“Did you ever drink more?” A detailed description of pregnant women’s drinking patterns

Women attending antenatal clinics at seven public hospitals located in Melbourne, Australia, between July 2011 and July 2012 were provided with information about the study by specially trained research staff and invited to participate. Pregnant women were eligible for inclusion if their pregnancy was less than 19 weeks gestation, they were 16 years of age or older, had a singleton pregnancy and spoke and read English. Women interested in participating in the study were invited to complete a consent form and a first questionnaire. All eligible women were invited until April 2012, after which abstainers, if this information was volunteered at recruitment, were no longer offered participation because the target number of the abstainer group had been reached. Of all 4788 approached, 27 % declined participation, 3035 women consented, 2046 completed at least one questionnaire, 73 % of whom completed all three pregnancy questionnaires and were eligible for this analysis (n = 1570). Additional details of data collection, participation rates and characteristics of women lost-to-follow-up, are described elsewhere [19].

The questionnaires were developed following a comprehensive literature review of existing survey measures of alcohol consumption during pregnancy, and identification of potential confounding, modifying and mediating factors. Focus groups with women of childbearing age augmented development of a set of questions and a pictorial ‘drinks guide’ to sensitively elicit accurate information about the type and amount of alcohol used during pregnancy [21]. Particulars of the full range of variables collected in the study are provided in the protocol paper [19].

Detailed information on the quantity and frequency of alcohol consumption for the three months pre-pregnancy and for each trimester was collected. For women who stopped drinking in the first trimester, information was gathered on when they stopped and women were classified as drinking only prior to pregnancy recognition (pre-aware) or drinking throughout first trimester (post-aware).

Women were provided with a pictorial drinks guide listing the most commonly-consumed types and volumes of alcoholic drinks including red and white wine, champagne, beer, cider, spirits, alcoholic sodas, pre-mixed spirits, port, sherry, and cocktails. Women were asked to use the drinks guide to identify what type of alcoholic drink(s) they usually consumed, with provision for up to five types of drink. For each beverage identified, women were asked how often they usually drank this type of alcohol (less than once per month, 1–2 days per month, 1–2 days per week, 3–4 days per week, 5 or more days per week) and how many drinks they usually consumed on each occasion (less than one drink, 1–2 drinks, 3–4 drinks, 5–6 drinks and 7 or more drinks). The next question asked women if there were any special occasions (or difficult times) when more than their usual amount of alcohol was consumed, the frequency of these occasions, the drink types, and number of drinks per occasion. If a woman reported consuming seven or more drinks on any occasion they were asked to provide the maximum number.

The level of alcohol consumption was derived from the drinks choices reported by the women, after conversion to standard drinks and then grams of absolute alcohol (g AA) per week. One standard drink in Australia is equal to 10 g of AA. The ‘low’ category was classified as ≤ 70 g AA per week and no more than 20 g AA per occasion. The ‘moderate’ group included women drinking ≤70 g AA/week, but consuming more than 20 g and less than 50 g per occasion. The ‘high’ group included women drinking >70 g AA/week, but never more than 49 g/AA per occasion. Binge drinking was classified as the consumption of ≥50 g AA per occasion.

Responses to the special occasion drinking question at each time point were converted to g AA and the number of drinking occasions and whether these were binge episodes or not were recorded. Estimates from special occasions were then combined with those obtained from ‘usual’ alcohol consumption to calculate a new maximum weekly intake.

At each of the five time points, a code was assigned to the woman depending on her weekly alcohol consumption level (none, low, moderate or high). If the woman had one or more binge episode during that time, ‘binge’ replaced the code for weekly alcohol consumption for that time point. Three PAE categories were created for women who abstained in pregnancy: lifetime abstainers; those who did not drink in the three months before pregnancy and those who drank in the three months before pregnancy. A small number of women were not categorised into any of these groups because of their unusual drinking patterns (e.g. abstained in trimester 1, then drank some alcohol at various levels later in pregnancy).

Final categorisation of prenatal alcohol exposure (PAE) is presented using a three-tiered approach: PAE tier 1 consists of women who were abstinent in pregnancy (not including lifetime abstainers) and women who drank any alcohol while pregnant; PAE tier 2 further categorises this latter group of women into those who only drank in trimester one and those who drank throughout; PAE tier 3 defines PAE in seven categories: women who were abstinent in pregnancy but not lifetime abstainers (control group); women who drank in the first trimester (either at low, moderate/high or at binge levels pre-aware) and were abstinent in trimesters two and three; and women who drank throughout pregnancy (either at low or moderate/high levels in the first trimester or at binge levels pre-aware).

A wide range of maternal demographic variables was collected. [19] The variables and the classifications used in this paper are maternal age (<30, 30–34, ≥35 years), parity (0, ≥1), marital status (partnered/un-partnered), ethnicity (Caucasian (e.g. White Australian, UK, other European), Asian and other (e.g. mixed race), education (secondary, diploma/trade, university degree), gross household income per year (up to $40,000, $40–100,000, >$100,000 AUD), current financial situation (living comfortably, doing all right, just getting by, struggling), region of residence (metropolitan, rural), smoking (no, yes) and planned pregnancy (no, yes). Maternal report of height and pre-pregnancy weight were used to calculate body mass index (BMI). To gauge possible individual variation in alcohol metabolism, the women were asked if they felt the effects of alcohol very quickly, quickly, normally, slowly, or very slowly. Women were also asked about their drinking history, how old they were when they first started drinking regularly, if they had ever been intoxicated after drinking alcohol (defined as slurred speech, unsteady on their feet, or blurred vision) and the age when they first became intoxicated from drinking alcohol. Responses were categorised by age into <18 years and ≥18 years to reflect the legal drinking age in Australia.

Data were prepared and analysed in Stata v14 [22]. Codes were assigned for mutually exclusive PAE patterns according to the pre-defined levels, timing and patterns of the exposure(s). Percentages of binge episodes disclosed as part of usual drinking patterns and after inclusion of special occasion drinking information were calculated with 95 % binomial confidence intervals.

Pearson chi square testing was used to examine the relationship between early maternal age at onset of regular drinking, or first intoxication (+/− 18 years) and each PAE group.

In the three months prior to conception, 269 of the 1570 women (17.1 %) declared a binge episode of drinking as part of their usual drinking. The special occasion question resulted in an additional 327 women reporting a binge episode, raising the total to 596 women (37.9 %) having consumed alcohol at binge levels prior to conception.

After conception but before pregnancy recognition, 193 of 1570 women (12.3 %) declared binge episodes as part of their usual drinking; another 6.2 % were identified with the special occasion question, resulting in almost one in five women (18.5 %) having binged prior to pregnancy recognition.

Following pregnancy recognition, five women declared a binge episode in trimester one, five women in trimester two and one in trimester three.

Of those who drank only in the first trimester, 87 % stopped once they realised they were pregnant (n = 280). The 329 women who continued to drink throughout pregnancy did so at low to moderate levels, 39 % of them drinking at binge levels in trimester one (n = 128), and 53 % drinking at low or moderate levels in trimester one (n = 174).

PAE tier 1 showed that most women who did not drink in the three months prior to pregnancy, remained abstinent while pregnant (90 %). The second PAE tier revealed that the majority of women who drank at high or binge levels before pregnancy drank throughout pregnancy (62 and 50 % respectively). PAE tier 3 showed a large percentage of women continuing to drink at the pre-pregnancy level, regardless of whether they ceased alcohol consumption in trimester one or continued throughout (e.g. 52 % of low pre-pregnancy drinkers continued to drink at low levels, 43 % of moderate pre-pregnancy drinkers and 64 % of high pre-pregnancy drinkers continued at these levels). A high percentage of women who drank at binge levels prior to pregnancy had binge episodes before they found out they were pregnant (44 %).

Of the 1115 women who reported their age at onset of regular drinking, 56 % drank regularly before 18 years of age (n = 626). There were no discernible differences in age of onset of regular drinking between pregnancy abstinence or alcohol use (PAE tier 1) and whether this influenced women drinking in trimester one only or throughout pregnancy (PAE tier 2). PAE tier 3 analysis revealed that women who reported drinking regularly before the age of 18 years were more likely to have binge episodes early in pregnancy and to continue throughout, than women who began to drink at a later age. (17.6 % compared with 12.0 %, p = 0.02).

Of the 1205 who reported their age at first intoxication, 54 % of women reported this occurring before 18 years of age (n = 651). This group of women was more likely to drink alcohol while pregnant (PAE tier 1: (74.8 % compared with 61.9 %)) and if drinking, more likely to drink alcohol throughout pregnancy (PAE tier 2: 44.7 % compared with 28.3 %, p < 0.001). Being underage at first intoxication also predicted early pregnancy binge drinking, especially in those who drank throughout pregnancy (PAE tier 3: 19.5 % compared with 6.5 %, p < 0.001).

Table 4 contains the results of the adjusted logistic regression analyses of nine maternal characteristics across PAE tiers 1 and 2. Table 5 shows the same analysis for the expanded, PAE tier 3 categories. Current financial situation, relationship status, maternal age and region of residence are not included as either no meaningful differences in the univariate analysis were noted between groups or group numbers were too small.

Women who drank any alcohol while pregnant were more likely to have an unplanned pregnancy (PAE tier 1: OR 1.89 (95 % CI 1.40–2.55)), smoke (PAE tier 1: OR 1.94 (95 % CI 1.36–2.77)), have a household income of over AUD 100,000 (PAE tier 1: OR 1.71 (95 % CI 1.15–2.54)). They were less likely to perceive the effects of alcohol quickly (PAE tier 1: OR 0.47 (95 % CI 0.37–0.61)), be obese (PAE tier 1: OR 0.53 (95 % CI 0.38–0.75)) or of Asian/other ethnicity (PAE tier 1: OR 0.38 (95 % CI 0.27–0.52)). The PAE tier 2 comparisons revealed that pregnancy planning was only an important factor in those who ceased alcohol consumption in trimester one (PAE tier 2: OR 2.05 (95 % CI 1.47–2.87)), whereas smoking (PAE tier 2: OR 3.05 (95 % CI 2.00–4.67)), higher income (PAE tier 2: OR 2.43 (95 % CI 1.45–4.04)) and obesity (PAE tier 2: OR 0.35 (95 % CI 0.22–0.54)) remained associated only with women who drank throughout.

The PAE tier 3 analysis presented in Table 5 shows that the maternal predictors found to play a role in pregnancy drinking mostly related to binge drinking in the pre-aware period (regardless of whether they ceased alcohol consumption in trimester one or continued throughout) and to moderate/high dinking levels in women who drank throughout pregnancy. None of the maternal characteristics examined were associated with low or moderate/high drinking in those who drank only in trimester one and only Asian/other ethnicity was a protective factor for low level PAE throughout pregnancy.

A strength of this study is the accuracy of the alcohol measure [21] and a focus on the most frequent patterns of drinking, rather than heavy and chronic alcohol consumption. We believe this has played a key role in high participation and low attrition, but most importantly, in providing PAE data of the highest quality possible. Although our study measured PAE prospectively and thus optimised our ability to measure frequency, dose and timing of exposure, the use of self-reported questionnaires runs the risk of reporting bias. However, our focus group research showed that if questions on alcohol in pregnancy are appropriately contextualised and include an option to report unusual drinking episodes, this will encourage more accurate reporting, [21] a finding which appears confirmed with the high number of binge episodes reported in response to the special occasion question.

Our methodology also identified two groups of women who abstained from drinking alcohol in pregnancy (not including lifetime abstainers), those who drank alcohol in the three months prior to pregnancy and those who did not. While we combined the two groups for our analyses, it is likely that the exposure pattern of alcohol consumption in the three months prior to, but not during pregnancy, includes some women who inadvertently drank before pregnancy recognition and have self-reported incorrectly. On this basis we believe the best ‘control’ group in future examinations of child outcomes comprises women who reported no drinking in the three months before pregnancy. Further, detailed information on pregnancy alcohol consumption patterns, such as presented in this paper, should contribute to better confidence in prediction of outcomes after low to moderate PAE. Along with comprehensive reporting of alcohol consumption in each of the three pregnancy questionnaires, women provided information on drinking history and demographics, characteristics that are potentially associated with various exposure categories. Together, these may help target health promotion messages to those at greatest risk.

The study included only English-speaking women from seven metropolitan public hospital antenatal clinics; therefore, women seeking care from private obstetricians, those from rural regions or those from culturally and linguistically different backgrounds may present with different patterns of drinking. However, the prevalence figures and predictors of PAE categories are specifically relevant to the target population which represents the majority of pregnant women in our State, where 70 % of women currently receive antenatal care in the public system. [45] Selection bias may have occurred if some women chose not to participate in the study because they did not drink alcohol while pregnant or because they were lifetime abstainers. Conversely, women drinking at higher levels may also have chosen not to participate or may have been more likely to drop out without completing all three pregnancy questionnaires needed to compute a PAE classification. All of this could affect our population prevalence estimates in Fig. 1. However, we achieved a sizeable number of women who abstained from alcohol (including lifetime abstainers), as well as a substantial number who drank throughout pregnancy, which we believe reflects realistic population drinking patterns.