HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Figures Only Full Text (PDF) All Versions of this Article: AJPH.2004.047399v1 96/9/1582    most recent Submit a response View Shopping Cart Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (7) Citing Articles via Google Scholar Google Scholar Articles by Patton, G. C. Articles by Bowes, G. Search for Related Content PubMed PubMed Citation Articles by Patton, G. C. Articles by Bowes, G. Related Collections School Health Prevention Surveys Alcohol Drugs Other Tobacco

George C. Patton, MD, Lyndal Bond, PhD, John B. Carlin, PhD, Lyndal Thomas, BSc(Hons), Helen Butler, BA(Hons), Dip Ed, G Dip in Adol Health, Sara Glover, PhD, Richard Catalano, PhD and Glenn Bowes, PhD

George Patton is with the Centre for Adolescent Health and Murdoch Children’s Research Institute, Melbourne, Australia. Lyndal Bond is with the Centre for Adolescent Health and Department of Pediatrics, University of Melbourne, Melbourne. John Carlin is with the Clinical Epidemiology and Biostatistics Unit, Department of Pediatrics, University of Melbourne. At the time of the study, Lyndal Thomas and Sara Glover were with the Centre for Adolescent Health and Department of Pediatrics, University of Melbourne. Helen Butler is with the Centre for Adolescent Health and Murdoch Children’s Research Institute. Richard Catalano is with the Seattle Social Development Research Group, University of Washington, Seattle, Washington. Glenn Bowes is with the Department of Pediatrics, University of Melbourne.

Correspondence: Requests for reprints should be sent to George C Patton, William Buckland House, 2 Gatehouse St, Parkville, Victoria 3052, Australia. (e-mail: george.patton{at}rch.org.au; lyndal.bond{at}rch.org.au).

	ABSTRACT

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION References

 

Objectives. We sought to test the efficacy of an intervention that was designed to promote social inclusion and commitment to education, in reducing among students health risk behaviors and improving emotional well-being.

Methods. The design was a cluster-randomized trial in 25 secondary schools in Victoria, Australia. The subjects were 8th-grade students (aged 13 to 14 y) in 1997 (n=2545) and subsequent 8th-grade students in 1999 (n=2586) and 2001 (n=2463). The main outcomes were recent substance use, antisocial behavior, initiation of sexual intercourse, and depressive symptoms.

Results. At 4-year follow-up, the prevalence of marked health risk behaviors was approximately 20% in schools in the comparison group and 15% in schools in the intervention group, an overall reduction of 25%. In ordinal logistic regression models a protective effect of intervention was found for a composite measure of health risk behaviors in unadjusted models (odds ratio [OR]= 0.69; 95% confidence interval [CI]= 0.50, 0.95) and adjusted models (OR= 0.71; CI =0.52, 0.97) for potential confounders. There was no evidence of a reduction in depressive symptoms.

Conclusion. The study provides support for prevention strategies in schools that move beyond health education to promoting positive social environments.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION References

 
Young people spend well over one third of their waking hours in school. For this reason, schools have often been a setting for preventive intervention. The use of school curricula to promote protective attitudes and skills has met with some success in reducing health risk behaviors.^1–3 However, there have been some spectacular failures, which have led to calls for a reexamination of current approaches to prevention in schools.⁴

One potential but currently neglected focus for prevention is the school’s social milieu.^5,6 There is much evidence that the school’s social atmosphere affects patterns of substance use, antisocial and disruptive behaviors, as well as how well students learn. Advocates of health promotion have argued that addressing organizational processes and social relationships are likely to be effective in bringing about behavioral change.^7–9 Despite the attractiveness of such a health promotional approach, relatively few strategies have been tested that use this approach.^10–12

The Gatehouse Project intervention was designed as a structured process to promote a sense of social inclusion and connection in secondary schools, building on the principles of the Health Promoting Schools Project.^7,13,14 The process involved use of information about a school’s social climate to assist in setting priorities for action. Intervention elements ranged from establishing an inclusive classroom environment to creating opportunities for student participation in school life beyond the classroom, and included implementing a student curriculum that teaches interpersonal communication and emotional management.^15,16 We examine the school-level effects of this intervention on indexes of health and behavior in lower secondary schools. measured at 2-year and 4-year follow-up.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION References

 
Intervention
The Gatehouse Project intervention strategies have previously been described elsewhere.^15–17 The intervention occurred during a 2-year period beginning in term 2 (April 19 to June 18), 1997. The structured process had 4 elements: (1) feedback from a student survey about security (a student’s personal sense of safety), communication with teachers, and broader participation in school life; (2) within each school, recruitment of staff involved in administration, student welfare, curriculum, or all 3 to a coordinating action team with a focus on school policies and professional practice of teachers; (3) consultation and training regarding specific intervention strategies; and (4) a curriculum element that focused on problem-solving in situations in which young people commonly experience emotional difficulties.

This strategy typically included the following sequence: (1) conduct a survey of the students, (2) provide feedback from the school social profile, based on the student survey, to the school-based action team, (3) consult with the school team regarding intervention priorities, and (4) train teachers in the selected strategies. The feedback process included an examination of the school profile of students’ school environment experiences, which was used to set priorities, facilitate discussion, and examine specific areas in need of action. Thus, for example, a school with students reporting a high level of deliberate social exclusion by peers might choose this particular area as a priority focus. Strategies varied between schools according to students’ perceptions of need; but, the implementation of school policy and curriculum elements that focused on social and emotional skills and strategies to promote inclusive relationships within the classroom were always addressed. The curriculum element of the intervention was designed to be taught in the 8th grade during a 10-week period in English, health, or personal development classes. The median number of lessons using the Gatehouse Project Intervention curriculum in the first year, as reported by the school liaison team, was 20 (approximately 15 hours of instruction). An average of 40 hours of professional development were provided by the intervention team to each school during the first and second years, respectively, with the time split between a focus on curriculum and whole school strategies.¹⁸

Design and Participants
A cluster randomization evaluation design was selected, with the school as the intervention unit. Twelve educational administrative districts were randomly sampled from 64 across metropolitan Melbourne. These were randomized to intervention and control status and the schools were pooled. The 2 pools were stratified by school administration so that 6 government and 6 independent/Catholic schools could be selected from each. This approximates the ratio of school types between these strata within the metropolitan area.

From the 10 nonmetropolitan school districts outside of Melbourne, 2 regional districts were selected and randomly allocated to intervention and control status. Four schools (2 government, 2 independent /Catholic) were then randomly selected from each pool of schools in these districts. From the total group (metropolitan and nonmetropolitan), 2 schools—1 in the intervention group and 1 in the control—declined to participate, citing involvement in other projects. Four further schools were unable to participate because of threatened closure or amalgamation.

Three cross-sectional surveys of 8th grade students (aged 13–14 y) in the participating schools were conducted at 2-year intervals. The initial survey took place in the school classroom between February 17 and March 28, 1997 (term 1), before the intervention. The survey was a self-administered questionnaire conducted on laptop computers provided by the research team. Absent students were surveyed at school at a later date or by telephone. New cohorts of 8th grade students were surveyed between April 19 and June 18, 1999 (term 2) and again between April 23 and June 22, 2001 (term 2); the items from the initial survey were identical but were administered in a pencil and paper format. Student participation on each occasion was voluntary and required written parental consent. School participation rates are shown in Figure 1. The final sample consisted of 11 schools in the intervention group and 14 in the control group; 1 school in the intervention group failed to provide complete behavioral outcome data in the 1999 and 2001 surveys and so was not included in the analysis. Participation rates were initially slightly higher in the intervention group but fell to a level similar to that in the control group at the last survey.

View larger version (19K): [in this window] [in a new window]   FIGURE 1—— Design of and participants in the Gatehouse Project Intervention, 1997, 1999, and 2001. aUnavailable because of proposed closure or merger in 1997.

Dichotomous variables were defined for estimating prevalence of use. Any substance use was defined as either having used alcohol in the previous week, tobacco in the previous month, or cannabis in the previous 6 months. Heavy substance use was defined as binge drinking, tobacco use on at least 3 days in the previous week, or cannabis use at least weekly. A composite outcome on 3 levels was defined for the purpose of multivariate analysis: no recent use of any of the 3 substances, recent use of at least 1, and recent heavy use of 1 or more.

Antisocial behavior was assessed with items from the Self-Reported Early Delinquency Scale covering property damage, interpersonal violence, and theft in the previous 6 months.²¹ Any antisocial behavior referred to at least 1 instance in the previous 6 months and frequent antisocial behavior referred to 2 or more instances. A composite antisocial behavior outcome on 3 levels was defined for each respondent for multivariate analysis: no reported behavior, 1 instance in the past 6 months, and more than 1 instance in the past 6 months.

Early initiation of sexual intercourse was assessed with a single item about ever having had sexual intercourse (sexual intercourse, "gone all the way").

In order to assess the overall effect of the intervention on health risk behaviors, 2 further composite 3-level outcomes were defined, incorporating the substance use indicators, antisocial behavior, and early initiation of sexual intercourse. Any risky behavior was defined on 3 levels as either none, 1 behavior (any substance use, any antisocial behavior or early initiation of sexual intercourse), or 2 or more behaviors at this level. Marked risky behavior was defined as either none, 1 behavior at the highest level (heavy substance use, report of multiple antisocial behaviors, or early initiation of sexual intercourse), or 2 or more behaviors at this level.

Emotional problems were assessed at baseline (1997) using a computerized revised Clinical Interview Schedule (CIS-R).²² The total scores were dichotomized at a cut-off point of 11/12.²³ In later surveys (1999 and 2001) the short Mood and Feelings Questionnaire was used to assess depressive symptoms with a cut-off point of 11/12, which was used to delineate high symptom levels.²⁴

School commitment was assessed with a questionnaire comprising 23 items and 5 subscales reflecting school attachment, student–teacher communication, perceived opportunities for participation, disincentives, and rewards for participation.²⁵

Data Analysis
Data analysis was designed to test the hypotheses that the prevalence of behavioral and emotional problems would be lower among students in the schools in the intervention group compared with those in control group at later surveys of 8th grade students. Power estimates for more common behavioral and emotional problems with a prevalence of 25% and an intraclass correlation of .05 suggested that the study should have 80% power to detect a 20% reduction in prevalence at a significance level of .05. Statistical analyses were conducted with the Stata 8.0 program (Stata Corp LP, College Station, Tex). All analyses were conducted using an intention to treat principle with the intervention categorized dichotomously. All prevalence estimates and measures of association have used robust "information-sandwich" estimates of standard errors with adjustment for clustering within schools.²⁶ To control for potential confounding effects attributable to imbalance between groups, logistic and ordinal multiple regression models were used. Ordinal logit models were fitted to 3-level composite end points (substance use and antisocial behavior). The 2 summary measures reflecting any or marked health risk behavior were also modeled as 3-level ordinal variables (none, 1, and 2 or more). A Wald test was used to test for deviation from the constant odds ratio assumption in the ordinal logit models.²⁷

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION References

 
Comparison of demographic characteristics between intervention and control groups in 1997, 1999, and 2001 revealed small between-group differences (Table 1).

View larger version (24K): [in this window] [in a new window]   FIGURE 2—— Prevalence (with 95% confidence intervals) of health risk behaviors among 8th-grade students in schools, intervention and control groups in 1997 (a), 1999 (b), and 2001 (c).

Intervention effects were formally tested in a series of binary and ordinal logistic regression models, depending on whether the outcome was defined on 2 or 3 levels (Table 2). At baseline in 1997, no differences in either specific or summary measures of health risk behaviors were apparent between schools in the intervention and control groups. In 1999, point estimates of risk for substance use appeared lower in the intervention group, but 95% confidence intervals for odds ratios did not exclude 1.0 in both unadjusted and adjusted models. Similarly, in regard to both any and marked health risk behaviors a nonsignificant trend was apparent for the odds to be lower in the intervention group. In 2001, a more consistent pattern of lower risk across all outcomes was found in the intervention group. The association between group and marked risky behaviors (OR= 0.69; CI= 0.5, 0.95) indicated a lower risk for each level of behavioral problems among students in the schools in the intervention group compared with those in the control group, a difference that is still evident after adjustment for gender, cultural background, and parental marital status as possible confounders. The Wald test for parallel regression did not suggest deviation from the assumption in either the unadjusted (₁²=0.08, P=.77) or adjusted models (₁²=0.03, P=.87).

Student Commitment
Mean scores on the subscales for student commitment to education did not differ between groups at baseline. No marked differences were found in the follow-up surveys, with the possible exception of a trend for improved student–teacher communication in 1999 (t = 1.9, P = .07).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION References

 
Four years after beginning an intervention to promote social inclusion within schools, patterns of health risk behaviors among students in intervention schools differed from those in schools in the control group. Marked health risk behaviors were reported by approximately 15% of students in the intervention school group after the intervention, compared with 20% of those in the control group, an overall reduction of one quarter. This difference arose from lower rates of substance use, antisocial behavior, and early initiation of sexual intercourse by students in the intervention schools. This group difference was not explained by changes in student emotional problems because these did not differ between the intervention and control groups before or after intervention.

These reductions in risky behaviors are greater than those found in most recent studies of health education.^9,10,28,29 These findings stand out in 2 further respects. First, the effect was apparent across a range of behaviors. Second, the changes were apparent in subsequent cohorts of students, a finding that is consistent with the intent of maximizing sustainability by nesting the intervention within normal school processes and curricula. Given the effects of adolescent-initiated health risk behaviors on health later in life, reductions of this kind could have major public health benefits if the approach were adopted broadly.

There were some study limitations. Cluster randomization took place at a school district level, but the intervention and analyses were at the individual school level. This design was necessary to reduce the risks of contamination but is a weaker randomization design. The number of schools that were available after randomization was diminished by nonparticipation of 6 selected schools, raising questions about the effectiveness of randomization, as well as study power. Similarities in the demographic profile of intervention and control groups across surveys are consistent with the integrity of randomization, but a possibility of unmeasured confounding cannot be totally excluded.

Although overall response rates were moderately high, some differential response between intervention and control groups in the first 2 surveys might account for our findings. Nonresponse tended to be greater in the control group, and because nonresponders have higher levels of health risk behaviors, differential nonresponse seems more likely to have led to an underestimation of intervention effect. Social desirability, a potential source of measurement bias in nonblinded trials, was limited by integrating the intervention within usual school curriculum and administrative processes.³⁰ The health education component was delivered by teachers through the usual curriculum, minimizing the chances that students were aware of being part of an intervention. It is possible that the first cohort—the 1997 baseline group—were aware of being a special group, but this is unlikely to have been the case in subsequent surveys. Finally, we cannot exclude chance variation as an explanation for our findings, although changes in health risk behaviors after intervention consistently moved in a favorable direction at follow-up surveys.

Our findings carry implications for concepts of how schools may influence student behavior. Measures of school engagement were strongly associated with health risk behaviors, but behavioral change in the intervention schools was not matched by equivalent changes on these measures. This raises a question of whether the constructs of school social inclusion and student connection to education were adequately operationalized to capture change in this study.

The intervention also had no clear effect on emotional problems. It is possible that the intervention was not sufficiently specific or sustained to produce a detectable effect, given that study power was marginal. Alternatively, key determinants of depressive symptoms may differ from those for health risk behaviors, perhaps operating outside the secondary school setting or at a developmentally earlier point.

Our study is a rare example of the implementation and evaluation of a complex systems intervention to improve health.³⁰ The elements of the intervention process—feedback on the social context, creation of a coordinating structure (i.e., the action team), and ongoing consultation—may be useful in other settings. The process coordinated health promotional work with varied foci, as well as provided a means for nesting a health agenda within a school’s policy and practice framework. The result was the development of interventions tailored to the needs of particular schools but differing between schools. Although there remains much to learn about the wider application of this particular approach, our findings support strategies to promote the social milieu of schools as a way of achieving better health and learning outcomes.^9,10

	Footnotes

 
Peer Reviewed

Contributors
G. C. Patton designed the overall study and supervised its implementation. L. Bond was the coordinator of research field work for most of the trial. J. B. Carlin and L. Thomas provided technical assistance with data analysis. H. Butler and S. Glover designed and managed the intervention in schools. R. Catalano provided advice on the conception and design of the intervention. G. Bowes assisted with the study design.

Human Participant Protection
The study protocol was approved by the Royal Children’s Hospital ethics in human research committee.

Accepted for publication September 23, 2005.

	References

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION References

 
1. Botvin GJ, Baker E, Dusenbury L, Botvin EM, Diaz T. Long-term follow-up results of a randomized drug abuse prevention trial in a white middle-class population. JAMA. 1995;273:1106–1112.[Abstract/Free Full Text]

2. Ellickson PL, McCaffrey DF, Ghosh-Dastidar B, Longshore DL. New inroads in preventing adolescent drug use: results from a large-scale trial of project ALERT in middle schools. Am J Public Health. 2003; 93:1830–1836.[Abstract/Free Full Text]

3. Ellickson PL, Bell RM. Drug prevention in junior high: a multisite longitudinal test. Science. 1990; 247:1299–1305.[Abstract/Free Full Text]

4. Peterson AV, Kealey KA, Mann SL, Marek PM, Sarason IG. Hutchinson smoking prevention project: long-term randomized trial in school-based tobacco use prevention—results on smoking. J Natl Cancer Institute. 2000;92:1979–1991.[Abstract/Free Full Text]

5. Rutter M, Maughan B, Mortimore P, Ouston J, Smith A. Fifteen Thousand Hours: Secondary Schools and their Effects on Children. London: Open Books, 1979.

6. Resnick MD, Bearman PS, Blum RW, Bearinger L, Harris KM, Jones J et al. Protecting adolescents from harm: findings from the National Longitudinal Study on Adolescent Health. JAMA. 1997;278:823–832.[Abstract/Free Full Text]

7. World Health Organization/UNESCO/UNICEF. Comprehensive School Health Education: Suggested Guidelines for Action. Geneva: World Health Organization; 1992.

8. Moon AM, Mullee MA, Rogers L, Thompson RL, Speller V, Roderick P. Helping schools become health-promoting environments: an evaluation of the Wessex Healthy Schools Award. Health Promotion Int. 1999; 14:111–122.[Abstract/Free Full Text]

9. Flay BR. Approaches to substance use prevention utilizing school curriculum plus environmental social change. Addict Behav. 2000;25:861–885.[CrossRef][Web of Science][Medline]

10. Lynagh M, Schofield MJ, Sanson-Fisher RW. School health promotion programs over the past decade: a review of the smoking, alcohol, and solar protection literature. Health Promotion Int. 1997;12: 43–60.[Abstract/Free Full Text]

11. Hawkins JD, Catalano RF, Kosterman R, Abbott R, Hill KG. Preventing adolescent health-risk behaviors by strengthening protection during childhood. Arch Pediatrics Adol Med. 1999;153:226–234.

12. Olweus D. Bully/victim problems among schoolchildren. In: Rubin K, Pepler D, eds. The Development and Treatment of Childhood Aggression. Hillsdale, NJ: Lawrence Erlbaum Associates, 1991.

13. Effective School Health Promotion—Towards Health Promoting Schools. Canberra: National Health and Medical Research Council; 1996.

14. McNeely CA, Nonnemaker JM, Blum RW. Promoting school connectedness: evidence from the National Longitudinal Study of Adolescent Health. J School Health. 2002;72:138–146.[Web of Science][Medline]

15. Patton GC, Bond L, Butler H, Glover S. Changing schools, changing health? The design and implementation of the Gatehouse Project. J Adol Health. 2003; 33:231–239.[CrossRef][Web of Science][Medline]

16. Patton GC, Glover S, Butler H, Bond L, Bowes G. The Gatehouse Project: Building the capacity for mental health promotion in secondary schools. Aust NZ J Psych. 2000;34:586–593.[CrossRef]

17. Bond L, Glover S, Godfrey C, Butler H, Patton GC. Measuring system-level change in schools: lessons from the Gatehouse Project. Health Educ Res. 2001;28: 368–383.

18. Bond L, Patton GC, Glover S, Carlin JB, Butler H, Thomas L et al. The Gatehouse Project: can a multilevel school intervention affect emotional well being and health risk behaviours? J Epi Commun Health. 2004;58: 997–1003.[CrossRef]

19. Hilton ME. A comparison of a prospective diary and two summary recall techniques for recording alcohol consumption. Br J Addict. 1989;84:1085–1092.[CrossRef][Web of Science][Medline]

20. Escobedo LG, Marcus SE, Holtsman D, Giovino GA. Sports participation, age at smoking initiation and the risk of smoking among US high-school students. JAMA. 1993;269:1391–1395.[Abstract/Free Full Text]

21. Moffitt TE, Silva PA. Self-reported delinquency: Results from an instrument for New Zealand. Aust NZ J Criminol. 1988;21:227–240.

22. Lewis G, Pelosi AJ. The manual of CIS-R. London: Institute of Psychiatry; 1992.

23. Lewis G, Pelosi AJ, Araya R, Dunn G. Measuring psychiatric disorder in the community: a standardized assessment for use by lay interviewers. Psychol Med. 1992;22:465–486.[Web of Science][Medline]

24. Angold A, Costello EJ, Messer EC, Pickles A, Winder F, Silver D. Development of a short questionnaire for use in epidemiological studies of depression in children and adolescents. Int J Methods Psych Res. 1995;5:237–249.

25. Arthur MW, Hawkins JA, Pollard RF, Catalano RF, Baglioni, AJ. Measuring risk and protective factors for substance use, delinquency, and other adolescent problem behaviours: the communities that care youth survey. Evaluation Review. 2002;26:575–601.[Abstract/Free Full Text]

26. Carlin JB, Wolfe R, Coffey C, Patton GC. Analysis of binary outcomes in longitudinal studies using weighted estimating equations and discrete-time survival methods: prevalence and incidence of smoking in an adolescent cohort. Statistics in Medicine. 1999;18: 2655–2679.[CrossRef][Web of Science][Medline]

27. Long JS, Freese J. Regression models for categorical variables using stata. Texas: Stata Press Publications; 2001.

28. Tobler NS, Stratton HH. Effectiveness of school-based drug prevention programs: a meta-analysis of the research. J Primary Prev. 1997;18:71–128.[CrossRef]

29. Thomas R, Bushby K. School based programmes for preventing smoking (protocol for a Cochrane Review). The Cochrane Library, ed. 1999. Oxford.

30. Campbell M, Fitzpatrick R, Haines A, Kinmouth AL, Sandercock P, Spiegalhalter D, et al. Framework for design and evaluation of complex interventions to improve health. BMJ. 2000;321:694–696.[Free Full Text]

This article has been cited by other articles:

				P. Hawe and L. Ghali Use of social network analysis to map the social relationships of staff and teachers at school Health Educ. Res., February 1, 2008; 23(1): 62 - 69. [Abstract] [Full Text] [PDF]

				C Bonell, A Fletcher, and J McCambridge Improving school ethos may reduce substance misuse and teenage pregnancy BMJ, March 24, 2007; 334(7594): 614 - 616. [Full Text] [PDF]

				E. G. Jessop Another public health triumph J. Public Health Med., December 1, 2006; 28(4): 297 - 298. [Full Text] [PDF]

This Article Abstract Figures Only Full Text (PDF) All Versions of this Article: AJPH.2004.047399v1 96/9/1582    most recent Submit a response View Shopping Cart Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (7) Citing Articles via Google Scholar Google Scholar Articles by Patton, G. C. Articles by Bowes, G. Search for Related Content PubMed PubMed Citation Articles by Patton, G. C. Articles by Bowes, G. Related Collections School Health Prevention Surveys Alcohol Drugs Other Tobacco