Maternal alcohol consumption during pregnancy and child’s cognitive performance at 6–8 years of age in rural Burkina Faso: an observational study

Study area, setting, study design and participants

Burkina Faso is a low income country located in the middle of West Africa; the population resides mainly in rural areas (70.1% in 2015), and the population aged 0–14 years was 46.3% in 2013 (INSD, 2016; UN Statistics, 2016). The literacy rate is very low and the mean years of education attained in women and girls was less than 3 years in 2013 (Patton et al., 2016). The official language in Burkina Faso is French. However, the country has more than 60 different ethnic groups. Several local languages are spoken in the study area Gouin, Karaboro, Dioula, Senoufo, Turka, Moore, and Fulfulde (Hama Diallo et al., 2012; Rossier et al., 2013; Ethnologue, 2016), which is a challenge when performing cognitive testing.

In 2006, a community-based cluster-randomised trial of children was conducted, the PROMISE Exclusive Breastfeeding (EBF) study. One of the sites was in rural Burkina Faso (Diallo et al., 2010; Diallo et al., 2011; Tylleskär et al., 2011; Hama Diallo et al., 2012). The sampling has been described (Diallo et al., 2010; Tylleskär et al., 2011). From 2013 to 2015, a cross-sectional follow-up study was conducted through the PROMISE Saving Brains study to assess the neuro-cognitive performance of the children aged 6–8 years old. We sought to re-enrol all children from the initial PROMISE EBF trial who were found to be alive and still residing in the study area.

Outcome measures

The Kaufman Assessment Battery for Children, 2nd edition (KABC™-II) is an individually administered cognitive test with verbal and nonverbal components which has been used across diverse cultural contexts (Boivin et al., 1996; Ochieng, 2003; Kaufman & Kaufman, 2004; Malda et al., 2010). In Africa, it has been used to study cognitive development and nutrition in Ethiopia (Bogale et al., 2013), Democratic Republic of Congo (Boivin et al., 2013; Bumoko et al., 2015) and South Africa (Taljaard et al., 2013; Rochat et al., 2016), among HIV infected children in Uganda (Boivin et al., 2010; Ruel et al., 2012; Brahmbhatt et al., 2017), and cerebral malaria in Senegal (Boivin, 2002), and Uganda (Bangirana et al., 2009). KABC-II has different sub-tests and is used in children aged 3–18 years. The sub-tests (Appendix A) used in our study were:

• Atlantis: a measure of memory

• Conceptual Thinking: a measure of visual and spatial abilities

• Face recognition: a measure of visual and spatial abilities

• Story Completion: a measure of pattern recognition and reasoning

• Number Recall: a measure of memory

• Rover: a measure of spatial scanning and reasoning

• Triangle: a measure of spatial abilities and visualization

• Block Counting: a measure of reasoning

• Word Order: a measure of memory

• Pattern reasoning: a measure of reasoning and visualization (Kaufman & Kaufman, 2004; Bangirana et al., 2009).

The Children’s Category Test Level 1 (CCT-1) is a widely used non-verbal test developed to evaluate problem solving in children; it is fast and easy to administer (Boll, 1993; Hundal & Morris, 2011; Goudis, 2014). It was used to examine the effect of different exposures including health conditions like traumatic injuries (Moore, Donders & Thompson, 2004; Donders & Nesbit-Greene, 2004; Horneman & Emanuelson, 2009), brain dysfunction (Allen, Knatz & Mayfield, 2006; Bello, Allen & Mayfield, 2008), diseases (Rosenberg et al., 2010), marihuana and cocaine (Fried, Watkinson & Gray, 2005; Ga et al., 2015), disabilities (Hinton et al., 2004), chemical products (Debes et al., 2006; Wright et al., 2006; Jurewicz, Polańska & Hanke, 2013), and alcohol (Mattson et al., 1998). CCT-1 is an individually administered standardized test for children from 5 to 8 years to test their ability to solve problems on the basis of corrective feedback. It is presented in booklet form and consists of five subtests. At the end of the test, the total number of errors is counted. Children with more errors are the one who performed worst (Boll, 1993; Moore, Donders & Thompson, 2004; Allen, Knatz & Mayfield, 2006).

The KABC-II and the CCT-1 were administered by a team of four trained psychologists who spoke the local languages. The children were randomly assigned to the psychologists for assessment. The assessors administered individually the KABC-II and the CCT-1 during a one-to-one session. The instructions of the measures were translated in the main local language (Dioula) commonly spoken in the study area. Independent back translations were done.

Exposure measure

Maternal alcohol consumption during pregnancy was the main exposure for this analysis. Information about maternal alcohol consumption during pregnancy was collected in a household interview with the caretaker prior to the neuro-cognitive assessment. Data collectors approached each child’s household to administer a questionnaire to the child’s caregiver during a one-to-one interview. Mothers were the primary respondents. A yes/no question of any alcohol consumption during pregnancy was asked. Of all the 554 caretakers, 518 were able to provide information on this question and 36 (6.5%) were not.

Covariates

In the interview, questions were asked about background characteristics and socio-economic status that may influence the child’s performance. These include the child’s age, child’s schooling, father’s employment, father’s education (dichotomized to educated = at least one year in school, or not educated), mother’s age, mothers’ employment, mother’s education, current maternal alcohol status (a yes/no question of any current alcohol consumption), mother’s depression status using the Hopkins symptom checklist (Sirpal et al., 2016) (dichotomized to depression = at least one symptom in the checklist, no depression = no symptom in any of the checklist), mother’s chewing tobacco status (a yes/no question of current tobacco chewing), and presence of latrine in the compound (a yes/no question). Questions regarding past hospitalizations since birth of the child were asked and anthropometric data (height, age) were measured according to standard procedures (CDC, 2007) by a paediatrician at the study site. Stunting was defined as below-2 standard deviations of height-for-age.

Before the starting of data collection, field-testing and piloting of all the instruments was conducted to calibrate and standardize the assessment of cognitive measures and the data collection. For instance, the stadiometer for height was calibrated according to the instruction of manual, and the psychologists underwent field training and refresher training to standardize the way to administer the KABC-II and CCT-1 on local children prior to the study participants.

Statistical analysis

Statistical analyses were conducted in several stages:

1. To examine within population variance of the sub-tests, the distribution of scores (mean, standard deviation, median, minimum and maximum) were used. Box-and-whisker plots per exposed and unexposed groups were used to illustrate the children’s performances on different sub-tests of KABC-II and CCT-1. Extreme scores were winsorized to discount the influence of outliers by replacing their values with the nearest scores within this range.

2. To examine the reliability of items of the sub-tests, split-half reliability coefficients were calculated for KABC-II (Kaufman & Kaufman, 2004; Malda et al., 2010) and Cronbach’s alpha coefficient was calculated for CCT-1 (Boll, 1993; Moore, Donders & Thompson, 2004; Allen, Knatz & Mayfield, 2006). The level of significance of the reliability coefficient was ≥0.7.

3. To examine the association between maternal alcohol consumption during pregnancy and cognitive performance of KABC-II and CCT-1, effect size differences using the Cohen’s d (Sullivan & Feinn, 2012; Cumming, 2014), and linear regression analysis were conducted. No validated norms of the KABC-II and the CCT-1 were available in Burkina Faso at the time of the study; we then used the raw scores instead of the scaled scores. However, all scores were standardized (Z) and analysis were conducted on both unstandardized and standardized scores. All the coefficients were adjusted for potential confounders including child’s age, sex, schooling, stunting, father’s employment, father’s education (Martinez et al., 2011; Falgreen Eriksen et al., 2012; Flak et al., 2014; Kesmodel et al., 2015) and the promotion of exclusive breastfeeding (‘intervention arm’ of the initial trial). A bivariate analysis between each covariate and the outcome was conducted (Table A1). STATA 13 was used to perform the analysis.

Ethical considerations

Written informed consent was obtained from all care-takers in the study and oral assent was obtained from the children. The study was approved by the Institutional Review Board of Centre MURAZ number 008-2013/CE-CM.

Study population

Of the initial 794 enrolled children in the PROMISE EBF trial in Burkina Faso site, 561 were alive, traced and re-consented for the follow-up study, 554 children completed the KABC-II and the CCT-1, and 518 children had information on their maternal alcohol consumption status (Fig. 1).

Of these, 51.7% (268/518) were boys, and 49.4% (256/518) were at school. The mean (± standard deviation, SD) age at assessment was 7.2 (±0.4 years), the median (interquartile range, IQR) was 7.2 (6.9–7.4) years and the range was 6.3–8 years. Of the mothers, 18.5% (96/518) reported to have consumed alcohol during the pregnancy and none of them had been more than 1 year in school. The mean (±SD) age of the mothers at assessment was 33.4 (±6.3 years). Of the fathers, 30.6% (156/510) had attended at least 1 year in school and 12.9% (67/518) had an employment. Three quarters of the compounds reported having a pit latrine 73.4%, (380/518) (Table 1).

On the KABC-II, sufficient variability (mean ± SD) of the raw scores was found for all the sub-tests except ‘Pattern Reasoning’ (Fig. 2 and Table 2). No child scored 0 in ‘Atlantis’, ‘Number recall’, ‘Rover’, ‘Triangle’ and ‘Word order’ (Fig. 2 and Table 2). The Split-half reliability coefficient was acceptable (>0.70) for all the sub-tests except ‘Story completion’, ‘Rover’, ‘Word order’ and ‘Pattern Reasoning’ (Table 2).

Maternal alcohol consumption and cognitive performance

Children whose mothers reported alcohol consumption during pregnancy performed significantly poorly for memory and spatial abilities tests from small effect size difference for ‘Atlantis’ (0.27) and ‘Triangle’ (0.29) to moderate effect size difference for ‘Number recall’ (0.72) compared to children whose mothers did not consume alcohol during pregnancy; the exposed children scored significantly higher errors with a small effect size (0.37) at problem solving (CCT-1) test compared to unexposed children (Table 3).

At unstandardized and standardized multivariable analysis, children whose mothers reported alcohol consumption during pregnancy performed significantly poorer for memory-‘Atlantis’ (p = 0.03) and ‘Number recall’ (p = 0.0001), and spatial ability tests-‘Triangle’ (p = 0.03); they scored significantly higher errors at problem solving CCT-1 test (p = 0.002); all the results were adjusted for age, sex, schooling, stunting, father’s education, mother’s employment and the promotion of exclusive breastfeeding (Table 4). No statistical association was found for visual abilities-‘Conceptual Thinking’, ‘Face recognition’, ‘Story completion’, and reasoning tests-‘Rover’, ‘Block counting’, and ‘Pattern Reasoning’ (Table 4).