The effect of maternal alcohol consumption on fetal growth and preterm birth

Authors

  • CM O’Leary,

    Corresponding author
    1. Division of Population Sciences, Telethon Institute for Child Health Research, Centre for Child Health Research, University of Western Australia, Perth, WA, Australia
    Search for more papers by this author
  • N Nassar,

    1. Division of Population Sciences, Telethon Institute for Child Health Research, Centre for Child Health Research, University of Western Australia, Perth, WA, Australia
    Search for more papers by this author
  • JJ Kurinczuk,

    1. National Perinatal Epidemiology Unit, University of Oxford, Headington, Oxford, UK
    Search for more papers by this author
  • C Bower

    1. Division of Population Sciences, Telethon Institute for Child Health Research, Centre for Child Health Research, University of Western Australia, Perth, WA, Australia
    Search for more papers by this author

Ms CM O’Leary, Division of Population Sciences, Telethon Institute for Child Health Research, Centre for Child Health Research, University of Western Australia, PO Box 855, West Perth, WA 6872, Australia. Email colleeno@ichr.uwa.edu.au

Abstract

Objective  To investigate the relationship between prenatal alcohol exposure and fetal growth and preterm birth and to estimate the effect of dose and timing of alcohol exposure in pregnancy.

Design  A population-based cohort study linked to birth information on the Western Australian Midwives Notification System.

Setting  Western Australia.

Population  A 10% random sample of births restricted to nonindigenous women who had delivered a singleton infant (n= 4719) in 1995–1997.

Methods  The impact of alcohol consumption in pregnancy on fetal growth (small-for-gestational-age [SGA] and large-for-gestational-age infants [LGA]) and preterm birth (<37 weeks of gestation) was assessed using multivariate logistic regression analysis and adjusting for confounding factors.

Main outcome measures  Odds ratios and 95% CI, attributable risk, and population attributable risk were calculated.

Results  The percentage of SGA infants and preterm birth increased with higher levels of prenatal alcohol exposure; however, the association between alcohol intake and SGA infants was attenuated after adjustment for maternal smoking. Low levels of prenatal alcohol were not associated with preterm birth; however, binge drinking resulted in a nonsignificant increase in odds. Preterm birth was associated with moderate and higher levels of prenatal alcohol consumption for the group of women who ceased drinking before the second trimester. This group of women was significantly more likely to deliver a preterm infant than women who abstained from alcohol (adjusted OR 1.73 [95% CI 1.01–3.14]).

Conclusions  Alcohol intake at higher levels, particularly heavy and binge drinking patterns, is associated with increased risk of preterm birth even when drinking is ceased before the second trimester. This finding, however, is based on small numbers and needs further investigation. Dose and timing of prenatal alcohol exposure appears to affect preterm delivery and should be considered in future research and health education.

Introduction

The evidence surrounding the effect of low to moderate intake of alcohol during pregnancy on fetal growth and preterm birth is inconclusive. While there is a large body of literature on the issue, the evidence base has many weaknesses limiting our ability to reach definitive conclusions. In their systematic review of the literature, Henderson et al. (2007)1 reported that many studies did not control for known confounding factors, such as cigarette smoking and ethnicity. In their more detailed report,2 the authors found that the studies that had adjusted for confounding factors had other limitations, which prevented their results from being generalised to the wider community.

In the few studies that have reported an association between low levels of prenatal alcohol exposure and fetal growth, the direction of the association has not been consistent. While the majority of studies have reported no association with less than 72 g of alcohol per week (equivalent to seven standard drinks per week in Australia; six in the USA, Canada, and Europe; and nine units in the UK) and low birthweight,3–9 intrauterine growth restriction,5,10,11 and preterm birth,8,10,12–17 a small number of studies found an increased risk at low levels18–20 and, conversely, others have reported a possible protective effect of low levels of alcohol consumption in pregnancy.10,13,18,21–23 At higher levels of prenatal alcohol exposure, the findings of an association between prenatal alcohol exposure and fetal growth are not consistent, with around half of studies reporting no significant association3,4,10,18,24–31 and half reporting a significant association.9,20,32–40 Uncertainty also exists about the impact of binge drinking on intrauterine growth and questions remain as to whether the increased risk from binge drinking, if one truly exists, is due to the pattern of binge drinking per se or rather a result of heavy alcohol intake.2

Using data from a population-based cohort study of nonindigenous women in Western Australia (WA), this study examines the impact of maternal alcohol consumption, taking into account the quantity per occasion, frequency of consumption, and total quantity consumed during the 3 months before and during each trimester of pregnancy on preterm birth and fetal growth.

Materials and methods

The details of the study used for this analysis have been described previously.41–43 Briefly, a 10% random sample of all women giving birth in WA between 1995 and 1997 was invited by letter at 12 weeks postpartum to participate in a postal survey of health-related behaviours and events during pregnancy and infancy (subsequently known as the RASCALS study), designed to survey health-related behaviours and events before and during pregnancy and early infancy. Data were collected using a postal questionnaire sent with the letter of invitation. Mothers whose infants were stillborn (n= 20) or given up for adoption (n= 5) were excluded. An 81% response rate resulted in 4861 completed questionnaires of which 4860 were able to be linked to their corresponding birth information on the WA Midwives’ Notification System, a statutory population-based surveillance system of all births in WA. The analysis reported here was restricted to women with singleton births (multiples n= 66) and nonindigenous mothers (indigenous n= 75), giving a sample size of 4719.

Comparison with data available for all births in WA in this period44 showed that the respondents were representative of mothers of all singleton live births with the exception of a slight underrepresentation of mothers with low-birthweight babies (5.3% overall versus 4.7% respondents) and mothers aged less than 20 years (6.0% overall versus 3.6% respondents; 2.5% in this sample). Ethics approval for the conduct of this study was granted by the Princess Margaret Hospital Research Ethics Committee and the WA Confidentiality of Health Information Committee.

Information about maternal alcohol consumption was collected retrospectively for the 3-month period prepregnancy and for each trimester separately. For each period, women were asked how often they drank alcohol (5 or more, 3–4, or 1–2 days/week; 1–2 days/month; less than once per month; or never) and the quantity consumed (e.g. number of cans, glasses, bottles) on a typical occasion for each of four types of alcoholic beverages (beer, wine/champagne, spirits/liqueurs, and fortified wines). Consumption frequency calculations used the lower of the days marked, for example 3–4 days/week was included as 3 days/week to calculate total weekly dose of alcohol. However, there were a small number of women (n= 7 in first trimester and n= 1 in third trimester) who reported a frequency of drinking of one to two times per week and who consumed two or more types of beverages each at less than 50 g per occasion, but with a total weekly consumption of 70+ g. As we could not be confident that the women had consumed only once per week, and therefore, at binge levels, we coded them as heavy drinkers. Where respondents used a tick mark instead of indicating a numeric value, a minimum level for the type of beverage, time period, and frequency was applied. Standard drink calculations were derived during the data analysis stage and covered a range of measures for each type of beverage;42 more details are available on request from the authors.

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. The level of alcohol consumption was categorised into five mutually exclusive groups; none, low, moderate, binge (less than weekly up to twice per week), and heavy (including women who binged more than twice per week) (Table 1). For the analyses, abstinence during pregnancy refers to women who reported not drinking at any stage during pregnancy. One standard drink in Australia is equal to 10 g of alcohol. The ‘low’ category was defined in line with the 2001 recommendation to women who are pregnant or might soon become pregnant set out by the Australian National Health and Medical Research Council alcohol guideline 11, which recommends that ‘If women choose to drink, over a week, should have less than 7 standard drinks, AND, on any one day, no more than 2 standard drinks’.45 To assess the overall impact of alcohol intake greater than the low level, moderate, heavy, and binge drinking were also combined.

Table 1.  Classification of maternal alcohol consumption
 Alcohol consumption (g)
LowModerate*Binge <weekly to one to two times per weekHeavy
Frequency per week<Weekly** to <daily<Weekly** to daily<Weekly** to 2 days2+ days
Grams per occasion***≤2020–<5050+>10–50+
Grams per week***≤60≤70≥5≥68
  • *

    Women reporting consuming 10 g of alcohol per occasion on a daily basis are included in the moderate group.

  • **

    <Weekly = one to two times per month to once every 8–10 weeks.

  • ***

    10 g = one standard drink in Australia and 50 g per occasion = binge drinking.

Prepregnancy
Mean9.321.560.3208.7
Median3.812.550.6141.0
Minimum0.52.15.068.0
Maximum60.070.0320.03685.0
Trimester 1
Mean6.216.658.0192.5
Median2.58.050.0150.0
Minimum0.52.15.071.0
Maximum60.067.5270.01453.0
Trimester 2
Mean6.014.648.0161.1
Median2.57.525.3120.0
Minimum0.52.15.075.0
Maximum60.066.0265.0540.0
Trimester 3
Mean6.015.247.0143.2
Median2.57.625.0105.0
Minimum0.53.05.074.0
Maximum60.070.0265.0540.0

The primary outcomes of the study were the effect of alcohol consumption in pregnancy on fetal growth and preterm birth. Appropriate fetal growth was ascertained using the proportion of optimal birthweight (POBW), where optimal birthweight was determined after taking into account infant sex, gestational age, maternal height, and parity. The POBW was then calculated by taking the ratio of observed birthweight to optimal birthweight.46 The population selected to define optimal birthweight was the total 1998–2002 WA population of singleton, Caucasian births, not exposed to factors known to influence fetal growth pathologically. We used a POBW score less than the 10th percentile to define small-for-gestational-age (SGA) infants. Preterm birth was defined as infants born at less than 37 weeks of gestation. Gestational age was estimated using an algorithm taking into account two independent estimates of gestational duration from routinely collected data (last menstrual period, expected due date, ultrasound fetometry, baby’s date of birth, and neonatal estimate of gestational age) by the WA Midwives’ Notification System.47

The effect of alcohol consumption in specific periods of pregnancy was examined by investigating infant outcomes for women who only drank in the first trimester, those who drank in the first trimester regardless of whether they stopped or continued drinking later in pregnancy, and the outcomes for women who drank in either trimesters 2 and/or 3 irrespective of whether they drank in first trimester, referred to as ‘late’ pregnancy. The maximum alcohol intake in each respective period was used to assign the level of drinking and where alcohol consumption was missing for the third trimester (n= 27), the second trimester alcohol consumption information was assigned.

This study had 80% power at a 95% level of confidence to detect a 50% increase in the odds of preterm birth or poor fetal growth (OR 1.50) for infants of women consuming a low level of alcohol and a 70% increase in the odds for the offspring of women consuming a moderate level of alcohol. Due to the small numbers in the specific categories, there was only limited power to detect a statistically significant difference for heavier drinking levels.

The association between alcohol consumption both before and during pregnancy and risk of SGA infants compared with infants of abstinent women was assessed using logistic regression analysis. In addition, we used the same method to assess the association between prenatal alcohol and large-for-gestational-age (LGA) infants, using a POBW score greater than the 90th percentile to define LGA infants. We also conducted generalised linear regression analysis of POBW to ensure we had not missed any information by categorising the variable. Cox regression was used to determine independent risk factors for preterm birth. The analyses were adjusted for potential confounders: maternal smoking and illicit drug use (tranquillizers, marijuana, ecstasy, amphetamines, heroin, methadone, cocaine, lysergic acid diethylamide [LSD], and volatile substances) during pregnancy, maternal age, parity, ethnicity, marital status and income (all self-reported), and maternal medical conditions (essential hypertension, pre-existing diabetes mellitus, asthma, and genital herpes); pregnancy complications (threatened miscarriage, threatened preterm labour, urinary tract infection, pre-eclampsia, placenta praevia, abruption, antepartum haemorrhage, prelabour rupture of membranes, and gestational diabetes); or pregnancy procedures (ultrasound, fertility treatments, cervical suture, chorionic villus sampling/placental biopsy, amniocentesis, and antepartum and intrapartum cardiotocogram), identified from the WA Midwives’ Notification System. As self-reported income was missing for 17% of the cohort, the relative Index for Socio-Economic Disadvantage48 was applied as a proxy for missing cases. Interaction terms were tested for but were not included in the analysis as they did not make a statistically significant (ρ < 0.01) contribution to the fit of the data. Data analyses were conducted using SPSS version 15.0, and results are presented as odds ratios with 95% CI.

The population attributable risk (PAR) and the attributable risk (AR)49 for preterm birth were calculated for alcohol exposure in the group of women who stopped drinking at moderate, heavy, and binge levels before second trimester.

Results

The quantity of alcohol consumed per week during pregnancy is presented in Table 1. Levels of alcohol intake decreased from the prepregnancy period to trimester 2 for each level of alcohol exposure, particularly the maximum intake for women with a binge or heavy drinking pattern. The median intake for women drinking at low to moderate levels decreased by approximately one-third between prepregnancy and first trimester but did not decrease markedly in late pregnancy. Conversely, there was little change in the median intake at higher levels of intake until late pregnancy when the median for binge drinkers halved and decreased by one-third for heavy drinkers. Around 9% of women drinking at heavy levels during prepregnancy consumed more than 400 g/week (data not shown).

The distribution of maternal alcohol consumption before and during pregnancy is shown in Table 2. While fewer than 20% of women abstained during the prepregnancy period, this increased to 57% in the first two trimesters, decreasing to 54% in the third trimester. Only 41% of women abstained throughout each trimester of pregnancy (results not shown). Approximately one-third of women consumed a low level of alcohol before and during pregnancy. There was a marked reduction at higher levels of consumption with the percentage of women drinking at heavy and binge levels dropping from around 10–3% between the prepregnancy and the first trimester.

Table 2.  Pattern of maternal alcohol consumption before and during pregnancy
Pattern of drinkingPrepregnancy, n (%)Trimester 1, n (%)Trimester 2, n (%)Trimester 3, n (%)
Abstinent919 (19.5)2707 (57.4)2688 (57.0)2537 (53.8)
Low1557 (33.0)1326 (28.1)1542 (32.7)1668 (35.3)
Moderate1282 (27.2)446 (9.5)367 (7.8)402 (8.50)
Binge ≤two times per week512 (10.8)131 (2.8)56 (1.2)43 (1.0)
Heavy449 (9.5)108 (2.3)66 (1.4)69 (1.5)

There were 421 (8.9%) infants who were SGA (POBW less than 10th percentile) and 265 (5.7%) infants were delivered before 37 weeks of gestation. Overall, there was little difference between the outcomes for infants of women who drank low levels of alcohol during pregnancy and those of women who were abstinent during pregnancy (Table 3). The percentage of growth-restricted infants was highest among those exposed to either maternal binge or heavy drinking during pregnancy, around 13% for each group. For preterm births, the percentage of infants born before 37 weeks was highest among those exposed to binge drinking during late pregnancy (9.5%) and for children born to mothers who drank heavily but stopped drinking before the second trimester (13.6%).

Table 3.  Distribution of SGA and gestational age by maternal alcohol consumption before and during pregnancy
 Abstinent, n (%)Low, n (%)Moderate, n (%)Binge ≤two times per week, n (%)Heavy, n (%)
  • *

    Second and/or third trimesters.

Drank during prepregnancy
Percentage optimal birthweight ≥10%844 (91.8)1398 (89.9)1177 (92.0)464 (90.6)410 (91.5)
SGA <10%75 (8.2)157 (10.1)103 (8.0)48 (9.4)38 (8.5)
Gestational age ≥37 weeks865 (94.1)1426 (93.9)1210 (94.5)489 (95.5)423 (94.4)
Preterm delivery <37 weeks54 (5.9)93 (6.0)70 (5.5)23 (4.5)25 (5.6)
Drank during trimester 1
Percentage optimal birthweight ≥10%1752 (91.2)1199 (90.6)408 (91.5)114 (87.0)94 (87.0)
SGA <10%170 (8.8)125 (9.4)38 (8.5)17 (13.0)14 (13.0)
Gestational age ≥37 weeks1812 (94.3)1252 (94.6)424 (95.1)121 (92.4)102 (94.4)
Preterm delivery <37 weeks110 (5.7)72 (5.4)22 (4.98)10 (7.6)6 (5.6)
Alcohol consumed in trimester 1 with abstinence in late pregnancy*
Percentage optimal birthweight ≥10%1752 (91.2)208 (90.0)80 (88.9)33 (89.2)20 (90.9)
SGA <10%170 (8.8)23 (10.0)10 (11.1)4 (10.8)2 (9.1)
Gestational age ≥37 weeks1812 (94.3)219 (94.8)82 (91.1)33 (89.2)19 (86.4)
Preterm delivery <37 weeks110 (5.7)12 (5.2)8 (8.9)4 (10.8)3 (13.6)
Drank during late pregnancy*
Percentage optimal birthweight ≥10%1752 (91.2)1640 (91.3)436 (92.4)55 (87.3)69 (86.3)
SGA <10%170 (8.81)157 (8.7)36 (7.6)8 (12.7)11 (13.8)
Gestational age ≥37 weeks1812 (94.3)1700 (94.6)452 (95.8)57 (90.5)75 (93.8)
Preterm delivery <37 weeks110 (5.7)97 (5.4)20 (4.2)6 (9.5)5 (6.3)

The distribution of maternal characteristics by alcohol use is shown in Table 4. The prevalence of smoking in the whole cohort was 24.6%, with 10.6% of women smoking >10 cigarettes per day. Women who reported smoking at any time during pregnancy were less likely to abstain from alcohol during their pregnancy and to comprise a larger percentage of the women who continued to drink during late pregnancy (27.7%). Predictive factors for continuing to drink during late pregnancy were maternal age 30 years and older, higher income, other drug use, Caucasians, and married women. A higher percentage of women who stopped drinking before the second trimester reported that the pregnancy had been unplanned (55%) than women who were abstinent during pregnancy (47%).

Table 4.  Maternal characteristics by alcohol consumption before and during pregnancy
 Alcohol consumption
Whole Cohort (%), n = 4719PrepregnancyDuring pregnancy
Abstinent (%), n= 919Abstinent (%), n= 1924Drank trimester 1 (%), n= 2011Stopped before trimester 2 (%), n= 380Drank during late* pregnancy (%), n= 2415
  • *

    Second and/or third trimesters. Note that numbers may vary due to missing covariate data.

  • **

    Smoking at any stage during pregnancy.

Maternal age group (n= 4717)
<202.52.434.42.33.41.8
20–2414.417.03.230.422.111.5
25–2932.334.816.613.230.930.6
30–4450.845.845.954.143.656.0
Smoking** (n= 4688)
None74.783.381.067.664.671.8
≤10 per day148.010.418.417.716.6
>10 per day10.68.88.614.017.711.1
Ethnic group (n= 4714)
Caucasian92.878.686.696.796.397.4
Other7.121.413.43.33.72.6
Marital status (n= 4700)
Married75.981.077.472.560.677.3
Defacto17.513.416.420.425.617.2
Single6.25.66.27.113.45.1
Parity (n= 4714)
First29.124.629.529.837.227.5
Second31.230.330.230.930.332.3
Third+39.645.240.339.332.440.2
Drug use during pregnancy (n= 4719)
No92.995.995.890.091.890.8
Yes7.14.14.210.08.29.2
Income (n= 4666)
Most advantaged >$40 00028.519.922.832.423.434
$25 001–40 00030.839.539.938.441.539.3
Most disadvantaged ≤$25 00139.639.536.228.234.025.7
Maternal medical conditions (n= 4717)
None81.380.880.982.380.381.8
Any18.719.219.117.719.718.2
Procedures and treatments during pregnancy (n= 4717)
None7.56.27.17.55.18.2
Ultrasound57.257.658.135.558.556.2
Other35.436.234.957.036.435.6
Pregnancy complications (n= 4719)
None72.371.171.873.169.473.1
One or more27.728.928.229.030.626.9
Unplanned pregnancy (n= 4395)
No46.946.247.143.838.341.2
Yes46.247.447.249.755.338.6

Table 5 shows the elevated odds of SGA infants with moderate to heavy alcohol consumption in pregnancy and how, after adjustment for smoking status, this effect is eliminated. There was, however, an increased odds of infants being born SGA following low levels of alcohol in the prepregnancy period (adjusted OR [aOR] 1.34, 95% CI 0.99–1.82) that attained significance in the fully adjusted generalised linear model (adjusted β−1.66, 95% CI −2.94 to −0.39) (results not shown), but this association was not observed at higher levels of alcohol in prepregnancy or in any other analyses investigating the other time periods during pregnancy. The odds of an infant being LGA were close to or below one for all levels of alcohol exposure in each time period; however, none was statistically significant.

Table 5.  OR (and 95% CI) for the association between alcohol consumption categories before and during pregnancy and SGA (POBW <10%)
Alcohol consumption and timing of exposureFrequencyUnadjustedAdjusted for smokingFully adjusted*
nSGAOR95% CIOR95% CIOR95% CI
  • *

    Adjusted for maternal age, smoking, ethnicity, marital status, parity, drug use, income, maternal medical conditions, procedures, and treatments during pregnancy and pregnancy complications.

  • **

    Second and/or third trimesters. Note that numbers may not add up to the total due to women who abstained in First Trimester but drank in Late Pregnancy (n=784, 17%) and those who abstained in Late Pregnancy but drank in First Trimester (n=380, 8%).

Prepregnancy
Abstinent91975Referent 1.0
Low15571571.270.95–1.691.240.92–1.661.340.99–1.82
Moderate12821030.990.72–1.340.960.70–1.321.050.75–1.46
Binge occasional512481.160.80–1.700.820.55–1.220.930.62–1.40
Heavy449381.040.69–1.570.810.53–1.240.830.53–1.30
Alcohol consumption during trimester 1
Abstinent1924170Referent 1.0
Low13261251.070.84–1.361.050.82–1.351.080.84–1.40
Moderate446380.960.67–1.380.850.58–1.230.870.59–1.29
Binge occasional131171.540.90–2.620.910.52–1.590.980.55–1.71
Heavy108141.540.86–2.750.840.44–1.590.820.42–1.59
Alcohol consumed in trimester 1 with abstinence in late pregnancy
Abstinent1924170Referent 1.0
Low231231.140.72–1.801.060.66–1.681.090.68–1.75
Moderate90101.290.66–2.531.110.56–2.201.100.55–2.21
Binge occasional3741.250.44–3.570.830.29–2.430.910.31–2.72
Heavy2221.030.24–4.450.750.17–3.310.720.16–3.26
Maximum consumption during late pregnancy**
Abstinent1924170Referent 1.0
Low18001570.980.79–1.240.960.76–1.211.000.78–1.28
Moderate472360.850.59–1.240.750.51–1.100.790.53–1.17
Binge occasional6381.500.70–3.200.860.40–1.880.860.39–1.92
Heavy80111.640.85–3.171.040.51–2.091.030.50–2.10

There was no evidence of an increased likelihood of preterm birth at low levels of alcohol after adjusting for covariates (Table 6). There was a nonsignificant increase in the odds of preterm birth with binge drinking in pregnancy, aOR 1.31 (95% CI 0.67–2.58) in first trimester and aOR 1.61 (95% CI 0.68–3.77) in late pregnancy, although the results lack precision due to small numbers. Combining all women who drank at greater than low levels during late pregnancy resulted in a masking of the association at higher levels, aOR 0.90 (95% CI 0.60–1.37).

Table 6.  OR (and 95% CI) for the association between alcohol consumption categories before and during pregnancy and preterm birth (<37 weeks)
Alcohol consumption and timing of exposureFrequencyUnadjustedAdjusted for smokingFully adjusted*
nPreterm birthOR95% CIOR95% CIOR95% CI
  • *

    Adjusted for maternal age, smoking, ethnicity, marital status, parity, drug use, income, maternal medical conditions, procedures, and treatments during pregnancy and pregnancy complications.

  • **

    Second and/or third trimesters. Note that numbers may not add up to the total due to women who abstained in First Trimester but drank in Late Pregnancy (n=784, 17%) and those who abstained in Late Pregnancy but drank in First Trimester (n=380, 8%).

Prepregnancy
Abstinent91954Referent 1.0
Low1557930.890.73–1.431.000.72–1.411.130.79–1.59
Moderate1282700.710.66–1.330.940.66–1.351.050.72–1.52
Binge ≤two times per week512230.770.47–1.250.750.46–1.230.830.50–1.38
Heavy449250.830.59–1.530.940.58–1.511.050.64–1.74
Alcohol consumption during trimester 1
Abstinent1924110Referent 1.0
Low1326720.950.71–1.280.960.71–1.301.040.76–1.41
Moderate446220.860.55–1.370.900.56–1.420.930.58–1.49
Binge ≤two times per week131101.310.69–2.511.270.65–2.491.310.67–2.58
Heavy10830.970.43–2.210.930.40–2.131.090.46–2.54
Alcohol consumed in trimester 1 with abstinence in late pregnancy
Abstinent1924110Referent 1.0
Low231120.910.50–1.650.910.50–1.650.910.49–1.66
Moderate9081.590.78–3.261.600.78–3.291.670.80–3.46
Binge ≤two times per week3741.890.70–5.111.730.63–4.791.650.58–4.69
Heavy*2232.590.82–8.152.260.71–7.232.300.69–7.72
>Low combined149151.811.05–3.101.731.00–3.001.781.01–3.14
Maximum consumption during late pregnancy**
Abstinent1924110Referent 1.0
Low1800970.940.72–1.240.950.72–1.240.990.75–1.32
Moderate472200.740.46–1.190.750.47–1.210.770.48–1.26
Binge ≤two times per week8061.650.72–3.751.520.66–3.511.610.68–3.77
Heavy6351.080.44–2.641.030.42–2.551.090.44–2.72
>Low combined615300.880.59–1.240.870.58–1.300.900.60–1.37

Women with a heavy pattern of alcohol intake in the first trimester and who stopped drinking by the second trimester had a nonsignificant, two-fold increased odds of preterm birth (aOR 2.30, 95% CI 0.69–7.72). There was a trend for moderate or higher levels of alcohol to increase the odds of preterm birth. Analysis combining moderate and higher levels of alcohol consumption yielded a 78% increased odds of preterm birth (aOR 1.78, 95% CI 1.01–3.14).

The PAR and AR for preterm birth for combined moderate, heavy, and binge exposure in the group of women who stopped drinking before second trimester were 11.4 and 19.4%, respectively.

Discussion

This study has been able to investigate the effect of different levels of alcohol consumption on fetal growth and preterm birth taking into account the quantity per occasion, frequency of consumption, and total quantity consumed. The findings of our study demonstrate that low levels of alcohol consumed during pregnancy at levels less than 60 g/week and not more than two standard drinks per occasion were not associated with preterm birth or SGA infants. Moderate to heavy alcohol intake resulted in an increased risk of preterm birth only in women who stopped drinking before the second trimester. There was no association between alcohol consumption during pregnancy and SGA infants after taking into account smoking status.

Our results highlight an increasing trend in the risk of preterm birth with increasing levels of alcohol exposure, although findings were imprecise due to small numbers. Many previous studies have not differentiated the pattern of alcohol consumption. By combining infrequent and weekly binge drinking with drinking at low levels (an average of less than one drink per day), the association was (likely to be) masked.1,40 The tendency for researchers to group together all women consuming an average of one drink or more per day also appears to conceal the association between dose and outcome.

The potential risk of preterm birth for women who ceased alcohol consumption before the second trimester was significantly increased when consumption at moderate and higher levels was combined. These results are similar to those reported by Jaddoe et al.40 who found an adjusted 2.5-fold increase in the risk of preterm birth with an average of one or more drinks of alcohol per day until pregnancy recognition. There are several possible explanations for the greater risk of adverse infant outcomes among women who ceased drinking before the second trimester that we have not been able to investigate in this study. Although we adjusted for maternal medical complications in general, there may be specific health problems for which women stop consuming alcohol and that predispose them to preterm birth that we were unable to investigate. Another potential reason may be that cessation of alcohol consumption before second trimester may trigger an inflammatory or other metabolic response resulting in elevation of inflammatory cytokines and thereby increasing the risk of preterm birth.50,51 Furthermore, it is possible that some or all the women reporting cessation actually continued to drink during late pregnancy, although given the lack of dose response observed in late pregnancy, this is unlikely to explain our findings.

It should be noted that the finding of increased odds of preterm delivery in women ceasing alcohol consumption before second trimester is based on small numbers and a type 1 error cannot be ruled out. It is, however, an issue warranting further investigation as our results indicate that, if this is a true finding, around 11% of preterm births would be attributable to this pattern of drinking. Prevention of high-risk drinking in early pregnancy could potentially minimise preterm birth and associated adverse developmental outcomes for the child and reduce costs on the healthcare system.

Our study highlights that fetal growth appears to be much more influenced by the effects of smoking than alcohol. The negative association between low levels of alcohol in prepregnancy and SGA infants was not observed with higher levels of exposure in prepregnancy. This association is likely to reflect the influence of unmeasured confounding factors rather than a true association. Our results are consistent with the majority of studies investigating low to moderate prenatal alcohol exposure and the power of our study to detect a difference at these levels was reasonable. While our study is in agreement with half of the studies examining higher levels of prenatal alcohol exposure, the lack of power in our study at higher levels of prenatal alcohol exposure precludes firm conclusions from being made for very heavy alcohol consumption. A significantly increased risk for women drinking an average of three or more drinks per day across pregnancy has been reported by one study.39 The relatively small numbers of women drinking at these levels in our study, particularly in late pregnancy, limited our ability to look at this group separately, so we cannot rule out that a higher risk is associated with this level of exposure.

The key strengths of this study are that it includes a randomly selected population-based cohort with a high response rate, thereby enabling the results to be generalised to the wider population, and we were able to adjust for a comprehensive range of known confounding factors including maternal behaviours and socio-demographic factors. While there may have been factors associated with heavy drinking that we were not able to account for, adjustment for known confounders such as smoking and socio-economic status is likely to have accounted, at least in part, for their effect. Our results have allowed us to address some of the main limitations identified in the recent systematic review by Henderson et al.,1 specifically, to examine the impact of low levels of alcohol consumption during pregnancy with sufficient power to detect a 50% increase in the odds of an effect for women drinking at low to moderate levels.

Underreporting of alcohol consumption is always a potential concern to epidemiological studies and although we cannot confirm if underreporting occurred in this study, there are a number of factors that we believe would have limited underreporting. In Australia, alcohol consumption during pregnancy is very prevalent with almost 50% of women in our study consuming alcohol during pregnancy; this percentage is similar to earlier Australian studies.52 Screening for alcohol use during pregnancy and information on the risks from prenatal alcohol exposure are not routinely undertaken by WA health professionals53 and there were no public health campaigns on this issue either before or during the study.

Reporting of prenatal alcohol consumption is influenced by the method and the timing of the questions.54–59 In this study, information on alcohol consumption during pregnancy was collected after the outcome of pregnancy was known so it is possible that recall bias may have occurred in some instances. Our data, however, were collected by self-administered questionnaire, which has been shown to elicit more valid responses on socially sensitive issues and reveal more unwanted behaviour than interviews.56 Studies have indicated that misclassification of prenatal alcohol consumption collected retrospectively is rare54,58–60 and thus is unlikely to have had a profound influence on our findings,58,60 although we accept this remains a possibility.

Although self-administered questionnaires have been reported to underestimate prenatal binge drinking,55 estimation of binge drinking in our survey was not obtained through a specific question on binge drinking. Instead, it was calculated from responses to questions on frequency, quantity consumed type of beverage, and measure (e.g. cans, glasses), which were asked together with questions on a range of prenatal maternal behaviours and family factors, so the focus of the survey was not exclusively upon alcohol consumption. However, if alcohol intake was underestimated in our study, the bias would likely be towards the null.

The trend to increased risk of preterm birth when mothers binge or drink heavily during pregnancy, even occasionally, highlights the importance of screening all women of childbearing age for alcohol use and promoting abstinence or less risky patterns of drinking. The increasing tendency for young women in general to drink at risky levels, including binge drinking,61,62 should be recognised as an important modifiable cause of preterm birth and efforts need to be made to reverse this trend especially before women conceive.

Conclusions

Our results highlight the importance of taking into account the pattern of maternal drinking when estimating risk. This population-based cohort study showed no evidence of an association between low levels of prenatal alcohol consumption and SGA infants or preterm birth and showed that smoking accounts for the association between alcohol intake and SGA births. There was a significant increased likelihood of preterm delivery in women who were drinking at moderate or higher levels in first trimester but stopped drinking before second trimester. Future studies with larger sample sizes and which take into account the pattern and timing of maternal drinking are necessary. In particular, to investigate the effects of heavy alcohol intake and binge drinking and the effect of ceasing alcohol intake after the first trimester on preterm birth.

Disclosure of interests statement

All authors declare that there are no competing interests and therefore we have nothing to declare.

Contribution to authorship

J.J.K. designed, obtained funding, and ran the original cohort study from which these data arose. Ms C.M.O. was the primary contributor to the paper in relation to the concept and design, analysis and interpretation of the results, and drafting of the article. N.N. and C.B. supervised Ms C.M.O. in the analysis of the data. All authors contributed to the interpretation of results, reviewing the article and provided final approval for the version to be published.

Ethics approvals

Ethics approval for the conduct of this study was granted by the Princess Margaret Hospital Research Ethics Committee and the WA Confidentiality of Health Information Committee.

Funding

The Western Australian survey of health-related behaviours and events during pregnancy and early infancy was funded by a grant from Healthway (the Western Australian Health Promotion Foundation 8016). J.J.K. was partially funded by a National Public Health Career Scientist award from the Department of Health and National Health Services Research and Development (PHCS022) when this analysis was conducted. This study was supported by the Australian National Health and Medical Research Council (NHMRC) program grant numbers 353514 (2005–2009), NHMRC Research Fellowship (353628) (C.B.) and NHMRC Public Health (Australia) Fellowship (404118) (N.N.).

Acknowledgements

The authors thank Margaret Wood, Peter Cosgrove, and Vivien Gee for maintenance of the databases.

Ancillary