Table 3.
Risk of bias assessment (Newcastle–Ottawa Quality Assessment Scale criteria).
| Study | Selection | Comparability | Outcome | Quality Score | |||||
|---|---|---|---|---|---|---|---|---|---|
| Representativeness of Exposed Cohort | Selection of the Non-Exposed Cohort from Same Source as Exposed Cohort | Ascertainment of Exposure | Outcome of Interest Was Not Present at Start of Study | Comparability of Cohorts | Assessment of Outcome | Follow-Up Long Enough for Outcome to Occur (Median Duration of Follow-Up ≥6 Months) | Adequacy of Follow-Up | ||
| Danielsson et al. (2011) [33] | Participants were truly representative of adolescents of Stockholm, Sweden. Participants covered low, middle and high socio-demographic profiles and participated from 6 districts (18 schools and 79 classes) of Stockholm out of 18 districts. ★ | Yes ★ | Students answered questionnaires in school | Yes ★ | Early alcohol debut (proportion of friends who drink, smoking, truancy, bullying, more than 300SEK to spend per month), protective factors (more than 6 h spent with parents on weekends, relationship to parents and peers), parental monitoring, school environment were adjusted for multivariable logistic regression. ★ | Adolescent self-report | Yes ★ | 87% of adolescents participated at the first data collection and after two years 85% participated at the second data collection. ★ | Good |
| Degenhardt et al. (2015) [36] | Adolescents were truly representative of the community. Schools were randomly selected from a stratified frame of government, independent private and Catholic schools. From each type of school the probability of selection was proportional to the number of students of that age. ★ | Yes ★ | Students completed questionnaires by computer at school | Yes ★ | Wave of observation, sex, school location, parental separation/divorce, frequency of parental drinking, smoking, adolescents’ smoking, cannabis use, antisocial behaviour and signs of anxiety and depression were adjusted for repeated measures discrete time proportional hazards models. ★★ | Adolescent self-report | Yes ★ | 87% participated at the 6-month follow-up, 84% at the 12-month follow-up, 81% at the 18-month follow-up and 79% at 24-month follow-up. ★ | Good |
| Komro et al. (2007) [38] | Participants were not representative of adolescents of Chicago, USA. Only Chicago public schools were selected and students were predominantly African American (44%) or Hispanic (39%) and low income (79%). | Yes ★ | Parents completed survey at home and students completed at school | Yes ★ | Race/ethnicity, age, gender and family composition, parent/child communication, family alcohol discussions, peer alcohol use, peers’ supply of alcohol, parental monitoring and alcohol communication were adjusted for generalized linear mixed-model regression. ★★ | Adolescent self-report | Yes ★ | Between 91% and 96% participated at each of the 12-month and 24-month follow-up. ★ | Fair |
| McMorris et al. (2011) [27] | Representative samples were recruited from seventh grade students of Victoria and Washington states of Australia and USA respectively. ★ | Yes ★ | Students completed questionnaires at classroom | Yes ★ | Gender, age, and socioeconomic status were adjusted for path models. ★ | Adolescent self-report | Yes ★ | 97% participated at 12-month follow-up and 24-month follow-up. ★ | Good |
| Strandberg et al. (2014) [35] | 40 municipal schools participated from 13 counties out of 21 Swedish counties. ★ | Yes ★ | Parents received questionnaires by post and youth completed questionnaires in school | Yes ★ | Multilevel logistic regression. What confounders were adjusted for was not clearly stated. | Adolescent and parent self-report | Yes ★ | 92% adolescents and 75% parents participated at the 12-month follow-up and 88% adolescents and 68% parents participated at the 30-month follow-up. | Poor |
| van der Vorst et al. (2010) [32] | Participants were representative of two biological parent households of 20 municipalities of Netherlands. There were inclusion criteria of participants that indicate “parents had to be married or living together, and the siblings and their parents had to be biologically related”. ★ | Yes ★ | Family members (both parents and two adolescent children) completed questionnaires at home in the presence of a trained interviewer | Yes ★ | Structural path analysis. What confounders were adjusted for was not clearly stated. | Adolescent self-report | Yes ★ | 416 families participated at the 12-month follow-up and 404 families participated at the 24-month follow-up. ★ | Poor |
| Warner & White (2003) [37] | Participants were representative of white adolescents (89%) who lived in metropolitan, middle-class and working environment. | Yes ★ | Parents and adolescents completed self-reported questionnaires at home during recruitment and later completed in the project site | Yes ★ | Gender, socioeconomic status, religion were adjusted for hierarchical logistic regression models. ★ | Adolescent and parent self-report | Yes ★ | 91% participated at 3-year follow-up, 6-year follow-up and 13-year follow-up. Participation rate is not specified at the 18-year follow-up. | Poor |
Good quality: 3 or 4 stars (★) in selection domain AND 1 or 2 stars in comparability domain AND 2 or 3 stars in outcome domain; Fair quality: 2 stars in selection domain AND 1 or 2 stars in comparability domain AND 2 or 3 stars in outcome/exposure domain; Poor quality: 0 or 1 star in selection domain OR 0 stars in comparability domain OR 0 or 1 stars in outcome/exposure domain.