What Influences 11-year-olds to Drink? Findings From the Millennium Cohort Study

Yvonne Kelly; Alice Goisis; Amanda Sacker; Noriko Cable; Richard G. Watt; Annie Britton


BMC Public Health. 2016;16(169) 

In This Article


The Millennium Cohort Study (MCS) is a UK nationally representative prospective cohort study of children born into 19244 families between September 2000 and January 2002.[21] Participating families were selected from a random sample of electoral wards with a stratified sampling design to ensure adequate representation of all four UK countries, disadvantaged and ethnically diverse areas. The first sweep of data was collected when cohort members were around 9 months and the subsequent four sweeps of data were collected at ages 3, 5, 7, and 11 years. At the 11 year sweep, interviews were conducted during home visits with cohort members and their carers, and questions asked about alcohol consumption, socioeconomic circumstances and family relationships. Cohort members filled out a self-completion booklet in a private place within the home. Interview data were available for 69 % of families when cohort members were aged 11.

Drinking at Age 11

In a question developed for the MCS survey, cohort members were asked "Have you ever had an alcoholic drink? That is more than a few sips?" (yes/no).

Parent and Friends' Drinking

Parents were asked about the frequency and amount of alcohol they drank. "How often do you have a drink that contains alcohol?" (4 or more times a week, 2–3 times a week, 2–4 times per month, Monthly or less, Never). "How many standard alcoholic drinks do you have on a typical occasion?" Response options on frequency and quantity of alcohol consumed meant it was only possible to approximate drinking categories as set out in guidelines by the UK Department of Health. The same categories were used for mothers and fathers as follows: None; Light/moderate - those who drank but were not heavy/binge drinkers; Heavy/binge - 4 or more times a week and drinks a minimum of 3–4 drinks per drinking occasion, or a minimum of 5–6 drinks per occasion. Separate categories were created for cohort members where information on parents' drinking behaviour was missing and when the father was absent from the household.

Friends' drinking was assessed by asking cohort members "How many of your friends drink alcohol?" Response categories were recoded: None of them as No; Some/Most/All of them as Yes; don't know was retained as a separate category.


Cohort Member and Family Characteristics. Gender; puberty, assessed from responses by the mother to questions (for girls - hair on body, breast growth, menstruation, boys - hair on body, voice change, facial hair); birth order (first vs subsequent); current socioemotional difficulties (normal vs high score);[22] antisocial behaviours ("Have you ever … been noisy or rude in a public place so that people complained or got you into trouble? … taken something from a shop without paying for it? … written things or sprayed paint on a building, fence or train or anywhere else where you shouldn't have? … on purpose damaged anything in a public place that didn't belong to you, for example by burning, smashing or breaking things like cars, bus shelters and rubbish bins?" categorised 0, 1, 2 or more); truancy (yes/no); cigarette smoking (yes/no); quintiles of equivalised family income; religious affiliation (none vs any of Christian, Muslim, Hindu, Jewish, other).

Potential Moderating Variables. Perception of risk due to alcohol was assessed by the question "How much do you think people risk harming themselves if they drink one or two alcoholic drinks nearly every day?" (no/slight risk, some risk, great risk). Positive expectancies towards alcohol were assessed by the following questions: "Drinking beer, wine, or spirits is a way to make friends with other people"; "It is easier to open up and talk about one's feelings after a few drinks of alcohol"; "Drinking alcohol makes people …worry less; …happier with themselves". Negative expectancies were assessed using questions: "Drinking alcohol … gets in the way of school work; … makes it hard to get along with friends"; "If I drank alcohol without my parents' permission I would be caught and punished". Items were summed and used as two separate scales.[23]

Parental supervision was assessed by questions about the weekday and weekend frequency of cohort member spending unsupervised time with friends (playing in the park, going to the shops or just 'hanging out'). Items were combined into a three category variable: rarely/never (at most occasionally at weekends/on weekdays), sometimes, often (unsupervised most weekends and at least one day per week).

Markers of family relationships were: frequent battles of will with cohort member (yes/no); mother-cohort member closeness (extremely/very close vs fairly/not very close); cohort member happiness with their family ("On a scale of 1 to 7 where '1' means completely happy and '7' means not at all happy, how do you feel about your family?" Responses corresponding to the top decile of the distribution were taken to indicate happy with family).[24]

Study Sample

Data on cohort member drinking were available for 12644 participants. Missing data reduced the sample to 10498 (83.0 %), as follows: friends drinking = 56; puberty = 1010; socioemotional difficulties = 475; religious affiliation = 34; antisocial behaviours = 27; perception of harm = 337; positive expectancies = 189; negative expectancies = 315; parental supervision = 87; frequent battles = 1106; relationship between mother and child = 762; happy with family = 91.

Statistical Analysis

To estimate the association between our exposures of primary interest – mother's, father's or friends' drinking with cohort member drinking, we ran three sets of logistic regression models adding covariates in stages. Boys were more likely to report drinking compared with girls (15.7 vs. 11.3 %), but as there were no gender differences in observed associations between parent and friends' drinking with cohort member drinking, we present analyses for boys and girls combined, and all models adjust for gender.

Model 0 is the baseline model which includes the primary independent variable (mother's, father's or friends' drinking) and gender.

Model 1 additionally adjusts for control variables: puberty, birth order, socioemotional difficulties, antisocial behaviours, truancy, smoking, income, religion and for other alcohol exposure variables e.g. when mother's drinking is the primary exposure, we add father's and friends' drinking to this step of the analysis.

Model 2 is fully adjusted adding in potential moderator and mediator variables, perception of harm due to alcohol, positive and negative expectancies, parental supervision and family relationships (battles, closeness, happiness with family).

All analysis was carried out using Stata version 13.1 (Stata Corp).