal Alcohol Syndrome (FAS) and Fetal Alcohol Effects (FAE): Final Report. Seattle University of Washington Publication Services1996; 3. Streissguth APMartin DCBarr HMSandman BMKirchner GLDarby BL Intrauterine alcohol and nicotine exposure: attention and reaction time in 4-year-old children. Dev Psychol 1984;20 (4) 533- 541Google Scholar 4. Knopik VSSparrow EPMadden PAFBucholz KKHudziak JJReich WSlutske WSGrant JDMcLaughlin TLTodorov ATodd RDHeath AC Contributions of parental alcoholism, prenatal substance exposure, and genetic transmission to child ADHD risk: a female twin study. Psychol Med 2005;35 (5) 625- 635PubMedGoogle Scholar 5. Alati RAl Mamun AWilliams GMO'Callagham MNajman JMBor W In utero alcohol exposure and prediction of alcohol disorders in early adulthood. Arch Gen Psychiatry 2006;63 (9) 1009- 1016PubMedGoogle Scholar 6. Griesler PCKandel DB The impact of maternal drinking during and after pregnancy on the drinking of adolescent offspring. J Stud Alcohol 1998;59 (3) 292- 304PubMedGoogle Scholar 7. Baer JSBarr HMBookstein FLSampson PDStreissguth AP Prenatal alcohol exposure and family history of alcoholism in the etiology of adolescent alcohol problems. J Stud Alcohol 1998;59 (5) 533- 543PubMedGoogle Scholar 8. Baer JSSampson PDBarr HMConnor PDStreissguth AP A 21-year longitudinal analysis of the effects of prenatal alcohol exposure on young adult drinking. Arch Gen Psychiatry 2003;60 (4) 377- 385PubMedGoogle Scholar 9. Linnet KMDalsgaard SObel CWisborg K Maternal lifestyle factors in pregnancy risk of attention deficit hyperactivity disorder and associated behaviors: review of the current evidence. Am J Psychiatry 2003;160 (6) 1028- 1040 KPubMedGoogle Scholar 10. Testa MQuigley BMDas Eiden R The effects of prenatal alcohol exposure on infant mental development: a meta-analytical review. Alcohol Alcohol 2003;38 (4) 295- 304PubMedGoogle Scholar 11. Huizink ACMulder EJH Maternal smoking, drinking or cannabis use during pregnancy and neurobehavioral and cognitive functioning in human offspring. Neurosci Biobehav Rev 2006;30 (1) 24- 41PubMedGoogle Scholar 12. O’Connor MJKogan NFindlay R Prenatal alcohol exposure and attachment behavior in children. Alcohol Clin Exp Res 2002;26 (10) 1592- 1602PubMedGoogle Scholar 13. Hill SYLowers LLocke-Wellman JShen SA Maternal smoking and drinking during pregnancy and the risk for child and adolescent psychiatric disorders. J Stud Alcohol 2000;61 (5) 661- 668PubMedGoogle Scholar 14. Knopik VSHeath ACJacob TSlutske WSBucholz KKMadden PAWaldron MMartin NG Maternal alcoholism use disorder and offspring ADHD: disentangling genetic and environmental effects using a children-of-twins design. Psychol Med 2006;36 (10) 1461- 1471PubMedGoogle Scholar 15. Scarr SMcCartney K How people make their own environments: a theory of genotype greater than environment effects. Child Dev 1983;54 (2) 424- 435PubMedGoogle Scholar 16. Rutter MPickles AMurray REaves LJ Testing hypotheses on specific environmental causal effects on behavior. Psychol Bull 2001;127 (3) 291- 324PubMedGoogle Scholar 17. Moffitt TE The new look of behavioral genetics in developmental psychopathology: gene-environment interplay in antisocial behaviors. Psychol Bull 2005;131 (4) 533- 554PubMedGoogle Scholar 18. Gottesman IIBertelsen A Confirming unexpressed genotypes for schizophrenia. Arch Gen Psychiatry 1989;46 (10) 867- 872PubMedGoogle Scholar 19. Silberg JLEaves LJ Analyzing the contribution of genes and parent-child interaction to childhood behavioral and emotional problems: a model for the children of twins. Psychol Med 2004;34 (2) 347- 356PubMedGoogle Scholar 20. Heath ACKendler KSEaves LJMarkell D The resolution of cultural and biological inheritance: informativeness of different relationships. Behav Genet 1985;15 (5) 439- 465PubMedGoogle Scholar 21. D’Onofrio BMTurkheimer EEaves LJCorey LABerg KSolaas MHEmery RE The role of the children of twins design in elucidating causal relations between parent characteristics and child outcomes. J Child Psychol Psychiatry 2003;44 (8) 1130- 1144PubMedGoogle Scholar 22. D’Onofrio BMVan Hulle CAWaldman ID et al. Smoking during pregnancy and offspring externalizing problems: an exploration of genetic and environmental confounds. Dev Psychopathol In pressGoogle Scholar 23. Lambe MHultman CTorrang AMacCabe JCnattingius S Maternal smoking during pregnancy and school performance at age 15. Epidemiology 2006;17 (5) 524- 530PubMedGoogle Scholar 24. Floyd RLO’Connor MJBertrand JSokol R Reducing adverse outcomes from prenatal alcohol exposure: a clinic plan of action. Alcohol Clin Exp Res 2006;30 (8) 1271- 1275PubMedGoogle Scholar 25. Substance Abuse and Mental Health Services Administration, Results From the 2003 National Survey on Drug Use and Health: National Findings. Rockville, MD Office of Applied Studies2004;NSDUH series H–25, DHHS publication SMA 04–3964 26. Rodgers JLCleveland Hvan den Oord ERowe D Resolving the debate over birth order, family size, and intelligence. Am Psychol 2000;55 (6) 599- 612PubMedGoogle Scholar 27. Baker PMott FL NLSY Child Handbook. Columbus, OH Center for Human Resource Research1989; 28. Chase-Lansdale PLMott FLBrooks-Gunn JPhillips DA Children of the National Longitudinal Survey of Youth: a unique research opportunity. Dev Psychol 1991;27 (6) 918- 931Google Scholar 29. Elliott DSHuizinga D Social class and delinquent behavior in a national youth panel. Criminology 1983;21 (2) 149- 177Google Scholar 30. Loeber RFarrington DPStouthamer-Loeber MVan Kammen W Antisocial Behavior and Mental Health Problems: Explanatory Factors in Childhood and Adolescence. Mahwah, NJ Lawrence Erlbaum Associates1998; 31. Moffitt TE Juvenile delinquency and attention deficit disorder: boys' developmental trajectories from age 3 to 15. Child Dev 1990;61 (3) 893- 910PubMedGoogle Scholar 32. Moffitt TECaspi ADickson NSilva PAStanton W Childhood-onset versus adolescent-onset antisocial conduct in males: natural history from age 3 to 18. Dev Psychopathol 1996;8 (2) 399- 424Google Scholar 33. Turley RNL Are children of young mothers disadvantaged because of their mother's age or family background? Child Dev 2003;74 (2) 465- 474PubMedGoogle Scholar 34. Achenbach TM The child behavior profile, I: boys aged 6-11. J Consult Clin Psychol 1978;46 (3) 478- 488PubMedGoogle Scholar 35. Peterson JLZill N Marital disruption, parent-child relationships, and behavior problems in children. J Marriage Fam 1986;48 (2) 295- 307Google Scholar 36. Lahey BBVan Hulle CAWaldman IDRodgers JLD'Onofrio BMPedlow SRathouz PKeenan K Testing descriptive hypotheses regarding sex differences in the development of conduct problems and delinquency. J Abnorm Child Psychol 2006;34 (5) 737- 755PubMedGoogle Scholar 37. Fergusson DMHorwood LJ Male and female offending trajectories. Dev Psychopathol 2002;14 (1) 159- 177PubMedGoogle Scholar 38. Lahey BBMcBurnett KLoeber R Are attention-deficit hyperactivity disorder and oppositional defiant disorder developmental precursors to conduct disorder? Sameroff ALewis MMiller Seds Handbook of Developmental Psychopathology. 2nd New York, NY Plenum2000;431- 446Google Scholar 39. Raudenbush SWBryk AS Hierarchical Linear Models: Applications and Data Analysis Methods. 2nd Thousand Oaks, CA Sage Publications2002; 40. Brennan PAGrekin ERMednick SA Maternal smoking during pregnancy and adult male criminal outcomes. Arch Gen Psychiatry 1999;56 (3) 215- 219PubMedGoogle Scholar 41. Streissguth APMartin DCMartin JCBarr HM The Seattle longitudinal prospective study on alcohol and pregnancy. Neurobehav Toxicol Teratol 1981;3 (2) 223- 233PubMedGoogle Scholar 42. Willford JLeech SDay N Moderate prenatal alcohol exposure and cognitive status of children at age 10. Alcohol Clin Exp Res 2006;30 (6) 1051- 1059PubMedGoogle Scholar 43. Mendle JTurkheimer ED’Onofrio BMLynch SKEmery RESlutske WSMartin NG Family structure and age at menarche: a children-of-twins approach. Dev Psychol 2006;42 (3) 533- 542PubMedGoogle Scholar 44. Neuhaus JMMcCulloch CE Separating between- and within-cluster covariate effects by using conditional and partitioning methods. J R Stat Soc Ser B 2006;68 (5) 859- 872Google Scholar 45. Little RJARubin DB Statistical Analysis With Missing Data. New York, NY John Wiley & Sons Inc1987; 46. Kim JMueller CW Standardized and unstandardized coefficients in causal analysis: an expository note. Sociol Methods Res 1976;4 (4) 423- 438Google Scholar 47. Schneider MLMoore CFKraemer GW Moderate alcohol during pregnancy: learning and behavior in adolescent rhesus monkeys. Alcohol Clin Exp Res 2001;25 (9) 1383- 1392PubMedGoogle Scholar 48. Allan AMChynoweth JTyler LACaldewell KKA A mouse model of prenatal ethanol exposure using a voluntary drinking paradigm. Alcohol Clin Exp Res 2003;27 (12) 2009- 2016PubMedGoogle Scholar 49. Wass TSMattson SNRiley A Neuroanatomical and neurobehavioral effects of heavy prenatal alcohol exposure. Brick Jed Handbook of the Medical Consequences of Alcohol and Drug Abuse. New York, NY Haworth Press2004;139- 169Google Scholar 50. Mattson SNSchoenfield AMRiley EP Teratogenic effects of alcohol on brain and behavior. Alcohol Res Health 2001;25 (3) 185- 191PubMedGoogle Scholar 51. Coles CDPlatzman KARaskind-Hood CLBrown RTFalek ASmith IE A comparison of children affected by prenatal alcohol exposure and attention deficit, hyperactivity disorder. Alcohol Clin Exp Res 1997;21 (1) 150- 161PubMedGoogle Scholar 52. Bobo JKKlepinger DHDong FB Changes in the prevalence of alcohol use during pregnancy among recent and at-risk drinkers in the NLSY cohort. J Womens Health (Larchmt) 2006;15 (9) 1061- 1070PubMedGoogle Scholar 53. Reich WTodd RDJoyner CANeuman RJHeath AC Reliability and stability of mothers' reports about their pregnancies with twins. Twin Res 2003;6 (2) 85- 88PubMedGoogle Scholar 54. Heath ACKnopik VSMadden PAFNeuman RJLynskey MJSlutske WSJacob TMartin NG Accuracy of mothers' retrospective reports of smoking during pregnancy: comparison with twin sister informant ratings. Twin Res 2003;6 (4) 297- 301PubMedGoogle Scholar 55. Alvik AHaldorsen TGroholt BLindemann R Alcohol consumption before and during pregnancy comparing concurrent and retrospective reports. Alcohol Clin Exp Res 2006;30 (3) 510- 515PubMedGoogle Scholar 56. Brookes KJMill JGuindalini CCurran SXu XKnight JChen CKHuang YSSethna VTaylor EChen WBreen GAsherson P A common haplotype of the dopamine transporter gene associated with attention deficit/hyperactivity disorder and interacting with maternal use of alcohol during pregnancy. Arch Gen Psychiatry 2006;63 (1) 74- 81PubMedGoogle Scholar http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Archives of General Psychiatry American Medical Association

Causal Inferences Regarding Prenatal Alcohol Exposure and Childhood Externalizing Problems

Download PDF
9 pages

Loading next page...
 
/lp/american-medical-association/causal-inferences-regarding-prenatal-alcohol-exposure-and-childhood-gFqOPzK4Cd

References (65)

Publisher
American Medical Association
Copyright
Copyright © 2007 American Medical Association. All Rights Reserved.
ISSN
0003-990X
DOI
10.1001/archpsyc.64.11.1296
pmid
17984398
Publisher site
See Article on Publisher Site

Abstract

Abstract Context Existing research on the neurobehavioral consequences of prenatal alcohol exposure (PAE) has not adequately accounted for genetic and environmental confounds. Objective To examine the association between PAE and offspring externalizing problems in a large representative sample of families in the United States using measured covariates and a quasi-experimental design to account for unmeasured genetic and environmental confounds. Design This study combines information from the National Longitudinal Survey of Youth and the Children of the National Longitudinal Survey of Youth. The analyses statistically controlled for measured characteristics of the mothers and families and exposure to other prenatal psychoactive substances. In the primary analyses, siblings differentially exposed to prenatal alcohol were compared. Setting and Participants Women were recruited from the community using a stratified and clustered probability sample and were followed longitudinally. The sample included 8621 offspring of 4912 mothers. Main Outcome Measures Maternal report of conduct problems (CPs) and attention/impulsivity problems (AIPs) during childhood (ages 4-11 years) using standardized assessments related to psychiatric diagnoses. Results There was an association between PAE and offspring CPs that was independent of confounded genetic and fixed environmental effects and the measured covariates. The CPs in children of mothers who drank daily during pregnancy were 0.35 SD greater than those in children whose mothers never drank during pregnancy. Although AIPs were associated with PAE when comparing unrelated offspring, children whose mothers drank more frequently during pregnancy did not have more AIPs than siblings who were less exposed to alcohol in utero. Additional subsample analyses suggested that maternal polysubstance use during pregnancy may account for the associations between PAE and AIPs. Conclusion These results are consistent with PAE exerting an environmentally mediated causal effect on childhood CPs, but the relation between PAE and AIPs is more likely to be caused by other factors correlated with maternal drinking during pregnancy. Prenatal alcohol exposure (PAE) has been associated with offspring externalizing problems, such as conduct problems (CPs),1 criminality,2 attention/impulsivity problems (AIPs),3,4 and alcohol disorders.5-8 A recent review of the literature9 and a meta-analysis,10 however, have raised questions about the strength of some of the associations. Given the current state of research on prenatal drug exposure, Huizink and Mulder11 recently highlighted important challenges for the research community. In particular, they noted the need to document deleterious outcomes associated with moderate levels of alcohol exposure, explore the adjustment of offspring across time, and delineate the effects of exposure to multiple drugs. Research also needs to adequately control for covariates that may confound the effects of PAE, such as family processes (eg, problematic parenting or family conflict) and parental characteristics, especially maternal substance use.1,12 Researchers also need to account for genetic liabilities that are shared by parents and offspring.13,14 A woman's genetic risk of substance use could be passed down to her children and subsequently affect their behavior.15 Research on the consequences of PAE, therefore, needs studies with large samples, with sufficient statistical power to detect small effects, using analytical methods and designs that can account for potential confounds, including factors that are not measured. Such studies could complement smaller, more intensive studies6,7,13 and provide the opportunity to find convergent evidence from multiple research approaches.16 Because of the myriad of variables associated with prenatal substance exposure, quasi-experimental designs are useful because they can test alternative causal inferences.16 Quasi-experimental designs rely on natural experiments that can pull apart processes that are typically confounded. Genetically informed approaches, one class of quasi-experimental designs, are important because they can separate the co-occurring genetic and environmental processes. The use of quasi-experimental designs, therefore, can help the study of putative environments, such as PAE, move from merely identifying risk factors to highlighting the underlying causal mechanisms of externalizing problems.17 Recently, a variety of quasi-experimental and genetically informed studies related to prenatal substance exposure have been reported. Two studies used the children-of-twins design18-20 to explore smoking during pregnancy (SDP), finding an independent association between SDP and birth weight21 and attention-deficit/hyperactivity disorder (ADHD)14 when shared genetic and environmental liability were taken into account. Sibling comparison studies have also been used to study SDP. The approach contrasts children exposed to more prenatal nicotine to their siblings exposed to less, a design that controls for characteristics of the mothers, including genetic risk associated with SDP. The results suggest that independent associations between SDP and psychosocial adjustment are either greatly reduced or absent.22,23 Only 1 quasi-experimental study, however, has been conducted on the consequences of PAE. Knopik et al14 found that PAE was not independently associated with offspring ADHD in a children-of-twins study. Given the somewhat contradictory findings of the association between PAE and offspring attention problems9 and the results of the 1 genetically informed study,14 a quasi-experimental study of the consequences of PAE on offspring externalizing problems is of critical importance. Inferences about PAE affect public health campaigns and prevention efforts24 in addition to informing theories about the etiology of CPs.17 The need for more rigorous studies of PAE is further supported by recent prevalence estimates that indicate that alcohol use during pregnancy is still relatively common.25 The overall aim of the present article is to explore whether PAE is related to offspring externalizing behaviors and, if so, whether the exposure is consistent with a causal association. The present study has 6 goals. First, these analyses explore the relation between PAE and characteristics of the mothers and families. Second, the study documents the statistical relation between PAE and child externalizing problems. Third, the analyses explore whether measured covariates account for associations with PAE. Fourth, the analyses compare siblings and cousins differentially exposed to prenatal alcohol to account for unmeasured confounds, including genetic and environmental factors. If PAE causes externalizing problems, the association will be found at all levels, especially within mothers (ie, siblings exposed to higher levels of prenatal alcohol would have more problems than their siblings exposed to lower levels).26 The comparison of siblings explicitly controls for maternal genetic confounds because PAE is not correlated with genetic risk within mothers. Fifth, the analyses investigate whether the relation was robust to the quasi-experimental approach and statistical controls. Finally, an analysis of a subsample of the children that includes measures of prenatal exposure to marijuana and cocaine further explores maternal polysubstance use. Ultimately, an environmentally mediated causal effect of PAE on childhood externalizing behaviors—a causal inference—would be supported if the relation remained robust to the measured covariates and quasi-experimental methods because these approaches account for many of the alternative explanations for the association between PAE and externalizing problems. This research was approved by the institutional review board at Indiana University. Methods Sample Mother Generation Sample The National Longitudinal Survey of Youth (NLSY79) is a longitudinal study of adolescents and young adults funded by the Bureau of Labor Statistics. The initial NLSY79 sample included a nationally representative sample of 6111 youths and an oversample of 3652 African American and Hispanic youth. The initial sample of youth used a stratified and clustered probability sample, which included all siblings in each household in the age range. The present study was conducted using 4912 females from the combined sample who had a child who was at least 4 years old by the 2004 survey. The initial NLSY79 assessment had a response rate of 90%. Participants were assessed annually between 1979 and 1994 and biennially since then. Retention rates were 90% or better during the first 16 waves of data collection. Probability weights for the sample are available for the sample of mothers, so the analyses could be weighted to produce population-based estimates. More details concerning the NLSY79 are available elsewhere.27 The characteristics of the females with children are given in Table 1. The sample was racially diverse; based on mothers' self-reports, 17.1% were Hispanic, 25.9% were African American, and 57.1% were non-Hispanic white. Offspring Generation Sample Starting in 1986, biennial assessments of the offspring of the women in the NLSY79 sample were performed, referred to as the Children of the National Longitudinal Survey of Youth (CNLSY) sample.28 The initial response rate was 95%, with subsequent assessments averaging 90%. Although 3 types of interviews are available on the offspring, the present project uses maternal report of the children's characteristics assessed between ages 4 and 11 years (n = 8621 children with reports of PAE). Studies using the offspring of the NLSY79 mothers have been plagued by selection bias because until all childbearing of the mothers is complete, offspring available for assessment were necessarily born to the younger mothers. Given that the overwhelming majority of the childbearing was completed by the 2004 survey and we use assessments of young children, the selection bias for the present analyses is relatively small (B.M.D., J. A. Goodnight, MA, C.A.V.H., I.D.W., J.L.R., P.J.R., and B.B.L., unpublished data, August 2007). Measures Prenatal Exposure to Psychoactive Substances Beginning in 1983, the NLSY79 females reported their prenatal substance use, including alcohol consumption and cigarette use, during each of their pregnancies. Most of the assessments of prenatal substance use reports were completed when the children were young, mostly within 2 years of birth.22Table 2 includes the response categories for alcohol and conversions to days per week for the sample of offspring with information on externalizing problems. A previous article22 analyzed the consequences of SDP. Owing to the small number of pregnancies during which the mothers reported drinking every day, the last 2 categories were collapsed. There are missing data for the prenatal substance use section. Most of the missing PAE data (92%) were due to an incorrect skip in the assessment protocol. Sensitivity analyses (replacing the missing values with zeros, the mean PAE for the mothers, or estimates based on multiple imputation) revealed that the missing values for PAE did not alter the results presented herein, which are based only on offspring with reports of PAE. The maternal assessment started to measure marijuana and cocaine use during pregnancy in the 1986 assessment. Therefore, the variables were available only for 3977 offspring with data on externalizing behaviors. Because of the low prevalence, exposure was coded as exposed or not for each drug. In the subsample, 1.6% of children were exposed to prenatal marijuana and 0.6% to cocaine. Maternal and Familial Characteristics Available characteristics of the mothers and families that could confound the effects of PAE were included in the analyses. Table 1 provides the distribution of the items. When the eventual mothers of the CNLSY sample were aged 15 to 22 years they were asked about their participation in 12 delinquent behaviors using the Self-Reported Delinquency Interview,29 a reliable and valid measure used in contemporary delinquency research.30-32 Symptom counts were regressed on the mother's age at which she completed the survey and were converted to a z score for ease of interpretability. Maternal age at the birth of her first child was also obtained, a variable that predicts offspring delinquency in the CNLSY (B.M.D., J. A. Goodnight, MA, C.A.V.H., I.D.W., J.L.R., P.J.R., and B.B.L., unpublished data, August 2007).33 Net family income was reported for mothers at age 30 years in 1986 inflation-adjusted dollars, a measure that includes income from all adults in the household at that time. Total number of years of schooling was also assessed for each mother. A composite score of the Armed Services Vocational Aptitude Battery was used as a measure of the mother's intellectual ability, which was converted to a z score. The 1994 assessment included an extensive assessment of the mother's alcohol use and alcohol problems (during her lifetime and not limited to her pregnancies). These questions assessed alcohol frequency (days per month), quantity (drinks per day), and the number of binge episodes in the past month, and 25 items assessed lifetime symptoms of alcohol abuse and dependence. A total of 32% of the women also reported having children by more than 1 man. Offspring Childhood Externalizing Problems Mothers rated their 4- to 11-year-old children on the Behavior Problem Index at each wave of assessment. The Behavior Problem Index included 13 items from the Child Behavior Checklist34 that had the strongest associations with Child Behavior Checklist factor scores.35 Mothers rated each child using a 3-point scale. Standard diagnostic approaches36 and previous factor analyses22 revealed that the items load on 3 factors: CPs, oppositional problems, and AIPs. The association between PAE and CPs and AIPs is explored herein. Trajectory analyses on the CNLSY sample36 indicated that the average number of CPs during ages 4 to 11 years was stable and represents a good measure of severity of externalizing problems reported by adolescents. Therefore, mean CP scores across ages 4 to 11 years were calculated by taking the mean of z scores for CPs at each assessed year of age, standardized within each age. The CPs at each age were highly correlated with the mean across ages 4 to 11 years (r = 0.74-0.80). The items assessing CPs used in the CNLSY also overlap substantially with those used in previous population-based longitudinal studies.32,37 The mean of z scores for AIPs was also stable and exhibited good external validity. The AIP scores at each age were highly correlated with the mean score across ages 4 to 11 years (r > 0.80 for each). Mean AIPs were also strongly related to standardized assessments of academic achievement performed at age 10 years, even controlling for CPs (βmath = −.17, P < .001; βreading = −.17, P < .001; and βword recognition = −.19, P < .001). Moreover, the AIP items are similar to those used in several studies that yielded results similar to studies that used full measures of ADHD consistent with the DSM-III-R.38 Statistical analyses Correlations were calculated among the maternal covariates to estimate the associations among PAE and maternal and familial characteristics to explore what factors predicted or were associated with PAE. All of the analyses in the present article were weighted to produce estimates from a representative sample of women in the United States. Hierarchical linear models (HLMs) were used to explore and characterize the associations between PAE and offspring externalizing behaviors because of the nested nature of the data.39 When the NLSY79 and CNLSY samples are combined, there are 3 levels of analysis: the offspring, maternal, and NLSY79 household levels.22 Five primary HLMs were fit to the associations between PAE and each measure of offspring externalizing problems. Model 1 calculated the unadjusted relation between alcohol exposure, measured in mean number of days exposed per week, and each dimension of externalizing behavior, controlling for offspring sex. The analyses included 8621 offspring of 4912 mothers. Although previous analyses suggested that offspring sex,6,40 maternal SDP,41 and family race/ethnicity42 may moderate the associations, the factors did not affect the magnitude of the relations between PAE and CPs or AIPs in the present sample. The analyses also revealed that linear models provided the best fit for CPs and AIPs because the parameters associated with a quadratic effect were small and statistically nonsignificant. Model 2 was fit to the relation between PAE and offspring externalizing behaviors controlling for measured covariates that could confound the association (listed in the “Measures” subsection of the “Methods” section). Model 3 compared siblings differentially exposed to prenatal alcohol (ie, the mother drank alcohol during one pregnancy and not another or drank more during one pregnancy than another). The within-mother association (ie, comparisons of the different offspring of each mother) provides a more stringent test of causality because the comparison controls for unmeasured genetic and environmental confounds that vary between unrelated mothers. The analyses were based on a subset of the families that included children with differing amounts of exposure (3447 offspring from 1258 mothers). Siblings were compared using contrast codes. Each offspring was assigned a value equal to the deviation of the individual measure of PAE from the mean PAE of all the children from the same mother. The contrast codes, therefore, reflect the amount of PAE relative (more or less) to the mean PAE of all the children in the family. The model also included comparisons of cousins born to NLSY79 mothers who are sisters. Although the comparison of cousins is not as strong as a within-mother approach, it can help provide converging evidence concerning causal processes.43 Cousins were also compared using contrast codes; cousins with more PAE, on average, were compared with their cousins with less PAE, on average. The model also compared unrelated individuals (mean PAE in all cousins from one extended family compared with mean PAE in cousins in a separate extended family). More details about how the contrast codes were calculated,22,43 and the statistical justification for their use,44 can be found elsewhere. Model 4 combined the use of measured covariates and the sibling and cousin comparisons, thus using statistical and methodologic controls for potential confounds. Model 5 tested whether PAE was associated with externalizing problems in the subsample of children that included information on prenatal exposure to marijuana and cocaine. The HLMs included prenatal exposure to alcohol, nicotine, marijuana, and cocaine and the sex of the offspring. Because the assessment of all the substances was available only on a subset of the sample, sibling comparisons could not be completed. Multiple imputation was used for all of the HLMs that included maternal and familial covariates to account for missing values. The analyses, therefore, use all available data rather than relying on families with complete maternal data. Given the assumption that the missing values are missing at random or completely at random, multiple imputation yields standard errors that account for uncertainty due to missing values and avoids bias arising from complete-record analysis.45 Unstandardized regression parameters were used because of the difficulty comparing standardized coefficients when exploring causal processes,46 and measurement of PAE is in meaningful units (days per week). Results Associations among mean alcohol use during pregnancy and maternal and familial characteristics Table 3 provides the associations among mean alcohol use during pregnancy and maternal and familial characteristics. Mean PAE across all pregnancies was slightly correlated with maternal delinquency and age at first childbirth. Mean PAE was more strongly related to substance use measures, including mean levels of SDP, alcohol frequency and quantity, and alcohol abuse and dependence items. Binge drinking Δ(Mbinge − Mno binge) = 0.20, where M represents the mean; P < .001) and having children with different biological fathers Δ(Mnot blended − Mblended) = 0.08; P < .001) were also associated with PAE. ASSOCIATION BETWEEN PAE AND OFFSPRING CPs The results of the HLMs exploring the association between PAE and offspring CPs are given in Table 4. Model 1 indicates that for each additional day of PAE per week, offspring had an increase of 0.09 SD in CPs (P < .001). Model 2 explored the association of PAE and CPs controlling for the main effects of race/ethnicity, prenatal nicotine exposure, and maternal covariates. Including the measured covariates slightly reduced the association between PAE and offspring CPs (b = 0.06, where b represents unstandardized regression weight; P < .001) compared with model 1, but the model still revealed a positive and statistically significant relation. The Figure presents the unadjusted and adjusted means for CPs at each level of PAE. The adjusted means are based on least squares estimates after controlling for the measured covariates. The Figure illustrates a positive association between PAE and CPs for the entire sample; children exposed to more prenatal alcohol had more CPs than unrelated children with less PAE, even after controlling for the measured covariates. Model 3 provided a statistical comparison of siblings differentially exposed to prenatal alcohol. The within-mother association with PAE was still significant (b = 0.05; P = .03) and was commensurate with findings from the comparison of cousins and unrelated individuals in the model. The last model included the measured covariates, and the statistical relation with PAE remained within mothers (b = 0.05; P = .04). The cousin comparisons in models 3 and 4 also suggests that PAE is independently associated with offspring CPs. The Figure also provides an estimate of the association between PAE and CPs based on the parameters from model 4. Children with more PAE have more CPs than their siblings with less PAE while controlling for the measured factors. The results of all the models are consequently consistent with a causal association between PAE and offspring CPs. ASSOCIATION BETWEEN PAE AND AIPs The results of the HLMs for AIPs are given in Table 5. Model 1 found an association between PAE and AIPs (b = 0.09 SD per days exposed per week). Model 2 controlled for measured covariates, and the statistical association with PAE remained (b = 0.06; P = .001). The results of model 3, the comparison of siblings (b = 0.03; P = .18), and model 4, the comparison of siblings controlling for measured covariates (b = 0.02; P = .24), revealed greatly reduced and statistically nonsignificant associations. The comparison of cousins in model 4 also revealed no association, further suggesting that the statistical association between PAE and AIPs is not causal. The association between PAE and AIPs, therefore, seems to be due primarily to characteristics related to PAE that were not included in the previous analytical models. Analysis of pae controlling for maternal polysubstance use during pregnancy Additional HLMs of the sample of children with data on exposure to prenatal alcohol, nicotine, cocaine, and marijuana were conducted. The results suggested that PAE was independently associated with CPs; the magnitude of the estimate (b ± SE = 0.04 ± 0.03; P = .16) was commensurate with the results of the full sample, but the parameter could not be precisely estimated in the smaller subsample. Consistent with the sibling comparisons, statistically controlling for all of the measures of prenatal substance exposure resulted in no association between PAE and AIPs (b ± SE = −0.01 ± 0.15; P = .93). Comment The structure of relationships within families in this sample provides a powerful natural laboratory for quasi-experimental designs of environmental risk factors.16,26 The results of the present study indicate that there is an association between PAE and CPs that was not due to genetic factors related to maternal alcohol consumption during pregnancy, to environmental factors correlated with PAE that vary between unrelated mothers, or to a variety of measured covariates, including maternal and familial characteristics and exposure to multiple drugs in utero. The findings thus support a strong inference that PAE causes an increased risk of offspring CPs through environmental processes. Compared with children who were not exposed to any alcohol in utero, the results suggest that offspring exposed to prenatal alcohol every day of the week had an increase of 0.35 SD in CPs (the parameter estimate in model 4 reflecting each additional day of exposure multiplied by 7 days). This finding is consistent with the previous research that found robust associations between PAE and offspring externalizing problems, including studies that controlled for characteristics of both parents7,8,13 and included more precise measures of PAE.6 In addition, the present causal inference regarding PAE and child CPs is consistent with experimental animal studies,11 including research in monkeys47 and mice,48 that found teratogenic effects of PAE. Therefore, prevention efforts should continue to target alcohol consumption during pregnancy.24 Although AIPs were statistically associated with PAE and the association remained robust to the use of maternal covariates, PAE does not seem to be a causal risk factor for AIPs. The comparison of siblings and cousins differentially exposed to prenatal alcohol while controlling for measured characteristics revealed no (or a very limited) association. If the present analyses relied only on the use of maternal covariates, we would have wrongly inferred that PAE was a causal risk factor for AIPs. The analysis of the subsample of children with information on prenatal exposure to multiple drugs further provides converging evidence in the study and suggests that maternal polysubstance use during pregnancy may account for the unadjusted associations between PAE and AIPs. The finding that PAE does not cause AIPs is also consistent with a recent review9 and 1 extant behavior genetic study on PAE.14 The contrasting findings for CPs and AIPs underscore the findings from imaging studies that different brain regions may be more vulnerable to PAE. Structures such as the basal ganglia, corpus callosum, and parietal cortex seem to be particularly susceptible to the effects of PAE.49,50 The brain imaging results and studies comparing children with high levels of PAE with those with ADHD51 highlight the importance of exploring the specific neuroanotomical processes that mediate or account for associations between each neurobehavioral problem associated with PAE. The present study has many advantages. Most important, the study used a quasi-experimental design that controlled for genetic and environmental liability shared by mothers and their offspring. The results also accounted for measured maternal and familial characteristics and maternal polysubstance use during pregnancy, a major limitation of many existing studies.11 The results of the study are also more highly generalizable than those of many previous studies because the present results are based on a sample of offspring of women in a large national study, which was weighted to be representative of all women in the United States. The present study has limitations that must be considered. First, quasi-experimental designs cannot prove causality.16 The analyses do not account for every possible variable that could confound the relation. The only plausible alternative explanation for the present findings is that an unmeasured factor that varied between siblings and was highly correlated with variations in PAE is the actual environmental causal factor. This might include variations in maternal depression at the time of each pregnancy. It could also include characteristics of different fathers when they differed across siblings (I.D.W., B.M.D., C.A.V.H., J.L.R., P.J.R., and B.B.L., unpublished data, August 2007),7 but the analyses at least partially ruled out that explanation by controlling for being in a blended family. A full exploration of the variation in alcohol use during pregnancy within mothers is beyond the scope of the present analyses, although some of the variation is related to the historical reduction that has occurred during the longitudinal NLSY79 study.52 Future studies will need to explore additional risks that covary with PAE within mothers. Second, the frequency measure of drinking during each pregnancy may not precisely measure PAE. The assessment of maternal alcohol use during pregnancy was based on reported frequency of alcohol consumption, which did not include the amount of consumption per day or the timing.6 The assessment was also completed retrospectively. However, the magnitude of the unadjusted associations is consistent with previous research,11 and there is evidence that retrospective reports of prenatal substance use are reliable53,54 and are perhaps even more valid than concurrent assessments.55 Third, the present design is based on variations in alcohol consumption across pregnancies.22 The present findings are relevant to the general population only if women who vary their consumption of alcohol across pregnancies are not different from mothers who consistently drank the same amount during each pregnancy. This assumption may be warranted in the present analyses, as most women (88%) who (1) ever drank during 1 pregnancy and (2) had more than 1 child reported variation in their alcohol use among their pregnancies. The difficulties in accurately reporting alcohol consumption for different children in a family by a mother could have affected the sibling comparison analyses. The fact that the sibling and cousin comparisons were comparable, however, suggests that inaccurate measurement of PAE within mothers cannot fully account for the findings. Fourth, the present analyses explored moderators (offspring sex, maternal SDP, and family race/ethnicity) that have been found to influence the effects of prenatal substance exposure,40-42 but other characteristics, including genetic factors56 and stressful events,47 may also moderate the effects of PAE. Fifth, the present analyses could not determine whether the confounding factors for AIPs were genetic or environmental in origin. Use of the NLSY79 and the CNLSY provides the opportunity to nest the sibling comparison approach within a children-of-twins/siblings design, but not all kinship relations are unambiguous.22 The most recent wave of assessments in the NLSY79 (2006) obtained more precise measures of genetic relatedness, so future analyses will explore the nature of the confounding factors with the full adult and offspring sample. Sixth, the results are limited to externalizing problems during childhood reported by the mothers. Future analyses will need to explore substance use outcomes or other characteristics of adolescents.5,8 Seventh, the study was conducted on a nationally representative sample in which the prevalence of children with extremely high levels of PAE was low. The finding of a linear association between PAE and offspring externalizing problems is consistent with previous research,49 but the sample may not have included enough children exposed to high levels of prenatal alcohol to explore whether there is a critical threshold. Likewise, the results exploring the effects of maternal polysubstance use during pregnancy must be viewed cautiously because of the low prevalence of exposure to cocaine and marijuana. Finally, smaller studies of PAE can conduct more extensive assessments of PAE, offspring mental health problems, and potential confounds. Thus, the present large-scale study complements but does not replace more focused and intensive studies. The present study, which was based on a representative sample, a rigorous design, and a sophisticated analytical approach to identify possible confounds, strengthens the inference that PAE has a causal effect on the risk of CPs in offspring. In contrast, the finding that unmeasured confounds related to PAE may be responsible for greater AIPs in offspring suggests that the research community needs to focus on putative environmental risk factors that are correlated with maternal substance use during pregnancy, particularly ones that vary across siblings. The results are consistent with research using different methods but further emphasize the need for additional quasi-experimental studies of prenatal substance exposure. Correspondence: Brian M. D'Onofrio, PhD, Department of Psychological and Brain Sciences, Indiana University, 1101 E 10th St, Bloomington, IN 47405 (bmdonofr@indiana.edu). Submitted for Publication: September 25, 2006; final revision received April 30, 2007; accepted May 1, 2007. Financial Disclosure: None reported. Funding/Support: This research was supported by grant R01-MH070025 from the National Institute of Mental Health (Dr Lahey). References 1. Sood BDelaney-Black VCovington CNordstrom-Klee BAger JTemplin TJanisse JMartier SSokok RJ Prenatal alcohol exposure and childhood behavior at age 6 to 7 years, I: dose-response effect. Pediatrics 2001;108 (2) e3410.1542/peds.108.2.e34PubMedGoogle Scholar 2. Streissguth APBarr HMKogan JBookstein FL Understanding the Occurrence of Secondary Disabilities in Clients With Fetal Alcohol Syndrome (FAS) and Fetal Alcohol Effects (FAE): Final Report. Seattle University of Washington Publication Services1996; 3. Streissguth APMartin DCBarr HMSandman BMKirchner GLDarby BL Intrauterine alcohol and nicotine exposure: attention and reaction time in 4-year-old children. Dev Psychol 1984;20 (4) 533- 541Google Scholar 4. Knopik VSSparrow EPMadden PAFBucholz KKHudziak JJReich WSlutske WSGrant JDMcLaughlin TLTodorov ATodd RDHeath AC Contributions of parental alcoholism, prenatal substance exposure, and genetic transmission to child ADHD risk: a female twin study. Psychol Med 2005;35 (5) 625- 635PubMedGoogle Scholar 5. Alati RAl Mamun AWilliams GMO'Callagham MNajman JMBor W In utero alcohol exposure and prediction of alcohol disorders in early adulthood. Arch Gen Psychiatry 2006;63 (9) 1009- 1016PubMedGoogle Scholar 6. Griesler PCKandel DB The impact of maternal drinking during and after pregnancy on the drinking of adolescent offspring. J Stud Alcohol 1998;59 (3) 292- 304PubMedGoogle Scholar 7. Baer JSBarr HMBookstein FLSampson PDStreissguth AP Prenatal alcohol exposure and family history of alcoholism in the etiology of adolescent alcohol problems. J Stud Alcohol 1998;59 (5) 533- 543PubMedGoogle Scholar 8. Baer JSSampson PDBarr HMConnor PDStreissguth AP A 21-year longitudinal analysis of the effects of prenatal alcohol exposure on young adult drinking. Arch Gen Psychiatry 2003;60 (4) 377- 385PubMedGoogle Scholar 9. Linnet KMDalsgaard SObel CWisborg K Maternal lifestyle factors in pregnancy risk of attention deficit hyperactivity disorder and associated behaviors: review of the current evidence. Am J Psychiatry 2003;160 (6) 1028- 1040 KPubMedGoogle Scholar 10. Testa MQuigley BMDas Eiden R The effects of prenatal alcohol exposure on infant mental development: a meta-analytical review. Alcohol Alcohol 2003;38 (4) 295- 304PubMedGoogle Scholar 11. Huizink ACMulder EJH Maternal smoking, drinking or cannabis use during pregnancy and neurobehavioral and cognitive functioning in human offspring. Neurosci Biobehav Rev 2006;30 (1) 24- 41PubMedGoogle Scholar 12. O’Connor MJKogan NFindlay R Prenatal alcohol exposure and attachment behavior in children. Alcohol Clin Exp Res 2002;26 (10) 1592- 1602PubMedGoogle Scholar 13. Hill SYLowers LLocke-Wellman JShen SA Maternal smoking and drinking during pregnancy and the risk for child and adolescent psychiatric disorders. J Stud Alcohol 2000;61 (5) 661- 668PubMedGoogle Scholar 14. Knopik VSHeath ACJacob TSlutske WSBucholz KKMadden PAWaldron MMartin NG Maternal alcoholism use disorder and offspring ADHD: disentangling genetic and environmental effects using a children-of-twins design. Psychol Med 2006;36 (10) 1461- 1471PubMedGoogle Scholar 15. Scarr SMcCartney K How people make their own environments: a theory of genotype greater than environment effects. Child Dev 1983;54 (2) 424- 435PubMedGoogle Scholar 16. Rutter MPickles AMurray REaves LJ Testing hypotheses on specific environmental causal effects on behavior. Psychol Bull 2001;127 (3) 291- 324PubMedGoogle Scholar 17. Moffitt TE The new look of behavioral genetics in developmental psychopathology: gene-environment interplay in antisocial behaviors. Psychol Bull 2005;131 (4) 533- 554PubMedGoogle Scholar 18. Gottesman IIBertelsen A Confirming unexpressed genotypes for schizophrenia. Arch Gen Psychiatry 1989;46 (10) 867- 872PubMedGoogle Scholar 19. Silberg JLEaves LJ Analyzing the contribution of genes and parent-child interaction to childhood behavioral and emotional problems: a model for the children of twins. Psychol Med 2004;34 (2) 347- 356PubMedGoogle Scholar 20. Heath ACKendler KSEaves LJMarkell D The resolution of cultural and biological inheritance: informativeness of different relationships. Behav Genet 1985;15 (5) 439- 465PubMedGoogle Scholar 21. D’Onofrio BMTurkheimer EEaves LJCorey LABerg KSolaas MHEmery RE The role of the children of twins design in elucidating causal relations between parent characteristics and child outcomes. J Child Psychol Psychiatry 2003;44 (8) 1130- 1144PubMedGoogle Scholar 22. D’Onofrio BMVan Hulle CAWaldman ID et al. Smoking during pregnancy and offspring externalizing problems: an exploration of genetic and environmental confounds. Dev Psychopathol In pressGoogle Scholar 23. Lambe MHultman CTorrang AMacCabe JCnattingius S Maternal smoking during pregnancy and school performance at age 15. Epidemiology 2006;17 (5) 524- 530PubMedGoogle Scholar 24. Floyd RLO’Connor MJBertrand JSokol R Reducing adverse outcomes from prenatal alcohol exposure: a clinic plan of action. Alcohol Clin Exp Res 2006;30 (8) 1271- 1275PubMedGoogle Scholar 25. Substance Abuse and Mental Health Services Administration, Results From the 2003 National Survey on Drug Use and Health: National Findings. Rockville, MD Office of Applied Studies2004;NSDUH series H–25, DHHS publication SMA 04–3964 26. Rodgers JLCleveland Hvan den Oord ERowe D Resolving the debate over birth order, family size, and intelligence. Am Psychol 2000;55 (6) 599- 612PubMedGoogle Scholar 27. Baker PMott FL NLSY Child Handbook. Columbus, OH Center for Human Resource Research1989; 28. Chase-Lansdale PLMott FLBrooks-Gunn JPhillips DA Children of the National Longitudinal Survey of Youth: a unique research opportunity. Dev Psychol 1991;27 (6) 918- 931Google Scholar 29. Elliott DSHuizinga D Social class and delinquent behavior in a national youth panel. Criminology 1983;21 (2) 149- 177Google Scholar 30. Loeber RFarrington DPStouthamer-Loeber MVan Kammen W Antisocial Behavior and Mental Health Problems: Explanatory Factors in Childhood and Adolescence. Mahwah, NJ Lawrence Erlbaum Associates1998; 31. Moffitt TE Juvenile delinquency and attention deficit disorder: boys' developmental trajectories from age 3 to 15. Child Dev 1990;61 (3) 893- 910PubMedGoogle Scholar 32. Moffitt TECaspi ADickson NSilva PAStanton W Childhood-onset versus adolescent-onset antisocial conduct in males: natural history from age 3 to 18. Dev Psychopathol 1996;8 (2) 399- 424Google Scholar 33. Turley RNL Are children of young mothers disadvantaged because of their mother's age or family background? Child Dev 2003;74 (2) 465- 474PubMedGoogle Scholar 34. Achenbach TM The child behavior profile, I: boys aged 6-11. J Consult Clin Psychol 1978;46 (3) 478- 488PubMedGoogle Scholar 35. Peterson JLZill N Marital disruption, parent-child relationships, and behavior problems in children. J Marriage Fam 1986;48 (2) 295- 307Google Scholar 36. Lahey BBVan Hulle CAWaldman IDRodgers JLD'Onofrio BMPedlow SRathouz PKeenan K Testing descriptive hypotheses regarding sex differences in the development of conduct problems and delinquency. J Abnorm Child Psychol 2006;34 (5) 737- 755PubMedGoogle Scholar 37. Fergusson DMHorwood LJ Male and female offending trajectories. Dev Psychopathol 2002;14 (1) 159- 177PubMedGoogle Scholar 38. Lahey BBMcBurnett KLoeber R Are attention-deficit hyperactivity disorder and oppositional defiant disorder developmental precursors to conduct disorder? Sameroff ALewis MMiller Seds Handbook of Developmental Psychopathology. 2nd New York, NY Plenum2000;431- 446Google Scholar 39. Raudenbush SWBryk AS Hierarchical Linear Models: Applications and Data Analysis Methods. 2nd Thousand Oaks, CA Sage Publications2002; 40. Brennan PAGrekin ERMednick SA Maternal smoking during pregnancy and adult male criminal outcomes. Arch Gen Psychiatry 1999;56 (3) 215- 219PubMedGoogle Scholar 41. Streissguth APMartin DCMartin JCBarr HM The Seattle longitudinal prospective study on alcohol and pregnancy. Neurobehav Toxicol Teratol 1981;3 (2) 223- 233PubMedGoogle Scholar 42. Willford JLeech SDay N Moderate prenatal alcohol exposure and cognitive status of children at age 10. Alcohol Clin Exp Res 2006;30 (6) 1051- 1059PubMedGoogle Scholar 43. Mendle JTurkheimer ED’Onofrio BMLynch SKEmery RESlutske WSMartin NG Family structure and age at menarche: a children-of-twins approach. Dev Psychol 2006;42 (3) 533- 542PubMedGoogle Scholar 44. Neuhaus JMMcCulloch CE Separating between- and within-cluster covariate effects by using conditional and partitioning methods. J R Stat Soc Ser B 2006;68 (5) 859- 872Google Scholar 45. Little RJARubin DB Statistical Analysis With Missing Data. New York, NY John Wiley & Sons Inc1987; 46. Kim JMueller CW Standardized and unstandardized coefficients in causal analysis: an expository note. Sociol Methods Res 1976;4 (4) 423- 438Google Scholar 47. Schneider MLMoore CFKraemer GW Moderate alcohol during pregnancy: learning and behavior in adolescent rhesus monkeys. Alcohol Clin Exp Res 2001;25 (9) 1383- 1392PubMedGoogle Scholar 48. Allan AMChynoweth JTyler LACaldewell KKA A mouse model of prenatal ethanol exposure using a voluntary drinking paradigm. Alcohol Clin Exp Res 2003;27 (12) 2009- 2016PubMedGoogle Scholar 49. Wass TSMattson SNRiley A Neuroanatomical and neurobehavioral effects of heavy prenatal alcohol exposure. Brick Jed Handbook of the Medical Consequences of Alcohol and Drug Abuse. New York, NY Haworth Press2004;139- 169Google Scholar 50. Mattson SNSchoenfield AMRiley EP Teratogenic effects of alcohol on brain and behavior. Alcohol Res Health 2001;25 (3) 185- 191PubMedGoogle Scholar 51. Coles CDPlatzman KARaskind-Hood CLBrown RTFalek ASmith IE A comparison of children affected by prenatal alcohol exposure and attention deficit, hyperactivity disorder. Alcohol Clin Exp Res 1997;21 (1) 150- 161PubMedGoogle Scholar 52. Bobo JKKlepinger DHDong FB Changes in the prevalence of alcohol use during pregnancy among recent and at-risk drinkers in the NLSY cohort. J Womens Health (Larchmt) 2006;15 (9) 1061- 1070PubMedGoogle Scholar 53. Reich WTodd RDJoyner CANeuman RJHeath AC Reliability and stability of mothers' reports about their pregnancies with twins. Twin Res 2003;6 (2) 85- 88PubMedGoogle Scholar 54. Heath ACKnopik VSMadden PAFNeuman RJLynskey MJSlutske WSJacob TMartin NG Accuracy of mothers' retrospective reports of smoking during pregnancy: comparison with twin sister informant ratings. Twin Res 2003;6 (4) 297- 301PubMedGoogle Scholar 55. Alvik AHaldorsen TGroholt BLindemann R Alcohol consumption before and during pregnancy comparing concurrent and retrospective reports. Alcohol Clin Exp Res 2006;30 (3) 510- 515PubMedGoogle Scholar 56. Brookes KJMill JGuindalini CCurran SXu XKnight JChen CKHuang YSSethna VTaylor EChen WBreen GAsherson P A common haplotype of the dopamine transporter gene associated with attention deficit/hyperactivity disorder and interacting with maternal use of alcohol during pregnancy. Arch Gen Psychiatry 2006;63 (1) 74- 81PubMedGoogle Scholar

Journal

Archives of General Psychiatry – American Medical Association

Published: Nov 1, 2007

There are no references for this article.