Health Education Research Advance Access published online on July 16, 2007
Health Education Research, doi:10.1093/her/cym030
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
School policies and binge drinking behaviours of school-aged children in Wales—a multilevel analysis
1 Public Health and Health Professionals Division, Office of the Chief Medical Officer, Welsh Assembly Government, Cathays Park, Cardiff CF10 3NQ, UK
2 Cardiff Institute of Society, Health and Ethics, Cardiff University, 53 Park Place, Cardiff CF10 3AT, UK
Correspondence to: * Correspondence to: C. Desousa. E-mail: desousac{at}cardiff.ac.uk
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussion and conclusionsFundingConflict of interest statementAcknowledgementsReferences |
|---|
Research has highlighted increased and earlier alcohol consumption among young people. This study aims to explore whether the type of school alcohol policy employed is associated with the frequent binge drinking behaviours of adolescents, after adjusting for known demographic and social factors. Integrated cross-sectional data were used from Welsh school surveys that assess the health behaviours of adolescents and school health policies. Frequent binge drinking was more likely to occur among older pupils, those living with one parent and pupils from more affluent backgrounds. Frequent binge drinking was also associated with pupils who bullied others, those with greater peer involvement and who felt pressured by schoolwork. The results suggested that strong parental and school bonds were protective factors against frequent binge drinking as were greater life satisfaction. Pupils who were bullied were less likely to have frequently binge drank. There was some evidence to suggest that written school policies are associated with lower likelihood of frequent binge drinking, in particular among boys and pupils with lower school attachment. However, there is a need for greater understanding of the differential population influence of school alcohol polices and an evaluation of their effectiveness.
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussion and conclusionsFundingConflict of interest statementAcknowledgementsReferences |
|---|
Research has highlighted increased and earlier alcohol consumption among young people [1]. A recent UK survey reported a 54% prevalence rate of binge drinking among 15- and 16-year olds, with particularly pronounced increases in binge drinking among young women [2]. While a range of explanations for such changes in consumption have been posited [3–5], an extensive body of literature has focussed on the influence of risk and protective factors on social development (e.g. [6, 7]). Risk and protective factors have been identified as biological, cognitive, social and environmental in nature. It is argued that these factors either provide greater potential for or inhibit the likelihood of a range of health risks, including alcohol misuse [8]. Hence, the greater the number of risk factors that cluster around individuals or particular groups in society, the more likely they are to engage in problematic alcohol use, although protective factors often have a crucial role at certain key points in the life course [9]. It has been suggested that such factors are frequently additive or interactive in nature and that they may act to both mediate and moderate behaviour [10, 11]. Particular attention has been paid to the relative health influence of the family, peers and school as contrasting or complementary environments.
It has been argued that the family plays a key part in preventing substance misuse by inducing risk, and/or encouraging and promoting protection and resilience [8, 12]. Aspects of both family structure [13, 14] and process [15, 16] have been examined to assess relative effects. It has been suggested that positive family relationships, along with bonding are associated with lower levels of alcohol misuse more frequently than family structure [12, 17, 18]. The influence of family processes on alcohol use, however, have been shown to be both moderated and mediated by the quantity and quality of social relations with peers. Their influence on alcohol consumption has received considerable attention [e.g. 19, 20] and it has been argued that variation in social relationships go some way to explaining observed gender differences in alcohol use [21]. In particular, it has been found that time spent with peers can reduce the effect of parental influences on alcohol use [14] and that higher levels of alcohol consumption are associated with greater involvement with friends, lower involvement in activities with parents and fewer pro-social family process [22].
In much the same way, it has been suggested that the school environment can both moderate and mediate family and peer processes as well as aggravate pupil alcohol and drug use, although few studies have assessed their relative influence [23, 24] or examined the policy environment. Increased alcohol misuse has been found to be associated with low perceived poor school satisfaction and disliking school [25] and aggressive behaviour such as bullying [9]. Less is known, however, about the provision and enforcement of school policies that promote supportive environments for health. Such influences are recognized in the concept of the health promoting school [26] and the ‘whole-school approach’ [27], which attempts to foster consistent environments that are conducive to helping children make healthy decisions [28]. Despite these developments, the evaluation of the impact of school policies on health risk has been limited [27, 29]. In the main, these have focussed on smoking and have, for example, assessed the mediating effect of school policies while adjusting for known family and peer influences [30].
Reviews of school alcohol policies, mostly in the United States, suggest that the implementation rather than the presence of a policy is associated with reduced consumption [25, 31]. In a European study, Maes and Lievens [32] found no association between alcohol consumption and health promotion policies, but did however observe a relationship between clearly formulated and communicated school rules and lower incidents of regular drinking. While a survey of schools in England reported that, in most schools, if pupils were found under the influence of alcohol, a letter would be sent home and if pupils were found drinking alcohol on school premises a note on the pupil's record or a verbal warning would be made. Pupils were referred to counsellors for repeated offences. However, the findings are limited since school policy data, as yet, has not been linked to pupil behaviours [33]. Furthermore, a report by the Office for Standards in Education found little evidence available to determine the impact of these school policies [34]. Within Wales no comparable data are available, although best practice guidelines are provided for schools in developing policies relating to substance misuse, including alcohol, substance misuse education and the management of substance use incidents [35].
This study aims to augment findings from the few studies that have evaluated school policies with regards to alcohol misuse in the United Kingdom. Using integrated cross-sectional data from Welsh school surveys that assess the general health behaviours and outcomes of adolescents and school health policies and promotion, and following previous approaches to the assessment of smoking policy [30], we sought to explore whether the type of school alcohol policy, employed by a sample of Welsh secondary schools, was associated with the frequent binge drinking behaviour of pupils, after adjusting for known demographic and social factors.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussion and conclusionsFundingConflict of interest statementAcknowledgementsReferences |
|---|
Data
The findings presented in this paper originate from the 2004 Health Behaviour in School-Aged Children (HBSC) study. The HBSC is a study conducted in collaboration with the World Health Organization Regional Office for Europe with cross-national data collected every 4 years and interim surveys conducted in Wales every 2 years. In the HBSC 2004 Welsh survey, one mixed ability class from each age group (11-, 12-, 13-, 14- and 15-year olds) was chosen from 80 randomly selected secondary schools in Wales, stratified by unitary authority and provision of free school meals. Pupils completed the questionnaire during a school lesson, following guidance from trained field workers. Questions included items on demography, general health behaviours, school experiences and pupils’ relationships with their friends and family. The survey design and data collection procedures are described in more detail in Currie et al. [36].
Data from the Health Promotion Policy and Practice (HPPP) in Welsh Secondary Schools Survey 2004 were also used [37–40]. These questionnaires were sent to all known secondary schools in 2004 in Wales and completed by headteachers or senior members of staff. Responses were received anonymously from 81% of schools (n = 225). The questionnaire assessed the ways in which health promotion issues were handled within schools and included items such as the type of school policies regulating both staff and pupil health behaviours at school and enforcement of these policies.
The HBSC and HPPP 2004 surveys were made available in English and Welsh and carried out between January and March 2004. HBSC 2004 pupil data were linked to HPPP school-level data by means of a unique identifier (schools = 54). A more homogenous sample of pupils was obtained by removing data for independent schools (n = 5), three schools with missing/do not know responses for policy relating to pupils drinking alcohol and all pupils of ethnic minority origin (7%, 374 pupils). In addition, only complete data were used, as required by MlwiN for multinomial analysis (schools = 46, pupils = 3882) and no imputation was used for missing covariates. There were no significant differences in gender between those respondents included and excluded from the final sample. However, those excluded from the final sample were significantly younger (P < 0.01) and hence less likely to have been binge drinking (P < 0.01). In the final sample, 48% of the pupils were boys and 52% girls with an average age of 14.00 years (SD 1.40).
For this study, binge drinking was defined as having five or more alcoholic drinks on the same occasion during the 30 days prior to the study. Based upon the distribution of frequencies, categorical responses (never, not in the past month, once, twice, three times, four times) were collapsed ‘a priori’ into ‘never’, ‘rarely’ (not in the past month or once) and frequently (two or more times) [41].
School characteristics
Time of school survey.
In 13 schools, the HBSC survey was carried out in January 2004. As a result, the responses to the binge drinking item may have been biased, relating to the Christmas and New Year period in the month prior to the time of the survey. In order to adjust for this possible bias, a dummy variable was created to represent the time of the survey (1 = survey carried out during January 2004, 0 = otherwise).
School policy
Headteachers were asked about the type of policy relating to pupils drinking alcohol on school premises (written policy/informal policy/no policy/don't know/not applicable). ‘Do not know’ and ‘not applicable’ were recoded to missing. The remaining responses were then dichotomized into ‘written policy’ and ‘no written policy’ (informal/no policy).
Pupil characteristics
Demography and family life.
- (i) Sex of pupil.
- (ii) Age.
- (iii) Living with parents.
- (iv) Family material wealth.
- (v) Home social environment.
- (vi) Parental support.
- (ii) Age.
The indicator ‘home social environment’ was an increasing scale based on the number of hours a child spent at home using the computer, watching TV or doing homework during the week and at the weekend (Cronbach's alpha, 
= 0.65). Parental support was assessed using a composite score derived from pupil's responses to how easy it was for them to talk to parents and/or step parents about ‘things that bothered them’ ( = 0.77). Responses ranged from ‘very difficult’ to ‘very easy’ and summed (min = 0, max = 12) [43].
Peer relationships.
- (i) Number of evenings spent with friends.
- (ii) Friendship support.
- (ii) Friendship support.
= 0.75; min = 0, max = 9).
School life.
- (i) How do you feel about school?
- (ii) Frequency of bullying someone in past couple of months.
- (iii) Frequency of being bullied in the past couple of months.
- (iv) Teacher support.
- (v) Classmate support.
- (vi) Pressure of schoolwork.
- (ii) Frequency of bullying someone in past couple of months.
= 0.86) and being bullied ( = 0.82) (min = 0, max = 28). Pupil's were asked to rate four statements (strongly disagree = 0 to strongly agree = 4) about how much help and encouragement they felt they received from their teachers ( = 0.82) [45]. Scores were summed with a higher score indicating greater teacher support ( = 0.81; min = 0, max = 16). Similarly for classmate support, a summation score was derived based on whether a pupil felt accepted by his/her classmates, whether classmates enjoyed being together, were kind and helpful and involved in organizing school events ( = 0.73). A dummy variable represented pressure of schoolwork with categories: ‘not at all’, ‘a little, ‘some’ and ‘a lot’.
Life satisfaction.
Life satisfaction was measured using a validated one-item scale depicted by a numbered ladder (0 = worst possible life, 10 = best possible life) [46]. Responses were centred about the median value of 8.0.
Model specification and estimation
The outcome variable, frequency of binge drinking, was treated as nominal data since categories were distinct and not equidistant (never, rarely and frequently). Moreover this study was concerned with explicitly modelling the odds ratios for factors associated with frequent binge drinking, and so the data were analysed using multinomial logistic regression. This is equivalent to estimating a set of binary logistic regression contrasts, which simultaneously compares each category with the reference group, in this case ‘never binge drinking’. Furthermore, multilevel or hierarchical modelling was employed in order to take account of the clustered structure of the data. Let yij represent the categorical response, frequency of binge drinking, for pupil i in school j, with pupils nested within schools. There were 46 schools with an average of 84 pupils per school (min = 22, max = 117). The response variable had s categories (s = 3) with the category t taken as the reference category, where t refers to ‘never binge drinking’. Thus, the probability of being in category s is 
and 
for each pupil, i.
 |
is the parameter estimate for each p explanatory variable and 
is a school-level random effect, assumed to be normally distributed with mean zero and variance 
. Separate coefficients of explanatory variables were specified for each contrast. If the parameter estimates were similar, then a common parameter for both contrasts was estimated, dependent on model convergence. Initial parameter estimates were obtained using a first-order marginal quasi-likelihood method which were then used as starting values for second-order penalized quasi-likelihood estimation [47]. Data preparation and univariate statistics, adjusted for clustering, were carried out in SPSS v.12.0 and SAS v.9 and multilevel regression analyses in MLwiN v.2.0.
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussion and conclusionsFundingConflict of interest statementAcknowledgementsReferences |
|---|
Two thirds of schools (n = 30) had a written policy for pupils drinking alcohol on school premises (Table I). Of note, only one of the 30 headteachers, who had stated that they had a written policy, felt that it was ‘quite successfully’ enforced. The remaining headteachers had felt that the policy was ‘very successfully’ enforced.
|
Forty-one per cent of pupils had never experienced binge drinking (41%, 1599), while 34% (1312) had rarely done so. However, 25% (971) of pupils reported having frequently binge drank on more than one occasion, in the month prior to the survey. Schools with and without written policies had approximately equal proportions of students who never, rarely and frequently binge drank, although schools with no written policy had a slightly higher proportion of pupils who frequently binge drank (
2 trend = 5.18, P < 0.10). Table II presents the frequencies (percentages) and means (SDs) describing the characteristics of pupils by frequency of binge drinking. There were statistically significant differences (P < 0.05) between pupils who never, rarely or frequently binged on alcohol across all explanatory variables except for gender, family wealth, home social environment, being bullied and classmate support indicators.
|
Coefficients of these pupil-level explanatory variables were simultaneously estimated in a multilevel multivariable multinomial logistic regression model (pupil-level only), adjusting for the time of the school survey. The corresponding odds ratios are shown in Table III. All variables were found to be statistically significant and retained in the model (P < 0.05), except for the home social environment indicator. All variable estimates, except for age, the number of evenings spent with friends and pupils’ perceptions of school, were similar across both logits. Hence, these parameter estimates were modelled as common coefficients for both rarely and frequently binge drinking. The school-level items relating to school policy and cross-level interaction terms between school policy and gender and pupils’ perceptions of school were then added to the model.
|
Table IV shows the odds ratios for the variables included in the pupil- and school-level model. The cross-sectional analysis indicated some evidence that the easier it was for a pupil to be able to talk to their parents about bothersome matters, the less likely they were to report binge drinking (OR = 0.87, P < 0.01, 95% CI (0.86, 0.88)). Similarly, pupils who felt relatively happy with their lives, as measured by their life satisfaction score, were less likely to frequently binge drink (OR = 0.83, P < 0.01, 95% CI (0.82, 0.84)). In contrast, pupils who reported fewer incidents of being bullied were also less likely to frequently binge drink (OR = 0.94, P < 0.01, 95% CI (0.93, 0.95)). Teacher support was negatively associated with binge drinking, with pupils who felt that they received encouragement and help from their teachers less likely to be involved in frequent binge drinking (OR = 0.90, P < 0.01, 95% CI (0.89, 0.90)).
|
As expected, the probability of pupils experiencing binge drinking increased with age. That is, for each 1-year increase in age, above the average age of
14 years, the odds of a pupil frequently rather than never binge drinking increased by more than two and half times (OR = 2.63, P < 0.01, 95% CI (2.47, 2.80)). Pupils who lived with only one biological parent were a third more likely to have been binge drinking than those living with both parents (OR = 1.36, P < 0.01, 95% CI (1.29, 1.43)). Increased family wealth was also associated with a greater probability of a child binge drinking (OR = 1.09, P < 0.01, 95% CI (1.07, 1.11)). Spending more time with friends in the evening increased the odds of binge drinking, especially frequent binge drinking (OR = 1.42, P < 0.01, 95% CI (1.37, 1.47)), as did being able to talk to friends about problems (OR = 1.19, P < 0.01, 95% CI (1.18, 1.21)). The greater the number of incidents of bullying others that a pupil was involved with, the greater the risk of frequent binge drinking (OR = 1.16, P < 0.01, 95% CI (1.15, 1.17)). Better classmate relationships was also associated with an increased probability of frequently binge drinking (OR = 1.15, P < 0.01, 95% CI (1.14, 1.16)). Pupils who felt ‘a little’ pressured by schoolwork were more likely to binge drink than those who did not feel pressured (OR = 1.14, P < 0.05, 95% CI (1.03, 1.27)).
Interaction terms and corresponding main effect terms
Understanding the implications of the cross-level interaction and associated main terms of gender and school policy requires recalculation of the odds ratios for each combination of categorical variables [48]. Figure 1 shows the predicted probability of frequent binge drinking by age, gender and type of school policy, based on cross-sectional analysis. Holding all other variables constant, the probability of frequent binge drinking was the lowest for boys who attended schools with a written alcohol policy. For example, the probability of a 15-year old boy frequently drinking was 0.25 compared with 0.31 if he attended a school with no written policy.
|
Figure 2 shows the predicted probabilities of frequent binge drinking by age, preferences towards school and school policy, holding all other variable constant. As with policy and gender, these predicted probabilities are calculated using the parameter estimates for both main and interaction terms. For pupils who had stated that they liked their school, either ‘a little bit’ or ‘a lot’, the predicted probabilities of frequently binge drinking were similar, irrespective of the type of school alcohol policy employed at their school. However, for pupils who did not like school and attended school with a written policy, the predicted probability of frequently binge drinking was lower compared with pupils who attended school with no written policy. This differential appeared to increase with age.
|
Finally, the positive covariance term (
= 0.13, P < 0.01) implied that schools with a higher (lower) than average prevalence of pupils rarely binge drinking also had higher (lower) than average prevalence of pupils frequently binge drinking. |
Discussion and conclusions |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussion and conclusionsFundingConflict of interest statementAcknowledgementsReferences |
|---|
There are a number of limitations to the present study that need to be considered. The results presented may have been biased due to the use of self-report data, with pupils overinflating reports in order to gain peer status or underreporting for fear of identification or recrimination [49, 50]. Also, it was not possible to determine the direction of causality based on the use of cross-sectional data. For example, in some schools, effective policies may have resulted in lower prevalence rates of binge drinking, while in other schools, the lower prevalence of binge drinking among pupils may have only necessitated the use of non-written policies. The lower age of respondents excluded from the analyses suggests that the results presented may be biased. However, without complete data, it was not possible to determine the extent of the bias. A further limitation was the remaining statistically significant school-level random variation (P < 0.01) and possible extra multinomial variation. This suggested that there were other unaccounted for school-level variables associated with pupils’ binge drinking behaviours and/or further school- or pupil-level random variation should be included in the model. In addition, the relatively small number of schools in the sample and the lack of convergence did not allow for further fixed main, interaction or random terms to be added to the modelling.
The restriction on the number of theoretically relevant school-level variables that could have been included in the modelling was due to the lack of variation among the responses to the HPPP survey. Overall, most headteachers had stated that their school had sufficient health-promoting policies and had successfully implemented procedures. Furthermore, questionnaire items may not have sufficiently assessed the complexities of school policy. For example, in response to the type of school policy in place, a few headteachers commented that they had both written and informal policies, which was not an available response option. Hence, more accurate measures of school environment may be provided by determining the level of pupils’ awareness of policies, teachers’ experiences of implementation, as well as assessments of curriculum coverage of alcohol and the broader aspects of school climate [23]. These issues underline the need for further work to improve the validity and reliability of generic questionnaires that seek to assess the dimensions and effectiveness of school substance misuse policies. Finally, caution should be exercised in generalizing these results from Welsh school surveys to other settings, especially where cultural differences may exist.
Despite these limitations, we believe that these results provide an informative, albeit tentative, insight into the association between school health policies and adolescent binge drinking in the United Kingdom. This study set out to explore the whether the type of school alcohol policy employed had a mediating effect on the frequent binge drinking behaviours of adolescents, after adjusting for known demographic and social factors. Older pupils, pupils who lived with one parent and pupils from wealthier families were more likely to frequently binge drink. Similarly, being able to talk to friends and spending more time with them in the evenings increased the risk of a pupil frequently binge drinking. Pupils who reported greater peer classmate support or those who bullied others were also more likely to engage in excessive drinking behaviour.
Being able to talk to parents acted as a protective factor. Pupils with stronger school bonds who liked school, felt supported by their teachers and generally felt good about their life were also less likely to binge drink. This supports previous research into risk and protective factors [8] and highlights the need for more intensive interventions for those engaged in alcohol misuse at an early age [51].
Adjusting for pupil characteristics and familial and peer influence factors, we found that the presence of a written school policy may have differential associations with pupil alcohol misuse, according to pupils’ gender and perceptions of school. There was some evidence that a written alcohol policy was associated with lower frequent binge drinking amongst boys but not girls. Similarly, a written policy was associated with lower likelihood of binge drinking among adolescents who did not like school than those who did like school. These findings suggest that males and those with weaker attachments to school may respond more favourably to clearly formulated regulations [32]. To further understand these findings, both in-depth qualitative and longitudinal quantitative studies are needed on how alcohol-related policies and curriculum initiatives are implemented and integrated into the school curriculum. In addition, a more detailed and ethnographic examination of the frequency of incidents involving pupils drinking alcohol on school premises is required, detailing how violations are dealt with and whether policies are effective in reducing the frequency of incidents [24, 31]. It may also be necessary to examine the relationship between the school context and wider community processes [23].
In conclusion, this paper is one of the few studies to have attempted to explore the association between school policies and alcohol misuse in relation to family, peer and school environments. For a school-wide approach, there was some evidence to suggest that written school policies are associated with lower likelihood of frequent binge drinking among adolescents, in particular among boys and pupils with lower school attachment. Our findings indicate that there is a need for greater understanding of the differential population influence of school alcohol polices and an evaluation of their effectiveness. Without this, there is a danger that whole population approaches may fail to reach particular groups and that policy may remain poorly integrated with curriculum education and school/community approaches.
|
Funding |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussion and conclusionsFundingConflict of interest statementAcknowledgementsReferences |
|---|
The Welsh Assembly Government and Cardiff University to CD.
|
Conflict of interest statement |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussion and conclusionsFundingConflict of interest statementAcknowledgementsReferences |
|---|
None declared.
|
Acknowledgements |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussion and conclusionsFundingConflict of interest statementAcknowledgementsReferences |
|---|
The views expressed in this paper are those of the authors and do not necessarily represent the views of the Welsh Assembly Government.
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussion and conclusionsFundingConflict of interest statementAcknowledgementsReferences |
|---|
1. Institute of Alcohol Studies. Adolescents and Alcohol Fact Sheet. Available at: http:// www.ias.org.uk. Accessed: 26 October 2006.
2. Plant MA, Miller P, Plant ML. Trends in drinking, smoking and illicit drug use among 15 and 16 year olds in the UK (1995–2003). J Subst Use (2005) 10:331–9.[CrossRef]
3. Goddard E, Higgins V. Smoking, Drinking and Drug Use amongst Young Teenagers in 1998 (1999) Vol. 2. London: The Stationery Office.
4. Sutherland I, Willner P. Patterns of substance use among English adolescents. Addiction (1998) 93:1199–208.[CrossRef][ISI][Medline]
5. Wright L. Young People and Alcohol: What 11-Year Olds Know, Think and Do (1999) London: Health Education Authority.
6. Hawkins J, Catalano R, Miller J. Risk and protective factors for alcohol and other drug problems in adolescence and early adulthood. Psychol Bull (1992) 112:64–105.[CrossRef][ISI][Medline]
7. Vakalahi H. Adolescent substance use and family-based risk and protective factors: a literature review. J Drug Educ (2001) 31:29–46.[CrossRef][ISI][Medline]
8. Communities that Care. Youth at Risk? A National Survey of Risk Factors, Protective Factors and Problem Behaviour among Young People in England, Scotland and Wales (2003) London: Communities that Care.
9. Murphy S, Bennett P. Lifespan, gender and cross cultural perspectives on health psychology. In: A Handbook of Health Psychology—Sutton S, ed. (2005) London: Sage.
10. Catalano R, Hawkins J. The social development model: a theory of antisocial behaviour. In: Delinquency and Crime: Current Theories—Hawkins J, ed. (1996) Cambridge: Cambridge University Press.
11. Bynner J. Childhood risks and protective factors in social exclusion. Child Soc (2001) 15:285–301.[CrossRef]
12. Velleman R, Templeton L. Reaching out—promoting resilience in the children of substance misusers. In: Secret Lives: Living with Substance. A Guide for Practitioners/Managers Who Work Directly with Children Who Live in Substance Misusing Families—Evans D, Harbin F, Murphy M, eds. (2005) Lyme Regis: Russell House.
13. Smith C, Nutbeam D. Adolescent drug use in Wales. B J Addict (1992) 87:223–7.
14. Miller P, Plant MA. The family, peer influences and substance use among UK teenagers. J Subst Use (2003) 8:19–26.
15. Piko B. Perceived social support from parents and peers: which is the stronger predictor of adolescent substance use? Subst Use Misuse (2000) 35:617–30.[ISI][Medline]
16. Sokolkatz J, Dunham R, Zimmerman R. Family structure versus parental attachment in controlling adolescent deviant behavior: a social control model. Adolescence (1997) 32:199–215.[Medline]
17. McArdle P, Wiegersma A, Gilvarry E, et al. Family structure and function and youth drug use. Addiction (2002) 97:329–36.[CrossRef][ISI][Medline]
18. Kuntsche EN, Kuendig H. What is worse? A hierarchy of family-related risk factors predicting alcohol use in adolescence. Subst Use Misuse (2006) 41:71–86.[CrossRef][ISI][Medline]
19. Honess T, Seymour L, Webster R. The Social Contexts of Underage Drinking (2000) London: Home Office.
20. Lilja J, Larsson S, Wilhelmsen BU, et al. Perspectives on preventing adolescent substance use and misuse. Subst Use Misuse (2003) 38:1491–530.[CrossRef][ISI][Medline]
21. Wilsnack R, Vogeltanz N, Wilsnack S, et al. Gender differences in alcohol consumption and adverse drinking consequences: cross-cultural patterns. Addiction (2000) 95:251–65.[CrossRef][ISI][Medline]
22. Kloep M, Hendry LB, Ingebrigtsen JE, et al. Young people in ‘drinking’ societies? Norwegian, Scottish and Swedish adolescents’ perceptions of alcohol use. Health Educ Res (1993) 16:279–91.[Medline]
23. Roski J, Perry C, McGovern P, et al. School and community influences on adolescent alcohol and drug use. Health Educ Res (1997) 12:255–66.
24. Aveyard P, Markham W, Cheng K. A methodological and substantive review of the evidence that schools cause pupils to smoke. Soc Sci Med (2004) 58:2253–65.[CrossRef][ISI][Medline]
25. Andersen A, Holstein BE, Due P. School-related risk factors for drunkenness among adolescents: risk factors differ between socio-economic groups. Eur J Public Health (2006) 17:27–32.[CrossRef][ISI][Medline]
26. Denman S, Moon A, Parsons C, et al. The Health Promoting School—Policy, Research and Practice (2002) London: Routledge Falmer Publishers.
27. Lister-Sharp D, Chapman S, Stewart-Brown S, et al. Health promoting schools and health promotion in schools: two systematic reviews. Health Technol Assess (1999) 3:22.
28. World Health Organisation. Promoting Health through Schools (1997) Geneva, Switzerland: World Health Organisation.
29. Flay B. Approaches to substance use prevention utilising school curriculum plus social environment change. Addict Behav (2000) 25:861–85.[CrossRef][ISI][Medline]
30. Moore L, Roberts C, Tudor-Smith C. School smoking policies and smoking prevalence among adolescents: multilevel analysis of cross-sectional data from Wales. Tob Control (2001) 10:117–23.
31. Evans-Whipp T, Beyers JM, Lloyd S, et al. A review of school drug policies and their impact on youth substance use. Health Promot Int (2004) 19:227–34.
32. Maes L, Lievens J. Can the school make a difference? A multilevel analysis of adolescent risk and health behaviour. Soc Sci Med (2003) 56:517–29.[CrossRef][ISI][Medline]
33. Fuller E, ed. Smoking, Drinking and Drug Use among Young People in England in 2000, National Centre for Social Research and the National Foundation for Educational Research (2005) London: TSO.
34. Office for Standards in Education (England) (OFSTED). Drug Education in Schools Hmi 2392 (2005) London: OFSTED.
35. National Assembly for Wales. Substance Misuse: Children and Young People (2002) Circular No: 17/02. Cardiff, UK: National Assembly for Wales.
36. Currie C, Samdal O, Boyce W, et al, eds. Health Behaviour in School-Aged Children: A WHO Cross-National Study (HBSC), Research Protocol for the 2001/2002 Survey (2001) Edinburgh, UK: Child and Adolescent Health Research Unit (CAHRU),University of Edinburgh.
37. Smith C, Roberts C, Nutbeam D, et al. The health promoting school: progress and future challenges in Welsh secondary schools. Health Promot Int (1992) 7:171–9.
38. Smith C, Frankland J, Playle R, et al. A survey of health promotion in welsh primary schools, 1993. Health Educ J (1994) 53:237–48.
39. Tudor-Smith C, et al. Health promotion in Welsh primary schools, 1998. Int J Health Promot Educ (1999) 37:81–8.
40. Tudor-Smith C, Roberts C, Parry-Langdon N, et al. A survey of health promotion in Welsh secondary schools,1995. Health Educ (1997) 6:225–32.
41. Kolbe LJ, Kann L, Collins JL. Overview of the youth risk behavior surveillance system. Public Health Rep (1993) 108((Suppl. 1)):2–10.[ISI][Medline]
42. Currie C. Health and well-being and their association with reported socio-economic circumstances among school-aged children in Europe and North America: the WHO HBSC Study. In: Child Well-being, Child Poverty and Child Policy in Modern Nations—Vleminckx K, Smeeding T, eds. (2000) Bristol, UK: Policy Press.
43. Smith C, Roberts C, Moore L. Parents and adolescent smoking. J Inst Health Educ (1995) 33:104–9.
44. Olweus D. Bullying among schoolchildren: intervention and prevention. In: Aggression and Violence Throughout the Life Span—Peters RD, McMahon RJ, Quinsey VL, eds. (1992) Newbury Park, CA: Sage Publications.
45. Samdal O, Wold B, Torsheim T. Rationale for school items: the relationship between students’ perception of school and their reported health and quality of life. In: Health Behaviour in School-Aged Children. Research Protocol for the 1997–98 Survey—Currie C, ed. (1998) Edinburgh, UK: Child and Adolescent Health Research Unit (CAHRU),University of Edinburgh.
46. Cantril H. The Pattern of Human Concern (1965) New Brunswick, NJ: Rutgers University Press.
47. Rasbash J, Steele F, Browne W, et al. A Users Guide to Mlwin Version 2.0 (2004) Bristol, UK: Centre for Multilevel Modelling, University of Bristol.
48. Chen JJ. Communicating complex information: the interpretation of statistical interactions in multiple logistic regression analysis. Am J Public Health (2003) 9:1376–7.
49. Percy A, McAlister S, Higgins K, et al. Response consistency in young adolescents’ drug use self-reports: a recanting rate analysis. Addiction (2005) 100:189–96.[CrossRef][ISI][Medline]
50. Shillington AM, Clapp JD. Self-report stability of adolescent substance use: are there differences for gender, ethnicity and age? Drug Alcohol Depend (2000) 60:19–27.[CrossRef][ISI][Medline]
51. Johnston K, Strader T, Berbaum M, et al. Reducing alcohol and other drug use by strengthening community, family, and youth resiliency: an evaluation of the creating lasting connections program. J Adolesc Res (1996) 11:36–67.[Abstract]
Received on November 20, 2006; accepted on April 25, 2007
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||