Maternal pre-pregnancy risk drinking and toddler behavior problems: the Norwegian Mother and Child Cohort Study

The Norwegian Mother and Child Cohort Study (MoBa) is an on-going prospective population-based pregnancy cohort study conducted by the Norwegian Institute of Public Health [31]. Participants were recruited between 1999 and 2008 and 277,700 pregnancies were invited. The participation rate was 40.6 %, and the cohort now includes over 109,000 children, 91,000 mothers, and 71,700 fathers. Non-participation was higher among pregnant women who were <25 years, living alone, smokers, and who had >2 previous births or previous stillbirths. MoBa mothers had somewhat higher education, a lower prevalence of preterm births and fewer babies with low birth weights compared with the general population of pregnant women in Norway in 1999 to 2008 [32]. Follow-up is still on-going through questionnaires at regular intervals and by linkage to national health registries. The study was approved by the Regional Committee for Medical Research Ethics in South-Eastern Norway, and has been performed in accordance with the ethical standards in the 1964 Declaration of Helsinki and its later amendments. Informed consent was obtained from each MoBa participant upon recruitment.

The current study employs information obtained at gestation week 17/18 and 30 (Questionnaire 1 (Q1) and 2 (Q2), and Questionnaire Father (QF)), and when the child was 6 months (Questionnaire 4; Q4), 18 months (Questionnaire 5; Q5), and 36 months old (Questionnaire 6; Q6). Additional information was obtained from the Medical Birth Registry of Norway (MBRN) which contains information about all births in Norway from 1967 and onwards [33]. Except from the MBRN and QF information, all information employed in the current study was obtained from mothers’ self-report.

The analyses in the current study are based on the two samples of participants who responded to the MoBa questionnaires when the child was 18 months (n = 56,682) and/or 36 months (n = 46,756). Mothers who answered items from the Child Behavior Checklist (CBCL) at 18 and/or 36 months, and responded to the T-ACE screening instrument in Q1, were included in the study. Mothers who had not answered T-ACE at Q1, who had incomplete responses to the CBCL items and who were currently pregnant were excluded from the samples. The exclusion processes are visualized in Fig. 1.

Maternal pre-pregnancy risk drinking was assessed retrospectively with the screening instrument T-ACE at gestation week 17/18. T-ACE assesses risk for current drinking among pregnant women, but is also a sensitive screen for both general risk drinking and lifetime alcohol diagnoses [34, 35]. T-ACE consists of four items designed to measure tolerance to alcohol and behavior associated with risk drinking; Tolerance: “How many drinks does it take to make you feel high?”, Annoyed: “Have other people irritated you or hurt your feelings by criticizing how much you drink?”, Cut-down: “Have you ever felt that you ought to drink less alcohol?” and Eye-opener: “Have you ever drunk alcohol in the morning to calm your nerves or to get rid of a hangover?”. In MoBa, one alcohol unit was defined as equivalent to 1.5 cl (12.8 g) of pure alcohol. Although the conventional cut-off for the Tolerance item is two drinks, a cut-off of more than two drinks has been found to give higher specificity [36]. Further, risky single-occasion alcohol use among women is also usually defined as the intake of approximately 40–60 g pure alcohol (≈3.1–4.7 in MoBa alcohol units) or more in one single occasion [5, 37]. Thus, in the present study, two points were given if three or more drinks were reported in the first item (T). Affirmative responses on the A, C and E items were given one point each. A total score ≥3 was used as cut-off point for risk drinking, which has been shown to give good sensitivity and specificity [38, 39].

The Child Behavior Checklist [40] version for preschool children (CBCL/1.5–5) is constructed to cover a range of emotional, social and behavior problems. CBCL/1.5–5 consists of 99 items describing a behavior in the child the preceding 2 months, and the respondent, usually a parent or other adult who lives with the child, rates each item on a scale from 0 “not true”, 1 “somewhat or sometimes true” to 2 “very true or often true”. CBCL/1.5–5 has several subscales that are included in two scales measuring internalizing and externalizing behavior problems. The CBCL version for older children has been validated in a Norwegian study [41], while CBCL/1.5–5 has been validated in Dutch and Danish samples [42, 43]. MoBa was designed to cover a range of areas relevant for child health and development, and the full CBCL/1.5–5 was not included due to space restrictions in the questionnaires. The 18 and 36 months questionnaires in MoBa contain CBCL/1.5–5 items that were intended to represent all CBCL subscales with two or three items, and to be clinically and theoretically relevant indicators of behavior problems. Thirteen of the CBCL items were included in both the 18 and 36 months questionnaire, and were employed in the current study as indicators of behavior problems. The items are presented along with their internalizing or externalizing scale affiliation in Table 1. Case cut-points at the 85th or 90th percentile on each subscale are recommended when the full CBCL is employed in population-based studies [41, 42]. We adopted this strategy in the present study, thus scores above the 90th percentile were defined as internalizing or externalizing behavior problems.

Information on child gender, birth weight and gestational age was obtained from MBRN and computed into a common z-scored variable indicating birth weight by gestational age. The number of older siblings was based on mothers’ report of previous pregnancies resulting in a live-born child in Q1. Maternal current civil status was assessed when the child was 18 and 36 months, and categorized as: (0) married/cohabitant, (1) divorced/separated, (2) widow, (3) single, and (4) other. Mother’s highest completed education was assessed at gestation week 17/18 and categorized as: (0) >4 years college/university, (1) ≤4 years college/university, (2) completed high school, (3) 1–2 years high school, and (4) 9 years compulsory school. Household income was based on maternal and paternal income if they were living together, reported at gestation week 17/18 and in QF. If the mother was not living with a spouse/partner, the household income was set equal to the mothers’ income. Six household income categories were generated, ranging from “No income” to “Over 1,000,000 NOK”. Finally, the mother was asked about recent financial problems at 18 and 36 months (yes/no).

The mother reported her alcohol consumption in each pregnancy trimester (Q1, Q3 and Q4) in terms of frequency: “How often do you consume alcohol?”, with alternatives ranging from “never” to “roughly 6–7 times a week”, quantity: “How many units do you usually drink when you consume alcohol?” with response categories ranging from “less than 1”, to “10 or more”, and binge drinking: “During pregnancy, how many times have you consumed five units or more of alcohol?”, with answer categories ranging from “never” to “several times per week”. The mother was also asked about her current frequency and quantity of alcohol intake when the child was 18 and 36 months old.

Mothers’ smoking habits during pregnancy were assessed in Q1, Q3 and Q4. Responses on these time points were computed into a common variable indicating smoking habits during the full pregnancy. The mother was coded as “daily smoker” if she reported to be smoking daily on either one of the assessments, “occasional smoker” if she reported occasional, and not daily smoking, on any of the assessments, and “never smoked” if she did not report daily or occasional smoking on any of the assessments.

Current maternal psychopathology included anxiety/depression assessed at 18 and 36 months and ADHD at 36 months. Maternal anxiety and depression were assessed with the eight items version of Hopkins Symptom CheckList (SCL-8) [44, 45]. The mothers were asked to indicate how much the presented emotions and cognitions had bothered them during the past 2 weeks, scored on a 4-point scale ranging from 1 (“not bothered”) to 4 (“very bothered”). The item scores were then summarized into a continuous variable, ranging from 8 to 32. Maternal ADHD was assessed in Q6 with the screening instrument Adult ADHD Self-Report Scale (ASRS) [46]. ASRS consists of six items measuring inattention and hyperactivity/impulsivity, with responses ranging from “never” to “very often”. The total scores of ASRS were summarized into a continuous variable, range 0–24, where higher scores indicate more ADHD symptoms.

Missing responses in the included covariates ranged from 0.1 % (maternal age) to 29.2 % (maternal ADHD in the 18-months sample). Multiple missing imputation procedures in Stata 12.0, using the multivariate normal approach with 10 imputation procedures, were employed for these variables before the main analyses were conducted [47]. Attrition between Q1 and Q5/Q6 was somewhat higher among participants who scored at or above the cut-point on T-ACE at Q1 compared to participants scoring below the cut-point (attrition Q1–Q5 (cut-point vs. below cut-point): 31.1 vs. 26.2 % (p < 0.001), attrition Q1–Q6: 48.9 vs. 43.0 % (p < 0.001)). The characteristics of the two samples (18 and 36 months) were examined with descriptive statistics. The associations between maternal pre-pregnancy drinking problems and internalizing and externalizing behavior problems in their toddlers aged 18 and 36 months were analyzed using logistic regression models. The findings are presented as (1) crude estimates, (2) estimates adjusted for single covariates (except from the alcohol consumption variables, which were included as a common block at each trimester in pregnancy, and when the child was 18 and 36 months), and (3) fully adjusted estimates. All estimates are reported as odds ratios (OR) with 95 % confidence intervals (CI). Stata 12.0 was used for all analyses [48].

The results from the present study should be considered in light of several important limitations. Firstly, the relatively low participation rate in MoBa (40.6 %) and attrition from Q1 to Q6 are likely to introduce selection bias [57]. Participation in population-based health studies is lower among individuals with alcohol problems [58]. Although selection bias is likely to have a greater impact on prevalence estimates than on association estimates [31, 57, 58], we cannot fully exclude the possibility that the associations identified in the present study underestimate the true relationships in a general population of mothers and toddlers. Secondly, under-reporting of alcohol consumption is a general challenge in health studies [59] and is particularly common in studies based on pregnancy samples [60]. Although a substantial portion of the mothers reported alcohol use during pregnancy, with falling tendencies from around 30 % in the first trimester, to 20 % in the second trimester and 10 % in the third trimester (data not shown), these numbers are lower than prevalence rates of pregnancy alcohol use from a previous Norwegian study by Alvik and colleagues [60], and may indicate under-reporting. However, bias due to residual confounding will only be present if under-reporting was more frequent among risk drinking than non-risk drinking women, and under-reporting has previously not been found to be higher among women scoring positive on T-ACE [60]. Thirdly, MoBa did not include the full CBCL 1.5/5 (for an overview of the items included, see Table 1). While the items indicating externalizing behavior problems at 18 months have been used in a previous publication from the MoBa database [61] the use of these abbreviated, and not validated versions of the original scales may threaten the validity of the results. Fourthly, the toddlers’ behavior problems were reported by the mother. Mothers with psychopathology or who engage in risk drinking may be more prone to both detect and interpret their children’s behavior in a more negative direction [51]. Maternal reporting of toddler behavior problems may thus reflect characteristics of the mother rather than the toddler. In contrast, Norwegian parents are in general found to under-recognize and under-report behavior problems in their children [62]. In the present study, mothers with psychopathology or risk drinking may over-report while “healthy” mothers under-report their toddlers’ behavior problems, resulting in falsely large group differences. Finally, many terms are used in the literature to describe harmful alcohol use, for instance “hazardous alcohol use”, “problem drinking”, “alcohol problems” and “alcoholism”, often without defining inclusion and exclusion criteria for the employed term. The choice of the concept “risk drinking” in the present study was based on this term being commonly used in studies employing T-ACE, although it lacks a theoretical rationale.

To conclude, we found that maternal pre-pregnancy risk drinking predicted toddler behavior problems. The associations were affected by other factors known to have importance in the relationship between parental risk drinking and child behavior problems. It is therefore likely that maternal pre-pregnancy risk drinking is not an independent risk factor for child behavior problems, but rather represents a proxy measure of other adverse characteristics which may influence her child’s behavior. A central issue for future research should be to identify and investigate such potential characteristics and how they in combination with each other and with risk drinking lead to adverse child outcomes. Further, developmental pathways in toddlers of mothers engaging in pre-pregnancy risk drinking should be investigated.