Parents can positively influence their children's alcohol use. One strategy they use is to provide their children with alcohol, believing it is the best way to teach their children how to drink responsibly. The impact of parental supply is not well understood and may be unintentionally harmful. This study will research the consequences of parental supply within the broader context of parent, child and peer relationships. It will help to determine how parental supply influences the different patterns of adolescent alcohol consumption over time, providing essential information to help parents prevent alcohol misuse in their children. Parents can play a pivotal role in prevention of alcohol misuse, but at present we don't know exactly how.

Title: The Australian Parental Supply of Alcohol Longitudinal Study (APSALS): Can parents teach their children to drink alcohol responsibly? Or, is one drop a drop too many?

Background: The Australian Parental Supply of Alcohol Longitudinal Study (APSALS) was established to investigate the short- and long-term associations between exposure to parental alcohol provision, early adolescent alcohol initiation, subsequent alcohol use, and alcohol-related harms, controlling for a wide range of parental, child, familial, peer, and contextual covariates. The cohort commenced with 1927 parent-child dyads comprised of Australian Grade 7 school students, and a parent/guardian. Baseline, one- and two-year follow-up data have been collected, and a three-year follow-up is underway. The data collected include child, familial, parental, and peer factors addressing demographics, alcohol use and supply, parenting practices, other substance use, adolescent behaviours, and peer influences. Baseline data show that only 5.8% of the Grade 7 adolescents had initiated alcohol consumption (of a whole serve of alcohol). Thus, the cohort is ideal for prospectively examining predictors of initiation and progression of alcohol use, which increases markedly through adolescence. Results to date have highlighted the importance of distinguishing between sipping and drinking of full serves of alcohol in the measurement of adolescent alcohol use as these represent distinct behaviours which occur in different environments.

Aims: The aims of this study are to determine if:

1. Parental supply is associated with the progression (acceleration/deceleration) in adolescent drinking over time;
2. The immediate and broader contextual factors mediate or moderate the relationship between parental supply and progression in adolescent drinking over time.

Design: A longitudinal study across three states in Australia (New South Wales, Tasmania, and Western Australia).

Recruitment: Catholic, Independent and Government school in NSW, Tasmania and Western Australia.

Data analysis: Logistic and binomial regression, and multi-level modelling will be used, and latent growth curve modelling (LGCM) within the framework of structural equations modelling (SEM). Moderating and mediating variables, both time variant (e.g., association with alcohol using peers) and time-invariant (e.g. gender) will be modelled. Outcome variables will be modelled using appropriate parametric distributions - Poisson distributions for count data outcomes and binomial distributions for categorical data outcomes. All analyses will be conducted in MPlus v5.2 or Stata. As recommended by Graham imputation of missing data will be done.

Sample size calculations: Due to the flexibility of the modelling approach taken (i.e., the number of parameters allowed to vary), there are a number of different scenarios that can be tested to determine an adequate sample size. The primary interest is in differences in drinking over time between groups who were supplied or not supplied alcohol. Thus, regression analysis will address drinking on a dummy-coded "parental supply" variable. It is hypothesised that parental supply of alcohol would alter the trajectory of harmful drinking by a regression coefficient of ±0.2. This regression coefficient is equivalent to a medium effect size difference between groups, a result which is not unexpected given past research. It was assumed that attrition would result in around 20% missing data on each subsequent measurement occasion. Based on simulation studies, and factoring in potential mediators and moderators, 600 children are required at a minimum. However, as small cell sizes for some associations may affect power, a final sample of ~1800 children and parents has been recruited.

Inclusion Criteria:

• Grade 7 High School enrolment
• Parental signed informed consent provided

Exclusion Criteria:

• none

The cohort includes 1927 Australian adolescents born from 1996-1999 (mean age at baseline: 12.9 years), and a parent or guardian (typically the mother (in 86.3% of dyads; mean age at baseline: 43.9 years)).