OBJECTIVE: This study examines the relationships between the dose, pattern, and timing of prenatal alcohol exposure and achievement in reading, writing, spelling, and numeracy in children aged 8 to 9 years.
METHODS: Data from a randomly selected, population-based birth cohort of infants born to non-Indigenous women in Western Australia between 1995 and 1997 (n = 4714) (Randomly Ascertained Sample of Children born in Australia’s Largest State Study cohort) were linked to the Western Australian Midwives’ Notification System and the Western Australian Literacy and Numeracy Assessment statewide education testing program. The records for 86% (n = 4056) of the cohort were successfully linked with education records when the children were aged 8 to 9 years. The associations between prenatal alcohol exposure and achievement of national benchmarks in school numeracy, reading, spelling, and writing tests and nonattendance for the tests was examined. Logistic regression was used to generate adjusted odds ratios (aOR) and 95% confidence intervals (CI), adjusting for potential confounding factors. The referent group included children of mothers who previously drank alcohol but who abstained during pregnancy.
RESULTS: Children were twice as likely not to achieve the benchmark for reading after heavy prenatal alcohol exposure during the first trimester (aOR 2.26; 95% CI 1.10–4.65) and for writing when exposed to occasional binge drinking in late pregnancy (aOR 2.35; 95% CI 1.04–5.43). Low-moderate prenatal alcohol exposure was not associated with academic underachievement.
CONCLUSIONS: The type of learning problems expressed depends on the dose, pattern, and timing of prenatal alcohol exposure.
- aOR —
- adjusted odds ratio
- CI —
- confidence interval
- OR —
- odds ratio
- RASCALS —
- Randomly Ascertained Sample of Children born in Australia’s Largest State Study
- WALNA —
- Western Australian Literacy and Numeracy Assessment
What’s Known on This Subject:
Children exposed to alcohol prenatally have increased risk of a range of developmental problems such as language delay, behavior problems, learning, and memory and cognitive deficits, all of which can have a negative impact on educational achievement.
What This Study Adds:
The expression of learning problems varied with the dose, pattern, and timing of prenatal alcohol exposure. Test scores below national benchmarks for reading were associated with first trimester heavy alcohol exposure and for writing after late pregnancy occasional binge drinking.
Prenatal alcohol exposure increases the risk of a range of neurodevelopmental disorders classified under the umbrella term fetal alcohol spectrum disorders (FASD).1 Children with such disorders can have a range of developmental problems such as language delay, behavior problems, and cognitive deficits in IQ,2 learning, memory,3–7 attention,6,8,9 and motor functioning,6 all of which can have a negative impact on educational achievement.10
Research has demonstrated an association between prenatal alcohol exposure and poor performance in preacademic skills that predict numerical problem solving and reading proficiency.11,12 Learning and memory deficits are reported to be the most common cognitive effects of prenatal alcohol exposure,5 with reports of deficits occurring in verbal,3,5,7 nonverbal,4 and visuospatial learning.5
The quantity and timing of prenatal alcohol exposure associated with deficits in child learning ability varies across studies and the type of outcome examined. Goldschmidt (1996)13 reported a linear dose-dependent relationship for deficits in arithmetic and a threshold effect with increased deficits in reading and spelling ability from exposure in the second trimester to an average of 1 drink per day or more. Overall, the pattern of maternal drinking during pregnancy that is most strongly associated with learning problems is binge drinking (5+ standard drinks/occasion) and the outcome has been shown to be related to the timing of exposure.4,5,10 Prenatal alcohol exposure at binge levels in the first trimester was reported to be associated with deficits in verbal learning and memory at 14 years of age,5 and reductions in reading ability at 10 years of age have been reported after binge exposure in the second trimester of pregnancy.14
In these studies, learning ability has generally been measured by using psychometric assessments4,5,10,13,14 or teacher’s rating of student performance10,14 which is unstandardized and open to subjective interpretation. It is important to clarify whether the reported deficits in learning are of a magnitude that results in underachievement in standardized school-based examinations, which are objective assessments, based on national benchmarks, and used to identify students not meeting the minimum expected standards.15 To our knowledge, only 1 study has used school-based achievement tests to examine this association.10 It is also important to consider that absenteeism can directly influence and/or be an indicator of educational risk and that factors stemming from the school/community environment, the student, family, or social issues can influence school attendance.16,17 School attendance is a parental responsibility so parental risk factors, such as maternal alcohol problems, may have an impact on attendance.16,18
This study examined the relationships between the dose, pattern, and timing of prenatal alcohol exposure and achievement in reading, writing, spelling, and numeracy at age 8 to 9 years in the third year of primary school and being absent for the test. Data from a population-based cohort of non-Indigenous children in Western Australia were linked with education data from the Western Australian Literacy and Numeracy school test assessments for children aged 8 to 9 years in their third year of school. We hypothesized that children exposed prenatally to heavy or binge levels of alcohol will have increased risk of not achieving national education benchmarks and/or being absent for the examination, that the fetus will be vulnerable to the effects of alcohol throughout pregnancy, but that the type of outcome affected will be dependent on the timing of the exposure. We did not anticipate any increased risk from low or moderate levels of prenatal alcohol exposure.
Ethics approval for the conduct of this study was granted by the Curtin University Human Research Ethics Committee and the Western Australian Department of Health Human Research Ethics Committee.
The full details of the cohort used for this study have been described previously.19–22 Briefly, at 12 weeks after delivery, a 10% random sample of all women giving birth in Western Australia between 1995 and 1997 were invited to participate in a postal survey of health-related behaviors and events during pregnancy and infancy (subsequently known as the RASCALS Study; Randomly Ascertained Sample of Children born in Australia’s Largest State study), designed to survey health-related behaviors and events before and during pregnancy and in early infancy. An 81% response resulted in 4861 completed questionnaires of which 4860 were able to be linked to their corresponding birth information on the Western Australian Midwives’ Notification System, a statutory population-based surveillance system of all births in Western Australia. For this study, the RASCALS data were restricted to women with singleton births (multiples n = 66) and non-Indigenous mothers (Indigenous n = 75), giving a sample size of 4719 participants.
The Western Australian Literacy and Numeracy Assessment (WALNA) is a statewide testing program undertaken in August each year for all students in all government and private schools in years 3, 5, and 7.15 The majority of students in each of these school years will be 8, 10, and 12 years of age, respectively. Students are tested in reading, writing, spelling, and numeracy. National benchmarks are agreed standards of performance that professional educators throughout Australia consider represent the minimum levels required. The benchmark cutoffs for each domain were provided by the Western Australian Education Department; reading ≤170, writing for years 2004 and earlier ≤174 and for years 2005+ ≤145, spelling ≤171, and numeracy ≤245.
The RASCALS birth cohort data were linked with the year 3 WALNA data15 by the Western Australian Data Linkage Unit. Cohort data were successfully linked with the education data for 4056 (86%) children. It was expected that between 10% and 15% of the cohort data would not link with education data, in part because of the use of probabilistic matching for the linkage of education data, which is limited when there are changes in maternal and child surnames and in part due to movement out of state. The majority (94.6%) of children in year 3 were 8 years of age, with 5.2% aged 9 years and the remainder aged 7 years. Information was also obtained on the percentage of children absent for the WALNA tests.
Examination of the characteristics of the mothers of children whose survey data did not link with education data was undertaken by using Pearson χ2 analysis. Linkage with educational data was less likely to occur for children of mothers drinking in first trimester at heavy or binge levels than for abstinent women (ρ = 0.02) or when languages other than English were spoken in the home (ρ < 0.01) and more likely to occur for women with less than a year 12 education (equivalent to high school graduation) than women with a degree/diploma/trade (ρ < 0.01). There were no differences in the proportions successfully linking for maternal alcohol consumption in prepregnancy or late pregnancy or for other maternal demographic variables, maternal age, ethnicity, marital status, parity, socioeconomic status, smoking, or other drug use.
Information about maternal alcohol consumption was collected at 3 months after delivery for the 3 month period prepregnancy and for each trimester separately. Details of the alcohol questions and coding have been published previously.23,24 To examine the impact of the pattern of drinking by women before and during pregnancy, we took into account the frequency, quantity per occasion, and total quantity consumed. Women who were abstinent throughout pregnancy were divided into “never” drinkers and “previous” drinkers, with the “previous” drinkers comprising the referent group. The categories for women drinking alcohol during pregnancy were low (1–2 standard drinks/occasion and <7 per week), moderate (3–4 standard drinks/occasion and no more than 7 per week), binge drinking (50+ g per occasion) less frequently than weekly (occasional binge drinking), and heavy drinking (>7 standard drinks per week, including binge drinking weekly or more frequently). One standard drink in Australia is equal to 10 g of alcohol.
The association between prenatal alcohol exposure during pregnancy and not achieving the benchmarks in numeracy, reading, spelling, and writing in year 3 of primary school was assessed by using multivariate logistic regression analysis. To examine whether any associations might reflect language delay, the data were restricted to children who had normal language development as assessed at 2 years of age (n = 1284; 31%) and reanalyzed.23 We also assessed the association between prenatal alcohol exposure and absenteeism on the day of the test. The alcohol categories for occasional binge drinkers and heavy drinkers were combined for the analyses of absenteeism due to small numbers. The referent group for all analyses was “previous” drinkers who abstained from alcohol throughout pregnancy.25
The analyses were adjusted for potential confounders: maternal smoking, illicit drug use (tranquillizers, marijuana, ecstasy, amphetamines, heroin, methadone, cocaine, LSD and volatile substances) during pregnancy, maternal education (degree/diploma/trade, year 12 high school graduation, less than year 12), languages spoken at home (English only, other languages spoken), and income (most advantaged, middle, most disadvantaged, and missing; 1%); all self-reported 3 months after delivery. Maternal demographic variables identified from the Western Australian Midwives’ Notification System were maternal age, parity, ethnicity, and marital status. Because self-reported income was missing for 17% of the cohort, the relative Index for Socioeconomic Disadvantage,26 an area based measure of deprivation, was applied as a proxy for missing cases.
Data analyses were conducted by using SPSS version 19.0, and results are presented as odds ratios (OR), adjusted OR (aOR), and 95% confidence intervals (CI).
The records for 86% (n = 4056) of the RASCALS cohort were successfully linked with year 3 education records. Of these, 40.9% of mothers did not consume alcohol during pregnancy, of which 33% were previous drinkers (Table 1). The percentage of women drinking at low levels increased from 33.6% in the first trimester to 41.1% in late pregnancy, whereas the proportion binge drinking occasionally or drinking heavily reduced during pregnancy. Approximately 17% of mothers did not drink in the first trimester but drank in late pregnancy, and 8.2% drank in the first trimester but abstained in late pregnancy.
Mothers <25 years of age were less likely to drink alcohol at low levels than women 25 to 29 years and more likely to binge drink in the first trimester (Table 2). Conversely, mothers aged ≥30 years were more likely to drink at low levels and to be moderate or heavy drinkers in late pregnancy. Compared with married mothers, women who were in a de facto relationship (cohabiting) or who were single were more likely to drink heavily and to binge drink. Mothers who had not completed year 12 (completion of secondary education) were more likely to binge drink occasionally in late pregnancy than mothers with a degree. Women in the low-income category were more likely to be never drinkers than women in the highest category and high-income mothers were more likely to be drinking at low and moderate levels (Table 2). Mothers who did not drink alcohol during pregnancy were less likely to smoke or use other drugs (results not presented).
A higher percentage of children of mothers drinking heavily during the first trimester of pregnancy did not achieve the benchmark for reading (9.0%), spelling (17.1%), and writing (15.2%) than children in the referent group: 4.0%, 15.1%, 12.7%, respectively (Table 3). Occasional binge drinking at any stage during pregnancy was also associated with a higher percentage of children not achieving the benchmark in spelling and writing. Conversely, a lower percentage of children of mothers drinking at low and moderate levels during pregnancy did not achieve the benchmarks in each of the 4 domains than the children of previous drinkers who abstained during pregnancy (Table 3).
The association between prenatal alcohol exposure and educational achievement at year 3 was influenced by the dose, pattern, and timing of exposure (Tables 3 and 4). Compared with children of mothers who previously drank alcohol but who did not drink alcohol during pregnancy, children of mothers drinking heavily during the first trimester had twice the odds of not achieving the benchmark for reading (aOR 2.27; 95% CI 1.11–4.67; Table 3). Conversely, children of mothers drinking at low levels during the first trimester of pregnancy were 27% less likely to be below the benchmark for numeracy than children of mothers who ceased drinking during pregnancy (aOR 0.73; 95% CI 0.54–0.98). Occasional binge drinking during late pregnancy more than doubled the odds of children not achieving the benchmark in writing (aOR 2.33; 95% CI 1.03–5.29; Table 4). The results did not materially alter when the analyses were restricted to the children assessed as having normal language development at 2 years of age (results not presented).
Children of mothers who were binge drinking occasionally or drinking heavily at any stage during pregnancy were more likely not to have sat the spelling test; the first trimester aOR 2.74 (95% CI 1.21–6.21) and late pregnancy aOR 2.36 (95% CI 0.85–6.56; Table 5). Of the children prenatally exposed to heavy levels of alcohol and who did not sit the spelling test, ∼60% were below the benchmark in writing, irrespective of the timing of exposure (results not presented). Prenatal exposure at binge or heavy levels in late pregnancy was also associated with not sitting the numeracy test; however, the estimate was imprecise due to small numbers (aOR 1.90; 95% CI 0.64–5.66; Table 5).
This study adds new, population-based evidence that prenatal alcohol exposure at binge and heavy levels increases the risk of specific learning problems resulting in these children not achieving the minimum expected academic standards in reading and writing and in being absent for the spelling tests. Our results indicate that reading ability was sensitive to first trimester exposure and writing ability to late pregnancy exposure, which supports our hypotheses that the fetus is vulnerable to the effect of alcohol throughout pregnancy and that the type of subsequent learning problem expressed may be related to the dose, pattern, and timing of exposure.
Exposure to heavy levels of alcohol during the first trimester of pregnancy increased the odds of not achieving the reading benchmark by more than twofold. Previous studies have demonstrated a dose-dependent relationship between prenatal alcohol exposure and phonologic processing,11 and verbal learning and memory deficits have been reported in 14-year-old children who were exposed to binge levels of alcohol in the first trimester.5 It is possible that these underpin the reading deficits reported in our study. Although the negative effect of heavy/frequent binge drinking on reading problems supports the findings of Goldschmidt et al,13,14 the sensitive period differs between the studies. The results of our study indicate that the first trimester is the sensitive period for reading problems, in contrast to the findings of Goldschmidt et al, which indicated that the sensitive period is the second trimester. This difference may relate to the prospective data collection used by Goldschmidt in contrast to the retrospective data collection for our study because it has been reported that antenatal alcohol surveys provide the most valid estimate to use when investigating neurobehavioral outcome.27 However, there were small numbers of children in our study exposed to binge or heavy levels of alcohol during late pregnancy who had not achieved the reading benchmark. The small numbers limited our ability to examine this association, and so we are cautious about drawing any firm conclusions from our results about the effect on reading ability from binge and heavy prenatal alcohol exposure in the second and/or third trimester.
Children of mothers who reported occasional binge drinking in late pregnancy were twice as likely to have writing scores below the benchmark than comparison children. This reflects the same pattern and timing of prenatal alcohol exposure that was associated with language delay in RASCALS children at 2 years of age.23 Restricting the analyses to children assessed as having normal language development at age 2 years did not materially alter the results, suggesting that the effect of occasional binge drinking in late pregnancy may be working through multiple pathways.
In contrast to an earlier study demonstrating an association between prenatal alcohol exposure in the second trimester and poor spelling ability,13 our study did not find any association between underachievement in spelling and prenatal alcohol exposure. This may be due to the high-risk students in this study being absent for the spelling assessment. School attendance may directly cause or act as an indicator of educational risk,16 and in this study, children of mothers who were drinking heavily or occasionally binge drinking in the first trimester of pregnancy were more than 2.5 times more likely to be absent from the spelling test. A similar estimate was observed for children exposed at these levels in late pregnancy, although the estimate was imprecise reflecting the small numbers. Approximately 60% of these absentees were below the benchmark in writing, so it is possible that they also had problems with spelling.
Our results did not demonstrate an association between prenatal alcohol exposure and numeracy ability, which is in contrast to some earlier studies.11,12 These other studies reported that alcohol-exposed children were deficient in premath skills based on testing that measured cognitive and intellectual functioning. However, measures of general cognitive ability only account for around half of the variance associated with academic achievement,28 which is what was assessed in our study. Other factors, such as the school learning environment, may have played a role in determining the numeracy results in our study, whereas the mean age of children in the Coles et al12 study was <6 years, and thus the children had fewer, if any, years in school than the children in our study. In our study, children exposed at heavy/binge levels in late pregnancy were almost twice as likely to be absent from the numeracy assessment than comparison children, although the estimates were imprecise because of small numbers. Although it is possible that absenteeism contributed to our null result, the small numbers of absent children make this unlikely to have greatly affected the findings.
Factors crucial to a child’s educational achievement include contributions from the child, home, school, teacher, curricula, and teaching approaches,29 and these need to be taken into account when examining educational achievement. In this study, exposure data were collected by a self-administered questionnaire about a range of antenatal maternal behaviors and family factors, so the focus was not solely on alcohol consumption. The use of self-administered questionnaires are reported to obtained more valid responses on socially sensitive issues such as alcohol consumption and reveal more unwanted behavior than interviews.30 The wide range of factors ascertained in our study enabled adjustment for potentially confounding maternal and family demographic factors and maternal risk behaviors during pregnancy. However, we were unable to adjust for maternal alcohol consumption and other risk behaviors through the child’s early years. These factors may be associated with less than ideal postnatal environment,31 which may also affect school test scores and absenteeism and may potentially have contributed to our findings.
As hypothesized, low and moderate levels of prenatal alcohol exposure were not associated with poor test score achievement. Although this is reassuring for pregnant women who have consumed alcohol at these levels before pregnancy awareness, these results should not be overinterpreted as indicating that drinking at these levels is not potentially harmful to the fetus. Neurodevelopment is sensitive to prenatal alcohol exposure and other studies have demonstrated fetal effects for other outcomes from moderate levels of alcohol.24,32 It is also possible that retrospective data collection may have limited our ability to detect an association at lower levels of alcohol exposure. Ideally, this study should be repeated by using prospectively collected data on alcohol consumption, with information on potential confounding factors such as maternal alcohol consumption and other risk behaviors occurring during the child’s early years and with larger numbers.
Strengths of this study include the use of data from a randomly selected population-based birth cohort with exposure information collected well ahead of the outcomes, which were derived from linkage with data from the WALNA program. The use of routinely administered school tests in this study allowed objective comparison of academic ability based on national benchmarks, making the results of this study generalizable to other similar, non-Aboriginal populations. Linkage of statewide education data enabled follow-up of the test scores for 86% of the RASCALS children at 8 to 9 years of age, overcoming possible reporting bias, loss to follow-up, and the potential problems and cost of following up and testing a large cohort of children. Although linkage was less likely for children whose mothers reported drinking at heavy/binge levels and who were potentially at risk of not achieving the benchmarks for reading, this would likely have biased our results toward the null.
This study provides new information about the relationship between the dose, pattern, and timing of prenatal alcohol exposure and school test achievement. Heavy prenatal alcohol exposure in the first trimester was associated with a reading score below the national benchmark, and occasional binge drinking in late pregnancy was associated with a writing score below the national benchmark. Children of mothers who drank at heavy or binge levels during pregnancy were more likely to be absent from some tests than children of mothers who abstained throughout pregnancy.
We thank Margaret Wood and Peter Cosgrove for maintenance of the databases. We thank the staff of the Western Australia (WA) Data Linkage Unit for access to the WA Data Linkage System and for their assistance in obtaining the data and the WA Health Data Custodians for access to the core health datasets. We are grateful to the parents who have participated in the RASCALS study and for the support of the Telethon Institute for Child Health Research, in particular, the RASCALS study team. Approval to access the Department of Education data were granted by the Developmental Pathways Project, funded by ARC Linkage grant LP100200504.
- Accepted April 29, 2013.
- Address correspondence to Colleen O’Leary BSc, MPH, PhD, Centre for Population Health Research, Curtin University, GPO Box U1987, Perth, Western Australia, 6845 Australia. E-mail:
Dr O’Leary designed this study, conducted the analyses, and drafted the manuscript; Dr Kurinczuk designed and conducted the original Randomly Ascertained Sample of Children born in Australia’s Largest State Study longitudinal study and reviewed the results of this analysis; Drs O’Leary and Bower provided expertise in alcohol and pregnancy; Drs Taylor and Zubrick provided expertise in child development and educational achievement; and all authors edited the manuscript and approved the final manuscript as submitted.
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: The Western Australian survey of health-related behaviors and events during pregnancy and early infancy was funded by grants from Healthway (the Western Australian Health Promotion Foundation; grants 94/2705, 96/49078, and 98/8016), to which we are most grateful. This study was supported by an Australian National Health and Medical Research Council (NHMRC) Public Health (Australia) Fellowship (594451) (Dr O’Leary), NHMRC program grant 353514 (2005-09), and an NHMRC Research Fellowship (353628; Dr Bower). Dr O’Leary was also supported by infrastructure grants from Curtin University and the Western Australian Drug and Alcohol Office. The funding agencies had no involvement in this study.
- ↵National Center on Birth Defects and Developmental Disabilities. Fetal Alcohol Syndrome: Guidelines for Referral and Diagnosis. Atlanta, CA: Centers for Disease Control & Prevention; 2004
- Howell KK,
- Lynch ME,
- Platzman KA,
- Smith GH,
- Coles CD
- Streissguth AP,
- Barr HM,
- Olson HC,
- Sampson PD,
- Bookstein FL,
- Burgess DM
- ↵Western Australian Department of Education and Training. Performance of Years 3, 5, and 7 students. Perth, Australia: WALNA; 2007
- ↵Western Australian Auditor General’s Report. Every Day Counts: Managing Student Attendance in Western Australian Public Schools. Perth, Australia: Government of Western Australia; 2009
- Gottfried MA
- Burke S,
- Schmied V,
- Montrose M
- Straker LM,
- Pollock CM,
- Zubrick SR,
- Kurinczuk JJ
- O’Leary CM,
- Bower C,
- Zubrick SR,
- Geelhoed E,
- Kurinczuk JJ,
- Nassar N
- O’Leary C,
- Zubrick SR,
- Taylor CL,
- Dixon G,
- Bower C
- ↵Kelly YJ, Sacker A, Gray R, et al. Light drinking during pregnancy: still no increased risk for socioemotional difficulties or cognitive deficits at 5 years of age [published online ahead of print October 5, 2010]? J Epidemiol Community Health. doi:10.1136/jech.2009.103002
- ↵Australian Bureau of Statistics. Socio-economic Indexes for Areas (Report No.: 2039.0). Canberra, Australia: Australian Bureau of Statistics; 2001
- Jacobson SW,
- Chiodo LM,
- Sokol RJ,
- Jacobson JL
- Hattie J
- Copyright © 2013 by the American Academy of Pediatrics